Low Temperature Environment Operations of Turboengines

0 Qo B n Y n 1c AGARD 2 ADVISORY GROUP FOR AEROSPACE RESEARCH & DEVELOPMENT 3 7 RUE ANCELLE 92200 NEUILLY SUR SEINE FRANCE

AGARD CONFERENCE PROCEEDINGS 480 Low Temperature Environment Operations of Turboengines (Design and User's Problems) Fonctionnement des Turborkacteurs en Environnement Basse Tempkrature (Problkmes Pos& aux Concepteurs et aux Utilisateurs)

processed I /by 'IMs ...... signed-...... date ......

NOT FOR DESTRUCTION

- NORTH ATLANTIC TREATY ORGANIZATION

Distribution and Availability on Back Cover

AGARD-CP-480

--I-

ADVISORY GROUP FOR AEROSPACE RESEARCH & DEVELOPMENT 7 RUE ANCELLE 92200 NEUILLY SUR SEINE FRANCE

AGARD CONFERENCE PROCEEDINGS 480 Low Temperature Environment Operations of Turboengines (Design and User's Problems) Fonctionnement des TurborLacteurs en Environnement Basse Tempkrature (Problkmes PoSes aux Concepteurs et aux Utilisateurs)

Papers presented at the Propulsion and Energetics Panel 76th Symposium held in Brussels, Belgium, 8th-12th October 1990.

- North Atlantic Treaty Organization --q Organisation du Traite de I'Atlantique Nord I The Mission of AGARD

According to its Chartcr, the mission of AGARD is to bring together the leading personalities of the NATO nations in the fields of science and technology relating to aerospace for the following purposes:

-Recommending effective ways for the member nations to use their research and development capabilities for the common benefit of the NATO community;

- Providing scientific and technical advice and assistance to the Military Committee in the field of aerospace research and development (with particular regard to its military application);

- Continuously stimulating advances in the aerospace sciences relevant to strengthening the common defence posture;

- Improving the co-operation among member nations in aerospace research and development;

- Exchange of scientific and technical information;

- Providing assistance to member nations for the purpose of increasing their scientific and technical potential;

- Rendering scientific and technical assistance, as requested, to other NATO bodies and to member nations in connection with research and development problems in the aerospace field.

The highest authority within AGARD is the National Delegates Board consisting of officiallyappointed senior representatives from each member nation. The mission of AGARD is carried out through the Panels which are composed of experts appointed by the National Delegates, the Consultant and Exchange Programme and the Aerospa.ce Applications Studies Programme. The results of AGARD work are reported to the member nations and the NATO Authorities through the AGARD serics of publications of which this is one.

Participation in AGARD activities is by invitation only and is normally limited to citizens of the NATO nations.

The contcnt of this publication has been reproduced directly from material supplied by AGARD or the authors.

ISBN 92-835-0618-9

Primed by Specialised Printing Services Limitd 40 Chigwell Lune, Loughton, Essex IGIO 3TZ Recent Publications of the Propulsion and Energetics Panel

CONFERENCE PROCEEDINGS (CP) Viscous Effects in Turbomachines AGARD CP 351, September 1983 Auxiliary Power Systems AGARD CP 352, September 1983 Combustion Problems in Turbine Engines AGARD CP 353, January 1984 Harard Studies for Solid Propellant Rocket Motors AGARD CP 367, September 1984 Engine Cyclic Durability by Analysis and Testing AGARD CP 368, September 1984 Gears and Power Transmission Systems for Helicopters and Turboprops AGARD CP 369, January 1985 Heat Transfer and Cooling in Gas Turbines AGARD CP 390, September 1985 Smokeless Propellants AGARD CP 391, January 1986 Interior Ballistics of Guns AGARD CP 392, January 1986 Advanced Instrumentation for Aero Engine Components AGARD CP 399,November 1986 Engine Response to Distorted Inflow Conditions AGARD CP 400, March 1987 Transonic and Supersonic Phenomena in Turbomachines AGARDCP401,March 1987 Advanced Technology for Aero Engine Components AGARD CP 421, September 1987 Combustion and Fuels in Gas Turbine Engines AGARD CP 422, June 1988 Engine Condition Monitoring - Technology and Experience AGARD CP 448, October 1988 Application of Advanced Material for Turbomachinery and Rocket Propulsion AGARD CP 449, March 1989 Combustion Instabilities in Liquid-Fuelled Propulsion Systems AGARDCP450,Aprill989 Aircraft Fire Safety AGARD CP 467, October 1989 Unsteady Aerodynamic Phenomena in Turbomachines AGARD CP468,February 1990 Secondary Flows in Turbomachines AGARD CP 469, February 1990 Hypersonic Combined Cycle Propulsion AGARD CP 479, December 1990 Low Temperature Environment Operations of Turboengines (Design and User’s Problems) AGARD CP480,May 1991

... 111 ADVISORY REPORTS (AR) Through Flow Calculations in Axial Turbomachines (Results of Working Group 12) AGARD AR 175, October 1981 Alternative Jet Engine Fuels (Results of Working Group 13) AGARDAR 181,Vol.l andVo1.2,July 1982 Suitable Averaging Techniques in Non-Uniform Internal Flows (Results of Working Group 14) AGARD AR 182 (in English and French), June/August 1983 Producibility and Cost Studies of Aviation Kerosines (Results of Working Group 16) AGARD AR 227, June 1985 Performance of Rocket Motors with Metallized Propellants (Results of Working Group 17) AGARD AR 230, September 1986 Recommended Practices for Measurement of Gas Path Pressures and Temperatures for Performance Assessment of Aircraft Turbine Engines and Components (Results of Working Group 19) AGARD AR 245, June 1990 The Uniform Engine Test Programme (Results of Working Group 15) AGARD AR 248, February 1990 Test Cases for Computation of Internal Flows in Aero Engine Components (Results of Working Group 18) AGARD AR 275, July 1990

LECTURE SERIES (LS) Operation and Performance Measurement of Engines in Sea Level Test Facilities AGARD LS 132, April 1984 Ramjet and Ramrocket Propulsion Systems for Missiles AGARD LS 136, September 1984 3-D Computation Techniques Applied to Internal Flows in Propulsion Systems AGARD LS 140, June 1985 Engine Airframe Integration for Rotorcraft AGARD LS 148, June 1986 Design Methods Used in Solid Rocket Motors AGARD LS 150, April 1987 AGARD LS 150 (Revised),April1988 Blading Design for Axial Turbomachines AGARD LS 167, June 1989 Comparative Engine Performance Measurements AGARD LS 169, May 1990

AGARDOGRAPHS (AG) Manual for Aeroelasticity in Turbomachines AGARD AG 298/1, March 1987 AGARD AG 298/2, June 1988 Measurement Uncertainty within the Uniform Engine Test Programme AGARD AG 307, May 1989 Hazard Studies for Solid Propellant Rocket Motors AGARD AG 316,September 1990 REPORTS (R) Application of Modified Loss and Deviation Correlations to Transonic Axial Compressors AGARD R 745, November 1987 Rotorcraft Drivetrain Life Safety and Reliability AGARD R 775. June 1990

iv Theme

The low temperature environment and its impact on aircraft propulsion reliability continues to be of major concern to military and commercial aviation, The Propulsion and Energetics Panel and Flight Mechanics Panel have sponsored Specialists' Meetings and Symposia in the past directed at problems and challenges conceming cold weather operation. This Symposium on Low Temperature Environment Operations of Turbo-Engines is particularly relevant now because in recent years engine and component design technology advancement permits better accommodation of cold weather variables in engine development and anti-icing design considerations.

This Symposium will address engine design and user's problems as well as new testing technologies. The Symposium encompasses four sessions:

- Session I deals with user's requirements and operational experience. - Session II addresses engine starting conditions, improvements in engine performance and reliability due to electronic engine control, life cycle management, and specific examples of anti-icing design methods and results. - Session Ill addresses fuels and lubricants and their performance at low temperature.

Icing conditions, analytical prediction models, testing technology, and recent test results are included in Session N.

In summary, this Symposium ties together the requirements and operational community and the engine design and testing community while presenting a balanced analytical and empirical view of the state-of-the-art.

Theme

L'euvironnement basse tempirature et son impact sur la fiabiliti des systkmes de propulsion des aironefs reste I'une des prioccupations majeures de la communauti de l'aviation civile et militaire.

Dans le passi, le! Panels AGARD de propulsion et d'Energetique et de la Micanique du Vol ont organid des reunions de spicialistes et des symposia sur 1es problkmes poses et les difis soulevis par le fonctionnement des moteurs d'avion par temps froid.

Ce symposium sur le fonctionnement des turboriacteurs en environnement basse tempirature est particulikremeut pertinente aujord'hui puisque les progrb rialises ricemment dans les technologies de conception des moteurs et des composants out permis une meilleure prise en compte des variables liies au temps froid daus le developpement des riacteurs et la conception des systemes d'antigivrage.

Le symposium examinera 1es problhes rencontres par les concepteurs et les utilisateurs des turboriacteurs, ainsi que les nouvelles technologies d'essai. II est organid en quatre sessions:

- La Session I porte sur les besoins des utilisateurs et I'experience opirationnelle - La Session II concerne les conditions de mise en route des reacteurs, les amiliorations apportees aux performances et a la fiabiliti des moteurs grHce a la r6gulation ilectronique riacteur et a la gestion de la durie d'utilisation, et propose quelques exemples spicifiques de methodes et de rbultats dans le domaine de la conception des systkmes d'antigivrage. - La Session III examine les carburants et les lubrifiants et leurs performances B basse temperature. - La Session N traite des conditions de givrage, les modkles pridictifs analytiques, les technologies dessai, et presente des risultats d'essais recents.

En resumi, ce symposium a pour objet de reunir la communauti des exigences opirationnelles et celle de I'itude et des essais des riacteurs, en presentant une synthbe iquilibrie, a la fois analytique et empirique, de I'etat de I'art dam ce domaine. Propulsion and Energetics Panel

Chairman: M. 1'Ing. Princ. de I'Armcment Ph.Ramette Deputy Chairman: Prof. Dr A. User Sociiti Europienne de Propulsion Middle East Technical University 24, rue Salomon de Rothschild ODTU

BP 303 ~ 92156 Suresnes Cedex Makina Muh. Bolumu France Ankara, Turkey

PROGRAMME COMMITTEE

Mr William W.Wagner (Chairman) Ing. de I'Armement Christophe Meyer Technical Director (Code TD) Service Technique des Programmes Naval Air Propulsion Center Aironautiques (Code PE 3), PO. Box 7176 4, avenue de la Porte d'Issy Trenton, New Jersey 08628-0176 00460 Armees United States France Dr Robert C.Bill Mr Manuel Mulero Valenzuela Chief, Engine & Transmission Div. Departamento de Motopropulsion y US Army Propulsion Directorate Energia, (INTA) (AVSCOM) Mail Stop 77-12 Crta. Torrcjon a Ajalvir, Km. 4 21000 Brookpark Road 28850 Torrejon de iirdoz, Madrid Cleveland, Ohio 44135 Spain United States Mr Don M.Rudnitski Major Ibrahim Corbacioglu Head, Engine Laboratory 1 NCI hava Ikmal Ve Bakim Division of Mechanical Engineering Merkeri K.ligi National Research Council of Canada Eskisehir Ottawa, Ontario K1A OR6 Turkey Canada Professor Dr Dietmar K.Hennecke Mr David J.Way Fachgebiet Flugantriehe PN2 Division Technischc Hochschule Darmstadt Propulsion Department Petersenstrasse 30 Royal Aerospace Establishment (Pyestock) 6100 Darmstadt Farnborough Germany Hants GU14 0LS United Kingdom Professor Rene Jacques Ecole Royale Militaire 30, avenue de la Renaissance 1040 Bruxelles Belgium

HOST NATION COORDINATOR Major L.Gabrie1

PANEL EXECUTIVE Mr Gerhard Gruher

Mail from Europe: Mail from US and Canada: AGARD-OTAN AGARD-NATO Attn: PEP Attn: PEP 7 rue Ancelle APO New York 09777 92200 Neuillv sur Seine France Telephone: 33 (1) 4738-5785 Telex: 610176 (France) Telefax: 33 (1) 4738-5799

ACKNOWLEDGEMENT

The Propulsion and Energetics Panel wishes to express its thanks to the National Authorities from Belgium for the invitation to hold this meeting in Brussels, and for the facilities and personnel which make the meeting possible.

vi Contents

Page

Recent Publications of PEP iii

Theme/Theme V

Propulsion and Energetics Panel vi

Reference

SESSION I - COLD WEATHER OPERATIONAL EXPERIENCE AND REQUIREMENTS

Low Temperature Environment Operation of Turbo Engines - 1 A Military Operator’s Experience and Requirements by MSummerton

Canadian Forces Cold Weather Experience in the Maintenance and Operation of 2 Figher Engines by C.Ouellette

Analyse des Problemes de Demarrage par Temps Froid avec 3 les Turbomoteurs d’H61icoptbre de Type ASTAZOU par W.Pieters

Avions d’ANaires Mystere-FALCON - 4 Experience Operationelle par Temps Froid par C.Domenc

Vulnerability of a Small Powerplant to Wet Snow Conditions by R.Meijn

Ice Tolerant Engine Inlet Screens for CHI 13/113A Search and Rescue Helicopters by R.Jones and W.A.Lucier

SESSION I1 -SYSTEM DESIGN CONSIDERATIONS

Cold Starting Small Gas Turbines - An Overview by C.Rodgers

Cold Start Optimization on a Military Jet Engine by H.Gruber

Cold Weather Ignition Characteristics of Advanced Small Gas Turbine Combustion Systems by SSampath and LCritchley

Cold Weather Jet Engine Starting Strategies Made 10 Possible by Engine Digital Control Systems by R.C.Wibbelsman

Cold Start Investigation of an APU with Annular Combustor and Fuel Vaporizers 11 by K.H.Collin

vii Reference

Control System Design Considerations for Starting Turbo-Engines 12 during Cold Weather Operation by R.Pollak

Cold Start Development of Modern Small Gas Turbine Engines at PWC 13 by D.Breitman and F.Yeung

Design Considerations based upon Low Temperature Starting Tests 14 on Military Aircraft Turbo-Engines by H.-F.Feig

Climatic Considerations in the Life Cycle Management ofthe CF-18 Engine 15 by R.W.Cue and D.E.Muir

Application of a Water Droplet Trajectory Prediction Code to the Design 16 of Inlet Particle Separator Anti-Icing Systems by D.L.Mann and S.C.Tan

Captation de Glace sur une Aube de Prerotation d’Entrke d’Air 17 par R.Henry et D.Guffond

Development of an Anti-Icing System for the T800-LHT-800 Turboshaft IEngine 18 by G.V.Bianchini

Engine Icing Criticality Assessment 19 by E.Brook

Ice Ingestion Experience on a Small Turboprop Engine 20 by L.W.Blair, R.L.Miller and D.J.Tapparo

SESSION 111 - FUEL EFFECTS AND LUBRICANTS BEHAVIOUR

Fuels and Oils as Factors in the Operation of Aero Gas Turbine Engines 21 at Low Temperatures by G.L.Batchelor

The Effect of Fuel Properties and Atomization on Low Temperature Ignition 22 in Gas Turbine Combustors by D.W.Naegeli, L.G.Dodge and C.A.Moses

Givrage des Circuits de Carburant des Turboreacteurs 23 par F.Garnier

The Influence of Fuel Characteristics on Heterogeneous Flame Propagation 24 by M.F.Bardon, J.E.D.Gauthier and V.K.Rao

The Development of a Computational Model to Predict Low Temperature 25 Fuel Flow Phenomena by R.A.Kamin, C.J.Nowack and B.A.Olmstead

Paper 26 withdrawn

SESSION IV - ICING CONDITIONS AND TESTING

Paper 27 withdrawn

Environmental Icing Testing at the Naval Air Propulsion Center 28 by W.H.Reardon and V.J.Truglio

viii Reference

Icing Research Related to Engine Icing Characteristics 29 by S.J.Riley

Modelisation Numerique de I’Evolution d’un Nuage de Gouttelettes d’Eau 30 en Surfusion dans un Caisson Givrant par P.Creismeas et J.Courquet

Icing Test Capabilities for Aircraft Propulsion Systems at the 31 Arnold Engineering Development Center by C.S.Bartlett, J.R.Moore, N.S.Weinberg and T.D.Garretson

Icing Test Programmes and Techniques 32 by E.Carr and D.Woodhouse

A Documentation of Vertical and Horizontal Aircraft Soundings 33 of Icing Relevant Cloudphysical Parameters by H.-E.Hoffmann

Developments in Icing Test Techniques for Aerospace Applications 34 in the RAE Pyestock Altitude Test Facility by M.Holmes, V.E.W.Garratt and R.G.T.Drage

Entrke d’Air d’H6licopteres: Protection pour le Vol en Conditions 35 Neigeuses ou Givrantes par X.de la Servette et PCabrit

1-1

LQW TKMPERATURB ENVIRONMENT OPERATION OF TURBO ENGINES - A MILITARY OPERATOR'S EXPERIENCE AND RgQUIm by Lieutenant Colonel M Summerton CEng MRAeS REME School of Aeronautical Engineering Army Air Corps Centre Middle Wallop Stockbridge Hampshire SO20 8DY

-SUMMARY manufacture. Above all, aircraft should be as reliable and .maintenance-free as possible, but The United Kbngdom's commitment to NATO includes where maintenance is unavoidable, it should be the regular use of Royal Marine and Army helicop- capable of being completed in the shortest ters in low temperature conditions. This paper possible time, with the minimum quantities specifically addresses the Operation of the of tools and equipment. TO understand why, Westland LYNX helicopter with its Rolls Royce consider how ow Royal Marine squadron operates in GEM engines during winter deployments in Norway Norway. where the near-arctic conditions present certain operating and working difficulties. Operating Conditions. Initially, the Marines operate from ships, flying off pitching decks This paper considers these difficulties both regularly doused with cold sea-spray. Having generally, from a human and physical point of view, landed all of their vehicles. equipment and and then more specifically with regard to the supplies, the road parties deploy inland to engines themselves. Finally, it concludes with a several different field locations, sometimes in few areas for improvement. with the emphasis on wooded areas, and sometimes in and around villages reliability, ease of maintenance, and effective or towns. Once in position they are joined by development and testing before entry into service. their aircraft, which tend to end up in deep, fresh snow, unlike the vehicles which are no~mally INTRODUCTION sited on, or close to, cleared tracks and roads. The aim of this paper is to help set the Once inland, away from the moderating influence of scene, as a prelude to more detailed and esoteric the relatively warm sea. and perhaps well above discussion of low temperature engine problems. sea level, they experience a variety of weather The papep has been kept deliberately short and conditions including: essentially practical. and does not seek to investigate and solve the relatively few problems a. Day-time temperatwes in the region -10 that the British Army experiences. to -15OC. b. Cold soak night-time temperatures down to The paper deals exclusively with helicopter -3O'C. operations, conducted from field locations with- C. Driving snow which drifts and builds up out the benefit of hangarage. on flat surfaces. d. Cloud, mist and rain. BACKGROUND e. Freezing fog and freezing rain. f. Wind. The British APmy has some 360 aircraft, the majority of which are LYNX and GAZELLE helicop- The P~~cTIc?~3ifficulties.. Operacing in rhc ters based in the United Kingdom and West Germany. above condirions, rherc arc many pracricill dif'li- This fleet includes a squadron of aircraft which culties. Keeping warm becomes a major preoccu- are operated by the Royal Marines who are a part pation, especially if there is any wind, where of the Royal Navy. Only a moderate breeze Can reduce a temperature of -7°C down to -24OC. At such temperatures, there This Royal Marine squadron, together with an is a progressively worsening risk of frostbite, Army flight of six GAZELLES, deploys from which at the lower end can t-esult in flesh freez- January to March each year to Norway where they ing within a minute. Such a threat requires participate in NATO exercises. engineers to work in pairs on a buddy-buddy system, to monitor each other for any signs of These deployments constitute the majority of our frostbite or cold BXPOSUPB. Low temperatures can cold weather experience but we also get a also cause instant adhesion of bare skin to very certain amount from our aircraft based in West cold surfaces, and cold burns. At temperatures Germany, though the conditions there rarely as 'mild' as -1O'C, bare metal contact is quite approach the severity of Norway. Neither of painful. these areas compare with places like Canada and the arctic regions, however, but with winter In Such conditions, effective protective clothing temperatures ranging from t5 to -30°C, with freez- is absolutely essential, but unfortunately, a ing rain, snow and ice, the climatic conditions are glove that is warm is guaranteed to be totally considered more than representative of the unsuitable for detailed technical work such as problems. adjusting engine controls and wire-locking components. THE NATURE OF WINTER OPERATIONS The obvious answe~to the problem is to move the When considering low temperature operations, it is aircraft under cover whenever you need to carry important for those associated with the specifica- out Significant work, but this can only be done if tion and design of aircraft. to understand the there are suitable large buildings close to the essential practical nature of Such operations. It landing site, and where the route to the buildings is important because winter conditions present is sufficiently flat, and clear of snow and ice. problems which are unique and which tend to aggra- In practice, the work involved in getting an vate any fundamental weaknesses of design and aircraft under cover, is not normally 1-2

justifiable, but it can be done if the need is The act of having to Pig up the heat source and great enough. the time it takes, however, is unacceptable.

The alternative is to leave the aimraft where Cold soak can also lead to the freezing and stiff- it is and rig a shelter over the area being worked ness of engine controls, especially teleflev on, and apply local heat. This still requires controls which may have got moisture inside them. considerable effort, however, and it creates a huge thermal signature which could badly compromise a unit's location. -Starting. Succe:;sful starting is almost exclusively a function of adequate levels of Besides personal protection, there is also a need electrical power. If we have well-charged air- to protect the aircraft, with covers. However, the craft batteries or we use an external DC supply, act of fitting the covers can represent quite a then we find the GEM engine relatively easy to difficult task in itself, especially if they happen start. If, however, electrical power is low then to be covered in snow and ice from previous use. we experience: The problem is compounded by the need to apply camouflage netting and by deep snow, where the a. Stagnation/failure to start. aircraft will settle on its skis at one height, but all ground activity is conducted 30 to 40 C~S b. Excessive light-up times, (30-50 S~CS lower. Pather than %% secs).

Other practical problems also pPeSent themselves: C. Wet starts and excessive turbine temperatures. a. It is difficult getting tools and equipment Out to the aircraft, especially Operating from woads, in deep snow, and pmbably in a full tactical setting where they will be scattered over a wide area, arpanging external widely dispersed around a location. power from either a vehicle or a ground Pig can at best be unacceptably slow and inconvenient, and at b. If tools and equipment are dropped in worst, impossible. Our' aircraft may also be the snow they can be very difficult to find. operating totally remote from such support. We therefore prefer to use internal battery power as C. Before working on the aircraft you the norm. Howevei-, after overnight exposure to have to remove portions of the camouflage temperatures of -LODC and below, or a 2U hour soak and protective covers. at even milder temperatures, we find that many of our batteries have only marginal power for suc- d. When you climb onto an aircraft you cessful starting. The only solution in this case must be careful to remove all snow and ice is to anticipate conditions and remove the battery from your' boots, otherwise you can slip very to a warn area, but this in turn can impose a easily. technical burden that is even WOPS~than the use of external power. e. Protective hoods tend to restrict peripheral vision and hearing, making you mope The problem of low battery power is almost cer- prone to dmpping tools and banging your head tainly aggravated by increased engine turning on aircraft structure. Likewise, the wearing resistance. We a~twayscheck our engines for free of NBC equipment is especially limiting. rotation in case

Of far greater concern is fundamental engine reli- In all cases of difficult starting, slight ability and maintainability, because, as illustra- advancement of tho engine speed select lever ted above, any technical work that has to be beyond the 'Ground Idle' gate normally helps, but carried out on an aircraft, in cold conditions, is TG limits need careful monitoring. extremely difficult. Snow and Ice The most significant limitation on Having put the subject in its comect context, let the use of helicopters in cold weather is the us now consider the problems that we do have. hazard caused by :;now and ice. In our Ministry of Defence, the various climatic conditions in which Cold Soak. When an aircraft has been cold-soaked our aircraft may be required to operate, are below -2G°C, we have to heat the engines and gear- defined in DEFENCIl STANDARD 00-970. box to bring the main rotor gearbox sump tempera- These standards zmdatorily require operation in ture back to -2GOC before we can attempt a start. 1-3 temperatures down to -26OC and no damage from cold engines, to minimise the mount of work that soak down to -4OOC. On the icing side, the stan- has to be carried out in low temperature dard defines nine different conditions that an conditions. aircraft may be required to operate in. These conditionscan be summarised as: b. Lower cold-start limits to avoid any necessity for having to pre-heat engines and a. Clear air ice. gearboxes.

b. Mixed snow and ice. C. Better batteries that are less p.wne to losing their charge, or an alternative design C. Falling snow. of autonomous starting system which is lesa dependent on battepy condition. d. Re-circulating enow (caused by hovering in ground effect). d. In conjunction with the above, even more dependable starting. e. Freezing fog. e. Fully integrated airframe and engine in- f. Freezing rain/drizzle. take design to pl'ovide full airframe, transmission and engine protection from snow, In practice, we find that unless the weather is slush and ice. almost clear blue skies, then we are greatly limited in when we can fly. These limitations are due f. More thorough specification, development to: and cold weather testing before delivery to service. a. A general lack of visibility. The state-of-the-art on helicopters may still not b. Ice accretion on rotors which destroys allow us to achieve, sufficiently economically, lift and causes severe vibration. the level of snow and ice performance to which we would ideally aspire, but what is quite clear to C. General ice accretion on the airframe us, is that between OUI. ministry and industry, we which, for example, might break off and dm- must specify and actually achieve, realistic age a tail mtor, 011 enter an engine as a requirements that sensibly reflect the operational foreign object. need. In the meantime, in the context of both temperate and cold weather operations, we must d. Ice accretion on or around engine in- continue to strive to make ow aircraft and takes, restricting air flow, or threatening engines as reliable and maintainable as possible. ingestion.

e. Snow and slush deposits which threaten Discussion ingestion. 1. W. Wagner, US Navy We are advised that an ordinary-sized snowball is capable of flaming-out a GEM engine, and that a Please address starting limitations concerning lubricants or 15cc lump Of ice can cause serious damage to the viscosity problems at low temperature. engine's axial flow, LP compressor. Which problems effect engines or power drive system components? The impression that we have gained, as operators, erroneously 011 not, is that we specify Author: mandatory temperature requirements, but we do not appear to specify and achieve particular We have no lubricant problem with the GEM engine in cold snow and ice requirements. AS a consequence, weather. Oil temperatures and pressures have to be when we conduct our cold weather testing, to monitored during start-up, and they take longer to stabilize, establish in-service operating limits, we end but this does not cause any problems. up having to restrict the aircraft's opera- We do not change engine oil when operating at low tion to what we find, rather than just con- firming what the designer has achieved in temperatures but we do change transmission oil to a lower satisfaction of a contract specification. viscosity type. A check of serviceability statistics reveals no detectable In the case of the LYNX, this has resulted in difference between temperate and cold weather operations. us having spent many years trying to develop a satisfactory snow and ice guard, capable of extending the aircraft's snow and ice clear- 2. W. Wagner, US Navy ances to reasonably acceptable operational Please identify the engine design problems which this limits. This has now been achieved. From assemblage of design and manufacturing community could this experience we would note the extent to which the engine designer is very much depen- address. dant on the aircraft designer, to achieve a satisfactory intake design, and further, we Author: would observe that an aircraft's snow and ice Fundamental reliability is the most important requirement clearance is only as good as its worst both in temperate and in cold weather conditions. feature. We hope that both our miniStPy and industry have learnt the appropriate lessons. A properly integrated airframe and engine inlet design is of equal importance, but whether you start with the CONCLUSIONS assumptions of an ice-tolerant engine or not is a question of philosophy. If you have a robust, ice tolerant engine, then the In COnCluSiOn, based on OUP experience of the integration of the engine into the airframe is not as critical, LYNX/GEM combination, we would like to see the following general improvements: and the protection afforded to the engine can be reduced. Neither of our two main helicopters have fundamental a. Even more reliable and maintainable engine design problems viz-a-viz cold weather operation.

2- I

LOW TEMPERATURE ENVIRONMENT OPERATIONS OF TURBO ENGINES

by Lt Christian Ouellette Mechanical and Propulsion Engineering Officer Base Aircraft Maintenance Engineering Organization Canadian Forces Base Cold Lake Medley, Alberta Canada

INTRODUCTION OPERATIONAL ROLE Good morning Ladies and Gentlemen, Bonjour Madames Our Air Force utilizes the CF-5 Freedom Fighter as our et Messieurs, my name is Lt Christian Ouellette and I am “Basic Fighter Pilot Trainer”. We operate about 40 CF-5 here representing the Operational and Maintenance aircraft out of 419 Sqn at CFB Cold Lake. The nucleus of community of the Canadian Armed Forces, and more our air defence posture is our front line Fighter and specifically the branch of Air Command. I am the Interceptor, the CF-18 Hornet. A total of 125 CF-18s are Mechanical and Propulsion Engineering Officer at spread amongst 8 squadrons, located in Baden-Soelligen- Canadian Forces Base Cold Lake, situated in the Central Germany, Bagotville-Quebec and Cold Lake- Alberta. Eastern portion of the province of Alberta. Our operational commitments are as follows: It is my distinct pleasure to share with you some of our (a) Basic Fighter Pilot Training at 419 Sqn on the CF-5, experiences in the maintenance and operation of Turbo and at 410 Sqn on the CF-18; Engines under cold weather conditions. @) NORAD peacetimealert roleand wartimedeployment Canada is a vast country, bordered by two oceans and the capability; arctic circle. Our Armed Forces, with its varied commitments of Peacekeeping, the North Atlantic Treaty (c) Air to Air Intercept; Organization (NATO) and the North American Air (d) Air Superiority and NATO training; and Defence (NORAD) plan, have the unenviable task of patrolling and protecting the vast reaches of the North (e) Air to Air/Air to Ground peacetime training. American Northern Region. At CFB Cold Lake, our 3 CF-18 Squadrons combined, over Never a dull moment it seems. From the time we first turn the winter months of November to February, fly an average over an aircraft at dawn, our missions vary from hunting of 860 sorties per month. 419 Sqn alone, with the CF-5, down the odd arctic Bear-H, to escorting MIG-29s who averages 490 sorties per month during the cold weather have a tendency to wander and lose their way. season. This morning’s presentation will cover the following areas: (a) The climate conditions which we are faced with in Canada; MAINTENANCE PROBLEMS AND PRACTICES (b) a summary of our operational role and commitments; Severe weather conditions pose a series of the challenging problems to the maintenance community. The types of (c) a brief look at some maintenance problems and obstacles encountered are as follows: practices associated with the cold weather environment; (a) Personnel Protection Against Cold When servicing/starting aircraft in extreme conditions, (d) the “Hung Start” problem associated with CF-18, safety of personnel is first and foremost. Crew GE-F404 engine; and members must protect themselves from the bitter (e) a quick review on the status of the infamous temperatures by adding layer upon layer of clothing, J-85-CAN-15 compressor stall problem. making it difficult at times to perform routine servicing tasks; CLIMATIC CONDITIONS (b) Power Take-Off Shaft Shearing Canadian Forces Base Cold Lake, appropriately named, is Special precautions must be observed on engines located along the 54th parallel. The following is a graph which drive exterior gearboxes (such as the accessory depicting the temperaturenormsoftheareaovera 12 month drive system on the CF-5). In extreme cold weather the period. Although our MEAN temperature low is only about oil in the gearbox gels and the drive shaft will shear on -18 deg C, we face ARCTIC type conditions for over 35% engine start. Usual precautions as follows: of the November to February time frame. Wind chill factors reach in excess of 1625 Watts per square meter, a point at (i) motor the engine prior to starting; which exposed flesh freezes. (ii) apply heat (portable heater) to the gearbox; (iii) run engine for minimum time before applying Inuvik, situated above the 68th parallel, is one of our four loads on generators, pumps. main Forward Operating Locations (FOLs). Weather conditions during the winter months are generally bitter, (c) Hydraulic Seal Leaks compounded further by constant prevailing winds from the Hydraulic system “0rings on cold soaked equipment East, driving temperatures down into extreme lows. The will tend to leak. The normal cure is to let the system outside temperature (excluding the wind chill factor) is warm up at idle, eliminating the leaks once the system below -20 deg C for over 150 days per year. warms up. 2-2

(d) Fire Extinguishers The following traces back the history of the stall problem, When the temperature drops to -40 deg C and below, outlining lessons learned, along with our present status: fire extinguishers may not work as advertised (particularly CO, extinguishers) for they do not Feb 75 Investigation by NRC showed that engine anti-ice generate enough pressure to adequately push out the increased the stall margin. cxtinguishant. At Cold Lake we have special extinguishers which we have identified by a broad blue Nov 76 The interrelationships between RPM and EGT band on the case. cutback, and T2 sensor outputs were discussed. The potential problem that incorrect T2 sensing (e) Fluid Servicing Carts could adversely affect stall characteristics was Servicing carts containing fluids may cause identified. contamination problems. Carts are generally stored in the hangar at a temperature of 70 deg F. While hangar Mar 78 Flight tests concluded and recommended a doors are often opened several times per day, the modification to activate anti-ice concurrent with temperature quickly drops to that of the outdoors. the Afterburner. During cold weather, the temperature of the cart shell (always metal) can be subject to temperature Jan 79 A temperature soaking procedure to prevent fluctuations of up to 100 deg F. This causes compressor stalls at very cold temepratures, mainly condensation within the cart and contaminates both during ground runs, was recommended and the cart and the system replenished by that Cart. In the implemented at 419 Squadron, until the Gas-Filled CF-5 world we get frequent complaints from SPAR, T2 sensor mod was installed. one of our 3rd line maintenance contractors, about water contamination in the CF-5 Accessory Drive May 82 Underwent flight testing of a modified T2 sensing System. The only solution to this problem is to check system, with a recommendation for relocation of the carts daily for water contamination, and drain the Resistance Temperature Detector (RTD) and water when found. adjustment of the MFCU to start RPM cutback at (0 Starting Problems -15 deg C. In general the gas turbine engine starts easier under cold weather conditions, however some problems do Nov 82 Testing of the automatic engine anti-ice occur. Failures to start can usually be attributed to engagement system, whereby anti-ice is selected either improper ignitor plug depth, burned outplugsor for 10 seconds following A/B initiation. Aircraft wetting of the plug (caused by introducing fuel into the were modified and a reduction in stall numbers was engine before firing the plug). recorded. (g) Overpower/Stalls Oct 84 MFCUs were biased by +30 deg F to compensate All gas turbine engines develop greater power in cold for inaccurate T2 sensing. Technique was temperatures, thereby becoming more prone to stall. somewhat effective in reducing stalls. While stall prevention measures will vary according to the equipment, general rules such as “No Erratic Oct 87 Modification CF-055, Gas Filled T2 sensor was Operation or Sudden Power Changes” will generally implemented. apply. HUNG START PROBLEM/F404-GE-400 ENGINE PRESENT STATtJS The CF-18, F404 Engine is experiencing periodic Hung Installed - auto engine anti-ice engagement system Start and Roll Back problems during cold weather - GFT2 sensor/modified T5 amplifier operations. As recently as last spring, while on deployment - various A01 and unit flying order to Inuvik, up to 6 hung starts per day were experienced, in an changes average temperature environment of -42 deg C. Present operating instructions call for motoring of the engine for 2 Modifications - automatic takeoff doors: automatically minutes prior to start. If a hung start is experienced, there are underway open for flight below M = 0.45 a series of checks to be carried out, verifying integrity to - P3 lag tube: damps out rapid electrical connections, density settings etc. If no fault found, fluctuations of the P3 tap, which is an motor engine for an additional 2 minutes and repeat input to the MFCU procedure until start occurs. If after repeated tries, the hung Projected - the definition of the true icing limits for start persists, remove and replace Main Fuel Control. The the aircraft, and the correct RPM situation in Inuvik often requires in excess of 15 minutes of cutback schedule engine cranking and throttle movement to achieve engine - considering a new replacement T5 start. amplifier Given the critical nature of operations at Forward Operating Locations, delayed aircraft starts are totally unacceptable. Further investigation into the Main Fuel Control start PRESENT STALL RATE SITUATION schedule is required and has been recommended. The following is a graph depicting the 5-85 Stall rate, per 1000 hrs, over a 3 year average. The total number of stalls CF-S/J-85-CAN-15 COMPRESSOR STALL HISTORY per year have gone from: Since the introduction into service of the CF-5 in the 1960s. the problem of suddcn compressor stalling has plagued the 1987 - 76 fleet. This phenomenon was first identified during cold 1988 - 53 weather evaluations of the aircraft in the late 196Os, and 1889 - 36 continues to be a problem to this day. 1990-evenless 2-3 Discussion 4. I? Sabla, GEAC What is the normal start-up procedure that permits plug wetting? 1. R. E. Smith, Sverdrup What were the fixes applied to make the fire extinguisher Author: units effective at low temperatures? Our present start-up procedure for the CF-5 is as follows - motor engines to 14.5% Author: - engage ignition The fire extinguishers acquired for cold weather conditions - advanced power lever, introduce fuel. are still CO, types, but with a greater built-in discharge pressure. How this is accomplished, I am not certain. Our wet start problem is not caused by an erroneous start However, we have identified these extinguishers with a wide procedure, but rather by pilot error in selecting power lever blue hand to ensure we do not use them in temperature too early. PILOT error will continue to be a problem, conditions above -26 C ambient. therefore perhaps the issue of a safeguard against accidental wetting can he addressed. 2. W. Alwang, Pratt and Whitney What was thenatureof theT2 sensing problemand how was How is failure to spark detected? it corrected? Author: Author: We rely on the inability of the engine to fire-up as an The T2 sensor was formally located near the under surface indicator of the aircraft where air is collected by two scoops, one for each engine, and is ducted to the MFCVs. The problem was 5. C. Rndgers, Sundstrand Power Systems that the total temperature of the sample air picked up by the Do all CAF aircraft fly from prepared bases? scoops was not representative for CIT, hence EGT and Do you provide portable heating for aircraft flying from engine cutback were noch occurring on schedule, increasing unprepared forward bases? the stall margin. The RTD sensor was relocated to an area in the engine bay Author: where a representative value of CIT could be obtained. Our fleet of CF-18 is based at prepared main operating Further sensing errors continued, so MFCV was biased to bases (FOLs) where we make use of facilities on site, which correct, and some improvement was seen. in the most cases are inadequate. In other words, storage Eventually the introduction of a gas-filled sensor eliminated facilities for both aircraft and equipment are in short supply. the sensing error and was successful in reducing the stall Although portable heat is available at FOLs, i.e. Herman rate. Nelson’s, conditions generally render them little or no use. Heating is generally used for personal rather than A/C. We 3. R. Toogood, Pratt and Whitney Canada prefer to conduct starting operations with cold soaked A/C, You have spoken at some length on the CFT stall problem. avoiding the problems of expansion and contraction Have you experienced any significant cold weather associated with introducing heat in severe conditions. problems as yet with the CF 18 aircraft? 6. W. Wagner, US Navy Author: You referred to hung starts at extreme low temperatures Our experience with the CF18 in cold weather is that the with CF 18 aircraft. Would today’s full authority digital aircraft performs better. Our only complaint is with the electronic controls eliminate this problem? scheduling of the MFCV during cold weather start-up. Our hung-start problem is critical during certain operations, and Comment by R. Wibbelsman, GE is considered totally unacceptable. This issue is apparently The floor indicated it would correct the situation, and the being addressed during paper 10. problem will be addressed in detail in paper 10 tomorrow.

3-1

ANALYSE DES PROBLEWS DE DEHARRAGE PAR TEMPS FROID AVEC LES TURBOMOTEURS D' HELICOPTERE DE TYPE ASTAZOU. par Lieutenant Ir. W. PIETERS Offr Maint 255 Cie Maint d' Aviation Leuere Flugplatz Butzweilerhof Butzweilerstrasse 5000 KOLN 30 RFA

RESUME

Pendant les periodes d' hiver assez severes au debut des annees 80. 1' armee helge a connu de considerables problemes de demarrage sur ses helicopteres de type ALOUETTE equipes de turbomoteurs ASTAZOU. Le rapport reprend les mbthodes de detection des phenomenes employees par les utilisateurs, les actions immediates prises au sein de 1' armhe et les solutions elaborees en collaboration avec les constructeurs ainsi que leurs consequences budgetairex.

SrrmW(Y During the heavy winter periods in the beginning of the 80th, the Belgian army had considerable starting problems on his helicopters ALOUETTE equiped with ASTAZOU turbo engines. The papers discuss the different detection methods of the phenomena employed by the users, the immediate actions undertaken by the army and the solutions worked out in collaboration with the constructors as well as their budgettary consequences.

NOUENCIATURE 255: Ahreviation utilis6 dans le texte, indiquant la 255 Compagnie Maintenance et Depat d' Aviation Legere, faisant 1' entretien des helicopteres de la Force Terrestre de 1' armhe belge. T4: Temperature a la sortie de la turbine. P2: Pression a la sortie du compresseur. RG: Revision gherale, allouant un nouveau potentiel au turbomoteur. VNIP: Visite NOn Interruptive de Potentiel, le potentiel du moteur reste le mOme apres son passage en usine. FB: Francs Belges. IT: Instruction Technique. FF: Francs FranCais (avec 1 FF=6,5 FB).

1. INTRODUCTION. a. situation de 1' armee belge. L' armee belge, stationnee partiellement en Allemagne de 1' Ouest et en Belgique, utilise des helicopthres pour repondre a 68s fonctions lui imposees dans le cadre des conventions de 1' 0TAN.Ainsi elle dispose actuellement de (situation claturee le 01 jan 90):

-FORCE TERRESTRE: 56 helicopthres ALOUETTE 11, dont 16 avec moteur ARTOUSTE et 40 avec moteur ASTAZOU.En utilisation depuis 1958, ils ont des missions d' observation et de 1iaison.Notons que 1' armbe a conclu recemment un contrat pour 1' achat de 46 helicopteres AGUSTA 109 dont 28 recevront une mission antichar et 18 auront une mission 6' observation et de 1iaison.Leur livraison est prevue des le mois de juin 1991.Une trentaine (32) d' ALOUETTES I1 resteront wand-m8me en service comme helicoptere de liaison , mais uniquement en version ASTAZOU. -FORCE AERIENNE: 5 helicopteres SEAKING preVUs specialement pour des op6rations de sauvetage en mer. -FORCE NAVALE: 3 helicopteres de type ALOUETTE 111, avec des missions de sauvetage en mer. -GENDARMERIE: 3 helicopteres de type PUMA. Les helicopteres de la Force Terrestre feront 1' objet de cet expos6.11~ sont utilises dans des conditions climatiques trhs variees, allant d' un climat maritime modere en Belgique a un climat continental plus vers 1' Est.En principe les conditions climatiques sont assez favorahles point de vue temperature , avec une humidit6 3-2

relativement Blev6e.Normalement, on ne connait suere de problemes dus a des temperatures extrhmement basses.Cependant, les annees 80 ont Bte marquees par une succession de quelques periodes d' hiver tres severes ainsi que quelques periodes nettement plus mod0rees.Pendant ces periodes d' hiver severes, des problhmes se sont manifest&. b.Position du probleme Pendant les periodes d' hiver au cours des annees 80, 1' armee belge a connu de considkrables problemes de demarrage sur ses helicopteres de 'type ALOUETTE I1 pourvu de turbomoteurs de type ASTAZOU I1 A2. L' expose reprendra toutes les phases du probleme, de sa detection jusqu' aux solutions elaborees ainsi que ses consequences budgetaires.

2. THEORIB ELEUENTAIRE DE LA PHASE DE DEMARRAGE DES UOTEURS ASTAZOU.

Le circuit de demarrage se compose de 4 parties (annexe A) -la micropompe (1) -le robinet 4 voies (2) -la prise d' air du carter turbine (3) -1es allumeurs-torches (4) Le fonctionnement est le suivant.D&s la mise en marche, la micropompe aspire le carburant et en premiere phase purge son circuit.Quand la pression s' eleve, le carburant est refoul6 vers le robinet 4 voies qui contient une bille.La pression du carburant repousse la bille sur son siege et ouvre ainsi la voie vers les allumeurs- torches en bouchant la sortie du kerosene vers la prise d' air P2 qui est une prise d' air totale a la sortie du compresseur centrifuge.Le kerosene est inject6 dans la chambre de combustion et enflamme par les allumeurs- torches. une fois que 1' allumage est realis&, la micropompe es:t coupee automatiquement; la pression carburant retombe B zero et la hille du robinet 4 voies est soulevee de son siege car d' un cat6 elle est soumise 2 la pression totale P2 et de 1' autre cat6 elle recoit la pression statique regnant au niveau des allumeurs-torches laquelle est plus basse.ce courant d' air seche les tuyauteries et Bvite 1' encrassage des allumeurs- torches par carbonisation du kerosene.

3. LA SURCHAUFFB AU DElULRRAGE. a.Temp6rature ideale de demarrage Le constructeur a defini dans son manuel de vol une temperature ideale de demarrage T4 de 450 OC pour moteur froid et de 550 Oc pour moteur chaud; on considere qu' un moteur est chaud quand la T4 residuelle est de plus de 100 et de moins de 150 0C.En respectant ces plages de temperature lors du demarrage, le bon fonctionnement du turbomoteur ne devrait pas Otre problematique b.Surchauffe au demarrage. Dans son manuel d' entretien edition D~c1987, le constructeur a, d' une manisre empirique, donne des temperatures de surchauffe au d6marrage.Ce sont des temperatures .i ne pas depasser lors du demarrage sous peine d' endommagement de la turbine.Voici les contrsles qu' il a prevu: SI T4 > 750 oc pendant t > 5 sec: Le moteur doit retourner en usine pour revision.

SI T4 > 750 OC pendant t < 5 sec: Le moteur peut Gtre maintenu en service apres ou T4 > 700 oc mais < 750 OC certains contrales specifiques la 255. La temperature de surchauffe depend principalement de la matiere de construction des aubes des disques de turbine. c.Les causes de surchauffe au demarrage Pour nos helicopteres, quelques causes "classiques" de surchauffe au demarrage sont connues. (1). Une tension de batterie trop faible provoque une tension primaire et secondaire d' allumage trop basse.L' allumage est retard6 tandis qu' un exces de carburant s' accumule dans le carter turbine.ce surplus de carburant: s' enflamme d' une maniere brutale lors de 1' allumage, il en resulte une temperature de demarrage excessive. (2). un fonctionnement deficient du robinet 4 voies.La bille se trouvant dans le corps du rcbinet peut se bloquer et perdre son etancheite par le givre ou par 1' encrassement.Ainsi, du carburant envoy6 par la micropompe peut entrer dans la chambre de combustion aussi bien par la prise d' air P2 que par les allumeurs- torches.11 en resulte un apport excessif de carburant par la prise de pression qui aboutit une surchauffe au demarrage lors de sa mise a feu. .(3). En dehors de ces causes primaires, il existe quelques causes secondaires de surchauffe au demarrage, notamment: 3-3

-La presence d' eau dans le carburant.L' Administration FedCrale de 1' Aviation indique qua une concentration de 30 ppm Peut Stre dangereuse car elle peut provoquer le blocage du robinet 4 voies lors du refroidissement du carburant. -Le bypass de la pompe de carburant mal regie pourrait provoquer Une pression de carburant trou &levee. -Un mauvais fonctionnement du regulateur barostatique qui permet la regulation du debit carburant en fonction de 1' altitude. -Me pression micropompe trop elevee. -Une entree d' air obstruee, mSme partiellement.

4. LBS SURCHAUPPES AU DBMARRAGE DANS LES ESCADR1LLBS.ANALYSB DES INCXDENTS ET ACCIDENTS. a.Les periodes d' hiver 1982/83 jusque 1984/85. L' annexe B represente un &tat recapitulatif des incidents et accidents pour lesquels un avis technique a et6 Btabli par la 255 pendant la periode 1982-1984.Sur un total de 70 helicopteres, on compte 23 accidents/incidents dont 5 dus aux surchauffes au d6marrage.Deux de ces incidents Btaient causes par un erreur de pilotage; les trois autres avaient une cause purement technique: le gel du robinet 4 voies par temps froid.Le coat d' une revision generale s' eleve B environ un million de FB (prix 85 non actualis&) sans compter 1' immobilisation de 1' helicoptere, les frais de depannage Bventuels, le main d' oeuvre....Il va de soi que des incidents successifs, comme ils se sont produits au debut 1985, sont tres lourds B supporter.Non seulement, ils consomment entre 15 et 20 % du budget prbvu pour 1' annee, mais 11s ont egalement une influence directe sur la disponibilite des machines et sur le planning des heures de vol. b.Analyse des problames. Les problbmes de surchauffe au demarrage etaient relativement bien connus.De temps en temps, des surchauffes se produisaient suite B une mauvaise manipulation du pilote ou B un defaut technique.Les deux cas mentionnes en 1982 en sont la preuve.La 255 avait d' ailleurs deja edit6 depuis le mois de juin 1978 une instruction technique pour attirer 1' attention du pilote sur ce ph&nom&ne.Ces instructions techniques Bmises par la 255, qui, en matiere de securite de vol, priment sur tout autre document (sans Otre en contradiction avec celui-ci naturellement), traitent des problemes specifiques rencontres sur nos helicopteres. L' IT 11 B4 no 3 en question parlait de la temperature ideale de demarrage, come defini ci-dessus et indiquait les causes possibles de son d6passement.En ce qui concerne les temperatures maximales autorides au demarrage, elle referait simplement au manuel de vol.Cette IT devait Stre portee B la connaissance des pilotes regulierement et pour marquer son importance, des briefings speciaux etaient donnes systematiquement. Les trois incidents de debut 1985 etaient consideres comme une suite plus ou moins aleatoire dont la cause presumee Btait le givrage de 1' eau de condensation dans le robinet 4 v0ies.C' est pourquoi que le commandement se limitait B envoyer quelques notes de rappel de 1' IT traitant du vol en atmosphere neigeuse et givrante.0n imposait egalement la mise en place des housses de protection des 1' arrSt de la machine.Les mesures prises paraissaient adequates car au cours de 1' annee, il ne se produisait plus de surchauffe. c.La periode d' hiver 1985-1986. Pendant la periode d' hiver 1985-1986 quatre surchauffes au demarrage se produisaient dans un delai de trois semaines (annexe C).C' etait un desastre, deux moteurs devaient subir une RG, les deux autres devaient subir une VN1P.Les frais de reparation s' Clevaient B environ 7,5 millions de FB.Les causes des incidents Btaient B chaque fois soit un robinet 4 voies gel&, soit une batterie trop faible, soit une combinaison des deux facteurs.Le resultat etait catastrophique car il ne restait plus aucun moteur de reserve pour le depannage des machines immobilisees. d.Analyse des problemes. Au debut de 1' hiver des briefings speciaux avaient encore ete donnes sur le probleme de surchauffe au demarrage; la reaction des autorites ne se faisait pas attendre.Des sanctions pecuniaires pour le responsable de 1' incident fUrent proppses et la 255 fat incitee & sortir des nouvelles IT; mais le premier pas envers la Solution du arohleme~ _.__. etait~~~~. la mise~ -~~ sur~~. *~~~Died d'~~ une ~~ ~ commission ~~~ ~~~~ d' enouete Dour traiter le Drobleme dans sa globalite.Debut mai 1986, la commission d' enquSte'sortait~ ses risu1tats.- Come premiere cause principale des problemes, la commission indiquait la faible charge de la batterie, elle proposait alors de faire particulierement attention au voltage et aux tests de batterie pendant les periodes d' hiver.Elle proposait egalement de faire remplacer la gendratrice DYNASTAR par une generatrice B courant de chargement superieur et de faire construire, en collaboration avec le constructeur, un circuit de protection pour couper automatiquement la sequence de demarrage si la tension batterie descendait en dessous du Seuil dangereux de 14 Volt. Comme deuxieme cause principale, le bon fonctionnement du robinet 4 voies etait mis en question.La commission proposait de chercher une solution en collaboration avec le constructeur ou bien en deplacant le robinet 4 voies ou bien en plaCant un systeme de prechauffage du robinet 4 voies.Elle proposait egalement de faire la mesure du dew6 hygrometrique du carburant en debut d' hiver. 3-4

En ce qui concerne le probleme de la batterie, il avait deja et6 remarque au debut 1985, suite 21 une consommation anormale d' elements de batterie que celles-ci Ctaient souvent insuffisamment chargees et que la methode de charge et de decharge par impulsions Pouvait Ctre amelioree.Ce probleme s' ag9raval.t en hiver car la batterie perdait de sa capacit&.Sur base de ces arguments, la d4cisj.on etait prise de former les responsables de la charge des batteries directement en usi:ne chez la firme VARTA et d' Ctre beaucoup plus severe lor6 du contr8le des 616ments.Les frais de remplacement des elements batterie pour la periode 1985/87 6' elevait 5,5 millions de FB.En outre, de nouveaux chargeurs de batterie etaient acquis pour une valeur totale de 600.000 FB.La proposition de placer un circuit de coupure du demarrage couple B la tension batterie n' etait pas etudie. Les defectuosites constatees au niveau du robinet 4 voies etaient doubles.Le mauvais fonctionnement provenait d' un encrassage de la bille suite aux differents demarrages, provoquant une augmentation Cvolutive des 1'4.Le deuxihe phenomene se faisait sentir en hiver, par temps froid et degre hygrometrique eleve, par le blocage du robinet suite au gel des vapeurs de condensation. Pour remedier 2t ces problemes, une nouvelle version d' une IT 11 B1 no 5A Sortait au mois de novembre 1985, apres contact telephonique av€!c la firme TURB0MECA.L' IT rendait obligatoire la mise en place des coiffes de protection sur un sol couvert de neige et lors d' une temperature de moins de 5 OC.Elle imposait Cgalement au pilote de refaire une mise en marche du moteur apres son arret jusqu, B une legere augmentation de T4, car cette operation permettait de remplir le robinet 4 voies de kCrosene et d' Cviter le gel de celui-ci.Elle prevoyait egalement le nettoyage et le sechage du filtre et du tuyau P2 ainsi que la bille et son siege du robiriet 4 voies B chaque visite multiple de 25 heures . Les trois inc.i.dents du debut de 1986 prouvaient que les actions entreprises ne suffisaient pas e0core.c' est pourquoi en mars et en novembre 1986 paraissaient des nouvelles versions des IT I1 B4 no 3 contenant des prescriptions plus claires et simplifiees quant aux temperatures de decisions ainsi que :Les actions a prendre par le pilote et les mecaniciens en cas de temperature ou tension anormales. La modification du robinet 4 voies Btait de la competence de la firme TURBOMECA.Les propositions de la 255 Otaient un rechauffement du robinet 4 voies par vole blectrique, une solution toute simple et efficace, et eventuellement une modification du systCme lui-mCme.La modification enfin proposee par la firme etait le deplacement du robinet 4 voies (plus pres du moteur, donc avec moins de risques de gel) et une Protection calorifique des tuyauteries du robinet 4 voies.La modifica1:ion coatait environ 7000 FF par turbine avec un delai de livraison de 8 mois.Quelques moteurs ont subi cette modification. e.La periode d' hiver 1986-1987.

Suite a 1' etude fait en 1985 et 1986, on pouvait esperer une nette amelioration des problemes pour 1' hiver suivant.En effet, deux surchauffes au demarrage Btaient remarquees mais les dCgats causes B la turbine Ctaient negligeables grace B de bonnes reactions des pilotes.Le fait que le probleme etait bien connu et que des mesures suffisantes etaient prises, avait porte ses fruits.

5. CONCLUSIONS GENERALES. Le probleme de surchauffe au demarrage Btait dti a deux phenomenes tout i fait ind6pendants.D' abord il y avait des causes purement techniques come la defaillance de la batterie et du robinet 4 voies, qui ont et6 resolues d.' une maniere satisfaisante mais pas tout a fait comPl&te.En effet, les solutions definitives, B savoir un circuit electrique de coupure du demarrage en fonction de la tension batterie ainsi qua un systeme de prechauffage du robinet 4 voies n' ont pas et6 :cetenues.En outre, le risque d' erreur de manipulation (facteur humain) a et6 r6du.it a un minimum par une sensibilisation et une instruction adequate. Enfin il ne faut pas non plus exagerer le phenomhne car depuis 1987 il n' y a plus jamais eu de problemes de surchauffe au demarrage dus au temps froid, d' un c8t6 parce que des conditions extremes de temperature sont devenues asfiez rares dans nos regions , peut-Ctre, comme pretendent les "verts" B cause d' un rechauffement de la terre par effet de serre mais surtout parce que les pilotes sont bien conscients du phenomene et en cas de situation anormale , ils ont B leur disposition des instructions simples et claires.

FZFERENCES

1.Manuel d' Entretien et Manuel de Vo1 S.A. 3180 ALOUETTE ASTAZOU,SUD AVIATION,Soci6t6 Nationale de constructions A6ronautiques,PARIS,FRANCE. 2.Documents classifies de la 255 Compagnie Maintenance et Dep6t d' Aviation LOgBre. 3-5

- CARWRAMfUEL) -. AIR

Rnnexe A:CIRCUIT DE CARBURANT DE DEHARRAGE

Nombre d' helicopt8res: 70 Nombre total d' incidents: 23 Nombre total de surchauffes: 5 Date Cause prdsum6e Prix Reparation (FB) 15 Fev 82 Non application des procedures 827.600 (RG) 17 NOV 82 Non application des procedures 1.254.800 (RG) 14 Jan 85 Bille robinet 4 voies gelee 691.637 04 Jan 85 Bille robinet 4 voies gelee 1.015.500 (VNIP) 14 Jan 85 Bille robinet 4 voies gelee 1.019.250

Annexe B:ACCIDENTS/INCIDENTS DE LA PERIODE 1982-1985

S.lLz&ilUf€enO.l Date: 05 Fev 86 ADDarei1:A 81 GTM no-ioo &-1322:30 hr Cellule: 3539:20 hr 1ncident:l surchauffe de 750 oc d' environ 2 sec Donnees supplementaires: Lieu:Vogelsang L' helicopthre etait stationne sans ses housses de protection sur un emplacement enneig6.Lors du passage d' une autre machine, cette neige a pu s' &lever du sol, entrer le moteur et s' accumuler dans le circuit d' alimentation jusqu' au robinet 4 voies.En plus, la procedure de remise en marche aprhs arr6t n' a pas 6th executee. Prix de reparation:219.784 FF.

Gi'M no-678 813:40 hr Cellule: 4612:35 hr 1ncident:l surchauffe de 700 *C, aprhs coupure demarrage 590 OC Donndes supplementaires: Lieu:Merzbruck (AIX) Demarrage apr8s une ventilation trop longue d' environ 15 sec. ce qui est nocif pour le demarreur,le relais ddmarreur et la batterie. cause pr6sumee:Gel du robinet 4 voies bien que la machine ne Stationna que 5 minutes dans une temperature ldgerement negative. Prix de r¶tion:150.656 FF, soit le prix de 1' expertise car la turbine n' a pas et6 endommagee. 3-6 alLchaLffstn5L3

Date! 25-. .FeV- . 86.. Apparei1:A 78 GTM no 551 a 1361:55 hr Cellule: 3570:40 hr Incident: Deux demarrages successifs anormaux de 630 OC, suivi d' une surchauffe de 750 OC! pendant 1 sec.Le 19 Fev 86 une premiere surchauffe de 730 pendant 1 sec avait deja eu lieu. Donnees supplementaires: La temperature exterieure etait de -5 V.Un contr6le le lendemain indiquait que le robinet 4 voies etait bloqub par le gel. Cause pr6sumee:Gel du robinet 4 voies et suite aux tentatives successives de demarrage, un affaiblissement de la batterie. Prix de reparation:171.549 FF plus remplacement de la DYNASTAR et le relais demarreur suite au refroidissement insuffisant entre les demarrages. surchauffe no 4 Date: 25 Fev 86 Apparei1:A 93 GTM no 376 b 1373:35 hr Cellule: 4505:40 hr 1ncident:l surchauffe de 800 OC pendant 1 sec. Cause pr6sI"e:R6action retardive de la part du pilote pour arreter la phase de demarrage, la batterie ne possedant pas le voltage minimum necessaire pour effectuer un demarrage correct. Lieu:Ecole d' Aviation Lbgere b BRASSCHAAT. Prix de r6paration:229.178 FF

Annexe C:ACCIDENTS/INCIDENTS DE LA PERIODE D' IXIVER 1985-1986

3-n 3-n 01 Date: 13 Jan 87 Apparei1:A 95 GTH no 718 Incident: Lors d' une premiere mise en route, la temperature de 700 OC a et6 atteinte, lors du deuxieme demarrage, elle Btait de 600 "C.Le robinet 4 voies a et6 prechauffe par un "crimp gaine heater" emprunte aux paracommandos.Apr6s cette operation, un dbmarrage normal etait possible. Donnees supplementaires: Lieu:schaffen Temperature extbrieure:-17 OC Grace a une intervention correcte du pilote, la turbine n' a subi aucun degit.

-0-2

Date: 15 Jan 87 Apparei1:A 79 Incident: A Cologne,un demarrage au moyen d' un groupe a et6 realis6 awes prechauffage du robinet 4 voies a 1' air chaud.Apr&s un vol d' environ 3 heures, la turbine a BtC coupbe et remise en route immediate,jusqu' b une augmentation de T4 de 300 OC.Les housses de protection ont et6 mises en place aussit6t.Apr&s 25 minutes, un nouveau demarrage a 6th tent6 selon la procbdure normale durant lequel une montee anormale de T4 jusque 600 "C a ete remarque.Apres prechauffage du robinet 4 voies,un demarrage normal a pu avoir lieu. Donnees supplementaires: Lieu:Soest Temp&rature:-lZ OC et vent d' Est assez fort et froid.

Annexe D:ACCIDENTS/INCIDENTS DE LA PERIODE D' HIVER 1986-1987 AVIONS D'AFFAIRES MYSTERE-FALCON EXPERIENCE OPERATIONNELLE PAR TEMPS FROID

C. DOMENC Kespons;ible Propulsion D,\SSAULI'- AVIATION Boite Postale no 24 33701 MERIGNAC CEDEX - FRANCE

FAMILLE DES AVIONS MYSTERE-FALCON

Brbe pr&sentationdes differents mod&les,de leur doniaine de vol, des caracttristiques de leurs moteurs, des systhes de dtmarrage.

EXPERIENCE DU DEMARRAGE PAR TEMPS FROID

Campagne d'essais jusqu'8 - 40°C dans le cadre de la certification, comportement des systhes de demarrage tlectrique i basse tempirature, procedures opkrationnelles correspondantes. Domaine de dkmarrage en vol.

EXPERIENCE EN CONDITIONS GIVRANTES

Br&vedescription des syst6mes de protection contre le givre, prklkvements d'air moteur, mtthode de certification, experience en service. 4-2

1. - INTRODUCTION

Dcpuis la certification en 1965 du MYSTERE-FALCON 20, la Societk DASSAULT a produit plus de 1000 avions d'affaires i riaction qui ont accumule 5 millions d'heures de vol en service. Cettc famille d'avions comprcnd les bi-riactcurs de type FALCON 20 et FALCON 10, ainsi que Its tri-reacleurs de type FALCON 50 et FALCON 900. Actuellement seuls les tri-riacteurs sont en production, mais un nouveau bi-rkacteur, le FALCON 2000, est en cours de diveloppement pour completer la gamme.

2. - FAMILLES DES AVIONS MYSTERE-Fm

A. - Caractiristiaues mincinales

TYPC Nbre de pasagen Motcur Rayon d'action Mach Maximum Altitude (avm rCselves NBM-IFR) Maximum .

FALCON 20 8/10 2 GENERAL ELECI'RIC 1400 Nm 0.85/0.88 42000 it CF 700 >

DERIVES DU FALCON 20 : noc/noo 8/10 2 GARRETT AT13 22W Nm 0.R6S 42000 ft FZO-5. 8/10 2 GARREIT 2200 Nm 0.85/0.88 42000 ft 'WE 731-SAR

FALCON 10 417 2 GARm 1600 Nm 0.87 45200 ft 1-FE 731-2

FALCON 50 8/14 3 GARREIT 3200 Nm 0.86 49000 ft lTE 731-3

FALCON 900 8/19 3 GARREIT 3900 Nm 0.87 s1000 it TFE 731-SAR

* Re-motorisation des EOen Service

Tous ces avions itaient de technologie tr&smoderne au moment de leur certification tant en ce qui concerne les systbmes, dont les moteurs, que I'airodynamique et la structure.

En raison de leurs performances ils sont tous entibrement iquipis de commandes de vol servo-commandkes. C'etait une premibre pour un avion civil 101s de la certification du FALCON 20, et ccla a contribuk B I'agriment particulier de pilotage de ces avions unanimement apprecii.

L'optimisation des voilures par le calcul a commence avec le FALCON 10 et a &ticomplitement rialisie pour le FALCON 50 et le FALCON 900 dont la voilure est optimisie en rigime supercritique. Tous ces avions ant des dispositifs hypersustentateurs dc bord d'attaque et de bord de fuite des voilures pour permettre I'emploi de pistes courtes utilisies par I'aviation d'affaires. 4-3

Les formes arrieres de fuselage ont et6 particulierement etudiees sur les tri-rkacteurs FALCON 50 et 900 pour reduire la trainee en croisikre.

Les voilure? sont a structure integrale sur tous les modkles. Un FALCON 10 est en service depuis quelques annees avec une voilure B caisson en composite carbone. Les matkriaux composites sont utilises sur le FALCON SO et surtout sur le FALCON 900.

B. ~ Domaine de vol twiaue

A titre d'exemple on trouvera ci-aprb le domaine d'emploi du FALCON 900 en temptraturejaltitude, avec la zone de decollage effectivement demontree jusqu'i 14000 ft (LA PAZ) et le domaine de rallumage en vol garanti jusqu'i 30 000 ft c. - &@g

Moteur Dicollage Croisiere Observations 4wW ft M = 0.8 PO"SS& Taux de Taux de Cs (Ib/lb/hr) Ib compression dilution (no" install&)

CF7W 4Mo/45W 6.5 2 0.967 Fan arritc ATF3 5200 21 2.9 0.80 Triple corps TFE 731-2 3230 13 2.7 0.81 Fan avant axe ridueteur TFE 731.3 3700 14.7 2.8 0.814 Fan avant avec reductcur TF'E 731-5AR 4500 14.7 3.6 0.759 Fan avant avee reducteur et tuyere a milangeur

- Remarques generales sur ces moteurs :

w le CFIOO Btait le premier petit rtacteur i double flux. Une roue turbinejfan ktait rajoutee derrii-re un generateur type (3610.Prefiguration des UDF actuels de GENERAL ELECTRIC.

I'ATF3 est le seul petit rkacteur double flux, triple corps, avec le corps haute pression juxtapost i I'arrikre d'un double corps fanjbasse pression.

W les TFE 731 ont it6 les premiers reacteurs utilisaut une rBgulation 6lectronique.

les taux de dilution mod&% permettent de limiter le diami-tre du moteur B des valeurs compatibles avec un montage sur le fuselage arrikre.

- Caracteristiques de ces moteurs lieees au demarrage, en particulier B basse temperature : 4-4

Pointe du couple rbistant, A - 40°F (- 40°C).

CF 700 731-5AR

Vitesse prise d6marreur (tours/minutes) 850

Couple rbistant (LB.FT) 65

Puissance m6canique correspondante (Watts) 7800 6777 6710

Ces puissances son1 compatibles avec un systkme de dkmarrage Clectrique. Le CF700, malgrt son faible tau de compression, est plus exigeant car le compresseur dans son entier est entrain6 par le dkmarreur, alors que seul le compresseur HP (centrifuge) de faible inertie et de taw de compression mod6r6 (2 i 2.5) est entrain6 sur I'ATF3 (N3) et sur le 731 (N2).

On peut noter que sur le TFE 731 le fan est coupl6 au compresseur BP par un r6ducteur qui nkcessite une quantiti d'huile augmentke, d'oD un impact sur le couple rtsistant, mais moins fort que si le dharreur devait entrainer I'ensemble.

Les dispositifs de gtom6trie variable d'entrte d'air (CF 700, ATF3) et les vannes de d6charge aident i diminuer le couple rCsistant au dtmarrage.

La rbgulation Electronique des ATF3 et TFE 731 dose le d6bit carburant au dkmarrage en fonctiou de la tempkrature ambiante. Un enrichissement antomatique est pr6w au-dessous d'une tempbature interturbine de 400°F.Une commande manuelle permettant de maintenir I'enrichissement au-deli de 400°F pour les dkmarrages i basse tempbrature est montEe sur les FALCON 10 et 50.

La charge des accessoires avion est faible. II n'a pas U6 nkcessaire dutiliser un by-pass des pompes hydrauliques. La g6nbatrice Blectriqne sert de d6marreur et ne retrouve sa fonction g6ntration qn'aprks la fin du dbmarrage. Selon Ies cas, il peut &re ntcessaire de temporiser la mise en ligne de la g6nQatrice pour laisser le moteur accElErer entre la fin de la stquence de d6marrage et le ralenti.

D. - Svsthne de d6marras

- Sur ces avions de bilan Llectrique modEr6, I'analyse i montr6 que le meilleur compromis en poids et en prix 6tait d'utiliser une g6n6ration 6lectrique en courant continu 28 volts, la g6n6ratrice servant aussi de d6marreur. Pour permettre le d6marrage autonome des moteurs, deux batteries (Nickel-Cadmium) sont n6cessaire. Les g6nQatrices (une par moteur) utilis6es sur les FALCON on1 une puissance de 9 Kw (sauf FALCON 200 : 10.5 Kw), les batteries sont de 36 Ah sur les FALCON 20 et 200, de 23 Ah sur les autres modkles. -Le couplage des batteries en skrie ou en paralkle au cours du dkmarrage est choisi selon leur rksistance interne de fason a maintenir une tension satisfaisante aux bornes du dtmarreur jusqu'B une temp6rature ambiante assez basse : . SCrie uniquement sur le FALCON 20 CF700. . Parallkle ou s&ie (a basse temphature, sur sblection du pilote) sur les FALCON IO, 50 et 200. . ParallBle uniquement sur le FALCON 900. . Une assistance par la gtnCratrice de I'APU, ou par celle d'un moteur dCjja dCmarr6 est possible sur les FALCON.

- L'ktat de charge des batteries et leur temphature sont des paramstres essentiels du dCmarrage a basse tempkrature.

3. - EXPERIENCE DU DEMARRAGE PAR TEMPS FROID

3.1. - FALCON 20 - CF 700

- Une campagne a eu lieu en FINLANDE et NORVEGE en 1966 jusqu'8 moins 28°C.

- Aprh exposition d'une nuit sans protection particulihe, les demarrages se font correctement avec les batteries de 36 Ah en skrie. Le sen1 ennui rencontre est I'obturation d'une tuyauterie de rkgulation, modifike ulthieurement par GENERAL ELECTRIC.

3.2. - FALCON 10 TFE 731-2

- Deux campagnes d'essais ont eu lieu la premikre en 1974 en ISLANDE et au CANADA (FROBISHER BAY) jusqu'h - 30T, la deuxi2me en 1975 au CANADA (FROBISHER BAY et YELLOW KNIFE) jusqu'a - 45°C.

- La prockdure de dkmarrage au-dessous de 5°C consiste coupler les batteries en strie pour dtmarrer le premier moteur et en paralkle pour demarrer le second.

DCmarrage sans protection particulihe des batteries ou des moteurs apris un skjour prolong6 au froid. .. Avec des batteries de 23 Ah le premier moteur dCmarre jusqu'i - 10"C/-15°Cc.Avec des batteries de 36 Ah cette limite passe 2 - 25"C/- 30°C : uu dkmarrage marginal 3 -30°C a durk 65 secondes avec allumage an bout de 45 secondes et ITT maximum 820°C pour une limite de 860°C.

..Le second moteur d6marre dans tous les cas, aprks un sijour prolong6 entre - 30°C et - 40°C. .. Le premier moteur a pu &tredkmarrt, avec les batteries de 23 Ah, aprks uu skjour de 4 heures i 30T, aprb deux teutatives infructueuses qui avaient rkchauffb les batteries de - 10°C B 0°C et diminui la poiute initiale de courant de 1000 A/12.5V B 825A/11.5V. 4-6

D6marragc du premier moteur avec systimes de rtchauffage branch& sur des pauneaux &alimentation exttrieure. .. Rechauffage des batteries par couverture tlectrique chauffante CSA lU)V/XOW. Aprb une nuit et une matinte passtes a - 4O"C/- 43T, la temperature des batteries se maintient 2 - 5'C et le moteur dtmarre. La temperature de I'huile etait de - 43°C. Le dtmarrage est difficile le fan tourne lentement, on considkre qu'au-del& de - 35°C il vaut mieux rtchauffer le moteur. .. Rechauffage des moteurs par ventilateur CTC 12OV/85OW. L'installation de ce ventilateur dans I'entrte d'air avec les caches entrte d'air et tuyauterie normaux ou avec une couverture molletonnke B trois tpaisseurs isolantes enveloppant toute la nacelle a permis de ramener la tempkrature d'huile respcctivement de - 43°C B - 27°C et B - 18°C. La couverture permet de gagner 10 B 20°C sans vent, U)B 30°C avec un vent de 10 Kts. Le dtmarrage a 6tt facile.

- Durant ces essais, des anomalies de fonctionnement des calculateurs Clectroniques du moteur 101s de la chute de tension en d6but de dtmarrage ont ament le motoriste B modifier cette version initiale des calculateurs. Avec des batteries de 23 Ah, la chute de tension initiale batteries en strie, si les batteries ne sont pas rtchaufftes, peut amener une insuffisance momentante de l'allumage et un passage en mode manuel du calculateur moteur. Tout rentre dans l'ordre quand le dtmarreur prend de la vitesse ce qui s'accompagne d'une augmentation de sa force contre-tlectromotrice, d'oii une remontte de la tension et une baisse du courant dtbit6 par des batteries.

- La pression d'huile, moteur non rtchaufft, peut atteindre 80 B 100 psi et demande 4 B 5 minutes avant de revenir dans la plage habituelle inf6rieure B 55 psi.

- un indicateur N2 qui se bloquait au froid a ttt modifit.

3.3. - FALCON 50 TFE 731-3

- Des essais ont eu lieu en FINLANDE, puis au CANADA (FROBISHER BAY - FORT CHURCHILL - CAMBRIDGE BAY) en 1980 jusqu'B - 39°C.

- Les moteurs et les batteries sont tris proches de ceux du FALCON 10, la g6nCratriceldCmarreur est d'un modkle difftrent, I'APU qui n'existe pas sur FALCON 10 est optionnel sur FALCON 50.

- La premiBre s6rie d'essais en FINLANDE a montrt la n6cessitt de rtchauffer les batteries au-dessous de - 10T/ - 15°C ce qui est homogine au FALCON 10.

Les couvertures chauffantes de SOW du FALCON 10 se sont av6rtes insuffisantes, des couvertures mieux adapttes de 120W et 150W ont &tiessaytes au CANADA. Les 150 W, lrop puissantes, donnaient des tltvations de temp6rature des batteries excessives (65 B 70°C). Les couvertures de 120W ont ttt retenues (tchauffement de 55°C) avec, dans le kit grand froid deduit des essais, line coupure de rtchauffage B une temperature batterie de 25°C. 4-7

- L'APU et les moteurs doiveut &trerbhauffks au-dessous de -35°C. Le fan ne peut &tretourn6 B la main sans rkhauffage B - 35°C. L'APU est rkchauffk par un ventilateur de 850 W plad dans I'entrCe d'air, obturateurs d'entrke d'air et tuyhe en place. Les moteurs ont kt6 rechauffks soit par deux ventilateurs (entrCe d'air et tuykre) de 850W, soit par un pr6lkvement d'air branch6 sur I'avion qui suppose I'APU ou un moteur en fonctionuement. Le rechauffage du moteur permet la rotation du fan, mais I'huile contenue dans le relais d'accessoires reste B la temperature ambiante.

- Un dharrage aprks 3H30 B - 27°C sans aucun rechauffage a Btk essay&.Les batteries etaient B -15"C, I'huile moteur B - 17'CC/ - 23T, le caisson APU B - 14°C. Aprh un dCmarrage correct de I'APU sa generatrice est coupke pour simuler I'avion sans APU. Deux essais de demarrage d'un moteur batteries en sCrie sont interrompus par le pilote B cause de la perte d'indicatiou N2 (tension trap faible). La troisihe tentative, batteries en parallkle est normale, les batteries Ctant maintenant B t 15°C.

- Avec les batteries rbchauffkes et une ambiance de - 31.5"C (tempBrature d'huile - 31°C) le demarrage est possible batteries en parallkle ou en sQie, mais dure 40 sec avec 1TT = 660°C en paralkle, pour 26.5 sec et In= 475°C en s6ie. Si I'assistance d'une generatrice moteur ou APU est utilide celle-ci fournit environ 30% du courant soit 460 Amp.

- Avec les batteries et les moteurs rechauffes (tempkrature infbrieure 8 - 35°C) le premier moteur est dBmarrB batteries en skrie, les autres batteries en paralkle avec assistance 6ventuelle des g6nCratrices A - 39"C, batteries B + %"C, huile B - 35°C (relais d'accessoires) le moteur dharre en 21 sec, ITT = 550°C.

- Des essais conparatifs d'huile MOBIL JET I1 et EXXON 2380 u'out moutrt aucune diffkrence perceptible des caractCristiques de demarrage malgrk leurs viscositbs diffkrentes. Un contacteur klectrique du circuit de dkmarrage a dfi itre rCchauff6.

3.4. - FALCON 9CKl- TFE 731-5AR

- Des essais ont eu lieu en 1986 et 1987 B FROBISHER BAY jusqu'8 - 39°C.

- Cet avion est proche du FALCON 50, les moteurs ont un peu plus d'huile, I'APU &ant de base sur cet avion le couplage des batteries est uniquement parall&le et I'assistance au dimarrage par le gCu6ratrice de I'APU est systhatique.

- Un essai a montr6 qu'il ktait possible de demarrer sans assistance et sans rkchauffage aprb 12H pasdes entre -15°C et -19°C.

- Le rkchauffage des batteries par les m&mescouvertures de 12OW que sur le FALCON 50 B des temperatures inferieures a - 15°C a 6tC utilis6 (hors I'essai ci-dessus) Le rechauffage des moteurs au-dessous de - 35T, au lieu d'utiliser les ventilateurs de 850W a Ctb fait 8 partir de groupes de piste permettaut de souffler de I'air chaud sur les zones B rkchauffer, en particulier sur les charniBres des capots avant d'ouvrir ceux-ci pour rtchauffer le reservoir d'huile et le relais d'accessoires ce qui u'ktait pas fait sur le FALCON 50. Un brassage du fan B la main d'uue quinzaine de tours assure son deblocage 4-8

- Le rtchauffage de I'APU, quand il est IubritX par de I'huile type I (ESSO 2381), n'est en principe pas necessaire. Mais le temps de dtmarrage observe conduit B recommander le richauffage a temperature infkrieure B - 35°C.

- Pour couvrir le cas de panne de I'APU, un moteur a et6 demarre sans assistance - 3YC, batteries B + 3O"C, aprPs un rkchauffage de 15 minutes (T huile = 35°C). Le dkmarrage est long (1 minute 48 secondes) sans surchauffe (ITT = 548'C), c'est un cas marginal. Avec assistance, le temps de dtmarrage est d'une trentaine de secondes.

3.5. - FALCON 200 - ATFJ

Une campagne d'essais a eu lieu en 1982 i YELLOW KNIFE jusqu'B - 48°C.

- Par rapport aux TFE 731 tquipant les FALCON 10,50 et 900, I'ATF3 n'a pas de rtducteur de fan ce qui est un peu favorable. L'avion est equip6 de batteries de 36 Ah (37 Kg contre 25 Kg pour la batterie de 23 Ah). Leur resistance interne plus faible donne moins de chute de tension initiale que les batleries de 23 Ah.

Sans protection particdiere il a et6 possible de dharrer jusqu'i - 25°C.

- Avec rkchauffage des batteries (temperature infkrieure B - 30°C) par les couvertures de 12OW, il a Btk possible de demarrer i des temperatures de I'ordre de - 45°C sans rtchauffer le moteur, aprh avoir tourn6 le fan B la main. Quelques stagnations de regime entre la tin de dtmarrage et le ralenti, rattrapables en avanGant la commande de gaz ont et6 observees et corrigees par le motoriste dans une version ulttrieure du calculateur.

3.6. - RESUME DES PROCEDURES DE DEMARRAGE PAR TEMPS FRm

- Pour les FALCON kquipts de moteurs GARRETT TFE 731 ou ATFJ il est recommand6 de tourner le fan B la main avant un dharrage a temperature negative.

- Jusqu'B - lO"C/- 15°C (batteries de 23 Ah) ou - 25°C (batteries de 36 Ah) aucun rtchauffage n'est ntcessaire, le dtmarrage autonome est possible en stlectionnant le dharrage LOW TEMP, batteries en shie, pour le premier moteur (sauf FALCON 900 ----> assistance APU). Pour une exposition prolongee au-dessous de ces temptratures, il faudra privoir un rtchauffage des batteries.

- Au-dessous de - 35°C il faudra tgalement rechauffer le rtacteur (modde TFE731) et I'APU de preference avec un groupe de piste (air chaud).

- Le deuxikme reacteur demarre toujours batteries en parallkle, le troisihe tgalement,

- Une pression d'huile forte peut &re atteinte, en particulier s'il n'y a pas eu de rkhauffage du relais d'accessoires.

- Les carburants ntilists au CANADA ttaient du JETAl ou du JET B,

. Le dispositif d'enrichissement manuel, pen utilis6 sur les F10 et F50 n'a pas tt6 reconduit sur le F900 4-9

3.7. - DOMAINE DE DEMARRAGE EN VOL

- Le domaine d'altitude dkmontrk sur les FALCON 20, 10,50 et 900 est 0 - 30000 ft et 0-35000 ft sur le FALCON 200. En cas de panne du calculateur klectronique l'altitude est IimitCe B 20 000 ft pour &iter les surchauffes (dkbit carburant plus fort).

- Le domaine vitesselrkgime de dkmarrage en moulinet est on non dkfini. C'est l'intkrtt du dkmarrage klectrique que d'ttre disponible B tout instant et pratiquement indkpendamment des conditions d'altitudejvitesse, une consigne, du genre N2 < 15% : utiliser le dkmarreur, peut suffire.

4. - EXPERIENCE EN CONDITIONS GIVRANTES

4.1. - DESCRIPTION SOMMAIRE DES SYSTEMES DANTIGIVRAGE

- Tous les FALCON sont kquipks de circuits d'anti-givrage par air chaud prklevk sur les rkacteurs. 11s protkgent I'intkgalitk des bords d'attaque des voilures, aiusi que les entrkes d'air des rkacteurs.

- La zone rkchauffke de I'entrbe d'air varie selon I'application :

. FALCON 20 CF 700. Entrke d'air double conduit (un pour le compresseur etun pour le fan arrikre). Rkchauffage des I&res entrke commune et eutrke compresseur, de mits structuraux du conduit de fan. du fond de ce conduit.

FALCON 10, TFE 731. Rkchauffage des lkwes et d'une portion aval du canal d'entrke d'air

. FALCON 200 (ATB), 50,900,20-5 (TFE 731). Rkchauffage des lkvres uniquemeut.

. Moteur central FALCON 50 et 900 : en plus des lkwes, rkchauffage des BOo supkrieurs du conduit en S.

La techuologie des kchangeurs de chaleur est la suivante :

. Pour le bord d'attaque (becs mobiles) des voilures F20/FlO/FS0-900, Yair chaud distribuk par un tube perfork drcule dans des canaux usinks dans la paroi externe et refermks par une paroi interne.

, Pour les Ikvres d'entrke d'air, il y a en kvolution : double peau (comme pour les voilures) sur la nacelle CF 700, simple caisson de bord d'attaqne avec distribution d'air par tube "piccolo"sur les autres nacelles, 6vacuatiou par une double pean rkchanffant un morceau du canal d'entrke d'air sur le FALCON 10, evacuation directe sur les autres avions.

, Pour les conduits en S des moteurs centraux FSO et F900, canaux en OMEGA rapportbs.

- Les prklkvements d'air (avec moteurs GARRETT) consistent en un mklange BP + HP pour les voilures et conduits en S, en HP pur pour les Ikvres d'entrkes d'air. Des vannes rkgulatrices de pression sont gknkralement utiliskes pour moduler le dCbit prklevk en fonction du rkgime rbacteur.

Par exemple pour un FALCON 900, en montke, on prklkve environ 1%de HP pour les lkvres d'entrkes d'air et 7% de mklange HP + BP pour les voilures et le conduit en S. 4-10

- Les moteurs eux-mimes ont, ou non, un systeme d'antigivrage. Le CF700 et les TFE731-2 et -3 initiaux (no - F10 - F50) ktaient kquipts d'unc vanne de prd6vement et de rtchauffage de l'enlrke compresseur sur CF 7M) (IGV) et du "spinner" elliptique d'origine sur le TFE.

Puis GARRETT a certifit des "spinners" coniques et supprim6 tout rtchauffage sur tous ses moteurs

4.2. - METHODFS DE CERTIFICATION

- Voilnres : calcul des zones de captation, essai tventuel d'un panneau en soufflerie de givrage, mesure des tempkratures de peau en vol et extrapolation aux limites, vol en givrage naturel. Determination par calcul des rbgimes rkacteur minimum nkcessaires et vkrification en vol.

- Nacelles : dossiers de calcul kventuels en conditions givrantes (GRUMMAN : TFE 731-2/F10 - ROHR : ATFJ/F;?oO), essais tventuels de I'installation motrice en soufflerie de givrage (CEPr SACLAY : nacelles CF7M)/F20,731-2/F10, manche a air centrale 731-3/F50), mesure des rempkratures de peau en air sec en vol, extrapolation aux limites du domaine, vol en conditions givrantes naturelles.

- Des essais de mise en service retardke de l'anti-givrage peuvent itre faits la demande des autoritts de certification soit en soufflerie de givrage pour l'installation motrice, soit en vol pour la voilure.

4.3. - EXPERIENCE EN SERVICE

- Hors panne de vanne de prkltvcmcnt, aucun cas de protection insuffisante n'a ktk rapportt.

- Des prkcautions d'emploi en jour chaud et en condition statique sont nkcessaires, en particulier pour kiter la surchauffe des bords d'attaque de voilure. 4-11

MYSTERE-FALCON 900

DOMAINE DE RALLUMAGE EN VOL

Altitude 1x1 OOOttl

Vlt.sI. lndlquk lkll

DOMAINE TEMPERATURE f ALTITUDE

rampiraturs ItatiqYe 1-1 FALCON 10 FALCON 10

INTEN51 TE5 INTENSITES OEMARRAOE RElCTEUR GAUCHE PPRES REPCTEUR DROIT 1Al 181 BATTERIES 23Wh EN SERIE A DBCharge B ext :-3O'c UPRES UNE NUlT II -4O'c I GENE Or B Batt !*Io'c B Huile :-36'c 1000 500 I

50C

500.

50( I000

TENSIONS U GENE G 1001

20 Charge t

TENSIONS (Vi TEIIPSli 0 0 U GENE C 5 IO 15 20 25

50 - 20

25 - IO

TEHPS (: TEMPS(: L 001 , * 0 5 10 15 20 5 IO 15 20 25 5-1

R. Meijn Manager Environmental Control and Ice Protection Systems

FoltwR Aircraft B.V. P.O. Box 7m 1117 w SChiFimldoat T%e Netherlawla

0. Bhatraft

Several temporary flame-out incidents were expsrienced in descent through light icing conditions and precipitation during regular scheduled flights. Previously, both engine and aircraft had been qualified against FAR 33/JAR E and FAR/JAR 25 including tunnel-testing and natural icing trials. Test-flights in natural icing conditions with video observation of the interior of the air intake revealed ice accretion at the interface between engine-casing and intake-duct in mixed conditions of cloud, snow, hail and rain while the aircraft exterior remained free of ice. Extensive ground testing of the engine indicated less tolerance to ice ingestion than bad been demonstrated in engine certification tests. Powerplant ice protection was enhanced by additional anti-icing of the engine flexible seal Figure la : Cross-sectional view of duct (split-up) by bleed air. and bifurcation (detail). This paper discusses factors influencing unexpected ice formation and associated uncertainties in the The leading edge is continuously heated and qualification process of a small turbprop temperature controlled, originally between 40 'C powerplant. and 50 "C. The group of continuously heated elements also comprises strips at either side of the gap; a strip 1. just helow the cavity of the seal of the intake and a strip an the adaptor flange. The aircraft is a twin-engined turboprop, designed Associated power-density is 12,5 W/in2. for operation in icing conditions to FAR/JAR 25, Appendix C. Figure lb shows the heater sections for the The engine is a three-shaft turbo-prop in the 2000 original duty-cycle down to -7,5 "C. horsepower-class with a maximum pressure-ratio of 15,2 (max cruise). The composite propeller is driven by the two-stage free turbine through a reduction gear-box. The engine air intake is excentric and is provided with a by-pass channel for foreign object separation. The engine has flexible mounting and consequently a flexible element is required between engine-casing and intake duct. A 0.2 inch (5 mm) gap with recessed compression seal (silicone elastomer) is located between a light alloy adaptor flange on the engine and a flange on the composite duct. Figure la gives a cross-section of the construction. The abovementioned gap runs acmss areas of water or 'snow impingement. The composite air intake is de-iced by an electro-thermal system, utilizing semi-conductor TO -7.SC switching. 0 8 SEC I3MlN It consists of five sections which are cyclically F3 8StCl 3MlN heated: the pmpllar blades and spinner, the inner 0 2r8SECi3MlN lip 1 and 2 surfaces, the duct floor and the duct 12 SEC 16 MIN side-walls. CONTINUOUS L0-5D'C The control of cyclically heated elements is dependent on outside air temperature. The power-density for the cyclically heated parts Figure lb : Engine air intake de-icing is 12,5 W/in2. configuration. 5-2

Powerplant de-icing is governed by two (w1/RH) 3. 'nci8ents De-icing Control Units, which monitor various input and output parameters and are capable of detecting During the winter of 1988 with more than 30 partial system malfunction. aircraft in service and about 50.000 hours of flight accumulated, several occurences were .. . reported of a sudden temporary loss of torque. 2. sts In -general. the so-called flame-out incidents occurred under the following conditions: The engine was qualified to FAR 33 and JAR E. - flight-phase: descent, power 10 to 35% torque FAR 33 addresses: water cloud icing conditions per - IOAT -2 to +3 "C Appendix C of FAR 25 and ice ingestion, in - altitude 8.000 to 14.000 ft particular ice from the air intake, a 1 inch - IAS 180 to 180 kts hailstone and water, 4% of the engine core massflow Analyzine~- a total of 8 confirmed occwrences the by weight. following common factors were recognized: - incidents occurred at relatively low Ice ingestion tests at sea-level with a typical air power-settings intake system yielded a maximum ice volume and - together with the altitude effect, this meant shape that could be ingested without damage to the that engine massflow was low engine compressor; an associated limit for ice - powerplant ice :protection system had been ingestion was declared in engine certification activated prior to the incident, together with documents. continuous ignition Since mechanical damage was considered the limiting - prior to each incident the aircraft had factor, most tests had been conducted at relatively encountered light icing conditions, while in most high engine speeds, i.e. high engine power- cases rain was reported by the flight-crew and in settings. some cases snow Operational limits for ice ingestion associated - however, no or little ice was seen on the with flameout were not defined, as by tradition airframe flame-out was supposed to he covered by sea-level - in each case ambient temperatures were near water ingestion tests. freezing level The powerplant was certified to FAR 25 and JAR 25. The abovementioned sudden losses of engine power FAR 8 25.1093 specifies water cloud icing were characterized by a drop in torque to zero conditions per Appendix C and "snow, both falling percent. followed by a drop in gas temperature. and blowing"; however, temperature, density, Next, when gasgenerator speed had dropped to the waterhce ratio and crystal structure are undefined level where relight is possible, power restored (ref.: AC 20-73 and FAA ADS-4). automatically. In the absence of definition, test-facilities and The aircraft did ;not leave its intended flight-path test-traditions in Europe, the paragraph addressing and flight safety was not impaired. snow has heen deleted from JAR 8 25.1093. ACJ 8 25.1093(b) recommends water cloud conditions Pceviously, there had heen several emsures to at altitude for qualification of the powerplant. severe water cloud icing conditions in which the powerplant ice protection system performed a8 For Drevious amlications__ of the endne,-. the designed and engine operation remained undiaturbed. powerplant had been subjected to icing tests in a sea-level tunnel-facility. In response to the first incidents operational Consequently, for subject derivative engine this restrictions were imposed, mainly addressing power tradition was followed with Liquid Water Content as a means of increasing engine tolerance to ice adjusted for the difference in airspeed. ingestion. The majority of the tests, including ice shedding However, the power needed to obtain the intended after delayed activation of the de-icing system, benefit was too high for regular operations. were performed at high power-settings, representing especially descent. prolonged conditions of ice encounters. Note that Am 9 25.1093(b), 2.4.3, 2.4.5 and 2.4.7 Evidently, an unknown phenomenon, not experienced suggest ice shedding at high engine power. during qualification testing, had to be identified At idle power-settings satisfactory ice protection before an effective solution could be defined. capabilities were easily demonstrated since electrical de-icing is independent on engine-speed. 4. In addition to abovementioned ice-tunnel testing a -- fully-instrumented prototype-aircraft was subjected Following aforementioned incidents, investigations to dry air tests and natural icing trials. were initiated to retrieve the cause of the power- Dry air tests comprised a wide range of interruptions. testconditions, including (simulated) failure A first series of flight-tests was conducted to conditions. analyse the engine tolerance at altitude to sudden Flight-tests in natural atmospheric icing water ingestion. conditions were executed to verify powerplant ice Water was used, because it could easily be injected protection system effectiveness, especially in repeatedly into the engine compressor in relation uitb system delayed activation, ice well-defined quantities at known rates. accumulation on unprotected areas and aircraft flight handling characteristics. The aircraft installation consisted of a large Continuous maximum icing conditions were found; hydraulic cylinder; total volime of water to be subsequent encounters with de-activated ice injected ranged up to 0.53 US gal (2 litres) protection system resulted in ice build-up that maximum at a flow-rats of more than 0.53 US gal/s would have accumulated in intermittent maximum h%5"). icing conditions. During flight-tests no anomalies marred; aot even More than 150 vatep injections were performed in when the engine ingested about 0.5 inch (13 mm) of various configurations at altitudes between 5.000 ice accumulated on the intake-lip. and 18.000 feet altitude at different It was concluded that the powerplant ice Protection power-settings (torque); a total of 73 flame-outs system performed satisfactorily. were induced. 5-3

Boundaries for flame-out were established at With video observation in natural snow conditions several anti-surge bleed (Handling Bleed) significant ice accumulation was found in the area massflows; it appeared that more bleed increased of the adaptor flange. the engine-s tolerance to water ingestion. The unheated slot into which a considerable portion However, from the testresults it was concluded that of ice keyed, formed a comfortable point of flame-out behaviour was rather different from that attachment for precipitation which otherwise would reported in the incidents, which therefore remained have been repelled from adjacent continuously unexplained. heated strips. Ice lodged in the cavity of the seal and bridged Later, comparative water and ice ingestion testa across the continuously heated strips to the were conducted on a sea-level testbed. cyclically heated side-wall. Ice appeared to be much more critical regarding Complete shedding waa inhibited by the ice being effect on engine behaviour, also on a basis of mass keyed; Figure 5b gives an impression of the way ice times enthalpy. mushroomed in the seal gap.

From the tests it WBB concluded that the flame-out ENGINE HEATING-STRIP IAOAPTORI incidents were unlikely to have been caused by ingestion of rain or water accumulated in the engine air intake. However, it was recognized that tolerance to water ingestion reduced considerably with increasing altitude independent of power-setting; restoration was achieved by alteration of anti-surge bleed valve control. ING-SIRlPlDUCTI The tolerance to water ingestion increased significantly with 8% Handling Bleed. Finally, it was concluded that engine behaviour with sudden water injection and increased SIOE-WALL CYILILALLY HEATED anti-surge bleed had no relation to that with ice ingestion (at sea-level); consequently, further ice ingestion tests were required. Figure 5b : Principle of ice accretion in the seal cavity. 5. Flieht-tests: video Ice accreted until aerodynamic forces prevailed over the adhesive forces between ice and duct wall As a result of the incidents, the engine air inlet and the strewh of the foundation provided by the de-icing system was re-evaluated to observe ice unheated slot. build-up and shedding characteristics inside the Periodic shedding wae frequently observed; the duct in natural icing conditions. central part disappeared in the by-pass duct, Ice accretion was recorded using three miniature whilst the remainder was sometimes ingested by the colour video cameras mounted at different positions engine. in the air intake of a prototme aircraft. This mechanism appeared to be dependent on the dimensions of the accumulated ice, whereas also The fields of view were chosen as follows: power-setting appeared to influence patterns of ice 1. rear of air intake duct including the shedding; Figure 5c shows the location of ice bifurcation and the lower half of the engine accumulation and the direction of shedding. inlet interface 2. floor of the intake duct 3. top (ceiling) of the intake duct and forward VERY LIGHT RUNBALK ACCRETION part of the engine inlet casing Figure 5a shows the positions of the cameras in the intake duct.

STICKING AREA GOES UP OR OOWN DEPENDING ON THICKNESS ORiANO POWERSETTING

BY-PASSOUIT

VIEWED FROM LAfiERA ON INLET LIP Figure 5a : Positions and fields of view of cameras Figure 5c : Location of ice accumulation and mounted in the air intahe duct. shedding pattern. Initially, exploratory flights were conducted in Apart from the seal area, ice accumulated on the atmospheric conditions of opportunity. including: (lefthand) rear siderall (influence of light cloud icing conditions, snow, hail and rain. propeller-swirl) where it accreted between the As a result of these video-observations it emerged heating-cycles to such an extent that it sometimes that significant ice accumulation developed at the resisted complete shedding. pear of the intake duct at the bifurcation. Further flight-tests were performed with only one In general, the observed ice accumulation in the video camera mounted on the intake lip, viewing the area where it could potentially enter the engine rear of the intake duct. under critical conditions was estimated to have a volume of 5 to 8 in" (82 to 98 una). 5-4

These figures are estimated by pmjecting the To completely eljminate ice accumulation in the sidewall and seal surface as a triangle of 30 inz vincinity of the seal gap, de-icing of the slot hy (196 cm2) with an average thickness of 0.16 to 0.20 means of engine compressor bleed air was inch (4 to 5 m). introduced; Figure 6a shows the basic principle. An asymmetrical sprw-pipe waa inserted in the seal The largest volume of ice which was seen to be cavity. covering an area reaching forward of the ingested by the engine was estimated to be 3 to by-pass duct range and extended at the 4 in3 (49 to 66 as); it did not produce a leftband-side to the forward radius of the engine flame-out condition. nor was engine operation aperture to account for any asymmetric ice noticeably affected. accretion due to propeller-swirl. However, it is quite conceivable that double that volume would occasionally have entered the engine, which is indeed in the order of the volume that is ENGINE CASING presently estimated to cause a flame-out at altitude. Note that ice accretion as described was formed in mainly wet snow conditions; OAT -7 to -2 'C at an altitude of 6.000 to 14.000 feet. with an airspeed of 150 to 170 knots (IAS). At the same time no or little ice was collected on windshield wipers. wing leading edges and air intake lip. These conditions match those associated with reported incidents quite well. Hence, it was concluded that the ultimate cause of the flame-out incidents had been traced. Ice protection capabilities in cloud icing conditions were judged satisfactory and quite Fimure 6a : Seal anti-icing by means of compressor similar to those previously qualified in sea-level bleed air impingement. tests. Protection against conditions of wet snow and mixed The spray-pipe haa a 3/16 inch (4,8mm) diameter, conditions was upgraded by modifying the air intake incorporating 88 holes of 0.04 inch (1 mm) at a de-icing system as will be outlined in the next pitch of 0.28 inch (7 ma). parsgraph. Total bleed air consumption under cruise conditions is about 2 Wmin (0,015 kg/s) at 150 "C Effectiveness was demonstrated during natural icing (1.0 lb/min.ft; 0,025 kg/s.m). trials with a prototype-aircraft with one The flexible seal was provided with slots to feed powerplant de-icing system modified, the opposite the spray-pipe by meanB of flat radial tubes from a system unaltered, both monitored by meam of video common manifold. observation. Figure 6b shows a bottom-view photograph of the Subsquently, the effectivity of the modifications production spray-pipe. was verified and confirmed during and in-service trial for about 210 days (1260 flight-hours; status The de-icine- flar is controlled bv a solenoid April 1990, trial is continuing) also utilizing a valve, while system operation is monitored by a small video camera on the air intake lip. pressure-switch. A flow-limiter protects against excessive flow in In general, video-recordings were of high quality the event of failure of the supply-pip. taking into account the variation of natural Figure €e shows the system general arrangement. (day-)light inside the air intake due to changing atmospheric conditions and changing aircraft heading and attitude; details of both intake duct 7. Conclusions.- and engine inlet could be clearly distinguished. The commercially available video system exhibited The following conments are relevant to the design excellent serviceability (EmCCD Calor Camera, and qualification process of the de-icing system of Model Bn 102; diameter 0.7 inch (17.5 mm), length a typical small powerplant: 2.1 inch (53 mm)). - Detailed advisory material for qualification to airworthiness requirements and traditional 6. I"y& ice DPO- testing may not address certain critical (natural) icing conditions. Engine de-icing system performance was improved hy In the case of the subject powerplant, wet mow raising the control temperature of the sensors appeared to be critical. embedded in the leading edge of the intake lip; these sensors also control the continuously heated - Small high pressure-ratio (turboprop) engines may strips at the edges of the gap at the engine be surprisingly sensitive to ice ingestion at interface. altitude in term of flame-out; traditional ice Control temperatures were raised to 50 "C falling, ingestion tests addressing mechanical damage do for activation of the heating-cycle and 70 'C not reveal engine tolerance to flame-out. rising, for de-activation. - An engine air intake having an internal Associated with this modification the bifurcation (foreign object separation) or any cycle-interval of the sidewalls was halved to once other stagnation area, may be prone to ice every tbee minutes. accumulation in wet snow conditions: OAT -10 to Originally, this cycle was determined to minimize 0 OC, TAT -5 to t5 "C. the chances of runhack ice in the engine-casing. These conditions are generally not identified hy which poaed a potential hazard of damage to the the flight-creu in the absence of external ice impeller upon ice ingestion. accretion. However, this was based on ice-tunnel testing rather than in-aervice experience. In addition, ice build-up in wet snow conditions - finally, it is concluded that commercially appears to be different from that in cloud icing available miniature video systems are quite conditions, which are subject of traditional suitable for observation of ice accretion inside qualification tests. an engine air intake during flight in natural icing conditions. - sudden water ingestion tests at altitude are Performance in test-aircraft as well as in an unlikely to produce the effects of ice ingestion; operational trial was excellent. these tests are no substitute

Figure 6h : Spray-pipe mounted on adapter-flange.

Figure 6c : General arrangement of bleed air supply to spray-pipe. 5-6

Discussion addition, the engine tested at NRC was basically a gas generator and only airspeed was varied to account for different flight conditions. In-flight video observations of 1. W. Grabe, NRC Ottawa the flow pattern inside the air intake duct indicated that, for Only light icing conditions were noted by pilots prior to instance, propeller swirl significantly influences the location flameouts, yet this problem was not encountcrcd under of stagnation area heavy icing conditions at NRC. How come? While problems were caused by wet snow, supercooled 2. R. Toogood, PI.dtt and Whitney Canada water icing at NRC should show the weak areas at the intake Would you care to comment on the differences in inlet split. Why did it not do it? prototype configurations used in the ice tunnel and that of the initial production configuration? Author: Would you agree that an important conclusion of this study At the moment of the power interruptions, the crew is that discontinuities in surface ice protection should he reported to encounter light cloud icing conditions. However, avoided at critical impingement locations? prior to the flame-outs, the aircraft had during a certain timc-interval been exposed to what can best he Author: The differences between the prototype characterized as mixed conditions of (wet) snow and configuration last tested in the icing tunnel and the initial (supercooled) water droplets. The conditions tested at NRC production configuration of the intake duct basically involvc during qualification only addressed cloud icing conditions (software) changes to the controller. However, these arc at sea-level in accordance with the requirements as laid subordinate to the introduction of spray pipe, which,is the down in IAREAR (ACI) 25. It appeared that engine result of testing in natural icing conditions rather than ice tolerance to ice ingestion at altitude when being exposed to tunnel tats. natural icing conditions (wet snow) may be quite different. One of the conclusions of the paper and the presentation is First of all, conditions tested at NRC AM and CM) may not indeed that discontinuities in stagnation areas which are he that severc as the mixed conditions experienced in exposed to impingement of precipitation involves a high natural precipitation conditions (i.c. wet snow, hail, rain). In design risk. ICE TOLERANT ENGINE INLET SCREENS FOR CH 113/113A SEARCH AND RESCUE HELICOPTERS

R.B.Jones Senior Project Engineer Boeing Canada Arnprior W.A.Lucier Manager Engineering Boeing Canada Arnprior Baskin Drive East Arnprior, Ontario KFS 3MI Canada

SUMMARY regimes, In icing conditions though, the CH113/113A screens had exhibited a tendency to become congested The Canadian Forces CH113/113A Search and Rescue with ice, and consequently the screens have to be Helicopters occasionally encounter unavoidable icing removed for winter operations. The unprotected conditions in their operating environment. The engines are exposed to increased risk of FOD during original engine inlet safeguards were not designed for, operations in this period. nor capable of sustained operations in icing environments, necessitating removal of the inlet The CH113/113A Helicopters are rated at a maximum screens in these conditions. This arrangement resulted all up weight of 21,400 pounds, and are powered by in unacceptable risk of foreign object damage to the twin General Electric T58-8F engines. The engine engine, and compromised operational safety. features a 10 stage axial in-line compressor, and variable inlet guide vanes to ensure efficient and stall Ice tolerant inlet screens have been developed as a free operation. The military power rating for each remedy to this problem. The flat faced, inverted cone engine is 1350 shaft horsepower. At normal SAR screens with a bypass opening accommodate operating weights and full fuel, the aircraft lacks progressive ice congestion during the various single engine capability with the present power operational modes with minimal engine performance available. Canadian Forces experience has shown that degradation. Comprehensive testing has confirmed the the inlet guide vanes and first stage compressor blades capability to operate in simulated severe icing of the T58 engine are particularly vulnerable to FOD conditions for a minimum of thirty minutes, meeting in the absence of protective screens I. the objective "get out of trouble" capability. Operational experience and testing has substantiated The inlet configuration for the CHI13 helicopter improved tolerance to the full range of meteorological consists of a straight cylindrical duct. This conditions for the fleet of CHI13 Helicopters. arrangement, with minimal obstruction, is known to offer virtually nndistorted inlet flow and the recovery Activity is continuing to equip the remaining CH113A of a large percentage of the ram air effect, although Helicopters with the ice tolerant engine inlet screens. offering little protection from sand and small particle Other international operators of the Boeing Helicopters debris. Electrical resistance heating and engine bleed Model 107 Type helicopters have expressed interest in air are employed to prevent ice build-up. The original the developments achieved, and independent particle separator system used with the CH113A evaluations are under way. variant provided improved resistance to debris damage, at the expense of a somewhat obstructed inlet. This latter arrangement also provided heating to breeze surfaces, although operational experience and 1.0 INTRODUCTION testing has shown that ice could develop at the separator in some conditions '. Small gas turbine engines generally lack the tolerance to foreign object damage (FOD) and ice ingestion Several factors have contributed to the need for problems afforded to large turbofan powerplants in improvements to the engine inlet protection systems commercial jetliners. As such, inlet screens and for the CH113/113A. The very nature of search and particle separators are employed as protection to the rescue (SAR) operations, particularly in light of vulnerable compressor blades in many helicopters and Canada's geography and environment, places a high smaller fixed wing aircraft. The Canadian Forces demand on availability and reliability. SAR missions CH113/113A Search and Rescue Helicopters, are most often conducted in coastal or mountainous derivatives of the Boeing Helicopters 107 series regions and frequently in adverse weather conditions. tandem rotor helicopter, were originally equipped with Although the aircraft was not designed, nor authorized conical wire mesh screens to minimize FOD problems for flight into icing conditions, tolerance to for its two internally mounted engines. The CH113A unexpected, unavoidable icing conditions, or marginal helicopter also employed a particle separator. These conditions is highly desirable in the SAR role and systems proved to he effective at minimizing engine particularly for a "get out of trouble" capability. The damage due to foreign debris under most operathg consequences of an aborted mission or a forced 6-2

landing could he serious due to both the role and a similar way to a mountain rescue involving altitude environment served by the CHI 13/113A. change on the West coast, and it is the variability of conditions: wet SIIOW, dry snow, cold weather icing, The limitations in single engine performance provides warm weather icing and perhaps most worrisome, further impetus for enhanced protection against FOD freezing rain that presents the real flight problem. and ice ingestion problems, underscoring the need for Trying to predict the onset and effect of potential icing full time safeguards for the engine. The loss of a conditions can he humbling even when the single engine is considered a critical emergency for consequences are less severe, such as when operating the CH113/113A. The risk of dual enginz failure can a personal automobile. It is not unusual to observe be significant in some circumstances of foreign debris trees and fences glazed with freezing rain, and yet no ingestion, particularly during any in-flight ice shedding accumulation will form on the windshield. or severe inlet congestion due to ice or w6t snow. Mysteriously, in what would seem to he comparable conditions to the casual observer, keeping even a Although the possibility of developing a modified inlet heated windshield clear of ice build-up will he a duct system was initially considered, this option was virtually impossible task. dismissed in favour of an ice tolerant inlet screen due largely to the limitations inherent to the internal The rate of icing on the different surfaces is important mounting of the engines. The problem addressed in for the type of screen considered and this introduces a the current work was therefore to develop a FOD ice further parameter i.e. the local air flow. A lesson screen to be installed externally in front of the straight learned at the start of testing invalidated an assumption duct. This approach was expected to yield a passive that the blade downwash air travelled directly device to he employed in all seasons with minimal downwards in ground effect hover and ground run. It change to the existing engine inlet and duct was found that the cross flow caused by the blade drag configuration. was very significant, especially for the ice accretion rate from the side of the engine inlet.

2.0 ICING CONDITIONS Earlier work with an ice detector installed on the helicopter did not yield successful results, leading to The logical start to an analysis of the rate of ice the removal of the detector. It was possible that it accretion is the definition of the icing conditions, and would mislead the air crew, especially as to the type herein lies an intensive challenge due to the variability of icing. The two basic types of ice formation of these conditions. While modelling the ice accretion experienced during tests were clear glaze ice and rime rate for one particular set of parameters may he ice (milky white). The glaze ice varied from hard to achievable, the certification requirement and the real soft, which introduced the danger of it breaking off, world icing conditions are subject to a broad range of hut the rime ice wfis always hard. This was of variation in the contributing factors, including: particular influence in the final detail design which atmospheric parameters, surface properties, and had to he refined because of the type of build-up on engine operating condition and flight parameters. different parts of the screen. The same also applied to the snow conditions: the wet snow often produced a Particularly difficult to define is the starting point, Le. soft glaze ice which could easily break off and enter the icing cloud and the snow cloud. The Coalescence the engine. Theory model for precipitation gives a mechanism of rain drop growth in warm cloud, whereas the The certification requirements, as defined in Wegener-Bergeron-Findeisen theory gives a cold cloud Ainvorthiness Regulations FAR 25 (which is not model in which supercooled water droplets and ice strictly applicable for military helicopters), do not coexist, hut coalescence takes place in such a cloud as completely clarify the icing conditions and accretion well ’. These two models are not mutually exclusive rates to be considered for the design case 4. The nor contradictory, therefore the liquidkolid water certification requirements also provide little guidance content is yet another variable. As well as this, the with respect to the failure criterion for different flight dependence of the type of icing cloud or snow conditions. There are obviously two extreme cases: conditions on the suspended atmospheric particles adds flame out due to air starvation and engine failure due more to the complex picture. This is also influenced to FOD resulting from an ingested ice piece. In either in ground effect hover with recirculating snow and ice case for the CH113/113A, a single engine failure laden air, resulting in a localized whiteout. would result in degraded performance and a mission abort. In the case of a dual engine failure, an The problem of variability of atmospheric conditions autorotation to the ground or sea surface would he for the Search and Rescue role for this helicopter in necessary unless power can he recovered. The regime Canada is aggravated by seasonal and geographical of flight and the local terrain or sea state would play realities. Canada has long coast lines on both the an important role in, the safe execution of the Atlantic and Pacific Oceans, and a vast interior autorotative landing and survival. It would certainly speckled by an array of freshwater lakes that are he desirable to work towards some intermediate failure among the world’s largest and most numerous. This, criterion, such as 70% screen blockage, versus the combined with our ranges of terrain and latitude, extremes raised with their inherent loss of affected contributes to a full spectrum of icing conditions that engine power. may he encountered. A rescue on the East coast far out into the shipping lanes can face weather changes in 6-3

As a result of the difficulties in rigorously modelling removed and replaced with a straight heated duct, the range in ice accretion rate, it was determined that comparable to the CHI13 configuration. This new testing and closely monitored prototype installations duct provided a improved heating capacity versus the would most effectively cover this aspect of the prior arrangement, and commonality for any development. Icing simulations were conducted at the subsequent modifications evolving from the present National Research Council Canada (NRC) Ottawa test investigation addressing the inlet screen icing problem rig facilities. No testing was done with artificial snow as previous studies reported unreliable results, and it A Boeing Helicopters proposal to the Swedish Armed was felt that natural snow, whether wet or dry, would Forces in 1974, and work done by Kawasaki Heavy yield more credible testing '. Design features Industries, describes a heated wire mesh screen similar influenced by the icing conditions would call upon the in geometry to the original conical screen. This test findings, as well as past experience in similar solution was burdened with prohibitively high developments, such as that for the CHI47 Chinook. manufacturing costs per screen on small production runs combined with excessive electrical power requirements: approximately 15 KW per screen to 3.0 OPTIONS TO REMEDY THE ICING anti-ice 40 per cent of the screen surface under PROBLEM extreme icing conditions for short periods (up to 15 minutes), and 5 KW continuous under moderate icing Several candidate solutions to the CHI 1311 13A engine conditions. These factors negated the feasibility of a inlet icing problems were considered. The number of heated variant of the conical screen. potential remedies evaluated was reflective of the difficulty in fully defining the scope of the design parameters, as raised in the preceding discussion on icing conditions. The parameters considered in the A "dog house" type of protective inlet cover has been evaluation of the candidates included: developed and used successfully by Columbia Helicopters. The simple arrangement is based on the a) icing performance concept of an enclosed space with a small entrance b) suitability for year round FOD parallel to the incoming air stream. Dry air can be protection, drawn in, leaving the moisture laden air and any c) inlet flow distortion, suspended FOD particles to rush past. This concept d) aerodynamic drag penalty, and was also studied b) Kawasaki Heavy Industries from e) electrical power demand. 1971 to 1981 for the 107 Type helicopter. The concept was dismissed because of potential for ice The solution commonly employed for inlet protection accretion and ingestion in severe icing conditions. has generally been integrated into the duct, such as an The Columbia experience however, has proved more S-duct, a vortex generator bank, or an axisymmetric positive and the dog house configuration remains in duct with swirl vanes 6. These and any similar service on their fleet of 107 Type helicopters. solutions demanding extensive modification to the engine inlet duct configuration were immediately A variation of this concept is a panel or screen cone in discounted due to physical space restrictions and front of the intake inlets pointing into the air stream so program constraints. It should be noted that these that the air has to go around the corner lip to enter the criteria dismissed several possible solutions that could engine. This concept was not studied in detail due to be of merit in other applications. Only solutions its similarity to the original cone and because of the external to the inlet duct were considered further in anticipated penalty in summer with the absence of the current investigation. FOD protection.

First among solutions considered were modifications to, or enhanced variants of, the original conical screen arrangement. This system had proven to be effective at preventing FOD ingestion in most environments, but operational experience and NRC ice test rig results x confirmed its vulnerability to ice congestion. The test results suggested that in some conditions, sufficient ice build-up could occur within a few minutes to cause the engine to flame out. The aircraft operating instructions reflected this limitation by directing that the screens be removed in possible icing conditions. With these screens removed the engines were open to IcelFOD ingestion, and engine damage occurred in this configuration.

It was suspected that the original particle separator utilized on the CHI13A was also responsible for a number engine failures caused by liberated ice developed at the separator. Testing in the NRC ice rig verified this possibility; the particle separator was 6-4

Kawasaki Heavy Industries developed an interesting In ground effect hover or ground running, the inlet air double screen with bypass. It is not known whether flow is less unidirectional, and the screen surfaces the concept had been prototyped, or if the large frontal would be expected to ice over more uniformly. area would introduce excessive drag. One interesting Continued operation with such ice development would aspect of Kawasaki's testing of this screen was the necessitate some form of open bypass to maintain a evaluation of Ice Repellant Coatings. Screens tested path to feed the engine air. The restricted and with and without the coating under simulated snow distorted air flow in this contingency mode of conditions unfortunately were almost simultaneously operation was expected to yield reduced but continued blocked by snow and ice. However, only a minimal engine performance. shedding force was required on the Ice Repellant Coated screen to remove the accumulated snow. In both icing and non-icing conditions, the flat faced In the 1950/1960's, Boeing Helicopters explored a design concept utilizing a flat face engine inlet screen. screen arrangement would provide full time protection against fod ingestion for particles larger than the wire The approach, like the dog-house concept, exploited mesh void size. Any bypass opening should also the expectation that an air stream would shed the provide shielding against direct ingress. majority of its entrained liquid water in flowing around a sharp obstruction. In this case, the ice The flat faced screen, including the suggested tapered congested flat face would force the axial flow of inlet sidewalls and a bypass feature, was selected over the air to accelerate radially inward around the face in alternate configurations to be the thrust of order to enter the engine through the side of the development. It has already been emphasized that this screen. Any airborne water droplets, snow or ice choice of final solution for this helicopter installation would safely pass the engine inlet. does not detract from the merits of other solutions for different configurations. The results of the initial Boeing Helicopters investigation proved the viability of the flat face The following table summarizes the options screen arrangement, and recommended that the considered, their merits and their drawbacks for the original cylindrical sides to the screen be tapered current application. Presented are the various screen toward the inlet, forming an inverted cone, to induce arrangements discussed herein, as well as the plenum greater separation of any moisture content. The flat inlet arrangement, and the option of going to an faced screen design, however, was never fully externally mounted engine with an inlet screen developed, nor tested for the Model 107 application. configuration similar to that of the Chinook Further investigation into the performance of the flat Helicopter. faced screen with cylindrical sides was conducted at the Naval Air Propulsion Test Center, Trenton, New Jersey, in November, 1974. Their tests in up to 20 minutes in severe icing conditions resulted in a fully congested screen face, but little ice build-up on the side mesh. Based on these positive findings, and the applicability to the CH113/113A, this solution concept was resurrected in the current investigation.

ICE, roo. DRY SHOW PMLOSS NOT FUSIBLE

The flat faced screen features accommodated a number of modes of operation. In fast forward cruise in icing conditions the front face ices over rapidly and the air I I has to turn sharply to enter the sides of the cone. The sides remain clear of deposit in forward flight, especially with the introduction of the cut back, and would thus provide the path for inlet air. 6-5

4.0 DEVELOPMENT OF THE ICE theoretically verify airflow characteristics through the TOLERANT INLET SCREEN air inlet to the engine, and allowed for testing to verify that the new design did not introduce any The process of designing and optimizing the ice detrimental turbulence to the airflow. tolerant inlet screens for the CH113/113A helicopter, based upon the flat faced inverted cone concept, was The determination of screen geometry and the sizing largely iterative. Although recognized as a protracted of the various elements was the first task undertaken approach, these iterations were necessary due to the in the development of the new screens. The basic difficulty in anticipating the icing, screen and engine strategy utilized was to size the inverted cone sides to performance in the range of meteorological and suit the geometry of the inlet duct while achieving a operating conditions to be encountered. flow area not less than that provided by the screen being replaced (Le. 418 sq. in.), even in the case of a The principle design features pursued in the fully congested front face. The diameter of the development were: inverted cone rear ring was chosen so as to enclose the streamlines of the air entering the intake duct. A flat front screen face perpendicular to The cone recess angle was chosen largely based upon the streamlines. During most (Le. non- physical constraints, although it compared favourably icing) conditions, engine intake air would to that earlier proposed by Boeing Helicopters. The he expected to enter the inlet via this angle was deemed suitable in that it was adequate to front face. induce water shedding but would not yield an excessively large screen face. The length of the An inverted screen cone with sides truncated inverted cone was thus governed by the parallel as possible to the stream lines of requirement for side flow area. By extension of the the air entering the engine. In the case geometry of the inverted cone, the face diameter was of a congested front face, engine intake established. air would be expected to enter via the side panels. The bypass dimensions were chosen so that the minimum open area shown in the table was set equal A shielded open bypass area in the to the normal open intake (i.e. 65 sq. in.). screen to facilitate continued, although reduced, engine performance in the extreme case of a fully congested inlet screen (i.e. both face and sides). OPEN MESH /I(SETN.) (SQ. IN.) A solid fibreglass attachment collar to ORIGINAL CONICAL 418 235 serve as a flush interface to the existing CH113: SIDEWALL 575 324 inlet fairing assembly. BYPASS NA H 65 The major design objectives were to: CH113A:SIDEWALL 575 BYPASS 65 I II,I I U a) minimize the aircraft performance penalty of aerodynamic drag induced by The basic principle that air loses its liquid water the inlet screens, content in bending sharply around the blocked flat face of the ice tolerant inlet screen suggests that the front b) minimize the engine performance penalty face should be normal to the streamlines of the resulting from intake air starvation or impinging local airflow. For a helicopter the direction distortion, of the streamlines varies with flight condition from hover to forward cruise. It was concluded that the ice c) minimize the probability of any ice accretion within the screen that would tolerant inlet screen's front face should be oriented normally to the forward cruise airstream to maximize introduce the risk of damaging ice ingestion, and its performance in the "get out of trouble" case. The optimum angle of incidence of airflow to the engine provide full time protection against FOD. inlet for forward cruise at 110 knots and blade d) downwash effects was derived as follows with vector addition of the primary airflow sources. This derivation excludes any influence of aircraft pitch, or 4.1 Inlet Airflow Requirements airflow disturbances induced by aircraft aerodynamics or local inlet effects. The project assumptions were that the existing air Forward Speed, V = 110 knots = 186 induction system with the original conical screens met all CHI 13/113A flight envelope requirements, and ft/sec.(corresponding to Vne at therefore, any new screen design, as a minimum, had maximum gross weight ') to maintain the diameter and area of the engine inlet opening, and screen open mesh area. These Air Density, p = 0.0023769 slugslft3 assumptions precluded the need to calculate and 6-6

The helicopter geometric data is shown below. This vertical velocity is for an ideal actuator disc, however, for a tandem rotor helicopter the disc efficiency is between 85% and 95% depending on the speed range and for the worst condition can drop as low as 70% with ice on the blades.

Consider the flow just from the front rotor which has a horizontal component due to the 9"tilt of the blade disk

As seen there are three flow regions below the two sets of blades; the first is outside the radius of both blade sets, the second is in the region where the blade Flow Through Disc set overlap, and the third is the area under each of the blade sets where they do not overlap. The engine intake air can be drawn from either of the last two regions depending on the forward speed so both cases will be analyzed. ft V, - 3 2 .9x tan@") = 5.21 - Case 1 INDEP DISC TOTAL AREA sec +) 50 x2 v,- v,+ v ft V-1lOk- 186- -3927 ft2 sec

Case 2 OVERLAP DISC TOTAL AREA ft VH- 186 + 5.2 1 191.2 1 - - sec

= ll0k = 186 ftlsec so the total horizont; velocity at engine inlet Where A;. sin20cos20~25~ = 186 + 5.21 = 191.21 Wsec = 191.21 ftlsec and the total verti,cal velocity at engine inlet = 32.94 200 ft2 Wsec.

:.A- 3927 -2x(273 -200) 191.21

= 3781ft2

From momentum theory for an ideal actuator disc

Vertical Lift = W = 2 AV;. mef. 8)

Where W = Weight of Helicopter Ob), 32.94 P = Air density (sluglf?), ;. tan I3 - -- A = Rotor Area (FP) 1.91.21 V, = Vertical air speed Ftlsec. 13-9.770

The angle of the flow to the ENGINE CENTRE LINE = 9.77 (For the case of ice build-up V = 1.3 V, , the angle is 12.62").(Ref. 9) Case1 Vv- 19500 2~0.0023769~3927 Therefore, the calculated optimum orientation of the -32.31- ft sec screen faces would be tilted upward approximately 10 degrees from the horizontal, while slightly divergent from the plane of the aircraft centreline. It is Case2 V,- 19500 interesting to note that the rear rotor set downwash is 1] 2~0.0023769~3781 only effective in the inlet region if the angle is 23 degrees. From geometry, if the forward flight speed -32.94- ft is greater than 45 knots, the engines inlet aimow is sec only influenced from the forward rotor. 6-7

The inverted front cone dimensions and the optimum Another interesting aspect in the design analysis was angle corresponding to cruise conditions enabled the the determination of the structural contribution of the location of the left front cone to be centred just in screen mesh, whose stredstrain relationship was front of the intake face. However, when integrating found by test to be inherently non-linear. the front cone to the attachment collar, the desired angle of incidence could not be maintained for the right hand screen due to interference with the 4.3 Prototype Development synchronizing shaft cover (tunnel) in front of the right hand intake. An inclination angle of 17 degrees was The earliest prototype developed was modular in the minimum allowable for the given screen concept. This design was applied to the CH113A, dimensions at the right hand side. This physical whose inlet configuration differed slightly from that of constraint superseded the optimum calculated the CH113. Unlike subsequent models, the bypass orientation. This angle of incidence would be aligned feature was not integrated into the side of the screen. to the expected inlet air flow at a forward cruise speed An optional bypass was achieved with interchangeable of 64 knots. front faces - one flush face providing no bypass (i.e. suitable during non-icing seasons), and one face Tufting tests conducted by AETE during the incorporating a forward bypass. certification phase of the final screen configuration substantiated the adequacy of the orientation for the Testing plans for this first design were comprised of screens. engine test cell work, intended to observe the flow distortions and engine performance at various stages of The above geometric constraints were employed as blockage, and icing tests. Due to program constraints, design objectives in the manufacture of the first difficulties in emulating the aircraft inlet, and test prototype screen, and in fact remained fixed facility problems; the findings of the engine tests with throughout the development and final design. CAD the first prototype were inconclusive. facilities were employed and very effective at implementing these critical geometric parameters in The simulated icing test plan was comprised of 28 preparing the engineering drawings for the designs soaies for a total of 14.1 flying hours. The results of completed. these icing trials suggested that'the flush faced screen did not provide a sufficient time margin in icing to enable the pilot to recover safely before engine flame- 4.2 Structural Considerations out due to blockage. Ice build-up was also found to occur in the yet unchanged particle separator duct, The basic materials selected for the build of the which led to a engine FOD incident. This arising prototype remained unchanged throughout the contributed to the decision to eliminate the particle development. The screen mesh for the various inlet separator for the CH113A helicopters. areas was type 304 stainless steel wire with 1/4 x 1/4 inch mesh. The underlying structure was This arrangement was abandoned in favour of an aft manufactured of welded type 321 stainless tube, plate bypass system for subsequent designs. and sheet. The attachment collar was fabricated with a fihreglass - vinyl ester composite moulded to conform with the inlet fairing of the helicopter. The supporting structure and aft most edge of the screen The bypass location chosen in the next iteration was at mesh were embedded in the moulded collar during the the top aft area between the front cone rear small ring manufacture. These materials enabled the designs to and the collar of the engine inlet screen. The bypass meet the various structural requirements including opening was covered by a piece of screen resulting in static strength, resistance to impact, and resistance to a shape like the open gill of a fish. This piece of vibration induced fatigue. screen, referred to as the "ear", was designed to provide shielding against direct ingress from the front This mention of structural requirements introduces a while introducing minimal frontal aerodynamic drag. very interesting aspect of the design Io. The impact This configuration was subject to extensive ice rig and load case for the front face of the screen, and engine test cell qualifications, as detailed in Section 5. consequently for the supporting structure right back to the support collar, was influenced by competing requirements. In order to reduce the impact load 4.4 Final Design Configuration case, the structure needed to be flexible to absorb the energy of impact. However, the structure also had to The final design configuration also utilized an aft be sufficiently strong to transmit the load developed bypass, hut with the opening situated at the outboard without failure. The design of the structure with side of the screen. The prototype of this configuration sufficient strength and flexibility to reduce the impact was subject to an extended flight test program, load resulted in a narrow window for structural detailed below, before fitment to the CH113 fleet. design. The loading on the front face was derived by Minor refinements were accomplished based upon the an iterative assumed load/deflection/resnltant load results of testing and operational experience with the calculation. This converged to the final impact load prototype. This "fine tuning" of the screens was a given by the rigid object strike to the front face. critical phase in the development. Sites of localized 6-8

ice accretion were eliminated by contouring air flows, being imposed by the progressive increase in inlet area some elements were strengthened, a localized heating restriction on the prototype screen design. In problem was remedied, and a removable bypass screen addition, it was desired to determine whether use of insert was developed to discourage bird nesting. the new inlet configuration would produce unacceptable inlet conditions which could affect the 5.0 QUALIFICATION TESTING condition of the engine.

The importance of the testing aspect of the To fulfill the objective of the engine test cell trials, the development of the ice tolerant inlet screen has been followine test points were defined as part of the final stressed through out this paper. The team for the various qualification tests of the final configuration was comprised of representatives from the Canadian Perform two initial standard performance Forces National Defence Headquarters (NDHQ), the runs to establish base line data and verify Aerospace Engineering Test Establishment (AETE), repeatability of data; the National Research Council (NRC) and Boeing. The availability of the unique NRC ice test rig and the Create various air intake flow restrictions technical support of the ice rig staff was crucial to the to simulate progressive screen blockage development and certification. The use of the due to icing; Standard Aero engine test cell and the support from Standard Aero made the engine testing possible. Test with the fairing without particle separator and with the existing CH113A fairing for performance comparison;

5.1 Tufting Test Conduct standard engine performance runs between main configuration changes A tufting flight test was conducted by AETE in to verify base line data repeatability; Ottawa to provide suitable airflow visualization throughout the CH113A operating speed range for Perform a one hour endurance run with confirmation of the optimum inlet screen angle. The "optimum" air inlet blockage. The test comprised two sorties for a total of 8.4 flying optimally blocked screen configuration hours with a test helicopter and a photo chase aircraft. involved 100%blockage of the top and Six inch long tufts were affixed to the test aircraft's front of the screen and 75% blockage of skin spaced every 4 inches for 7 ft along the top of the the bottom of the screen with the bypass aircraft in front of both engine inlets and along the aft area open. This configuration provided pylon within 3 feet of the engine inlets. In addition, a an element of swirl in the inlet flow hoop positioned 32 inches in front of the engine inlets conditions creating the worse inlet allowed the attachment of three 12-inch long tufts condition. floating freely in the airstream for the purpose of indicating the direction of the airflow entering the engines. Perfoim '"slam" accelerations as well as "load chops" and "load bursts" with the screen totally blocked to demonstrate the The behaviour of the tufts was filmed from the photo engine's ability to neither stall nor surge chase helicopter. The film of these flights was with the screens completely iced. difficult to analyze and showed rapid oscillation of the tufts, especially on the long tuft located to give the Monitor engine vibration levels angle of airflow at the engine inlet. The estimated throughout the test as an indication of average air flow angles were observed to be inlet flow distortion caused by the approximately 10 degrees for the left hand screen, and various degrees of screen blockage. 15 degrees for the right hand screen. These figures, although subject to wide scatter in the test results (i.e. In summary, the engine cell performance tests fl-8 degrees), substantiated the calculated screen indicated a two percent (2%)power loss with front orientations. face completely blocked and 75% of sidewalls blocked. Thirtyseven percent (37%) power loss with both front face and sidewalls completely blocked - 5.2 Engine Test Cell Qualifications engine air entering solely through bypass opening. In the unlikely situation where such extreme inlet screen The ice tolerant inlet screens incorporating the aft top blockage might be experienced for both engines, there bypass feature were tested in December 1986 at the would be sufficient power remaining (Le. 1700 HP) Standard Aero engine test cell with a production for cruise. During the tests the engine did not stall or (T58-SF) engine. The main objective for testing the surge during all '"slam accelerations", "load chops", prototype screen and new fairing configuration in a and "load bursts". test cell environment was to determine whether or not any detrimental effects on engine performance were 6-9

The results of this testing verified that the installed nozzles below the mid-mast position and the helicopter engine performance matched that of the original inlet positioned approximately 100 feet away from and configuration. The inlet distortion was monitored by facing the array of spray nozzles. For in ground recording the vibration of the first set of compressor effect (IGE) hover the nozzle array was raised blades using the zoom feature of a spectrum analyzer, completely to the top of the mast, however, for the and no adverse effects were noted in testing with the tandem rotor helicopter a wind velocity of at least new induction system. The conclusions drawn at the 15.5 to 24.9 miles per hour was required to obtain a end of this test phase were that the ice tolerant inlet good cloud coverage. screens and the new H46 inlet system performed satisfactorily. Even with all of the screen blocked and The conclusion from this testing was that the prototype the bypass partially blocked, engine performance was inlet screen allowed sufficient airflow for normal acceptable. The design was subsequently approved for operation of the engines for up to 30 minutes at flight fight testing by Bceing and NDHQ technical idle under severe simulated icing conditions. There authorities. was evidence of ice on the front of the struts which indicates that some of the air entering the front face Finally, based on the analysis of the engine test cell exits through the bypass which is a very desirable trials, the proposed inlet screen design was approved, feature, however, this was not confirmed. It was with respect to the propulsion system, for use during concluded that the bypass must be moved more to the actual icing trials. It was recommended to proceed side and that all struts in the bypass area must be with flight trials. relocated. It was recommended that further testing be conducted to investigate the capability of a modified CH113/113A helicopter to operate in all types of snow 5.3 Simulated Icing Trials conditions and in visible moisture in the range of 32°F to 39°F. The Icing Spray Rig of the Low Temperature Laboratory of the Division of Mechanical The recommended modifications were embodied on Engineering, NRC, was employed for the simulation the subsequent prototype, and the qualifications, with of icing environments. The rig, located in Ottawa, these revisions, resumed in genuine meteorological consists of an array of 161 steam atomizing water conditions versus the simulated environments. nozzles producing an icing cloud 16.4 feet deep and 75.5 feet wide. The array can be rotated about a single supporting mast 75.5 feet high to take 5.4 Performance fight test. advantage of wind direction. Liquid water content (LWC) can be varied over the range of 2.0 x IO-' to Performance tests were conducted by AETE and 424 1.5 x IO4 slugs per cubic foot (slug/@) and droplet Squadron personnel March 1987. The tests were to sizes ranging from 1.2 to 2.4 thousandths of an inch determine and quantify any significant degradation in depending on temperature and wind velocity. engine performance with the prototype bypass ice The second stage of testing in the NRC ice rig with tolerant inlet screen and modified H-46E air induction the ice tolerant inlet screen and top aft bypass was the system installed. The tests comprised seven sorties for most crucial part of the test program. Basically, the a total of 2.2 hours of ground testing and 7.3 hours of performance objective was for 30 minutes of operation flight testing. in severe icing conditions. The thrust of this testing was ice accretion, and no rigorous measure of any In summary, the helicopter's performance with the engine performance degradation was conducted at this protective inlet screens installed was acceptable. test phase.

AETJ3 was tasked to evaluate the ice tolerant inlet screen and develop the icing trial methodology. This 5.5 RainlSnow Flight Trials. methodology was based on knowledge from previous ice testing and a logical choice of cloud conditions. Results of the icing trials were sufficiently promising The FAR 25 stratiform cloud case was chosen rather to warrant further investigation of the capabilities of than the cumuliform because this case is related to a the prototype screens 'I, AETE was tasked to develop larger horizontal extent (10.8 miles) of cloud giving the test program and perform the flight testing. Tests more accretion time. Similarly the choice of droplet were conducted to investigate different types of size of 1.2 thousandths of an inch, which was accretion on the test aircraft and on the prototype ice suggested by NRC personnel to represent a good tolerant inlet screens in the following three average value for FAR testing, was considered meteorological conditions It 13: prudent. These decisions enabled the test method to become simply a matter of controlling the accretion of a) cold rain in the range of 32°F to 39°F ice as a function of the LWC and exposure time. The OAT: maximum LWC available from the rig, corresponding b) wet snow in the range of 26.6"F to to the selected test parameters, was chosen. The 35.5"F OAT; and available range of LWC was from 3.9 x IO7 to 1.4 x c. dry snow in the range of 15.8"F to IO4 slug/ft-'. With the helicopter on the ground the 28.4"F. best icing cloud coverage was obtained at winds of approximately 10 miles per hour, with the spray 6-10

Ground idle, (IGE) hover and low speed fonvard flight tests were conducted at CFB Gander in cold FEATURE TESTING rain, wet snow, and dry snow conditions. Cold rain Front Cone Proved by extensive tests were conducted only at flight idle and no testing. accretion of ice was observed. The worst cases in wet Bypass Area Final deaign same are; and dry snow were experienced during the 30 minute as screen tested in IGE hover flights. Unlike the original CH113/113A engine call. inlet screens, the prototype CH113/113A ice tolerant Frontal Drag h Performance flight inlet screen, now equipped with the outboard rear Engine Performance test. bypass, was capable of operating during these trials in Engine Inlet Tested in engine cell. visible moisture below 39°F OAT. >istortion CH113A NO anomalies reported during ground and flight testing. The results of the flight tests with the side bypass rngine Inlet No anomalies reported were summarized as follows: Fleet fitment of the side lietortion CH113 during ground and bypass ice tolerant inlet screens was recommended flight testing. with the following refinements incorporated in production versions: Sufficient testing has been completed at the NRC ice The fibreglass attachment collar was to rig, the engine tesl cell and during free flight in icing be extended in the area beneath the left and snow conditions to establish both the airworthiness hand ice tolerant inlet screen assembly to and effectivity of the ice/FOD deflector installation on cover the exposed heated surfaces of the the CHI 1311 13A. engine air inlet fairing assembly in this area. 6.0 OPERATIONAL EXPERIENCE

All surfaces of the fibreglass attachment The CHI13 fleet of 6 aircraft has been flying with the collar extending inside the ice tolerant ice tolerant inlet screens for over a year with no inlet screen assembly were to be trimmed engine inlet icing or ice FOD related problems and all acute comers were to be reported. It has been difficult to ascertain the extent smoothed out to reduce the collection of of any ice development on the new screens in normal snow or ice. use. The engine inlets are not visible to aircrew in flight, and any acclretion would be expected to melt An additional 1 inch of the lower during post landing op” and shutdown. The quadrant of the mesh screen was to be positive performance of the new ice tolerant inlet embedded in the fibreglass attachment screens has contributed to improved pilot confidence collar to delay ice accretion inside the in the aircraft capability, particularly in adverse mesh screen. weather conditions,

These modifications to the screen assemblies further The first CH113A helicopter modified has also been reduced or eliminated their susceptibility to the flying for a year with the screens installed with no formation of ice inside the screen assembly. The two reported operational problems. sets of CH113 prototype ice tolerant inlet screens were flown in normal operations for the duration of the One set of screens has also been supplied to Kawasaki icing season. The screen performance was acceptable. Heavy Industries for evaluation by the Japanese Self No engine performance degradation was reported. Defence Forces. To date, there has been no feedback Therefore, the commitment was made to proceed with regarding any operational performance results. the CH113 fleet fitment of production inlet screens. 7.0 CONCLUSIONS 5.6 Qualification Summary An ice tolerant engine inlet screen has been developed The following chart summarizes the ice tolerant inlet to provide protection against foreign object damage screen project with respect to the method used to and resistance to inlet congestion. The design features achieve non-turbulent engine inlet airflow, minimum a flat mesh face, an inverted conical side and a degradation in engine performance and minimum shielded bypass opening. The findings of a aerodynamic drag. comprehensive test :program proved that the screens would provide a miinimum of thirty minutes of operation in severe icing conditions, and satisfactory performance in wet snow or dry. snow.

These positive results have led to an improved tolerance to unexpected and unavoidable icing conditions for the engines of the CH113/113A Search and Rescue Helicopter fleet, and improved aircrew confidence in the aircraft’s reliability in inclement weather. 6-11

During engine cell tests for the ice tolerant inlet 11. AETE PROJECT REPORT 87/30, screen, a unique method of monitoring any influence "CH113/113A ENGINE INLET ICE FOD of detrimental inlet flow distortion was derived. This SCREEN FOLLOW-ON TESTING", 9 AUG technique employed a spectrum analyzer to record the 1988. vibrational characteristics at the blade passing frequency for the compressor first stage. The result 12. AGARD-AR-223, Advisory Report,Rotorcraft was a relatively simple and real-time verification of Icing- Progress and Potentia1,Sept 1986. any undesirable compressor blade excitation when operating in the presence of the various inlet 13. FAA Advisory Circular No. 29-02 configurations. Change,Certification for Transport Category of Rotorcraft, May 1985.

8.0 ACKNOWLEDGEMENT Discussion We would like to thank The National Research Council, Department of Mechanical Engineering for 1. H. Saravanamuttoo, Carleton University sponsorship in publishing this work. In addition, we Was this system designed to allow flight in known icing acknowledge the support of the individuals and conditions or to permit continued flight if unexpected icing organizations contributing to the design and tests was encountered? conducted, specifically the Canadian Forces Project Officers, the Aerospace Engineering Test Author: Establishment, the staff of the National Research The system was designed and tested to permit continued Council Ice Rig facility, and the staff of the Standard flight if icing is encountered in flight and no attempt was Aero Engine Test Cell. made to certify the helicopter for deliberate flight into icing. Essentially this is a get out of trouble capability. There may be a next stagein the project for continued icing certification 9.0 REFERENCES together with the re-introduction of blade de-icing.

1. AETE PROJECT REPORT 85/52, 2. W. Grabe, NRC Ottawa Anti-icing screens have been around for a long time, e.g. "CHI 13/113A Helicopter Engine Inlet Ice FOD Sikorski gave a paper on it at a NAPTC conference some 20 Screens", 17 NOV 1987. years ago. What is different about this screen? How is its design special? 2. AETE PROJECT REPORT 79/55,Leigh MK 12A Ice Detector Trials on CH113A Author: Helicopter, Cold Lake, Alberta, 22 Dec 1980. Yes, there are other good screens in service: the sister tandem rotor helicopter the Chinook has externally 3. Vincent J. Shaefer, John A. Day. A Field mounted engines and conical screens with a removable Guide to the Atmosphere 1981. bypass screen which is effective in icing and is a good 'Tad" screen. The design in this paper is special only because of its 4. FAR 25 Airworthiness Standards, Appendix geometrical shape which was developed to be ice-tolerant C,1974. and the concept originated from studies 20 years ago. The side bypass and the cutback on the cone sides produce the 5. X. De la Servette and P. Cahrit. Helicopter Air unique multi-mode operation for cruize and hover. Intake Protection Systems, AGARD Lecture Series 148 1986.

6. J.Ballard, Impact of IPS and IRS configurations on engine installation design,AGARD Lecture Series 148 1986.

7. CFTO C-12-113-000/MB-000 Aircraft Operating Instructions CH113/113A. 1987-08-31.

8. E.L.HOUGHTON AND A.G. BROCK,Aerodynamics for Engineering Students, 1960.

9. BOEING ARNPRIOR ENGINEERING REPORT DE-89-091 Rev. B, Aerodynamic Report 27 Jnn 1990.

10 BOEING ARNPRIOR ENGINEERING REPORT DE-89.025, "STRESS REPORT", 27 FEB 1989.

7- 1

COLD STARTING SMALL GAS TURBINES -AN OVERVIEW (Demarrage Temps Froid des Petites Turbines a Gaz)

by C.Rndgers Chief Concept Design Sundstrand Power Systems 4400 Ruffin Road San Diego, California 92123 United States

ABSTRACT The requirements to operate aircraft gas turbines over a large range of environmental conditions prove particularly demanding to the systems designer, especially when rapid starting of a cold engine is stipulated at sub-zero ambient temperatures. As a consequence the occurrence of cold climatic extremes are discussed and a trend is observed toward designing aircraft for specific areas and deployment, rather than worldwide usage. Cold engine cranking torque characteristics are basically controlled by the lubricant viscous drag in the mechanical drive train and accessories. To further complicate matters, this viscous drag is dependent upon the magnitude of the applied start torque. Either actual viscous drag test data or realistic methods of predicting engine cold cranking torque levels with high lubricant viscosity may be required before definitive starter specifications can be issued. Experience with start systems for small gas turbine Auxiliary Power Units (APU's) has shown that the total weight required for successful rapid starting at -54'F can approach the weight of the APU powerhead itself. As a consequence, most cold start requirements are relaxed to -4OOC or higher. Briefly addressed, but of critical importance, is the role of the combustion designer who provides a key input in designing and developing a combustor capable of rapid consistent light-off over a wide operating range at the earliest possible speed, thereby permitting the transition from negative (cranking) to positive engine assist during start. Methods of reducing APU viscous drag and start energy requirements that deserve future study are the all electric gearbox-less APU, and the possibility of a self-start pulse combustor concept.

ABREGE Les demandes d'utilisation des turbines d'avion daus une gamme etendue de conditions d'environment s'avkrent de plus en plus &gentes pour le concepteur des systltmes; ceci est specialement le cas du demarrage rapide d'un moteur froid par temperatures trks basses. La probabilite de rencontrer des conditions tris froides doit stre discutie; on observe une tendance a concevoir des avions pour des regions spicifiques et des conditions particulikres, plut6t que pour une utilisation universelle. Les couples characthistiques d'entrainement d'un moteur froid dependent issentiellement de la trainee visqueuse de I'huile dans les boites d'entrainement et les boites accesssoires. La question se complique un peu du fait que la trainee visqueuse depend de la grandeur du couple de demarrage applique. Avant d'icrire les specifications difinitives du demarreur, il faut soit des resultats d'essai mesurant la trainie visqueuse rkelle, soit des mithodes realistes de prevision des couples d'entrainement du moteur froid. L'experiences a montri que pour le dimarrage de petits.APV, la masse totale necessaire pour reussir un demarrage a -54°C pouvait approcher la masse du generateur de puissance lui-mhe. En consiquence, la plupart des exigences de demarrage temps froid se limitent a des tempiratures supirieures ou egales a -40°C. Nous mentionnerons briivement mais en soulignant son importance le role du concepteur du systeme de combustion; celui-ci apporte une contribution decisive en concevant et diveloppant une chambre de combustion capable d'allumage rapides dans un domaine d'utilisation tris etendu, a des vitesses de rotation les plus basses possible; ceci permet de passer rapidement d'un couple rbistif a un couple moteur pendant le dimarrage. Parmi celles qui mbitent des etudes plus approfondies, les solutions permettant de reduire la trainee visqueuse dun APU et les besoins en inergie elictrique sont I'APU tout-elktrique sans hoite d'entrainement et le concept possible d'une turbine a volume quasi-constant. NOMENCLATURE

Area Reference Area AMAD Aircraft Mounted Accessory Drive ATS Air Turbine Starter ATSiM Air Turbine Starter Motor APU Auxiliary Power Unit D Diameter ECS Environmental Control System EGT Exhaust Gas Temperature EPU Emergency Power Unit F Axial Load JFS Jet Fuel Starter JP Jet Petroleum HP Horsepower IP Polar Moment of Inertia KPa Kilo Pascal M.E. Main Engine \J Rotational Speed PASS Pneumatic Air Start System P Pressure Q Oil Flow SLS Sea Level Standard TIT Turbine Inlet Temperature T Torque VSCF Variable Speed Constant Frequency VLSI Very Large Scale Integration U Tip Speed Vref Combustor Reference Velocity W AIRFLOW vo Isentropic Spouting Velocity I* Viscosity P Density

SUBSCRIPTS b Burner 7-3

1.0 Introduction 2.0 Cold Weather Extremes Cold weather and cold environmental starting The design of cold weather start systems for of aircraft main propulsion engines represents a aircraft gas turbines is critically dependent upon significant design burden for self-sufficient fighter the design specification, specifically, the low tem- aircraft operation from unimproved forward bases. perature extreme soak condition. Low temperature Although military and commercial aircraft typically extremes for worldwide usage (excluding the air, operate from permanent bases with both personnel land and ice shelf areas south of 60 degrees) are and equipment preheating capabilities, emergency shown on Figure 1, and set forth in MIL- in-flight power outage may require start up of the STD-210B and in the revised updated version aircraft secondary power system after extensive MIL-STD-21OC. Both standards do not give defi- soaking at -57-C conditions. nite limits. but thev offer design criteria and starting points' for engineering anaryses to determine The many elements of the start system and the design criteria. synergistic effect of cold weather are often difficult to accurately predict. A conservative cold start The detailed requirement values used in MIL- analysis may present an unpalatable system weight STD-210B represent risk values. It would often penalty. A more palatable analysis can mature to be cost prohibitive and/or technologically impossi- prove disastrous when into the final qualification ble to design military equipment to operate any- stage. In retrospect, lessons learned would deem where in the world under the most extreme envi- improved definition of the synergistic effects of fuel ronmental stresses for all but a certain small per- and lubricant viscosity effects, thermal soaking, and cent of the time. The design criteria for operations transient performance during preliminary design. It in MIL-STD-210B are based where possible, on is the intent of this presentation to briefly expose hourly data from the most extreme month and most of the known pitfalls in cold weather starting, area in the world. From hourly data it is possible such that the designer is provoked and appreciative to determine the total number of hours a given of the task involved. value of a climatic element is equaled or surpassed. If a value of a climatic extreme occurs (or is sur- In preparation for addressing the cold weather passed) in about 7 hourly observations in the 744 problem, a review of climatic weather extremes and hour observations of a 31-day month, then this previous cold start technology is presented. value occurs roughly 1 percent of the time. The risk values pertaining to low temperature are as follows:

Low Temperature Risk Values

Risk Value 1% 10% 12% 50% ? I Temperature ("C) -61 -54 -5 1 -46 -40

Figure 1. Climatic Extremes 7-4

The one percent extreme is used as the design F-15 was most likely the first aircraft to be quali- criteria for all climatic elements except low tem- fied at -4OOC. Many of the justifications used to perature and rainfall rate. The 20 percent extreme relax the cold temperature requirement for the is used for low temperature. F-15 pertain to innst or all of the other aircraft as NATO has adopted a Standard Agreement well. which covers climatic environmental conditions af- The feasibility of starting an aircraft at -51°C fecting the design of material for use by NATO was considered poor using the technology available forces operating in a ground role. This document, and the expense required to obtain -51°C capabil- STANAG No. 2831, is similar to MIL-STD-210C. ity was considered unrealistic. Batteries were con- Five categories pertain to cold weather and are sidered to be unacceptable for use below zero de- listed in the following table. Where a range of grees. Hydraulic start systems were also limited in temperature is given, it reflects the high and low cold weather due to poor viscosity characteristics temperatures during a 24-hour period in which the and while increased volume would provide the nec- severe temperature occurs. These values are for essary energy, the weight penalty was unacceptable. operational conditions corresponding to the values Lack of full human functionality at -4OOC and given in MIL-STD-21OC which serve as starting lower was used as justification, and the standard points for engineering analyses. itself was often deemed to be unrealistic. The five locations of these climatic categories The results of the Reference 1 survey are sum- are: marized as follows: Mild Cold - Areas which experience mildly 1. MIL-STI)-210B is an interpretable docu- low temperatures such as the coastal areas ment which offers guidelines that allow sig- of Western Europe under prevailing mari- nificant deviance from the intended design time influence, Southeast Australia and the criteria. Lowlands of Xew Zealand. 2. MIL-STD-210C will make no attempt at Intermediate Cold - Areas which experi- setting any design criteria. ence moderately low temperatures such as 3. The U.S. Air Force sets cold weather re- Central Europe including South Scandina- quiremen1.s on a system-by-system basis. via, Japan and South Eastern Canada. 4. Start systems appear to be a limiting factor Cold - Colder areas which include North- with cost and spacelweight penalties being ern Norway, Prairie Provinces of Canada, main drivers. Start system technology has Tibet and parts of Siberia but excludes not kept pace with other areas. those areas detailed in the two colder categories. 5. The designing trend leans toward aircraft for use in specific areas rather than for use Severe Cold - The coldest areas of the worldwide. North American Continent. Extreme Cold - The coidest areas of It was emphasized that MIL-STD-210 was an Greenland and Siberia. attempt to apply the lessons learned in World War I1 to future aircraft designs. Future systems must The values listed in STANAG 2831 represent not be designed for the conditions they experience the air temperature, which, on average, was atin peacetime, but for the conditions that they may tained or exceeded for all but approximately lper- encounter in the ;advent of war. cent of a month during the coldest month of the year. 3.0 Review Of Previous Start Technology A detailed evaluation of MIL-STD-210B, Extensive design data and design techniques for -210C and STANAG 2831, plus cold weather aircraft gas turbine engine start systems have been starting specification requirements for USAF air- published by the Society of Automotive Engineers craft was made in Reference 1. (SAE) Starter Committee AE6, most notable of The A-10, F-I5 and F-16 aircraft are all which are listed in References 2 through 8. The qualified to -40°C. The minimum ground opera- SAE Starter committee was first formed in 1962, tion temperature specifications for the C-17 and and meets biannually to formulate standards for the B-IB are -40°C and -34°C respectively. The aircraft engine starting systems.

Cold Temperature Extremes for NATO Operational ConditionsI Category Lowest Recorded Temperature Temperature

Mild Cold “C -6 to -19 Intermediate Cold “C -21 to -31

Cold OC -37 LO -46 Severe Cold “C Constani -51 ---

Extreme Cold “C Conslant -57 -7 1 1-5

References 9 and 10, Starting Systems Tech- 2. Pneumatic Link - With small integral bleed nology, SAE special publications SP 598 1984, and or load compressor auxiliary power units SP 678 1986 movide latest design data for most (Figure S), either mounted on-board, or aircraft start‘sktem comvonent< including gas tur- from external ground carts. bine auxiliary power uniis, emergency powe? units 3. Electrical - From aircraft batteries, on- (EPUs), air turbine starters (ATS), and air turbine board APU, or ground power supplies. starter motors (ATSM), and hydraulic and electric Starters 4. Hydraulic - From on-board APU or ground hydraulic chart. The author and his affiliation (References 11 - 16) have devoted efforts toward the optimization of 5. Windmill - With possible assist from APU. the start systems for both prime propulsion engines 6. Cartridge - Solid propellant breech firing and small auxiliary power units. Previous Start pa- into. the high temperature air turbine starter pers published by the author are found in Refer- or direct rotor impingement. ences 11 - 16, and were motivated by a primary goal to improve start system reliability, particularly for self-sufficient jet fuel starters (JFS), which may provide the only means of aircraft dispatch. 2.0 I Constant Speed 4.0 Starting Methods Constant Turbine Inlet TemD A frequently used method of starting aircraft prime propulsion gas turbine engines is by means of a small auxiliary power unit providing com- pressed (bleed) air to an accessory drive gearbox- mounted air turbine starter. This APU may be carried on-board or mounted externally on ground support equipment. The majority of these APU’s deliver a bleed pressure ratio of approximately 4.0 at standard day conditions, as constrained by oil auto-ignition temperature limits no higher than 25OOC for mostly commercial aircraft applications. The general effects of ambient temperature and altitude on small gas turbine performance at rated conditions are shown in Figure 2. The basic power lapse rate towards lower power on hot days is dic- APU Air Inlet Temp ‘C tated by the decrease of cycle efficiency with turbine inlet temperature (T.I.T.) to ambient temperature ratio, air density with hot days decreasing output power, and cold days increasing output Figure 2. Typical Power Lapse Rate Single power. The temperature and altitude lapse rates Shaft Radial Gas Turbine shown are typical of single-shaft APUs with centrifugal compressors operating at rated TIT and constant rotational speed. Slightly different hot I day power lapse rates may be exhibited with variable speed operation. Note that the power lapse rate change with ambient temperature may be approximated by: t tridge

It is observed’in Figure 2 that rated output power decreases about 30 percent from SL 15OC LO SL 52OC conditions. Normally APU’s are sized to meet full aircraft secondary power requirements at critical hot day conditions, and are therefore capable of delivering access power at cooler ambient temperatures, as dependent upon load change with ambient temperature. Usually, 90 percent of APU operating time is spent at ambient temperatures below 35°C and therfore exposure to maxi- A- Ground Cart mum TIT’S may be limited. Six common methods of starting aircraft prime propulsion gas turbine engines shown in Figure 3 are: I. Mechanical Link - With small gas turbine jet fuel starters of the two-shaft power turbine, or single-shaft with torque converter, configuration, (Figure 4). Figure 3. Engine Start Methods LF300m2-1 7-6

LP900032-4

Figure 4. Jet Fuel Starters 7-7

LP900032-5

Figure 5. Pneumatic APUs 7-8

Since each of these start methods may also be control actuation, the energy source is convenient, applied in turn to start the on-board APU, caution besides which accumulators can serve dual roles in is advised during start system integration discussions providing both braking and back-up start assist. to clarify which starter is being addressed, the main Back-up starting is also possible using a hydraulic engine, or the APU ground cart, or hand pump for medium place helicopters. A summary of the various start systems options is listed in Table 1 and a more detailed compari- The three primary %art systems for small gas son can be found in Reference 17. turbines (excluding expendable turbojets which use cartridges) are: Start system survey data for se1ected'U.S. military aircraft and helicopter is shown in Table 2 Hydraulic which identifies Lhe aircraft, main engine, on- board APU or JFS, and APLIJFS start mode. Pneumatic Examination of Table 2 data indicates the domi- Electric nance of hydraulic start option. Since hydraulics are usually routed throughout the aircraft for flight

Table 1. Summary Comparison of Secondary Power System Design Approaches

SYSTEM TYPE ADVANTAGES DISADVANTAGES AND USAGE - Pneumhtic Link. APU can be remotely located for Low system efficiency, resulting Large Commercial Aircraf flexibility in installation location in relatively large APU power ra and environment ing required EFA Similar ECS airflow requirements Pressure ratio limited by bleed a ATF temperature auto-ignition hazard B1 System is compatible for ground B2 cart backup main engine starting

Mechanical Link. High system efficiency results in Requires close APU coupling to relatively low APU power rating main engines F-15 required Significant penalty for backup F-16 Power turbine or torque con- main engine start capability verter provides high stall torque s Mechanical coupling equipment and controls tends to be comple: Incompatible with ECS

Electric Link. APU can be remotely located for Relatively high-power starter1 flexibility in installation location generator technology not and environment demonstrated Offers increased system integra- No backup main engine starting tion capability and simplification from standard ground carts with all electric aircraft Incompatible with ECS

Hydraulic Link. APU can be remotely located for Power transfer hydraulic system Most Helicopters flexibility in installation location tends to be complex and environment Backup starting from ground carts is difficult since aircraft hydraulic system must be penetrated Incompatible with ECS

Windmill Possible with clean inlet and Limited operating envelope exhaust installaiion Slow acceleration

Cartridge IHigh power density Poor start reliability under cold conditions Excessive smoke logistics Fail to fire breech danger Table 2. Starting Data - U. S. Military Aircraft

AIRCRAFT F-15 F-16 F-18 B-1 KC-135 C-141 Black Hawk AH-64 vlain Engine Model F-100 F-100 GE404 F-100 CFM56 TF33 T700 T700 Ip lb.fi.sec2 3.2 3.2 2.65 5.92 5.92 8.9 0.11 0.11

:ore N krpm 13.5 13.5 14.5 14.5 14.5 9.0 43 43 1~x2x10-5 583 583 557 1244 1244 720 203 203

GTCP GTCP GTCP GTCP dodel JFS190 T62JFS 36-200 165-9 T40LC 85-106 T40 36-150 p Ib.ft.sec2 ,0054 ,0032 ,0071 022 ,0045 ,026 ,0031 ,0068 i krpm 65 61.6 62 38.0 64.3 41 61.6 62 pN2 X10-8 22.8 12.1 27.3 31.8 15.6 75.6 11.8 26.1 ipprox. Weight Ib 100 84 160 230 190 320 96 100

I APU START SYSTEM1

Type Hydraulic Hydraulic Hydraulic Hydraulic Hydraulic Hydraulic Hydraulic Hydraulic Weight Ib 75 47 51 110 63 119 56 46

-ow Temp OC -40°C -40°C -40°C -29°C -4O'C 0 -40°C -32T

Starter Displacement 0.3 0.42 0.37 0.60 0.65 0.37 0.68 0.365 inVrev

\ccnmulator 2x2 I5 2x200 290 1200 500 2x400 200 299 Volume in3 ipprox. System 75 50 51 110 61 119 62 40 Weight Ib 7-10

4.1 APU Hydraulic Star1 4.3 APU Electric Start Until recently, most quick start military gas tur- Increased avionics, electrical systems, and bine APU installations used hydraulic start systems, power levels has created renewed interest in the as oppbsed to electric systems. The reasons full electrical aircraft. Advanced VSCF type gen- stemmed from compatibility with the aircraft hy- erators, capable of operation as starters are being draulic system, and cold weather starting. Electri- studied for both large prime propulsion and APU cal systems usually require battery electrolyte heat- gas turbines. Switched reluctance starter-genera- ing at temperatures below -28"C, but can be tors have also been identified as a potential future started at lower temperatures with battery shorting, technology approach. These starter-generators if time is not critical. Hydraulic start systems are generally operate at 150 percent of rated power mainly in use for sub-Arctic environments, but during a start to meet the engine start torque re- nevertheless, are also prone to increasing fluid vis- quirements, and may also require a dual ratio gear- cosity pressure drops effects, and thus reduced box. None of them have yet demonstrated satis- starter output torque. Hydraulic start system sizing factory starts in an actual aircraft. Battery charge for -54°C conditions, in particular, require large considerations under cold weather conditions as hvdraulic accumulator volumes and therefore De- typified in Figure 7 is a fundamental constraint. nalize aircraft gross take-off weight. As with bat- Reference 19 describes a batterv shortine techniaue teries, APU preheating may be necessary. used to make sui:cessful -54'C'start withY a smali gas turbine. The engine ECU incorporated a The effect of soaked temperature conditions on "WINTER" mode which initiated a pre-program in the performance of a typical hydraulic starter using the microprocessor to periodically connect and dis- MIL-H-83232 fluid is shown in Figure 6. For connect the batterv from the starter windines in temperatures less than -40°C the use of MIL-H-5606 fluid is recommended. from its own cell resistance.

I I I 5.0 Engine Starting Characteristics It is difficult to estimate with any reasonable accuracy the star1.ing characteristics of small gas turbines because of the synergistic nature of the problem. Engine starting characteristics are influenced (among other intangibles) by: Temperature and altitude e Initial engine design selection Engine performance Engine fuel control scheduling iMechanical features and associated viscous 20 1 1 I ! 1 I drag 0 20 40 60 80 100 % Max Speed Acceleration rate

LpIDmll.8 Starter selection and torque characteristics

Figure 6. Typical Hydraulic Starter Performance

4.2 APU Pneumatic Start High pressure pneumatic air start systems (PASS) (see Reference 17 and 18) are being pro- *ON70 posed and used to circumvent the low temperature problems of the electric and hydraulic approaches. Stored air in high pressure bottles is expanded across an air motor attached to the APU gearbox

or~ directlv ~~~~~~~ imoineed on the turbine rotor. Current ,~ ~ system p;essure levels-~~ are 1,000 psig, with higher levels, 4,000 psig, being considered to reduce stor- ape volume. Further air storage volume reduction ispossible with develppment 07 small high pressure 1 fuel rich air/JP combustors~ ~~~ caoable of heatine the cold expanding air (sometime; at -73-C or IGwer) j38 to 900°C. The cryogenic temperatures experi- 0 50 100 150 200 250 300 enced with PASS components and prolonged high Discharge Current pressure recharge compressor times^ offer challenges to match the high start reliability standards of hy- LPs00011.1 draulic starters.

Figure 7. Typical Battery Characteristics 7-11

Due to the complex interrelation of these fac- During the initial start phase (below approxi- tors, engine starting characteristics are realistically mately 20 percent design speed), the aerodynamic only obtained by actual test under controlled con- and viscous drags at normal ambient temperatures ditions with a specific start system. Although ex- may be relatively small in relation to the basic trapolation of these characteristics to other operat- starter applied torque. In such cases, the starter ing conditions and start systems without actual test applied torque is approximately equal to the basic is not recommended, it is often necessary. flywheel acceleration torque as determined from the rotating assembly inertia and acceleration. The three major torques during engine acceleration shown in Figure 8 are: Under these conditions, examination of several small gas turbines with different start systems has . Unfired cranking torque shown that the starting difficulty (or amount of Viscous torque starting energy required) can be correlated with the product of the rotating assembly polar moment of Fired torque inertia and the square of the rotational speed, IpN2. Since the start system weight is a function of the stored starting energy, general trends have been established for start system weight as a function of IpN2 as shown in Figure 9. These trends are useful in early engine design to indicate the potential effect of compressor and turbine geometry and design speed upon start system weight for equal start times at normal ambient conditions. Both APU and main engine (core) IpN', data are shown on Figure 9 for a wider correlation base of gas turbine engine types. The use of the core inertia and -Fired Acceleration speed is a fundamental correlating parameter for turbofans, particularly with the trend to higher bypass ratios and three spool configurations.

Figure 8. Idealized Engine Torque Characteristics

K, m2 rpm2 108 109 I I 1000 I

500

100 E .-al g 100 E c.a 50 $

1c 107 5 108 5 109 5 10'0 I, N2 Ib ft sec2 rpm' 1 LF3OOOJZ-9

Figure 9. Start System Weight Correlation 7-12

A review of the hypothetical starting times for a typical main propulsion engine using a pneumatic link start system comprising a small APU, ducting, and air turbine starter was described in Reference 15, where relative sea level start times for the idealized case with only aerodynamic and no viscous drag effects considered showed

Ambient OC Relative Time %

15 100

34 130

-54 63

LPIOO012-'0

The longer main engine start time for the SL Figure 10. Four Quadrant Turbine Operations 130°F is typical of pneumatic link systems, and consequently often sized the APU. The computed reduced start time at -54T is purely hypothetical and contrary to normal sub-zero trends, where Analysis of low speed unfired cranking torques viscous drag becomes dominant and causes relative for small gas turbine APUs indicates that the ap- start time to increase substantially above that of SL proximate magnitude of the net aerodynamic 15OC. These start times reflect the basic power cranking torque conservatively relates to that of the lapse rate of the smaller gas turbine APU with out- torque required to drive the compressor (only). put power increasing significantly at lower ambient For fixed compressor geometry and zero bleed temperatures. Both the main engine and APU, conditions this can be simply expressed as: are, however, impacted adversely by cold weather drag effects. At -54OC ambient conditions, hydraulic start systems weight for small APUs may approach that of the APU power plant itself, forcing the APU Aerodvnamic Crankine Toraue designer to scrutinize all possible sources of reducing rotor inertias. Such scrutiny is common practice Compressor Design HP x 5250 2 in turbocharger design, for example, where fast e accelerations are required for response and smoke abatement. Rotor blade number and disk configu- Design Speed rpm rations are whittled to the minimum is necessary for structural integrity at design or overspeed conditions. The APU engine turbine blade number tradeoff is extremely weight effective, and turbine efficiencies of up to three percentage points may In effect, this presumes the turbine operates be sacrificed for non-continuous duty operation. near runaway conditions producing no expansion torque and requiring minimal input. 5.1 Unfired Cranking Aerodynamic cranking torque, is best deter- Engine performance behavior during the initial mined from actual torque measurements, but may cranking phase, prior to light off, are often pre- be derived from APU speed deceleration traces dicted using steady state compressor, combustor after a cold unfired starter crank. and turbine component characteristics, yet it is well known that the transient and steady-state compo- 5.2 Viscous Drag nent performances may be different. For example, the first start of a gas turbine normally requires a The mechanical features of the encine starter longer time than subsequent starts. drive train and support bearings are possibly the most dominant single factor controlling cranking at Difficulties encountered in computing low speed sub-zero ambient temperatures. Inadequate clear- "unfired" cranking torque are: ances and dissimilar metal contraction rates have Accurate low speed characteristics been known for example to cause interference definition of the compressor, combustor binding preventing starting. High oil viscosity and turbine. throughout the train and bearings increases drag torque by an order of magnitude. Parasitic losses Operation of the turbine near or beyond in the oil and fuel pumps increase also with the "runaway" conditions, as illustrated in change in viscosity. Figure IO. At normal ambient temperatures, the mechanical losses in the drive train, bearings, and accesso- I HP loss/bearing = 0.80 (pQ)’” I ries are on the order of 1 to 4 percent of the design power rating dependent upon train reduction This relationship is shown plotted on Figure 11 ratio, type, number of bearings, and accessories. Additional parasitic losses may stem from engine During preliminary engine design, a mechanical driven equipment such as cooling fans, generators design configuration is selected after many design and hydraulic pumps. Miniature gas turbines (less iterations to best compromise all the design criteria than 20 hp) tend to exhibit higher mechanical and constraints. The creative designer will gener- losses on the order of one-tenth of rated power, ate an efficient design in which mechanical losses since it is not always cost effective to scale engine are minimized to avoid impacting such factors as driven accessories, especially when ultra high rota- cost, performance and oil cooler requirements. tional speeds are employed. Although these mechanical losses at normal ambi- High speed rolling element bearing losses for ent temperatures and design speed may be small small gas turbines can be estimated using the data (2-5 percent) and seemingly negligible relative to of Rekrence 20 approximated to give: the aerodynamic drag, oil viscous shearing at start initiation results in larger mechanical than aerodynamic drag. The T-47 gas turbine APU (Figure 5) gearbox was subjected to extensive cold crank tests during initial development. The effect of oil sump tem- HP loss/bearing = 0.20 (GQ)’.’ perature on its cranking torque at 15 percent speed is shown in Figure 12. (D6) 2 t .0002D ( N) ‘”F The cranking torque ratio shown is that com- posing mechanical drag from the starter drive train via the main reduction gear bearings, accessories, and aerodynamic drag. Total mechanical losses at design speed for this engine were approximately four percent of the total engine output. Light aircraft type oils, MIL-7808 and MIL-23699 were Where the bearing bore, diameter D is given in used but nevertheless cranking torque increased by mm, and the viscosity p is in centistokes. For light a factor of 4.0 at -54’C with higher oil viscosity. axial loads the power loss is dominated by viscous Cranking tests results with the gearbox deprimed shearing, and with typical “DN” values of 2 x lo6, are also included and indicate up to a 20 percent can be further approximated to yield. reduction drag.

Litre/Min 1.o 5.0 10.0

MIL-L-7808 Oil

@F2 UJ UJ 0 4 P = 1.0

0.5 ( 1 0.5 1.o 5.0 10.0 Oil Flow GPM LF300032-11

Figure 11. Bearing Losses 7-11

Applications of external starting torque direct to the high speed shaft itself, with un- "C coupling of all but necessary viscous para- -50 -25 0 25 4 sitic shear sources. Advancement in applied tribology to reduce viscous shear effects in gas turbine accessory and main shaft components. 3 MIL-7808 Deprimed .-0 -- - ze MIL-23699 Primed 0 3 92 I- m .-c 1 c ;1

Oil MIL-L-7808 0 -..m In.c.. Excludes Hydra! Pumps, nerators, Fa1 Oil Sump Temperature "F LC IF$owI1-II

Figure 12. Effect of Oil Sump Temperature on Cranking Torque

Examination of the apparent oil viscosity variation (based on ambient temperature) and the increase in cranking torque indicated a relationship similar to that presented in Reference 21 based upon cold starting of diesel and automotive engines where: - S.L. 15% Rated Power ipJwD,i~,, Cranking Torque

Cranking Torque Reference Figure 13. Typical APU Breakaway Torques

5.3 Fired Acceleration The reference conditions define the cranking torque and apparent oil viscosity at design ambient. The transition from unfired cranking torque The exponent n varies from 0.2 to 0.3. This rela- resisting acceleration, to fired torque assisting accel- tionship may be applied to project mechanical drag eration, occurs upon ignition and combustor light at sub-zero ambient conditions providing: off with fuel burn scheduled by the engine control unit (ECU). The synergestic starting effects culmi- Mechanical drag at the reference condition nate during this transition phase. With instantane- is known to be reasonably accurate by esti- ous thermal and fluid response this transition would mation or direct measurements. hypothetically mal.erialize by a vertical shift from Aerodynamic torque is assumed negligible unfired to fired torque levels at constant speed. In in the range of 0 to 15 percent speed. practice the transition takes a finite speed interval as dictated by the system response. In the event that a reference torque condition Thermal transient response effects become is unavailable, the representative breakaway torques even more significant during fast accelerations. (torque at zero speed) in Figure 13 may be used in Substantial increases have been observed in tran- a first cut start analysis. Higher breakaway torques sient fuel flow demands for small gas turbines at at -54°C have been experienced resulting in stored the author's affiliation, directly relating to heat energy start system weights that can approach the losses to the "cold" castings and flame quenching dry APU powerheacl weight itself. The same APUs effects on combustion efficiency. Heat exchanged when employed as a starter for larger prime pro- gas turbine engines are notorious in this respect as pulsion gas turbines are burdened with the incom- the large mass of the heat exchanger acts as ther- patibility of starter torque output dependent upon mal reservoir absorbing heat during start and dissi- the square of air density (Figure 2), yet main en- pating heat (with possible bearing soakback prob- gine cranking torque is basically dependent upon lems) on shutdown lubricant viscosity. Fortunately, a main engine restart at altitude occurs with the lubricant already The phenomena of increased viscous drag with preheated. Allitude and sub-zero starting condi- higher applied starler torque and faster accelera- Lions, however, will continue to require excessively tions is discussed in Reference 14. The effect is heavy start systems until viscous shear losses can be related to the increase in oil viscosity under high reduced bv either. pressure shear rates. 7-15

Cold starting characteristics of a load comores- The static pressure at the exhaust must never SOT type APU are shown on Figure 14 at S.L.' be greater than the static pressure at the inlet when -4OT conditions, with 0.95 and 0.68 cu. in. hy- a start is attempted since this can result in flame- draulic start motors. Acceleration fuel scheduling back out of the APU inlet. Although high inlet of the EGT topping type was the same for both. ram recovery can increase APU power, it increases Increasing the start stall torque by approximately velocities through the combustor and may interfere 50% (0.95 cu. in. motor) increases the initial en- with APU starts, which is discussed in the next sec- gine resisting torque by a similar percentage. The tion. Starting above the tropopause with low pres- increase of APU resisting torque with acceleration sures and temperatures results in compressor opera- rate comolicates start svstem ootimization and stan- tion at lower Reynolds numbers and higher relative er selecti'on and necesskates experimental determi- Mach numbers which can reduce surge margin. nation of its magnitude. The EGT topping fuel Advanced ECU's are now capable of identifying schedule eventually produces the same APU assist- surge and responding by re-metering the start fuel ing torque characteristics after acceleration past schedule to avoid surge confrontation. starter cut-out. Small gas turbines are particularly sensitive to 6.0 Combustor Design Considerations tip clearance effects since clearance gap to blade height ratios tend to he larger than those of larger The previous effects are relatively small when gas turbines. In most instances, the minimum en- compared with the transient behavior of the com- gine operating clearances are established by tran- bustor. Transient combustor efficiencies as low as sient engine tests to determine limiting rub clear- 20 percent have been experienced for small short ances gaps. Some tuning can be accomplished by reverse flow annular combustors on cold starts fol- matching thermal responses of the rotors and sta- lowing ignition and light-off, as indicated on Figure tionary shrouds and by appropriate material selec- 15. tion. Computation of engine transient performance using steady-state component performance characteristics does entail small errors due to the effect of -Steady State COnditions transient clearance effects on both component effi- -Transient ciency and pressure flow characteristics.

6 SL-4O'C

Percent Design N ,--I

Figure 15. Transient Burner Efficiencies

Of primary importance to increasing engine acceleration assisting torque is early ignition and good fuel atomization, which affects the transition from engine resisting to positive acceleration torque. L"*.l4 characteristic.. A major constraint for reliable early ignition under a wide range of operating conditions is in the design of the fuel injection (atomization) Figure 14. Effect of Higher Starter Torques system. All these transient effects can be simulated with sophisticated analytical models, but the acid test is start demonstration under exposed cold soak 5.4 In-Flight APU Starting conditions. In-flight starting of the APU after cold soak The capability to fill the fuel manifold faster, conditions is dependent upon: combined with higher ignition spark rate, and im- The inlet and exhaust duct pressure proved start nozzle fuel atomization, is required to recovery or loss, provide an earlier transition from unfired cranking to positive engine assist torque. Under Arctic con- Aircraft location of the inlet and exhaust, ditions, high fuel viscosity and poor fuel atomiza- Gearbox breakaway torque after cold soak, tion may cause a delayed light-off, which combined with higher viscous parasitic drag can result Combustor characteristics under ram in an aborted or hung start. conditions. 7-16

Low combustor reference velocities and load- The typical effect of engine speed and flight ings are preferred for reliable ignition (especially at Mach number on unfired combustor reference ve- altitude), fuel air ratio stability limits, and low com locity is shown in Figure 17. Higher Mach num- hustor pattern factors. Combustor reference veloc- bers, or more specifically, engine inletlexit pressure itv is defined as: ratios, increase the reference velocities and eventually may cause blow-out upon ignition attempts.

V,,I = Wbl(A,,r . 0) AnnulusMoule Area Ratio- 10 where W b = Combustor airflow ATef = Combustor reference area P = Combustion gas density

The combustor reference area is typically calculated from the mean diameter and chamber dome height since most gas turbine combustors are of the reverse flow or axial through-flow types. Combustor maximum diameter becomes governed by the necessity to minimize frontal area compatible with that of the compressor and or turbine. The only remaining “free” combustor sizing vari- Figure 17. Typical Effect of Ram Pressure ables are usually dome height and length. Choice Ratio on Vref of length (volume) influences the combustor heat release rate and loading, which are additional important combustor design criteria. Note that under ram conditions, flow is being rammed through the engine at zero speed and may actually decrease slightly as the engine initially breaks away and picks up speed, as dependent upon the turbine flow characteristics at low pressure ratios and speeds. Combustor Loading = Wil I Combustor reference velocities are determined also by the ratio of the combustor reference area to the turbine nozzle area. Thus, for a given combustor area, higher engine pressure ratios with corresponding smaller turbine nozzle areas provide Lower combustor reference velocities permit lower combustor reference velocities. wider fuellair ratio limits during start in the manner shown in Figure 16, and therefore tend to promote As mentioned previously, good fuel atomization improved fuel atomization and blow out limits unis critical for small gas turbines. In this respect der high altitude operation. Figure 18 shows the result of tests on a series of swirl pressure atomizers, of differing sizes appropriate to 200 hp APU, using two different fuels of 4 and 25 centistokes viscositv. Usine 4 centistokes is appropriate to typical JP-4, at an ambient of -Sea Level -40°C. A viscosity of 25 centistokes is appropriate to a heavy diesel oil near its cloud point, and represents a worst cas, appropriate, for example, to Air Force ground carts burning diesel fuels. The two curves define the fuel oressure reauired for 5 m/Sec 10 adequate atomization and any chosen he1 flow, when using a 0.8 Joule ignition system. The scale effects of swirl pressure atomization is clearly shown by the impossibility of providing adequate fuel atomization for fueling rates of 5 and 1 kglhr .01 t+--f&L= or less, for viscosities of 25 and 4 centistokes respectively as, at such fuel flows, the fuel pressure required is infinite. On the other hand, fuel flows I I much above 15 kgihr show little effect of fuel viscosity, and fuel pressure required for adequate fuel atomization is modest at about 35 psi. Thus, a very significant difrerence in flame performance using swirl pressure atomizers, as between large and small gas turbines, exists. This focuses, in the small gas turbine, more strongly than in the large turbine, on those fuel properties affecting fuel at- Figure 16. Typical Small Burner Ignition Limits omization, with emphasis primarily on fuel viscosity and, to a smaller degree, on surface tension. 7-17

7.0 Control System Once combustion is accomplished, the magni- Kglhr tude of engine assisting torque is governed by: Engine performance and temperature and 2001 temperature limitations. Compressor surge margin. Fuel control scheduling Parasitic applied torques. 2 Transient thermal mass affects. Y Larger, higher pressure ratio gas turbines use 1001 complex controls systems to skin the compressor surge line and accelerate within predefined mechanical operational limitations. Smaller, low-cost APUs have, until recently, used simplified “open loop” type acceleration control systems. The advent of electronic controls for small gas turbines has permitted the utilization of closed loop acceleration fuel control scheduling where fuel may 0 30 L20 be topped as a function of a single or multiple con- Fuel Flow (PPH) trol variable. Most recent APUs provide higher starting torques by metering fuel close to the design tran- Figure 18. Minimum Fuel Pressure for sient T.I.T. limitations. Typically, two fast re- Good Atomization sponse thermocouples are positioned in the exhaust to measure exhaust gas temperature (EGT) which is a speed dependent indicator of T.I.T. Two EGT thermocouple outputs are normally averaged Fuel atomization effects are dominant on igni- to provide the control input signal. However, if tion as the fuel evaporation rate is inversely pro- there is a disagreement of a predetermined differ- portionate to the square of the fuel droplet size. ential between the two readings, the lourer one is Fuel volatility has a secondary important influence, disregarded and the higher of the two is used for but hydrogen content and end boiling point have, control. If, by certain predetermined criteria, nei- in comparison, little influence. This is well illus- ther probe is providing valid data, then a shutdown trated in the small Gemini APU (Figure 19) where is initiated. unusually excellent ignition is obtained, with very viscous non-volatile diesel fuels, in the extreme A comparison of APU starting torques with conditions of Arctic operation, by use of extremely open loop and EGT topping control systems is fine atomization from a novel rotating cup fuel in- shown on Figure 20. Torques at low-end speed remain essentially unchanged, being established by jector. light-off speed and viscous effects. Above 32 per- In addition, fuel viscosity has a further impor- cent speed, the EGT topping control exhibits tant influence on ignition since the fuel spray angle higher torques than open loop control, until at 80 is significantly reduced with viscous fuels. With percent speed, accelerating torque is approximately this reduction, the fuel spray may not be in correct doubled, providing faster Start time. relationship to the ignitor spark and ignition will not occur. A separate start injector makes it easier to take acount of this, so that in the worst case, most viscous fuels such as JP-5 having up to 16.5 centistokes viscositv at -40°C. and No. 2 diesel having up to 25 clntistokes viscosity at -12”C, can be ignited. 7-18

LP900032-19

Figure 19. 10 Kw Turboalternator 7-19

8.2 Hot Gas Vane Motor The inability of most current front-line military aircraft to provide self-sufficient starting at -34’C conditions has prompted the USAF to sponsor the development of a high torque, high power density “Hot Gas Vane Motor” program described in Ref-

erence 23.~~ The~~. concent~~ is to nrovide the hot motor (Figure 21) as a F116 ~airc;aft retrofit Arctic start kit, for the existing hydraulic Stan motor (Fig- ure 3). Demonstration starts have been completed using’both hydrazine and airiJP gas generatois, and development is continuing with a flight weight unit for eventual overall installation. Small high pressure airlJP gas generators as depicted in Figure 22 have found several recent applications in high energy start and emergency power systems. ---Open Loop -Egt Topping Axial Cross-Section

LPJ00012-20

Figure 20. Effect of Closed Loop Control

8.0 Alternate Approaches It has been shown that the critical phases in cold weather starting are fuel atomization and ignition, plus the transition from unfired to fired torques as dependent upon viscous drag and engine Exhaust Porting thermal response. Alternate starting methods cur------rently under study and development to improve conditions are described as follows.

8.1 “All-Electric” APU The development of a small “Gemini” Turboal- ternator APU is described in Reference 22 where the gas turbine directly drove a high speed solid a. Lundell alternator at 93,500 rpm. Several accessory drive systems were studied, but cost and reliability considerations finally constrained the configuration to that shown in Figure 15, with a conventional reduction gearbox arrangement and lower speed accessories. inlet iorting LP900032-21 The increasing demand for electrical secondary power and dramatic electronic power conditioning component technology advances are requiring a reappraisal of the Turboalternator concept, without Figure 21. Hot Gas Vane Motor a gearbox, where in the alternator could also be used as a starter. The accessory fuel pump, oil pump, etc., would be driven electrically from the conditioned alternator output. Viscous drag during cold starting could be significantly reduced in this arrangement. A disadvantage of this approach is that the alternator inertia may be similar in magnitude to the gas turbine rotating assembly thus pro- longing the start time, and the requirement for battery preheating under Arctic conditions. The replacement of metallic components in the small Gemini gas turbines by the ceramic components is described in Reference 16. The replacement permitted the demonstration of an extremely fast start of 2 112 seconds duration from zero to 100 percent rated speed. 7-20 n

L /- Air Motor Fuel Rich Air/JP Start System e Vane B-- Turbine or lmpingemer L

Figure 22. Fuel Rich AirlJP Start System LP900032-22

8.3 Self-Start APU Concept The gas expansion does not reverse flow into Blackburn Reference 24 fully describes a pulse the engine compressor because flapper valves be- combustor method and apparatus for starting a gas tween the compressor and the combustor act as a turbine engine, (Figure 23). The first process of check valve. The absence of compressor through- the concept method is to inject fuel into the quies- flow as the combustion gases are expanding cent combustion chamber of an at-rest engine and through the turbine minimizes the power required to drive the compressor. Hence, most of the turbine output torque is utilized to accelerate the iner- the chamber volume expands through the turbine tia of the compressor-turbine rotor. Once expan- and imparts rotational energy to it. sion (i.e., blow-down) is completed, the flapper valves automatically open to allow the now-rotating compressor to scavenge and recharge the combustor volume with fresh air. Fuel is then injected I Cantroller I again and ignited to commence another cycle of SCV Combustor the "ratcheting" start process. \ Attempts to develop constant volume type gas turbines have so far been unsuccessful essentially due to combustor valving inadequacies, heat losses, and non-optimum choice of combustor volume to turbine nozzle area ratio. Exploratory research in the combustor process has been initiated at the author's association with a goal of eventually developing a small self-start APU of the type shown in Figure 23 which potentially would not require a stored energy (battery or accumulator) start assist system.

L"1.U

Figure 23. Self-start APU Concept 7-21

8.4 High Altitude, High Speed Start High energy starters and sophisticated electronic controls may not have as much immediate With the use of significantly high starter torque payback as simply reducing the rotating assembly input it is possible to accelerate the engine to near inertia. Advanced material development however, full (100 percent) speed where the combustor inlet can be equally expensive and time consuming. pressure and temperature are substantially higher Thus, all options should be examined in detail, than the ambient conditions, as dependent upon including the all electric gearbox-less APU with compressor design pressure ratio. Combustor load- switched reluctance high speed generator. ing is also reduced and fuel requirements increased making combustor and fuel spray conditions more The requirements to operate aircraft gas tur- ideal for light-off and operation. bines over a large range of environmental conditions prove particularly demanding to the designer, Often specialized start injectors are a require- especially when rapid starting of a cold engine is ment for high altitude ignition, i.e., a means of stipulated at sub-zero ambient temperatures. fuel atomization specific to high altitude low speed ignition. By means of high speed ignition such Cold engine cranking torque characteristics are complexity is eliminated. The difficulty with the basically controlled by the lubricant viscous drag in high speed light-off concept is the high starter the mechanical drive train and accessories. To torque requirement, even though the unfired win- further complicate matters, this viscous drag is de- dage torque is reduced at altitude. Two potential pendent upon the magnitude of the applied start methods of providing the required high starter torque. Either actual viscous drag test data or re- torque are: alistic methods of predicting engine cold cranking torque levels with high lubricant viscosity may be Hydraulic starter motor with increased ac- required before definitive starter specifications can cumulator volume and capability to operate be issued. Experience with start systems for small at up to 150 percent overspeed. aas nirhine APlJs had shown that the total weight --.- ~ Hot gas generator driving a vane starter required for suicessful rapid starting at -54°C :an motor or direct impingement on the rotat- approach the weight of the APU powerhead itself. ing assembly [Figure 22). It is evident that starting torque characteristics of small gas turbines are a function of many inter- Additionally, the use of variable compressor dependent criteria and thus, quite synergistic. In- inlet guide vanes can help reduce high unfired win- deed...~~, isolation~~ of the individual and effect of

dare. torque. mmv of-. the.~~. criteria~ ~involved mav be intangible. In summary, a high altitude start method is Nevertheless, an effort has been’made to dkscribe feasible where the engine light-off is scheduled to these criteria and their apparent effects which may take place at about 80-100 nercent rated speed, prove sufficiently intriguing to simulate renewed rather than the conventional Dractice of about interest in the development of improved starting 5-40 percent speed. prediction procedures. Hopefully, the considerations presented will be 9.0 Conclusions employed by the gas turbine mechanical or systems engineer to more realistically appralse the complex Specific power levels of gas turbines continue starting problem issue, and may serve to make him to increase paced by continuing advancements in more cognizant of the effects of engine parasitic key technologies such as aerothermodynamics, viscous drag sources on starting during the critical computational fluid mechanics, and high tempera- engine preliminary design phase. Briefly ad- ture materials. Power sections consequently con- dressed, but of critical importance, is the role of tinue to diminish in size and increase in speed for the combustion designer who provides a key input a given power rating. Associated technology devel- in designing and developing a combustor capable of opments in engine driven mechanical equipment rapid consistent light-offs over a wide operating such as reduction gearboxes, generators and hy- range at the earliest possible speed thereby permit- draulic pumps proceed at a more modest rate. ting the transition from negative (cranking) to posi- This technological disparity precipitates dispropor- tive engine assist during start. tionate sizing of the engine powerhead and driven equipment, spawning increased viscous drag and Methods of reducing APU viscous drag and potentially reduced cold start reliability. start energy requirements that deserve future study are the all electric gearbox-less APU, and possibil- Increased viscous effects are mitigated to some ity of the self-start pulse combustor concept. degree by advanced engine ECUs with inherent capability to tailor the starting sequence for all environments. The nemesis of the advanced ECU is 10.0 ACKNOWLEDGMENTS fixed cost which can represent more than half the The author would like to acknowledge the ef- cost of an advanced small APU. Implementing forts of the following people in preparation and VLSI technology will serve to deter disproportion- editing of the manuscript: Jack Shekelton, ate cost trends and should be pursued to realize Mechele Wilson, Patti McNally and Trudy Lentz, cost effective advanced digitally controlled APUs. and to Sundstrand Power Systems for publication authority, and the combined efforts of the SAE Starting Committee in their continuing function to improve aircraft gas turbine starting technology. 1-22

REFERENCES Discussion 1. AIR 781A - Auxiliary Power Sources for 1. H. Saravanamutton, Carleton University Aerospace Applications. What difficulty do you have in obtaining realistic 2. AIR 781 - Guide for Determining Engine compressor data at very low speeds, say 30-404/, ofdesign? Starter Drive Torque Requirements Compressor characteristics are very seldom available below SO%, a typical idle speed. 3. ARP 906A - Glossary, Aircraft Engine Starting and Auxiliary Power Author: 4. AIR 944A - Pneumatic Ground Power - Reference to Figures 2 and 12 show that viscous power Supplies for Starting Aircraft losses increase more rapidly than thermo-dynamic 5. ARP 949A - Turbine Engine Starting System power increase, with decreasing ambient temperatures. Design Requirements - We test our compressors to as low as 20% speed. I am 6. AIR 1174 REV A - Index of Starting System not knowledgeable of low speed efficiency levels of Specifications and Requirements larger more sophisticated multible-spool axial 7. AIR 1467 REV A - Gas Energy Limited Start. compressors. ing Systems - Removal of misfired cartridges is indeed a dangerous 8. MIL-STD-21OB - Military Standard Climatic task. Extremes for Military Equipment 9. SAE SP-398 - “Starting Systems Technology.” 1984. 2. P. Sabla, MGFL CAD GEAC Why in figure 9 does weight go up by the factor 5 (data 10. SAE SP-678 - “Starting Systems points) for the same energy level. Is this driven by design Technology 11.” 1986. requirements or by external requirements? 11. Rodgers, C., “Starting Torque Characteristics of Small Gas Turbines and APUs,” ASME Author: 79-GT-95. 1979. The weight factor (see Figure 9) is actually less than two, and 12. Rodgers, C., “Impingement Starting and results from the correlation of many different APU and Power Boosting of Small Gas Turbines.” main engine configurations produce a relatively wide ASME 84-GT-188. 1984 correlation band. 13. Rodgers, C., “Secondary Power Unit Options for Advanced Fighter Aircraft.” 3. R. Wibbelsman, GE AIAA-85-1280. 1985 The moment of inertia depends very much on the design of the engine and is also a function of the mission of the aircraft. 14. Rodgers, C., “Fast Start APU Technology” SAE 86-1712/SP-679. 1986. At high Mach engines the rotor is stronger and heavier and impacts on the torque. 15. Rodgers, C., “Pneumatic Link Secondary Power Systems for Military Aircraft” SAE 88-1499. 1988. 16. Rodgers, C., Bornemisza, T., ”Fast Stan Ceramic APU” SAE 89-2254. 1989. 17. Rhoden, J. A., “Modern Technology Secondary Power Systems for Next Generation Military Aircraft.” SAE 84-1606. 1984. 18. Gazzera, R. W., “Advanced Pneumatic Start Systems for APUs.“ SAE 86-1713. 1986. 19. Drury, E.A., “Low Temperature Starting of the VI Battle Tank.” ASME 82-Gt-190, 1982. 20. Trippett, R.J., “A High Speed Rolling Element Bearing Loss Investigation” Trans ASME Vol. 100 Jan. 1978. 21. Meyer, W.E., DeCarolis, J.J., Stanley, R.L., “Engine Cranking at Arctic Temperatures,” Society of Automotive Engineers Transactions, Vol. 63, 1955, Pg. 515. 22. Rodgers, C., “Performance Development His- tory - 10 KW Turboalternator”, SAE 740849. 1974. 23. Dusenberry, G., Carlson, D., “Development of a Hot Gase Vane Motor for Aircraft Starting Systems.” SAE 861714. 1986. 24. Blackburn, R., Moulten, J., “Semi Constant Volume Pulse Combustor for Gas Turbine Starting”. AIAA 89-2449. COLD START OPTIMIZATION ON A MILITARY JET ENGINE by H. Gruber Hans-Sachs-Str. 16b 8038 Grobenzell Germany

1. INTRODUCTlON Cold-start testing at temperatures of approximately minus 40'C (233 K) was performed on 2 RE 199 engines at a West German altitude test facility.

The engines were of the same build standard with exception of the seal configuration (labyrinth or brush), and running times.

One part of the test was performed with F34 fuel, the other with F40.

The Fiqure 2: Cooling-Air Flow in m facilities Test Cell . test methods and test results will be presented in this report. After the air had passed through the compressor, the temperature was reduced to approximately 40°C by a water-air cooler. The cooling-air tem- 2. TEST FACILITIES perature was then reduced by another 65'C to minus 25'C by an additional The general layout of the altitude downstream brine cooler before enter- test cell in which the test series ing a cooling turbine, which then were performed is shown in fiqure 1. lowered the temperature to the level The engine test set-up was largely required for test-cell induction, in identical to that of sea level test this case minus 40'C to minus 45'C. configurations. The induction air temperature remained at this level from starting to idle speed. To maintain this, it was necessary to bypass a certain amount of air in relation to the specific engine operating conditions.

Up to idle speed, pressure in front of and behind the engine was approximately 100 kPa. A movable damper installed behind the engine prevented the rotation of the engine spools during the cooling phase. Additionally, the high-pressure spool was blocked by internal resistance from the Slave-Loading-System. The intermediate-pressure spool would not turn when both high and low-pressure spools were static. Fiqure 1: Altitude Test Facility An additional locking pin was installed radially to block the fan, thereby preventing the low-pressure The induction air path through the spool from rotating when the damper test facility and test cell is de- was opened. scribed in fiuure 2. A steel pipe holding approximately 25 liters of fuel was installed in the test cell directly ahead of the fuel backing pump. This portion of fuel very quickly reached the required minus 40'C during the test cell cooling phase. 8-2

shows the temperature drop with respect to cooling time.

After approximately 4 hours, the engine as well as the fuel and oil had been completely cooled.

The damper (figure 2) was opened approximately 1 minute before starting; the fan-locking pin was removed as the starting sequence was initiated. After reaching idle speed, the engine was to be accelerated to maximum power. Fiqure 3: Temperature Orop During Cooling Phase During the starting phase, load (50 kW at the high-pressure spool) was applied to, and bleed-air (approximately 3. STANDARD 0.22 kg/sec. from high-pressure com- ENGINE pressor area) was taken from the en- The general layout of constructive gine. features of the three-spool engine RB 199 is shown in fjqure. The following main parameters were to be monitored:

Time

Temperatures and pressures in the test cell

m Engine speed Turbine-area temperatures

The testing comprised

11 starting attempts at minus 30'C using F34 5 starting attempts at minus 30°C using F40

Figure 4: 12 starting attempts at minus 40°C Cross-Section of Three-Spool using F34 Engine RBI99 16 starting attempts at minus 4O'C using F40

The assemblies incorporating labyrinth All of these attempts were carried out seals during the first test phase and with a high margin of success. brush seals during the second test phase are specially identified.

The exhaust gas temperature, which supplied important information for engine monitoring during the starting 5. TEST RESUIS phase, was measured by rakes located behind the low-pressure turbine. 5.1 Combustion chamber iqnition characteristics It was d-ifficult to reliably evaluate the ignition characteristics of the 4. TEST METHODS combustion chamber during the individual starting phases with the in- The cold-starting tests were to be strumentation avai 1ab1 e. Combustion performed in accordance with the chamber ignition generally means a standard starting procedure (at 15 "C) rapid increase (1 50"/second) of the shown in fioure 5. turbine outlet temperature. During testing, this rapid increase was not observed until the low-pressure spool began to rotate (12% NL). This may have been caused by an improvement of the flow and combustion 5.1.2.1 Starter jet flow numbers characteristics in the combustion (greater pressure drop) and turbine Starter jets with flow numbers 0.3 and areas once the low-pressure spool be- 0.6 (FN = c/m were available for gan to rotate. Unusually high tempe- the test engine. ratures at the high-pressure compressor outlet indicate a blockage of Starts were performed using both star- airflow in the combustion chamber and ter jet sizes and both F34 and F40 turbine before the low-pressure spool fuel. The starts using FN 0.6 were has begun to rotate. successful in every case, whereas only half the tests using the smaller jets 5.1.1 Influence of the igniter plug offered the desired results. location The major advantages of the larger Fiqure 6 shows a simplified drawing of jets were, as expected, the increased the combustion chamber head featuring rate of engine acceleration to the the arrangement of the starter jets, point of main fuel jet activation, igniter plugs and vaporizers. improved heating-up of the combustion chamber and, consequently, improved combustion propagation of the main fuel. In order to prevent excess fuel delivery and the corresponding danger of hotstarts, the flow number 0.45 was determined in an additional short test phase as being optimal for all further tests.

All starter jets were removed and cleaned after every third or fourth cold start. The flow values dropped by as much as 5% between starts. The spray angle remained practically unchanged.

Fiqure 6: Combustion Chamber Detail According to general experience, combustion propagation should have been better with F40 than with F34. This, however, could not be confirmed by the test data. The differences between F34 and F40 were within test result scatter.

5.1.2.2. Starter jet configuration

The plus-sign indicates: At the beginning of testing it was observed that the starter jet fuel had - the igniter plug protrudes into ignited on only one side of the com- the combustion chamber bustion chamber. Fisure 7 shows the exhaust gas temperature profile ob- The minus-sign indicates: tained from rake measurements during starting phase 1. It is plain to see - the igniter plug is located "Out- that the maximum value was measured at side" the combustion chamber the right-hand starter jet. This was documented by a video-camera, as well. The test evaluations did not reveal By the time the main fuel is ignited, any tendencies regarding ignition cha- approximately 23 seconds after the racteristics even though other tests starter jets are activated, both star- indicated that the location of the ter jets are burning. It was assumed igniter plug did indeed influence that the uneven ignition of the star- ignition characteristics. ting fuel during the initial starting phase was due to a difference in tim- ming caused by the different lengths of fuel-feed tubes to the port and Influence of the starter jets starboard starter jets. The equal iza- 5.1.2 tion of the pipe lengths led to an The influence of the starter jets is increased average temperature level, greatest during the first phase of which meant that an higher amount of starting, since their proper function- energy was present within the combus- ing is critical to the ignition of tion chamber and the turbine area. the main fuel. 8-4

This required that the main fuel feed be activated at a certain point in time rather than at a specific engine speed. The point at which the main fuel jets were activated corresponds to 12 - 15% high-pressure spool speed at minus 30'C and minus 40°C. During ambient starts (engine-intake temperature = 15'C) 19 - 23% high-pressure spool speed was reached, compared to 25% during hotstarts (engine-intake temperature = 50°C).

5.3 Toraue profile Fiaure 7: Exhaust Gas Temperature Distri HgWd shows the torque profiles for bution During Cold-Starting cold-starting of the test engines with different seal configurations.

It is suspected that the starter jet feed lines contained residual fuel vapors, whose effect on fuel delivery increases with an increase in fuel line length. '"Wet cranks", performed prior to the starting tests would su- rely have helped preclude any such difficulties.

The exhaust gas temperature, which is a primary factor in evaluating the effect of pipe-length equalization, subsequently showed a greater rate of increase during the starting tests that followed.

5.2 Startins Procedure optimization Fiaure 8: Torque During Cold-Starting The starting procedures for cold - starting were substantially modified compared to the sequence at 15 'C (fi- It must be taken into consideration gure 5). that the engine equipped with the During starting at minus 30°C and brush seals had been operated for ap- minus 40'C. a sufficient high-pressure proximately 180 hours prior to spool speed (19.23%) could not be cold-start testing and could therefore reached with normal starter jet assi- not be representative of a new brush- stance alone. It would have been help- seal equipped engine. ful to keep the starter jets activated for a longer period of time to augment The comparison between the different the thermal energy available in the engine builds showed that: combustion chamber, thereby improving the conditions for the ignition of the m the differences in the resistance main fuel. This, however, would have curves for both engines were small. resulted in a very short overlap of starter engagement and main fuel jet m the low-pressure spool breakaway of activation. Experience shows that a the brush seal engine was not affec- long period of starter engagement with ted by engine intake temperature the combustion chamber "lit" or par- between minus 30'C and minus 40'C. tially "lit" is required for success- The low-pressure spool breakaway oc- ful cold-starting. This could be ac- curred at approximately 19% high - complished by varying the starter en- pressure spool speed, which is gagement time in relation to engine slightly higher than that of the intake temperature, as is the case engine equipped with labyrinth with other engines. seals. It appears that the labyrinth configuration was indeed affected by The following starting sequence was engine- intake temperature. found to be optimal for the test engine: It is assumed that the friction of the cold brush seals was lower than that of starter jet open: after 2 to 5 seconds the cold labyrinth seals. This may be main jets open : after 20 + 5 seconds due to the relatively long running time of the brush seal engine and the fact that the brush seal pins did not come into contact with any rotating parts at these low temperatures. 6. CONCLUSIONS Discussion These results are based on tests with 1. C.Moses, Southwest Research Institute two engines of different seal configura- You stated that there was very little difference in the starting tion and with different running times. characteristics of the two fuels. Was this also true when you They are certainly useful as a basis for used the smaller fuel nozzles? general assertions concerning the behaviour of those engines; they do not, however, allow any final conclusions to Author: be drawn, Further extensive testing The situation was comparable with both flow nozzle sizes. would be necessary. 2. P. Sabla, GEAE The following points of interest, rele- Was the inlet air dry? How was the moisture removed? vant to the RBI99 engine cold- starting were noted; Author: The inlet air was not dried. We ran all the tests with The differences between the igni- unconditioned air, that means with normal ambient tion propagation properties of fuels F34 and F40 were within the measure- conditions. ment scatter. 3. D. Hennecke, Technische Hochschule Darmstadt Do your test cell arrangement and your test procedure The starting characteristics of the engines equipped with brush and ensurc that you have steady-state temperature conditions labyrinth seals differed only when you start the engine? slightly. The relatively long running time of the engine equipped Author: with brush seals prior to testing Figure 3 of my report shows the temperature drop with must be taken into account. respect to cooling time. During this phase a low amount of wol air was passing through the engine although the damper The configuration and size of the was filled behind the engine. The spools did not rotate. The starter jets is of fundamental si- temperature level in and around the engine was controlled gnificance to the ignition propaga- all the time by temperature probes. After approximately 4 tion in the combustion chamber and, subsequently, to the success of the hours, the engine as well as the fuel and oil had been cooled starting process. down to the required -40 C.

4. C.Rodgers, Sundstrand Power Systems m The optimization of the starting process entails changing from an Was ignition of the starter jets programmed at 5% n? engine-speed-determined main fuel Did a revision of the cold start fuel schedule require supply activation point to a time- reprogramming of the ECU? determined point. Author: m A further, marked improvement in The ignition of the starter jets was initiated at 5% n by the ~ starting characteristics may be ach- engine driver. ieved by generally changing to an The tests were not performed with a DECU, but the intake temperature-related-starter- exverience mined during the tests later on was introduced in engagement sequence, as is the case ~ I with a number of other engines. fuel scheduling of another engine programme

COLD WEATHER IGNITION CHARACTERISTICS OF ADVANCED SMALL GAS TURBINE COMBUSTION SYSTEMS

I. Critchley, P. Sampath, F. Shum Pratt & Whitney Canada Inc., 6375 Dixie Rd., Mississauga, Ontario, Canada L5T 2E7

friction in bearings and gears, accessory loads and lubricant viscosity effects, tend Low temperature and high altitude to increase the initial drag at low starting requirements of present day small temperature as seen in Figure 1. Lower aero-gas turbine engines are discussed from battery voltages during cold soarlaltitude the viewpoint of their influence on the operations and the higher drag therefore design of the combustors and ignition tend to reduce engine cranking speeds and systems. Use of electric starters, comnon air flows during start-up and hence reduce in small engines, creates particular the combustor air pressure drop available challenges to starting especially under for atomization and mixing within the cold soak sea level and altitude start up combustor. conditions. FIG,: TYPICAL STARTING TORQUE REQUIREMENTS The main factors in combustion system design affecting starting performance are !I discussed; including combustor sizing, fuel placement, fuel atomization, fuel scheduling and igniter selection. Experience of Pratt and Whitney Canada (PPWC) with small engines for different applications are also covered. Low emission requirements may adversely affect starting performance. necessitating use of elaborate fuellignition systems, some recent developments are described.

Nomenclature

AP = Pressure drop SMD = Sauter Mean Diameter To = Ambient temperature T3 = Combustor air inlet temperature Adding to these difficulties is the T5 = High pressure turbine exit temperature increased viscosity of the fuel at low TTLTiL = Time to light temperatures which will result in a more la= Air temperature coarsely atomised fuel spray and hence be Tp = Fuel temperature more difficult to light. Low pressures and temperatures adversely affect both the INTRODUCTION initial ignition process and the subsequent flame propagation because of slow The ability of an aircraft gas turbine evaporation and reaction rates. Even after to light-up and accelerate easily and a stable flame has been established its reliably is of crucial importance both for efficiency and therefore the ability of the ground and altitude starts. The engine to accelerate are significantly achievement of this goal under adverse lowered under extreme conditions of low conditions of low soaking temperatures, or temperature and pressure. Reduction of the at low temperature and pressure during combustion efficiency occurs at a time when altitude operations, imposes severe increased engine drag necessitates more constraints on the combustion system. energy from the combustion system. Typical cold start requirements for small aircraft gas turbines would be for air All these factors combine to increase temperatures down to 220~and with fuel the difficulty of starting under adverse VlscOSitieS up to 12 centistokes. Altitude conditions and can result in the inability relighting capability is needed typically to light, accelerate or lead to torching up to 11,000 meters for main power plants and consequently turblne damage which may and in excess of 12,500 meters for the new not be imnediately apparent. Other generation of Auxilliary Power Units (APU), possibilities are hung starts or extremely which may be required to start and rapidly long times to light and idle. Solutions to generate emergency power after cold soaking these problems impose restraints on the at high altitude. combustion system and increase the complexity of the fuel and ignition Small aircraft gas turbines are systems. This is particularly undesirable tYpiCally equipped with electric starters. on small aero gas turbines where these the starting torque being set to be greater systems can be a significant contribution than the maximum drag torque during to the total engine cost and weight. cranking conditions. However all the major drag sources; compressor aerodynamic load, Advanced engines with their demand for windage in compressor P turbine discs, lower specific fuel consumption and 9-2

specific weight operate at higher engine FIG 3: FUEL iNJECTOR MOUNTING ARRANGEMENTS pressure ratios and turbine inlet temperatures. These requirements have driven combustor designs to be more compact ...... - for durability, to have lower combustor ...... &a++flf e. 7qy-5: pressure drop for performance and to operate with leaner primary zones for smoke -,,e7 /j' and emissions control. All these ...... requirements have made the problem of designing for cold and high altitude starting more challenging than ever before.

DESIGN CONSIDERATIONS ...... The first requirement of the combustor is good ignition performance, necessitating a combustion system with adequate fuel atomization, optimum fuel ...... :..I...... placement and a reliable and effective ignition system. However, the combustor Primary zone must be sized with sufficient volume for reasonable air loadings [1]. The combustor primary zone flow structure and considerations. The number selected will fuel - air ratio must be Set to achieve influence combustor exit temperature rapid ignition and subsequent flame distribution as well as ignition and stabilization and propagation around the starting performance. Engine manufacturers combustor annulus. A recirculating primary generally use empirical correlations zone flow is COmnOrIly used. The structure developed to suit their own combustion of the recirculating flow will vary with systems. Figure 4 shows the minimum different designs and may be driven by ignition fuel flow for an airblast fuel swirling air through the fuel nozzles or, injector system with different numbers of as is more comnon in small engines, a fuel injectors: the minimum ignition fuel Single toroidal vortex, Figure 2, driven by flow required decreases with increasing wall cooling flows around the front end of number of fuel in.jectors. Combustor. The fuel nozzles are positioned to spray the fuel into the combustor so as Primary zone fuel air ratio control is to maximize the residence time of the fuel an important aspect of combustor design in the primary zone. The nozzles may be affecting ignition, combustion efficiency mounted radially spraying upstream or and emissions performance. One of the axially through the dome of the combustor means of reducing emissions is to design or tangentially as shown in Figure 3. the primary zone to be fuel lean, however, a rich primary zone is usually required for FIG 2: TYplCAL COMBUSTlON SYSTEMS FOR GAS TURBINE ENGINES good ignition and flame stability. Elimination of local fuel rich pockets in the primary zone will reduce smoke but may adversely impact ignition, Figure 5. In most cases detailed combustor hole ...... pattern development and fuel spray optimization are necessary to meet both ..,.. cold ignition iind emissions requirements. FIG 4 MINIMUM IGNITION FUEL FLOW FOR DIFFERENT NUMBER OF FUEL INJECTORS ...... \ \--- I.. ,,..,... .,.,...- .... ,.,. L...... L,,r" ,.,,..I...,.... .e..

The igniters (usually 2 to provide redundancy) are positioned in a fuel rich region of the primary zone. However, care must be taken to prevent excessive fuel wetting which can lead to premature igniter failure or spark quenching. Circumferential positioning of the igniters will depend on the degree and direction of any residual swirl from the diffuser air outlet which may be present in the primary zone and the trajectory of the fuel spray. In practice, igniter position and FUEL DELIVERY SYSTEM penetration are determined by development testing. Optimum penetration will be a The fuel system design, including fuel compromise between improving ignition and injector, manifold and scheduling plays an excessive igniter tip temperatures as important role in cold ignition penetration increases. performance. Good atomization is crucial for good ignition. Figure 6 shows a The number of fuel injectors is comparison of droplet size between pressure generally selected as a compromise to and airblast a.tomizers, showing the satisfy both combustor performance superior performance of airblast injectors requirements and cost and weight except under starting conditions. 9-3

W 5: REU7k?+4SHIP BETWEEN UIN!MUM IGNmON FUEL FLOW will light but fail to accelerate. In this AN0 SWEEMlSSWNS case, the primary fuel inJectors may have low combustion efficiency under adverse conditions. Ifthe engine then fails to develop sufficient air pressure for rhefuel control to schedule an increase in fuel flow, the secondary fuel injectors may not have sufficient fuel pressure for good atomization. This can be overcome by increasing the starting fuel flow as shown in Figure a. The engine hangs just below 260K with the lowest starting fuel flow; increasing starting fuel flow eliminates hanging and the engine always accelerates at temperatures above its lighting limit. However, this is not always a viable solution since over-fuelling can lead to excessive exhaust temperatures during light-up. The hanging can also be eliminated by use of a single manifold (unstaged) system, Figure 9. However, such a system usually has a fairly high ambient Pressure Atomizers temperature ignition limit, although this can be lowered by increasing the starting Pressure atomizers have been used fuel flow. successfully for many years on gas turbine engines. However, on engines with a high FIG T; VARIATION OF PEAK TEMPERATURE turn-down ratio (ratio of mximum to DURING STARTING W!TH minimum fuel flow), low fuel pressure and NUMBER OF PRIMARY FUEL INJECTORS thus poor atomization will result at the starting conditions. This problem can be overcome to some extent by staging of the fuel whereby some fuel injectors near the igniters receive fuel at a higher pressure which will improve their ignition performance. Staging 1s achieved by use of a flow divider valve whlch my be single or multi-port providing several levels of staging to aid a gradual flame propagation.

FIG 8: COYPARISON OF SPRAV SAUTEA MEAN OIAMmRS FOR PRESSURE JET AND PURE AIRBUST ATOMIZERS

FIG 8: EFFECT OF STARTING FUEL FLOW MI HwGulG

Staging of the fuel can result in torching or flame propagation difficulties. Correct proportioning of the fuel to the various stages is required to prevent long manifold fill-up times with consequent fuel dribbling and pooling which can lead to torching during starts. Recent PT6 engines use a 2 stage system with primary and secondary fuel injectors. Figure 7 shows how the peak transient combustor exit temperature can be reduced by optimizing the number of primary and secondary fuel injectors. An optimum fuel flow split between stages exists when the temperature peaks associated with the light up of each stage are equalized and neither predominates. In this case this was achieved with 10 primary injectors. Engine hanging can also occur as a result of fuel staging where the engine 9-4

An alternate approach to improve PawC has successfully demonstrated ignition starting performance and to eliminate with airblast fuel injectors at combustor hanging is use duplex fuel injectors, pressure drops down to 38 mn of water at Figure 9; lighting should not be a problem fuel and air temperatures of Z30K. although the limit was not established in this testing. However, the complexity of Piloted fuel systems are used to duplex injectors and the need to maintain enhance starting and to alleviate torching adequate fuel passage Sizes limits their from overfuelling at low temperature and use to low turndown ratio applications. pressure conditions. Piloted systems usually employ a hybrid fuel nozzle or a torch igniter. A torch igniter consists Of FIG 9: COMPARISON OF STAGED AND UNSTAGEO a combined igniter and pressure jet FUEL FLOW ON HANGING atomizer. Problems of coking and blockage can be minimized by mounting the atomizer in a relatively cool region or by purging. Application of torch igniters is most comnon on vaporizing fuel systems but has been used in some airblast applications. Hybrid fuel injectors consist of a pressure atomizing primary and an airblast secondary nozzle. Fuel is introduced to the primary fuel circuit first for initial light-up and subsequent propagation to the remaining airblast fuel injectors. Figure 10 shows a comparison of minimum ignition fuel flow for a pure airblast and a hybrid airblast ~ystemat cold conditions. For the hybrid system, the minimum IgnitiOn fuel flow is independent of combustor pressure drop. With the pure airblast system. time-to-light is governed by the manifold fill-up time and the achievement of favourable conditions at the igniter. This can result in much slower ignition with possible pre-ignition fuel accumulation and torching at ignition. With the hybrid system, ignition can be achieved at extremely low combustor air pressure drops and with a much quicker Airblast Atomizers time-to-light. Ignition is usually instantaneous and a controlled light-around Advanced gas turbine engines with without any torching is achieved. their reouirements for better sorav oualitv and longer life generally use airbiait fuei FIG 11: STARTING CHARACTERISTICSOFA pT6 ENGINE injectors. This type of injector is WITH VAPORIZING AN0 PRESSURE particularly suited to small engines since ATOMIZING FUEL SYSTEMS fuel passages can be larger than for pressure atomizers which are prone to blockage In small engines. As a result of the dependence on air pressure drop across the Combustor for atomization, droplet sizes are usually coarse at starting conditions, particularly at low temperatures when the fuel is more viscous and cranking speeds are low. For small gas . , . . , , turbine engines, under extreme conditions, L..* ." ... ," .,. ... .* ,_ the combustor air pressure drop could be as .1.....,." ...... "..- low as 25 mn of water. Development of airblast fuel injectors for small engines has concentrated on improving atomization at these low pressure drops. Testing at

FIG 10: COMPARISON OF IGNITION PERFORMAHCE BETWEEN HYBRID AND PURE AIRBUSTSYSTEMS

-- -- I".,... .,.I.""..."..,"

Vaporizers

Vaporizing fuel 5ystems are widely used in small gas turbine engines; this type of system requires the use of an external flame source like a torch igniter for lighting. Once developed, lighting performance is governed by the torch igniter and is very similar to pressure atomizing systems as shown in Figure 11. However, problems may occur with flame propagation around the combustor depending detrimental to igniter durability, Figure on vaporizer design and spacing. Slow 13. The final choice will be the minimum flame propagation may cause unburned fuel energy to meet cold and altitude ignition to accumulate during start-up and can lead requirements so as to maximize igniter to high exhaust temperatures, Figure 11 and life. long starting times. FIG 73 EFFECTOF IGNITION ENERGY ON IGNITION PERFORMANCE FUEL MANIFOLD AND CONTROLS AND IGNITER DURABILITY Fuel manifold design plays a role in the success of the starting system. A small diameter manifold can give excessive line pressure losses which can ultimately affect the combustor outlet temperature distribution. A large diameter manifold will require a long filling time leading to dribbling and torching. It is necessary to find a compromise between these two requirements. Spiking the initial fuel flow can reduce the manifold filling time and the torching tendency, a solution Which has become more viable with the advent of electronic fuel controls. The accurate fuel scheduling off any required engine operating parameter provided by electronic controls is also of great benefit to starting performance since it is possible to operate with a small margin between sub-idle surge and the acceleration demand. Alternate means of ignition have also been used or are under study. The Air assist systems use high pressure concept of plasma jet ignition has been air which can be introduced into the fuel studied by various researchers 121. In this manifold to aerate the fuel thereby type of igniter, Figure 14, the arc does assisting atomization during light-ups. not occur at the tip of the igniter but is This kind of system has been used for heavy confined in a small cavity. The gases (and fuels in industrial applications. liquid fuel) in the cavity are heated by the arc and a pressure rise results. The IGNITION SYSTEM pressure is allowed to exhaust through an orifice resulting in a jet of energetic Most small gas turbine engines use plasma which is propelled into the ignition system energies in the range 1 to combustor. The concept was developed for 4 joules. Both low tension (L.T.) automotive applications but can offer semi-conductor (approx. 4kV) and high advantages to gas turbine designs since it tension air gap (typically 24 kV) systems will allow the spark kernel to be projected are employed. For a given output energy, away from combustor wall cooling flows into both systems give similar ignition that part of the primary zone most amenable performance as shown in Figure 12. Air to ignition. The principle can be applied cooling to the igniter can also affect to both high tension and low tension ignition performance. In this case, Figure systems. 12, the provision of air cooling aided Fmeroencu Dower units far aoecial ignition which suggests that rich -r-~.- conditions prevailed near the igniters. applications can have a requirement for Igniter cooling air is not always very fast light-up and acceleration at very beneficial to starting and the effect high altitudes. A one second time-to-light depends on the environment adjacent to the at altitudes up to 15,000 meters with pure igniter. airblast fuel injectors has been demonstrated by use of oxygen injection at FIG 12 EFFECT OF IGNRER NPES ON IGNITION the fuel injectors. The recent development of solid state LT exciters has improved the efficiency of the energy transfer process so that a given energy can be delivered at the igniter tip with a smaller, lighter, cheaper exciter. This type of device can also offer improved control of spark rate which can be maintained independently of input voltage and ambient temperature. This is particularly important for continuous ignition applications where a low spark rate is desirable. Future developments could allow the adjustment of spark rate or energy to suit ambient or engine condition, I diagnostic capability will also be "0 IO lb 100 111 ,so available to warn operators of impending EOY.U.100,IIO~"~~DlOl ignition system failure. ,In or WLIER)

rhe most dominant effect on ignition FUEL EFFECTS performance is the energy SUpplied at the igniter tip, increasing energy level will Small gas turbines, even for aero improve ignition Derformance but will be applications, have to operate wlth a wide 9-6

variety of fuels such as Jet A, Jet Al, influencing Starting performance include JP4, aviation gasoline and diesel fuels. number and type of fuel injectors, These fuels have widely differing stoichiometry of the primary zone and the properties giving very different efficiency of the combustor. Sub-idle atomization and ignition characteristics at performance of rotating components also low temperatures. affects starting performance.

FIG

Some requirements for low emissions may adversely affect coldlal titude starting, requiring the use of more .elaborate fueljignition systems to achieve required performance. Ability to model transient per,rormance of engines and combustion sy:;tems while simulating Ignltion performance is governed starting, can enable better understanding mainly by fuel viscosity and volatility. of. and improvements to, this very Decreasing temperature reduces volatility important aspect of engine performance. slowing evaporation rates and also increases viscosity deteriorating atomization quality. Data on these effects FIG 16: EFFECT OF FUEL TYPE ON TIME TO IDLE for a variety of fuels were obtalned by P6WC in a research program sponsored by CDND and USAF [3]. Testing was carried out on a PT6A-65 engine at low temperatures with a wide variety of fuels. The effect on time to light is shown in Figure 15 and on time to idle in Figure 16. The results clearly indicate that the more volatile, less viscous fuels like JP4 have better starting characteristics than the high aromatic content fuels.

FIG IS: EFFECTOF FUEL NPE ON TIMETO LIGHT

CONCLUSIONS Reliable cold day and altitude start-up of modern aero-gas turbine engines requires careful optimization of combustor internal flows, fuel atomization/placement, ignition sourcellocation and fuel schedule durlng tne starting cycle. Other factors REFERENCES:

1. Lefebvre, A.H., "Gas Turbine Combustion", McGraw-Hill Book Co.

2. Zhang, J.X., Clements, R.M., Smy, P.R., "An Experimental Investigatlon of the Effect of a Plasma Jet on a Freely Expanding Methane-Air Flame", Combustion & Flame, 50. (1983).

3. Gratton, M.. Critchley, I., and Sampath, P., "Alternate Fuels Combustion Research", AFWAL-TR-84-2042, July 1984.

Discussion

1. C. Scott Bartlett, Sverdrup Is consideration given to use of gaseous fuel injection or a gaseous fuelled pilot torch for cold start assistance? Is the primary reason that gaseous fuels are not used due to the complexity of the additional systems required?

Author: Thequestioneris right. Although thebenefit ofgaseousfuels for ignition is well known for static applications it would normally not be considered for commercial aero- applications due to the complexity and logistics of providing and refuelling such a system. However the idea could he considered for military applications to aid, for example, relighting at very high altitudes. In fact, work has been carried out at Pratt and Whitney Canada using oxygen injection through igniters and fuel injectors which demonstrated a significant extension of the relight envelope to high altitudes.

2. D. Hennecke, Technische Hochschule Darmstadt If I understand correctly you emphasized the importance of smoke emission during start-up. Is this not unimportant compared to smoke emission at full power conditions?

Author: During my presentation I was referring to high power emissions. Therefore I would agree to this comment.

. senior staff bggineer General Electric CFmpanY Feebles Test Cpration 1200 Jaybird Rmd Feebles. Ohio 45660 USA iz!m!xY me subject of a-td and adaptive jet engine SMing strategy is,a ocaoplicatd subject involving many intricate sets of conflicting -, bath fmn the starxipint of the engine envimmmtd situaticm as well as the engine health situation. In the past, many Wign had to be made which dtdin limiting the engines ability to adueve 100% starting success, particularly in the "off desiqn" rqiom. '5e advent of the ccanputing of digital controls mw rake it pcssible to achieve a major step forward in the control system ability to cope with the mltiplicity of situatiom confronting the engine starting deslw. lhere are lrany different strategies that auld be employ&. "is paper presents the one used by the a an3 CIMT ccwoercial Family of lqehim bypass ratio turkofan engines. Kunwsws variatiom of this basic mnoept muld be employed. It is hopes that this paper will prcanste a new vision of what is pmsible to achieve in the way of:

- s~lutionsto the present design dilennms facing the engine desiqer. - Better reliability at the environrwtal extremes. - Wre reliable starting of deteriorated engines. - mre reliable starting under extrem ~EIY- situations. - mke starting mre certain deradverse battle damage situation. we have just kqun to scratch the surface. lrrt us mw continue to exploit this concept on the next generation of jet engins..

Table of contents

1.0 Intrxduction

1.1 Basis of mis Pap 1.2 General operation 1.3 Manual versus Autamatic

2.0 overview of System and Its ~esponseto Unusual Events

2.1 GrourdStartWm 2.2 AiI start sequ- 2.3 Arctic weather oesiqn mideratiom

3.0 Stall Sensing d-erE-

3.1 GxmrdstartStdllSenSing 3.2 AbStartStallSensing 3.3 FGeseqU€xEingDletoStall

3.4 kiaptive strategy for mines with Repeatea rround Start Stall History 4.0 Light Off Sensing and Resequencing

4.1 LiQlt Off Sensing 4.2 CXW.ln3 Start Light Off -9 4.3 Arctic Gxmrd startkg Resequenclng 4.4 Air SM Liqht Off Resequencing

5.0 Fuel Sdkxiuling Feab.1~5

5.1 FADM: "dl 5.2 Nonral rround Start Strategy 5.3 Arctic Gmurrl start Strategy 5.4 Air start strategy 6.0 Starter Usaqe strategy Curing Air Starting Full Authority Digital 0"l system are amkg into Wi& use in the 1-e osmexcial jet engine inhztq, ani a~estarting to wxin the military engine market. Sane of these engines employ ahticstart ypence features, which s-ly a-te the wentimal hyd?m-I"ru cal mtml ccnoept, Weothers take advKltage of the ccmpning pcwer of the digital amchand incorp3mte adaptive system features. ?he CFMI ani Gewal Electric large Cnmen2ial engines, sx31 as the CM6-5 and CF6-8CC, fall into this later category.

"is appmadl is largely being driven by 6mno.i~amsidexations because in wrld-wide operaticsl, aircraft wing the% engines calimlly land in relatively wsrgprtd airports. under those c' , if the engine will not start the psenqem nust be unloaded, wemight hatel amcwrrdations ani dspmvided, while another aircrafl, is ferried in. All of which entail "Me ani dissatisfaction. Iqengines are basically capable of being SMed mier adverse C- C- , but are inhibited by the inberent limitations Of the fuel ccnrtml schedules, i.e., not custsnu'zed to thme specific unhwine health c- . The adaptive system erployed on the" GE engines assuze that the engines will start (with sare sacrifice in start th) if ware- y capable of being SMed. The adaptive fea- qloyed effectively custsnu'ze the fuel and ignition schedules an3 ssymzes 50 as to match the specifics of '1 partidar qine pmblem. The des19 mnoept was Medto allow the aimft to rerrain in service unhil it could be scheduled into an w&ul facility where the abmmality could be mrrected. lhese same mncepts/featun?s also make the ccnrtml mre adaptable to airstart sibti-, and uare imprtantly (fm the stanapoint of this disxsion) make the -1 adaptive to the arctic engine starting unrirmrment. Allof this was amwplished with sensors ptwiwsly applied for other reasons.

1.1 Basis of this Pam

Rqine SMing systens are specifically custsnu'zed to the characteristics of each engine. Different CZMC ard GE engines qloy samnrhat different stratqies. "is paper represents a ccsnp3site of the different apprmches and is not inten363 to represent any one specific application.

-re, the systems arrently in use represent the current @mse of the evolutioMzy mture of this ccnoept. Sane of the apprmches disasa3 are not myin use as they are deem3 "not rqukecY at this tinrr, but have been tri&/develOpea, an3 are knwn to work, if so rquired, in respnse to a specific set of cinznmtances ascciated with a partidar engine.

1.2 General oaeration Ihe F'ADEC autcamted start sequerwe intqrates and autcmates all of the elemnts of the SM sequenoe in respsnse to a single aa"l fmthe nilot. 1. Initiates starter 2. Initiates Ignition 3. Initiates Fuel 4. "s off Ignition tation. 5. mInS Off starter 6. mxpm ?he Appmpriate Rlel Flar 7. p" ?he nppmpriate Start Bleed t/ Manual versus AutcaMtic The aircraft syst5n allcw for toth lnanual ani autorratic SM &. In the manual mde the nil& rmst -a rnrmber of switches. levers.

Much of the F'ADEC lcgic involves the pnper squeming and appropriate action to iMppmpriate pilot manipulation of the thmttle, aircraft fuel shutoff lever, etc. 1.0 X ON tcont‘d) 1.3.1 lhis pper concentrates on the unique fea- appropriate for starting engines * eriv~~am3 health carditions midl presented diffidt design dlicts/tradeoffs associated with amventioMl h” h” ‘Cal “1 systems. 1 me system is inm&to be adaptive to miax or special situaticm which are “ally not -, but if expriemxil, usually result in an aborted start. It is the intent of these fa- of the GE FADEc systw to circvrnvent the abnodity and still prcduce a suc~ssfulstart, ala- at a dlsacrifice to start th.

2.1 General -intion Of GITQIK~ Start Svstem Secnwce me following presents an overview of the sequenae of events am3 aaaptive strategy employ&. Details of this lcyic are demihed later in this paper. when a start is initiated the follcuing sequenoe of events is follcwed:

1. Starter Initiated

2. Ipition Activated At nFpmpriate RFFl (before fuel admission) 3. Iow starter Assist check -~~ForMini”starter ~ssistjust prior to fuel adnksion

4. mel Initiated At Desired Fhh3F?R4

5. For Limff

- ~esequenoeme3 and Izplition If Lightoff IS Not oetected 6. check For stall After Lightoff (am3 thmughout start) - sEs€qlm If stall Detected 7. check For Fan Rotation -fore Fed- Idle 8.checkForsMerAirpresflue~m

- warn of possible high speed starter resl-r~aemnt.

9. stater shutoff and Ignition Off at ’ m. 10. store In bmuzy Any corrective Action - If correctl‘on 011 Successive Series Of Starts, ~emwtersew Aril Initiate Folldng Starts Usirq stored Exprim 2.22 m me air restaxtnKde is Similar to the gmnd SM-, but is different in several asps.ts of its operaticn. 2.2.1 SMer: Different of Operatian are eoployea, ie., with or withak sbrter assist, upn oore wimbnil~imgrpn am3 other fa- ie., arch as:

a. Fliqht apzLiticsl b. stable or qxolirg &am c.starteraash-p- ‘ (interld) if requirdky starterlmnuf-. 2.2.2 Ipitim Initiation mint: Ignition is initiated at any rpn if the start is a wh&ii~lk start, but delayed until a ’ mininumqxnisaacfLievedifassistis emplW&. 2.0 ovmvnwmmm"m"mzm~~(ux@g

2.2.3 Fuel Initiation mint: Fuel is initiated at any wirCanilling RFM if the SM does not employ the starter, but is delayed until a mini" firing REM is achieved if the starter is tc be employed. m either wse, fuel is never initiated until the ipition is initiated.

2.2.4 Light off detection/resequencing and stall det5ztioq&zsqxmitq can be different fmgmurrl operation 2.2.5 Autamtic fuel shutoff philosophy, in the case of light off failure, ard stall -cry squexing can be different in the air versus on the gmurd depenlirrJ up~nthe airplan= "ufacbrsrs preferenoe.

-2.3 Arctic Cold Weather oesiqn Considerations

me arctic gmurrl starting sequence of events is th~,same as the mnml grcmn3 starting sequenoe, but "ly scwe of the design considerations of air starting as well as ground starting. 2.3.1 Both -at cold ambient tstpmtures mrdtions, ie., -40 to -6OOF. - Gmund starting can Q~QVwith bth mld an3 wazm fuel de- qmn the cold soak th. Air starting rarely mtersextrrmely mld fuel due to elevated oil cmler temperatures - "is nems that fuel vismsity can vaq over a wide range ard mnsquently the "bustion effects can vary over a 50%to 100% mnge - Wines can have radically different the& corditions depending upn tine since shutdm ard wkbnilling tine sine shutdmn.

2.3.2 me follcwing table shm the varigty of corditions that could exist at an engine inlet temperahlre of -65 F:

AIR-. MFIZLZ.". .mARrsIlwATIm

-65OF Very Cold Cold Start -65% Very Cold Refuel with wam Fuel iifter Cold soak

-65OF Warm IlesM After Aborted SM -65OF Warm I5eS.tal-t After Refueling -65OF w. Wann '-1 m Restart -65OF Ed. Cold Ekterx3d Win3nilling

inem 'colaracteristics~easc~assicald ard 2.3.3 % fasth-ion efficiently wirg apgmj"1y 98%. Eut, under arctic corditioxs, aar3xlstion cltaracteristics can vaq dmnntically as a fuxtion of fuel vismsity and its effect on fuel nozzle spray pattern. "is meaxs that: - Lightoff fuel Wtscan charqe by 50-100%. - After li@toff -ion efficiency can vary by 50%.

2.3.4 hl scwe erqines mnpressor stall can vary as a fllnction of campressor metal temperature histoly. 2.3.5 Oil viscmity increases erqh macharu'Cal drag, and dkap%xx asa Wonof Mal of revolutions. This oil vismus drag is not p-esent during air staaiq, M the effects of abxaff p~wer extraction drag can be dstantial duriq air stnrtiq prhdarly at low cwji7e inlet pressure flight dtions.

3.1 stall sensing

start -stall is=nsgi?y--dep*upcnthe stall si-- -&enstics of the qezific -me ard deperrlinguprm-wt.in3ernri"ettt (gmndvezs~s in-flight). 3.0 matZ,~lN2AND~(mnt Id)

3.1.1 Gravd Start Stall Sensing

c,round SM stall Sensing ccsnprisg tw ' t sensing systems, 4 d"/.r, ani a 'lslidhp m lh? 3.1.1.1 The dd'$& system senses a 5cdden dez33ss in are mrmzt&l rp acceleration rate and capares it to a stall ref- sch&ule as a fundion of corrected are speea. Stall Refertlnce Change in Schedule Core Acceleration Starter Cut Out Rate x< x< Idle Governing

ircraft Horsepower Aitraction

a. since engine acceleration towe is a "corrected" pmreter, the A dy&is biased as a . function of engine inlet P-. P-. b. me key to this ccnoept is to &bin an REH versus time sigml wh- mise content is dficiently low so that a geed first and seoond derivative can be &tad.

In this reqest, '%myhave tried but few have succeeded". ?he GE mEC qlloys a very sophisticates mthenntical appmch to signal acquisition ani prooessing which makes such an approach pible.

c. "is apprmch is applicable if the charaderistic of the stall reaction si@ is sufficiently strcng so as to distinpish between the other events which also prd~cesimilar signals, i.e., starter orrcut Ainraft Aocessory Lmd ppplication Idle RFM Govenmr Action d. "is concept can detect stall within 0.5 sec. of its ani thus quickly prevent the adverse sbsss/overtapram ozmnnly associated with stall.

3.1.1.2 EDT versus Cxxe Speea as a stall sem- si-.

me system as0 senses corrected ECT (EC7/0) ani it to a &&e of stall m v- mmmre m( ~tr (Referred to as the "Slidirq EGI'" limit]. a. zhis schedule is desiw to be himthan that erpmcted chlring l~~llpalSMing, tut, well Mar the tenpram a55ociated with a "hot start". b. me 0 fa- autcwtically bias the schedule for hot ard mld day mrditions.

C. "is reference d-eiule is further biased to a"t for anY residual EDT caused by previous qh Wry.

Hot Start Limit EGT "Sliding Limit" Corrected Exhaust Gas Temp.

EC7 4 b L Corrected Core Speed 3.0 STALLSEWIN3AwD-I mnt"

3.1.1.3 Other alternative approaches to stall sensing which mist of oxbiratiom of parameters could te enploy& axing groun3 xd/or air starting &ti- w?wre erqh idicsy-cnisies preclude the sole use of the A dd/& or the sli- ~cp:Limit approah. For -le:

- "ately hot E%p and very slow acceleraticm rate.

- low acceleration rate, "te Ecp, an3 elevated fuel sch€ciulimq.

-3.2 AirStartStallSenskq

Air Start Stall Sensiq is acowplished diffexfntly, ' vn the M- of the stall pinch pint.

3.2.1 Lightoff Stall

wing air starting lightoff stall, the core RIM can continue to increase due to wMlling assist or due to starter assist. Because of the altitcde effect on acceleration rate, the stall sigmture can be very low thus mering the A dq& -r inpractical. likewise, the sliding Ecp, refercsxe schedule (referenae ParagraFh 3.1.1.2) can be relatively lm due to the effects of DX/o at the cold ambient temperature -iated with high altitude, low aircraft rrachine aperation. 'mus the sliding Ecp limit my also be an inadequate stall indicator.

3.2.1.1 ?he high residual Ecp inmrsaiately follming an in-flimt fl-t, mupled wibi the lm Ecp reference frcan the sliding ECT limit, can &ims pmiuce an erroneous iniimtion of stall. In that case the added intelligexe pzrduced by abmnmlly low core acceleration rate can prwide the intelligence rquired to di- in-flight compressor stall.

3.2.2 "After Light-Off" Typ Stall

Ihe sensing of stall after light-off can he acsx,..plish& by sensing d dN/& as per paragraph 3.1.1.

However, in flight, this is amplimte3 when aircraft p3wer extraction is a@i& hecause airnaft plner extraction has a much -er impact on core acceleration at attitude than it does at SIS.

If such is the case. the OCBlceDt of the slidim ECT limit. biased by the resim m; and ql'&with low core-aderatibn rate, is an appropriate imlication of stall.

3.3 stall &coverv seuwmiq

3.3.1 Grunxi Stall Reowery Sequencing

If a gmund start stall is -, tbe follmiq sequence is enployed:

o Intermpt the fuel for a short duration in order to break the stall. mnthto a~plyignition and sMer assist. cnre ~pn may either amtime to accelerate, hang up or Merate denznairg upn the relatidp bet*sen unfired engim pmpirs toque and starter toqlE.

0 I5.+ertbeWF/FS3CJa fuel-eapre-determu7ed . anKunt. o Reinitiate fuel flm am3 dtor for liqht-off: 3.0 Sl!&LL8EM3IN3ANDVfcont'd)

3.3 stall Ret" 8pnarmeing I ccmt 'dl

3.3.1 rrand startlm' stall Reoovery~im (cont'd)

0 Ifthe?qxaisakwelwrmalstarteroRalt,~tothe starter crash enagement limits, ie., bo not re-eqage the starter if m > 20%, M allow qine to & dam to the safe starter -qemznt m. initiate Fuel flow o after liwt off is senss3, again mN- for stall. o If stall is again sensed, hterqk the fuel and further lcwer the fuel schaiule.

o If rquixd, repeat the stall ~~owe~ysequenoe for a pre.-detemind "rnrmber Of fuel -. If stall persists, abrt the start. Ihe pilot can then minitiate the -sequence. 3.3.1.1 ArRi m --up dm to dem"te3 fuel edmiule

If stall free acceleraticm is even+xally achiared, erg& acceleration with the (stall free) decremerrtsd fuel schedule can be sluggish, particularly in the -ion akove starter oR out m, or above aircraft load application ?qxa. If this coxus, the anti-hanq up feature is activate3 which then slowly rstwxs the fuel edmiule to its original value.

3.3.1.2 me canhination of the stall dez?xmnt stratqy and the anti-hanq up straw have the effect of repro3ramning the fuel schedule so as to "skirt aranrl' an unusual stall "bucket". 3.3.2 Airstart Stall Recovery sequenoe lbe airstart stall rec0Vex.y squalce employd can differ ciue to prefererme of the airfram "facture=. o one aircraft nmufacture= does not autaMtically resequenoe for any reason. - Wir Basic Flimt Safety FRilosophy is:

- If a stall is sense3, it is so miat& to the pilot and rquires pilot action to the system. - No adaptive fuel n%luction features are employed, hut fuel enrichwnt for lean fuel qxn hang-up is available. o Another aircraft manufacturer does enploy resequenCing fed- due to their philosophy to '- the System NlyAutcsnatic" and consistent with grcund "peration. 3.3.2.1 operatioral conflicting wquhamt

lbe followirq opemtional situations can be e"tere3 &img air starting.

-Ecp Accel Rate Ass"3 cause Corrective mion

very Any stall Resesuence werate merate No stall Not Rqui?.W

I"&iate law stall Resequence law law Raise Nel schedule 3.0 Sl!ALL SEWIS Aw tmnt 1dl

3.3.2.2 Air start stall sequerck Strategy employed by FADE is:

AfterlStStdll Interruptfkl, lower fuel schedule

After 2rd stall Repeat first resequence (with additiond fuel reiiuction)

After 3rd stall

Suocessive stalls continue to resqwwe per 2rd stall

3.3.2.3 Anti-hanqul, feature in mniunction with stall ressm€Xing

If the fuel dKxar2nt fluresSful1y clears the stall, an3 suksqumt lean m --up is -, the fuel schedule is SlWlY inrreased until such pint where sufficient accel rate is established. 3.4 €&Wive Stratezw For mkWith A Relx?d ted Ground SM Stall

cepending upon the prefmof the ablenwufactun?x, adaptive fea- wn be added which shortens the nlrmber of resequencing events requir& for each start. feam is inten%d for use in c"srcial aircraft until the aircraft can be scheduled into an appmpriate arerhul station. ixXae%Y Each tim? an engine is requir& to be resequencEd &e to stall, the FAD= Logic.

a. Determines the nmker of resequence tries nmfor SU-. b. Stores this in n"?.y. c. hWm a suksquent start is initiated the rnnrber of tries is recalledardthenap~p~decisionismade~the fuel schedule decranent to start out with.

several neans of sensing lightoff have teen e!pI.oycd an3 are SaEWhat different between the two "GE" engine p7xqraxs. 4.1.1 Seme a sudden, alnast itaease in engine acceleration rate. - since a sudden, almst instantanxm increase in engine ameleraticm rate m at lightoff, the samt! A Ju/d4 signal used in stall sensing can be employed. lhis cwncqt 1s applicable to bth an3 air stating which bth prcduce appruximtely the saw signal level.

4.1.2 Seme a significant an3 sudden itaease in m. - a sxx%ien junp in m *e the residul m level. me -system the~Xoredetemirs the reskiual at the initiation of the start sequerre an3 then 1- for significant m increases above the resichml valw. Ihe GE asnnercial engines emplay ignition smrcs for both excitnrs and ignitors.

0 Normally, for lqeviiq RaSKn3, only me ignitor fires &ring a given start. TIE system alternates the excitor an3 ignitor pair after each suooessful start.

o after the fuel is initiated, the ligilt off detector c-t imNtors for liqht-f f. - If light-off is not detected within a preaetenained time, the system shuts Off the fuel, motors the engine then fires both ignitors. - me system then reinitiates fuel and mNtors for light-off. o If failure to detect YO is again encountered, the fuel off, notor, and fuel on sequence is (using both ignitors) .

Tnis sql€x.x is repeated several tiES dep?&uq' pairplane mnufactxrer preference) after which the system declares a "false start" and the start terminated. o -use longer than noma light off delays can mutinely &ring air start- &ring arctic starting, additional -"th and to light off" is alldunder these c-w.

4.2.1.1 me GE -ial engines do not require additional light off fuel flnr to for light off failure. However, other alte?xatives auld easily be applied deperding Upn, the specific light off characteristics:

- pdditional light off fuel flow auld be employed on each wivestart by: a. pdditiona3 incremental step in "fuel flnr b. Th?2mp hm2sass "I fuel flow after each YO

4.3 Cold c,rourd Starth Liuht Off Reseauencinq

o !&primto date on both cxxmzcial engines has shown that the elevated f~elassociated with the nom (r/~/B3@'sched~ling along with the increased allcwable light off delay this adequate when used with the reseqym describe3 in paragraph 4.2.1.

o mre sophisticated lqic circuits employing incrpmental imrases in light off fuel are p?xq"& into FADE for futare use, hut aredlsanwd at this th. If rquhxcl at a future date these strategies auld be easily activated.

4.4 Air Starth Liuht Off Resesuencins Ligilt off res0pnzing can differ between ensine Pro3rans ale air start liqht off characteristic diffw

o -fly neither GE engines ennploy any adaptive air start light off fea-. - air starting can be an emergency sidtion, bth ignitors are Mately employes. - muse of the wim?nilling airflow effect, no fuel pursing is w. - If light off is not deteded within 20 seconds, limt off failure -iated and the q&en relies on pilot action to resequence- 4.0 LI”0m s!iW3m (mntldl

4.4.1 other adaptive feabxes that carld be employed At extrem flight cwditions, air start light off can, on sans engines, involve design mnflicts be- the rewinmmts for elevated light off fuel flow and the -t lower fuel rquird duetothe--rstallchnracteristics~tdevelopafter flame pmpagation. The ability to detect light off within 0.5 seconds (by the use of the delta ameleration rate sensor) nwks

5.0 FUEL In the past, the design of fuel &€&ling assmiated with h”33mJU “1 control systens has been a ccanprcwise between different conflicting cm- because

1.441~~3eN versus &!G was used as the fuel sing primuy stall avoidance pamwter.

SinOe W P is not a direct sensor of --r stall, it is subjectd to3~er influences.

- combustion efficiency variations due to fuel viscosity, altitude ~~essure.emine netal etc. . ~~ .1 temoerature. - Varying ccxpmr smll &racreristics ptTduced by crqk themdl history ssmkwiffi. - ‘Ihermodynanc cycle variations due to ran prcEsure racio ctf& (i.e., &ring air starting). Unusual ambinations of thq situations can ccar at cold air temperatures, ie., below -20 F ambient. (reference. parasraph 2.3.2).

mus, scheduling W P as a function of core qmand engine inlet temperature is msdficient control intelligexe to do the opti“ job that is p3ssible with digital controls.

5.1 FADM: General mmch

me mrrp3uting and scheduling flevibility of the FADE 1l0w mkes it possible to do a better job of bringing all of these elements tagether. lhus pramting a better mre reliable sy.st5n mnoept, as follows:

1. &sign for the nom1 (or rcutine) situation Mcountered during aperation in the three major emrirorrments. - NomlGroundStart - AzrticGroundStart - Airstart+

2. Use special override features uder non-norm1 circumstances. 3. Where appropriate, or whexe design conflicts still exists, or where stall risk still exists, use the stall sens~/resquaing,or failed lightoff resquexing and anti-hang up features to override the fuel scheauling features.

Thus, the fuel bemmes very foqiving by adopting the philosophy of &@&ling the prune paamster, biasing it for additioral Cinwstances, and then if all else fails, using the lightoff/stall resequencing as a final authority override.

5.2 Ground Stall StrateqY mmnl ambient teiw ranael

1. use conventioml dp/p.I~”fuel &me coupled with a fixed mini” fuel flow as the primary liqhtoff fuel criteria and dur+ the initial RFFI region. 2. nbwe this use aoceleration rate versus corrected core speed as the prime de. 3. hrerride the acceleration rate scheduling with the WF/FS3 dverses H& topping sch&ule, design33 prkily as a stall protection cimuit. 4. Wign the topping schedule per 6OoF day an3 we mmal d A40 , factors to ccanpensate for ambient pressur= ard temperature variations. 5.0 FUEL E5m"x FEmmSs (cont'dl 5.2 G& Stall ShtKN 1mnral anbient t" ramel (mnt'd) damwanl to -te for mn-mmd effects of

- scme engines have a dep- m"preSs0I stall line when the engine is at an int"?& 'amy thdetions (ie., not mld, not hot, but warm). 5.3 Arctic Ground Start Stxateqy 5.3.1 resign the stall pmtection portion of the fuel s&&uling cinvit hi- at very mld teparatmx days, i.e., arctic mdtions based on thepr- 'on efficiency, associated with mld anbient and mld fuel. Assume the engine is therually stable at mld sm)r mdtions.

Arctic Acceleration Fuel Temperature Schedule Corrected WFlPS3 Ambient Temperature

Corrected Core Speed

5.3.2 Provide an additioral bias over-ride features so that the base schechlle is modifisd dcwmw3 to aomunt for the residual qine therrral effects Of recent engine nlnnilq such as: - Increased ccanbustion efficiency asuziated with warm fuel PIS&& by a residual of warm oil in the oil cooler. - hgine lhalml Histoly The signals associated with establishing the engine thdcondition ard oil tapram, ana/or fuel tapratwe are available fmother portions of the FADE system.

5.3.3 Increased lightoff fuel flows are usually requird, ard are subject to the same variety of envirornrrntal and operatioral mnsiderations defined in pamgraph 5.0. Tne details of these requirements are to be a function of the specific engine fuel injedion system design.

a. GE FADM: systens have been successful with the hcreasd lightoff fuel flow p?xciuc& by the raised "=-e., nowever, other engines my need additioral lY clrullts. (referenoe. paragraph 5.3.2).

b. meSe "pensating circuits can take varims form.

- Raise "I fuel flow as a function of fuel oil mler tapra- or other in3ication of fuel viscasity. - Ramp incrsase mini" fuel flar as a function of time (only with very mld fuel wrxlitions) . -Ramp incrsase, OT time step imrease themin. fuel flm lxlt MaWyreduce it to a lcuer level (i.e., within 1/4 - 1/2 secord after limbff 50 as be a"a*with stall-free aperation.

c. There are a variety of other S&G%ES that muld be employed dep-db qxm the specific idicsyraasies of the engine design. 5.0 ETJEL ---(mnt'd) 5.3.4 puite often, the precise mnkustion efficiency effects of fuel twpemture an3 air tapra'axe are pmrly defhd. In such situations the following strategy can be adopted: - Error in the -ion of rich fuel scheauliq - If stall is €zxo.m-, employ the stall serc;ing an3 resequencing feature

L Enploy the air start anti-hangup cckoept if nc-. - mich the fuel if the E6T is cml an3 rpn acceleration rate is slow.

5.4 Air start stra*

5.4.1 ?he saxe mnflictirq factors miate3 with arctic gmund starting exist in the air start region, but in different canbinatim: - Cmhstion efficiency during air startiq is usually "1- to that asscciated with mrmal qrwn3 starting than it is to arctlc gmund starting Wen th- Similar engine inlet are encountered. - Cold fuel tenpxa'axe has a mter influence on ccsnbustion efficiency than does cold air temperature - Fuel -ture is determined by oil cooler temperature which is always wa~mduring air starting corditions as contrasted with cold oil during arctic grcunj statinq. - The fuel schedule design straw employe3 is to design for the arctic si-tion which infers a higher function which lowers the schgfule as a inlicative of fuel tapzrabre Wor flight cordition.

5.4.2 Lightoff is generally the mst stringent constraint during air starting.

- Lightoff fuel flm is generally set by the mini" fuel available at the special flight cordition. - F'ADE pmvides a mqe of "I fuel flows a% a function of flight codtion. to arhiwe pmper

flame p-qation. Fhysical min WF can excessively high corrected fuel due to the effects of engine inlet pressure, thus reducing lightoff stall nayin. - me ~-11% effect of aircmft ram pressure ratio (PYP ) has the Wficial effect of raising the apparent corrected fuel fl&r -red for stall an3 helps mapemsate for decreasej. lightoff stall ". 5.4.2.1 The GE FADK: establishes lightoff fuel flow by the limiting item of one Of trm feab.lJx5: .. a. Muu" fuel flow &e%led as function of wine flight conlitions

b. Mini" €53 multiplier flmr schedule. - me pr- control lcgic - Conventional value of the P multiplier cinvit called the flmr schedule. - mis Sam2 "&t is employed by FmKC.

PS3 Multiplier Schedule Bias with PS3 Flight Conditio Multiplier Floor Schedule

I Sensed Burner Pressure t'd -1 -1 - FADEC provides the capability to manipulate the flwr schedule and thus liqhtoff fuel flow. ?he GE cx"ial erqines anwkly do not a a bias of the flwr SdIEiUle, M FADE is capable of doing so, if rqim3 in other applications. 5.4.3 .me corrected fuel flow W+/SS" mquird because of liqhtoff atcwization mnsideratims (flame propagation) my be exoessive after lightoff due to stall considentiom. &cause of the availability of the extmmdy rapid adon liqht-off Sensar, it is possible to rapidly reduce liqhtoff fuel flow after liqhtoff ht before stall is developed. mis fea- is not currently employel, M available for activation, if rquire3 by a specific engine.

5.4.4 slar acceleration or "RRI hangup" can oscul in the air start -ion ciue to two factors:

1. IMdequate fuel hecause of: - Aircmft power extraction effects - In-exact knmle2ge of the rqim3 fuel whicb can axlv at flight extremes and/or atscure dtions.

2. Ccwpressorstall

5.4.5 mti-hang up straw

5.4.5.1 If slow aderation is encountered, feab.mZ5 can be employed that detc" mether the slow aderation is &e t0 i~dquatefuel flar or ampressor stall.

5.4.5.2 If the slar ac€xzleration is prcrhd by low fuel flar, the anti-hangup circvit slowly increasg fuel flow, werridinq the other fuel ccanputations, until a "I mre RFM accel rate is prcducea.

5.4.5.3 If the rpn hangup is caused by ampressor stall, the adaptive stall remvery stratqy is activated.

m: mis feature has the potential for significantly increasing the vnassista3 portion of the air start envelqe. 6.0 StartarUSaqeStrataWaunnJ. Airstarting Air starting is usually divided into two regions: wirchnilliq an3 starter assisted.

0 SMer assist is required when WiIKinilliIq rpis less than 10-20% depdinq upcn the engine. If Wirchnilling RRI is low, the starter is activated inmdiately, M icplition and fuel are delay& until the mine aozlerats to a safer firing RPM. o If wirdnilling RFM is greater than zO%, SMer assist is not employed. o The siixation is further amplicated by the engines state and the starter energy supply availability when the start is initiated

- The starter my or my not be available dep"g' upon the status of its energy supply. - ?he engine can be wi"illing (stabilized) or spooling darn dep"g' how som after shutdoan that the start is initiated. - In either case, wirchnilling v- spool darn restart, the lcgic does not apply starter assist unless the ?.pn is below the safe starter Wtm. PI-01 11-1

COLD START INVESTIGATION OF AN APU WITH ANNULAR COMBUSTOR AND FUEL VAPORIZERS

K.H. Collin Thermodynamics and Performance Department K. Pie1 Consultant KHD Luftfahrtteehnik GmbH D 6370 Oberursel, Germany

Summary

The paper deals with the combustor of the APU for the "Tornado" fighter aircraft. As this APU has to cope with the narrow space in the fuselage it must be of small size. An annular combustor is Pavourable as it is short and can be integrated into the envelope of the outer diameter. The fuel vaporizer system is chosen because of its great advantages with combustion.

The paper describes the ignition process which is difficult as no fuel is actually vaporized when the start is initiated. Theoretical background and experimental steps of a development programme are reported. The result was perfect starting of this system down to -40 OC and a very high "First Start Reliability" which means no false start leading to several start procedures.

1. Introduction

An APU (Auxiliary Power Unit) is installed in an aircraft to provide secondary energy for one or more of the following tasks.

- On ground an APU has to deliver shaft energy for electrical generators, for hydraulic and fuel systems. In addition it has to deliver bleed air for starting the main engine as well as for air-conditioning.

- As soon as the aircraft is airborne some APU's have to be able to work as standby emergency power sources.

- With the initiation of the landing procedure the APU may again - with some applications in case of emergency it has to - take over all secondary energy in order to re- lieve the main engine of secondary loads.

Several facts have influence on the operation of an APU. The main aspects are:

- Design of the APU

- Design of the secondary power system of the aircraft

- Operational requirements for the APU

- Requirements for special types of fuel and oil

- Influence of ambient climate conditions and altitude.

Regarding low temperature environment conditions we have to consider two phases of operation of an APU

- Starting the APU even at extreme low temperature conditions

- Running the APU 11-2

During the Prototype-Testing of this aircraft,the first phase - starting the APU at low temperature conditions - showed that sometimes two or three start attemps had to be conducted before a successful start was reached. For Production aircraft this was contrary to the user's requirements which stipulated:

- high start reliability which meant no false starts leading to several start procedures: "First start re1 iabi 1 i ty".

- no special preparations before starting like

. pre-cranking of the rotor . pre-heat(ng of the oil . pre-pressurization of fuel

- short starting and run up time

For the second phase - running the APU at low temperatures after successful starting - no problems have been experienced: as the air intake of the APU T 312 is located under the body of the "Tornado" aircraft and the opening with the hydraulic operated flap is showing to the ground, there is no danger with ingestion of hail and only small amounts of rain and snow enter the APU. Experience has proved perfect running of the APU. There- fore this Paper deals with the development and experience connected with the first phase of APU operation, i.e. starting at low temperature environments.

2. Auxiliary Power Unit (APU) T 312

2.1 General Description of the APU

The requirements of the secondary power system of the aircraft,which are important for defining start and operation of the APU are:

- small size and weight of the APU

- limited battery capacity in the aircraft

- short starting time for immediate aircraft star

- high starting reliability: "First start reliability"

- same types of fuel and oil as for the aircraft

- imediate delivery of oil with high pressure to a dry friction clutch connecting APU and the gearbox of the aircraft

- no in flight operation required.

For ground operation such as pre- and aft-flight check of the aircraft systems, for stand-by and for starting the main engines via a hydraulic torque converter, the APU delivers the necessary energy as mechanical shaft power. This is transmitted into the gearbox by the clutch which will be automatically engaged as soon as the APU has nearly reached its 100 % speed after start up. To compliment these design features t.he APU T 312 has been defined as a single shaft gas turbine engine. Besides other characteristics, for example the constant output speed, such a single shaft APU has the great advantage of low weight and small size.

Fig.1 shows a cross-section with the main components: A two stage axial-radial compressor, reverse flow annular combustor, two stage turbine, exhaust duct and a planetary type gearing which reduces the output speed to 8.000 rpm. The APU has its own oil pumps, which deliver oil for lubricating the bearings and gearing as well as supply high pressure oil to the clutch. The oil reservoir is in the aircraft mounted gearbox to which the APU is connected. The fuel supply is controlled by a hydraulic governor. The starting sequence as well as overload and overspeed protection are achieved by an integrated Electronic Control Unit. 11-3

Dimensions of the APU are: Overall length is 510 nun, including all accessories the diameter is 380 mm. The rated shaftpower is 105 kW for continuous running with a 10 % short time contingency power. The ratio of rated power to weight is 2.7. [I]

2.2 Description of the Combustor DesiRn

AS the combustor of the APU is of great importance to the starting procedure, it will be depicted here in more detail. The combustor, see Fig.3 , is of the reverse flow type. This type allows short engine dimensions.For small gasturbines like the APU T 312 an evenly spray pattern of the fuel cannot be reached with atomizer nozzles since a considerable amount of fuel will be spilled onto the walls of the combustor. Fan sprayers as they were used with some gasturbines may be one solution. At a number of other gastur- bine manufacturers greater experience, however, were gained with vaporizer nozzles. The- refore this type of fuel injection was chosen.

Advantages of vaporizer nozzles:

- low fuel pressure system

- excellent combustion because of pre-vaporization of fuel in the vaporizer canes by hejt transfer from the primary zone in the combustor

- good maintainability

- low cost

A serious disadvantage, however, exists:

- bad starting at low temperatures and in altitude. The reason for this is poor or even no fuel vaporization at these conditions.

As can be seen in Fig. 2 the dome of the reverse flow annular combustor carries 12 vaporizer nozzles and 3 starter nozzles in 4''. 8'' and 12'' position. The compressor outlet air enters the combustor through bores directly. For film cooling of the walls an air-layer is directed through rows of small holes combined with deflector rings . After an elbow bend of 180 ' the stream of hot gasses leaves the combustor and enters into the first turbine nozzle. Fuel is injected into the I2 vaporizer nozzles through thin pipes. During the starting period of the APU ignition is initiated by three ignition nozzles, i.e. one torch igniter in 12O0 position and two stabilizer nozzles in 4'' and 8'' position.

A torch igniter is shown in Fig.4.It consists of an atomizing swirl type spray nozzle combined with an electrical high energy spark plug. An oval shielding tube reaches through the turbine housing into the wal1,of the combustor. The ignition spray nozzle and the igniter plug are assembled on the shielding. A Stabilizer nozzle consists of an atomizing swirl type spray nozzle only and is shown in Fig.5. The design is similar to the torch igniter.

3- Theoretical Analysis of Starting

3.1 Balance of Torques

The cold starting condition depend upon the "balance of energies" which prevails during the starting phases.

The balance of torques is described as:

Driving Torques - Drag Torques = Acceleration Torque

In Fig. 6 and 7 the torques on the turbine shaft versus turbine speed are shown for the early design of the APU with only one ignition nozzle, i.e. a torch igniter at 8'' position. 11-4

Curve 1 : torque of starter motor Curve 2 : turbine torque Curve 3 : acceleration torque

Curve 2 is the combination of all drag torques and - after the combustor has lighted - the sustaining torque of the turbine wheels. The acceleration torque, curve 3 reveals a great difference between normal temperatures, 115 'C and lower ones, e.g. -26 OC. At 115 OC ambient temperatures there is sufficient surplus torque for acceleration (0.8 Nm). At -26 'C this surplus torque is already very 1itl:le (0.1 Nm), At temperatures below -26 "C there will be no acceleration torque, that means a "hanging start".

The elements which influence the balance of the torques are following:

Driving Torques

- Starter motor This torque is dependent on the size of the starter motor and capacity and charge of the battery which is aboard the aircraft. Both their sizes, i.e. of starter and battery, are limited by the overall design and weight limits.

- Combustion of fuel Tests have shown that a completely "lighted up" combustor will result in a sufficient energy release. However with decreasing temperatures light up procedures of the combustor becomes more and more difficult. Drag Torques

- Ventilation of compressor and turbine These losses cannot be reduced once the basic design has been frozen. All the worse, with decreasing temperatures the ventilation losses of the compressor increa8es because of higher density and mass flow of the air due to the laws of similarity ( N / T ) and (m T /P).

- Bearings These losses can definitely be reduced by use of ball and roller bearings instead of sleeve bearings. In the APU all bearings with the exception of pump bearings are ball or roller bearings,compared with sleeve bearings their losses raise with increasing viscosity less.

- Sealings It was already the intention with the basic design to minimize losses by use of labyrinth-sealings wherever possible. With decreasing temperatures, however, a certain shrinking as well as in other places rubber sealings has been considered.

- Gears Here the losses are negligible, with lower temperature they may increase a little because of the oil viscosity.

- Fuel pump With the choice of vaporizer nozzles a low pressure fuel system already results in rather small losses. Even for spray nozzles like the igniter and stabilizer nozzles the pressure need not be higher, especially as the flow range is limited. to the starting phase. The vaporizer system including the igniter and stabilizer nozzles requires only 15 bar pressure compared with approximately 60 bar for a fully swirl type system.

- Oil pumps Here the viscosity influence is significant. The losses are proportional to the delivery pressure which increases when flow resistance raises wirh higher vjscosity, Therefore deloading the pressure oil pumps would be helpful. 11-5

3.2 Combustion Techniques

3.2.1 General Theories

During the start - especially at low tempertures - the sustaining torque of the combustion energy plays the important roll. The aim must be an immediate ignition and a high efficiency of the combustor.

The ignition of combustors with a vaporizer system is initiated by a torch igniter and sometimes stabilizer nozzles. Combustion efficiency of this type of combustor is influenced by the quality of atomization.

Another influence is the type of fuel. As soon as the torch igniter and stabilizer nozzles are burning the vaporizer nozzles should also start burning. Below a certain temperature the heat for evaporation has to be induced by the torch igniter and stabilizer nozzles.

Another influence of the ambient temperature in connection with the fuel/air ratio of the APU is shown in Fig.8.The turbine speed represents the amount of air delivered from the compressor. Here again the inflammability of fuel depends on the torch igniter and stabilizer nozzles.

3.2.2 Combustion Techniques of APU T 312

Applying the general principles of the ignition theories to the APU T 312 it can be said that reliable starting at low temperatures requires early ignition of all three igniter nozzles to warm up the vaporizer nozzles and inflame there as much fuel as possible. The envelope which allows combustion, however, is limited by local conditions of lean or rich fuel/air ratio. The total fuel flow to the ignition and vaporizer nozzles is scheduled by the fuel control dependent on the CDP (compressor discharge pressure). This fuel flow is shown in Fig. 9. It can be seen that at a CDP up to 1.5 bar a constant quantity of 24 l/hr is scheduled. It shall be noted, that an engine speed is assigned to the compressor discharge pressure 1.5 bar, which depends on the compressor inlet temperature and corresponds to the following values:

Compressor inlet temperature To LOCI -26 +25 Engine speed N [%I 44 51

The scheduled fuel of 24 l/hr is split by means of a flow/ pressure divider within the fuel control unit into 14 l/hr to the ignition nozzles and 10 lihr to the vaporizer nozzles. Furthermore it has been found by testing that some of the fuel injected through the ignition nozzles is sprayed on the walls of the combustor as can be seen from Fig.10. Calculations have been performed to find out the fuel/air ratio. Only those inlet openings in the combustor which are in the spray area contribute air to the fuel of the ignition nozzles. This air together with the fuel results in a local fuei/air ratio which is dependent on turbine rotor speed. These ratios have been calcula.ted as well as the fuel/air ratio in the vaporizer nozzles. Results are represented in Fig.11. Curve "A" reflects the conditions in the vaporizer nozzles, but this curve does not apply to the first phase of the starting cycle, because the fuel is not vaporized yet. Curves "B" and "C" are related to the ignition fuel nozzles. It shall be noticed, that with insufficient fuel/air mixture, caused by wetting of the combustor walls, the combustible range is reduced by leaning-out.

In summary it can be said that for successful starting it is necessary to reduce the drag torques and to increase the driving torques: Therefore two tasks have to be solved to guarantee at low tempertures reliable starting of this APU with the vaporizer system in the combustor: - Improvement of good and complete light up of the combustor during the early starting phase.

- Deloading the pressure oil pump as long as possible but safeguarding lubrication to the bearings. 11-6

Some essential markings have been entered in Fig.11.

1) Range, in which the maximum fuel pressure at the ignition fuel nozzles and the nominal fuel flow is reached,

2) Maximum cranking speed without ignition

3) Switching to fuel control algorithm and end of range in which fuel quantity is constant

4) Shut down of ignition fuel nozzles

5) Combustion range to be expected at homogeneous fuel/air mixture.

6) Cut out of starter motor and electrical ignition

It can be seen from Fig.11 that exceeding a certain speed without a pilot flame has been ignited (because of late fuel supply etc.) the fuel/air mixture fails to be ignited due to leaning-out. So it shall be considered as essential, that either

- the maximum fuel presssure for the ignition nozzles will be reached at low speed, or

- the local fuel/air ratio of the ignition nozzle at 12" position will be increased.

4. Test Results

The tests had to cover the whole operational range of the AFU, which are:

Ambient temperatures -40 OC up to t52 OC, Fuel temperatures -40 OC up to +IO OC (in some cases up tot90 "C) Type of fuel: AVTUR (Kerosine type, Jet A-1, F 35) AVTAG (Wide range, JP 4, F 40) Type of oil: Synthetic lubricating oil, 5 CSt, MIL-L-23699

Main emphasis had been given to testing the starting ability at low temperature environment conditions. AS an APU cannot get any easment preceding a start -e.g. pre-cranking- test conditions had to be assessed as realistic as possible. This meant for each start at low temperature that the APU had to be soaked long enough these ended up in rather long cold-soaking-periods, at -40 'C, this were 3 hrs minimum.

On the other hand all design changes or changes in adjustments, e.g. change of fuel setting, had also to be tested at high temperature environment to ensure that no detrimental effect could occur. As an example: influence of different fuel settings in regard to compressor surge or stall.

Besides the above mentioned limitations, for all test procedures and subsequent design changes the basis was defined by the analysis as explained in the preceding chapters.

Low parastic losses, i.e. low drag torques during start up.

Optimization of the combustor, i.e. correct fuel/air ratio for igniter and vaporizer nozzles

In order to give a condensed information two test configurations in the sequence of development will be presented together with their design status.

1. First experimental design with only one igniter nozzle

2. Production standard design with three igniter nozzles 11-7

4.1 Optimization of Drag Torques

It was already reported in chapter 3.1 that there could not be a reduction of the drag torques.Except the oil pump torque could be reduced by deloading the oil pressure.

According to the system specification the APU had to deliver an oil flow with a pressure 30 bar in order to serve the pressure for the elutch.This clutch is necessary because a single shaft APU cannot he loaded during the start.

The viscosity of the specified type of oil, MIL-L-23699, increases by a factor of 150 when the temperature decreases from 115 OC to -40 'C. As a result of this the drag increases.Fig.12 displays the rotor drag lines for both temperatures. The performance of the electric starter motor is nearly Constant over this temperature range resulting in steady-state cranking speeds of 3 % at -40 "C and 20 % at 115 'C. Due to the APU control system the fuel does not begin to flow before 8% speed is reached. Therefore additional steps had to be taken to safeguard APU light up with temperatures down to -40 'C. Tests had shown that the oil pressure of 30 bar already builds up at 2 % speed,because the pipes had to fill up. Therefore a device was developed to deload the pump during the starting phase, still ensuring full pressure in time for clutch engagement. During the whole procedure, however, a level of pressure is maintained in order to guarantee the lubrication of the APU. With the modified oil system the oil pump is deloaded during the starting phase by a valve which is electrically switched to "load conditions" at 48 % APU speed. The result will be a greater acceleration torque on the turbine shaft as is shown in Fig.13 (comparison with Fig. 7). With earlier and complete light up of the combustor the ascending part of curve 2 (Turbine Torque) shifts to the left, curve 2a. The accelerating torque, curve 3 raises to 3a. With this a reliable start and shorter starting time will be reached.

4.2 Optimization of the Combustor

It was explained that reliable starting at low temperatures requires early - i.e. at low APU speed - ignition of all three igniter nozzles. These will then warm up the vaporizer nozzle and inflame the fuel.

Most aggravated conditions were chosen for the tests in order to cover all environmental and operational conditions.

- for cold and normal temperatures lowest battery charge,l Volt/35 Amps,i.e. min power supply to the starter motor,

- for high temperature max. battery charge,l Volt/?O Amps.

- Fuel pressure head 0,019 bar representing lowest level in aircraft tank.

- Fuel type . Jet A-1 for cold and normal temperatures . JP4 for high temperatures.

These fuels were selected for the different temperature ranges because of the data of volatility and vapor-pressure. The underlined values in the following table indicate the most adverse properties:

Begin of volatility vapor pressure

Jet A-1 0,04 bar

JP 4 + 66O 0.64 bar 11-8

Jet A-I has worse condition for starting the engine at cold ambient temperature because there is a large difference in fuel and volatility temperature. JP4 is critcal at high ambient temperature because vapour pressure is in the same order of suction pressure of the fuel pump.

For some tests 6 thermocouple-probes had been installed in the combustor as is shown in Fig.14. With this arrangement the light up sequence in the combustor could be well assessed.

TL1

Test conditions:

Ignition: only torch igniter at 8'' position Ambient Temperature: -40 'C, 3 hrs soaked Fuel: Jet A-I Battery: min charge

The test traces, Fig.15 show temperatures of the 6 probes TC1...6 and their position in the combustor, average exhaust gas temperature TE and turbine speed N.

Discussion of test result.

The turbine speed raises very slowly: after 25 seconds the speed is only 22 %, although TE is already 900 "C. The temperature indications of the six probes, however, show the actual situation within the combustor. Initial ignition is after 1.5 seconds. Looking at the positions of the torch igniter 8'' and the six probes (below right in Fig.15) we see that probes 1, 2 and 3 indicate immediate increase, 1 and 3 stagnate, 2 and 6 climb furtheron. For probe 1 there is no full inflamation, because the spray angle of the 8'' nozzle is too narrow: the vaporizer is not heated up enough.

Probes 1, 3, 4 and 5 are very slow, slowest is 5: it does not reach a temperature increase of 100 "C before 38.5 seconds. This figure has been taken as an indicator: Ignition Delay Time 100°: IDT 100 = 38.5 seconds.The explanation for the increase of TE after 25 seconds can also be clearly explained: fuel is not burning within the combustor but is atomized lateron by turbine disks and burns in the exhaust duct where TE probes are positioned. The analysis of this test showed that the initial flame of 8'' torch igniter did not heat up the far off side (3, 4, 5) of the combustor, air/fuel ratio of homogenous mixture was very lean. Therefore first start reliability was poor. Once the combustion chamber had warmed up with the first start, the second or third start was 25 seconds.

The conclusion of this test was:

Three igniter nozzles-instead of one equally spaced at ?OO, 80°, lZoo with the torch igniter at 12'' position because of maintenance reasons

Test conditions:

Ignition: 3 starter nozzles Ambient Temperature: +I0 'C Fuel : Jet A-1 Battery : min charge

This test series was now conducted under normal ambient temperatures to find out the effect of the three igniter nozzles. 11-9

The test traces Fig.16 shbw the temperature indications of the six probes (note different positioning compared with test I!). The delay time of probe 5: IDT 100 = 11 seconds.

The start up time of 24.5 seconds as well as "First Start Reliability" was still marg i na 1.

The analysis of this test showed that the air/fuel ratio of the igniter nozzles were not sufficient. Therefore a rather wide test series for optimizing these conditions were performed. The result was:

- Increase of the fuel flow to the igniter nozzles

- Change of arrangement of the spray nozzle and the igniter plug of the 12" torch igniter.

The design features were already shown in Fig.4.

Test3

Test conditions:

Ignition: 3 starter nozzles, optimized flow and optimized arrangement Ambient Temperature: +15 'C Fuel : Jet A-1 Battery: min charge

The test traces Fig.11 show much faster ignition and complete light up of the combustor. The delay time, probe 6, is IDT 100 = 6 seconds, start up time is 12.5 seconds. A very important characteristic of this optimized design was, that also at max. dry cranking speed (20 %) ignition was possible.

With this configuration (Test 3) cold and hot chamber tests were performed. Since these tests were "Official Qualification Tests" the APUs were not equipped with the six probes in the combustor. Therefore only the general results can be reported.

-4 -4

Test Conditions

Ambient Temp. Fuel Battery Start time to charge 100% speed Test 4.1 -40 'C Jet A-1 min 24 seconds Test 4.2 110 OC JP 4 max 12 seconds

Test 4.2 was necessary to proof successful operation at high tempertures. All test results of the complete programme are compiled in Fig.lE.First start reliability including all endurance tests was in the range of 98 %. This was achieved mainly to the fact that even with late fuel supply, e.g. because of air bubbles in fuel supply line,the ignition at max. dry cranking speed was successful. With normal fuel conditions the specified max start up time of 30 seconds was also achieved.

5. Conclusion

A combustor with a fuel vaporizer system cannot be started without an auxiliary ignition system because there is no vaporization during the starting phase. With the APU for the "Tornado" fighter aircraft a programme was initiated for a thorough improvement of the start capability at low temperature environment. Theoretical analysis combined with a test programme resulted in successful starts of the APU down to -40 'C. Besides this the user's requirements for "First Start Reliability" could be met.

References

(11 K Piel: Experience with the KHD APU T312 for a modern Fighter Aircraft, AGARD Conference Proceedings No.324 Engine Handling. 11-10

Fig.1 APU T312 Cross section

TORCH IGNITER 12"

STABILIZER NOZZLE 8''

Fig2 Dome of combustor 11-11 =rHW Fig.5 Stabilizer nozzle

I- 3 0 c a V a W c a

50 60

U TURBINE SPEED(%) n 2

Fig.6 APU Start at 15'C - E8 z v

w -4z m X 3 c z2 0 w

TURBINE SPEED(%) -2

Fig.7 APU Start at -26'C

/' I / / A 10-1- - AMBIENT TEMPERATURE - 288(K) -0

4 1 51 COMBUSTION 2ANGE AT _J iy 233110 53 HOMOGENEOUS U 'i FUEL/AIR MIXTURE / i 0 10 20 30 43 jn 60 70 EO TURBINE SPEED (%)

Fig.8 Fuel/air ratio at the vaporizer nozzles 7 0 1: + 2 3 4 5 Compressor Discharge Pressure (bar1

Fig9 Fuel control schedule

\- IGNITION FUEL NOZZLE

Fig.10 Dome of combustor spay area of ignition nozzle 12''' 11-15

ROTOR SPEED I

Fig.11 Ignition investigation

__------APU ROTOR OILPUMP DRAG

LOADED

-40°C STARTER PERFORMANCE

W V z 4 S KC3 Leou KO W &e 0 STEADY-STATE LTC 0 YO CRANKING SPEED CK LT ua CL me

i 0 23 8 10 20 30 APU ROTOR SPEED %

Fig.12 Starter performance and APU rotor drag TORQUE ON TURBINE SHAFT (Nm) I Tu a rv P m ca

-3. 5 11-17

, I 5 10 15 20 25 M 35 40 45 50 55 TIM [SI

Fig.15 Test 1 Ambient temper. = -40 C Torch igniter 8""

Fig.16 Test 2 Ambient temper. = +lo c Torch igniter 12'" I I 1 I 15 20 25 30 TIME [SI

Fig.17 Test 3 Ambient tempcr. = +I5 C Torch igniter 12°i1

9( 8( 7c 4.2

-10

-20

-30 -40.

77 I I I I 8 10 12 14 16 18 20 22 24 26 28 30 32 34

START TIME (s)

Fig.18 Start time vs temperature CONTROL SYSTEM DESIGN CONSlDERATiONS FOR STARTING TURBO-ENGINES DURING COLD WEATHER OPERATION

BY ROBERT R. POLLAK DEVELOPMENT ENGINEER PRAn & WHITNEY GOVERNMENT ENGINE BUSiNESS P. 0. BOX 109600 WEST PALM BEACH, FLORIDA 33410-9600 USA.

ABSTRACT WF/PT2: Ratio of Combustor Fuel Flow to Engine Starting turbo-engines at climatic extremes has inlet Pressure, Mass Flow/Absolute always presented challenges to the systems engineer. Pressure The wide range of both ambient and engine internal FTIT: Fan Turbine Inlet Temperature temperatures experienced by many influential variables increase the complexity of the startup ne: Combustion Efficiency Within Main Engine process both on the ground and in the air. The Burner content of this paper provides the current status of FADEC: Full Authority Digital Electronic Control advanced control methods designed. SDecifically to address combustor ignition and quick, stall-free TQ/& Corrected Torque acceleration to idle. Sensitivity of combustor ignition limits to cold conditions as well as fuel types has TQ,& Engine Total Aerodynamic Torque been accommodated by both the combustor fuel TQ,,,,: Engine internal Parasitic Torque delivery system and control system design. Specific attention is also given to starting at cold altitude TQenslne: Engine Torque Available for Acceleration conditions with extremely hot as well as extremely cold internal engine temperatures. Successfully TOen: Airframe Torque Extracted from Engine meeting these requirements has been accomplished bv desianino the control svstem to automaticallv TOstart,; Starter Cranking Torque I_ monitor external influential variables as well as engine TQ~~~~,:Requested Acceleration Torque internal parameters both prior to and during the actual ,rlo".,nct~ ,#"\1"'1L, startup cycle and using these data to continuously adjust fuel scheduling to obtain optimum startup TQ~~~~,:Actual Acceleration Torque Sensed characteristics. (Actual) By the Control

LIST OF SYMBOLS ip: Polar Moment of Inertia N2: High Compressor Rotor Speed, Revolutions Fuels: JP-4, Equivalent to NATO F-40 Der minute JP-8. Equivalent to NATO F-34, F-35 JP-5, Equivalent to NATO F-44 pT2: Engine Inlet Face Absolute Total Pressure

TT,: Engine Inlet Face Absolute Total INTRODUCTION Temperature The process of ignition that occurs during cold e2: Ratio of Engine Inlet Face Absolute Total weather condltions within a Turbo-Engine has always Temperature to Sea Level Standard Day presented a challenge to the combustor designer and Temperature systems engineer. The multitude of variables that influence the initiation of combustion under these TQ: Torque conditions have certainly been encountered by 6: Ratio of PT2 to Sea Level Standard anyone given the challenge to assure ease and Absolute Pressure consistency of starting.

N2C2: Corrected High Compressor Rotor Speed. Surveying any fuel properties refererice manual quickly shows that the vapor pressure of the fuel N2/G drops dramatically as fuel temperatures approach WF/PB: Ratio of Combustor Fuel Flow to Combustor sub-zero conditions while viscosity similarly increases. Pressure, Mass FlowIAbsolute Pressure With the high viscosity levels encountered. the 12-2

process of atomization for any combustor can be AS the fuel flow for ignition is increased. the resulting significantly altered relative to warm day conditions pressure due to combustion rises proportionally. The The less-desirable fuel droplet distribution together maximum fuel ratio that can be delivered to the with less available vapor surrounding each droplet combustor is limited by the maximum pressure minimizes the useful fuel-air ratio to the point that capability, or stall limit, of the compressor. Similarly, ignition of the mixture is repressed. during the actual acceleration process, the stall limit generally varies and is a function of the characteristic Since fuel vapor available for combustion is restricted shape of the compressor stall line as it varies with under extremely cold conditions, introduction of more rotor speed. Additional variation of this maximum fuel fuel flow is usually the only alternative available to ratio limit is often due to the actual combustion compensate short of the use of exotic systems such efficiency of the burning process during the start as hypergolic ignition or the use of specially designed cycle. This variable becomes especially significant fuel injectors that

Whether the control is hydro-mechanical or fully electronic, the acceleration schedule is usually defined in terms of the already-described WflPb ratio WF - units or a similar term such as Wf/Pt2 (fuel flow PS ratioed to engine or compressor inlet pressure) scheduled against physical or corrected compressor rotor speed. Ignition limit a 0.002 - 0.003 F/A Engine testing ha:; verified that the parameter WflPb

NI presented as wfpb e2 against corrected Figure 7. Starting Window Defined by Ignition and compressor rotor speed, NZ/&, correlates the stall Stall limit as a single line at different ambient temperatures. While combustion efficiency may vary from ignition to For standard day ambient conditions, the ignition limit idle, the stall limit remains consistent for a significant should be quite low. The overall fuel-air ratio within range of ambient temperatures, thus providing a F100-Design combustors generally range from 0.002 consistent relationship between corrected fuel ratio to 0.003. To provide quick, reliable ignition with units and corrected rotor speed. typical ignitor systems, the actual level of fuel flow used for lightoff is usually three to four times higher As fuel vapor pressure is reduced with lower ambient than that actually required to obtain a stable flame. temperature, the fuel-air mixture available for ignition 12-3

likewise drops. Consequently, more fuel is required to temperatures. the design task is to provide required provide sufficient vapor tor ignition. The characteristic fuel flow to initiate a timely lightoff. in the usual case, is manifested as an increase of required fuel ratio the fuel flow adequate for warm day ignition is too low units as shown on Figure 3. The amount of increase for extreme cold day operation. The general for a given temperature level is, of course, a function technique includes a bias with engine inlet sensed of the fundamental ignition characteristics of the temperature, Tt2. or an equivalent compressor inlet combustor design. Test data under controlled temperature behind the fan or low compressor. In any conditions have indicated that the rate of increase of case. as the temperature drops. the fuel schedule is ignition limit is a function of the level of the warm day adjusted to provide lightoff flow commensurate with ignition. if the limit is low. the rate of increase of the the ambient temperature as shown in Figure 5. limit with temperature reduction is likewise low. Characteristics for a typical lightofl rotor speed

schedule requires WF P&:' '%/Typkxl lightoff i .. fuel fm )"ition'--- : . !mils :\. ----,.Cold fempemtUTB >: -- *Vis" fuel/low vaw wssure "F -60 0 60 120 Nt/K ("C) (-51) (-18) 1161 (49) Figure 3. Lightoff Window Influenced by Fuel Ambient temperature Characteristics Figure 5. Temperature Bias Used to Optimize Lightoff Depending on the specific design of the combustor, combustion efficiency can drop off significantly at As mentioned previously, combustion efficiency can Sub-zero temperatures. As a result. the amount of drop significantly as ambient temperature drops. This fuel required to drive the compressor into stall is especially true at low rotor speeds just after ignition likewise increases. If this characteristic is such that during which low vapor pressure and high viscosity the stall limit increases to a greater level than the stili exist. This low combustion efficiency minimizes ignition limit. a useful start window will remain. if the energy available for acceleration such that a higher increase of the Ignition limit is larger than the Stall ievei of fuel flow is required to meet required-to-run limit, it is conceivable that the two will intersect, thus operation. Although difficult to measure or calculate, relegating the engine to an unstartable condition, available heat from combustion is dissipated into the Figure 4. This condition, of course, requires a fuel surrounding metal thus contributing to the low nozzle or combustor redesign, or a compressor with effective efficiency. As such, the fuel schedules at increased Stail margin to provide a useful operational low rotor speeds must be increased a proportionate window amount to maintain acceleration margin, Figure 6. Typical lightoff rotor speed

Fuel required for @nition initiates Stail at lightolf -WF Ped"

I .... . when mid OF -60 0 60 120 increased heat LOWllS High 'is ("C) (-511 (-18) (16) 149) Ambient temperature N*G Figure 4. Lightoff Window Varies with Design and Figure 6. Combustion Efficiency Drives Acceleration Temperature Requirements As the rotor proceeds toward idle, the internal DESIGN PROCESS conditions of the engine are rapidly heated to the Assuming that a useful starting window exists at all point that the combustion Process becomes more 12-4

efficient. Further, less heat is dissipated to internal has been successfully demonstrated at ambient parts thus releasing more energy for acceleration. temperatures as low as -40" F (' C) with both JP-4 ConseQuently.as rotor speed approaches idle, the and JP-8 fuels. initially high fuel schedule required to maintain desirable acceleration is no longer necessary Whenever a start is attempted during a hot-iron condition (an engine with residual heat) the warm Similarly, the amount of fuel required to stall likewise FTlT provides the control with the capability to diminishes due to the process of increasing internal schedule a lower value of lightoff fuel flow to temperatures and increased combustion efficiency. As compensate for the higher combustion efficiency that a result of the lower required-to-run line and stall Prevails at these conditions even though ambient air limit. the resulting schedule can be lowered while still temperature is cold. This feature therefore, provides meeting acceleration requirements, and must be optimum ignition capaDility at sub-zero air lowered to avoid stalling the compressor. temperatures while assuring maximum compressor stall margin when engine internal temperatures are The rate of combustion efficiency change during the warm. acceleration to idle is a reasonably consistent process at cold temperatures as long as the engine Additional lightoff protection is included to address internal temperature during any given start is initially low-grade fuel which may further degrade ignition the same. Difficulty can arise if the combustion capability via extra-low vapor pressure or high efficiency at low rotor speeds is significantly different viscosity. The control monitors the time after fuel is from one start to another at a given ambient introduced to the fiiel manifold and senses whether temperature. That is, for a cold-soaked engine or ignition has taken place. If ignition has not taken "cold-iron" engine, the efficiency is low, but, for an place within the prescribed time period which includes engine still hot after normal shutdown, or "hot-iron". expected time to fill and light, the control begins to the efficiency is high. Should available compressor increase fuel flow in an attempt to establish an stall margin be sufficiently high to accommodate acceptable ignition fuel-air ratio, Figure 8. A these differences, no difficulties will occur. However, reasonable limit is included for obvious safety many high performance compressors may have reasons. limited margin to accept these differences. This is Lightoff not Maximum limit far true for lightoff fuel flow requirements as well as the detected within saiety reasons acceleration schedule. 1 prescribed time ADAPTiVE CONTROL Fuel Initial lightoff The now-common Full Authority Digital Electronic flow - fuel itow PPh Control (FADEC) used on high-technology engines at Pratt & Whitney utilizes internal core engine Throttle Ramp initiated to temperature to schedule lightoff fuel flow to obtain advance establish kner FIA successful ignition during cold day operation with both

"cold-iron" and "hot-iron" engine conditions. Shown Time - seconds on Figure 7 is a typical lightoff fuel schedule for an Figure 8. Additional Lightoff Protection Provided for operational engine. Fuel flow is scheduled with Low-Grade Fuels control-sensed Fan Turbine Inlet Temperature, or FTIT. During warm day operation the fuel is AS shown in Figure 6, the level of fuel schedule scheduled at relatively "low" values which provides required to accelerate can be significantly higher at quick lightoff capability while maintaining adequate cold temperatures during low speed operation due to stall margin. As ambient temperature and likewise low combustion efficiency. Since this process is more FTlT drop. the scheduled fuel flow is increased to physical rather than aerodynamic, and the physical compensate for the more difficuit ignition process of combustion can change depending on the characteristics discussed previously. This technique grade or type of'tuel. designing the fuel flow schedule for the acceleration to idle can often be a hit or miss process. The design is further complicated since warm engines murit also be started at cold 450 temperature, which will also alter the amount of fuel required to maintain a desired level of acceleration. The FADEC has completely eliminated the need for second-guessing continually changing variables during the starting process. ACCELERATION PROCESS Fan t"" ,",et tempratvre The concept of describing turbo-engine starting Figure 7. Lightoff Fuel Flow Set by Engine Internal characteristics in terms of acceleration torque of the Temperature high compressor rotor has been fundamental for 12-5

years. It accurately describes the power of the engine sm1 at a specific speed and relates directly to the aerodynamic capability of compressor and turbine characteristics.

The relative relationship of acceleration torque to compressor operation is shown on Figure 9. The level of torque available for acceleration is proportional to the compressor operating line and the temperature generated within the combustor to provide turbine power. The resulting airflow. pressure, and temperature together with the compressor and turbine efficiency characteristics provide the net excess Margin from st811 limit power for acceleration. As speed increases. airflow and compressor pressure capability increase, and the Acceleration Torque levels at temperature required to reach this pressure likewise torque increases, thus increasing net power. ~talllimit --..-

I I II

Corrected rotor speed Figure 10. Acceleration Limits Defined by Compressor Stall / r' . Net a~~elerati~nTO16 Total torque (fm sea Iwel. STD day)

~ 8,

Fire out Fire on

Increasing power I Acceleration torque .-- ' Friction. pumps, windage, etc.

corrected ,010, speed Figure 11. Internal Influences Incorporated Within Corrected rotor speed Control Schedules Figure 9. Compressor Fire-on Line Relates to corrected high compressor rotor speed. N2C2, and Acceleration Rate represents total engine power which includes net measurable acceleration torque and that absorbed For any given rotor speed, fuel flow can, of course, internally within the engine for friction. pumps, be increased to the point that the compressor will windage, etc. This relationship has worked stall, as shown on Figure IO. At this point on the successfully at ambient temperatures ranging from compressor map, there is a corresponding point of -40 'F ("C) to over 125 "F (52%). maximum acceleration torque that represents in physical terms the aerodynamic capacity of the To construct the physical acceleration characteristics engine. These relationships circumvent any variations of the engine, the TqIS schedule is uncorrected by of combustor efficiency due to fuel grade or internal the following relationships: engine temperatures, such as a hot engine starting on a cold day. N2=N2C2*&, Tq = Tq/8 * 8 The FADEC on Pran 8, Whitney's FlOO family of engines has been using this technique SuCCeSSfUllY As part of the overall acceleration rate of the engine, for years. The actual allowable level of acceleration the starter cranking characteristics are also torque for any engine model is programmed into the programmed as a function of ambient temperature control as Torque18 and represents the maximum and pressure altitude (elevation) to represent the desirable aerodynamic torque available for a Start at starter's contribution to acceleration. Further. the a given temperature, such as that shown on Figure aircraft accessory loads extracted from the engine 11, Corrected torque in terms of TqIS is used against are also included as shown on Figure 12. The 12-6 combination of these three major variables comprises closed-loop function described mathematically within the torque available for acceleration and becomes the the programming of the FADEC.' This capability is actual requested parameter that continuously changes useful when different types of fuel are used, such as with ambient temperature and pressure conditions. JP-4 and JP-5, i?specially when fuel type adjustment on the fuel control is not available. While the fuel scheduling used for the two starts will be different, the acceleration rates will be virtually identical together with identical compressor stall margin consumption.

Consequently. as ambient temperature decreases to sub-zero ieveis, engine acceleration is not altered from the desired, automatically-programmed values. Since fuel flow is continuousiy-adjusted rather than pre-programmed. the control system adjusts fuel flow to meet the desired start characteristics as influential variables change.

CLlMATiC DEMONSTRATIONS N* Figure 72. External influences Define Net Acceleration As mentioned above. all of the FlOO family of engines Capability currently in production incorporate the closed-loop control mode which has eliminated the difficulties The control calculates the desired net acceleration normally encountered during sub-zero starting. This torque from these three variables (including engine was successfully demonstrated during climatic testing parasitic drag) on a continuing basis as a function of of an FIOO-PW-220 in the Climatic Hangar at Elgin NZ, Figure 13. Rate of change of N2 with time is Air Force Base in Florida. The entire aircraft was sensed by the control and converted to a torque installed within the hangar and tested under value by way of compressoriturbine moment of temperature extremes of -10°F to - 40°F using JP-4 inertia. This actual torque is compared to the and JP-8 fuels. Each cold test was preceded by at requested value to provide an error which is used to least a 10 hour soak at the specified temperature. set the value of fuel flow increase via a gain. The The soak time was begun after critical aircraft end resuit is a control system that delivers only the gearbox and engine oil temperatures had reached fuel flow required to produce the desired aerodynamic ambient temperature. The resulting start times are acceleration of an engine rather than scheduling fuel shown on Figure .I4 which demonstrate that the flow on an open-loop basis and obtaining an potential start time of an engine is actually faster acceleration rate that can vary significantly with instead of becoming longer as ambient temperature is uncontrollable influential variables. With the exception lowered. of the initial portion 01 the start during which the engine is accelerating on open-loop ignition fuel flow, 1. United States Patent No. 4,274,255 and Foreign the acceleration times to idle are virtually identical for Patents, R. R. Poiiak. "Control For Startup of a Gas all engines. Any fuel control hardware tolerances are Turbine Engine." Assignee: United Technologies Corporation, Hartford Conn. automatically eliminated since fuel flow is a

L 1

- trim to m -,. + (Request) - engine Nzl!xY- II

U Figure i3. Starting Components Assembled to Control Fuel Flow 12-7

F-15 engines were recorded with the FADEC in the F-15 aircraft mode which uses the power extraction loads representative of the F-15 aircraft. Similarly, the F-16 engine start times are representative of the loads extracted by the F-16. During field operation, the load schedules are automatically selected by the location of pins in the aircraft-to-engine control cable pin connector for the particular model aircraft. Since L the loads are larger for the F-15, the start times are 50 about two seconds longer while the F-16 times are shorter due to the lower levels of extracted loads. Figure 74. Quick Start Time Capability Provided at This automatic process provides a consistent ievei of Cold Temperatures compressor Stall margin by keeping the engine power The time to obtain ignition was held approximately output virtually the same while adjusting the ObSeNed constant at ail ambient temperatures since ignition net acceleration rate by considering external loads. fuel flow is adjusted as the temperature drops, More recently, a similar test was conducted on the especially in the sub-zero range. Actual ignition and FIOO-PW-229 engine in the same climatic test start times during the test were: facility. During these tests, the entire aircraft with the AMBIENT FUEL IGNITION TIME START TIME, REF. 1 engine installed was soaked at the sub-zero 18econdEI TEMP 'FIDC1 -TYPE (seoondol temperatures from 11 to 24t hours after the aircraft central gearbox and engine oil temperatures had -10 (-23) JP-4 8 31 reached the desired ambient temperature. Again. the 30 -25 (-32) JP-4 9 start times at sub-zero temperatures were faster than -40 (-40) JP-4 7 24 those demonstrated at standard day conditions as -40 (-40) JP-4 7 23 shown on Figure 16.

-40 (-40) JP-4 7 23 100 r I 138 -40 (-40) JP-8 8 25

IGNITION TIME is dt ed between throttle advance combustor lightoff. START TIME is delined as the total time taken between throttle advance to 95% Of idle high compressoriturbine rotor speed, N2, at the test temperature. With non-closed-loop controls, longer start times are 4 !o ,b ;o io 50 A-51 typical at the colder temperatures and are due either stan me. SR to long ignition times. fuel schedules that should be Figure 16. Cold Temperature Capability is a Design richer due to poor combustion efficiency, especially Process with low-grade fuel, or control tolerances which can Similar to the -220, the times to obtain ignition were affect the acceleration capability if too lean. held to a minimum by the automatic adjustment of fuel flow by the control as the temperature was The consistency of the closed-loop control system reduced. This method again provided reliable and can be seen on Figure 15. Start times for 670 consistent ignition. FIOO-PW-220 production engines are compared to AMBIENT FUEL IGNITION TIME START TIME, REF. 2 the delivery requirement for ambient temperatures that TEMP 'FIT) TYPE lsecondil lseoondol range from 90 "F (32 "C) to about 25 "F (-4 "C). - The sample includes 384 F-15 model engines and t75 (r24) JP-4 7 35 286 F-16 model engines. The start time range of the -10 (-23) JP-4 8 31

38 -25 (-32) JP-4 6 24 '"EBo 27 -40 (-40) JP-4 5 25 18 -40 (-40) JP-4 3 22 4 "bant air -40 (-40) JP-4 5 25 -7 le" - *C -18 -40 (-40) JP-8 9 22 -40 (-40) JP-8 4 23 -40 (-40) JP-8 4 23 -5, -0 10 20 30 40 so €a -Stan time. nec Figure 15. Consistency Provided Through Closed Loop The extremes of engine temperatures during starting Control are found while attempting to start the engine in the 12-x

air. Depending on the requirements specified by the significantly longer and converts into lost altitude for 8 customer, procedures can be significantly different single engine application. Obviously, if the shutdown from one application to another. event occurs at a low altitude, recovery time is critical. In multi-engine aircraft applications. the classical windmill airstart may be the desired procedure since, The same technique used during ground starts that for the most part, the time factor for restarting the delineates cold vs hot engine conditions is used in engine may not be critical. Further. designing for a the air. Fuel flows for ignition are shown on Figure 18 single procedure offers simplicity and minimizes as a typical schedule that accommodates hot engine development costs. conditions at altitude. The schedule uses FTiT as the engine temperature parameter to delineate internal in such cases, the engine is usually in a cold gaspath conditions prior to start initiation. Note that at condition due to the relatively long engine-out time at the upper scale of temperatures, fuel flows are sub-zero engine inlet temperatures. Under these scheduled lean to provide stall-free ignition. while at conditions. fuel flows to obtain quick, consistent the colder. sub-rero temperatures that represent ignition together with a fuel schedule to provide windmill conditions, high fuel flows similar to ground adequate acceleration can be designed without too starting on cold days are used. This method of much difficulty. In some cases a single fuel flow for scheduling fuel flows provides the capability of starting lightoff together with a single acceleration schedule, under any set of engine conditions from cold e.g.. WfiPb. can suffice for all windmill airstart windmills to extremely hot, quick restarts. requirements. As mentioned above, at cold temperatures, starting requirements may be met only by rich fuel flow scheduling for both ignition and Gmud stms acceleration. As such, the predescribed procedure must be adhered to in detail to restart the engine *intam during an in-flight shutdown. Should the engine be 'MLQhtOl,PPH Ibr . 350 hot. however, the windmill-designed fuel scheduling ::by

The accuracy of the closed-loop control provides not Time only consistent airstart characteristics, but a larger Figure 17. Spooldown Reduces Engine-Out Time airstart envelope. This includes hot as well as cold engine temperatures at both warm and cold engine SUMMARY. inlet air conditions. Since the late 1970% Pratt & Whitney has With an open-loop control system, hardware incorporated the closed-loop starting process in its tolerances and mechanical wear influence the FADEC engine controls. The design technique Of capability of achieving successful airstarts especially incorporating all the primary parameters that make up at low airspeeds and reduced rotor speed conditions. the starting process is fundamental to the success of While some control units might be able to meet low cold-weather starting. Fuel flows for ignition are airspeed requirements, some units, whether too rich tailored specifically to the combustor and the desired or too lean will be unable to accelerate the engine to acceleration characteristics of the engine are built into idle successfully. While it is desirable to initiate the the control rather than a pre-determined acceleration restart as soon as possible at higher rotor speeds. fuel schedule. The pre-established starter cranking the pilot does not always have this luxury due to Capability and aircraft extraction loads provide ail aircraft departure, etc. By the time the pilot necessary external information for the control to determines that a restart is required. rotor speeds produce successful and consistent starts. The fact may have decelerated to extremely low values at that the control senses whether proper acceleration is which very little compressor stall margin is available. taking place allows for the complete flexibility Of adjusting the level of fuel flow to meet the capabilities Under these conditions, accurate scheduling of fuel of the engine. This design provides on-line flow is mandatory if low airspeed recoveries are to be instantaneous adjustment to compensate for achievable. Maintaining low airspeed during the entire low-grade, low vapor pressure, or highly viscous fuels recovery is usually desirable due to more optimum as weii as sub-zero temperatures. Further, the control times aloft or glide distances available. is capable of modifying fuel flow delivery to compensate for engine restarts with hot internal FLiGHT DEMONSTRATIONS temperatures and hot fuel while operating in a sub-zero ambient temperature environment. This is The incorporation of all the requirements mentioned especially desirable during altitude restarts which may above into one control system was accomplished in include both warm and cold inlet air as well as cold Pratt & Whitney's F100-PW-220 used in both the as well as extremely hot gaspath temperatures. With F-15 and F-16 aircraft. Prior to the incorporation of these features, engines can now demonstrate the aforementioned FADEC into the FlOO family of consistency and reliability which assure the customer engines, the minimum airspeed required to provide that special handling and special procedures during 100% reliability for all enginelcontrol combinations was sub-zero operation are no longer necessary. 250 knots indicated airspeed, kts. While some engines had 200 kt. Capability with the open-loop REFERENCES control mode, tolerances within the control hardware necessarily required higher airspeed for others. 1. R. Scrivner. United Technologies - Pian & Whitney, "F100-PW-220 Cold Start Preliminary Data Engines with the closed-loop control mode within the Review," 21 June 1985, Attachment 2. FADEC now provide consistent airstart capability at 200 kts up to 40.000 feet (12,000 meters) pressure 2. M. J. Pfeiffer, T. J. Norton, J. A. Patterson, United altitude as shown on Figure 19. The procedures used Technologies - Pratt & Whitney. "F100-PW-229 during the flight test included initiating the restarts at Climatic Laboratory Starting and Acceleration Test compressor rotor speeds that ranged from 25% to Report", 1989, Report Number FR-20490. Table 4, 50% of cockpit tachometer reading as well as page 43. steady-state windmill conditions. The majority of the starts were initiated after a shutdown from a high power condition in order to place the engine in a condition of minimum available compressor staii margin.

.3yI S" S" airstam in F.15 LBgend 0 25% A 30% Allifude - 0 40% 1000 11 v 50% 0 Windmill Ope" - SucCeSSl"I Closed .U"S"cCe5s1YI

0.2 0.4 0.6 0.8 1.0 Mach number Figure 19. Improved Start Capability with Closed Loop Start System 12 -10 Discussion Author: While power extraction does vary from one engine to another in multiple engine aircraft, the load characteristic 1. K. Piel, BMW-RRAeroengines programmed into the control represents the worst case With reference to figure 19 of the paper, what is the figure of conditions. In that way, actually applied loads will always be steady-state windmill conditions? equal to or less than those used in the control. As such, compressor stall margin is always maintained at the minimum design value or more, thus maximizing reliability Author: The windmill limit is defined as the airspeed required to 3. W. Bouwman, Netherlands MOD obtain an "2-rpm of 11.4% Tach speed. This is defined as Is the worst case also used concerning TQ because of no the guarantee rpm required to perform a windmill airstart. temperature dependence occurring? In most cases starts can and have been performed at lower rpm. Author: No temperature bias is included because the effect of cold temperature is less significant for larger engines than in 2. R. Wibbelsman, GE smaller size engines, such as those used for helicopters. A/C TP, loads can vary dependent on A/C load However, should a situation exist that the parasitics are distribution. This makes rate control schemes more difficult. significant at colder temperatures, the variation of the How do you handle this? characteristic can easily be programmed into the FADEC. COLD START DEVELOPMENT OF MODERN SMALL GAS TURBINE ENGINES AT PRA" & WHITNEY CANADA by DSBreitman and F.K.Yeung Pratt & Whitney Canada Inc. 6315 Dixie Road Mississuaga Ontario Canada L5T 2El NOMENCLATURE high pressure from the auxiliary power units (APU). N, compressor rotational speed P, compressor delivery pressure Power to an electric starter is supplied hy AP pressure drop across combustor either batteries or a ground cart. After the 0 relative temperature ratio TdT,,f engine starts, the starter, drawing power from the engine, will act as a generator to recharge ABSTRACT the batteries. This type of starter is usually Engine cold start capability is essential for air- applied on an APU or in small gas turbine air- crafts in Arctic or winter operations. Demon- craft. The selection of an electric starter stration of this capability is part of the engine requires careful evaluation of the starter1 development and certification requirements. torquehattery match for the application. Variables such as the combustor design, the igni- Fipm I: &flat of Tcmprmiure on PW30.5 Drog Torgur tion system, the fuel nozzle design, the diffuser 70, exit flow characteristics, and the compressor performance at sub-idle conditions all affect the cold start capability of an engine. This paper describes briefly how these factors are usually optimised, and presents an overview of the successful PW305 Engine cold start development (yith an electric starter). The PW305 is a new turbofan engine from Pratt & Whitney Canada in the 5000 lb (22.2 kN) thrust range. INTRODUCTION J z (I to The starting cycle of a gas turbine engine is the X Nz acceleration of the turbomachinery from 0 to idle speed. The starting procedure generally The main factor affecting the selection of the consists of 3 stages: cranking, ignition, and starter combination is the torque requirement. acceleration to idle. During a typical engine Figure 1 illustrates how the engine drag torque start, many different events are occuring in the (at N, lower than light-off speed) increases compressor, combustor, and turbine, simultante- with decreasing temperature. The starter ously and in sequence. Most of these events are delivery torque (which decreases with decreas- inter-related, and sometimes counteract each ing temperature as shown in Figure 2) must be other. Furthermore, excellent engine starting able to overcome this drag at the lowest temp- characteristics may be at the expense of other erature within the operating range, and must engine requirements. To ensure relatively be able to crank the engine beyond the light- healthy engine starts over a full range of operat- off speed. ing conditions and environments, considerable development work is necessary to obtain the best Figure 2: Effeccr of Tempermure on PW305 Sinrrrr Delivery Torqur possible compromise, taking all of the engine requirements into account. CRANKING STAGE In this. stage, a starter provides the power to crank the turbomachinery from 0 speed up to a pre-determined light-off speed, at which fuel is introduced into the combustion chamber. The most important consideration for the cranking stage is the selection of the starter. Selection of Starter

The 2 most comon types of starters are air starters and electric starters.

The power of an air starter comes from a turbine which requires high pressure air to drive. After the torque requirement is finalised and The main application for aircraft operations the starter chosen, the minimum voltage snp- with air starters is to start main engines with ply for the starter to deliver this torque can be 13-2

determined from the starter characteristics. not properly atomised because of the low fuel The batteries must be able to provide this nec- flow. At combustor AP beyond the stability lim- essary voltage. it the air velocity may be too high for ignition, or for any initial flame to stabilise. The preference During development of a typical PWC fan is a wide ignition range, which can provide ade- engine, the power supply was the electrical quate margin to ensure smooth and fast ignition simulation of the batteries at the lowest oper- during every start. Obviously, the local flow con- ating temperature. Only when all parameters ditions at the igniters and fuel nozzles are impor- affecting starting had been optimised were the tant factors affecting ignition. The optimum actual batteries tested to confirm the devel- combustor and fuel injector configuration, taking oped data. all the engine requirements into consideration, must be determined experimentally during development. The other factors that also affect igni- IGNITION STAGE tion are:

During cranking, as the engine reaches the light- 1)operating temperatures off speed, fuel is introduced into the combustion 2)altitudo chamber and ignition occurs. Fast ignition is 3)ignitiorr fuel flow desirable in order to avoid fuel pooling at the 4)ignition system bottom and unburnt fuel accumulation in the 5)fuel nozzle and igniter Idcations down-stream turbine stages. Hence a long time 6)selection of the light-off speed to light will, very likely, cause torching and damage to the turbine.

Prior to full engiue starting tests, the ignition Figure 4 illustrates how the ignition range is range (ie. optimum fuellair ratios for ignition) of affected with decreasing temperature. At low- the combustion system has to be determined er temperatures increasing fuel viscosity through-out the operating range of the engine, adversely affects fuel atomisation. To com- including the lowest operating temperature and pensate, higher fuel flow is necessary to the most adverse altitude relight conditions. A ensure successful ignition. graphical representation of the ignition range is given in Figure 3. The upper limit is the fuel-rich

Figure 3: Typicof PW305 Ignirion Range

- $ 3.0- Y 0 k - 2.5- 3s LL 2.0- u.3 L $ 1.5-

1.0-

0.5(-! 0.5(-! 200 250 0 50 100 150 At altitude, a stalled engine will windmill COMBUSTOR AP lN/m21 because of the aircraft's forward speed. This will cause a high pressure drop ( AP ) across the combustor. At higher combustor AP's the limit above which the fuellair ratio is too high stability limit for ignition is reduced (see Fig- for combustion to occur, or that any initial flame ure 5). A wide ignition stability range is will be extinguished by fuel. The lower limit is essential at altitude where fast relights are the fuel-lean limit, below which, either the fuel/ expected over a range of aircraft speeds (ie. air ratio is too low for combustion, or the fuel is windmilling and thus combustor AP). 13-3

Fi~ure5: Eifecr of Aitirude on PW305 Ignilion Range achieved mostly at the expense of igniter durability. Therefore, the selected igniter box should have the minimum output energy level \ and/or pulse rate required to achieve a rea- E 2.5- CI sonable ignition range at tbe lowest tempera- L '\' ture and the highest altitude in the operating 9 , \ k 2.0- -P--- envelope. I ' 9200m RANGE

!!Y

~ . 1.5- YI 3LL

5 1.0- h

0.5 1000 t500 0 ' 500 ' 2000 COMBUSTOR AQ rNim21

&nition fuel flow

The main concern in the fuel system design is fast ignition and light-around. This is essential to provide a good circumferential temperature distribution following ignition, in order to 0.5 avoid a non-uniform back pressure onto the 0 500 1000 1500 2000 COMBUSTOR CNtm21 compressor, which can force it into stall. AP

After the ignition range is determined for the various operating conditions, the starting fuel Although there are usually 2 igniters on an flow for each condition can be selected in con- engine, regulations specify that the engine junction with the combustor AP (which is must he able to relight at altitude with 1 igni- determined by the light-off speed). Because ter only. of the narrower ignition range at cold temperatures and at altitude, the control on the accu- Fuel nozzle and inniter locations racy of the fuel flow is just as important as the actual amount itself. Otherwise a long time to The locations of the fuel nozzles and the igni- light will result. ters can significantly affect the ignition range, and are determined in relationship to the com- The ignition fuel flow normally consists of a bustor. In additon to ignition, the other con- pre-determined amount of "pilot" fuel (from siderations are combustion efficiency, combu- the pressure atomisers) and "main'' fuel (from stor exit temperature profiles, combustion the air-blast nozzles), flowing simultanteously. system durability, maintainability, installation This is to facilitate a fast light-around to and access. After their axial locations are produce a uniform temperature for reasons fixed, the circumferential locations of the pilot cited in the previous section. nozzles and igniters have to he decided carefully. knition system A pilot nozzle and an igniter must be in the The ignition system consists of an electrical same proximity for ignition. Two of each are power supply, an exciter box, and usually 2 required for redundancy. The 2 pilot nozzles igniters. The electrical power supply comes may be side-by-side, or they may be separated, from the batteries. The main criteria for igni- depending on the following: ter selection are durability, low loss (Le., the ability to transmit the energy from the exciter 1)air-flow around the fuel nozzlehgniter box to the igniter tip efficiently), and compat- region, including the air swirl ability with the combustor. 2)fuel manifold connections, including the position(s) of the fuel supply The choice of exciter boxes is dependent on 3)the intended circumferential fuel flow distri- the output energy level and the pulse rate. bution Figure 6 is an illustration that the higher this 4)the intended directions of flame propagation energy level and/or pulse rate is, the wider the after ignition ignition range becomes. The energy level is a more dominant factor. This is very beneficial Selection of liaht-offweed for low temperature ignition and altitude relight. However, this performance is The selection of the light-off speed is depen- 13-4

dent on the optimum combustion AP, the Figure 7 is a sketch of a typical compressor torque demand, and compressor stall charac- map at very low speeds. Below a certain teristics. These requirements are sometimes speed the compressor running line is beyond contradictory, and a careful compromise is stall. The compressor stall characteristics add necessary. to the engine drag significantly, and may prevent a successful start. At these speeds, a For a given engine (where the compressor flame in the combustor may simply hack pres- running line has been fixed by the turbine vane sure it further, forcing it deeper into stall, and size, and the combustor APlP, being constant) increasing the drag torque tremendously. As the absolute combustor AP is a function of the illustrated in Figure 7, the compressor running compressor corrected speed, NJa. From a line is in the stall-free region above a certain speed. Preferably, the light-off speed should plot of the ignition range, it would appear that be above this N,, otherwise the starter must the optimum combustor AP is the lowest pos- cany the engine through this regime. sible.

For air-blast nozzles, a higher AP generates If compressor handling bleed is desirable, the better fuel atomisation, which is essential for light-off speed should be sufficiently high for fast light-around after ignition. Since finer fuel this bleed to be effective, droplets require 1es.s evapouration energy, good atomisation is even more important at ACCELERATION TO IDLE STAGE low temperatures, hence this requires a higher AP. Fortunately, as the temperature decreas- After ignition, a relatidy fast and smooth accel- es, for a given engine speed (N2), the correct- eration to idle is desirable. This is controlled by ed speed of the compressor (NJa)is higher the engine drag, rate of acceleration, fuel sched- than that at normal temperatures, hence a ule, compressor performance, and temperature higher compressor delivery pressure P,, and a schedule. higher combustor AP.

As discussed previously, the engine drag torque (at N2 below the light-off speed) As discussed previously, the engine net drag increases with decreasing temperature. With (turbine delivery torque - total engine drag) an electrical starter, the delivery torque increases with Nz during the cranking stage. decreases with decreasing temperature, After ignition the turbine extracts power from because of the lower delivery power from the the combustion gas, hence the net drag starts batteries. This delivery torque also decreases to fall off. As Nz increases, the turbine deliv- with increasing N2, as illustrated in Figure 2. ery power will exceed the total engine drag at This would suggest that a low light-off speed a certain speed, above which the engine self is desirable. The light-off speed must be sustains, as illustrated in Figure 8 (the 0-drag selected such that the starter delivery torque is line). Normally, the starter motor is applied always higher than the engine drag over the until the engine speed exceeds this self- complete temperature envelope of operation. sustaining speed.

Figure 7: Typical PW3OS Compressor 111 Low Speeds

%OF DESGN CORE FLOW 13-5

Rate of acceleration tion of the how close the compressor is from stall. It is desirable to have this fuel margin The control of the engine acceleration rate to be as wide as possible. This can he from ignition to idle speed is very important. achieved by the improvement of the compres- Too fast an acceleration can force the com- sor. pressor into stall, and very likely will result in an unsuccessful start. This will compromise safety for altitude relights. Too slow an acceleration would pro-long the exposure of the Figure IO: Ty#col PW30S Fuel Morgin turbine stages to the hot combustion gas at 4o 1 sub-idle speeds where cooling is inadequate. This will very likely cause hardware damages. Furthermore, since the engine normally still requires starter assist below the self-sustaining N,, very slow acceleration may cause the starter motor to overheat and result in damages.

Figure 9: Effecr of Acceleration Rare on Compressor Choraeteristicr

!? !z w The fuel schedule must be between these 2 2 YIv) limits. A fuel schedule closer to the upper a limit corresponds to a fast acceleration, while a schedule closer to the lower limit implies a very slow acceleration. The final schedule should be governed by the selected acceleration rate, which was discussed in the previous section. AIR FLOW The fuel flow on the engine “demand line” at the self-sustaining speed is usually taken as the minimum fuel flow in the starting fuel Figure 9 is an illustration of the relationship schedule. between the acceleration rate and the compressor characteristics. It is obvious that the Compressor preference is a moderately fast acceleration. The main control over this is the fuel sched- During the start cycle, most of the engine drag ule. is contributed by the high compressor, which in the case of most PWC engines, consists of Fuel schedule several axial stages followed by a single centrifugal stage. This drag increases significantly as Fuel scheduling provides the most direct con- the compressor is in stall. Therefore it is par- trol on engine starting. It affects all the other amount to eliminate, or at least minimise the parameters, and therefore has to be set care- extent of the compressor stall. At low sub- fully. Prior to the setting of the fuel schedule, idle speeds, stall usually occurs at the first the fuel margin as sketched in Figure 10 must axial stage rotor. At higher speeds, stall is be determined experimentally. The upper lim- controlled by the diffuser downstream of the it is the over-temperature and stall limit, centrifugal stage. The speed where the first above which the inter-turbine temperature is rotor and diffuser stall lines cross each other exceeded and/or the compressor stalls. The depends on the stall margins of these 2 com- lower limit is the engine ”demand line” where ponents. The most commonly used method the net engine drag = 0, helow which the for stall control at these speeds are variable engine cannot self sustain. The minimum dif- inlet guide vanes, compressor handling bleed, ference between the 2 limits is the fuel margin. diffuser sizing and alignment. Since this margin is proportional to the compressor stall margin, it is a very good indica- The variable inlet guide vane (VIGV) consists 13-6

of airfoils set at certain angles upstream of the compressor stall. If the "fuel margin" (dis- high compressor. Its function is to introduce cussed in the "Fuel Schedule" section) is very swirl into the air flow, which delays compres- narrow, a carefully controlled temperature sor stall at very low sub-idle speeds. In our schedule can he used to "navigate" through the experience, the VIGV alone may not be ade- awkward speed range. This temperature quate. It is usually employed in conjunction schedule is usually determined experimentally with compressor handling bleed at sub-idle during development, and Figure 12 is an illus- speeds. tration of several schedules that were evaluated during the PW305 development. For a The handling bleed is air taken off the com- modem PWC fan engine, the start cycle is pressor gas path through bleed slots. The controlled by a sophisticated electronic engine bleed serves as an air flow sink that helps control system (EEC) to follow a temperature increase the air flow through the first stages of schedule. the compressor, and pulls it away from stall. Figure 11 is an illustration how bleed affects the compressor stall margin, as indicated by the fuel margin. For this bleed to he effective, the light-off speed should be sufficiently high, such that the compressor pressure ratio at the slots is adequate for air to be bled off.

OTHER CONSIDERATIONS

During PW305 development all the above parameters were optimised, and successful starts at low temperatures were demonstrated. More tests were conducted in search of improvements, partly to demonstrate safety, and partly to determine ultimate limits.

1)More tests at extreme temperatures, even below the operating range, with alternative fuels and oils. The purpose of this was to The diffuser throat sizing, while having a sig- determine whether the engine starting per- nificant effecton the sub-idle stall margin, has formance could be improved further with low- an even greater effect on performance and er viscosity fuel andlor oil. stall margin at full speeds. Since this geometry cannot be vaned as readily as the VIGV 2)Starts conducted with deteriorated batteries to and the handling bleed, it is decided usually determine the worst conditions these compo- with more consideration given to the full nents could he operated at, without compro- speed performance and stall margin. mising safety.

Temperature schedule 3)Starts demonstrated in the test facilities with water saturated fuel at the worst icing condi- To avoid damages to the turbine stages, inter- tions. turbine temperature limit schedule for sub-idle speeds must be determined. This temperature 4)Altitude relights at extended operating enve- schedule depends mainly on the amount of lope demonstrated in the test facilities. This cooling available for the turbine stages. At was later confirmed in flight tests. any given speed, this temperature should be sufficiently high for the turbine to extract 5)Successful altitude relights with single igniter enough power for reasonably fast acceleration, and single pilot nozzle beyond the operating yet low enough to avoid turbine hardware envelope demonstrated in the test facilities, damages and/or over-acceleration causing and later confirmed in flight tests. 6)Starts with deteriorated engine to simulate the Discussion hardware conditions after an extended period of operation. 1. C. Meyer, French MOD Quelles sont les possibilitb en cas de redemarrage avorti sur PW 305? Y-a-t-il des procedures en des precautions particulihes?

CONCLUDING REMARKS Author: The starting sequence for the PW 305 both on the ground This has been a brief description on the major and in the air is as follows: parameters affecting starting, the problems that usually arise, and the techniques applied during (i) Continuous ignition selected on the development of the PW305 Engine to over- (i) Thrust lever angle (TU)selected to idle position (iii) Starter motor switch on come these problems systematically. Part of this experience may be readily applied to other small As the engine accelerates through 10% N2 @ugh pressure gas turbine engines with an electric starter. rotational speed), the electronic engine control However, each engine is designed for its own automatically introduces fuel. Since the igniters are already unique application, and hence some of the proh- on, ignition- takes place and the control accelerates the lems may require considerations on an individual engine to idle. basis. In the event of an aborted start. at altitude, windmillinn will scavenge pooled fuel and the process (i-iii) is repeated. On the ground a motoring cycle must be carried out to ensure scavenMg of excess fuel. Then the start procedure is repeated. ACKNOWLEDGEMENT This procedure is repeated until a start is successful or it is felt that further investigation of the situation is necessary The authors would like to thank Pratt & Whit- (burnt out igniter plugs, plugged fuel nozzles., etc). ney Canada Inc. for the permission to publish this data, and to express special appreciation to 2. I? Sabla, GEAE Mr. Alan Wheatley (of the PWC Combustion Curve nr. 3 shows delta p off by a factor of IO? Dept.) and the test crew at the National Research Council for their support during the Author: cold start development. This is in fact incorrect, scale should read 0-2500

Design Considerations based upon Low Temperature Starting Tests on Military Aircraft Turbo Engines by H.-F. Feig Wehrtechnische Dienststelle fur Luftfahrzeuge (WTD 61) Flugplatz, 0072 Manching, Germany

SUMMARY Test experience on engine low temperature starting was obtained in the course of multinational and national trials to assess weapon system performance. The objective of the trials was to recommend a clearance for the weapon system. In order to carry out these tests adequately the operational role of the weapon system had to be considered and the operational limits of the engines and associated systems had to be known. Parameters influencing low temperature start capabilities were reviewed and experience gained from the tests was discussed.

1. INTRODUCTION

Engine start is an interaction of various systems. Consequently, not only the engine has to be considered - of equal importance are the starting system and the capability of monitoring the start process. The evidence that the solution provided by the contractor complied with the requirements of the Services had to be demonstrated by official tests. Conclusions resulting from these tests are summarized in a recommendation stating the extent to which the weapon system can be operated by the Services. The assessment of low temperature start performance represents a specific and very important field in the overall testing of the weapon system, since low temperature engine starting has a considerable effect on the value of the weapon system. a) The objective of the low temperature engine start test has to be an assessment of whether the technical solution meets the requirements of the Services. This official task can be defined in more detail as follows: - checking of the engine and engine starting system standards with regard to performance and characteristics required by the specifications - checking the validity and applicability of the operating instructions - assessment of the weapon system start up activity from the military operational point of view - gathering of information for checking and validating the national weapon system and engine documentation - recommending a clearance for Service use. b) The mission for which the weapon system was designed together with resulting performance aspects must be considered with regard to their influence on the test programme. In order to meet operational requirements, each weapon system design - including engine design and starting system design - comprises many technical features which together contribute to a satisfactory weapon system operation. Knowledge of this features is required for the assessment of low temperature engine start capabilities. This information has to be obtained from specifications and design documents and from development test experience available from airframe, engine and system manufacturers. c) Operational limits, which must not be exceeded, togehter with proper interaction of the systems involved, form the technical basis for the assessment of the test results. 14-2

2. MILITARY SYSTEMS ENVIRONMENTAL REQUIREMENTS

During the weapon system definition phase environmental requirements and system capabilities were established. Guidelines for the selection of requirements and the scope of testing necessary are provided by: - NATO STANDARDISATION AGREEMENTS [STANAG) - MILITARY STANDARDS - MILITARY SPECIFICATIONS

List 1 shows an abstract of current documentation relevant to definition and evaluation of system capabilities.

3. TEST ENVIRONMENT 3.1 CLIMATIC CHAMBER TESTING

In order to achieve independence from natural meteorological influences, environment simulation is used. In many cases the environmental systems of test chambers are not capable of delivering the airflow demand of the engine. Therefore, during start up, outside air is mixed into the test chamber. In general, outside air has a higher temperature and a higher total amount of humidity than the air of the test chamber. This results in a direct fall out of humidity and rapid ice build up on the cold soaked test object. It must be ensured that the free water content of the air and the ice accretion does not result in any interference with the test objective. 3.2 TESTING IN NATURAL ENVIRONMENT

Complete weapon system testing in a natural environment is the approach desired, sometimes, however, it is the reliability of the weather forecast that is tested! As shown in Fig: 1, a typical temperature survey of a 94-hour cold start test conducted at WTD 61 airfield during winter 1985 is provided. The lowest temperature reached was -21°C and the highest temperature observed -9'C. Fluctuation of temperature is remarkable in the course of a day. In order to give an impression of the cool down capability of the natural environment, the frequency distribution of the temperature survey is shown in Fig: 2. The reference soak temperature achieved on aircraft during this test was -14°C.

3.3 COLD SOAK EFFECTIVENESS

A so called "Full Cold Soak" is only completed when all systems affected during the test are cooled down to the same level of temperature. Because of the unequal mass concentration on weapon systems long soak periods were required in order to obtain a balanced temperature. A proper test instrumentation must be installed in order to measure system temperatures.

4. INFLUENCES IN ENGINE COLD START CAPABILITIES 4.1 VISCOUS SHEAR EFFECTS One of the main influences aggravating engine cold start up are viscous shear effects depending on: 4.1.1 TYPE OF OIL Modern engines and weapon system designs demand sufficient lubricity and thermal stability under high operating temperatures. The following oils are used in modern German military systems:

0 - 160 according to UK-Specification DERD 249?/3

and

0 - 156 according to MIL-L-23699 C Amd 1 or DERD 249911 Amd 2

These oils are of the 5.5 mm"/s Viscosity Class at 1OO'C. As expressed in Ref: 1, the viscosity of the above mentioned oils increases by a factor of 150 when the temperature is lowered from +15'C to -4O'C. This increase in viscosity is the main driver in system drag increase. The use of oils with a lower base viscosity,which are more suitable for cold operations, is discarded due to the climatic conditions prevalent in Central Europe and because of logistic reasons as: - restrictions in oil types - abandonment of oil changing activities necessary for a very short period of the year only.

4.1.2 SYSTEM DESIGN

Geartrains and bearings cause the highest portion in start up drag. A reduction can be achieved by:

4.1.2.1 SEPARATION: Reducing the number of components operated during cold start up by:

- multi-spool engine design - use of clutches and freewheels.

4.1.2.2 DELOADING: - Engine oil pump deloading during start phase - Accessory hydraulic pump deloading during start phase

4.1.2.3 MINIMIZATION OF GEARTRAIN LOSSES BY - stacked design of accessories such as fuel and oilpumps - avoidance of niches and depots where oil can be hidden - dry sump oil system design and shielding of the geartrain in order to reduce churning losses.

4.1.2.4 ROTOR DRAG REDUCTION Appropriate clearances/material contraction rates need to be provided in order to avoid brushing or blocking of the rotor. In order to reduce the rotational moment of inertia, separation has to be considered by: - multi-spool engine design - use of clutches and freewheels. In order to reduce the torque demand of the compressor, deloading by use oE inlet guide vanes/bleed valves should be considered.

4.2 FUEL AND AIR MANAGEMENT Matching of all components under consideration of various influences must be ensured in order to enable a reliable light up of the engine. AS effects are analysed in detail in Ref: 2, Ref: 3 and Ref: 4, indication of the main subjects is considered to be sufficient in this paper. 14-4

4.2.1 TYPE OF FUEL Introduction of the fuel Nato Code F34(JP8) with a flamepoint of at least C38OC and a destillation range of 130’C to 300-C as a basic engine fuel for German Forces instead of Nato Code F40 (JP4) with a flamepoint of below - 20°C and a destillation range of 50°C to 25OPC (Wide cut) aggravated the engine demands in low temperature starting.

4.2.2 SYSTEM DESIGN Capabilities of cold start up are mainly influenced by - compressor delivery characteristic - combustor design - fuel vapouriser-/atomiser-capabilities - number and position of starter jets - start up and acceleration fuel control schedule - number and position of ignitor plugs - ignition energy provided by a soaked and loaded battery as power source.

4.3 STARTER ASSISTANCE AND ENGINE RESISTANCE/ASSISTANCE The most important aim of the overall process called engine start is to enable a reliable and a stable circumferential light up of the combustion chamber. Ref: 5 describes an annular combustion chamber with a wide fuel/airflow operational range of 10 to 40% of engine speed, for example. The start window of each individual engine needs to be met by tailoring the engine starter characteristic. Fig: 3 shows a typical engine resistance/assistance characteristic with a starter characteristic and Fig: 4 typical different engine resistance/assistance characteristics of various engines in cold operation. Increasing engine speed leads to higher torque demand on the engine up to the light up point when combustion takes place and turbine energy starts to assist the engine run up. At higher speed the resistance reaches zero and a positive torque of the engine commences. The typical speed-torque relation of an engine starter is a decending, linear slope with maximum torque at zero speed. These characteristics are well known from turbines and some types of electrical motors. Starter size and characteristic must meet the requirements made by: - maximum drag torque of the engine - maximum drag torque of the accessories - start assistance of the engine - load characteristics of the accessories when hydraulic pumps and generators are getting on line - rotational moment of inertia of the engine - rotational moment of inertia of the accessories. On the basis of the knowledge of the resistance/assistance characteristic and the sum of the rotational moments of all components related to one shaft (for example the power take off shaft), the required starter assistance can be calculated in order to meet the weapon system requirements. Dependent on weapon system task and environmental range, different start systems are in use.

4.3.1 DIRECT MECHANICAL SYSTEMS Main engine starts are carried out via one or two shaft APUs, reduction drives, and on some weapon systems by torque convertors. Because the APU power output characteristic is dependent on air density and fuel control characteristic, power output in cold environment increases and is therefore able to cope with th.e higher demand of power required. 4.3.2 PNEUMATIC SYSTEMS 4.3.2.1 PRESSURE BOTTLE OPERATED Provision is made for impingement operation of the turbine or of the compressor wheel, or a separate starter motor is provided. These are the systems often used for small engines. Low temperature performance can only be ensured, when especially extreme care is taken of the water extraction system of the bottle filling compressor. The operating media air is stored under high pressure. Rapid expansion will not be followed by the humid portion of the alr. The resulting free water can freeze and can create trouble in pressure control valves, control valves and nozzles.

4.3.2.2 APU OR GROUND CART OPERATED The power available is substituted by the characteristic mentioned in Para 4.3.1 and an advantage in respect of less run up drag in comparison to direct mechanical drive systems due to the almost gearless design is provided. 4.3.2.3 ELECTRICALLY OPERATED. The main power source is the aircraft battery. Because of the decreasing power output in a cold environment, battery-powered systems can only fulfil the system requirements when the battery is properly sized and service usage allowance and charge factor are taken into consideration. A common factor for service usage allowance is 0.8 and for charge 0.9. In other words, 0,8 times 0.9 = 0.72 , which means a battery of 12% capacity shall be considered for sizing and demonstrating system performance.

Because of the limited power available, battery powered starting systems are commonly used only up to the 1500 KW/15 k Newton engine power class. 4.3.2.4 HYDRAULICALLY OPERATED A high amount of energy can be stored in hydraulic accumulators and recalled even in extreme cold environment. The special effort required is normally applied only when appropriate utility systems are already on board and extreme environmental conditions have to be met.

5. ENGINGE STARTING SYSTEM PERFORMANCE ASSESSMENTS 5.1 PNEUMATICALLY AND ELECTRICALLY POWERED As an denotation of different engine start system capabilities, the respective speed ratios, each calculated from dry crank speed achieved at ISA temperature divided by dry crank speed achieved at -15°C. are shown in Fig: 5. The WILLIAMS WR 2-6 engine is equipped with a pneumatic starting system. The KHD T312 and LARZAC 04 engines are equipped with a battery powered electrical starting system.

5.2 DIRECT MECHANICALLY POWERED

As shown in Fig: 6 and already described in Ref: 6, the TORNADO aircraft is equipped.with a mechanical secondary power system. For engine starting, APU or main engine power is used to operate a torque converter, the turbine section of which is connected to the main engine power take off shaft. cold temperature testing revealed a temperature dependency of the torque output as shown in Fig: 7.

The warming up behaviour was investigated. Fig: 8 5hOWS the warming up curve for the left hand and right hand gearbox of the secondary power system respectively. Temperature increase at the end of the curve was caused by torque converter operation. The different warming up behaviour of the gearboxes is mainly caused by the integrated oil system of the right hand gearbox and the APU when hot return oil flow of the APU shortens system warm up. It was determinded from test results that in order to fulfil main engine assistance demand, an oil temperature of at least +30°C was required. A warming up procedure was created and recommended to the Services. The influence of viscosity in torque converter output torque was eliminated in the meantime by modification action mainly by improving the suction conditions of the scavange oil pump.

ENGINE PROTECTION DURING START UP

As much as engine resistance increases and starter assistance is degraded in cold environment, engine assistance has to take over a higher portion in order to accelerate the engine properly up to idle speed according to the system requirements. That will result in a higher thermal loading during the start phase. In order to protect the engine from thermal overloading special arrangements were built in. Typical differences in speed and temperature during start up of an engine under ISA and cold conditions are shown in Fig: 9 and Fig: 10. The protection system has to cope with this wide range of engine parameters. In order to safeguard the engine start under extreme environmental conditions, the following measures were created: SCOPE OF PROTECTION PROTECTION AGAINST TOO LOW STARTER ASSISTANCE Too low starter assistance can be caused by seizure inside the transmission train or degradation of the power source, as: - exhausted battery - exhausted accumulators or air pressure bottles - flow starvation of the power transmission media.

PROTECTION AGAINST "NO LIGHT UP" Achieved by minimization of fuel injected in o:rder to avoid overtemperature and to save starter system energy.

6.1.3 PROTECTION AGAINST HUNGSTART A hungstart is caused mainly by incomplete light up of the combustion chamber or low starter assistance. 6.1.4 PROTECTION AGAINST OVERTEMPERATURE The necessity has already been mentioned in Para 6.

6.1.5 PROTECTION AGAINST INCORRECT RESTARTING Proper starter engagement and sufficient drainage time have to be ensured for consecutive engine starts.

6.2 APPLICATION OF ENGINE PROTECTION SYSTEMS 6.2.1 FIRST GENERATION OF ENGINE PROTECTION DURING START Manual protection is given for early developed systems. Speed and temperature gauges have to be observed by the operating personell according to the operating instructions, and hJ.gh pressure fuel cock opening/closing and starter shut off have to be initiated manually. Because of the slow beginning and very dynamic behaviour at the end, the outcome of the start, especially in cold operation, depens to a great extent on the skill of the operators.

6.2.2 SECOND GENERATION OF ENGINE PROTECTION DURING START Combined, automatic and manual protection is provided. Engine start up operation within the limits is enabled by the use of some threshold values functioning as watchdogs and the assistance of the operator. Electronic Control Units of discrete or integrated technology were used. Some typical threshold values which are logically combined in the Electronic Control Units are explained below:

6.2.2.1 PROTECTION AGAINST TOO LOW STARTER ASSISTANCE This protection is provided by speed observation at the end of the starter only acceleration prior to initial ignition of the engine. Because of the low acceleration customary in cold environment, the threshold value has to be tailored to these conditjons.

For example:

6 seconds after start initiation the engine speed must to be equal or greater than 8%.

6.2.2.2 PROTECTION AGAINST "NO LIGHT UP" such protection is provided by temperature or speed observation of the engine.

6.2.2.2.1 TEMPERATURE OBSERVATION A temperature threshold value indicates the begin of the light up of the combustion chamber. For example: 12 seconds after start initiation the exhaust gas temperature must be equal or greater than 150'C.

6.2.2.2.2 SPEED OBSERVATION A speed threshold value indicates the begin of the light up of the combustion chamber.

For example: 15 seconds after start initiation the engine speed must be equal or greater than 25%.

6.2.2.3 PROTECTION AGAINST HUNGSTART This protection is provided by engine acceleration or speed observation.

6.2.2.3.1 ACCELERATION OBSERVATION An acceleration threshold value indicates a minimum acceleration.

For example: 20 seconds after start initiation and up to idle speed the acceleration of the engine must be equal or greater than 0,5% per second.

6.2.2.3.2 SPEED OBSERVATION

A speed threshold value indicates the acceleration above a typical hungstart area.

For example:

20 seconds after start initiation the engine speed must be equal or greater than 40%. 14-8

6.2.2.4 PROTECTION AGAINST OVSRTEMPERATURE As shown in Fig: 10 the temperature level is much higher during cold start up of an engine than during all other starts. Because of the very dynamic nature of the engine parameters during the start up, an overtemperature protection has to be provided. In order to meet stringent acceleration requirements, metering of the fuel is scheduled accordingly. This results in a rapid increase of the gas temperature during start up to a high level.

6.2.2.4.1 PROTECTION AGAINST OVERTEMPERATURE BY TEMPERATURE OBSERVATION

6.2.2 4.1.1 TEMPERATURE LIMIT A temperature threshold value allowing a safe operating range for start UP.

For example: Engine exhaust temperature must be less than 800°C.

6.2.2 4.1.2 TEMPERATURE/TIME LIMIT A temperature versus time threshold value is built in For example:

Excursions of the exhaust gas temperature up to 11OO'C are allowed for 2 seconds duration.

6.2.2.4.2 PROTECTION AGAINST OVERTEMPERATURE BY ACTIVE CONTROL

6.2.2.4.2.1 BY TEMPEWTURE OBSERVATION The fuel scheduled to the engine is uneffected up to the maximum allowed gas temperature. When the gas temperature threshold is reached, this results in fuel bypassing in order to limit the gas temperature to the maximum allowed. Because of the lower portion of fuel metered tO the engine, acceleration is decreased.

6.2.2.4.2.2 TEMPERATURE OR ACCELERATION CLOSED LOOP CONTROL An exhaust gas temperature or acceleration schedule versus engine speed is provided. The fuel is metered according to this schedule dependent on the environmental conditions. This method demands the greatest effort, ensurlng in this way most careful treatment of the engine during starting operation.

6.2.2.5 PROTECTION AGAINST INCORRECT RESTARTING Threshold values for recycling of a start are built in. For example:

- For a restart engine speed must be less than 2%. - For a restart temperature must be less than 150'C. - Intervals between successive starts must be greater than 2 minutes.

6.2.3 THIRD GENERATION OF ENGINE PROTECTION DURING START

Fully automatic control is provided. After initiation of the start sequence by the operater or programmed start up, observation and protection were performed automatically. 14-9

7. CONCLUSION Tests of Official Test Centres are different to those of the contractor. Test techniques are generally the same but test objectives are different. These are not aimed at research or development, but are addressed to a final product offered to the government for Service use.

Knowledge of the boundaries of engine parameters during cold start up as well as engine observation and protection facilities and their interaction are most important for the planning and conduct of engine cold start tests.

8. RECOMMENTATION

Evaluation of the proper functioning of observation and protection facilities is required. Any interference of the observation and protection system during start up must be displayed to the ground crew for quick correcting action, to enable the aircraft to be returned to flight status as soon as possible.

It should be noted that under cold or ISA conditions speed and exhaust gas temperature build up in substantially different ways. A threshold based observation and protection system should therefore be capable of adapting its threshold values accordingly.

Fully automatic A/C run up systems should be supported by a sub-program capable of at least 3 subsequent engine start attempts. It must be ensured that a reset of the automatic preflight sequence of the airborne system is not required when a failed start occurs. Minimization of thermoshock loads of the sensitive components of the engine can be achieved by use of advanced observationlprotection systems.This method can be applied not only to cold operation but throughout the whole operational range of the weapon system, resulting in increased life. This means that the availability of the system is increased and reduction of cost will be achieved.

REFERENCES

Pirl, K.j. Experlence with the KHD APU T312 for a Modern Fighter Type AGARD-CP 324; Page 12-1 to 12-15

Alternative Jet Engine Fuels AGARD-AR-181-VOL.11

Combustion Problems in Turbine Engines AGARD-CP-353

Combustion and Fuels in Gas Turbine Engines AGARD-CP-422

Collin, K.H.; A Small Annular Cumbustor of High Power-Density, Wide Operating Range and Low Manufacturing Cost AGARD-CP-422, Page 42-1 to 42-12

Hausmann. W.; Pucher, M.; Weber, T.; Secondary Power System For Fighter Aircraft Experience Today and Requirements For A Next Generation AGARD-CP-352, Page 3-1 to 3-15 14-10

LIST 1

DOCUMENTATION TO DEFINE AND EVALUATE SYSTEM CAPABILITIES

STANAG 2895 Extreme Climatic Conditions and Derived Conditions For Use In Defining Design/Test Climatic Conditions For NATO Forces Materiel

STANAG 3518 Environmental Test Methods for Aircraft Equipment and Associated Ground Equipment

MIL-STD-210 Climatic Extremes for Military Equipment

MIL-STD-810 Environmental Test Methods And Engineering Guidelines

MIL-E-5001 Engines. Aircraft, Turbojet and Tur'bofan, General Specification for

MIL-A-87229 Auxiliary Power System, Airborne

MIL-P-85573 Power unit, Aircraft, Auxiliary, Gas Turbine, General Specifiaction for. TEMPERATURE IC1

-26 HOURS

FIGURE 1 94 HOUR COLD SOAK TEST MANCHING A1RFIEI.D 11 .-15.01.1985

TIME [HI 20 I

U -9 -10 -11 -12 -13 -14 -16 -16 -17 -18 -19 -20 -21 TEMPERATURE iC!

FIGURE 2 TEMPERATURE FREQUENCY DISTRIBUTION ASSISTANCE I

I RESISTANCE ENGINE SPEED

A STARTER CHARACTER. -C- ENG.CHAR.ISA

FIGURE 3 TYPICAL ENGINE RESISTANCE/ASSISTANCE AND STARTER CHARACTERISTIC

I RESISTANCE ENGINE SPEED

- ENG.CHAR.COLD +ENG.CHAR.COLD 4- ENG.CHAR.COL0

FIGURE 4 TYPICAL ENGINE RESISTANCE/ASSISTANCR CHARACTERISTIC OF DIFFERENT ENGINE DESiGNS TN COLD OPERATION CRANCSPEED RATIO

'II

WILLIAMS WR 2-6 0KHD T312 LARZAC 04

FIGURE 5 CRANKSPEED RATIO ISA/-lS'C

PTO pm

FIGURE 6 DIRECT MECHANICAL DRIVEN SECONDARY POWER SYSTEM OF TORNADO AIRCRAFT TORQUE I

HP-SPOOL [RPM]

-MODIFIED / WARMED UP +UNMODIFIED

FIGURE 7 TORQUE CONVERTER CHARACTERISTIC

TEMPERATURE IC1 8o I

L" TIME

-L/H Gearbox +R/H Gearbox

FIGURE 8 GEARBOX WARMING UP CURVE SPEED

TIME

-ISA +COLD

FIGURE 9 TYPICAL ENGINE START SPEED RELATION

TEMPERATURE 1

TIME

-IS4 +COLD

FIGURE 10 TYPICAL ENGINE START TEMPERATURE RELATION 14-16

Discussion Author: Adequate configuration control has to be obtained. The 1. C. Meyer, French MOD minimum Derformance is soecified and has to be Un problime qui se fair giniralement en essais de demonstrated. Performance of items of series production qualification est celui de la dispersion des performances has to include the production scatter and has to be above the entre matiriels d’une mime sene. Prennez vous en compte minimum specified. For demonstrating system cet aspect pour la qualification des capacitis dedemarrage a performance, for example the battery, the minimum basse temperature? performance required is used. 15-1

CLIMATIC CONSIDERATIONS IN THE LIFE CYCLE MANAGEMENT OF THE CF-18 ENGINE

by Capt. R.W. Cue Canadian Forces National Defence Headquarters Ottawa, Canada, K1A OK2 Attn: DFTEM 6-3-2

D.E.~ ~ MUir ~~ GaSTOPS Ltd. 1011 Polytek Street Ottawa, Canada, K1J 9J3

SulMARy The Canadian Forces have developed an Engine Parts Life Tracking system (EPLTS) to define the scheduled maintenance requirement of CF-18 aircraft engine components. Up to 64 components are tracked by this system, 26 of which are life limited on the basis of eight different Life Usage Indices defined by the engine manufacturer and evaluated during each operational mission by the aircraft's Inflight Engine Condition Monitoring System. Data on the rates of component life consumption collected by the EPLTS during a full 12 month time span have been analyzed. The manner and extent to which seasonal effects might influence these life consumption rates and hence the life cycle management of the engine are present.ed and discussed.

NOW3NCLATURE AOF - Aircraft Data File N2 - Compressor Rotor Speed CFB - Canadian Forces Base N2F - Full N2 Cycles CR - Count Rate N2P - Partial N2 Cycles EFTC - Equivalent Full Thermal Cycles OCM - On-Condition Maintenance EPLTS - Engine Parts Life Tracking P3F - Full Ps3 Cycles System EOT - Engine Operating Time p3P - Partial Ps3 Cycles

ELCF - Equivalent Low cycle Fatigue Ps3 - Compressor Delivery Static Pressure HPC - High Pressure Compressor PLA - Power Lever Angle HPT - High Pressure Turbine RS - Relative Severity IECMS - Inflight Engine Condition SRF - Stress Rupture Factor Monitoring System IRP - Intermediate Rated Power TAMP - Time at Maximum Power LCF - Low Cycle Fatigue TMT - Turbine Metal Temperature LCMM - Life Cycle Maintenance Manager T1 - Engine Inlet Temperature LUI - Life Usage Index tR - Time to Creep Rupture LPT - Low Pressure Turbine EC - Creep Strain MDRM - Maintenance Data Recorder dec/dt - Creep Strain Rate Magazine MOB - Main Operating Base 15-2

INTRODUCTION The air el-ement of the Canadian Forces operates 124 CF-18 fighter aircraft from Main Operating Bases located at Cold Lake, Alberta: Bagotville, Quebec: and Baden- Soellingen, West Germany. The operational role of the CF-18 requires that it operate out of bases in various geographical and climatic areas. For the CF-18's Canadian sovereignty and North American Air Defence (NORAD) role, the aircraft operates from far north deployments at Inuvik, Yellowknife, Rankin Inlet and Iqaluit, North West Territories to training deployments at various bases in the southern U.S. The Canadian Forces' NATO commitment has the CF-18 operating out of the European theater. This puts the CF-18 in an operating environment which ranges from the very cold weather in the far north, to the moderate climate of Europe, to the extreme heats of the southern U.S. The acquisition of the CF-18 by the Canadian Forces has also resulted in a new approach to fighter engine maintenance. Whereas previous engines were overhauled on a fixed time interval basis, the General Electric F404-GE-400 engine is for the most part maintained on an "on-condition" basis. In particular, the life usage of critical components within the engine is continuously evaluated by the aircraft's Inflight Engine Condition Monitoring System (IECMS). A ground-based Engine Parts Life Tracking System (EPLTS) has been developed by the Canadian Forces to record the accumulated life usage, configuration status and maintenance history of these conponents. The EPLTS also is used to forecast the fallout or removal time for each lifed component based on historical life usage accumulation rates, This paper describes the CF-18 IECMS and EPLTS and examines the effects that climatic variations can have on the rates of F404 component life usage accumulation. The influence of these effects on the life cycle management of the engine are also discussed.

BACKGROUND The CF-18 aircraft is powered by two General Electric F404-GE-400 engines, a low bypass, twin-spool turbofan fan engine with mixed flow exhaust and afterburning. The F404 is the first engine in the Canadian Forces to be maintained under a formal On- Condition Maintenance (OCM) program. Under this program, engine components are replaced or repaired based upon their condition. For the replacement of life limited parts, "scheduled" maintenance is mandatory when the life limit is reached. For the remaining components, "unscheduled" maintenance is performed when an in-service problem has occurred or when degradation is observed. The OCM maintenance concept therefore requires that an engine condition monitoring capability be developed to monitor and predict component life usage rates and to provide early warning of progressive engine deterioration. To meet these requirements, each CF-18 is equipped with a fully integrated Inflight Engine Condition Monitoring System which automatically records data on engine life usage, limit exceedances and take-off performance for post-flight analysis. The IECMS logic wa5 developed by General Electric and is implemented as software on one of two aircraft Mission Computers. As the aircraft performs each mission, engine life usage is continuously calculated by the IECMS using eight different Life usage Indices (LUIS) related to the low cycle fatigue, thermal fatigue and creep damage experienced by the engine components. AS such, these LUIs provide a basis for establishing individual component life limits, rather than the general and more conservative life limits applied to older engines. Table 1 presents a brief description of the CF-18 engine LUIs. In addition to various other aircraft mission data, the running sums of each LUI are stored in the Mission Computer memory and are recorded twice per flight to a removable Maintenance Data Recorder Magazine (MDRM). The gathering and processing of MDRM data is accomplished by a network of ground-based processors as illustrated in Figure 1. Each squadron is equipped with a Ground Data Station which stores a copy of the MDRM contents in an Aircraft Data File (ADF), erase5 the tape for return to service and provides hard copy reports of specific data records. The ADFs are automatically trans- ferred from each squadron to a central Base computing Facility and eventually archived onto 9-track tape. Prior to archiving, IECMS data are automatically extracted from each ADF for the EPLTS and other engine application programs.

The Engine Parts Life Tracking System is an application program developed by the Canadian Forces 'to process and analyze IECMS life usage data and to aid the maintenance decision making process of retiring critical engine components from service before they fail. The EPLTS is a computerized database system which includes a suite of computer programs for: automated usage data entry, interactive entry of maintenance data, validation of maintenance data by configuration and compatibility checks, and interactive data retrieval for reporting purposes. The primary component of the EPLTS is its database, which contains the component part numbers, serial numbers, their relationship to other components and the life usage status of each part. The system tracks by serial number the location and current accumulated LUIs of 24 life-,limitedand 40 logistically critical components of each F404-GE-400 engine. 15-3

The EPLTS also tracks LUI accumulations for each mission flown and, by using historical LUI accumulation rates, the system can forecast lifed component "fallout" from 1 month to 30f years. The historical LUI accumulation rates have been found to be dependent on three primary factors: 1. mission-to-mission variations 2. pilot-to-pilot variations 3. climatic variations Although mission-to-mission and pilot-to-pilot variations are the predominant causes of scatter in the LUI accumulation data, the mean or average accumulation rates of certain LUIS also exhibit significant variations due to climate. As previously noted, the CF-18 operates from main, forward and deployed operating bases whose locations encompass most of North America and Western Europe. The aircraft and its engines therefore experience considerable variations in climatic factors such as ambient temperature, ambient pressure, humidity and air quality. Each of these factors can be expected to influence engine component life usage rates: however, the factor which is by far the most dominant for CF-18 operations is ambient temperature. During winter months the aircraft routinely operates from Cold Lake and Bagotville at ground level ambient temperature conditions of -20 to -30°C. Conversely, during the summer months the ambient temperature at these bases is typically 20 to 30'C. Enaine cvcle temoeratures (and to a lesser extent rotor soeeds) are deDendent on tive of the extremes of CF-18 operation, show the estimated TI variations over the CF-18 flight envelope on MIL-STD-21OA POLAR (-26.5'C) and TROPICAL (32.1-C) days. It is evident from these figures that the engine inlet temperatures experienced on the POLAR day are significantly less than the TROPICAL day temperatures and that TI ranges from a maximum of approximately 12O'C (TROPICAL day, high speed) to a minimum of -30'12 (POLAR day, low speed).

CLIMATIC EFFECTS ON ENGINE LIFE USAGE The gas path components of a military aero engine are subject to extremely high cyclic stresses which result in a finite life for many of these components. The cyclic stresses arise primarily from the rotor speed, temperature and pressure variations which occur within the engine: thus, engine inlet temperature variations of the magnitude described above can be expected to affect life usage rates, particularly for hot end components whose lives are governed by thermal stresses or hold times at elevated temperatures. The traditional measure of life used for aero engines is Engine Operating Time (EOT). Experience has shown, however, that EOT is a relatively poor measure of life used for engines subject to variable mission requirements such as fighter engines. To this end, the CF-18 IECMS evaluates a number of Life Usage Indices which provide a more direct measure of the low cycle fatigue, thermal fatigue and creep or stress rupture damage experienced by the engine components. Table 2 identifies the lifed components of the F404-GE-400 and the LUIS which are used to define their life limits. It is evident that low cycle fatigue due to rotor speed excursions governs most of the component lives: however, the HP turbine blade lives are determined by thermal fatigue cycles and are a significant concern of the engine life cycle managers due to their relatively high cost and short service lives. The following sections of this paper describe the thermal fatigue, stress rupture and low cycle fatigue life usage algorithms employed by the CF-18 IECMS and the effects that ambient temperature variations will have on the LUI accumulation rates. a) Thermal Faticrue Thermal fatigue or stress cycles caused by temperature variations are evaluated by the Equivalent Full Thermal Cycles (EFTC) index. The equation used to assess the relative severity, RS, of each thermal cycle is:

RS = 10" (1) where: a = -0.00001058 TMT2 + 0.03076 TMT - 19.211 (2) and: TMT = HP turbine trailing edge blade metal temperature ('C) The HPT trailing edge blade metal temperature is, in turn, determined by the magnitude of a given throttle movement and by the relationship between TMT and Ti shown in Figure 3. This relationship corresponds to the TMT obtained during a throttle advance to the Intermediate Rated Power (IRP) position. For throttle movements to positions less than IRP, a Power Lever Angle (PIA) correction is applied to the TMT calculation. Above IRP, the engine's temperature limiting system maintains TMT at approximately the relationship shown in Figure 3. 15-4

Table 3 summarizes the relative severity of the HFT thermal cycles as predicted by equations (1) and (2). The extreme sensitivity of the EFTC counts to TMT is clearly evident. For example, a cycle to TMT = 960'C is over 20 times more severe than a cycle to 840'C. Figures 4 and 5 present the variations of TMT and EFTC counts with maximum PLA for a single throttle excursion at an altitude of 20,000 ft. and Mach number of 0.8 on both POLAR and TROPICAL days. It is evident from these figures that throttle excursions at the warm inlet conditions result in higher blade metal temperatures and significantly more EFTC counts than on colder days (in this case, by a factor of over 3:l). The sensitivity of EFTC counts to the magnitude of the throttle movement is also apparent. The number of counts rises sharply to a maximum or limiting value at approximately the IRP throttle position. Beyond this position, hot end temperatures are limited by the engine control system as previously noted.

It is evident from Figure 3 that, if the engine's temperature limiting system is functioning properly, the maximum number of EFTC counts for a single throttle excursion is obtained at a T1 of approximately 76'c and a TMT of 946-c. From equations (1) and (2), this maximum value is 2.6 counts. Considering that. the average number of EFTC COUntS for a given mission is roughly 20 and that the maximum number of EFTC counts per mission can be as high as 150-200, it is evident that the number of throttle excursions is the primary determinant of the EFTC accumulation rates.

b) Stress RuDture Turbine stress rupture or creep damage is caused by prolonged operations at elevated cycle temperatures. Creep damage is generally measured by the accumulated or time-integrated creep strain, E,=: < eC = ,/(dr,/dt)dt (3) where the creep strain rate, dc,/dt, is assumed constant for a given temperature and stress level. Additionally, the time to creep rupture, tR, is related to the strain rate by:

dec/dt = l/tR (4) The IECMS Stress Rupture Factor (SRF) algorithm evaluates a count rate, CR, which is proportional to the creep strain rate using the following equations:

CR = ea (5) where: a = -0.oo00321 TMT~+ 0.1038 TMT - 78.13 (6) As in the case oE the EFTC counts previously described, the stress rupture count rate is a function of the HP turbine blade trailing edge metal temperature which, in turn, is determined by throttle position, engine inlet temperature and altitude. An additional correction is applied to the TMT calculation for SRF if the engine exhaust gas temperature exceeds a reference value.

Table 4 gives the variation of stress rupture count rate with TMT as predicted by equations (5) and (6). It is evident from this table that the SRC rate is also extremely sensitive to turbine blade metal temperature and is therefore likely to be affected by ambient temperature variations. For instance, Figure 4 indicates that the value of TMT will vary between 885'C and 925'C from a POLAR to TROPICAL day at PLA = Intermediate, altitude = 20,000 ft. and Mach No. = 0.8. This would result in the SRC rate on a TROPICAL day being over 6 times the POLAR day rate at these flight conditions. c) Low Cvcle Fatique The CF-18 IECMS evaluates full and partial compressor rotor speed, N2, cycles to assess the low cycle fatigue (LCF) accumulations for many of the engine's rotating components. The full and partial cycles are defined as follows; N2 Full (N2F) Cycle: 59-92-59% N2 N2 Partial (N2P) Cycle: 76-92-76% N2 where NZ = 59% corresponds to the Ground Idle speed of the engine. The N2F and N2P counts are combined by the Engine Parts Life Tracking System to give an Equivalent LcF count :

ELCF = N2F + K . N2P (7) where K varies for different components. The expression for ELCF given in equation (7) appears to be based on a linear LCF damage law, where the Value of K is a weighting factor used to assess the damage of a partial cycle relative to a full cycle. Furthermore, since only compressor speed is used to assess LCF, it is implicitly assumed that the fan rotor speed is proportional to N2 and that the effects of temperature on LCF damage to both the high and low pressure turbines is also proportional N2. 15-5

Since the effects of cycle temperature are not directly accounted for in the IECMS N2F and N2P algorithms, it is difficult to predict the influence that ambient temperature variations might have on LCF accumulation rates. As shown in Figure 6, which gives the variation of N2 with Ti at the IRP throttle position, the engine will operate at lower N2 speeds on colder days: thus, LCF damage should be less on colder days (as the cyclic stresses are proportional to the square of rotor speed). However, as indicated on the figure, the present LCF threshold of 92% N2 is set such that only under extremely cold inlet conditions would the number of N2P and N2F counts be affected by ambient temperature. d) Operational Data Analvsis The effects or ambient temperature variations on F404-GE-400 engine life usage have been investigated using operational data collected by the Engine Parts Life Tracking System over a period of 1 year (1989) from two CF-18 Main Operating Bases: CFB Cold Lake, Alberta and CFB Baden-Soellingen, West Germany. The results of this investigation are presented in tabular form in Tables 5 and 6 and in graphical form in Figures 7 and 8. Seasonal or ambient temperature dependent variations are clearly evident in the monthly average accumulation rates of EFTC/EOT and SRF/EOT for both bases. In the case of the thermal fatigue cycle accumulations, the average EFTC/EOT rate at CFB Cold Lake ranges from about 7-10 in the winter months to 17-22 in the summer months: or by a factor of roughly 2:l. At CFB Baden-Soellingen, the EFTC/EOT variations are somewhat less pronounced (because ambient temperature variations are more moderate): nevertheless, a seasonal variation is still apparent. The stress rupture damage accumulations also exhibit significant seasonal variations at both Cold Lake (approximately 4:l variation) and Baden-Soellingen (approximately 3:l variation). Certain deviations in the monthly or season patterns of the LUI accumulation rates are evident at both bases. These are most likely due to variations in types of missions flown during these months. At both bases, there is no apparent seasonal variation in the low cycle fatigue accumulation rates, as indicated by the N2P/EOT data. As previously discussed, this result is to be expected because the upper threshold for N2P cycle definition (N2 = 92%) is likely to be exceeded during major throttle excursions under all but very cold engine inlet conditions. The average annual LUI accumulation rates for the two bases are compared in Table 7. It is evident that an engine at CFB Cold Lake, which has a substantially lower mean ambient temperature as compared to CFB Baden-Soellingen, will accumulate EFTC counts at approximately 75% of the rate of an engine at CFB Baden-Soellingen. However, engines at both bases appear to accumulate low cycle fatigue and stress rupture damage at approximately the same rate. .The similarity of the low cycle fatigue rates indicates that the engines experience similar rates of throttle movement at each base. In view of the colder mean annual temperature at Cold Lake, the SRF results are unexpected. It is possible that the engines at Cold Lake spend comparatively longer periods of time at elevated temperature than those at CFB Baden-Soellingen, even though the rates of throttle movement are similar.

LIFE CYCLE MANAG-NT CONSIDERATIONS In order to effectively manage an engine maintenance program, the Life Cycle Maintenance Manager (LCMM) must have an accurate method to forecast the fallout of life- limited parts. Traditional engine maintenance required that an engine be overhauled periodically based on an accumulation of engine hours. The third line Repair and Overhaul contractor would remove life-limited components and return the engine‘s performance to specification level. Thus, for the maintainers of the engines and the LCMM there was only one component to track and its ‘life‘ was based on aircraft flying time. AS the F404 is maintained under an OCM concept, the LCMM must track the life usage of 64 Components on each operational engine. %Ln the Canadian Forces, there are some 300 engines which amounts to approximately 19,000 tracked components. Without an EPLTS, a considerable burden would be placed on the LcMM of the F404 engine as each component has its own life limit and each engine accumulates LUIS at a different rate. The main method of engine repair under an OCM concept is the replacement of modules within the engine. Although OCM requires fewer spare ‘engines‘,it does require a greater number of spares modules be available to repair the engines. The high technology engine components of today with their high cost and complex manufacturing Processes have led to delivery times of months or even years. with the long lead times, the LCMM for the F404 fleet must look at the requirements to support F404 maintenance throughout the life of the CF-18 program. One of the primary tasks of the engine LCMM is to identify spare part requirements long enough in advance to place orders which will ensure deliveries when required. In performing F404 maintenance at the CF-18 Main Operating Bases (MOBS), the engine bay requires an accurate forecast of maintenance workload for up to the next 6 months. The base must know what they need in terms of manpower allocation and spares so that adequate resources are available to ensure engines are repaired and returned to service as soon as possible. The LCMM and the MOBS must know what affects usage accumulation so that micro-management of assets can be carried out to meet the present and short term change outs (less than one year). 15-6

There are presently no components on the F404 which are life-limited in terms of SRF counts; as such, the only component fallouts which are affected by ambient temperature variations are the High Pressure Turbine blades (lifed by EFTCS). If a base uses a mean annual EFTC rate to project their fallout during the warm months, the blades will be time expired earlier than forecasted. For the MOB, this could mean an additional rotor change out per month. If the MOB doesn't have the spares or the manpower available, delays repairing the engines are encountered. The effect on forecasting is opposite when using the mean annual EFTC rate during the cold months. In this case a forecast could show one more rotor change out per month than would actually occur. Although the reduced workload does not impact operations, it does tie up valuable resources which could be used elsewhere. Over a period of approximately 1 to 2 years, seasonal variations in the LUI accumulation rates are no longer apparent in the long term life usage rates for a given base; therefore, for long term planning at a particular base, climatic effects are minimal. However, since aircraft are not routinely rotated from base to base, fleetwide planning for the procurement of long lead time spares must take into account the base to base variations in EFTC accumulation rates due to mean annual ambient temperature differences.

CONCIUSION The predominant climatic factor affecting the life usage of CF-18 engine components is the variation of ambient temperatures resulting from aircraft operations ranging from the extremes of hot and cold weather experienced in North America to the relatively moderate climate of Europe. The influence of ambient temperature variations on CF-18 engine life usage has been investigated by examining the sensitivity of the life usage algorithms employed by the aircraft Inflight Engine Condition Monitoring System to engine inlet temperature variations and by the analysis of operational data collected by the CF-18 Engine Parts Life Tracking System. The results of these investigations indicate the following: 1. Seasonal variations in ambient temperature have a significant effect on the mean monthly accumulation rates of the IECMS thermal fati.gue and stress rupture life usage indices. These variations arise from the extreme sensitivity of the IECMS EFTC and SRF algorithms to engine inlet temperature.

2. Ambient temperature variations have little or no effect on the mean monthly accumulation rates of low cycle fatigue counts. The full and partial N2 cycles computed by the IECMS do not directly assess the effects of cycle temperature levels on the low cycle fatigue of engine components. Furthermore, the upper threshold for the N2F and N2P cycle definitions is set at an N2 level which is likely to be exceeded during a major throttle excursion under most engine inlet conditions. 3. Significant variations in the mean annual rates of thermal fatigue cycle accumulation can be found between operating bases with significant variations in mean annual temperatures. In the life cycle management of the F404 engine, thr! LCMM must take into account the effects of climatic conditions on LUI accumulation rates in order to ensure the accuracy of maintenance forecasts. If there is no significant variation between bases in the mean annual ambient temperature, then the long term management of the engine fleet is not adversely affected by seasonal changes in LUI rates. In the month to month management of the F404 fleet, the LCMM and the personnel at the MOBS must take into account the effects of seasonal mean ambient temperature variations on the LUI accumulation rates in order to ensure that micro-management of assets can be carried out to meet the immediate and short term maintenance requirements. The conclusions of this paper are based on life usage algorithms developed by General Electric. Although the derivation of these algorithms is proprietary information to General Electric and specific to the F404-GE-400 engine, it is evident that other engines will also experience thermal fatigue, creep and possibly low cycle fatigue damage rates which are also sensitive to ambient temperature variations. It is important therefore that the maintenance managers of these engines take into consideration these climatic effects when forecasting maintenance requirements. 15-7

LU I SYMBOL DESCRIPTION 1. Full N2 RPM Cycles N2F - Counts the number of 59-92-59% N2 cycles

2. Partial N~ RPM Cycles N2P - counts the number of 76-92-76% N2 cycles

3. Equivalent Full Thermal Cycles EFTC - counts the number of HPT blade metal temperature cycles - Weighted according to the magnitude of the cycle

4. stress Rupture Factor SRF - Temperature-weighted time at temperature count - Rate of count accumulation is dependent on HPT blade metal - temperature 5. Time at Max Power TAMP - Monitors amount of time spent at - PLA 2 Intermediate 6. Full Ps3 Cycles P3F - counts the number of 70-407-70 psia pS3 excursions - These counts generally occur only at at high speed/low altitude flight - conditions 7. Partial pS3 Cyles P3P - counts the number of 70-340-70 psia - P~?excursions 8. Engine Operating Time EOT - Records the total time the engine oDerates at or above Ground Idle

Table 1 - IECMS Life Usage Index Algorithms

PART LIFE LIMIT PARAMETER FAN MODULE 1. Stage 1 Fan Disk ELCF* I 2. Stage 2 Fan Disk 3. stage 3 Fan Disk 4. Fan-Blades 5. Aft Fan shaft HPC MODULE 6. Stage 1-2 Spool ELCF 7. Front Shaft ELCF 8. Stage 3 Disk ELCF 9. Stage 4-7 Spool ELCF 10. Combustion Casinq P3F 11. Fuel Nozzle Set ELCF HPT 12. ELCF 13. ELCF 14. ELCF 15. ELCF 16. ELCF 17. ELCF 18. EFTC 19. EOT 20. No. 4 Bearing LPT MODULE 21. Conical Shaft 22. Forward Rotor Seal 23. LPT Disk 24. LPT Blade

* ELCF = N2F + K.N2P Table 2 - F404-GE-400 Engine Life Limited Parts 15-8

TRAILING EDGE BLADE METAL RELATIVE SEVERITY, RS TEMPERATURE. TMT (Emations (1) and (2,L 840 OC 0.15 870 *C 0.35 908 'C 1.00 930 'C 1.75 960 "C 3.67 Table 3 - Relative Severity of F404-GE-400 HPT Blade Thermal Cycles

STRESS RUPTURE TRAILING EDGE BLADE METAL COUNT RATE, CR TEMPERATURE. TMT ('C) (Equations (5) and (6)L

840'C 0.13 10-5 870'C 0.54 10-5 900'C 2.23 10-5 930'C 8.62 960'C 31.42 10-5 Table 4 - Sensitivity of F404-GE-400 Stress Rupture Count Rate to HP Turbine Blade Metal Temperature

I JAN I FEB MEAN GROUND I LEVEL TEMP- 1-17.6 1-13.4 ERATURE ( OC) MEAN THERMAL FATIGUE COUNTS 111.8 6.9 1 (EFTC/EOT) 1 MEAN LOW CYCLE FATIGUE COUNTS 5.9 4.9 (N2P/EOT)

MEAN STRESS RUPTURE COUNTS .09 .05 1 (SRF/EOT) I I

Table 5 - seasonal Variations of CF-18 Engine Life Usage Accumulation Rates: CFB Cold Lake

._ __ 1 JAN I FEB I MAR 1 APR I MAY I JUN 1 JUL AUG iEPT - - MEAN GROUND (LEVEL TEMP- 1 0.6 1 1.4 I 8.9 I 9.7 116.2 117.4 120.7 0.2 .6.0 ERATURE ('C).. I I I I I I I ._ - MEAN THERMAL FATIGUE COUNTS 15.5 12.9 17.2 14.9 20.2 21.4 21.7 3.6 13.9 (EFTC/EOT) ._ - MEAN LOW CYCLE FATIGUE COUNTS 5.4 5.3 6.0 5.5 6.0 6.6 5.8 6.4 6.5 (NZP/EOT) __ - MEAN STRESS RUPTURE COUNTS .13 .07 .ll .10 .15 .18 .20 .20 .19 (SRF/EOT) __ -

Table 6 - Seasonal Variations of CF-18 Engine Life Usage Accumulation Rates: CFB Baden-Soellingen 15-9

CFB COLD LAKE CFB BADEN-SOELLINGEN

1.4 10.8 MEAN GROUND LEVEL TEMPERATURE (~C) 14.3 19.0 MEAN EFTC/EOT 5.6 5.9 MEAN NZP/EOT 0.16 0.15 MEAN SRF/EOT

Table 7 - Average Annual CF-18 Engine Life Usage Accumulation Rates

Figure 1 - CF-18 Ground Data Processing 15-10

POLAR DAY

MACH NUMBER

Figure z(a) - CF-18 Engine Inlet Temperature Variations - POLAR Day

TROPICAL DAY

Figure 2(b) - CF-18 Engine Inlet Temperature variations - TROPICAL Day 15-11

0 loo0I

L I h I 750 -50fi 0 50 100 150 ENGINE INLET TEMPERATURE, T1 (DEG C)

Figure 3 - HP Turbine Blade Trailing Edge Blade Metal Temperature at Intermediate Rated Power

0 950 - TROPICAL v

kI 600 80 90 100 110 120 MAXIMUM PIA FOR THROTLE EXCURSION

Figure 4 - Variation of HPT Blade Metal Temperature with Power Lever Angle for a Single Throttle Excursion 15-12

2.0

TROPICAL

1.5 : 3 0 0 1.0 i

0.5

0.0. 80 90 100 110 120 MAXIMUM PIA FOR THROlllE EXCURSION

Figure 5 - Variation of EFTC Cycles with Power Lever Angle for a Single Throttle Excursion

102

100

84 I -80 -60 -40 -30 -20 -10 0 10 20 30 40 60 N.32 TEMPERATURE, T1 (DEG C)

Figure 6 - Variation of Compressor Rotor Speed with Engine Inlet Temperature at Intermediate Rated Power CTB eao ME - 20 ,

-20 ' JFMAMJJASOND MOMH

30.0

25.0

6 20.0 s E; 15.0 3 10.0

5.0

0.0 JFMAMJJASOND MONTH

WB COLD UKL 0.30

0.25

6 0.20 w 2 % 0.15 2 = 0.10

0.05

0.00 JFMAMJJASOND MOMH

8.00 7.0C 6.0C 5.0C 2 2 4.0( z 2 3.0( 2.0( 1.0(

0.01 JFMAMJJASOND MONTH

Figure 7 - Mean Monthly Temperatures and Life Usage Accumulation Rates for CFB Cold Lake Y 20.0 2 5 15.0 cY 2 g 10.0 0

0_j r 5.0 0

0.0 JFMAMJJASOND MONTH

JFMAMJJASOND MONTH

1.11 JFMAMJJASOND MONTH

+ 0.20 >2 6 0.15 2 0.10

0.05

n on JFMAMJJASOND MOMH

Figure 8 - Mean Monthly Temperatures and Life Usage Accumulation Rates for CFB Baden-Soellingen 15-15

1 Discussion 1. E Kuiper, MOD Netherlands uses historical engine usage data for an individual base. For The Canadian forces calculate their short term forecast per fleet planning purposes, EPLTS uses average data for the 1 base, How is the forecasting performed for long term fleet. It depends on the magnitude of the accumulation rate 1 forecasts for the whole fleet? Also average LUTs per base or difference. If one date is different by a factor of l.l/l, then averages over the entire fleet. an average between the two should be sufficient for long term planing. If the difference is by a factor of 2 or 3/1, then I Author: long term forecast should be done separately (using 1 For short term and long term forecasting at a base, EPLTS individual base rates) and then added together.

16-1

APPLICATION OF A WATER DROPLET TRAJECTORY PREDICTION CODE TO THE DESIGN OF INLET PARTICLE SEPARATOR ANTI-ICING SYSTEMS

by D.L.Mann* and Dr S.C.Tan** *Rolls-Royceplc, Leavesden,Watford W27%Z, United Kingdom **Cranfield Institute of Technology,Cranfield, Bedford, United Kingdom

ABSTRACT engine and scavenge paths (Figure 1). In terms of the engine power performance Over the past five years a dust penalty associated with the fitting of an particle trajectory code has been jointly IPS, the necessary heating energy input may developed by Rolls-Royce and Cranfield. The be 80% higher than for a conventional paper describes recent work on the code to intake duct. include an ice accretion prediction model suitable for use as a design aid for a wide Development of an anti-icing system for variety of gas turbine engine inlets, but such an intake by conventional 'cut and particularly for particle separator try' methods is expensive, time consuming geometries. The calculation of the local and will always tend to lead to a design heat transfer coefficient is seen to be which is energy inefficient in achieving critical to the success of the ice its required performance. Energy efficiency accretion prediction. The paper describes here applies to whatever the chosen the incorporation of a suitable model and anti-icing mechanism is, whether that be by shows that a series of validation tests, a hot gas engine bleed or by some carried out on a full scale rig, have electrical means. Current state of the art satisfactorily verified the code. A second IPS technology gives an anti-icing heating series of validation experiments, carried requirement which represents a not out in an icing facility, further shows the insignificant engine power penalty. prediction model to be appropriate. This paper describes the development of a predictive anti-icing system design code aimed specifically at reducing the expense NOMENCLATURE and time of development by 'cut and try' methods and, from a power penalty point of Drag coefficient view, to reduce current level heating Particle's cylindrical requirements by 50%. coordinate Gas and particle velocity The foundation of the ice accretion vectors prediction model has been software component gas velocities in developed during the course of the the cylindrical coordinate generation of a particle separator design system system (Reference 2,3). The principle component particle velocities components of the model are a Navier-Stokes "' "" "" in the cylindrical coordinate flow solver, the Moore Elliptic Flow system Programe (MEFP) (Reference 4), and the Particle and gas densities joint Rolls-Royce/Cranfield developed QP' 09 particle trajectory prediction program (Reference 5).

INTRODUCTION validation of the prediction code to date has comprised two series of rig tests The susceptibility of rotorcraft to ice carried out on a full scale IPS. The first accretion problems in icing conditions has set of experiments were aimed specifically been observed almost since the aircraft at measuring local heat transfer type first appeared. Rotor blades have been coefficients along the gas passage of the cited as the prime area of concern and much IPS and hence validating the calculation activity has been initiated to investigate method derived during the course of the the icing phenomena in this area. In work, and the second took place in a particufar, the need for suitable Rolls-Royce icing facility and was aimed at predictlve tools to ease the burden of validation of the water droplet necessary design, development and trajectory/ice-accretion calculation certification testing efforts has produced a number of prediction models, an excellent methods. review of which is contained in Reference 1. THEORETICAL MODEL In comparison to the rotor icing problem, the problems of the rotorcraft Flow-field Prediction engine manufacturer may appear to be relatively minor, but with the simultaneous MEFP has firmly established itself at trends towards more aerodynamically the core of the particle separator design efficient (and therefore more damage prone) package used at ~011~-~oyceon the basis of turbomachinery and the fitting Of inertial the accuracy of its predictions and its type inlet particle Separators (IPS). the computer C.P.U. efficiency. The code is job of engineering a suitable Powerplant described in Reference 4 and its use within anti-icing system is becoming more the particle separator design role is difficult. The advent of the IPS makes described in Reference 3. anti-icing more difficult due to the fact that internal surface area must inherently The IPS under consideration during the increase as the incoming flow is first course of the work described here, being a forced to travel over a bend and is then direct engine mounted device, is near bi-furcated as the flow is split into axi-symmetric in form and hence in order to 16-2

reduce computation time MEFP was run in its droplet size over a wide range of values 2D form. Grid generation was fully within the code, for the purposes of this automated and a typical calculation grid study a size distribution was used comprised around 4000 node points, as shown comparable with that measured on icing in Figure 2. tests carried out to UK Def Stan engine certification requirements (Reference ll), that is, with a mean droplet size of 20 microns. Water Droplet Trajectory Prediction The particle equations of motion were derived by assuming the drag force to be HEAT TRANSFER COEFFICIENT (HTC) MEASUREMENT the only force of interaction. This -TESTS assumption is valid if particle density is more than three times greater than that of Experimental Set-up the surrounding fluid medium (Reference 6). In cylindrical co-ordinates, these The experimental set-up for the HTC equations may be written as follows:- measurement tests is illustrated in Figure 3. The rig consists of three detachable sections:- 1. An upstream section which consists of an intake nozzle, followed by a diffuser with an adapter flange at one end. 2. The test section which consists of the IPS inner and Outer wall assembly. The two are connected to the upstream section by a 'v' clamp at the flange. 3. The downstream section which consists of the scroll and splitter lip assembly, where G is the force of interaction = followed by the scavenge and mainline pipe. The scavenge pipe is joined to the mainline pipe just upstream of the suction fan. A throttle valve is located on the mainline pipe and an ISA 1932 nozzle (for flow metering) is set in the scavenge pipe. The centrifugal and Coriolis acceleration terms are represented by the The downstream section was mounted as a end terms on the right hand side of permanent feature of the whole rig. The equations (1) and (2) respectively. The detachable upstream and test sections force of interaction depends on the allowed film probes to be 'stuck' to the relative velocity between the particle and inner and outer walls with considerable the gas flow, and particle size, shape and ease. density. The drag coefficient, C,, is obtained from the standard drag curve for spherical particles. The trajectory of a particle is calculated by a numerical AERODYNAMIC MEASUREMENT solution of the three equations of motion obtained through the Kutta-Felhberq method (Reference 7).

ICE ACCRETION MOE The ice accretion model chosen as most suitable for this study has been that developed by Cansdale and Gent (Reference 8). The method is based on the solution of a set of heat balance equations (kinetic heating, convective cooling, etc.) and is used to calculate the rate of ice growth. HOT FILM PROBE? The water flux distribution is calculated by collecting droplets in a series of The hot film probes specified for the 'compartments' located along the IPS wall local heat transfer measurements were surfaces. In the model, it has been assumed manufactured in Denmark by DISA and were of that water droplets do not rebound upon the 'glue-on' type (ref: 55R47). These impact, however, previous experience in probes are more normally used for skin icing tests has shown that significant friction measurements. The probe consists bounce can occur under certain conditions, of a polymid foil substrate (16 x 8mml on when sufficient impact energy exists to which a thin nickel film (0.9 x 0.1 x (Reference 9). It is intended that 0.001mm) is deposited. The film ends are experiments to investigate water droplet joined to two nickel/silver plates on to restitution ratios will be carried out in which a pair of capper wires (O.lmm the near future to overcome this potential diameter) are soldered. The resistance of shortfall. the probe wire usually varies from 10.0 to 18.0 Ohms and lead (copper wire) resistance The calculation is primarily dependent is usually about 0.2 Ohms. on the local heat transfer coefficient (HTC). The method of Jayatilleke (Reference In order to minimise IPS wall heat 10) has been selected as most suitable to conduction effects, the appropriate walls this application. of the IPS had shallow channels cut into the sheetmetal, these channels were then Whilst it is possible to vary water filled with a low conductivity material and the hot film probes attached, by means of from:- double-sided tape, so as to sit flush with the IPS wall. Q = 12R - 12R(no flow) Therefore, the HTC, h, is given by the HEAT TRANSFER MEASUREMENTS formula:- Copper wires from the probe were h = Q/(Aeff(Ts - T+)) soldered to 5m length cable and plugged directly into a constant temperature where Aeff is the effective heating area of anemometer. The operation of the anemometer the probe and Ts, T are sensor and air is based on the operation of a Whetstone temperatures respectively. The effective bridge with an amplifier which feeds heating area parameter will be further unbalanced signals (caused by the current discussed in a later section. changes in the sensor) to the sensor in order to maintain a constant wire temperature. A combination of two types of anemometer were used:- RESULTS AND DISCUSSION i) 55M system consisting of 55N01 and Figure 5 shows a plot of the predicted 55M10 units, velocity vectors in the IPS produced by the MEFP code. The predicted flow recirculation ii) 56C system consisting of 56C01 and in region A and the.stagnation and 56C17 units. separation in region B have been previously seen during flow visualisation tests and The construction in both systems are are hence true phenomena. The wall static similar but the M-type is more expensive pressure readings for the IPS design point and normally used for high performance flow are shown in Figure 6 alongside the meaurements. corresponding NEFP prediction. The pressure values have been non-dimensionalised The heat transfer measurements were against atmospheric pressure. Agreement is carried out with 6 probes connected to 4 seen to be very good. M-type and 2 C-type anemometers. The output from the anemometers were connected to an The comparison between measured and RMS digital voltmeter via a 55D65 scanner. predicted XTC at the wall locations shown The probe wire temperature is Set from the in Figure 4 is plotted in Figure 7. required overheat ratio, a, which is Predicted HTC’s come from the Jayatilleke defined as: model. The results for the inner wall (Figure 7a) shows good agreement except in a = (R, Rc )/Rc the region of re-circulating flow, A, where - the measured HTC was higher. This where R, and Rc are the hot and cold sensor difference is to be expected because the resistances respectively. The sensor turbulence in a re-circulating flow region resistance, R is then calculated from usually serves to enhance heat transfer rates (Reference 10, 13). R = R2@ (1 + azo (T - 20)) The comparison between experimental and where R is the sensor resistance at predicted values on the outer wall (Figure 20.0deg6: azo is the temperature 7b) also show good agreement except around coefficient of resistance at 2OdegC and T the stagnation region, 8, just after the is the required sensor temperature. Hence maximum diameter point. Again, separation the hot resistance can be calculated as is thought to increase the effective heat follows:- transfer in this region. R, - (1 + a)RaO + Rcable + RL) where Rcable and RL are the cable and lead EFFECTIVE HEATING AREA resistances. The average effective heating area of The overheat ratio was set and the 0.3”’ employed in the calculation of the anemometers weKe balanced after the probes local HTC was found by curve fitting of the had been glued onto the IPS wall. Figure 4 experimental data. This would give an shows the location of the probes on the effective width of the sensor wire as test section. A staggered arrangement was 0.33mm instead of the actual value of adopted in order to avoid disturbances in O.lmm, assuming that the heat is conducted the flow caused by the wires (and heat) in into a rectangular area. This, however, is the upstream probe from affecting about three times the geometrical area of downstream probes. the sensor wire, which has an actual area of 0.09mm’. Many researchers (Reference 14, 15, 161 have encountered the uncertainty of the effective heating area because it CALCULATION OF THE LOCAL HTC depends on the amount of heat conducted into the substrate materials and the The current, I passing through the immediate flow over the sensor. It is probe is calculated from Reference 12 as generally recognised as a difficult and follows:- time consuming problem, and almost impossible to determine in practice due to I = V/(R, + R + Rcable) the small component sizes involved. Previously, it has been concluded only that where V is voltage, R and Rcable are the effective heating area is greater than resistances of the sensor and cable, R, is the geometrical area. the anemometer top resistance, which is given as 50 and 20 Ohms for the M- and C- Defining a factor, Fa, as the ratio of type systems respectively. The heat the effective heating area over the actual transfer to the flow, Q, is calculated geometric area then the work of the above 16-4

previous investigators has shown that Fa is RESULTS AND DISCUSSION greater than unity; thus setting a lower bound value for the factor. Figure 9 shows predicted water droplet trajectories within the IPS for the test In an attempt to try and locate an conditions. The distribution of droplets at upper bound to the factor, a model of the the inlet was assumed to be uniform from sensor using a conventional heat transfer hub to shroud with 155 droplets being code was constructed. The effects of heat launched for each water droplet size. A convection and radiation were ignored and range of water droplet sizes, compliant conduction was assumed to be steady state. with the required 20 micron mean diameter, The non-linear thermal properties of the based on a previously measured Kapton material were accounted for. By characteristic was used. In all, the making these assumptions, a high limit on analysis comprises some 1240 droplet the effective heating area was able to be trajectories. Further assumptions made were calculpted. This value was found to be that the droplets do not bounce and that 0.43" and so (Fa)max becomes:- they do not coalesce. (Fa)max = 0.43/0.09 - 4.78 Figure 10 illustrates the resulting predicted ice accretions on the IPS walls and so, Fa may be said to lie in the (shown dashed) alongside the measured range:- accretions. The three areas of actual accretion: inl-et hub front face, outer wall 1.0 < Fa < 4.78 and splitter 1.ip top surface have all been picked up by 1:he prediction model and Curve fitting of the experimental data generally speaking the agreement in ice gave a value of Fa of 0.3/0.09 = 3.33, shapes is good throughout the IPS. which is consistent with the stated band of valid values The main area of discrepancy occurs at the splitter lip accretion where the It is intended that the analysis using prediction shows a pronounced 'hill' of ice the heat transfer code will be extended in at the splitter lip stagnation point the near future to include the effects of whereas the tests showed a flat covering of convection and radiation and hence allow ice. The difference is believed, in part, for a more precise justification for the to be due to the fact that the actual IPS use of the 3.33 value used in the splitter flow - where the flow is turned experimental curve fitting exercise. 180 degrees into the scroll collector system (Figure 1) - was not simulated in the model. 1t should be stated, though, that if anything, the experimental results ICING TUNNEL TESTS at the splitter lip was a surprise in that other tests on different splitter lip Experimental set-Up shapes have produced the predicted 'hill' of ice at the stagnation point. Overall, the results would appear to indicate that the theoretical model is an appropriate one.

CONCLUSIONS The Hucknall facility set up is The method of heat transfer measurement illustrated in Figure 8, which shows the using hot film sensors has been

... ~~ ~ set-u~)~Ifor~~~ the ~~ ~ IPS ~~ work. The facilitv is a successfully applied to a helicopter engine re-circulating, blowing one ~iiwhichathe IPS intake and has validated a viscous flow test section is held within a closed cell. prediction code and the Jayatilleke HTC The IPS was full scale and operated at its model. Discrepancies in areas of design point flow conditions. The required re-circulating Elow were as expected and flow split into 'engine' and 'scavenge' have been explained. Tests on an unheated lines was achieved using a valve at the IPS in an icing facility have successfully discharge from the engine section. validated the water droplet trajectory Super-cooled water droplets were produced model incorporated into the code. from a spray grid upstream of the IPS inlet and droplet size was calibrated to comply with the UK Def Stan (Reference 11) 20 micron mean diameter. The nozzle FUTURE WORK arrangement to achieve an even distribution of droplets was carried out by trial and At the time of writing, the IPS ice error. accretion studies are on-going. The next stage of work involves the development and The tests were run at -10degC with a validation of a model to predict the heat water droplet concentration of 0.6g/m3. The inputs required through a geometry to duration of the trial was established by prevent the formation of ice. The trial and error to give a satisfactorily validation exercise will be done through a measurable ice accretion whilst not second series of icing tunnel tests, this significantly altering the flow field time employing i heated IPS model. It is within the IPS. A maximum permitted ice hoped, subsequently, to design and test a accretion depth of 6mm was chosen and this full IPS anti-icing system in order to resulted in a test duration of around 10 demonstrate the stated desire to reduce minutes. heating requirements over current levels by 50%. with the available evidence, there is no reason to believe that this objective will not be met. 16-5

15. Hodson, H.P., 'Boundary Layer REFERENCES Transition and Separation Nearing the Leading Edge of a High Speed Turbine 1. AGARD, 'Rotorcraft Icing - Progress Blade', ASME 29th International Gas and Potential', AGARD Advisory Report Turbine Conference and Exhibition, No. 223, September 1986. Amsterdam, Paper no. 84-GT-179, June 1984. 2. Tan, S.C., Elder, R.L., Mann, D.L., Thorn R.I., 'Study of Particle Trajectories in a Turbine Intake,, 16. McCroskey, W.J., ourbin, E.J., 'Flow Angle and Shear Stress Measurements Ninth International Symposium on Air Using Heated Films and Wires', ASME Breathing Engines, Athens, September Paper 71-WA/FE-17, July 1971. 1989. ACKNOWLEDGEMENTS 3. Mann, D.L., Tan, S.C., 'A Theoretical Approach to Particle Separator The authors would like to express their Design', Ninth International Symposium thanks to colleagues at Rolls-Royce and on Air Breathing Engines, Athens, Cranfield for the invaluable assistance and September 1989. advice given during the course of this work. Special mention must go to Joan Moore 4. Moore, J.G., 'An Elliptic calculation for her advice on the use of MEFP. ~inaiiy, ProcedureLecture Series 3DNO. 140; 3D ComputationAGARD andthanks the areUK M~D,also theoffered two tosponsors Rolls-Royce of this plc Techniques Applied to Internal Flows work. in Propulsion Systems, June 1985. 5. Tan, S.C., 'A Study of Particle Trajectories in a Gas Turbine Intake', PhD Thesis, Cranfield Institute of Technology, Cranfield, Beds, UK, 1988. 6. Rudinger, G., 'Flow of Solid Particles in Gas', AGARD - AG-222, October 1976. 7. Ledermann, W., 'Handbook of Applicable Mathematics, Volume 111, John Wiley & Sons. 8. Cansdale, J.T., Gent, R.W., 'Ice Accretion on Aerofoils in Two-dimensional Compressible Flow - A Theoretical Model', RAE Technical Report 82128, 1983. 9. Parker, G.J., Bruen, E., 'The Collision of Drops with Dry and Wet Surfaces in an Air Atmosphere', Proc. Inst. Mech. Eng. (London), 184 (1969-1970), Pt.3G(III), pp57-63. 10. Jayatilleke, C.L.V., 'The Influence of Prandtl Number and Surface Roughness on the Resistance of the Laminar Sub-layer to Momentum and Heat Transfer' Progress in Heat and Mass Transfer, volume 1, Pergammon Press. 11. UK Military Icing specification, DEF Stan 00971. 12. DISA, 'Type 55M10, Instruction Manual, DISA 55M system with 55M10 CTA Standard Bridge', DISA Elektronic A/S. DK-2740. 13. Gooray, A.M., Watkins, C.B., Rung, W., "Numerical Calculation of Turbulent Heat Transfer Downstream of a Rearward Facing Step', Proc. of the 2nd conference on Numerical Methods in Laminar and Turbulent Flow, Venice, Italy, July 1981. 14. Matthews, J., 'The Theory and Application of Heated Films for the Measurement of Skin Friction', PhD Thesis, Cranfield Institute of Technology, Cranfield, Beds, September 1985. 16-6

Fig.1 Section Through IPS

Fig.2 Typical MEFP Calculation Grid Fig.3 HTC Measurement Rig Set-up

Fig.4 Static Pressure Tapping and Hot Film Probe Positions Ih-X

Fig.5 MEFP Predicted Velocity vector plot

REGION "E"

REGION "FI"

Fig.6 Measured Versus MEFP Predicted Wall Static Pressures

'"3 16-9

Fig.7 Measured and Predicted Local Heat Transfer Coefficients

Fig.8 Hucknall Icing Facility

15 In ICING TUNNEL DIAGRAMMATIC LAYOUT

TRICHLORETHYLENE TANKS

KENT CONTROLLER

'SUNVIC' TRIC. FLOW REGULATOR ).LOUVRE VALVE FOR FINE TRIYMINC

HOT hlR INPUT TO TEST .- MODELS VAlVE Fig.9 Predicted Water Droplet Trajectories x10-I

AXIAL DISTANCE (mm)

Fig.10 Predicted Versus Measured Ice Accretions

____ : Predicted '. : Measured \ 16 -11

Author: Discussion At each of the heat transfer probe positions, the experimental data capture technique comprised a half hour 1. D. Breitman, Pratt and Whitney Canada What is the IPS bypass ratio? probe stabilisation period followed by a 30 minute test Do you analyse particle trajectories that are not parallel to period, during which time readings were recorded on circa the engine ccntreline? How sensitive is your design to this? 20 occasions. Thus, a data point on figure 7 describes a mean of the 20 readings taken at a position. Author: Ice accretion tests were run with the IPS operating at At its design point, the IPS of the RTM 322 passes a flow approximately 75 4: design point flows, air temperature was equivalent to around 18% of the core flow through the -10 C and water droplet concentration 0.6 dm’. Water scavenge system. droplet size was set according to UK Def Stan requirements. A comprehensive run of the trajectory code will consider a The coalescing of water droplets which initiated runback is wide range of inlet particle velocities, both in temis of something which is not currently modelled by the code. direction and speed. This has found to be important in Along with water droplets break-up (pre and post impact) replicating experimental data obtained from a cloud-fed and bounce restitution ratios, this is an aspect which is to be inlet. the subject of experimental activities in the near future, Initial particle directions significantly different from that aimed at generating a programmable model. The motion of parallel to the engine axis, do exhibit a trend to be less well single water droplets along the surface is something which separated on the RTM 322 IPS. However, for inlet the code is already capable of handling. trajectories representative of a cloud-fed dust this effect has been seen to be marginal.

2. G. Bianchini, Allison 5. F. Siegmund, Fokker Aircraft Have you validated your predictions with only one Is it the intention to make a fully evaporative anti-icing configuration of inlet particle separators? Do you feel that system or will the required power be such reduced that only the model is flexible enough to be used with other water runs back into the separator? configuration IPS? Are there any future plans to validate the model with Author: different configurations of IPS? Design philosophy, given to current certification requirements, is to prevent any ice formation on gas passage Author: surfaces leading into the engine core passage. The liquid The ability of MEEP to calculate heat transfer coefficients water generated from such a system has not been seen to has beenvalidated, withinRollsRoyce,foradiverserangeof create a runback problem on the RTM 322 IPS. engine components. The trajectory code validations have Other designs may not exhibit such a characteristic and so occurred over a range of RTM 322 IPS geometry standards caremust be takento ensure that theengineis tolerant to the only A more generalized validation exercise is about to take ingestion of potential large quantities of water, and that the place during the heated rig tests, where a range of IPS geometry provides the maximum possible water geometries will be evaluated. separation efficiency. In the second case this may be The design system is considered to be appropriate to the full achieved by suitable wall profiling in conjunction with other range of IPS geometries, both axisymmetric and means. asymmetric, being considered for the next generation of IPS technology

3. C. Scott Bartlett, Sverdrup Technology 6. R. Toogood, Pratt and Whitney Canada The droplet trajectory and impingement prediction (figure Could you please elaborate if your accretion analysis 4) shows no droplet impingement on the splitter plate, incorporates a time-stepping or is it a simple impingement however the following figure (10) shows a prediction of ice location prediction? accumulation. Would you please comment on this apparent Are you considering a building and a sudden process of ice discrepancy? or do you require completely clean surfaces in the RTM 322 IPS? Author: Unfortunately, the trajectory prediction done to produce Author: figure 10 did not generate a trajectory plot. The trajectories The current accretion model is not time-stepping. An ability presented in figure 4 are from an earlier run, performed on to be able to predict ice accretions while continuously an IPS geometry different to that later used on the trajectory accounting for changes being made to HTCs and the flow validation tests. field is not appropriate for the design of anti-icing systems where the objective is to keep surfaces free from ice, as on 4. A. Sutton, BAe the RTM 32. Looking longer term, where controlled How many data points were collected for the determination shedding devices may become available, then a time- of heat transfer coefficients? stepping code would have to be considered. What were the conditions of the ice accretion validation Apologies for the short fall, but the main point of figure 9 test? was merely to highlight that such plots are produceable and Should the impact of water droplet model take into account to graphically illustrate the large number of trajectories the runback of the water which is not frozen? which are involved in a calculation.

17-1

CAPTATION DE GLACE SUR UNE AUBE DE PREROTATION DENTREE DAIR Par R. HENRY et D. GUFFOND 0. N. E. R. A, Direction des Etudes de Synthhe 29, Avenue de la Division Leclerc 92320 CHATIJLON, France

RESUME tridimensionnel axisymiuique, type Euler compressible [2] calcule le champ a6rodynamique dans un &age Lorsque les moteurs fonctionnent au ralenti, la complet (aubage fixe et mobile). Dans notre cas, seules protection des aubes directrices de pr6rotation contre les les donnees dam l'aubage fixe sont n6cessaires. effets du giwe est rendue delicate en raison du calage L'koulement amont ne poss&depas de compsante de important des aubes et de la faible 6nergie dispnible. giration. Utilisant la symktrie de rkvolution, le champ La connaissance de la s6v6rit6 du givrage dans ces est &termin&entre deux aubes voisines, dans un repire configurations ainsi que la masse de glace susceptible (R, Z, 0)(Figure 1). de se dkposer est nkcessaire pour optimiser la r6partition et l'intensite du flux de chaleur utilis6 pur la protection des aubes. La mod6lisation des uajectoires des gouttes d 'eau surfondues, r6alis6e d'abord dans un champ axisymeuique bidimensionnel, puis sur une nappe de courant B proximite de l'aube, permet de calculer l'etendue et I'intensiti de la captation. Un bilan thermodynamique sur la paroi de l'aube permet enfin dkvaluer la forme du givre diposk.

INTRODUCTION

Un des moyens de proteger les aubes de prirotation Fig. I :Ecoulement entre dew aubes :choir du rep2re dentree d'air des moteus conue le givrage consiste ales rechauffer par de l'air pr6lev6 en sortie du compresseur centrifuge (jusqu'B 1,7 95 du debit). Cet air chaud circule Le calcul tridimensionnel des trajectoires de gouttes est B l'interieur des aubes. Lorsque le moteur fonctionne au tr&s coDteux en temps de calcul et n'est pas justifi6 ici. ralenti, la quantit6 de chaleur et la temp6rature sont plus En effet, on peut distinguer deux zones distinctes dans faibles et donc la protection moins efficace sur certaines l'kcoulement : la zone situee en amont de l'aubage, peu parties des aubes. perturbie par la pdsence des aubes, et la zone mouillant L'6tude de ces phenomtnes par la seule voie les aubes. Dans la premBre, on peut considerer expkrimentale itant u&sonereuse, il est interessant de l'koulement bidimensionnel axisym6trique dans le plan simuler nudriquement la formation de givre sur ces (2,R) dam une coupe d'angle Q constante (Figure 2a). aubes afin de determiner l'etendue du dipat, son Dans la deuxihe, l'ecoulement peut &tre6tudi6 sur une intensit6 et sa forme, sans apprt de chaleur, dans un nappe de courant dans le plan (Z,R*O) dans une coupe premier temps. de rayon R (Figure 2b). Les faibles valeurs de la La chaine numkrique de calcul bidimensionnel de forme troisihne compsante des vitesses dans chacun des cas de givre sur profil, d6veloppie I'0.N.E.R.A [l]est justifient cette approche. adapt5e et appliqu6e a ce cas. Elle est composCe de 3 parties distinctes : un calcul a6rodynamique donnant le champ de vitesse et pression autour de l'obstacle, un calcul de trajectoires de gouttes deau dans l'koulement aboutissant au coefficient local de captation qui caracterise l'etendue du givrage et, enfin, un bilan thermodynamique la paroi fournissant Epaisseur et la forme du dep8t. L'application de ces diffkrents calculs B la captation sur une aube est presentee dans ce papier ainsi qu'un exemple de resultat obtenu.

CHATNE DE CALCm Pig. 2a :Mailloge dans le plan (28)d B m'din. 1- Chamo a6rndvnamiaa La simulation numkrique de I'koulement dans un itage Les rCsultats du code aercdynamique sont donc exploit3 de turbine a 6t6 largement apprt5hendie par la Direction successivement dans deux plans perpendiculaires pour de l'Energ6tique B I'0.N.E.R.A. En particulier, un code chaque zone de l'icoulement ainsi definie. 17-2

Fig4 :Approximalion au bard d'attaque Fig. 2b :Maillage de proximiti dans le plan (Z,R*O) au niveau :dupied daube. du rayon moyen, de la tEte .. dhube. -res des

Les figures 3a et 3b nionhent le champ de vitesse dans Eouatlonseene' I& ces deux plans dhomm6s "canal" et "profil - aube", Les equations de I'ecoulement diphasique sont situ& chacun dans une coupe mediane dans la 3" appliquees avec les hypothkses suivantes : dimension. On peut noter que les lignes de courant sont - I'ecoulement dest pas modifie par la presence des trks peu devices par les aubes. Le choix du maillage gouttes (faible nombre). conduit B une imprkcision de calcul au niveau du bord - les gouttes sont parfaitement spheriques d'attaque de l'aube : en effet, un "becquet" doit Stre - la goutte n'est soumise qu'8 la force de trainee. ajout6 B I'avant du profil (Figure 4). Cest un des Le bilan de force applique B une goutte s'ecrit : inconvhients de ce code.

Ft=Mg% (1)

avec : Fg = force de haink Mg= masse de la goutte Vg= vitesse de la goutte t= temps

En exprimant la force de trainee en fonction du coefficient de trainee, (1) s'krit :

avec : pa = masse volumique de I'air Pg= masse volumique de la goutte Fig. 3a : champ de viresse dans leplan '%anal" vr = va-vg (ZR)d 0 midian Rg = rayon de la goutte Ct= f(Re) Cwfficient de trainee fonction du nombre de Reynolds de la goutte Le calcul des hajectoires est effectue de maniere explicite [l]:B pmir de la vitesse de la goutte 8 l'instant t, on calcule I'acceleration (2) et en la supposant constant pendant dt on intkgre (2) et deduit la vitesse de la goutte B I'instant t+dt ainsi que sa position jusqu'i I'impact sur l'obstacle.

Les figures 5a et 5b montrent les trajectoires dans les deux plans definis prkckdemment : dans le plan "canal", le calcul met en evidence un Iiger resserrement des hajec.toires ; dans le plan "profil-aube",situ6 dans une coupe mediane de rayon, les gouttes sont peu dkviies et suivent les lignes de courant. La zone Fig. 3b : champ de vitesse dans leplan "prof/-aube" d'impact s'6tend h I'intrados jusqu'au bord de fuite. (Z,R*O) d R midian 17-3

Fig. 6 :Co@cient local de captation

La courbe representant la position & I'origine en fonction de l'abscisse curviligne dimpact par rapport au pied de pale est repode sur la Figure 7, pour un Fig. Sa :Resserrement des trajectoires dans le plan nombre de aajectoires variant de 10 B 30. "canal".Mach amOnt= 0.33. DiamEtre mkdiun = 20 p La d6rivb de cette fonction donne les valeurs de pC aux diff6rents points dimpact Une fonction de cubic-spleen interpole les valeurs en fonction du rayon et calcule en particulier le facteur de surconcenmtion aux rayons correspondants aux differents plans de coupe du calcul dans le plan "profil-aube" (Figure 8). En raison de l'interpolation de la &rivee, les valeurs varient en fonction du nombre de trajectoires choisi. Cependant, les oscillations sont gonunks pour un calcul effectu6 avec la valeur moyenne de 15 trajectoires, couramment utilisk. On constate que la captation est nulle au niveau du pied de l'aube, ainsi qu'au sommet et que les valeurs dkroissent, pour des Fig. Sb :Trajectoires aufourde l'aube. Rayon mkdian rayons croissants.

Coefficient local de cautation dist. ariqine(m1 Dune mani®6drale, le coefficient local de 1m-0, captation, not6 P, est le rapport, en suivant un tube de i / courant d6fini par deux trajectoires voisines, de la densiti de flux massique deau mi impactant localement, i la densit6 de flux massique rencontr6e par une surface perpendiculaire & l'6coulement 2 l'infini amont I& (Fig. 6).

iam-mr , oraim mt-m I.IIU-OI imt~jwm (186i-m swoi 6.m-m m, Fig.7 :Positions initiale des goulles d l'origine en En 6crivant la conservation du d6bit massique entre fonction de leur position d'impact sur le bord d'attaque deux aajectoires voisines, (3) s'6crit dans un plan : dans le plan "canal". beta 1.30 7 dy- (4) P=, - 1.10 - -.. avec : y.. = ordonnee de la goutte B l'infini amont 1.00iiW - s = abscisse curviligne de I'impact sur le 9.Wi-01 - profil. 8.OOi-01 - 1.00t-OI - On obtient donc P en calculant la d6riv6e de la fonction WUi-01 - reliant, pour chaque goutte, l'ordonn6e B l'infini amont B sa position dimpact sur le profil. Le calcul de trajectoires dans le plan "canal" permet donc le calcul dun premier coefficient local de captation Pc sur le bord d'attaque de l'aube, en fonction du rayon. I.Wi-Ol J obrc.curu(m) O.OOt+W CaractQisant l'effet de resserrement des gouttes leur o.wt+oO 1.m-02 1.WI-02 J.D3-02 rm-(12 iOm[-Ol 6.wOtM arriv6e sur le profil, il s'agit en fait dun coefficient de Fig. 8 :Coefficients locaus surconcentration pC dum surconcentration dont les valeurs peuvent &Ue de le plan "canal".Influence du nombre trajectoires. supkrieures i 1. de 17-4

De la m&memanikre, dans chaque plan " profil-aube", situ6 2 une coupe de rayon differente, on peut calculer Les valeurs de p sont representees en fonction de des coefficientslocaux de captation sur le profil de l'abscisse curviligne pour trois coupes de rayon situees: I'aube (pi) .La position B l'origine est representee en au pied de l'aube, a mi-hauteur et au sommet (Figure fonction de l'abscisse curviligne dimpact sur la Figure 11). L'origine est situ& au point d&t. La corde &ant 9, dans une coupe de rayon m6dian ;l'abscisse 0 diffkrente pour chacune des coupes, la captation debute B correspond ici au bord d'attaque. La courbe croit I'intrados i des abscises differentes. Les valeurs sont fortement i pa& de ce point : la captation sera faibles a I'intrados (=O,l) et augmentent fortement au maximale. La figure 10 montre les valeurs de pi voisinage du bord dattaque pour atteindre des valeurs correspondantes, pour diffirents nombres de trajectoires. sugrieures i 1 en raison du facteur de surconcenhation. La valeur est faible a I'intrados, maximale au bord La captation est nulle sur l'extrados. d'attaque et devient nulle 2 l'extrados. Les oscillations sont minimales pour un nombre de uajectoires &gal 2 bet<; 20. 1.1 Finalement on peut definir dans chaque coupe de rayon, un coefficient local de captation global p &gal au produit des coefficients pi et pc.

p =Bc* pi (5) , , ,

Ce coefficient prend en compte $tla fois la dkviation des gouttes par le profil dans le plan "profil-aube"et le 0.4 resserrement des gouttes dans le plan "canal". 0.3 0.2 0.' ,- .--.-~.-- 0.0 ' -0.~4 -0.03 --a02 -a.oi o.00 0.01 0.02 0.03 0.04 *m.ooio._ - j_u .o/o, I - abrc.c"r"(m) ?.*.o,,7m -----

Fig. I1 :Coefficient local de captation global pen fonction de l'abscisse cwviligne. pow des rayons : de pied dhube. .midian. et de &e Jade.

3 - Bilan thermidvnamiaue

Le bilan thermodynamique permet d'ivaluer la forme du deptit de glace en &terminant la fraction deau captee qui se congkle et la temgrature de la paroi. Fig.9 : Positions initiale des gouttes d l'origine en Lorsque cette dernikre est nettement inferieure a 0" C, la fonction de leurposition d'impact sur l'aube dam une congelation est quasi instantank :la forme se coupe de rayon median. developpe autorx des points de fort coefficient de captation. beto Par contre, autour de 0" C, une fraction deau ruisselle de part et dautre du point dam& et se congkle en aval ; le &@t s'etend pour donner une forme die en come".

La surface de l'aube est dkoupk en facettes suppasees isolees entre elles et n'khangeant de la chaleur qu'avec l'exterieur. Les 6Lhanges thermiques sont lies a la If>[-01 captation d'eau surfondue, a la convection et aux changements de phase eadglace. Deux bilans sont donc 6tablis en surface :un bilan massique et un bilan thermique [l]. Bilao marsh= -040 - 030 -02C -.0!0 0.000 0.O:C Chaque facette est susceptible de recevoir de I'eau provenant de la captation (mi) et du ruissellement de

Fig. 10 :Coefficient local de captation autour de I'el6ment precedent (mrs). Une partie s'kvapore (mv& l'aube. Influence du nombre de trajectoires. 5.OiiOl - h(w/mZ/K) I.Ii+03 -

4.OE+OS - une fraction se congkle (mg)et une autre ruisselle vers 1,litOl- la facette suivante (mB) (Figure 12).

La conservation du bilan de masse donne :

' mi+---"" - mrs t mVls + mg (6) As As

Captation:; i / Vaporisation : ni - I Sur chaque facette, on suppose le bilan en equilibre: 7

(7) j=1

Fig.12 :Bilan massique au niveau de laparoi On ddtermine de manikre igrative la temperature de paroi et la fraction d'eau qui se congele, atin que les deux &pations de bilan (6) et (7) soient verifiees. Ce Bilan thermiane calcul suppose les khanges instantanes et ne prend pas De la m&ne mani&re,un bilan thermique est effectue en compte les khanges par conduction dans la glace et sur chaque facette. On peut distinguer les pertes et les la paroi. gains (Figure 13). Dans le cas d'une temperature de Enfin, on en deduit un taux de croissance et donc, une paroi dgative, les pertes proviennent : forme de glace B un instant donne. - de la convection (qd A ce stade, un nouveau calcul du champ ahdynamique - de l'~vaporation/sublimation(qVls) prenant en compte les perturbations induites par le d+t de glace est gen6ralement effectub Ce n'est pas

~ du dchauffement de l'eau d'impact (hi, possible ici dans la mesure oh le maillage utilis6 dans le calcul aerodynamique ne permet pas de modification Les gains proviennent : au bord d'attaque. La forme obtenue constitue donc une predere evaluation. - de la chaleur latente de congelation (qg)

- du refroidissement de la glace (qf) RESULTATS ET DISCUSS ION du refroidissement de l'eau de ruissellement - (4) Plusieurs simulations sont effectuees 2 partir du m&me - de l'hergie cinktique des gouttes (qd champ aerodynamique CONqOndant h un nombre de Le signe de certains termes peut &treinversd selon les MACH d'environ 0,3 en amont de l'aube et B un calage configurations. des aubes nul, c'est B dire B un fonctionnement en regime avec volets ouvem. Dans un premier temps, on fait varier la temperature

d'dtde ~ 15 "C ?I t 3'C pour une teneur en eau liquide Evaporation: qvls moyenne de 0,6 g/m3, un diadtre volumique median Refroidissement dela glace :(r Energie cinktique :( ,, de gouttes de 20 pn ;la durk de captation est de 60 secondes. Les resultats sont examines dans une coupe Congaation: 4g ,\ 1 ,,/,/:ti " de rayon m6diane et reprdsent6s sur deux aubes Convection : h, voisines. Ruissellement :qr - Influence de la temtxirature

Pour les temperatures de -15 et -10 'C (Figures 15 et Fig. 13 :Bilan thermique d laparoi 16), la glace rmuvre essentiellement le bord d'attaque, atteignant une epaisseur de 3 mm environ. Alors qu'une fine couche (4mm) recouvre l'intrados, l'extrados reste Le coefficient d'dchange par convection est determine ?I propre. partir d'un calcul de couche limite avec rugositk de paroi A partir de -5 "C (Figure 17). la forme se mcdifie : elle (Figure 14). s'aplatit au brdd'attaque pour kvoluer vers une forme 17-6

"en come" 0 "C (Figure 18). Le point darr&tn'est plus recouvert mais l'eau se congde de part et dauue. La largeur de la forme atteint 7 mm. Elle reste confinee autour du bord dattaque.

La glace est encore presente pour des temperatures darr0t IigSrement positives car les temperatures sont toujours nigatives i I'intrados et 2 I'exaados (Figure 19 et 20). Une ipaisseur relativement importante subsiste au debut de I'exaados et I'eau ruisselant sur I'intrados se congkle plus loin vers le bord de fuite. En risum6, on constate une 6volution importante des Lf_ctn_l formes avec la temperature. Fig. 18: - 60 s de captation - iwc = 0.6 glm3 - dmv =: 20 pn - tar = 0 @C-Rayon midian -

L!YJ Fig. 15: - 60 s de captation - lwc = 0.6 glm3 - dmv = 20 prn - tar = -15 "C -Rayon median t3

Fig. 19: - 60 s de captation - lwc = 0.6 glm? - dmv = 20 - tar = + 2 OC - Rayon m6dian

L'"I

Fig. 16: - 60 s de captation - lwc = 0.6 glm3 L""1 - dmv = 20 prn - tar = -10 "C -Rayon median Fig. 20: - 60 s de captation - iwc = 0.6 glm3 - dmv = 20 pm - tar = t 3 "C - Rayon midian

Le code permet deffectuer les calculs de forme dans chacune des 15 coupes de rayon definies dans le code airodynamique, du pied daube ?ila ti@. Dune manikre ginerale, le calcul de aajectoires a monm5 qu'aucune gout@ n'impactaii sur les extrimites :les limites de la zone captation sont donc situkes en retrait denviron 4 mm par rapport ail sommet et i la base. Entre ces deux limites, la comparaison des formes pow trois coupes diffirentes ne r&v&lepas de diffcirence notable (Figure Fig. 17: - 60 s de captarion - lwc = 0,6 glm3 21). La forme n'hvolue que trks piu avec le rayon de

~ dmv = 20p-tar = -5 Oc -Rayon midian - I'aube. 17-7

avec le rayon :l'kpaisseur est identique de la base au sommet de l'aube. L'influence du calage des aubes n'a pu 2tre Btudiee car sa modification nicessite un nouveau calcul akrodynamique 2 partir de nouvelles donnees de profil. Selon l'exemple trait6 ici, l'application de cette chaine -Pied d'awbe - de calcul 2 une configuration de fort calage (60") permettra une meilleure- localisation des dBp6ts de givre. La chaine numerique de calcul de forme de givre sur -Rayon midian. profil dBveloppk ?I l' 0.N.E.R.A a et6 modifiee et adaptee au cas de captation sur un aubage de pdrotation dentrBe dair de moteur. Le champ aerodynamique tridimensionnel est exploit6 successivement dans les deux plans perpendiculaires (ZP) et (Z,@ afin de dBterminer les coefficients locaux de captation dans ces deux dimensions. Un bilan thermodynamique est ensuite effectuB dans une section de l'aube et la forme de - Ttie daube glace Bvaluk.

Une simulation est effect& dans une configuration de Fig. 21: - 60 s de captation - lwc = Obglm3 calage nul (volets ouverts). En gBnBral, la zone de - dmv = 20pn - tar = -5 'C - captation maximale est situ6e autour du bord d'attaque dont il conviendra donc de privil6gier la protection. Influence de la teneur en eau lipide L'intrados est recouvert d'une mince couche de glace, et de la taille des gouttes alors que l'extrados n'est pas atteint.

Une simulation B 0 "C avec une teneur en eau liquide de Les quantites apparaissent insuffisantes pour conduire B 1 g/m3 montre, comme on pouvait s'y attendre, que la une obstruction par recouvrement de plusieurs aubes quantitk de glace est plus importante que dans le cas B voisines. Toutefois, on retrouve l'Bvolution classique 0,6 glm3. Par contre, la forme reste inchangBe (Figure des formes qui tendent B s'6largi lorsque la temperature 22). La severit6 du givrage augmente avec la teneur en augmente et avoisine 0 "C. Le rble prBponderant de la eau liquide. L'influence de la taille des gouttes se dvkle temperature est ici confirm&. n&gligeable.Ceciest en contradiction avec les rBsultats obtenus sur profil d'avion mais s'explique par les fortes La simulation de configurations differentes, en valeurs du coefficient d'inertie en raison des tr2s faibles particulier avec un calage important peut Stre rBalisBe dimensions de l'aube. avec ce code. Elle permetaa de mieux apprehender les problhmes de protection contre le givrage renconm5s n dans ces conditions.

REFERENCES

[l] D.GUFFOM) 0.N.E.R.A - France Captation de glace par une surface non protBgk au cours d'un vol en conditions givrantes. Recherches et etablissement dune methode de calcul - 1981 - R T no 315146 SY Fig. 22: - 60 s de captation - lwc = I glm3 [21 J. P VEUILLOT et G. MEAUZE 0.N.E.R.A - dmv = 20 pm - tar = 0 "C - Ruyon midiun - A 3D Euler Method for Intemal Transonic Flows Computation With a Multi-Domain En r6sum6, l'application du code 2 une configuration Approach AGARDIPEP - Lecture Series no prkise et une &tudenon exhaustive de parmetres 140, Paris, 13 - 14 Juin 1985 mettent en Bvidence la presence de givre, essentiellement autour du bord d'attaque. Pour un [31 L. BRUNET champ de vitesse identique, la prBponderance du Conception et discussion d'un modele de paramhtre temperature est confirm6e :elle influe formation du givre sur des obstacles varies. largement sur la forme du d6pat de glace dans une 1985 - These de doctorat - Universitk de section donnee. Par contre, la forme parait peu Bvoluer CLERMONT 2 -France 17-8

Discussion 3. A. Sutton, BAe Is there any possibility of using this code on rotor blades as well as on stator vanes? 1. K. Broichhausen, MTU My question is also directed to the authors of paper 16. Both Author: codes, the Moore code and the Veuillot code, work either No, this code cannot be used on rotor blades. In this case, with cusps or have some problems in the stagnation zone. the problem is a really 3-D flow field code one. Droplet On the other hand, the icing effects are concentrated near trajectory codes are not developed with effect of giration. the leading edge. Could you please comment on this? 4. E. Brook, Rolls Royce Author: Does the code re-calculate the flow field to take account of That is the problem of the flow field calculation. These the accretion? codes were not developed for this application. The main purpose of the codes is to know the flow field at the exit of a Author: stage. ONERA/energetic direction is developing a new code No, in this application the flow field is not re-calculated. The with a mesh. I will take this problem into account. approximations at the leading edge do not permit it. The final ice shape is not important. The main question is to determine the areas of accretion. 2. S. Riley, Rolls Royce How does the heat transfer coefficient at the ice surface 5. P. Sabla, GEAC warm transiently and how is the time stepping achieved? What inclement weather corrosion does 0.6 g/m’ drop represent? Author: The transfer coefficient is calculated with an integral Author: boundary layer code. The calculation is done on the blade These conditions represent the standard conditions of LWC roughness and introduced by an equivalent sand height and DVM by FAR 25 law (Appendix C). Other conditions roughness. Transition appears with a critical Reynolds can be simulated. Accretion increase with the LUC, but number. The boundary layer is not re-calculated. The there is a weak effect of the DMV because of the value of the calculations are steady, and we calculate an ice growth rate. inertial coefficient. DEVELOPMENT OF AN ANTI-ICING SYSTEM FOR THE T800-LHT-800 TURBOSHAm ENGINE by Gary V. Bianchini Allison Gas Turbine Division, General Motors Corporation P.O. Box 420 Indianapolis, Indiana USA 46206-0420

ABSTRACT parmerships through the Preliminary Flight Fating (I”) phase with competitive procurement from the winning The T800-LHT-800 is a modem technology 1200 hp partners following the Qualification Testing (QT) phase. (900 kw) class turboshaft engine developed for the U.S. by’sLH helicopter and various civil applications. One Following the PFR contract award, the Light Heli- of its significant features is an integral inlet particle copter Turbine Engine Company (LHTEC), a partnership separator (IPS). The presence of an IPS significantly between Allison Gas Turbine Division of General Motors complicates development of an anti-icing system for Corporation and Garrett Engine Division of Allied SignaV protection against the hazards associated with ice forma- Signal Aerospace Company, began a 3-year development tion during operation in environmental icing conditions. program competition. LmCwon the downselect and is currently in the QT phase of the program. Following the Characteristically, inlet particle separators expose completion of QT, scheduled for 1991, Allison and large areas to the incoming air, with the potential for widespread impingement of supercooled cloud droplets Garrett will compete for the T800 Government procure- and resulting accretion of ice. Thermal anti-icing of such ment. Both LHTEC partners will, however, be awarded a an inlet system. with the limited amount of compressor minimum order in each production lot to ensure their bleed air available, presented a significant design chal- continued competitiveness. lenge, which has been addressed through three design iterations, an engine test involving a thermal survey of the In parallel with the LH application engine develop- protected surfaces, and an engine test in environmental ment, other engine applications have been explored. icing conditions. A second thermal survey test and a final Commercial ventures including T800 installation in an environmental icing test, which will satisfy military and Agusta A129 and Westland Lynx are currently being civil certification requirements, will be run on the pursued. In conjunction with the U.S. Coast Guard, the resulting system late in 1990. Army has issued a T800 contract modification providing proof concept flight demonstration for Aem- Theanti-icing system that has evolved from this for a of an series of designs and tests is simple and light weight, spatiale HH65 aircraft re-engined with T800 propulsion imposes a performance penalty of less than 5% in power units. and specific fuel consumption, and protects against harmful formations of ice throughout the envelope ENGINE DESCRIPTION defined by the Military and the Federal Aviation Admini- stration (FAA). In addition, the system provides protec- The T800-LHT-800 engine is a 1200 hp (900 kW) tion for the vulnerable surfaces of the IPS scavenge class, modular design, 310 pound (141 kg), turboshaft system so that separation capability is available to aid in engine. The engine’s core consists of a dual-stage preventing core ingestion of ice shed by the air vehicle. cenmfugal compressor, an annular flow combustor, two- stage gas generator turbine, air cooled with the exception This paper describes the T800 engine, discusses the of the second-stage blade, and a two-stage front drive anti-icing system requirements, design evolution, and power turbine. The engine was initially designed for the validation testing, and presents the final anti-icing system U.S. Army which required a high level of maintainability configuration resulting from the development effort. and battle field hazard resistance. The engine features a PROGRAM DESCRIPTION dual channel full authority digital electronic control, which utilizes a Mil Standard 1553 bus. Other outstanding The T800-LHT-800 engine is under development for features of the engine include attitude capability to 120 the US. Army‘s LH helicopter and has significant poten- degrees noseup, 90 degrees nosedown, and 45 degree roll; tial for commercial application. The Army’s development 6 minutes loss of oil capability; and an engine-mounted program was structured as an effort for two competing IPS. The engine is shown in Figure 1. 18-2

INLET PARTICLE SEPARATOR The IPS system is illustrated in Figure 2.

The engine IPS consists of an inner dome, which The IPS has been developed from detailed flow forms the inner flow path, and the IPS casing assembly, analysis and over 100 hours of both aerodynamic and which forms the outer flow path and also contains the sand ingestion rig testing. The IPS system sand ingestion flow splitter, scavenge vanes, and scavenge scroll. An separation efficiency has been demonstrated on both the engine-mounted blower, which is driven by the accessory rig and engine to be 86%with AC coarse and 96% with gearbox, pulls air through the IPS system to create the Mil-C Spec sand. Additionally, foreign object damage aerodynamic separation. The flow path of the IPS forms a POD)ingestion tolerance, by either separation or foreign 'Y' with one leg leading to the inlet of the compressor object retention, has been demonstrated on the rig. and the second leg connecting to the IPS scavenge system. This flow path shape coupled with the aerody- Although the IPS system has a demonstrated high namic separation helps protect the engine core from separation efficiency, the large surface area, inherent to foreign object ingestion. Foreign material in the engine IPS design, results in the potential of widespread super- inlet is separated and pulled through the IPS and dis- cooled water impingement on the flow-path surface, and charged overboard. The IF'S Creates a large surface area the subsequent ice accretion while operating in environ- exposed to incoming airflow, which is a characteristic of mental icing conditions. This, coupled with the budgeting panicle separating systems. The scavenge vanes and the of anti-icing energy source to reduce performance impact, scroll account for a significant portion of the surface area. created a significant anti-icing system design challenge.

ANTI-ICING SYSTEM REQUIREMENTS

The engine anti-icing system was designed to satisfy both the Military and FAA anti-icing requirements. The T800 Engine System Specification (military) requirements state that the engine's anti-icing system shall prevent ice accretion on all core flow-path surfaces of the front frame and all flow path surfaces of the inner dome. On the IPS casing assembly, all the flow-path surfaces upstream of the scavenge vanes 20% chord point (Figure 3) must be ice free, except light frosting will be allowed on the outer shroud upstream of its point of maximum diameter. These areas are shown in Figure 3. Addition- ally, ice accretion or shedding on surfaces other than the Figure 1. T800-LHT-800 turboshaft engine. described areas shall not: cause damage to any engine components, cause IPS scavenge flow degradation greater than 10%at 10 minutes in the icing condition, cause airflow disturbances that excite harmonic compressor frequencies, cause secondary damage due to reduced clearances between rotating and stationary components, result in stall or surge, adversely affect the power setting parameters, or cawre engine flameout.

The engine pei+ormance loss associated with operating in icing conditions with the anti-icing activated cannot exceed 5% loss in delivered power available at all conditions above Maximum Continuous power and 5% increase in SFC at all conditions above 50% Maximum Continuous power. Engine transient response must be demonstrated to be within specification limits while operating in icing conditions.

The icing conditions that are required to be tested by the Engine System Specification represent a ground fog point (at idle only), a 23'F (-5"C) condition, and a -4" F (-1Y C) condition. Watsrcontent, drop diameters, and exposure times at each of these conditions are shown in Figure 2. Inlet particle separator system. Table 1. 18-3

minute, immediately followed by anti-ice actuation. The engine anti-ice system is manually activated, so this requirement simulates the probable delay in anti-icing system actuation since an encounter with an icing condition may not be immediately recognized by a pilot.

Table 2. FAA icing test points. .. Condmons -1 2 a Engine inlet total temp-- -4 +23 +29 Airflow 'F ('C) (-20) (-5) (-2) Mean effective drop diameter-microns 15 2s 40

Liquid water content -- &' 1 .o 2.0 0.6 IPS scavenge vane cross-section Exposure time -- minutes 10 10 30 Figure 3. Required ice free surfaces. The military requirements preclude the use ofengine oil as the primary anti-icing energy source. During the icing test, it is required that the engine inlet oil tempera- The FAA, similar to the military, requires the engine ture be maintained at or below the minimum oil thermo- operate in the particular icing condition with anti-ice stat setting. Also, the ground fog icing condition (Table 1, actuated for 10 minutes, except the FAA also requires that condition 3) must be run with the engine inlet oil tempera- the engine continue to run in the icing conditions until any ture maintained at or below the engine inlet air tempera- ice accretion has stabilized. This is a significant require- ture, 23" F (-5" C), for a 60 minute duration of cloud ment for engines with inlet particle separators, since ice exposure. These requirements ensure that any oil heat formation in the IPS scavenge flow path is allowable, and transfer effects are minimized, due to the large surface area inherent to an IPS, almost unavoidable. Table 1. Military icing test points. Other requirements, not specifically documented in Conditions the specification, are self imposed to ensure the anti-icing -1 2 3 system is reliable, simple, light weight, maintainable, and Engine inlet total temp-. -4 +23 +23 easily producible. "F ('C) (-20) (-5) (-5) INITIAL ANTI-ICING SYSTEM DESIGN AP- Mean effective drop PROACH AND RATIONALE FOR REDESIGN diameter-microns 20 20 30 Several anti-icing concepts were evaluated. The Liquid water content -- selection criteria of the anti-icing system included consid- dm3 1.o 2.0 0.4 eration of satisfying the Military Specification and FAA requirements, cost, weight, reliability, maintainability, Exposure time -- minutes 10 10 60 and producibility.

The FAA imposes anti-ice system requirements very The initial anti-icing system concept was to anti-ice similar to the military, with some exceptions. The icing only the inner dome and front frame flow path. Using this test conditions specified, as shown in Table 2, differ from criteria, several schemes were proposed and evaluated, the test conditions specified by the military. This is insig- including compressor discharge air anti-icing, engine oil nificant to the design criteria because the anti-icing anti-icing, and compressor interstage anti-icing, the one system is designed to prohibit ice accretion during any chosen for the T800. icing condition encountered within the engine operating and icing envelopes, shown in Figures 4 and 5. During the engine development program, the IPS splitter nose was extended to increase sand separation The FAA also requires that the engine operate in an efficiency (Figure 6). This resulted in the need to actively icing environment without anti-icing activated for 1 anti-ice the flow splitter, instead of relying solely on heat 18-4

Ple91We anlude - km TE90~2188 Figure 4. Continuous icing envelnpi:.

Allitude 1.22 10 6 71 km 114.000-22.000 I!/ horizonlsl exten, 4 82 km 13 "1 30

25 E :20 -cm -5 1.5 i" 9a 1.0 05

Medun drodet dmmeler- microns

Figure 5. Intermittent maximum icing conditions.

conduction from the front frame. Since this forced an anti- power turbine support for wheel cooling. This is sche- icing system redesign, and in order to further reduce risk, matically shown in Figure 8. This concept was rejected it was decided to also actively anti-ice the IPS outer flow because at several flight conditions the flow circuit path and the scavenge vanes. The anti-iced engine flow pressure drop reduced power turbine cooling airflow paths of the initial design and the redesign are compared excessively. Also, the size and weight of the plumbing and illustrated in Figure 6. resulted in weight and maintainability penalties.

Using the redesign concept, four anti-icing schemes The third concept evaluated was a compressor inter- were evaluated. An air and electrical system (Figure 7) stage dual pass system. The anti-ice air would be deliv- was studied. The air would anti-ice the IPS flow splitter, ered from compressor interstage to the front frame struts. front frame, and inner dome, and electrical heaters would The air would then pass through the struts and to the anti-ice the IPS outer flow path and scavenge vane leading edge of the inner dome. Then it would flow aft leading edges. The electrical heater power could be along the inner dome flow path surface, then through the provided by the engine driven PMA, which would require IPS flow splitter, outer flow path, and scavenge vanes. a 50% increase in output capability. This scheme was This concept, as shown in Figure 9, was rejected because rejected due to cost, weight, and maintainability concerns. a significant increase in strut and inner dome passage area would be required to accommodate the increased total The second concept consisted of integrating the flow. This would impose cost and weight penalties. power turbine buffer air with the anti-ice system. Buffer air is supplied from compressor interstage to the rear The fourth concept investigated, shown in Figure 10, support to cool the power turbine wheels. This anti-icing was chosen for the PFR configuration of the T800. It is a concept would use compressor interstage air to anti-ice compressor interstage split flow system. The anti-icing air the inner dome and IPS flow splitter, flow path, and is delivered from compressor interstage to the front frame, scavenge vanes. The portion of the air used in the IPS The air then passes forward thru five tunnels in the front would exit the IPS casing assembly then be routed to the frame to the struts. The airflow splits at this point. Part of 18-5

(CAD), heat transfer analysis, flow system analysis, and engine performance analysis were utilized during the design. The design point of the system was established, using these analytical tools, as -4" F (-20" C), static, sea level. All power conditions were considered. Heat transfer analysis was used to determine the required anti-ice air temperature and flow rates to provide flow path surface temperatures at a sufficient level to retard any ice formation of the super cooled water droplets impinging on flow-path surfaces. The analysis accounted for heat transfer due to conduction, convection, and evaporation. The analysis also predicted and accounted for anti-ice air heat loss due to deliveIy losses through the anti-ice valve and delivery tube. Engine performance loss is directly PFR design extends flow splitter. Flow spliner, outer shroud, and scavenge vanes related to the amount of compressor interstage air used for actively anti-iced. anti-icing. This coerces the heat transfer analysis to accurately predict the anti-icing airflow rates required. Flow analysis was used to determine anti-ice passage size to ensure the proper airflow rates. Using the heat transfer analysis predictions of anti-ice air temperature, the flow analysis accounted for pressure losses to determine the metering orifice sizes. Engine performance analysis was used to predict engine power loss and SFC increase due to anti-ice bleed extraction.

The analytical tools were integrated to provide a basis for the design. It was determined, based on engine performance predictions that a variable anti-ice flow rate was TE~O-2190 required, to ensure minimal engine performance loss Figure 6. PFR anti-iced surfaces. impact. Compressor interstage air temperature increases with engine power level, so more anti-icing energy is available in the air at high power than at idle. Therefore, less flow is required at the higher power to obtain acceptable anti-icing.

The anti-ice flow is controlled by a self modulating, engine mounted valve. The valve is schematically illustrated in Figure 11. The anti-icing valve is an integral part of the engine's acceleration bleed valve which is located on the bottom the engine. It is mounted onto an engine housing which has slots to access the compressor interstage area. The anti-ice valve consists of an elemi- cally actuated solenoid, a pressure switch, and a piston and modulating spring. The anti-ice valve is fail safe, Figure 7. Air/electric anti-ice system concept.

the air is directed through the struts, anti-icing the struts Turbine leading edges, to the inner dome. This portion of the air cooling air then flows forward near the flow path surface of the inner dome and exits through slots at the front edge of the dome, then mixing with inlet air. The other portion of the air is directed to the IPS flow splitter then through the scavenge vanes and along a portion of the IPS outer flow path before exiting into the IPS scroll. This system was chosen because it best fit the selection criteria, and concept rejected: imposed the least impact on hardware changes from the . Reduced power turbine cooling original design. at some flight conditions f S8ze andweightof required plumbing exce~we iE90~2192 Once the anti-icing design concept was established, several analytical tools such as computer-aided design Figure 8. Buffer air anti-icing system concept. 18-6

m lower pressure of compressor interstage air and a spring causing the valve to close, disallowing aimow through the valve. When anti-ice is selected on, the solenoid is un- powered and blocks compressor discharge air from entering the piston. This, in turn, allows compressor interstage air pressure to act against the spring and modulate the valve. Compressor interstage air pressure, Concept reiected similar to temperature, decreases as engine power level . Excessive increase in SINIand decreases. The lower interstage pressure results in inner dome passage area increasing valve area, as seen in Figure 11. A pressure .Cost increase *Weight excessive switch senses pressure in the anti-ice flow circuit and provides a cockpit indication of anti-ice actuation.

INITIAL THERMAL SURVEY TEST AND PRE- LIMINARY ENVIRONMENTAL ICING TEST

The anti-ice system design was supported with several analytical tools; however, due to the complexity of the flow circuit and heat transfer effects, and to demonstrate the system as required by specification, .'.q&> hardware flow tests and two test programs were per- '\L formed. TE90-2193 The flow tests were designed verify the pan fabri- Figure 9. Dual pass anti-ice system concept. to cation and design integrity by ensuring that the engine requiring power to actuate the solenoid and tum anti-ice hardware was capable of flowing the design flow rates. off. When anti-ice is selected off, the solenoid becomes Flow tests were run on the anti icing valve, the inner powered and allows compressor discharge air to enter the dome, the front frame, the IPS casing assembly, and the valve's piston. The pressure of the air acts against the entire assembled anti-icing circuit.

Inlet particle separator

Front Frame Sectbn A-A

TE90-2194

Figure 10. PFR configuration anti-icing system. - Compressor discharge pressure (P3)

The first test program run was an anti-ice thermal fold in the front frame, resulting in lower temperatures of survey. Thermocouples were installed at various circum- the anti-icing air in both the front frame, inner dome, and ferential and axial locations of all anti-iced surfaces. the IF'S casing flow circuit. Figure 12 illustrates the location of thermocouple placement. The thermocouples were imbedded in the flow-path Heat transfer analysis, used to model the test data and surface to reduce convection errors and provide a me provide possible solutions, focused in the aft manifold surface temperature. The engine was then run throughout area. As the air enters the aft manifold from the delivery the power range with dry inlet air at 23" F (-5" C) and -4" tube, it is forced to flow circumferentially in one direction F 1-20' C), since these are demonstration points required by a blockage wall in the manifold (Figure 10, Section A- by the specification. Early in the test it was found that the A), thus ensuring that the air is distributed evenly. The anti-ice flow rate was lower than predicted values. The analysis verified that removal of the blockage could result valve was modified to obtain higher flow rates and thus in less temperature drop. Furthermore if the wall had a provide meaningful temperature data. properly sized opening, acceptable flow distribution could occur. Later in the engine development program, prior to The temperature data revealed that the surface tem- the QT phase, the front frame was redesigned and the aft peratures were lower than expected, as shown in Figures manifold was completely eliminated. This will be 13 and 14. Review of the test data showed a significant discussed, in detail, in a later section. temperature drop, about 100" F (56" C) more than exwted, from the anti-icing valve inlet to the aft mani- The temperature data from the the most severe condition, -4O F (-20" C) inlet temperature, revealed that all surface temperatures were above freezing except for the IPS outer flow path and the leading and trailing edges of the scavenge vanes.

The thermal data provided a preliminary evaluation of the anti-icing system effectiveness, however a preliminary icing test was required to determine flow-path response to icing conditions and to further evaluate engine performance loss due to anti-icing operation. The test unit was configured as a gas generator, @ewer turbine removed), during the icing test in order to simplify the test equipment adaptation of the icing spray system through elimination of the load absorption system. This did not compromise the test objectives because the entire

.Metal temperalure anti-icing system was intact. Engine performance during .static pressure these icing runs was synthesized based on actual perform- Figure 12. PFR anti-ice thermal survey ance data from this same engine prior to removal of the instrumentation. power turbine. 18-8

,--- / 't ,- I' I I I 9 (161- 60

(10) - 50

(4~.40

(-1) 30 IPS Scavenge Front shroud van3 frame Nose lrilet Innerdome A A A A A A Figure 13. PFR thermal survey data; 23"F,25% MCP. TE90-2197

The test stand was converted from an allitude test required a sophisticated deicing and defogging system. facility to an icing simulation test stand for this test. Additionally, the borescopes and support equipment were Spraying Systems 1/4J-IA nozzles were used to produce mounted to the engine and had to survive the engine and the water spray. The variable conditions were set by icing environment changing the nozzle water flow rate and the atomizing air pressure. Each test condition (cloud liquid water content Additional documentation of flow-path surfaces was and droplet diameter) also required that the number of obtained with videotaped borescope inspections immedi- spray nozzles be varied because of the limited operating ately following the icing runs and still photographs of the range of water flow and atomizing air pressure of each engine inlet. Cold airflow was maintained during these nozzle. The uniformity of the icing spray, which if not post-test inspections to prevent any ice melting for the correct could significantly bias test results, was estab- duration of the inspections (usually 20 minute duration). lished by collecting ice on a screen located at the end of the engine supply air duct After nozzle radial location The simulated icing conditions that the engine was was varied to optimize spray uniformity, the irpray system exposed to conformed to the Military requirements de- was calibrated using laser probes. The calibration not only lineated by Table 1. They were: identified the required water flow and air pressure set points for each operating condition, but also established (1) 23" F (-5' C) inlet temperature, 2.0 g/" liquid water sensitivity to variations of these parameters. content, 20 micron drop diameter (2) 4'F (-20' C) inlet temperature, 1.0 p/m' liquid water Intemal on-line video recording of a front frame strut, content, 20 micron drop diameter the IPS flow splitter, and an IPS scavenge vane was (3) 23' F (-5' C) inlet temperature, 0.4 g/m3 liquid water provided from borescopes mounted in the IPS casing content, 30 micron drop diameter, idle only assembly. This on-line monitoring was used as a diagnostic mol and as a safety feature, to ensure engine damage The test points run during the icing test were exten- due to ice build up and shedding was avoided during the sive. Power points from idle to maximum power were run test The borescope system was designed and imple- with simulated conditions (1) and (2). The engine was mented due to the inability to visually inspect and exposed to the icing condition for a 10 minute duration. document the 'Y' shaped engine inlet flow path, as can be After 5 minutes of exposure a rapid acceleration to seen in Figure 2. The video monitoring systeni design maximum followed by a rapid deceleration to the test criteria were very demanding. The borescopes tips were point was run to demonstrate transient response. Immedi- installed flush with the engine flow-path surface and ately following thc full 10 minute exposure period, the T800 engine 1808 8U4 Metal tenperature -4OF (-20%) amb 25% MCP (%) OF ,/ T

engine was shutdown and the post test inspections were The output power loss and SFC increase, resulting from performed. anti-ice operation in icing conditions were both within the specificationlimits. Condition (3) was run for a 60 minute duration at idle only. During this period engine oil was maintained at 23" Additional development work was performed to F (-5" C), to eliminate any heat transfer effects from determine if the engine could operate in the icing condi- engine oil. tion until ice accretion stabilized, as required by the FAA Anti-icing flow rates were adjusted by means of an specification. Ice formation in the IPS scavenge vanes external valve, throughout the test, to optimize the continued to build but did not significantly reduce the required anti-ice flow rates to prevent ice formation. This scavenge flow of the IPS. However, the ice buildup on the data pmved to be invaluable during the system redesign IPS outer flow path caused test point termination prior to for the QT configuration (reference next section). accretion stabilization.

Performance data were obtained while the engine was It had already been determined prior to the conduc- operating in icing conditions. Direct measurement of tion of this PFR preliminary icing test that the QT version output power was not possible since the test article was of the IPS casing assembly would have scavenge vanes configured as a gas generator. The dry air performance constructed of aluminum instead of stainless steel, as in data obtained previously on the complete engine was used the PFR version. This design change was made primarily in conjunction with the gas generator performance data to to reduce engine weight but it offered a benefit to the anti- synthesize output power determination. The performance icing system due to increased heat transfer characteristics data was ultimately used to establish the performance of aluminum. An early configuration IPS casing with alu- degradation, output power loss and SFC increase, associ- minum vanes was available near the end of the prelimi- ated with operating in an icing condition with anti-ice nary icing test. This IPS casing assembly was installed on activated. the engine and evaluated in simulated icing conditions. The ice accretion on the scavenge vanes was significantly The test results of the icing simulation points reduced with this IPS casing assembly, without making revealed some ice formation along the IPS outer flow any other system modifications. path, forward of the anti-iced surface, and on the pressure surface of the IPS scavenge vanes. The engine's core flow The thermal survey and the preliminary icing tests path remained free of any ice accretion during all points. were not intended to be qualification tests but rather to 18-10

identify areas of the anti-icing system that needed further above that required to inhibit ice accretion. This coupled enhancement. Both of these tests were very effective in with the icing problems in the IPS outer flow path and providing this data. scavenge vane areas led logically to a redistribution of the anti-ice flow rate splits. It was imperative to maintain the ANTI-ICING SYSTEM REDESIGN BASED ON melevel of total anti-icing flow rate to ensure minimal TEST RESULTS engine perform'ance loss penalties. Therefore, the flowrate in the front frmdinner dome anti-ice flow circuit was Following completion of the engine PFR develop- reduced 10% by decreasing the size of the inner dome ment program, LHTEC was awarded the follow-on anti-icing inlet holes, which meter the flow. The redistri- contract for QT development (and engine production). bution, coupled with slight modifications to anti-ice flow Several engine hardware redesigns were inifiated to orifices and pasrages in the IPS casing assembly, in- address problems encountered in the PFR phase as well as creased the anti-icing flow to the IPS by 10%. thus the to enhance engine reliability, producibility, and decrease overall anti-ice flow rate was unchanged. weight and cost. Part of this redesign was focused on the anti-icing system. The test results from the mti-ice In addition IOthe increase in anti-icing flow to the thermal survey and the preliminary icing test were very IPS flow split& and IPS scavenge area, other IPS casing valuable in highlighting the areas of the anti-icing system assembly changes were also implemented. The anti-icing that required design changes. The original anti-icing flow circuit on the outer flow path was extended forward, concept of the compressor interstage dual bypass system as shown in Figure 16. This change addressed the outer was maintained; however, the anti-ice flow circuit was shroud ice formation problem encountered in the icing modified to increase [he anti-icing system effectiveness. test. The anti-ice holes in the scavenge vanes were modified to provide more uniform heat transfer to the The front frame was modified significaritly by relo- vane surface. As planned near the end of the PFR phase, cating the oil supply and scavenge passages. This change the vane material1 was changed from steel to aluminum to was made primarily due to a compressor modification, reduce weight. This increased the heat transfer character- however it allowed beneficial rerouting of the anti-ice istics of the vanes. airflow. As discussed previously, the anti-ice air temperature decreased significantly in the aft tunnel of the front The anti-icing valve was redesigned to provide the frame anti-icing air manifold. Relocating the oil tubes anti-ice flow rate changes determined from the prelimi- allowed this manifold to be eliminated, thus lheoretically nary icing test As pan of the initial development test the reducing a large portion of the 100°F (56" C) temperature anti-ice flowrates were controlled with an external valve drop. With the new design, schematically shown in Figure to provide variability not available in the engine valve. 15, the anti-ice air enters the front frame from the delivery This flow data defied the design flow rates for the QT tube and flows directly to the forward manifold. anti-icing valve.

Both the PFR theimal survey temperature data and the preliminary icing test verified that the inner dome and front frame strut surface temperatures were at a level well

inlet panicle separator

PFR anti-ice System

Anti-iced sudace extended forward I fT--'\\

An manifold eliminated Single passage 10 toward manifold

OT anl-lce s)'sIem

TE90-2200 Figure 16. IF'S outer flow path anti-icing design Figure 15. QT anti-icing system.. change. 18-11

The modifications made to the engine anti-icing tavirtdly eliminate any blockage since the leads will be circuit addressed all of the concems resulting from the routed through some of the anti-icing flow passages. In thermal survey and icing test. These changes did not order to allow access, the thermocouples located on the impose major setbacks to engine hardware fabrication IPS scavenge vanes are installed during the basic IPS schedules because they were coordinated with hardware casing fabrication process. changes which enhanced the other features of the engine (reliability, cost, weight, prcducibility). The modifica- Thermocouples will also be located near the shroud tions, made on the basis of the development data, proof the IPS scavenge vanes. The purpose of this insmmen- vided a high degree of assurance that the effectiveness of tation is to ensure that the epoxy used during the IPS the anti-icing system would be increased to a Level that casing manufacturing process remains below the allow- both the Military and the FAA certification tests could be able temperature limit during the hot day anti-ice failed successfully completed, including the extended exposure condition. requirement of the FAA. The engine will be operated throughout the power range with dry air (no icing cloud) at each of the follow- VALIDATION OF REDESIGN BY THERMAL SUR- ing conditions; 59'F (WC), 40°F (4.4"C), -4'F (-2O'C). VEY TEST and 13 1°F (55°C). Each of these conditions will be run with and without anti-ice actuated and with customer The QT anti-icing system will be validated with bleed extraction. The source of the customer bleed air, hardware flow tests and a second thermal survey of the like the anti-ice air source, is compressor interstage. Bleed protected surfaces, scheduled to be performed in 1990. extraction provides a more challenging condition for anti- icing system operation, due to the increase in temperature The anti-icing circuits of the redesigned inner dome, and pressure loss occurring when a higher flow rate is front frame, and IPS casing assembly, and anti-icing valve delivered. will be flowed to verify design flow capacity. The parts will then be instrumented to allow anti-icing air tempera- Following the completion of the dry air runs, the ture and flow path surface temperature measurement 40'F (4.4OC) inlet temperature run will be repeated while during engine calibrations, which will be run with simulating an icing cloud at the engine inlet. The 4OoF variable inlet temperatures. (4.4T) inlet temperature was chosen to eliminate the complex test equipment provisions required to prevent ice Instrumentation during the thermal survey test will be formation in water supply lines and the spray nozzles that very similar to that used during the PFR development would occur with below freezing inlet temperatures. phase thermal test, except for slight changes to thermocouple location based on experience and design differ- The purpose of this water injection test is to provide ences. Figure 17 illustrates the location of the instrumen- data so the heat transfer analysis model can be further tation. As before, the thermocouple sensing tips will be enhanced, and then extrapolated, to provide confidence in imbedded in the surface to provide a true surface tempera- actual icing conditions. Slight variations between the PFX ture indication unaffected by convection. The size of the thermal survey test results and the PFR preliminary icing thermocouple wire used during this test will be minimized test results occurred because the location of the icing cloud water droplets on engine flow-path surfaces could not be accurately predicted. The water significantly affects the heat msfercharacteristics of the surface, therefore determining its location and quantity becomes a critical part of the analysis. Predicting the location of the water droplets is challenging due to the complex engine airflow and aerodynamic separation.

Surface temperature variations between the dry air run and the simulated icing cloud run will provide the temperature data required to accurately predict the location and quantity of the water droplets on flow-path surfaces. Following this, the heat transfer model can be used with a high level of confidence to predict surface temperatures during icing condition operation.

The thermal survey test is intended to provide data so that the heat transfer model can be enhanced. Ultimately, Figure 17. QT thermal survey instrumentation. the thermal survey testing will provide a high level of 18-12

confidence that the anti-icing system can provide accept- incoming area and provide potential for ice accretion. able engine icing protection throughout the: icing envelope This challenge was met by designing and developing an prior to running the final environmental icing test. anti-icing system that exceeds all military and FAA requirements. FINAL ENVIRONMENTAL ICING TEST The anti-icing system design evolved from the initial The QT environmental icing test will take place late concept to the final configuration through the use of in 1990 at the Naval Air Propulsion Center (NAPC) in analytical tools, a thermal survey test, and an environ- Trenton, New Jersey. The test will be identical to the mental icing test. It has undergone three design iterations, icing test performed in the PFR phase excelpt the test which were integrated with other modifications as the article will be configured as an engine instead of a gas engine matured, so as to reduce schedule and cost impact. generator. The system utilizs compressor interstage air as the energy source and delivers the air through a dual bypass NAFC has performed several icing tests on various arrangement. The airflow rate is regulated by a self modu- engines, including the Allison T406. NAPC’s facility, lating valve to prevent excess air delivery at higher engine equipment availability, and experienced personnel made powers and thereby minimize performance loss associated their test site very attractive. LHTEC has the. capabilities with the use of the anti-icing system. to perform the test but the cost associated with a one time Effectiveness of the system has repeatedly setup were not an acceptable option to the NAPC test site. been demonstrated with analytical and actual test results during various test programs. Final verification of the system The preliminary work completed on the: engine’s will result from a second thermal survey. Military and anti-icing system provides LHTEC with confidence that FAA flight certification of the system will be obtained the engine will successfully complete the icing test It is through a final environmental icing acceptance test at the planned that the testing completed at NAPC will be ap- Naval Air Propulsion Center in Trenton, New Jersey, proved by the US. Amy as completion of the required scheduled to be performed late in 1990. demonsmation of the anti-icing system. LH’IEC will also build on the test program to include additional test points The anti-icing system is effective, meeting all re- that will satisfy the FAA requirements. quirements. It is light weight, low cost, reliable, easily maintained, and provides the pilot with engine icing SUMMARY protection by activation of a single switch. The anti-icing system development, as well as the T800 engine develop- The T800-LHT-800 engine anti-icing requirements ment, represents many thousands of hours of dedicated imposed a design challenge due to the presence of an work from LHTEC personnel to providing a state-of-the- integral inlet particle separator (IPS). Characteristically, art helicopter engine that meets or exceeds all require- inlet particle separator systems expose a large areas to ments and design goals. 18-13

Discussion Author: Yes.--, the laser orobes measured both varticle size distribution and density distribution. The probes used were 1. R. Wibbelsman, GE called a Fowaro scattering spectrometer prohe (FSSP) and Have you checked the surge margin under the conditions of an optical array probe (OM). These probes are anti-icing or STIJ inlet temperature? commercially available.

Author: The engine surge margin has been completely characterized 5. D. Mann, Rolls Royce under all operating conditions, including operation with Have you made any separation efficiency measurements on anti-ice on. water droplets? How impoprtant do you bklieve is the waterhe separation 2. R. Toogood, Pratt and Whitney Canada efficiency to the configuration of the anti-icing system? What is the IPS bypass/core airflow ratio? Have you determined that the change from steel to Author: aluminium for the IPS turbo shafts will maintain adequate We have not made water droplet separation efficiency erosion capability? measurements. Did you consider an IPS design which did not include I believe that water separation efficiency is very critical to turning vanes in the scavenge duct? the anti-icing system. I do not think that increased water separation efficiency is critical to anti-icing system design, Author: but rather, identifylng the location of the water droplets is The IPS scavenge flow is in excess of 20% of the engine inlet the critical aspect relating to anti-icing design. flow. We have run the IPS with aluminium scavenge vanes on the sand ingestion rig to evaluate erosion of the aluminium. The result of this test showed no significant erosion. 6. W. Rearson, Naval Air Propulsion Center To the best of knowledge, a design without turning vanes was Besides PFR icing, the engine was configured as a gas not considered. generator. How was the performance deterioration determined in this configuration? 3. A. Spirkel, MTO How much anti-icing air flow is used at high power ratings? Author: Prior to the removal of the power turbine (configuration of Author: gas generator) the engine performance was characterized. Approximately 2.5 of the engine core flow is the quantity Power turbine inlet total pressure and inlet temperature was of anti-icing air used at idle, ahout half of this amount at measured during this. When the engine was run as a gas maximum power generator, the power turbine inlet pressure measurement and power turbine inlet temperature measurement were 4. W. Alwang, Pratt and Whitney used to compare the complete engine data. Therefore, gas Did the laser probe measure both particle size distribution generator performance loss was determined by analyzing and density distribution? and comparing the gas generator and complete engine Was it a commercially available probe? performance data.

19-1

ENGINE ICING CRITICALITI ASSESSMENT by E. Brook Rolls-Royce plc PO Box 31 Derby DE2 8BJ United Kingdom

SUMMARY unacceptable performance loss or engine damage. The assessment can be carried out Assessment of an engine design for icing by breaking down the engine into areas risk is important at both the design stage according to type of risk. Figure 1 and for development and certification illustrates the areas of concern on a high testing. Icing must be included with bypass ratio turbofan. Each area is aerodynamic and noise constraints during considered in detail in the following the design phase to minimise the risk of sections. design change during development, and the compromise tested must be tested at the 3. NON-ROTATING COMPONENTS UPSTREAM OF extremes of the atmospheric icing, and THE FIRST ROTOR aircraft and engine operating envelopes most appropriate to the particular 3.1 Intakes components. This paper addresses the type of assessment necessarv illustrated mainlv~~ - The engine intake is often treated as part by reference to high b;pass ratio of the aircraft rather than the engine. turbofans. Aerodynamic considerations determine the shape, but anti-icing considerations are The approach to identifying critical one of the factors affecting structure and conditions is presented and areas where materials. Much data is already available research can provide basic data for the and analytical techniques are used development of design methods are routinely, confirmed by testing on the discussed. aircraft. I do not propose to cover intake anti-icing design in this paper, however, 1. INTRODUCTION ice shedding following delayed selection of nose cowl anti-icing, or caused by In evolutionary engine design where incomplete anti-icing, can directly affect relatively small changes distinguish engine the design of the first rotor stage due to types, it is easy to ignore icing as a potential damage. Therefore, collection of design constraint and accept that the __.ice and __~~size of~~ ice ~~ shed is an enaine- modest changes will not cause problems when design consideration that must be tested in icing conditions for considered, along with other debris certification, or in service. However, ingestion questions, when designing icing constraints are often directly compressor blading. opposed to other design considerations, such as noise, and anti-icing use is a 3.2 Intake Ducting performance penalty. It is therefore necessarv to review modifications for For the modern pod mounted high bypass efficti-on icing-at an early stage, most of ratio engine, intake ducting is minimal and this review being by comparison with offers no surfaces for collection of ice. existing satisfactory designs. However, this is not true for all, and shaped inlet ducting (Fig. 21 such as used Revolutionary designs lead to a requirement on turboprops and on some tail mounted for more analytical techniques. These must engines, or where inlet particle separators combine experience from past engine testing are used to prevent grit and debris with research for basic data acquisition to entering the engine can offer surfaces on establish methods that may allow which ice collection is likely. extrapolation beyond the range of existing designs. Prediction of requirements for anti-icing and identification of areas of concern Assessment of engine designs at an early requires detailed flow prediction stage can also identify evidence to be programmes for three-dimensional flow more obtained during development testing to complex than those necessary for intake lip validate analyses. anti-icing design. At a later stage in the engine development Heated surfaces within a shaped intake must process, the engine must be assessed be assessed against propensity to against the Certifying Authorities' accumulate ice in ice crystal conditions requirements to determine the details of where surface temperature and local airflow the icing testing for engine certification. velocity are directly relevant. Consideration must also be given to 2. BASIS FOR ASSESSMENT drainage to prevent ice forming whilst the aircraft is parked that could be sucked The assessment must bring together the into the engine on start-up. details of the engine design, the aircraft flight envelope and engine operating 3.3 Instrumentation envelope with the atmospheric icing envelopes defined by the Certifying Pressure and temperature probes for engine Authorities. control or for measurement of thrust mounted in the engine intake are normally The design must then be assessed to continuously anti-iced on modern engines. identify areas of concern for production of The adequacy of this anti-icing needs to be 19-2

addressed on the engine as local flow least partly, the rotating nose cone or effects can produce areas of enriched water spinner. With no anti-icing ice must content within the intake, although accumulate on this and the low centrifugal detailed testing of the isolated probe in forces acting close to the engine icing conditions will have been carried out centre-line may allow an excessive by the supplier. Prediction of water accumulation before melting or droplet behaviour around the intake lip can out-of-balance forces cause shedding. be used to assess the acceptability of the probe immersion. There are two approaches to nose cone design; either accept the shape determined Modern temperature and pressure probes are by compressor intake flow design and design usually electrically anti-iced to minimise an anti-icing system adequate to prevent their performance penalty. For these the ice build-up, or determine the shape of

~~~~ ~~ ~~~ critical icino~ conditions ~ ~~~ are likclv1 to~ be ~~~ nose cone that minimises ice build-up and maximum engine p3wer conditions a: minimun ensure that this is established as the altitude and either 3c minimum temperatures, basis €or design. Departures from this if heat removal is dominated by the ideal design, which may still not require airflow, or at about 0°C total inlet any deliberate anti-icing, can then be temperature, where water input is maximum assessed provided data on ice catch and ice due to the nature of the atmospheric icing shedding is available, similar to that used envelopes, if heat removal is dominated by on fan blades discussed in Section 4.2. the requirement to evaporate the supercooled water droplets caught by the If the methods to assess the design are not probe surfaces. available, specific testing of a new design may be necessary once criteria for On some older engines hot air anti-iced acceptability, e.g., maximum ice build-up probes are used and critical conditions in weight or thickness, have been then depend on the hot air supply. established. As a last resort, engine

testing in icinq conditions will establish~~ ~~ ~~~~ The design of the probe anti-icing system the acceptabilify of a particular design, should be such as to prevent any j.ce at a cost. forming on the probe either by heating the whole probe body, or by providing full The critical conditions for ice build-up on evaporation of all water caught by the a spinner are likely to be extended probe on the critical areas. Probe designs exposure at minimum temperature conditions that are not fully anti-iced must be such as occur d.uring hold as this produces assessed against the damage that could be the StrOngeSt ice bonding, and is therefore caused by shedding of the maximum size of likely to ."Ulate more ice than higher ice particle. liquid water contents at higher temperatures which will shed earlier. 3.4 Inlet Guide Vanes and Nose Com For aircraft that normally cruise at lower Figure 3 illustrates an engine with altitudes, within the continuous maximum non-rotating engine hardware immediately icing envelopes, higher engine powers may upstream of the first rotor stage. This give rise to a worse situation due to the type of hardware must be anti-iced, and higher energy imparted to the shed ice. because of the danger of damage caused by ice shedding and the distortign caused by A further aspect to be considered in blockage the inlet guide vanes at least spinner designs is the possibility of should be continuously anti-iced. critical asmetric ice accumulation. Uneven gaps conditions will depend on the type of around fairings can give rise to ice anti-icing used, whether it is independent accumulations sufficient to give of engine power, like electrical heating, or unacceptable out-of-balance and, hence, dependent such as the use of engine oil or engine vibration. Spinner drainage must hot air. also be addressed for the same reasons. 4. THE FIRST ROTOR STAGE 4.2 The Fan ~llsurfaces ahead of the first rotor stage Ice cannot be prevented from accumulating are potential sources of ice debris that on the fan blades, the first rotating stage could damage the aerofoils or casing in a high bypass ratio engine. Assessment linings of the first, and subsequent, of an engine design, therefore, reduces to rotating stages. This type of damage can prediction Of the quantity of ice that will be minimised by appropriate application of ~CCUmUlatebefore shedding, how this ice anti-icing or de-icing, i.e. prevention of Will shed, where the ice debris will go and ice accumulation or removal of ice before how to Prevent t.his ice debris causing an unacceptable quantity has accumulated. damage. Once into the rotating stages it becomes virtually impossible to provide anti-icing Current fan designs are unlikely to suffer and it is, therefore, necessary to appraise from unacceptable PerfOm"e loss due to the design for its propensity to accumulate flow Passage blockage by ice, but it cannot ice and the likely consequences of the be ignored for revolutionary designs if accumulation. It is then necessary to blade spacing is significantly reduced, determine how this ice will shed and especially at the blade root where consider the design of the surfaces against distortion generated would directly affect which it will impact. the core engine. 4.1 The Nose Cone Ice accumulation on fan blades is a function of the fan blade relative air There is one further part that it may be temperatures. This leads to a maximum possible, and desirable to anti-ice at radius above which ice cannot adhere for a 19-3 given set of ambient conditions and of excessive quantities of ice can cause rotational speed, and a gradation in ice surae or flame-out as well as damaae to bond strength to the blade surface rot& and stator blades or rotor pith decreasing as radius increases. liners. This bond strength is an important Methods to predict ice accumulation and parameter if analysis techniques are to be shedding from stator vane sets based on ice established to Dredict ice sheddina from catch and bond strength data may be fan blades, and-also nose cones or-any possible, although prediction of the other surface. aerodynamic forces responsible for sheddina.-. aiven- the hiahlv The energy of the ice debris can then be three-dimensional form-of-the flow, considered for design of the fan track especially after ice has started'to liner and casing panels downstream, and accumulate and is causing local blockage prediction of worst case scenarios for dependent on ice shape, is extremely engine icing testing for certification. difficult. It is also necessary to For example, is the maximum accumulation of consider the interaction between ice on ice occurring during extended exposure at adjacent vanes as ice structures bridging low power conditions likely to produce more between vanes are strong and do not rely on damage than the limited accumulation at bond strength to the individual vanes to high power conditions where much of the fan prevent shedding, but require a failure blade will be ice free? For the low power within the ice structure. condition ice shedding during engine acceleration must also be considered. Stator vane spacings that permit ice to bridge between vanes, given the icing In most current high bypass ratio engine conditions that actually occur within the designs it is not possible for ice shed engine, and realistic exposure times are from the fan blades to enter the core unlikely to be acceptable due to blockage, engine. Reduction in fan blade trailing and hence compressor inlet distortion . edge to core engine inlet spacing may permit ice to shed into the core engine. Empirical data is needed to permit This could cause core engine damage, or assessment of the propensity for OOWR~I. ._ loss~.__ due to commessor surae or unacceptable ice accumulation on stator flame-out. Methods to-predict ice vane sets as a function of vane spacing, trajectory once released from the blade stagger angle and temperature, as reduced surface are, therefore, necessary for spacings are often favoured for noise design assessment. reduction reasons. 5. ENGINE COMPONENTS DOWNSTREAM OF THE Temperature is an important parameter as FIRST ROTOR STAGE the ice form and quantity is dependent on the temperature of formation. Icing The assessment of the components downstream conditions aivina worst blockaae need to be of the first rotor stage in icing idenclfied for ccrcificacion tcscing. conditions is fundamentally different from Highcsc air temperacures give che least components considered in Sections 3 and 4 aerodynamic ice forms, but also the wcakese as the temperature and pressure rise ice a compromise temperature must, through the first, and subsequent, therefore, be identified. compressor stage alters the envelope of icing conditions. Engine power becomes a Ice catch on surfaces is also important in major factor in the exposure of components identifying realistic airflow water to icing. At high engine powers in flight contents to be considered once the fan ambient conditions with supercooled water blades have removed a proportion of the droplets present may never exist combined atmospheric liquid water content. with temperatures downstream of the first rotor stage at which ice can accumulate. 5.2 Fan Bypass Stream Outlet Guide Vanes Ice catch on the fan blades and nose cone Ice accumulation on fan outlet guide vanes is also an important parameter in follows the same basic rules as that on predicting the liquid water contents in the engine section stators, but is usually less airflow that must be considered. critical as spacings are usually wider and Therefore, it is not usually necessary to there are no components downstream that can anti-ice components downstream of the first be affected by either the blockage rotor stage, and it is often impossible to generated distortion or the ice that is provide anti-icing on the aerodynamically shed. However, excessive blockage can give determined components due to their size or rise to pressure loss through the outlet the problems of supply of hot air or guide vanes or loss in aerodynamic electrical power. efficiencv leadina to~ excessive ~~~ ~~ ~ swirl ~~~ in ~~~ the bypass stream airflow, giving reduced However, there are a number of areas of engine thrust if not compensated for by the particular concern that should be engine control system. considered at the design phase and addressed during engine certification icing An assessment of the envelope of engine testing . thrust, aircraft operating conditions and atmospheric conditions in which icing can 5.1 Engine Section Stators occur downstream of the first rotor stage is, therefore, important. The engine section stators are the first set of components within the core engine. 5.3 Intercompressor Ducting Ice accumulation on these can cause core engine inlet blockage, leading to Intercompressor ducting usually has a very distortion and compressor surge. Shedding limited envelope when icing may occur depending on the number of stages in the 19-4 upstream compressors. For ducting with Instrumentation most likely to be affected major changes in direction of flow, such as by icing is that in the bypass duct. high radius ratio swan-necked ducts, Pressure rakes used to determine engine consideration must be given to accumulation thrust should be anti-iced if it is of ice during encounters with ice crystals, necessary to guarantee their proper which can occur at much lower ambient operation in all icing conditions. Where temperatures than supercooled water engine pressure ratio is used as the droplets. The effect of local du.ct surface major control parameter for the engine it is temperatures due to passage of en.gine preferable to avoid danger of inaccuracy in services such as oil pipes must be icing conditions by using hot stream rakes considered as this can give rise to only. surfaces heated to the correct temperatures for ice crystals to melt initially and For instrumentation icing and maximum depth provide adhesion to the surface if of penetration the critical engine aerodynamic forces are not sufficient to conditions will be minimum power, when prevent it. compressor temperature rises are minimum, combined with minimum flight speed. Experimental data on ice crystal accumulation as a function of surface 6. CONCLUSIOIE temperature and local velocity is necessary for detailed prediction. For both assessment of engine design to reduce risk and for identification of 5.4 Depth of Penetration of Icinq conditions that should be addressed during certification testing in icing conditions Assessment of the depth into the core the engine and intake design must be engine that icing conditions can occur at considered in stages according to the all normal operating conditions, needs to parameters affecting icing, and the be carried out. Although the liquid water following quest:ion answered: content~ ...... ~~is ~~~~~~~~~~reduced at~~~ each ~~~~~ staae-- bv ice accumulation, icing penetration into high 0 How much ice will accumulate during an pressure compressor stages where spacings icing encounter? and vane sizes become smaller must at least be considered and, if outside the range of o What effect will the accumulation have current experience, development testing on the component and those downstream should be recommended to reduce the risk of of it? failure during identified worst case conditions tested during certification o HOW will the ice be removed? icing tests. 0 Where will. the ice go once it is Consideration of icing of bleed ports is removed and what subsequent effect can also necessary if blockage could cause this debris have? compressor surge. Collection of data of either fundamental 5.5 Instrumentation ice properties, such as ice bond strengths, or component specific data, such as ice Core engine instrumentation must also be catch on surfaces and stator vane passage considered relative to depth of penetration blockage, are important to permit these of icing. Blockage of pressure probes or questions to be answered. incorrect reading of thermocouples due to ice can affect engine control such as bleed valve and variable guide vane operation.

Key 1 Nosecowl 2 Intake probe 3 Spinner 4 Fan blades 5 Inner barrel

I, 6 Engine section stators ,I 7 Guidevanes 8 Bypass duct instrumentation 9 Core engine ducting

Figure 1. Engine and nacelle icing considerations NOSET INTAKE GUIDE VANES

Bifurcated trifurcated (in-line gearbox)

/OUT& TO NOSE COWL

Figure 3. Anti-iced inlet guide vanes and nose Cone

Bifurcated

Figure 2. Example of convoluted intake duct 19-6

Discussion except indirectly through the charge in the engine as the ice accumulates. It should be noted that no loss in thrust occurs as the ice accumulates since the engine is usually controlled 1. M. Holmes, RAE to an overall pressure ratio which means a direct To what extent are altitude test facilities used by Rolls Royce measurement of engine thrust. for icing qualification tests compared with flight tests behind an aircraft carrying water droplet spray equipment? 4. D. Way, RAE Author: Which of the icing problems that you have decribed is most RR use altitude test facillities for engine icing tests troublesome, in the sense that it is least amenable to extensively and consider an altitude test facility the best way protection or experimental model testing early in the engine of performing engine icing tests. Aircraft icing tests are only development programme? contemplated where problems occur during flight testing or where the certifying authorities of the airframe constructor Author: insist. However, in the future, the limited size of the altitude The worst problem recently has been icing of engine section test facililties available may force a change as engine size stators. Engine section stator designs have been giving increases. reduced gaps and increased blade numbers for noise reasons, and ice building on these can lead to problems due 2. W. Grabe, NRC Ottawa to the pressure loss through the blockage, or flow distortion, Have you estimated energy requirements for anti-or de-icing or ice shedding into the core. This can produce compressor the Contrafan nacelle lip on account of its larger diameter? stall or surge and combustion extinction. Without testing of Suggested is not to reject the cyclic de-icing which is done the full engine the interaction between core inlet icing and now quite successfully, providing accretions are not allowed the operation of the core compressor is very difficult to to grow too large? predict, especially as problems are likely to occur at descent idle powers. Author: If problems occur during engine testing the only solution is Engines like the contrafan, which has both a large diameter to apply some restrictions on engine operation in icing fan cowl and a gas generator cowl ahead of the aft mounted conditions which is undesirable. fan, plus struts and inlet guide vanes which require protection from icing, puts a demand on core compressor blade flow for full anti-icing which is unacceptable. I do not 5. M. Holmes, WAE know the exact numbers but they are such that the bleed You suggested that your icing assessment and prediction demand would give an unacceptable effect on gas generator methods were aimed at minimizing or even eliminating the operation. The same applies to many of the designs need for expensive testing. Do you believe you will ever currently being proposed for ultra-high bypass ratio reach that goal, and what evidence have you obtained so far engines, and alternatively anti-or de-icing of engine inlets which gives you (and the airworthiness authorities) such must be considered, such as electro-impulse de-icing. confidence? However, any de-icing method gives an additional problem in that the engine must then be proved to be capable of Author: accepting the regular inputs of the ice debris generated. This The assessment and prediction methods have arrived at should not be an insuperable problem but it increases the eliminating expensive testing, but they have also arrived at erosion problems already generated by rain and hail. ensuring that the testing carried out is the most effective one. Having identified areas of risk relative to operation in icing 3. A. Sutton, BAe this is accepted by the airworthiness authorities already as What amount of blockage have you found with ice accretion evidence in adjusting test conditions or eliminating test on the fan blades? requirements. The evidence of the success of the approach is in the excellent operation record of Rolls Royce engines in Author: service both in term of safety and in-service behaviour with We have not measured the ice blockage on the fan blades icing. 20-1

ICE INGESTION EXPERIENCE ON A SMALL TURBOPROP ENGINE

L.W. Blair, R.L. Miller, D.J. Tapparo General Elecbic Aircraft Engines 1000 Westem Avenue, Lynn, Massachusetts 01910

Introduction The compressor operates behind an integral inertial particle separator (IPS). The separator system, shown in Premise Figures I-IA and I-lB, is integrated into the power unit oil Modem high technology turbine aircraft engines often cooling system and provides aimow to cool the electronic employ high rotor speed compressors with thin advanced engine control. Originally designed as a sand separator for blading designs to achieve better performance. The engine the military helicopter environment, the system incorpo- designer is faced with a tradeoff between optimum compressor performance and on-wing durability. During the TO EXHAUST EJECTOR engine/ aircraft development stage, certain assumptions are \ made regarding the icing environment and the testing CONTAMINATED required to confirm compatibility with it. Often the true COMPRESSOR I impact of the design trade-off is not realized until the engine is exposed to its service environment.

Despite successful engine test cell and aircraft natural icing certification tests, in General Electric Aircraft CONTAMINA 1984 SCROLLCASE Engines Company began to experience an unacceptable INLETAIR level of foreign object damage (FOD) caused by ingested ice with its CTl-5/-7 family of turboprop engines. INLET FRAME FFAME AFl FORWARD STRUTS (5) PulDose STRUTS (6) I The purpose of this paper is: Figure I-1A Engine Inlet Flow Diagram (1) to address the issue of Stage 1 compressor rotor blade ice FOD in the CT7 engine, (2)to explain the methods and techniques used in assessing the icing environment, (3) to explain the lessons learned from test and analysis, and (4) to define the final resolution of the compressor maintenance problem which simultaneously created accelerated performance deterioration for the engine.

The paper is divided into two parts, the fnst dealing with the airframe icing environment and its impact on the engine inlet system, and the second concentrating on the design improvement and durability testing of the Stage 1 compressor blade. I Figure I-1B Inlet Frame, Mainframe, IGV Casing Assembly PARTT rated swirl and deswirl vanes to optimize efficiency. It has Basic Eneine Design since been, changed from a swirl concept to a straight axial The CTl turboprop engine is a free-turbine modular bypass to reduce pressure loss and improve performance. design derivative of the T700 turboshaft engine. It employs a five-stage variable stator axial and single stage cenaifugal Inlet and Installed Icing Environment compressor with a combined pressure ratio of 18:l. The airframe-provided inlet design is a conventional S-duct with a throat area optimized to aircraft specifications. An inlet protection device (IPD) is incorporated to meet bud ingestion certification requirements. 20-2

Responsibility for inlet aerodynamics was assumed by ENGINE ICING CERTIFICATION TEST CONDITIONS the engine manufacturer. The airframer assumed responsi- (REF AC 20-73) bility for the mechanical design, structural integrity, anti- ENVIRONMENTAL ICING TEST CONDITIONS icing, test and procurement. Condlan Number 1 2 3

Almorphenc Temperalure. OF 29 23 4 Liquid WaerConfent 0.3 2.0 1.0 Other areas forward of the engine, which are subject to (minimum). gramimelci 3 the icing environment are shown in Figure 1-2. Propeller Mean EnWbe Diop Diameler. mlCron$ 40 25 15 spinners on both applications zeparabolic profile and Icing Candaio" P0werSe"ingr Ground Idle Flight Idle Fqht Idle 50% Max Con1 50% Ma* Con, unheated (neither anti-iced nor de-iced). Original ice-shed 75% Max coni 75% Max Coni trajectory analyses showed no ingestion into either inlet Takedl Takeoll Duration e each 01 the power 30 10 10 design. Later analytical refinements reversed this finding Se,,ingr &red *"e. Iminules) I for accumulations on the spinner nose. The two propellers used, were each 4-bladed with de-icing provided in alter- Table 1-1 nating pairs. Propeller cuffs have been observed to accrete Inlet testing indicated fairly good results, however, the ice which was not shed with blade de-icing. In the upper duct was not ice-free, as required by the engine manufac- inlet highlight-to-spinner boundary layer divener area, one turer. Accretions up to 50 grams on the inlet to IPD splitter inlet, which is integral with the forward nacelle cowl, were noted. Testing was conducted under the conditions incorporates anti-icing. The other inlet, which moves shown in Table 1-2. Since both inlet and installation were relative to the forward nacelle cowl, is not ami-iced in this subjected to significant natural icing certification flight area. Both diverters have been observed to accrete ice. tests, and not a single ice FOD event was observed, the CT7 entered revenue service in 1984. By the end of that year however, the ice FOD rate was 1.2/1000 EFH and considered unacceptable.

3 LIP

Figure 1-2 Installation Ice Accretion Certification Testing The CTi turboprop was successfully certified according to the requirements of FAR 33.67, FAR 25 Appendix C, pG7-] -30 MINUTETOTALTESTTIME and Advisory Circular 20-73 in January of 1983, at the Table 1-2 Flight Icing Encounter Pattem for Test Condi- General Elecmc Test Facility, Evendale, Ohio. Test conditions 1 & 2 tions are shown in Table 1-1. The Evendale facility is a sea &&Inlet Svstem Develonment level test cell capable of providing a controlled icing cloud from a free jet pipe 30 inches in diameter at a maximum Follow-on kine Tests velocity of 88 knots. Testing was conducted using an anti- From 1984 to 1988. six additional inlet tests and two iced bellmouth. No installation specific inlet testing was engine tests were conducted to troubleshoot actual or sus- required or performed. During all phases of testing, engine pected anomalies that might have caused ice FOD. Several surfaces remained ice-free and discemible ice ingestion, no of these tests were conducted on the initial prototype de- causing either Stage 1 blade damage or parameter fluctua- sign. Later tests on production ducts revealed additional tion, occurred. problems that requirrxl further optimization and proof tests. Inlet ducts for both installations were certified at the Inlet component tests were run at the GE Evendale inlet manufacturer's facility. This facility is a closed tunnel facility late in 1984 in an attempt to find a correlation capable of simulating airspeed up to 195 knots and full between the engine certification testing conditions and Appendix C icing conditions to -30" C. No engine hard- those for the inlet at the inlet manufacturer's facility. waTe was included in this testing. 20-3 I

Because the GE open facility has limited viewing of the test ing intent of the test was to see if propeller effects had any specimen, a technique was developed to allow real-time impact on ice accretion within the duct. While no new inlet remote viewing and recording through the inlet duct wall ice was found, ice accumulated on both spinner nose (700 g during the icing test. This technique proved useful in max) and prop blade cuffs (20-30 8). .These new sources subsequent tests for viewing both of the IPD and engine prompted new analysis and efforts, described later, to face areas (See Figures I-3A and I-3B). reduce these potential threats.

Figure I-3A I 1 Figure 1-4 Engine Ice FOD Rate 3 Month Rolling Average

Figure I-3B

The icing was still apparent and a second test, parallel - ing the Evendale engine test conditions, was conducted in Figure 1-5 Outdoor Crosswind Test Facility January 1985 at the inlet manufacturer’s facility. Improve- ments made as a result of that testing were proof tested Ground tests in heavy blowing snow were conducted there later that year. in Cincinnati, Ohio and Erie, Pennsylvania in the same time period. These tests consisted of 30 minutes at ground idle Although the ice FOD rate seemed to be coming down followed by accel-to-takeoff power. No ingestion events as additional fleet hours were accumulated (See Figure 1-4), occurred. Upon shutdown of the engines, little or no ice operators, particularly in the wet and cold winter climates was found in the IPD, yielding no correlation to the field (Sweden, Switzerland and Midwest US.), were still experi- reports. encing engine performance loss, associated with compressor blade maintenance and causing some premature engine It was clear that the missing ingredient was airspeed removals. In addition reports of increased maintenance and it was the inlet manufacturer’s facility that was capable being required to keep the IPD from filling up with ice of testing airspeed along with ice crystals (simulated snow) prompted more testing. and slush-like environments. Because the engine inlet and separator frame has a large frontal area of low velocity, like Late in 1985, at GE’s outdoor crosswind test facility in the IPD, a combined inlet and engine frame test was pro- Peebles, Ohio, an inlet icing test was conducted using an posed in February 1986. The inlet facility, however, was entire SAAB 340 aircraft (See Figure 1-5). The aircraft not capable of running a CT7 engine. As a result, the test was tied down in front of a bank of 15 fans. These fans rig consisted of the inlet duct, inlet and separator frames, created an icing cloud that enveloped the entire propeller and the engine accessory gearbox, fitted with starter genera- and spinner in the right-hand installation. However, the tor and IPS scroll casing (See Figure 1-6). fans only provided 60 knots of ram airspeed. The underly- Other Amroaches to Ice FOD Reduction Turboprop spinner icing experience was reviewed and a conical spinner design was service evaluated in an effort to reduce accretions and thereby alter trajectories. There had been success in the past on turbofan applications for such designs: Figure 1-7 depicts the turbofan test results upon which the turboprop service evaluation was based. The results of this survey yielded no improvement. Spinner coating experiencl: was examined with no obvious benefit / Rm uncovered. A program to anti-ice the spinner nose was -YLz--- carefully reviewed, however, a shortfall of aircraft electrical - \4wnn - ...c=-c2 ,...... power eventually ruled out such a program, as well as any -c. -c. /- ...... potential for increased de-icing of propeller blades/cuffs. maw

Figure 1-6 InletiFrame Test Setup

Extemally heated oil was supplied to the gearbox and circulated by motoring on the starter generator. This provided engine frame anti-icing for those surfaces normally protected by oil temperature. Hot air was generated externally to anti-ice those engine surfaces which rely on bleed air. Skin temperatures obtained from a simultaneous engine Relative Ice Accrelion anti-icing test at Evendale were maintained to simulate anti. GQ&k@ Mm iced engine heat rejection. S.L.S3'C. 2gm/m: 301 1.00 .06 5000 M 25 MACH -1PC 2.2 gmim? 2Op .43 .or In addition to the ice crystal and mixed conditions, the standard test liquid water content (LWC) was factored up. 5000 M 25 MACH -20% 1.7 gm/m*, 2Op 30 .05 This accounted for the additional water impingement cre- igure 1-7 Spinner Ice EliminationIReduction ated by the aircraft normal cruise speed of 260 knots vs the tunnel maximum speed of 195 knots. Three things were Simultaneous to these source reductionielimination learned: (1) the engine frames did not ice or pack slush, (2) efforts, GE embarked on an extensive program to evaluate the IPD required additional flow-through ventilation to and improve the effectiveness of the engine separator reduce packing of slush and aerodynamically assure that the system for target size pieces of ice. Trajectory analyses, slush would not enter the engine, and (3) the factored LWC using the standard axi-symmetric model of the separator, put additional load on the inlet anti-icing system requiring were conducted for varying degrees of adverse particle further improvements. One of these improvements was a preconditioning imposed by the turboprop S-duct. Baseline heated IPD exhaust chute. cases were correlated with factory test results for 3 gram pieces of ice (.75 inch dia). During the above-mentioned second engine test at Evendale, this time with aircraft inlet installed. a single Several flowpath modifications were analyzed for FOD event was observed. Using the inlet video cameras, efficiency improvement and the most promising was se- the event was traced to ice build-up inside the inlet tem- lected for model testing (See Figures I-8A through I-8D). perature sensor duct, from runback along the inlet frame. Modified inlet and separator hardware model testing was As the ice built up in the sensor duct, it gradually was conducted in the GE Lynn component facility, using separa- sucked out by the compressor. A reworkable fix, demon- tor scavenge flows confirmed from flight test measure- strated later in this test, was implemented immediately ments. The mixed test results (see Figure 1-9) showed no across the fleet and in production. consistent net improvement in ice protection. It was concluded that the range of inlet duct exit trajectories is too As duct improvements were gradually incorporated broad to achieve substantial IPS performance gains within into the fleet the ice FOD rate stabilized at about 0.2/1000 the confines of a compatible installation. While gains may EFH. This, however, was still an unacceptable perform- be realized for a given preconditioned trajectory, a compa- ance and maintenance penalty for the small commercial rable loss is likely for another trajectory of equal probabil- operator. As a result, a target goal of .075/1000 EFH was ity. While inlet separators are effective in minimizing the established. While further inlet improvements continued to number of damaging objects reaching the compressor, they be addressed, the possibilities of ice FOD coming from ad- are at best statistical devices which cannot be 100 percent ditional sources, and measures needed to deal with these effective. When faced with a steady input of instantaneous possibilities, were explored. damage-producing objects (as opposed to a time dependant 20-5 I

7.5 PPS CORE 10 PPSCORE

IPS FLOW -% WWRl Figure 1-9 Separation Efficiencies with Production Inlet

Figure 1-SA erosion process), separation efficiency improvements may be misleading, At this point, attention was tumed to the Stage 1 compressor blades themselves.

Eneine Inlet Svstem Develooment Conclusions 1)To ensure a consistency in the application of icing test requirements, the aircraft inlet and engine should be tested together in the most severe environment available. This environment should include ice crystals and mixed conditions. 2)Successful operation in natural icing tests may not be sufficient to preclude in service icing problems. 3)Follow-up icing testing of the final Production inlet design is required to ensure proper implementation of prototype test results. Figure I-8B 4)During the initial installation design phase, the complete inletipropellerlspinner icing environment should be assessed as a system to ensure optimum distribution of anti-icing energy.

Ice FOD-Resistant Comoressor Blade Develooment

Stage 1 Comoressor Blade Clipuing Even with imoroved inlet ducts, the ice FOD rate requiring maintenance action for CV-5 powered aircraft was about .20/1oOO EFH; still beyond customer expecra- tions. the FOD appeared as a curled stage 1 blade tip as shown in Figure 11-1, and it caused an audible compressor whine and loss in engine temperature margin. The maintenance action required removal of an axial compressor casing half for access to the damaged stage 1 compressor Figure I-8C blade(s); clipping the deformed blade material and a similar portion from an opposite blade to maintain rotor balance; and hand benching a leading edee on the cliooed airfoils.

Figure-I-8D 20-6

This action eliminated the compressor whine but did not causing erratic flight and inconsistent impact areas. The ice fully restore the engine’s original performance level k- balls were targetted to pass cleanly between the IGV’s and cause the material removal adversely impacted compressor impact the blades as close as possible to the tip without efficiency and airflow. contacting the shroud. A speed trap measured the object velocity just before entering the inlet casing. The largest Test Set-un ice chunk would Ikthe circumferential space between the A key part of developing a more rugged compressor inlet guide vanes (about I-inch at the casing) and would blade was to determine the degree of the ice FOD which permit an 8 gram sphere to pass through when the vanes could result in Stage 1 blade damage. A rig was designed were fully open (i.e., axial at casing O.D.). with a row of inlet guide vanes (IGV) ahead of the Stage 1 blisk driven by a high speed spindle at engine speeds (See The test events were recorded with a high speed cam- Figure 11-3). Due to the possibility of FOD wedging action era system at 22,000 frames/second. This provided insight between the blade and casing, a shroud over rhe blades was on the impact phenomenon and verified the impact area of used to simulate the casing. A pneumatic gun was used to the objects. High speed video (2500 frames/second), which propel the FOD objects at various velocities up to 400 ft/ had the advantage of instant replay to monitor testing, was sec (120 m/sec). The ice FOD tests however, were con- used to verify the impact area and ice velocity. The test ducted at 85 - 90 ftisec (27 m/sec), the highest possible speed was 42,000 RPM, a typical altitude cruise speed at velocities determined for ingested ice objects released which most ice FOD events occur. during flight. It is possible that ice chunks impacting on the inlet S-duct could enter the compressor inlet at slower && speeds, but the impact of these objects was considered less The initial blisk testing was done una CV-5 blade. It severe thdn the test simulation. The approximate ice veloc- was found that a 2.8 -eram ice obiect caused initial bending.-. ity was calculated by equating the acceleration of a spheri- and a 4.5 gram ice object caused a single event blade tip cal ice mass to the drag of free steam flow using a drag curl similar to field damage (See Figure 11-1). Smaller coefficient of 0.55. Figure 11-2 shows the calculated veloc- chunks of ice &e., 3 grams) can cause a blade curl from ity at the Stage 1 blade as a function of ice ball size. successive hits. Successive hits occur on a particular blade because it is bent forward of the plane of the other blades. The ice objects used for the test were spherical in The testing also determined the vulnerability of a blade shape because the initial small right cylinders tumbled, which is adjacent to a clipped blade. A clipped blade al-

12 lows more ice penetration prior to impact by the adjacent blade. This causes full tip curls from smaller ice objects.

M 10 This correlated with field observations where compressors E E with clipped blades had higher rates of maintenance action Fm 8” after the first clip. The high speed photography also re- ~ Y I c 0 vealed that a large ice object can impact and bend a blade 8TO 85 4 on its initial hit and then, due to a heavier impact on the , 0 > ~ protruding blade, cause a full curl on the next revolution. 3- 42 Y Large ice chunks (5 to 8 grams) were also fired at the pitch

v ~ sections causing minor bulging. This was not typical of ob- YI served field damage. Therefore, it was concluded that the damdging ice objects were naveling along the outer casing

ICE edLLYIEIGHT (GRAMS) wall and impacting in the blade tip area.

Figure 11-2 Ice Ball Parameters at IGV Inlet A sample of operator ice FOD data was evaluated to determine the extent and frequency of the blade curls (See Figure 11-4). The data included the bend radius, since it is the amount of mateiial removed prior to benching a new leading edge radius. About 75% of the events required a clip of 0.3 inches along the blade tip chord and 0.8 inches down the leading edge.

It was judged that a cutback of .25 inches along the tip and .65 inches down the leading edge along with a reshape of the leading edge to improve the aerodynamics would eliminate most of the problem associated with tip curls. Furthermore, a swept leading edge would present a more aerodynamic shape to minimize performance loss. This was called the “mat clip” (See Figure 11-5) and is subject ‘Iof a pending patent application. This “smart clip” was Figure 11-3 Component Test Setup 20-7

TABLE 11-1 TEST RIG ICEBALL FOD SUMMARY Object which Blisk Object which Causes Extensive Confieuration Initiates Damage Damaee or Tiu Curl CT7-5A grams grams Baseline 2.8 4.5 ~ ...~ .I2 Cutback >4 5 .18 Cutback 6 >6 25 Cutback >8 >8 Baseline m-5A 1.5 2.8 next to Field Clip CT7-5A 25 Cutback 6.5 >6.5 next to Field Clip CT7-9 Cutback >8 >8 I I Test Conditions: Figure 11-4 Operator Ice FOD Data Rotor Speed 42,000 rpm Ice Ball Velwitv 85 ftlsec evaluated in the test rig in steps of .12, .18, and 25 inches Impact Area @ Blade Tip of tip cutback. The test results in Table 11-1 show that the Initial blade cutbacks were’canied out at the opera- .25 inch cutback eliminates any damage that could be tor’s shop or during on-wing maintenance that required caused by an 8 gram iceball passing between the inlet guide access to the compressor. After the “smm” blade cutback, vanes at 85 ft/sec. Although these larger objects travel virtually all ice FOD problems were eliminated with that more slowly when accelerating in an engine inlet, these engine, As the field engine population increased in cutback tests were conservative because they held velocity constant compliance, corresponding reductions in maintenance at the upper limit of smaller objects. action occurred (See Figure 11.6). As of December 1989, the rate has been reduced by over 30 times to less than .005/1ooO EFH with 98% compliance. The performance effect of the ‘‘smart clip” has been small particularly when carried out with other compressor refurbishment which, in itself, offsets the cutback Stage 1 blade performance loss.

a24 O.m om

Figure 11-5 “Smart Clip” Stage 1 Blisk Figure 11-6 CT7-5 Turboprop Ice FOD Reduction From “Smart Clip” Compliance A significant reduction in the baseline blisk damage resistance was observed for a blade positioned next to a Additional ice testing and further review of the high clipped blade. The results show that the initial damage speed films was done to understand whether blade leading threshold was reduced by objects from 2.8 to 1.5 grams. edge thickness or sweep back geometry was influential in This is also where major curl damage was reduced by improving ice FOD resistance of the “smm clip” blade. objects from 4.5 to 2.8 grams. Also, the 25 inch cytback Blade leading edge shapes of the test hardware were photo- showed initial damage occumng at 6.5 grams weight, graphed, measured and correlated with the test results. This whereas a uniformly cutback blisk was above 8 grams (See follow-on investigation which included tests of swept back Table 11-1). blades with thin leading edges showed that good correlation could be obtained between damage resistance and thickness measured .04 inches back from the leading edge at a location 0.5 inches down from the blade tip. Figure 11-7 is a plot of ice ball size and leading edge thickness showing a 20-x

line of demarcation between damaged and undamaged This investigation has demonstrated the importance of airfoils. For this plot, damage was assumed LO be any de- leading edge thickness in preventing ice damage to blades. tectable plastic deformation of the blade leading edge. This It allows the engineer to design the best aerodynamic blade plot points out the sensitivity of the damage resistance to configuration including leading edge contour and assure blade leading edge thickness. good ice damage capability by controlling the leading edge thickness. This assumes that the remainder of the airfoil has sufficient smength to accelerate the ingested ice pa- ticles up to wheel speed without deformation.

Ice FOD-Resistant Comoressor Blade Develooment

Conclusions

1)Blade leading edge thickness is the most important parameter for resisting ice FOD damage. 2)A full tip curl can be caused by a single large ice object, or by a succession of smaller objects which, initially, only partially deform the airfoil. 3)Airfoils which protrude forward of the plane of the other blades are likely to be hit repeatedly and more severely. Similarly, airfoils following more deeply cutback UE THICKNESS 1 OW CHORD, 3.5 RADIUS1 -!A REFERENCE blades are more vulnerable to damage caused by deeper ice Figure 11-7 FOD Resistance Correlates Well With L.E. penetration prior 10 impact. Thickness 4)Increased ice velocity and size increase blade dam- The blade leading edge thickness is felt to be critical age. However, larger panicles accelerate slower and have since damage is initiated at this location. More dramatic lower velocity at blade impact. blade curling occurs later due to the increased exposure of 5)As a rule to durability, the stage 1 compressor blades the blade because of the fonvard deformation of the leading should be capable of withstanding impact from any piece of edge. The high speed films show the ice ball is shattered by ice that can pass between the inlet guide vanes. the initial blade impact which causes a significant load on 6)Cutback modification to the leading edge of the the leading edge. However, blade curling is caused by the CTi-5 Stage 1 blade has an ice FOD resistance that will high loads imposed on the blade as it accelerates the in- eliminate most field damage. gested mass of ice up to wheel speed. The measurement location of .04 inches from the leading edge is somewhat arbimary but does provide a quantitative measure of the lead edge bluntness where the initial ice impact occurs, Figure 11-8 shows enlargements of the leading edge shape of two airfoils and how the relative thickness at .04 inches characterizes the leading edge shape. Velocities of the ice ball and blade are such that the ice is ingested at a rate of approximately .07 inches per blade at 80 feet per second. This adds further justification to a thickness measurement near the leading edge.

BLUNT SHARP I BLADE BLADE I

Figure 11-8 Thickness at 0.04 Inches From L.E. Character izes L.E. Bluntness 20-9

Discussion 3. R. Jones, Boeing Arnprior Canada What is the mechanism by which the cut back on the leading edge of the blades prevent fad from one inch pieces of ice? 1. H. Saravanamuttoo, Carleton University Could you please indicate how many compressor blades Author: were subject to damage in a typical FOD incident, and could The cut back increases the leading edge thickness resulting you give some quantitative idea of the penalties in in a stronger blade. Figure 11-7 illustrates that only modest compressor efficiency or engine temperature margin? increases in thickness are required.

Author: 4. M. Holmes, RAE Most ice-fad events involved a single blade which resulted in What lessons were learned as a result of the CT?-5 ice fod an audible whine from the compressor and a 5- 10 degree C problems in service on repeat of qualification of future shift upward in T5 temperature at constant torque. The engines? repair, which clipped off the bent blade material and a similar amount from the opposite blade, costs about 1 point Author: in compressor efficiency and about 6 C in temperature First, compressors should he capable of handling ice objects margin. large enough to pass cleanly between the inlet guide vanes. Second, ice accretion in the inlet system must be neglegible 2. D. Mann, Rolls Royce or limited to sizes easily handled by the compressor blades. Did you at any stage operate your trajectory code in a This issue of qualification (certification) often focusses on reverse mode, in an attempt to assess where the FO could the safety of flight issue which was not the problem for CT?. have come from, upstream of the damaged parts of the Our problem was unscheduled maintenance from ice-fad compressor blades? and I do not think these kinds of issues are easily It appears from figure I-SA that during the course of demonstrated in an engine qualification programme. assessment of different designs, no consideration of changes to the splitter lip was made, despite the fact that this is an area which has a significant effect on separation efficiency. 5. C. Scott Bartlett, Sverdrup Technologies Could you comment on this? Would you please comment on considerations given to the type of the ice used during your ice ingestion tests and if you Author: feel there is a difference in damage due to ingestion at This is an interesting suggestion but we did not evaluate this different types of of ice? approach. We suspect the ice objects are much slower than the air stream and they tend to slide along the outside walls. Author: The tests with cameras viewing the inlet duct demonstrated We felt that ice near freezing would be less brittle and may this behaviour. cause larger energy transfer to the blade because it is less Splitter lip changes were to extensive for consideration as a likely to scatter upon impact. Therefore, we allowed the ice field fix. Major redesign certification programmes would be to warm up slightly before test so that there was some liquid needed to implement those kind of changes. on the surface of the iceball.

FUELS AND OILS AS FACTORS IN THE OPERATION OF AERO GAS TURBINE ENGINES AT LOW TEMPERATURES by G L Batchelor Eng 2a Procurement Executive, Ministry of Defence St Giles Court, 1-13 St Giles High Street London WC2H 8LD United Kingdom

Two factors strongly influence the low temperature behaviour of aero gas turbine fuels and oils. They are Viscosity, and State or phase change - i.e. whether the material is liquid or solid. In fact the question is whether solids are, or are not, present because, although the whole may cease to flow at some designated temperature, the lighter components of hydrocarbon and other organic mixtures are likely to he liquids under all natural circumstances. Ternis such as Freezing Point, Pour Point, and the like, will be familiar enough. This presentation examines the chemical and physical realities underlying such parameters and considers their impact on aero gas turbine engine performance. For the purposes of this paper fuels and oils will be treated quite separately.

Part I : FUELS

FUEL TYPES impact. They are water itself, and the fuel system icing inhibitor (FSII) added to All ordinary aviation turbine fuels are all military fuels to combat any free water variants of kerosine. The Wide-Cut fuel forming in those aircraft not equipped with F-40, or JP4, is kerosine plus naphtha. engine fuel filter heaters. The current European Theatre primary fuel Water is present in solution in all fuel F-34, or JP-8, is Kerosine per se. The at equilibrium concentrations of the order naval High Flash fuel F-44, or JP-5, is a tens of parts per million. FSII is added slightly higher boiling kerosine cut which at tenths of a percent by volume. can, on occasion, satisfy either F-34 or The water occurring naturally should not F-44 requirements. Such differences may be be confused with that which may arise manifest in fuel low temperature behaviour. through contamination of the fuel supply. The latter is a logistic concern not to be addressed here. FUEL COMPOSITION

From the low temperature performance FUEL COMPOSITION EFFECTS aspect there is need to take account only of those organic compound classes present (A) Phase Change in "bulk" proportions; i.e. whose percentage contribution will be at least five, and WATER. Apart from the fact that its will likely be numbered in tens. Namely:- equilibrium solubility varies marginally with hydroc,arbon class, water will behave Paraffins (normal and iso) almost independently of fuel type. When in solution, water is no problem. But it is a Naphthenes (or cyclo paraffins) low temperature performance issue due to the simple natural fact that from about 0°C Aromatics (single ring) and on downwards the dissolved water will crystallise out as ice. Although that but only such members of each class as are process is virtually complete at -30oC, the allowed by the boiling range of any given ice crystals tend to remain in suspension. fuel. Those which do reach the tank walls form an ice film and are thereby inactivated until Two trace components have significant that film melts. Another wholly natural 21.2

source of water is precipitation out of the the fuel will be wholly liquid in terms of aircraft fuel tank ullage air. its hydrocarbon make up. 'The function of the F'SII is that of an FLOW POINT Flow Point addresses the "anti-freeze" which so modifies water:fuel threshold of flow. Under the title "Pour partitioning as to obviate the blockage of Point", it is a common enough parameter for engine filters by ice. There is a popular oils and diesel fuels. In the context of niisconception that E'S11 operates on the jet fuel it has specific meaning . Freezing Point (see below) of the fuel When fuel temperature continues to fall itself. IT DOES NOT. below the Freezing Point, more and more hydrocarbons crystallise out until a point So much for water. is reached a.t which the solids form a stable matrix. Precisely at that point the HYDROC,ZRRONS Generally speaking, the still liquid components remain fully mobile heavier nwmberE: of a given series can be but very soon thereafter even they become regarded as being dissolved in the lighter entrapped in the growing matrix, and the ones. Ccrtaitily it is the heaviest which fuel then becomes a de facto solid. crystallise first on cooling. adding to The jet fuel criterion of flow concerns the lighter end of a fuel cut (if that be the ability of the crystal matrix to nn option) allows greater flexibility at collapse at its unsupported edges under its the other end. This is well demonstmted own weight such that the two phase mush by wide-cut fuel which has a Freezing Point can still flow of its own volition to the 10 degrees or more below that of ii kerosine aircmft tank lbacking pump; thereafter, any but hns the highor final boiling point. mechanical work on the fuel ensures that Straight chain normal paraffins are the the phase balance cannot again worsen. really "waxy" types which most readily and Flow Point has attracted attention *ram distinctly crystallise out. The phenomenon time to time (chiefly at times of perceived becomes increasingly less definitive as the fuel shortagel but has yet to attain pract- molecular pattern compacts through ical significance. branched structures to saturated rings. Without there having been cause for full There are also inter-specie effects. consensus on the issue, Flow Point has thus Normal fuels being always comprised of far been defined in terms of recoverable mixtures, the low temperature behaviour of fuel under designated test conditions. any particular fuel will be a function of the balance of its components. h mixture VISCOSITY In an aviation fuel context of two fuels of differing composition will the parameter is the kinematic variety. not necessarily average the properties of the two individuals. This is not a practical concern except for the rare event MEASUREMENT of both being at the specification ma.rgins. FREEZING POINT Determination entails ( B I Viscosity cooling a fuel sample in glass apparatus at a controlled rate to just beyond the point So long as they remain wholly liquid, of hydrocarbon crystal formation. Normally most fuels are fully Newtonian in their this is readily distinguishable from the flow. 'That. is true for the more paraffinic ice crystal haze mentioned above. The fuel examples even to below their Freezing is then allowed t.o warm until the last Points. But this is open to some question hydrocarbon crystal melts. This procedure with the extremely cyclo-paraffinic types is repeated until the temperatures of yielded by modern hydrocracking processes appearance and disappearance differ by no .. a point to which I shall return later. more than some prescribed amount. With normally paraffinic fuels all is stmightforward and operator experience is THE FUEL PARAMETERS not critical. When, as with some of the newer cyclo-paraffinic fuel varieties, the FREEZING POINT Rest known of all the transition points are difficult to observe, fuel low temperature parameters, Freezing and temperature differentials cannot be Point addresses the transition between a met, subjectivity increases markedly. fuel being wholly liquid and thex being solid hydrocarbons present. Thermodynamics ALTERNATIVES TO FREEZING POINT The dictate that both the first appearance of fact that Freezing Point determination is crystals on cooling, and their final dis- labour intensive, coupled with the fact of appearance upon warming, shall take place its increasing element of subjectivity, has at different temperatures. The definition driven a search for automation. Any auto- of Freezing Point centres on disappearance; matic equipment capable of truly replic- it marks the lowest temperature at which ating the Freezing Point has thus far 21-3 proven to he extremely sophisticated. But parameter viability) is far from being there are now several simpler, portable, valid. devices capable of indicating the Freezing Point, or something closely related to it. (B) Viscosity The importance of these does not lie in their specification potential but in the Because it is a design criterion that possibilities they offer for greater oper- fuel viscosity shall not exceed 12 cSt at ational flexibility. At present, although any functional point in the engine system, Freezing Points are frequently well within the parameter is not generally seen as a specification, the absence of an actual performance or operational consideration value at the point' of refuelling means that other than when the aforesaid criterion is, aircraft flight plan must presume a value for some reason, not met. Certainly there at the specification limit. A "plane-side'' have been spectacular instances of starting Freezing Point capability would be a difficulties with kerosine fuels under considerable operational boon. unusually cold conditions. Certain hydromechanical devices apart, FLOW POINT The only test method thus viscosity at the burner spray or atomiser far formally published is that developed by is the main concern. the UK Institute of Petroleum in the late 1950s when it was thought that there might be insufficient fuel of low enough Freezing SPECIFICATION Point to meet the requirements of the then dawning inter-continental commercial jet It is quite usual to think of fuels in aircraft traffic. It entails a vertically terms of their specified Freezing Points: cylindrical test vessel at w,hose bottom end viz .. is a conical valve capable of sudden release such as to allow the fuel to flow - -40 ["CJ for Jet-A; -47 for Jet-A1 if it can. and JP-8; -58 [-7ZTI for JP-4: etc. However, the purpose of this short section OPERATIONAL and PERFORMANCE IMPACT is to briefly review the manner in which specifications address the low temperature (A) Phase Change requirements.

FREEZING POINT The parameter is often (A) Phase Change seen as being an airframe concern inasmuch that any resultant constraints applied to FREEZING POINT Although its definition aircraft routeing arise essentially out of is centred on crystal disappearance, the the relationship between meteorologically latter is only the indirect basis of limits projected fuel tank temperatures and set by fuel specifications. understood fuel Freezing Points. The reality is best illustrated by the outcome of the lengthy ASTM debate which But, in the final analysis, it all comes attended the lowering of the Jet-A1 limit back to the engine. There seems to he a from -504: to -474: in response to the fuel universally accepted stipulation to the supply crises of the late 1970s After due effect that fuel shall he delivered by the consideration of data from the airlines and airframe to the engine three degrees above from Boeing's studies of its 747 aircraft, the Freezing Point - the latter again taken the consensus finally settled at minus 47 as being at the specification limit. on the basis that ...

Such a rule presumably purports to The best overall number for in-flight ensure single phase flow regardless of the minimum fuel temperature was -43oC. precision of the Freezing Point test method ADD the 3" margin for the engine. employed (of which aspect more anon). ADD 1 more for good measure!

FLOW POINT The parameter, as it has so It might appear, in the light of the far has been defined, is perhaps a little data entailed, that there is over-caution pessimistic in that vibration, coupled with here. But, the data is only as good as the attitudinal g forces, will very probably measurements made in those aircraft tanks keep a crystal matrix unstable below the wherein measurement was possible. temperature indicated by static laboratory testing. Caution is not misplaced. The protagonists However, the blithe assumption that of specification change are usually looking ordinary aircraft might somehow be able to only at the fuel side of the equation. One utilize semi-frozen fuel (thus giving the hears tell of "difficulty" in getting fuel 21-4

out of wing tip tanks in cases where the Again I sub-divide into already noted operators have sailed close to the Freezing categories; but I discount Flow Point as Point wind. even a medium term recourse.

FOR SO LONG AS AERO GAS TUREIINE FUEL FREEZING POINT This is certainly a simple SYSTEMS ARE TO BE DESIGNED FOR SINGLE and convenient parameter but must we PHASE HYDROCARBON FLOW ONLY, THEN SO remain forever locked-in to the concept of LONG WILL FREEZING POINT REMAIrV THE KEY nil solid hydrocarbons? Cannot engine FUEL LOW TEMPERATURE PARAMETER. systems handle just a few? Admittedly that begs the question of FLOW POINT There is little that can be defining a “few“ but let fuel specification said, in a specification context, of a fuel writers worry on that score for the moment. parameter that has yet to he specified. From a hardware aspect the solution to such a problem may already be to hand. It (B) Viscosity may simply be a matter of the batting order for the fuel system’s components. Are they Although the 12 cSt hardware limit is always in optimum sequence? Once an engine independent of temperature, a specification is running heat is plentiful. Indeed, a must stipulate a temperature of testing. current concern is the limiting capacity of This will preferably be some convenient fuel as a heat sink. laboratory norm. For aviation kerosine If the heal. exchanger can be as far up- that turns out to be -2OoC whereat an 8 cSt stream as possible, then only at start-up maximum satisfies the operational need. would there be need for auxiliary energy to Wide-cut fuel is, by its very nature, raise the incoming fuel through a probably well inside the viscosity mark; but, even very small temperature increment. were that not so, its volatility would put routine measurement beyond reach. Test methods coming under the heading ALTERNATIVES TO FREEZING POINT may well offer a route to definition of acceptable HARDWARE DESIGN IMPLICATIONS solid hydrocarbon levels - they may even have on-line potential. (A) Airframe

The aircraft fuel system must respect FRESHSTUDY Flow Point in order to deliver fuel at all, and respect Freezing Point in order to VISCOSITY may be deserving of detailed deliver fuel of acceptable quality. reappraisal in the aero engine context. Warming a just flowing fuel mush to an engine-acceptable condition has to entail Hitherto, we have been accustomed to aircraft fuel tank heating .. with the fuels whose flow has remained consistently necessary heat deriving ultimate1.v from the Newtonian albeit that their Freezing Points engine. have been relatively high. Now we see Boeing, and others, have carried out hydro-cracked fuels whose Freezing Points, numerous paper studies - often with seeming though hard to measure, are very low. enthusiasm. But the outcome, he it initial Is enough known about these liquids at installation or retrofit, always carries so those temperatures? Might their reluctance many, and such varied, penalties that I am to crystallise be accompanied by some form forced to the view that such action would of nucleation in solution? Do perceived be set in train only by the most traumatic viscosity temperature relationships hold of fuel eventualities. for them? Above all, how might any such aberrant behaviour impact on combustor (B) Engine performance?

On the engine side, there are far less In recommending this as an area ripe traumatic options which can, and perhaps for study I would point out that, to should, be looked at in the shorter term as be meaningful, such studies would a safeguard against any future fuel supply entail the most careful fuel eventualities. modelling. PART I1 : ENGlNE OILS

Aviation fuels come directly from naturally available resources. They are subjected only to the minimum of processing and/or additive treatment essential to ensure specification compliance. The oil situation is very different. Petroleum based mineral gas turbine oils today form a minority category; they will not be addressed here. The majority of current oils are "synthetic" in that their primary constituents are synthesised from raw materials of animal or vegetable origin. Such oils are in fact designed with specific user engine types view. Their successful manufacture is as much an an art as it is a science.

OlL TYPES with shorter chain acids - principally esters based on .. Aero gas turbine synthetic oils are mostly categorised by notional viscosities (2) Tri-methylol-propane (TMP) at 409: .. viz: or, 3 cSt eg .. Nato 0-148, or 0-150 (3) Penta-erythritol (PEi 5 cSt eg .. Nato 0-156, or 0-160 The tu;o last named allow a measure of 7+cSt eg .. Nato 0149 product tailoring through variation of the esterifying acid or acid mix. Multiple designations under a single category arise out of differing national qualification processes or out of differing ADDITIVES performance levels (0-160 has a higher load bearing capability than 0-156). These regulate thermal and oxidative There is no specific categorisation with stability, load bearing capability, hydro- respect to low temperature performance. It lytic stability, corrosivity, foaming, and is, of course, easier to achieve a desired so forth, but not, it should be noted, low low temperature capability with a thinner temperature performance per se. oil. Oil viscosity is as much an engine The low temperature concern is that design parameter as it is an oil one albeit additives, particularly any which happen to that some engines can use oils of differing be at the threshold of true solution, might viscosities. precipitate irreversibly - especially when exposed to prolonged soak. This is as much a problem between oils OIL COMPOSITION as it is one within an oil.

A synthetic aero gas turbine oil is usually characterized by the type of ester PHASE SEPARATION and VISCOSITY making up the bulk of oil volume and thereby giving the product its underlying That final comment re additives said fluidity and elasto-hydrodynamic quality. virtually all that there is to tell about To this base will be added various low temperature phase separation in oils. additives whose purpqse(s) will be to confer, or enhance, specific properties of VISCOSITY is the key. And the key is to the finished lubricant. engine starting. I exclude from this discussion circum- ESTERS are the reaction products of stances in which aircr$t heating, or other alcohols and organic acids. They can be external aids to starting, are a recognised looked upon as falling into three classes:- part of an operational scenario. Given such exclusion, interest centres (1) [The first on the scene] Long chain on starting torque engendered by the oil. acids coupled with relatively simpler That will be primarily an engine design alcohols .. generally termed di-basic acid parameter but consequential oil design will esters - eg di-octyl sebacate, or its depend also on the lower operating temp- azeleic or adipic counterparts. erature envisaged for the user aircraft. Such limits typically vary between -25% Thereafter come complex alcohols coupled and -54oC. 21-6

The real problem with attaining approp- At the type approval stage it is usual riate low temperature viscosity is that of also to confirm low temperature homo- matching this particular need with the geneity empirically in the laboratory. parallel one of having sufficient viscosity at normal running temperatures. In fact, as a dominant factor, the oil OIL PRODUCTION low temperature requirement has but brief duration. Once the engine is up and The raw materials for synthetic aero running, oil quickly takes on the crucial engine oils derive from natural renewable r6le of a heat transfer medium. resources. They are, therefore, subject to climatic and other environmental vagaries. Moreover, aviation is only a minor user OIL COMPOSITION EFFECTS of them and in no position to dictate the shape of the market in any fundamental way. It is not really realistic to set out to As a consequence, oil suppliers prefer examine composition as an option vis-a-vis to have a range of optional compositions oil low temperature performance. Such are approved for any product, albeit that some the conflicts with the high temperature options may be superior others in the final imperatives, and such is the increasing analysis. weight given to the latter, that there is little option other than comproniise at the low end. ENGINE DESIGN and OPERATIONAL Certain advanced engine systems (using CONSIDERATIONS specialised lubricants) have resorted to oil dilution .. itself a fairly venerable In the matter of oil superiority there dodge in the piston engine world! Others very definitely are "Horses for Courses". allow for higher than normal viscosity (at Not all oils perform equally well in all low temperature) in relatively conventional the engines for which they are notionally oil types. suited. A sometimes necessary reverse comprom- Different engines do not rank the same ise (from a lubricant design standpoint) is oils in a single order. Indeed, different constraint on additive addition where an parts of the same engine may rank a given otherwise optimum treatment level might oil set differently. And this even more result in unacceptable low temperature true when the ranking is done with regard characteristics. to differing oil performance parameters.

It is a not unreasonable generalisation There is, therefore, a need perhaps to to state that , for any given oil type, rank the parameters. We might observe at whilst it may be possible to move the range once that the exigencies of aircraft perf- band of high to low temperature capability ormance seem already to have ranked the oil up or down the temperature scale, it is high temperature requirement above the low. very difficult to widen that band. That beint: so, there is need to examine the low temperature requirement itself. SPECIFICATION and MEASUREMENT After start-up it involves a very small part of the fight envelope other than for The only quantified oil low temperature those APUs and other accessories which may parameters are Viscosity and Pour Point. be exposed to static cold soak on long flights; they might be allowed to windmill, VISCOSITY is measured in the conven- or be started-up periodically, to maintain tional manner. Specification limits have their readiness. generally been predicated to a functional Furthermore, low temperature starting kinematic viscosity of 13,000 cSt at -40%. occupies only a small part of the total However, as implied above, high temperature operational envelope; one ponders the performance demands are forcing a re-think extent to which it should be allowed to and numbers such as 14,000 cSt at -20°C are predicate overall oil design? being mooted for some impending aircraft projects. FUTURE OIL DEVELOPMENTS POUR POINT is an empirical observation of the limiting temperature at which an oil The viability of alternative "chemis- just flows of its own volition. The spec- tries" for the lubrication of aero ification limit is usually -60%. It is a gas turbines in the broad inventory batch quality control and is also used in is a subject of current and on-going screening new products. interest and review. 21-7

3. F. Gamier, SNECMA Discussion Pour les turboriacteurs, on a par16 aujourd'hui des huiles de 3 et 5 cst, 5 jusqu'a -40 C, au dessous mais on parle aussi 1. R Sabla, GEAE maintenant de plus en plus de nouvelles huiles de 4 cst. Que Should surface tension he considered as a controllable seraient les avantages de ces nouvelles huiles par raport aux parameter to enhance the atomization at low temperatures? autres, auront-elles les mhescomportements? Author: An additive approach would he viewed with caution as it carried unacceptable penalties in other directions. The Author: desirable first step is a deeper understanding of the These standard viscosity classifications are predicated to relationship between physical effects, such as atomization +40 C. I doubt that a 4 cst oil would necessarily offer and fuel composition. improved low temperature performance above 5 cst unless specifically designed with such an end in view. 2. M. Holmes, RAE What is the fuel property that most limits the low temperature operation of turbo engines? Is this a serious limitation and, if so, does your organization have any 4. D. Way, RAE programmes to find a solution? At long distance aircraft, the ranges are becoming longer Author: and longer, and the danger of the fuel coming to lower temperature increases. this a concern? At the moment freezing point, in the future perhaps also Is viscosity, as the aircraft criteria are presently defined, are absolute limitations (goho go). The fuel man can offer little service. Any solution open to him must necessarily limit fuel Author: availability or increased fuel price or both. Limits can be It does on transatlantic flight with slower aircraft. altered only with the consent of the aircraft design side. The With normal aircraft, it is somewhat compensated by fuel side does attempt to maintain a data base for current aerodynamic heating. The problem increases with tip tanks and foreseeable fuels. Perhaps that needs to he reconsidered with a smaller volume to surface ratio and a thin skin. For with design options for the eventuality of a fuel availability extremely long distances there must he a compromise shortfall. between the commercial and the technical interest.

22-1

THE EFFECT OF FUEL PROPERTIES AND ATOMIZATION ON LOW TEMPERATURE IGNITION IN GAS TURBINE COMBUSTORS by D.W. Naegeli, L.G. Dodge and C.A. Moses Southwest Research Institute 6220 Culebra Road San Antonio, Texas 78228 U.S.A.

ABSTRACT INTRODUCTION

Experiments were conducted in a T63 engine The fuel qualitylavailability ratio has become combustor to gain a better understanding of the complex in recent years because of the uncer- role played by volatility and atomization in low tainty in the supply of petroleum. Altematives, temperature ignition. Eight test fuels were used, such as synfuels, tend to have different boiling some of which were specially blended to vary ei- point distributions and increased viscosities when ther viscosity or volatility while holding the compared with their petroleum equivalents. This other constant. Six atomizers were used to vary trend toward reduced volatility and increased vis- the fuel spray characteristics, and average drop cosity could cause degradation of low tempera- sizes, represented by Sauter mean diameter ture ignition, altitude relight, and flame stabiliza- (SMD), were measured. Air temperatures were tion in gas turbine engines. varied from 239 to 310K. Ignition comparisons were made by the minimum fuel-air ratios neces- Several aspects of the ignition process in gas tur- sary to achieve ignition. Significant results in- bine engines are not clearly understood. Combus- cluded: 1) viscosity, which determined atomiza- tor rig tests have been effectively used to demon- tion characteristics, was more important than strate the effects of varying fuel properties on volatility in the ignition process; 2) ignition de- ignition and flame stabilization.('-") Anal tical models have been developed to correlate (K2-17) pended more on achieving a critical drop size than on reaching .the lean-limit fuel-air ratio; and data from several combustors. However, the mod- 3) fuel temperature was found to be more impor- eling efforts have been seriously hindered by the tant than air temperature for low-temperature ig- absence of information on the effects of fuel vol- nition, an effect due principally to viscosity and atility, and atomization. atomization rather than evaporation. A practical implication is that fuel heating would give a It has been found(18) that ignition models are much greater improvement in cold-start perfor- quite sensitive to the drop-size distribution in the mance than heating the combustor inlet air. fuel spray, characterized by the Sauter mean diameter (SMD), but this sensitivity has not been NOMENCLATURE properly measured in the combustor data available to date. Previous studies (13-15)have em- reference velocity ployed correlations that were based on measure- fuel-air ratio ments of SMD in sprays from atomizers that Sauter mean diameter, micrometers were similar, but not sensitive to the subtle dif- kinematic viscosity, cSt ferences in the design of the actual fuel atomiz- surface tension, dyneskm ers used in gas turbine combustors. The purpose of the present study was to provide a clearer def- flow number, kgh -6a inition of which fuel property, viscosity or vola- mass flow rate, g/s tility, plays the more important role in the igni- coefficient of determination tion process. Rosin-Rammler Parameters for: R cumulative volume fraction of spray EXPERIMENTAL FACILITIES AND D droplet diameter METHODS X droplet size parameter N width of distribution General Description - This work was performed in the combustor facility of the Belvoir 22-2

Fuels and Lubricants Research Facility (BFLRF) the radiation sensor used to detect the onset of a at Southwest Research Institute (SwRI). The visible flame in the ignition measurements. combustor facility was specially designed to study fuel-related problems in the operation of The T63 combustor employs an igniter that pro- gas turbine engines. The air supply system produces sparks at a rate of 8 per second with ener- vides a clean, smooth flow of air to the combus- gies of about 0.87 joules, and a dual-orifice pres- tion test cell with mass flow rates up to 1.1 sure-swirl atomizer. The primary orifice is used kg/s, pressure to 1620 kPa (16 arm), and temper- principally for tlie atomization of fuel in the igni- atures from 239K (-30'F) to 1089K (1500'F) tion process. The secondary orifice supplies the (unvitiated). For cold start ignition tests, the com bulk of the fuel but does not engage until the bustor inlet air is cooled by injecting liquid nitro- fuel flow rate is above that required for ignition. gen and gaseous oxygen through a manifold To further examine the effects of atomization on placed upstream of the combustor. This system ignition, an adapter was constructed to allow use is capable of lowering the teFperatuze of the of single-orifice, Delavan pressure-swirl atomiz- burner inlet air supply to -34 C (-30 F) for the ers with flow capacities ranging from approxi- mass flow rates (0.09 to 0.27 kg/s or 0.2 to 0.6 mately 15 to 30 litershr (4 to 8 gal./hr) at a dif- Ibh) used in the T63 combustor ignition tests. ferential fuel pressure of 689 kPa (100 psid). Turbine flowmcters and strain-gage pressure transducers are used to measure flow properties Test Fuels - The test fuels described in TABLE of the air and fuel. Themoocouples are refer- 1 were chosen for the ignition study because of enced to a 338.5K (150.0 F) oven. Data reduc-

tion may be performed on-line with test summa- TABLE 1. Fuel Prooerties~~ ries available immediately; these summaries -1 provide average flow data as well as standard deviations (typically less than l percent of average value) of inlet temperature and pressure, exhaust temperature, flow rates of fuel and air, emissions data, and combustion efficiencies. Further detail on the facilities and procedures described below may be found in Reference 19.

T63 Combustor Rig - The combustor used in this study was fabricated from T63 engine hardware. This combustor has been used in previous programs to study the fuel effects on ignition, combustion stability, combustion efficiency, exhaust pattern factor, radiation, and smoke. Figure their broad range of viscosities, and volatilities. 1 is a schematic of the combustor rig showing To independently study the effects of viscosity the locations of the fuel atomizer, igniter, and and volatility on the ignition process, Fuel 4 RADIATION was blended with a viscosity equal to that of SENSOR JP-5 and a front end boiling point distribution similar to JP-4. Fuel 5 was blended with a vis-

IGNITER ,/ cosity equal to that of JP-4 and a front end boiling point distribulion similar to gasoline. The viscosities of gasoline and methanol are essentially the same, but their 10 percent boil-off temperatures and heats of vaporization are very dissimilar.

Droplet Size Measurement - Atomization is a crucial factor in the ignition of fuel sprays in gas turbine engines. Ignition failure occurs with diminished atomization quality because of the de- / 'BURNER creased fuel surface area available for evapora- COMBUSTOR CAN tion and the increased spark energy required for HOUSING droplet evaporation. For a given engine design Figure 1. T63 combustor 22-3 poor atomization is caused, for the most part, by The Delavan atomizers were single-orifice pres- increased fuel viscosity; consequently, higher vis- sure-swirl (simplex) nozzles differing only in cosity fuels and lower fuel temperatures lead to flow rate capacity. Compared to the standard higher fuel/air ratio requirements for ignition. It T63 atomizer, the droplet-size measurements in is a general rule that when fuel viscosities ex- sprays from the Delavan atomizers were rela- ceed 12 cSt combined with lower burner inlet tively straight forward. air temperatures, the probability of ignition in gas turbine engines is greatly reduced at any The test fluids and test conditions used in the at- fuel-air ratio. omization measurements are given in TABLE 3.

To better understand the ignition test results in Table 3. Test Fluids for T63 IgnitiodAtomiza- the T63 combustor, droplet-size measurements tion Study to Generate the Correlation of Sauter were made on hollow cone sprays produced by Mean Diameter (SMD) with Fuel Properties and the atomizers described in TABLE 2. Flow Conditions

Table 2. Fuel Atomizers \Flaw Cawcity*/ Flow I cone Description @ 689' kPa Number* Angle (g/s) (kg/s-@a) (degrees I Factory T63 Dual Orifice Pressure Swirl )clavan Simplex 5:; 5:; x 1v6 Pressure Swirl 1 1 7.143.83 x 1 ;; Mavan Simplex 4,01 4.83 x 90 Pressure Swirl

'elavan smplex 6.18 x The first three of these fluids, JP-4, JP-5, and Pressure Swirl I ::E 1 1 NDF are test fuels described in TABLE 1; Ielavan Simplex 7.08 x HMGO is a heavy marine gas oil selected be- Pressure Swirl :: cause of its relatively high viscosity. The flow Ielavan Simplex 7,00 8.43 x lo6 90 rates ranged from 1 to 7 g/s depending on the at- Pressure Swirl omizer and fuel combination examined. Capacity and flow number for Jp-5 *Primary nozzle only Drop-size data were obtained with a Malvern Model 2200 particle sizer based on the diffraction angle produced by drops when illuminated The standard T63 atomizer is a dual-orifice noz- by a collimated beam of mono-chromatic, coher- zle that uses a flow divider valve internal to the ent light from a HeNe laser. A 300-mm focal nozzle body to determine the fuel flow split be- length fn.3 lens was used to collect the scat- tween the primary and secondary nozzles. At tered light. The laser beam diameter was 9 mm low flow rates, such as those used in ignition, with a Gaussian intensity distribution truncated the valve completely restricts the flow of fuel at the edge by the 9-mm aperture. Data were re- through the secondary nozzle. As a result, the corded at an axial distance of 38.1 mm (1.5 in.) droplet-size measurements were made only on downstream of the nozzle tip. sprays originating from the primary nozzle. Two procedures were necessaly to ensure the The standard T63 atomizer is also equipped with proper calibration of the particle sizing instru- passages to carry burner inlet air through the ment. The drop-size distribution is computed nozzle to prevent deposit buildup on the nozzle from the relative scattered light intensity mea- face exposed to combustion gases. However, the sured at different scattering angles by a set of airflow was also found to affect atomization 30 annular ring photodiodes. and, therefore, had to be accounted for in the droplet-size measurement. To simulate this air- For an ideal optical and electronic system and flow, a special manifold was designed to allow uniform responsivity between photodiodes, the the separate flows of fuel and air through the at- signal intensity may be used to compute the omizer. drop-size distribution without resorting to exter- 22-4

nal calibration standards. Tests at this labora- clusions are drawn concerning the relative roles tory have shown that the nonidealities of the sys- of atomization and evaporation on low tempera- tem are negligible except for the detector ture ignition in gas turbine engines and the po- responsivities that must be determined for each tential use of fuel heating as an effective starting instrument in order to get accurate results. The aid. first procedure(") was used specifically to assure proper calibration of the particle sizing in- Droplet-Size Results - Ignition in gas turbine en- strument. gines is strongly dependent on fuel atomization because it is the droplet size that determines the The second procedure('l) was necessary to cor- rate of fuel vaporization. The Sauter mean diame- rect for multiple light scattering off droplets in ter (SMD) of the spray at the spark-gap depends very dense fuel sprays obtained at the higher on the atomizer, fuel properties, and the flow fuel flow rates. This procedure was used in only conditions within the combustor. The SMD's a few instances to correct the data recorded were measured in fuel sprays produced by the when dense sprays were encountered. standard T63 dud-orifice pressure-swirl atomizer and five Delavan simplex atomizers at several All tests were performed at atmospheric condi- flow rates using five test fluids (see TABLE 3). tions in a test chamber of square cross--section The measurements were made along the center- 30 cm on a side and 76 cm long, with air line of the spray about 38 mm from the nozzle pulled through the chamber at a velocity of tip. The SMD obtained by this approach was an about 2.1 m/s by an explosion-proof exhaust fan. average through the spray centerline; no account A set of twisted metal screens in the eyhaust was taken of the radial drop-size distribution in duct removed the fuel mist from the air before the spray. Figure 2 shows the effect of the fuel exhausting to the atmosphere

All spray data were reduced assuming a Rosin- Rammlcr drop-size distribution, which is specified by two parameters X and N, defined by,

R = exp(-(D/X)N) (1) where R is the cumulative volume (or mass) fraction of the spray contained in drops whose diameters are larger than D; X is a size parameter and N indicates the width of the dislribution. Large values of N imply narrow distributions NOS. INDICATE NOZZLE and vice versa. Fuel sprays were characterized CAPACITY IGALlWRl AT 115 PSlD by an "average" size based on the volume-area mean diameter, more commonly called the Sau- ter mean diameter (SMD), which may be com- LA puted from the Rosin-Rammler parameters.(22) 1QQ.l 1 5 10 FUEL MASS FLOW, g/r RESULTS AND DISCUSSION Figure 2. Effect of fuel flow rate and atomizer capacity on SMD l'or JP-5 fuel The purpose of this study was to gain a more flow rate on the SMD for all the atomizers ex- basic understanding of the ignition process in amined. Clearly, the results show that the capac- gas turbine combustors. The results of this work ity of the atomizer has a significant effect on are presented as two parts. The first part de- the flow rate required to achieve a desired scribes the droplet-size measurements in the SMD. It is found that for a given fuel flow rate, sprays of the six atomizers used in the ignition the measured SMI> decreases in almost direct measurements and the correlation of SMD's with proportion as the capacity (or flow number) of atomizer flow conditions and fuel properties. In the atomizer is reduced. the second part, the actual ignition measurements are presented; also, the effects of combus- The break in the curves at low fuel flow rates tor flow conditions, fuel properties, and SMD's shown in Fig. 2 is probably due to the Weber on the limiting fuel-air ratio are examined. Con- number effect observed by Simmons and Har- 22-5 ding.(23) The Weber number, We, is the ratio of zles producing larger drop sizes at equivalent inertial forces to the surface tension forces. The conditions. The average drop size as repre- SMD is much more dependent on flow rate sented by the Sauter mean diameter (SMD) when We < 1, Le., at lower flow rates. could be correlated with the fuel nozzle capacity (flow number), the fuel flow rate, and the fuel This effect was not important in the present viscosity and surface tension as follows: study because the minimum fuel flow rates for 1.18 0.21200.457wj1.16 ignition always occurred at We < 1. SMD = 4.052 FN F (2) Figure 3 shows the effect of he1 type on the In correlating the SMD data from the T63 atom- 500 I I izer, it was necessary to account for the airflow directed across the nozzle face used to reduce carbon deposition. This airflow was considered because it had a significant effect on the atomization process. The ignition tests in the T63 combustor were performed at three airflows, 91, 182, and 273 g/s. SMD’s were measured with air pressure drops across the T63 nozzle corresponding to combustor airflows of 91 and 273 g/s; the atomization at the intermediate airflow was assumed to be an average of the two extremes.The SMD correlation at the 91 g/s airflow condition for the T63 atomizer was,

SMD = 112.6p0.399 wf -1.21 (3)

and at the 273 g/s airflow condition, 1ogL.1 1 5 10 FUEL MASS FLOW, gls 0.334 -1.09 Figure 3. Effect of fuel flow rate and fuel type on SMD = 99.7p wf (4) SMD produced by the standard T63 atomizer Equations 2 through 4 were used to calculate SMD’s measured in sprays produced by the T63 SMD’s of fuel sprays at the ignition limits and atomizer. The results show that fuel properties to ascertain the effects of atomization on the ig- have a significant effect on the SMDhass flow nition process in the T63 gas turbine combustor. rate correlations. The fuel effect is most pronounced for HMGO at ambient temperature and Low temperature ignition experiments were 274K. Othenvise, the JP-4, JP-5, and NDF fuels conducted at several operating conditions in a tested in the high mass flow rate regime (We z T63 combustor rig; TABLE 4 shows the range 1) gave very similar SMD’s. Basically, the ef- of test conditions used. To simulate a sea-level fects on SMD’s measured in sprays from the start, pressures were kept close to one atmo- Delavan simplex atomizers were similar to those sphere. Burner inlet temperatures were vaned found for the T63 atomizer. However, the fuel from ambient to -34T (-29.2OF). The mass flow effecrs on SMD appeared to be more pro- rate of air was varied to change the reference ve- nounced in sprays produced by the Delavan at- locity in the combustor. Fuel temperature was omizers. This fuel effect was especially apparent vaned in order to examine the effects of viscos- among the JP-4, JP-5,and NDF fuel sprays gen- ity, surface tension, and the SMD of the spray erated at the higher mass flow rates that were used in the actual ignition studies. Considering the fuel properties in TABLE 3, it is apparent Conditions for Ignition that the relatively high SMD’s determined for HMGO are most probably attributed to increased viscosity.

Atomization Correlations - The five Delavan simplex atomizers with different flow capacities performed similarly, with the larger capacity noz- 22-6

on the limiting fuel-air ratio for ignition. Fuel fuel evaporates more rapidly. As a result, the temperature was varied both as a result of the minimum fuel-air ratio for ignition is expected variation of inlet temperature, and independently to decrease as the SMD of the fuel spray is low- of inlet temperature by using an ice bath. Fuel ered. temperature was monitored by a thermocouple inserted into the fuel line and located at the en- The fuel effects on the minimum fuel-air ratio trance to the atomizer. for ignition are very apparent in Fig. 4 Fuels such as NDF with high viscosity and low volatil- The ignition limits of the test fuels were mea- ity require a higher minimum fuel-air ratio for sured in terms of the overall fuel-air ratio re- ignition than the more volatile and less viscous quired to achieve ignition at a given set of oper- fuels such as P-4. Fuels with similar viscosi- ating conditions. These tests were performed by ties such as JP-8, JP-5, and NDF/25-percent gas- first establishing the desired air flow conditions oline have similar minimum fuel-air ratios for ig- in the combustor. The igniter was turned on, and nition. It is important to note that Fuel 4, the fuel flow was started at a flow rate below NDF/25-percent gasoline, was blended with the the ignition limit. The fuel flow was gradually same viscosity as E'-5, but with a front-end vola- increased, using a motorized valve until ignition tility (Tioq,) similar to JP-4. Fuel 5, NDF/70 per- occurred. At the point of ignition, the motor turn- cent gasoline, was blended with the same viscos- ing the fuel valve was stopped, and the mini- ity as JP-4, but similar front-end volatility to mum fuel flow rate for ignition was recorded. gasoline. By inspection of Fig. 4 it is apparent that Fuel 4, NDF/25-percent gasoline, has a mini- Most of the ignition measurements were made mum fuel-air ratio for ignition similar to JP-5. on Fuels 1 through 6 in TABLE 1. Limited data Also, Fuel 5, NDF/70-percent gasoline, has a were obtained on methanol and gasoline. Figure minimum fuel-air ratio for ignition similar to 4 shows the effect of reference velocity on the JP-4. The gasoline data are not presented, but its minimum fuel-air ratio, in fact, falls significantly .03 1 below that of JP-4. These results show that

,025 ~ fuels of equal viscosity require similar mini- 3 mum fuel-air ratws for ignition, and that viscos- - q .02 ity appears to be more significant than volatil- z ity in the ignition process. I ,015 - z3 z .01 - Figures 5 and 6 show correlations of the mini 1 mum fuel-air ratio for ignition with viscosity

1 U 163 ATOMIZER I , .0°5 t ""L dl~l.l.,.l.,,,.,., 0 2 4 6 8 10 12 14 16 18 REFERENCE VELOCITY (m/s) Figure 4. Effect of reference velocity and fuel type on minimum fuel-air ratio for ignition with the standard T63 atomizer

minimum fuel-air ratios for ignition of Fuels 1 ,006 1 w' I through 6 at a fuel and burner inlet temperatures of 300K using the standard T63 dual-orifice at- .002\ , , , , , , , , , , , 1 omizer. Similar results were obtained using the 0 1 2 3 4 5 Delavan simplex atomizers. It was found that KINEMATIC VISCOSITY, cSt the minimum fuel-air ratio for ignition decreased Figure 5. Correlation of minimum fuel-air ratio as the reference velocity was raised. This de- for ignition with viscosity for the standard T63 crease is explained in part by the fact that the atomizer and the K5 liter/hr Delavan simplex mass flow rate of fuel delivered to the atomizer atomizer must increase to maintain the fuel-air ratio when and 10 percent boil-off temperature, respectively. the mass flow rate of air through the combustor These data were obtained with the standard T63 is raised. When the fuel flow rate is increased, dual orifice atomizer and the 15 liter/hr Delavan the SMD of the fuel spray decreases and the atomizer.The results clearly show that the mini- 22-7

-0- T63 ATOMIZER + OELAVAN ATOMIZER If indeed a critical SMD is the important criteria ,016 ' for ignition, then it should be relatively indepen- O/ I dent of the flow number of the atomizer. Figure 8 shows the deuendence of the SMD at the mini-

,006 1 a/

,0021 , , , , , , , , , , , 1 250 300 350 400 450 SO0 10 PERCENT BOIL-OFF TEMPERATURE Figure 6. Correlation of minimum fuel-air ratio for ignition with volatility (10 percent pt) for the standard T63 atomizer and the 15 liter/hr Delavan 40 simplex atomizer 3 4 5 6 7 8 '3 FLOW NUMBER (kg/s-Pa'/') x IO' mum fuel-air ratio correlates more favorably Figure 8. Effect of atomizer flow number on the with viscosip than front-end volutility. SMD at minimum fuel-air ratio for ignition of JP-5;data obtained at reference velocities of 5, 11, The effects of atomization on the minimum fuel- and 16 m/s air ratio for ignition have been apparent in the results discussed abcve. Figure 7 shows the ef- mum fuel-air ratio for ignition versus the atom- fect of atomizer flow number on the minimum izer flow number for JP-5 fuel at three reference fuel-air ratio for ignition. The data were mea- velocities. The SMD's were calculated from sured using JP-5, a bumer inlet temperature of Eqns. 2 through 4. At the 5 m/s condition in 300K, and reference velocities of 5, 11, and 16 Fig. 8, the SMD's ranged from 99 to 122, a fac- m/s. The six points on each line represent the tor of 1.2 while in Fig. 7, the corresponding fuel- six atomizers examined. If ignition depended air ratio at ignition varied by a factor of 2.6 only on achieving a critical fuel-air ratio, there over the range of flow numbers. At 11 m/s and would be little or no dependence on atomizer 16 m/s, the respective SMD's ranged from 56 to flow number.The fact that the minimum fuel-air 88, a factor of 1.6, and 43 to 69, a factor of 1.6 ratio for ignition increases with atomizer flow whereas the corresponding fuel-air ratios at igni- number means that the average drop size of the tion vaned by 3.2 and 2.7. spray must be less than some critical value if ignition is to be successful. For the higher capac- In previous combustor studies, the effect of the ity atomizers, this critical SMD is reached at a bumer inlet air temperature on ignition has been higher fuel flow rate than for the lower capacity difficult to discern because of the effect of fuel atomizers. temperature. In the present study, the effect of fuel temperature was accounted for in the viscos- 025 U VT = 5 m/* ity and surface tension terms of the SMD corre- +v, = 11 m/s -eV, = 16 m/s lations. Figures 9 and 10 show the effect of burner inlet air temperature on the required SMD at the minimum fuel-air ratio for ignition. These data were obtained with test fuels listed in TABLE 1 using the respective T63 dual orifice and 15 literhr Delavan simplex atomizers with a reference velocity through the combustor of 11 m/s. The vend lines through the data indicate that smaller SMD's are necessary for ignition when air temperatures are reduced. The SMD's required for the more volatile fuels such as JP-4 appear to vary more strongly with air temperature than the less volatile fuels like JP-8 Figure 7. Effect of atomizer flow number on the and P-5. In fact, the SMD's required for the minimum fuel-air ratio for ignition, fuel; JP-5 less volatile fuels are relatively independent of data obtained at reference velocities of 5, 11, and 16 mls 22-8 . air temperature, and also because smaller drops will be more easily camed to the igniter by the A JP5 airflow and will evaporate more readily. to To reduce the SMD of the spray, it is necessary only to heat the fuel to reduce the viscosity. This can be seen in Eqns. 2, 3 and 4. Figure 11 shows the effects of fuel tem erature on viscosity for several aviation f~e1s.k~)In Fig. 10, the ignition characteristics at an air temperature of 273K (OOC) for JP-8 can be improved to that 2 l5 LL:.-LL,d-Aat 305K (32°C) by reducing the SMD by the 220 230 240 250 260 270 280 290 300 310 ratio of the minimum SMD requirement AIR TEMPERATURE (K) or about 8 percent. Using Eqn. 3, this would mean Figure 9. Effect of inlet air temperature and fuel a viscosity reduction of about 33 percent. From type on SMD required for ignition using standard Fig. 11, reducing the viscosity from 2.5 cSt to T63 atomizer and reference velocity of 11 m/s 1.7 cSt would mean heating the fuel from 0°C 100 to 18T. This concept would be expected to be m JP8 JP4 applicable to full scale engines starting in cold JP5 n air, but in general the amount of heating would 752 NDF. 252 Can 30% NDF. 70% Gor be different for each engine design. For new air- NDF v craft, a fuel tank heater would greatly improve 4. > 70 the starting characteristics of cold weather aircraft. To avoid the cost of retrofitting existing 60 aircraft, perhaps loading the aircraft with z "warm" fuel, e.g., room temperature, 25"C, just 50 prior to starting would suffice; for quick re- 4. 1 0 40 sponse time though, a small heater in the fuel tank that the engine draws from would be better. 240 250 260 270 280 290 300 310 AIR TEMPERATURE (K) It should also be noted that these results refer Figure 10. Effect of inlet air temperature and fuel only to combustor ignition; there may be other type on SMD required for ignition using 15 low temperature effects on engine and gear box literhr Delavan simplex atomizer and reference lubricants that further inhibit engine starting. velocity of 11 m/s the bumer inlet temperature. It is also important to note that the NDF/25-percent gasoline fuel blend which had a viscosity close to JP-5, but a volatility similar to JP-4, seemed to require SMD's for ignition nearer to that required by JP- 5, JP-8 and NDF. These observations tend to favor fuel viscosity over volatility and !suggest that fuel vaporization depends, for the most part, on droplet size.

It should be noted that during engine start up, there is very little compression of the air so that the bumer inlet air temperature is essentially the -50 -30 -10 10 30 50 io 93 110 uo 1% 110 ambient temperature. Thus the significance of Temperature. "C these results is that at lower temperatures, Figure 11. Typical dependencies of viscosity on changes in the air temperature will not have a temperature for alircraft fuels significant effect on the ignition characteristics of low-volatility fuels such as JP-8 and JP-5. It CONCLUSIONS is more important to reduce the SMD of the fuel spray. This is because the evaporation of Extensive experiments were carried out in a T63 the fuel is caused by the spark energy, not the gas turbine combustor to determine which fuel 22-9 property, viscosity or volatility, is the most criti- ESL-TR-80-46 (November 1980). cal in low temperature ignition in gas turbine engines. 7. Reider, S.B., Vogel, R.E. and Weaver, W.E., "Effect of Fuel Composition on Navy Aircraft Ignition depended more strongly on achieving a Engine Hot Section Components," Report No. critical average drop size (SMD) than on reach- DDAEDRlll35 (September 1982). ing the lean-limit fuel-air ratio. Fuel viscosity, which determines atomization characteristics, 8. Beal, G.W., "Effect of Fuel Composition on was found to be more important than volatility Navy Aircraft Engine Hot Section Components," in the ignition process. The importance of viscos- Report No. PWA/GPD/FR-16456 (August 1982). ity was particularly apparent in the less volatile 9. Rutter, S.D., "Effect Fuel Composition on fuels such as JP-5 or JF-8. Volatility effects of were most apparent in JP-4 and fuels containing T53L13B Hot Section Components," Report No. gasoline, Low temperature ignition performance D12-6-032-83 (March 1983). appeared to depend more so on the effect of 10. Rutter, S.D., "Effects of Fuel Composition fuel temperature, i.e., the fuel viscosity, rather on Navy Aircraft Engine Hot Section Compo- This obser- than the bumer inlet air temperature. nents, Lot 11-Component Test," G.E. draft report vation has the important practical implication prepared under Contract No. NOO140-79-C-0483. that heating the fuel or using a lower viscosity fuel would be a much more efficient way to im- 11. Ball, I., Graham, M., Robinson, K., Davis, prove low-temperature ignition than beating the N., "l76 Altemative Fuels Final Report," Garrett inlet air, a guideline particularly suited to com- Turbine Engine Company, Report No. 21-4744 bustors employing pressure-swirl atomizers for (August 1983). ignition. 12. Ballal, D.R. and Lefebvre, A.H., "A Gen- REFERENCES eral Model of Spark Ignition for Gaseous and Liquid Fuel-Air Mixtures," 18th Symposium (In- 1. Moses, C.A., "Studies of Fuel Volatility Ef- temational) on Combustion, The Combustion In- fects on Turbine Combustor Performance," Pre- stitute, Pittsburgh, pp. 1737-46 (1981). sented at the 1975 Joint Centralmestem States Section Spring Meeting of the Combustion Insti- 13. Peters, J.E. and Mellor, A.M., "A Spark Ig- tute, San Antonio, Texas. nition Model for Liquid Fuel Sprays Applied to Gas Turbine Engines," Journal of Energy, 6 (4),- 2. Gleason, C.C., et al., "Evaluation of Fuel p. 272 (1982). Character Effects on FlOl Engine Combustion System," Final Technical Report AFAPL-TR-79- 14. Peters, J.E. and Mellor, A.H., "Cbaracteris- 1018, CEEDO-TR-79-07 (June 1979). tic Time Ignition Model Extended to an Annular Gas Turbine Combustor," Journal of Energy, 6 3. Vogel, R.E. and D.L. Troth, "Fuel Character (6),- p. 439 (1982). Effects on Current High Pressure Ratio, Can- type Turbine Combustion Systems," Final Techni- 15. Peters, J.E. and Mellor, A.H., "An Ignition cal Report AFAPL-TR-79-2072, ESL-TR-79-29 Model for Quiescent Fuel Sprays," Combustion (April 1980). and Flame, (38).- p. 65 (1980).

4. Oller, T.L., et al., "Fuel Mainbumerlrurbme 16. Derr, W.S. and Mellor, A.H. "Characteristic Effects," Final Technical Report AFWAL-TR-81- Times for Lean Blowoff in Turbine Combus- 2100 (May 1982). tors," Presented at the Fall Meeting of the West- em States SectionKhe Combustion Institute 5. Gleason, C.C., et al., "Evaluation of Fuel (1986). Character Effects on J79 Engine Combustion System," Final Technical Report AFAPL-TR-79- 17. Naegeli, D.W., Moses, C.A., and Mellor, 2015, CEEDO-TR-79-06 (June 1979). A.H., "Preliminary Correlation of Fuel Effects on Ignitability for Gas Turbine Engines," ASME 6. Gleason, C.C., et al, "Evaluation of Fuel 83-JPGC-GT-8. Character Effects on J79 Smokeless Combustor," Final Technical Report AFWAL-TR-80-2092, 18. Moses, C.A., et al., "An Alternate Test Pro- 22-10

cedure to Qualify Fuels for Navy Aircraft, Phase Messrs. R.C. Haufler and M.G. Ryan conducted I1 Final Report, Appendix D. Ignition and Alti- the experimerital measurements. Ms. S.J. Hoover tude Relight," SwRI-5932-3, NAPC-P.E-145C and h4s. L.A. Pierce performed the manuscript (August 1984). preparation with the editorial assistance of Mr. J.W. Pryor and Ms. C.G. Wallace. Mr. S.J. Lestz 19. Naegeli, D.W., Dodge, L.G. and Moses, provided assistance in contract management. C.A., "Effects of Fuel Properties and Atomiza- tion on Ignition in a T63 Gas Turbine Combus- tor," Interim Report BFLRF No. 235, December 1987. Discussion

20. Dodge, L.G., "Calibration of the Malvem 1. R. Pollak, Pra1.t and Whitney Particle &er,'' Applied Optics, 23, p. 2415 Have you examined tF,effect of small quantities of lower - viscosity/higher volatility fuels, say JP5, mixed into more (1984). viscousAow volalility fuels, say JP4, to obtain better ignition characteristics at cold temperatures? 21. Felton, P.G., Namidi, A.A., and &gal, A.K., Author: "Multiple Scattering Effects on Particle Sizing We have conducted experiments such as this several times by Laser Diffraction," Report No. 413HIC (Au- over the years including the results reported here where two gust 1984). of the test fuels were blends of gasoline in diesel fuel to tailor the viscosity and volatility properties. In earliest studies (at 22. Allen, T., Particle Size Measurement, 3rd ambient air temperatures) we used blends of pentane with Ed. Chapman and Hall, New York, p. 139 both Jet-A and diesel fuel. In all cases, the addition of the lighter materials improved the ignition characteristics of the (1981). combustor. We have not, however, conducted a thorough study of the type you mentioned in the sense of developing a 23. Simmons, H.C. and Harding, C.F., "Some fuel sensitivity model for low-temperature ignition. The Effects of Using Water as a Test Fluid in Fuel significance of the more recent studies, of which the work Nozzle Spray Analysis,'' ASME 80-G'I-90 reported here was a part, is that we can now do these (1980). combustor experiments at low temperatures and make detailed measurements of atomization, air-fuel ratio and 24. Handbook of Aviation Fuel Properties, CRC velocities at the spark gap, and spark energy to isolate the fuel variables. Document No. 50, p. 34, Coordinating Research Council, Atlanta, Georgia (1983). 2. l? Kotsiopoylns, Hellenic Air Force Academy I would like to know if you have any comments to make about the effect of the change of fuel composition, for instance from JP4 to JPX, on the hot engine components as a ACKNOWLEDGEMENTS result of the change of the viscosity - provided that the mass flow rate is kept constant by adjusting the fuel control system. This work was performed by the Belvoir Fuels Have you observcd any phenomena like carbon particle ad Lubricants Research Facility (BFLRlF) located formation at the hot parts etc? at Southwest Research Institute (SwRI), San An- Author: tonio, Texas under Contract Nos. DAAK70-85-C- Fuel effects on hot-section durability are due to increascs in 0007 and DAAK70-87-0043 with the U.S. Army the flame radiation or changes in temperature distributions, Belvoir Research, Development and Engineering i.e. hot streaks. The major differences between JP4 and JP8 which could cause there problems are lower hydrogen Center (Belvoir RDE Center). Funding was pro- content and higher viscosity. Lower hydrogene content vided by the Naval Air Propulsion Center increases the sool. formation and flame radiation in the (NAPC) through a Military Interdepartmental primary zone leading to higher liner temperatures. This soot Purchase Requisition, and also in part with can also deposit on the liner surfaces or atomizer face and funds from the SwRI Fuels and Lubricants Re- disturb the flow pattern leading to hot spots. Higher viscosity means larger drops in the fuels spray and could search Division. mean inadequate air-fuel mixing or even impingement on the liner. Such problems are generally eliminated in the design of the combustor and most combustors are designed Mr. EA. Katpovich was the principal NAPC to operate satisfactorily on both fuels; however, it is possible staff member providing Program direction. Mr. that in an older, perhaps marginally designed combustor, a F.W. Schaekel of Belvoir RDE Center (STRBE- change from P4to JPX could result in inadequate air-fuel VF) sewed as the Contracting Officer's represen- mixing and/or create an increase in deposit formation in isolated areas and lead to durability problems. In the tative, and Mr. M.E. LePera, Chief of Fuels and combustor studies conducted by the US Air Force, there Lubricants Research Division (STRBE-VF), were no hot scctioii durability problems identified due to a served as the project technical monitor. change from JP4 to JPX. 23-1

GIVRAGE DES CIRCUITS DE CARBWDES TURBOREACTEURS par Francis GARNIW IngGnieur expert- en sysths d'huile SNECMA Direction technique Centre d'essais de Villaroche 77550 ml1SSY-CRAWsYF.L (FRANCE)

RB- RB- sur les avions civils, il n'est pas de pratique courante d'ajouter d'additifs anti-givre dans le carburant. I1 s'ensuit aux tres basses temperatures des risques de depots de givre dans certains composants des circuits moteur a partir de l'eau ineluctablement contenue dans le carburant. Ce phenomene de givrage peut affecter tous les moteurs a la fois au moment du decollage, entrainer des perturbations importantes touchant leur pilotabilite, pouvant mSme aller jusqu'a leur extinction. pour Bviter le givrage, la solution la plus simple est de rechauffer suffisamment le carburant avant son entree dans les composants sensibles pour qu'il soit toujours a temperature positive; la difficult6 etant de connaitre avec suffisamment de precision la quantite de calories disponibles sur le circuit d'huile dans les conditions extremes froides des le debut des etudes de developpement d'un nouveau moteur. NOUS avons donc et6 amen& 3. developper, mettre au point et valider un nodele mathematique capable de determiner les equilibres thermiques huile/carburant dans tout le domaine d'exploitation du moteur. A partir des analyses effectuees grace 5. ce modele, il a et6 possible de definir une solution simple qui, sans rien changer B l'equilibre du circuit d'huile, a permis de rechauffer suffisamment les circuits d'asservissement. Au cours de cet expos&, nous decrivons ce modele, les principales analyses effectuges, l'installation d'essai et les rdsultats obtenus en conditions extremes de givrage qui montrent que la solution retenue avec rechauffeur huile-carburant des servo&canismes permet un fonctionnement correct jusqu'i - 45 OC au lieu de - 30 OC pour des conditions identiques sur le circuit d'huile.

Le carburant consomme par les moteurs d'un avion est stock6 dans un systeme de reservoirs situ& dans les ailes et la partie centrale du fuselage de l'avion. Apres un systeme de transferts entre reservoirs, il est pomp6 dans une "nourrice" et achemine vers les moteurs. Pour des raisons de skurite, pollution de l'environnement en particulier, ce carburant est tres souvent stock6 sur les aCroports dans des cuves non enterrees. En consequence, par temps froid, on peut considerer que le carburant pomp6 vers le moteur est i la temperature ambiante qui peut atteindre, dans certains pays, des valeurs excessivement basses. Malgre toutes les precautions prises par les compagnies petrolieres qui assurent l'avitaillement des avions, de l'eau peut subsister dans le carburant sous forme dissoute et libre. Aux temperatures negatives, l'eau libre se trouve en suspension sous forme de micro-paillettes de glace qui seront pompees avec le carburant. Ces cristaux de glace peuvent obturer les circuits carburant: les filtres, mais aussi et surtout les systemes de regulation qui dosent le carburant a injecter dans la chambre de combustion ou qui commandent la position des vannes de decharge et des stators i calage variable. I1 peut s'ensuivre de graves perturbations pouvant affecter la pilotabilite du moteur et conduire, B la limite, B l'extinction des moteurs. Des moteurs, en effet, car le phenomene de givrage peut les affecter tous B la fois puisque la cause est commune. Si on montre par les calculs que la condition la plus critique est celle du regime de decollage, on comprend alors la gravitir de situation que peut engendrer le phenomene de givrage des circuits carburant. Si en exploitation militaire, le probleme est resolu par l'adjonction au carburant de produits anti-givre, cette pratique n'existe generalement pas en exploitation commerciale. I1 revient donc au motoriste la charge de definir et d'installer sur le moteur un systeme permettant d'eviter le phenomene de givrage ou, pour le moins, d'eviter ses consequences. 23-2

NOUS vous proposons au cours de cet expose de presenter : En lirepartie : - les diverses solutions qui avaient 6th imaginees lors de la conception du CFM56 et plus particulierement celle qui a et6 retenue - les analyses et les calculs qui ont permis d'aboutir h ce choix En 2"* partie : - les moyens d'essais mis en oeuvre - les essais de validation effectues au banc de simulation En 3'ne partie : - une actualisation

1 - LE CIRCUIT CARBURANT DE BASE I1 comprend, pour ce qui nous interesse : - les reservoirs de carburant (partie avion) - la pompe basse pression du moteur lpompe B.P.) - un filtre qui protege la pompe haute pression Zi engrenage (pompe H.P.) - la regulation du moteur - la chambre de combustion avec ses injecteurs - un Bchangeur de chaleur destine a refroidir l'huile de lubrification du moteur

2 - ELEplEHTS DU CIRCUIT POWANT ETRFi AFFECI'ES PAR L!3 GIVRAGE En conditions givrantes, ler; cristaux de glace v8hiculi.s par le carburant peuvent affecter le fonctionnement de certains organes du circuit carburant.

Bien que l'huile du moteur soit chaude, les tubes et les plaques frontales de l'echangeur peuvent Gtre ?I temperature negative, favorisant l'accrochage et l'accumulation des cristaux de glace. I1 va s?ensuivreune diminution de la section libre pour le passage du carburant et consecutivement une augmentation de la perte de charge de l'echangeur; la limite &ant l'obstruction totale. A partir d'une certaine valeur, le clapet de derivation va s'ouvrir, pouvant a la limite deriver tout le carburant. La reduction du debit carburant dans la matrice conduit B une augmentation de la temperature de l'huile, des plaques frontales et des tubes. Le degivrage de la matrice va s'operer progressivement jusqu'a refermeture du claoet. Ce ohenomene d'ouverture fermeture Deut s'ooerer olusieurs fois mais sans jimais pekturber reellement le fonctionnekent coriect d; moteur, si les systemes en aval peuvent accepter du givre, bien siir. 2.2 Les premieres manifestations du phenomene de givrage se produisent sur le filtre a carburant cause des temperatures de paroi p:Lus basses d'une part et des sections de passage unitaires plus faibles d'autre part. Dans les cas de colmatage extreme par le givre, le moteur peut Stre amen6 a fonctionner clapet ouvert, c'est a dire sans filtration, pendant une duree importante du v01; le moteur fonctionnant neanmoins correctement, seule la duree de vie de la pompe H.P. risque d'6tre affectee. 2.3 REGULATEUR Les servomecanismes du regulateur qui dosent le carburant et fixent la position des systemes a geometric variable (VSV et VBV) constituent la partie du moteur qui peut entrainer des consequences graves en cas de givrage. En effet, le colmatage de ces organes peut conduire a des perturbations importantes sur le fonctionnement du moteur: delai d'obtention important voire non obtention du taux de poussee affiche par le pilote a la manette des gaz; ces perturbations pouvant aller jusqu'a l'extinction du moteur, situation particulierement grave puisque, come nous allons le voir, la condition la plus critique se situe a pleine puissance du moteur, c'est a dire au regime de decollage et de montee donc B tres basse altitude. Les analyses ont donc porte plus particulierement sur oette partie du circuit carburant. 3 - LES REGLEMENTS Les reglements demandent que le moteur soit capable de fonctionner correctement dans tout le domaine d'utilisation pour lequel il est concu avec du carburant initialement sature d'eau i 27OC (80OF) et ayant au moins 0,75 cml d'eau libre ajoutee par US gallon de carburant (2OOPPM) et refroidi jusqu'i la condition la plus critique du point de vue givrage pouvant Stre rencontree en exploitation. 23-3

4 - ANALYSE En l'absence d'additifs anti-givre ajoutes au carburant ou de systemes capables d'enlever tous les cristaux de glace pouvant &re contenus dans le carburant, la premiere solution qui vient B l'esprit est de rechauffer suffisamment les circuits pour que, dans la condition de vol la plus severe, les pieces et le carburant lui-meme soient 6 temperature positive. En effet, il ne suffit pas que le fluide soit B temperature positive pour &iter tout probleme lie au givrage : les cristaux de glace oui n'ont vas eu le temvs de fondre Deuvent adherer aux vieces B temverature neaative-~ it obturer-les circuits- la temp8rat;re des pieces depenzant des Cchinges thermiques avec le carburant d'une part et avec l'air de l'environnement (nacelle) d'autre part. 11 est clair que la condition la plus critique se rencontrera lors de fonctionnements aux temperatures les plus basses specifiees. Les etudes ont porte plus particuligrement sur les cas de vol suivants: - decollage, maxi continu et maxi montee en IsA -124-F (Tamb=-54OC) - maxi montee et maxi continu depuis le sol jusqu'B une altitude de 35000 ft en atmosphere extrsme froide. Pour mener 2 bien ces etudes, un modele de calcul comportant plusieurs modules a &ti: Cree et valid6 sur des resultats d'essais provenant des moteurs en developpement. Le modele complet comprend les modules suivants : le "module technoloqie" dans lequel sont introduites toutes les donnCes geometriques permettant de decrire le moteur : . g6ometrie des roulements, des pignons . surfaces d'echange des enceintes-huile . definition geometrique des joints a labyrinthe . description des rotors susceptibles d'entrainer de l'air ou de l'huile en rotation . diametre et longueur des tuyauteries . circulation des fluides au sein du moteur . liaisons mecaniques entre le moteur proprement dit et ses equipements - le "module thermodvnamique" permettant d'introduire les donnhes du cycle pour les points que l'on desire calculer : . vitesse de rotation des corps . temperatures, pressions et debits 2 des points caracteristiques des flux primaires et secondaires - le "module circuit d'huile", le plus important, contient les methodes de calcul. C'est lui qui calcule la puissance calorifique recueillie par l'huile qu'il delivre sous la forme d'une courbe - le "module circuit carburant", il contient la technologie et les methodes de calcul propres au circuit carburant. C'est lui qui calcule l'hquilibre et delivre les temperatures - le "module Bchanaeur" fait l'interface entre les modules "circuit huile"

~~ ~ ~~~ ~~~ et "circuit carburant" et contient les chamvs d'efficacith des echanoeurs~~ ~~~ provenant d'essais ou d'anaiyses; c'est~Bdire l'efficacite thermique en fonction des debits et des temperatures 4.1 DETERMINATION DE LA PUISSANCE CT3LORIFIQLlE DISPONIBLE DANS L'HUILE PAR LE MODELE "CIRCUIT D'HUILE" Pour une condition de vol et un fonctionnement moteur donne, la puissance calorifique Wh rejetge par le moteur sur le circuit d'huile ne depend plus que de la Temperature d'Huile 6 1'Entri.e du Moteur. Le rBle du module "circuit d'huile" est donc de determiner les courbes Wh = f (THEM) pour tous les cas de vo1 8tudie.s. La puissance totale recueillie par l'huile est la some de toutes les puissances BlCmentaires rejetees par les composants du moteur en contact avec l'huile. Ces puissances sont calculees analytiquement et non pas par une identification avec des resultats obtenus sur le moteur lui-meme.

environnement (convection + rayonnement + conduction au travers des supports de palier) - de l'air de pressurisation introduit dans les enceintes au travers des joints B labyrinthes - des centrifugations d'huile cr68es par les rotors contenus dans les enceintes - des puissances mecaniques necessaires 6 l'entrainement en rotation du groupe de lubrification (circuit ferme) - du frottement de l'air sur les parties tournantes situires dans les enceintes : friction sur le voile des engrenages, friction entre rotor 23-4

et stator dans les labyrinthes, ... - des (.changes de chaleur, principalement par convection, entre les Bquipements et l'air nacelle : reservoir d'huile, boite d'cngrenages, tuyauterics, pompes B huile, ... - des &changes de chaleur par convection et rayonnement entre lcs tuyauteries dans la traversee des parties chaudes du moteur : traversees des bras du carter d'echappemcnt par exemplc Ce module, le plus complique, comporte environ 1400 lignes d'ecriture en FORTRAN 4.2 DEITCRMINATION DU NIVFdU D'EWILIBRB THERMIQUE ENTRE L'WILE ET LE CARBURAWT PAR LE ~DULE"CIRCUIT camma"* La puissance calorifique recueillie par l'huilc est transmise au carburant par les Bchangeurs de chaleur. Selon leur efficacite thcrmique, cette transmission se fera B des nivcaux plus ou moins eleves sur chacun des circuits : * efficacite elevee : - niveau bas sur l'huile - niveau &lev& sur le carburant * efficacite faible : - niveau eleve sur l'huile - niveau bas sur le carburant Comme le module precedent, ce module comporte un sous-programme permettant de calculer les puissances calorifiques generees au sein du circuit carburant lui-mgme par les pompes basse pression et haute pression, la recirculation de carburant dans la boucle par degradation de 1'Bnergie de pression, les diverses pertes de charge jusqu'aux injecteurs. Par ailleurs, il prend en compte les &changes thermiques entre les composants du circuit et ].'air nacelle. Enfin, come nous l'avons d&jB vu, il determine 1'6quilibre par application du premier principe de la thermodynamique.

4.3 VALIDATION DU mDELE

~ ~ ~ ~~~~ ~~~~~ ~ ~ ~~~~~ ~~~~ ~~~~~~ ~~~ ~~~~~~~~~~ Avant d'entreurendre~~~~ des calculs .~~~~Dour determiner les temD6rature.s- susceptibles d'stre atteintes en extreme froid, il convi.ent de valider lcs resultats produits par le modsle compose des deux principaux modules dCcrits plus haut, dans un domaine de fonctionnemcnt explork en essai le plus large possible, de faCon B faire varier tous Xes paramstres influents sur une plage la plus large possible. Car, il est bien evident que si les conditions de basses temperatures peuvcnt Stre reproduites en essai partiel sur le circuit carburant, elles ne pcuvent l'stre sur un moteur complet en fonctionnement, particulierement au regime du plcin gaz. 4.3.1 MODULE CIRCUIT D'WILE Pour ce faire, 101s du developpement du CFM56-2, les circuits d'huile et de carburant du moteur 004 ont Bt.6 fortcment instrument& et un programme d'essai specifique a kt6 realis&. Un amenagement des circui.ts au niveau des Gchangeurs de chaleur a BtB necessaire de faCon B explorer une plage de temperature d'huile plus large que celle rencontree normalement au banc sol : . adjonction sur le circuit d'huile d'Bchangeurs supplementaircs huile/cau et huile/azote liquide de faFon B abaisser la tempCrature d'huile B l'cntree du moteur aussi pres que possible des temperatures precalculees par le modele lors d'un fonctionnement stabilise au sol sur avion en extreme froid. . adjonction d'un systeme de vannage, sur le circuit de manisre .ideriver de l'bchangeur tout ou partie de l'huile de manisre B obtenir des temperatures d'huilc Blevees. Ces essais ont permis d'obtenir un nombrc important de resultats couvrant toutes les vitesses de rotation possibles sur une plage de temperatures d'huile B 1'entri.e du moteur variant de 20 B 120 OC. L'effet des prelsvements d'air et mgcanique a Bgalement kt6 wantifie. C'est B partir de ces champs de valeurs que le "module circuit d'huile" a kt6 verifie et valide. I1 est bien evident que de nomhreuses retouches ont dii Gtre effectuees. En particulier, il s'est aver6 que les puissances calorifiques obtenues sur moteur aux faibles temperatures d'huile Ctaient plus faibles que celles obtenues par le calcul; ceci provenant principalement du choix de:; exposants sur le terme de viscosite dans la mBthode utiliske initialement pour le calcul des puissances calorifiques g6ni.rGes par lcs roulements. D'autre part, des amCnageinents ont dii stre apportes en ce qui concerne la determination des puissances calorifiques generbcs par barattage de l'huile dans les enceintes confinecs ou l'effet de la pression dans ces mGmes enccintes : effet 23-5

du frottement de l'air sur les rotors. Une boite d'engrenages fonctionnant avec une pression interne de 1,s bar absolu dissipera plus de puissance qu'h 0,5 bar absolu. L'effet des ventilations externes au moteur a dii Bgalement Gtre affine pour tenir compte de l'environnement : moteur au banc sol sans nacelle ou moteur avionne avec des nacelles de types de ventilation differents. Les correlations calculs/essais obtenues in fine sont les suivantes : . au banc sol sur le moteur 004/4 La comparaison des puissances met en evidence l'excellent accord calculs-essais pour tout le champ regimes-temperatures explore au banc. L'ecart quadratique moyen est de 6 % ce qui se traduirait au sol par un kcart maximum de 2 OC sur le niveau d'equilibre des temperatures . en vol Turbanc volant "Caravelle". Moteur 00612. * sur avion YC15 avec le moteur 003/2. Le but de ces recoupements etait de s'assurer que les effets de moteur, d'altitude et de ventilation nacelle (effet du mach de vol) etaient correctement restitues. 4.3.2 MODULE "ECHANGEURS DE CHALEUR" Dans sa version finale, ont BtC introduits dans ce module les resultats

YNOLDS, d'epaisseur de couche limite . 4.3.3 WDULE "CIRCUIT CARBURANT" Le mGme processus que 4.3.2 a et& applique, c'est h dire des champs de performances provenant des essais partiels realis& sur les pompes H.P. et B.P. et le circuit carburant au complet mais depourvu de calories provenant du moteur et transmises par les echangeurs. 4.3.4 MODEtE COMPLET Chaque module ayant et6 prealablement teste et valid6 separement, il ne pouvait subsister que des problemes d'interface entre les modules; c'est h dire des problemes d'dcriture informatique mais nullement de simulation mathematique des systemes ou des phenomenes physiques. La precision finale, en terme de temp6rature est meilleure que 6 OC. A la suite de quoi le modele a Cte declare valid6 et applicable h la determination des temperatures d'equilibre en particulier en extrhe froid. 5 - SPECIFICATIONS Un moteur d'avion doit Stre capable de fonctionner correctement sur une plage tres etendue de temperature ambiante : depuis - 55 OC jusqu'h + 55 "C environ au sol; le carburant, come nous l'avons vu, Btant Ggalement supposC Stre B cette temperature l'entree du moteur. On conGoit que les systemes de refroidissement d'huile/r&chauffage de carburant ne pourront Stre qu'un compromis pour satisfaire les valeurs des temperatures limites souhaitees sur chacun des circuits. D'oii la necessite de specifier ces valeurs, qui peuvent resulter pour certaines des proprietes physiques des fluides et pour d'autres de considerations fonctionnelles inherentes au moteur lui-m0me : experience acquise, essais partiels, analyses ... I1 est ainsi apparu necessaire apres analyse : . que le debit de carburant qui traverse les asservissements soit rechauffe suffisment de maniere que la temperature du fluide soit toujours superieure h 10°C et non pas 0 "C afin de tenir compte des effets de convection autour du regulateur d'une part et d'echanges thermiques B l'interieur du regulateur entre le carburant principal froid (C 0 "C) et le carburant chaud des asservissements d'autre part (cas du rechauffage du debit d'asservissement) . que les temperatures extrhes pouvant ktre atteintes sur les circuits d'huile restent B l'interieur des limites acquises par l'experience, tant h froid qu'B chaud; ce qui a pour effet, dans les cas qui nous interessent, de limiter le prelevement de puissance sur le circuit d'huile . clue les temDBratures maximales dans .le circuit carburant restent. en atmosDhere chaude, infsrieures h certaines valeurs pour &?iter tout probleme de cavitition des pompes, de "vapor-lock" ou de cokefaction. Par ailleurs, il Btait demand6 que les circuits soient les plus simples possibles. En particulier, il etait conseille d'eviter tout systeme de vanne thermostatique ou command6 par calculateur. En un mot, Bviter toute geometric variable. 23-6

6 - DETERMINATION DES TEMPERATURES D'EQUILIBRE EN CONDITIONS EXTREMES FROIDES Le modele etabli et valid6 come nous l'avons vu, va nous permettre maintenant de simuler les conditions extrsmes :froides en introduisant dans le "module thermodynamique" les conditions aux limites representatives des cas que l'on veut etudier; valeurs determinees en amont par un autre modele.

Dans les conditions de temperatures au sol ISA - 124 OF lTamb= - 54 "C) et nombre de Mach de "01 s6v6re du point de vue ventilation autour du moteur, la puissance calorifique recueillie par le circuit d'huile du moteur depend du niveaux de poussee demand6 par le pilote. C'est cette puissance qui va d'abord stre utilisee pour rechauffer le carburant et les pieces, car elle est gratuit.e. Elle ne depend pas du systeme de rechauffage de carburant que l'on va &tre amen6 B adopter : un ou plusieurs Cchangeurs, huile/carburant, huile/air, air/carburant ou au.tre, ainsi que nous l'avons dejk souligne au paragraphe 4.1. Elle ne sera donc calculee qu'une seule fois par le "module huile"; ce qui a l'avantage d'eviter des iterations entre ce module et le "module carburant" et d'accroitre la vitesse de calcul. 6.1 EQUILIBRE OBTENU SUR LE CIRCUIT DE BASE (SANS RECHIAUFFEUR) Ce circuit est le plus simple. Helas, s'il est satisfaisant dans la plupart des cas, il ne l'est pas B firoid. La temperature de 10 OC minimum i l'entree des servomecanismes est loin d'iitre satisfaite. Dans la condition de decollage extrsme froid lTamb= - 54 "C), la temperature du fluide B l'entree des servomecanismes atteint 0 OC pour une temperature du carburant de - 30 "C B 1'entri.e du moteur. Pour satisfaire la condition des + 10 OC, il ne faudrait pas que la temperature du carburant pomp6 soit inferieure a - 15 OC. Ceci montre tout simplement l'incapacite du moteur B rechauffer suffisamment tout le carbura.nt qu'il consomme. L'energie calorifique provenant du circuit d'huile et du circuit carburant llaminages, rendement des pompes) ne permet d'echauffer toute la masse consommee que de 25 OC i 30 OC I- 15 OC --> + 10 OC et -30 ., --> 0 "C); la difference de 5 "C provenant de la pente negative de la puissance calorifique delivree par le moteur en fonction du niveau d'equilibre THEM. Pour satisfaire la condition des 10 OC, il faudrait apporter au carburant consomme environ 3 fois plus de puissance que le moteur n'en dispose. Les solutions consistant h apporter ce manque de chaleur par des systemes electriques ou par prelevement d'air chaud sur le compresseur Iechangeur air/huile ou air/carburant) ont BtC envisagees, Btudiees, puis abandonnees pour des raisons diverses: complexite, s&curit&, fiabilite, masse, performance du moteur, etc... Des considerations prbcedentes, il ressort que la voie B suivre pour proteger le circuit carburant contre le givrage, en utilisant la seule puissance thermique disponible sur le circuit d'huile, est la suivante : . rechauffer suffisamment le debit carburant qui traverse les asservissements du regulateur (debit 4 fois plus faible que celui consomme) de maniere i ce que la temperature du fluide soit superieure B 10 QC . completer le refroidissement de l'huile jusqu'aux niveaux de temperature desires par un refroidisseur adapt6 situ6 en lieu et place de celui qui existe dans le circuit de base . demontrer par des essais partiels que la partie du regulateur non suffisamment richauffee est capable de fonctionner correctement pendant des temps tres longs en condition givrante C'est cette solution qui a et6 Bgalement utilisee avec succes par notre partenaire GENERAL ELECTRIC sur ses moteurs CF6-50 et CF6-80. 6.2 EQUILIBRE OBTENU SUR LE CIRCUIT awc UN RECHA~URDES SERV~CANIS~S Tout d'abord, de tres nombreux calculs ont et6 neoessaires pour definir le meilleur couple lrefroidisseur d'huile , rechauffeur des servomecanismes) du point de vue de leurs efficacites respectives, et du choix de leur emplacement dans les circuits. Mais aussi pour s'assurer que, dans tous les domaines d'exploitation possibles du moteur, les temperatures d'huile et de carburant ne depassaient pas les limites fixees : effet des regimes de rotation, de l'altitude, des conditions ambiantes, froides mais aussi chaudes, des diverses phases de vol, des prelevements, de l'irtat du moteur et de ses equipements: (influence du debit d'huile delivrir par la pompe d'alimentation, d'une erreur sur l'estimation de la puissance rejetee sur l'huile, du debit d'asservissement, de la nature des fluides utilises : huile type I ou tType 11, carburant JETA, JP4, etc ...). Hormis pour la temperature du carburant ;1 l'entree des servomecanismes, cette optimisation a permis de rendre le niveau general d'equilibre pratiquement insensible B la presence ou lion du rechauffeur des servomecanismes. C'est B dire que la presence du rechauffeur n'abaisse et n'echauffe que de 1 B 2 OC le niveau des temperatures d'huile et de carburant i l'injection dans la chambre de combustion. 23-7

Les evolutions des temperatures calculees en fonction de la temperature du carburant TCa B l'entree du moteur font apparaitre : . un classement des temperatures inverse de celui de 1'6nergie calorifique recueillie par le circuit d'huile. Ceci provient du fait que, dans la condition decollage par exemple, le debit carburant consomme a relativement plus augmente que l'energie reoueillie par l'huile : plus le taux de poussee du moteur est eleve, plus les niveaux d'equilibre thermique des circuits sont bas . en extrOme froid un rehaussement de la temperature du carburant des servomecanismes de 25 OC par rapport B celle B l'entree du regulateur et un niveau d'equilibre qui se situe autour de 10 OC. Rappelons qu'en l'absence de rechauffeur de carburant des servom6canismes, la temperature B l'entree de ceux-oi serait celle B l'entree du regulateur. 7 - CONCLUSION DES ETUDES Les nombreuses analyses effectuees ont montre qu'en condition extrOme froide, le circuit d'huile du moteur n'est pas capable de rechauffer tout seul la totalite du carburant consomme jusqu'i un niveau de temp6rature tel que toute cause de givrage puisse Stre evitee sur la totalit6 des Bquipements mouilles par le carburant. si pour certains, le givrage ne peut entrainer que des dysfonctionnements locaux n'affeotant pas la marche normale du moteur, pour le regulateur et ses asservissements, on peut &re confront6 B des ennuis graves affectant la pilotabilite m@me du moteur, pouvant aller B la limite jusqu'i son extinction par mauvais dosage du carburant ou mauvais contrale de la position des stators variables. La recherche d'une solution simple nous a conduit B rechauffer separ6ment les servomecanismes, partie la plus sensible, jusqu'i un niveau de temperatures Bvitant tout risque de givrage et le circuit principal du regulateur jusqu'au maximum possible autorise par le oircuit d'huile du moteur; la temperature dans le regulateur restant cependant fortement negative en extrSme froid. Le domaine de fonctionnement correct de ce concept ne pouvait qu'Stre explore et valid6 par des essais partiels reproduisant le plus fidelement possible les conditions de givrage susceptibles d'@tre rencontrees en exploitation. C'est l'objet de la deuxieme partie.

abmePiXRTIE - LES ESSAIS EN COlVDITIONS GI-TES

Les essais de simulation ont et6 effectues au Centre d'ESSaiS des Propulseurs (CEPr) de SACLAY (FRANCE), centre disposant de moyens d'essai et d'investigation tres importants mis B la disposition des constructeurs de moteurs aeronautiques mais aussi des avionneurs, en particulier pour ce qui concerne les simulations de vol en altitude. 1 - GENERaLITES SUR LES CONDITIONS DE SIMULATION Des analyses presentees en premiere partie, il ressort que les conditions de decollaae sont les ~luscritiaues et couvrent Dar leur severit6 tcus les autres cas de vol possibles. En consequence, nous nous somes attaches a faire subir au cirouit carburant oomplet les conditions qu'il pourrait etre B m@me de rencontrer sur avion dans des conditions de temperatures extrgmes : Ll6Collage Z= 0, M= 0.6. ISA= - 124 OF. 1.1 TENEm EN EAU NOUS avow deja vu que les reglements demandent que la teneur en eau libre ne soit pas inferieure B 200 PPM au cours des essais; le carburant B 27 "C ayant et6 prealablement sature en eau dissoute. Ce qui, compte tenu des courbes de solubilite de l'eau dans le carburant, conduit B un rapport en masse de 0.0078 % d'eau dissoute dans le carburant, soit environ 100 PPM en volume. Dans la realite sur avion, cette eau dissoute peut se liberer totalement grke B l'abaissement de la pression avec l'altitude. En definitive, le carburant peut contenir, B la limite, jusqu'i 300 PPM d'eau libre : cas d'un avion long courrier decollant d'un pays chaud et humide (zones Bquatoriales) et contraint de redeooller en fin de mission suite 2 un atterrissage avorte. En consequence, l'ensemble des essais a et6 effectue avec une quantite d'eau en volume Bquivalente B 300 PPM au moins d'eau libre. 1.2 PUISSANCE CmORIFIQUE Un parametre tres important H simuler est la loi de puissance calorifique transferee par les Bchangeurs, du oircuit d'huile moteur au circuit carburant: loi qui n'est pas constante mais, come nous l'avons vu en premiere partie, decroissante quand la temperature d'huile B l'alimentation du moteur augmente. Le debit d'huile Btant constant et fixe, le plus simple Btait de restituer par un systeme de regulation la difference de tempCrature (sortie moteur - entree moteur) en fonction de la temperature entree moteur. 23-8

1.3 ENVIRONNEMlWT on s'est efforce de recreer autour des equipements, les conditions aerothermiques les plus defavorables susceptibles d'stre rencontrees sur avion respect de la temperature ambiante et du coefficient d'echange par convection. qui a necessitb d'installer tous les materiels en essai dans une enceinte climatique calorifugCe. 1.4 TEMPESfATURE CARBUPAN'I! De faGon i determiner les limites de fonctionnement correct des syst&mes,des essais de 5 en 5 degrds ont dtC effectues depuis - 25 OC jusque vers - 5C OC. 1.5 DIJREE DE CHAQUE PHASE D'ESSAIS La majorit6 des essais a it6 effectuee pendant des durees continues de 60 et 90 minutes en conditions extrtmes de givrage. 1.6 CONFIGURATIONS 2 configurations ont BtB testbes : - sans rechauffeur des servomecanismes - avec rechauffeur des servomecanismes 1.7 FLUIDES * CARBURANT Pour tous les essais, le carburant utilis6 etait du TR-0 (JET Al) sans additif anti glace. I1 repondait i la norme AIR 3405/C. *- L'huile utilisee sur le circuit des echangeurs etait de l'huile synthetique repondant aux exigenaes de la norme AIR 3514 (type I). z - ~TERIELSEN EssaI La configuration d'essai etait en tout point identique au montage des circuits carburant et huile du moteur (equipements, diamstres de canalisation, boucles d'asservissement ...1. Les ecarts par rapport aux modsles definitifs ne portaient, que sur l'aspect physique exterieur, dus aux modes de fabrication, mais ne remettaient pas en question leur fonction. Par ailleurs, les liaisons mecaniques et hydrauliques itaient identiques i celles utilisees sur moteur. 3 - INSTWATION D'ESSAI La definition, la realisation et la mise au point de cette installation ont BtC effectudes par le CEPr de SACLAY sur la base des spCcifications d'essai Cmises par la SNECMA . Elle se composait des parties principales suivantes : * un circuit carburant * un circuit d'huile * un circuit d'air reguli. * une enceinte climatique * un dispositif d'entrainement mecanique 3.1 LE CIRCUIT CARBURANT Parmi les differentes procedures d'essais de givrage appliquees au CEPr sur les circuits de carburant et leur composants , c'est celle du "refroidissement prealable et d'ingestion d'eau proportionnelle" qui a et& retenue, car elle permet de realiser des dosages precis et des essais de longue duree : plus de 9 heures dans notre cas. Dans ce type d'essais, une des difficult& vient du fait que le debit d'eau i injecter dans le carburant est souvent faible :quelquer; litres par heure. En consequence, on Cree 2 circuits : * Le circuit principal constitue d'un reservoir de grande capaoitb contenant le carburant necessaire B l'exbcution de l'essai, refroidi B la temperature demandee et envoye vers les materiels i tester. * un circuit secondaire comportant un reservoir d'eau et un reservoir de carburant. I1 permet d'obtenir un melange eau + carburant i la tempsrature ambiante. Ce melange est introduit dans le circuit principal par un injecteur qui delivre un debit correspondant i la concentration souhaitbe. 23-9

Cinq fonctions doivent Stre assurees : * reserve de carburant froid et sec * injection d'eau * recuperation du carburant contamin6 par l'eau * rincage des circuits au carburant sec * sechage du carburant 3.1.1 RESERVP. DE CARBVRA" FROID ET SEC Le carburant d'essai prealablement "sech6" par un filtre coalesceur etait stock6 dans deux reservoirs de 4,5 m3 chacun, comuniquant en leur point bas. Lors de la preparation du carburant froid, ils etaient raccordes en circuit ferme, successivement sur deux refroidisseurs : * l'un utilis6 pour refroidir jusquIB - 35 OC, constitug par un groupe moto compresseur B freon capable de compenser un apport thermique de 75 kw a -35 oc. * l'autre, utili& pour descendre au dessous de - 35 "C, constitue par une installation B evaporation d'azote liquide, capable de compenser un apport thermique de 10 kw aux plus basses temperatures des essais. ce carburant froid et sec filtr6 B 50 pm, etait achemine aux materiels en essai par une pompe capable de 5 ml/h sous 2,5 bars relatifs; la pression et le debit etant reglables par un systeme de vannage. 3.1.2 INJECTION D'EAV Le melange, B raison de 20 % d'eau et 80 % de carburant, etait prepare automatiquement par pompage, filtration, et reglage des debits dans deux reservoirs B temperature ambiante : * un reservoir d'eau d&min&ralisee de D,1 m1 * un reservoir de carburant de 0.4 m1 Apres regulation et mesure du debit, la partie de ce melange correspondant .i la concentration recherchee est injectee dans la veine de carburant froid; l'excedent etant renvoye dans un reservoir de recuperation de 0,5 m'. L'injecteur, soumis exterieurement aux basses temperatures du carburant froid, etait pourvu d'un dispositif de rechauffage B huile chaude pour &iter la formation de glace et l'obstruction. A l'endroit de l'injection, la veine Btait transparente pour permettre de regarder l'etat de l'injecteur lui-meme mais aussi la formation des paillettes de glace dans le courant froid. La pulverisation de produits anti-givre sur la paroi externe de la veine etait frequente pour ameliorer la visualisation. 3.1.3 RECUPERATION DV CARBURANT CONTAMINE PAR L ' EAV A la sortie du circuit carburant moteur en essai, le carburant froid, charge de particules de glace, etait dirige vers un reservoir d'une capacite de 11 m1 pour stockage temporaire avant traitement. 3.1.4 RINCAGE DES CIRCUITS AV CARBVRA" SEC Juste apres chaque phase d'essai en conditions givrantes, les circuits en essai 6taient aliment& avec du carburant sec B temperature ambiante provenant d'un reservoir de 0,72 m'. Apres chaque passage dans le circuit du moteur, le carburant etait &charge de son eau par passage dans le filtre coalesceur avant de retourner au reservoir. 3.1.5 SECHAGE DU CARBVRA" D'ESSAI De m6me que precedemment pour le rincage des circuits, l'operation de sechage du carburant avait lieu apres chaque essai. Elle s'effectuait en 2 phases : * Premiere phase Rechauffage du carburant stocks dans le reservoir de 11 m' par circulation en circuit ferme. * Deuxieme hase Transvasezent du carburant du reservoir de 11 m1 dans les deux reservoirs de 4,5 m3 apres passage au travers du filtre coalesceur charge de separer l'eau libre du carburant. Des analyses effectuees suivant la methode de KARL-FISCHER B l'aval du filtre coalesceur permettaient de contreler la teneur residuelle en eau libre. Elle se situait aux environs de 80 PPM. 3.2 LE CIRCUIT D'HUILE Ainsi que nous l'avons d&j& signale au paragraphe 1.2, ce circuit etait charge de restituer exactement les conditions des transferts de chaleur du circuit d'huile moteur vers le circuit carburant. 23-10

11 se composait des principaux elements suivants : * un circuit comprenant : un reservoir, un vase d'expansion, une pompe de circulation. * un dispositif de chauffage de l'huile. * Un circuit vehiculant l'huile chaude vers les ichangeurs du circuit carburant en essai comprenant : une pompe vol.um6trique, un systeme de vannage pour restituer le debit et la pression d'huile que l'on aurait sur moteur. I1 convient d'apporter quelques pr4cisions sur le dispositif de chauffage. I1 Btait constitue par : * un ensemble de 4 giroupes de resistances Blectriques immerges dans le reservoir d'huile et dont la puissance totale Btait de 60 kw (6+12+18+24). * un ensemble de regulation de temperature B partir d'une sonde (thermocouple) constituC par un regulateur B modulation de temps du type Proportionnel - Integral De plus, afin de simuler au mieux l'evolution de la puissance transferee au carburant par l'huile moteur du fait d'un colmatage Cventuel par le givre de l'echangeur principal , un boitier de regulation de la chute de temperature d'huile B travers l'echangeur, donc la puissance transferee au carburant, avait et6 intercale dans le circuit principal de regulation de cette temperature. La loi de regulation etait representative du cas de fonctionnement simulk; la constante de temps choisie (2 min B 63% de 1'Bchelon) correspondant B celle du moteur CFM56 en ce qui concerne les phenom6nes de rejection de puissance dans l'huile . 3.3 SIMULATION DE LA PRESSION-CCMPRESSEUR L'indication de la pression statique sortie compresseur H.P. "PS3" au regulateur etait recrGe par une batterie de 2 bouteilles d'air cornprim6 B 200 bars equipees chacune d'un detendeur 200 - 40 bars. Cet air dCtendu alimentait un regulateur de pression permettant l'ajustement et le maintien de la pression B la valeur fixCe par le programme d'essais. 3.4 L'WCEINTE CLIMATIQLIE L'ensemble de regulation, comprenant les pompes, le filtre, Les Bchangeurs et le rhgulateur lui meme, Btait install6 dans un caisson,climatique,B l'exclusion du moteur hydraulique d'entrainement des V.B.V. et du verin de commande des stators variables du compresseur H.P. qui Ctaient situ& B l'exterieur du caisson. Cette enceinte confectionnee, B l'aide de panneaux isolants en laine de verre etait refroidie par un Bvaporateur B azote liquide alimente par un reservoir de 0,6 m' de N2. L'homogeneite de temperature dans l'enceinte etait assuree par un ventilateur brassant continuellement l'ambiance; la vitesse autour des Bquipements etait reprksentarive de celle obtenue sur avion. La regulation de la temperature .il'interieur du caisson etait assuree par l'ouverture et la fermeture d'une Blectrovanne assurant l'alimentation en N2 liquide et commandee B partir d'une sonde de tempQrature et d'un regulateur. 3.5 LE DISPOSITIF D'ENTRA1"T MECANIQUE Le dispositif d'entrainement du regulateur et de ses pompes Btait compose par les deux eliments essentiels suivant : - un moteur electrique B courant continu B vitesse variable d'une puissance nominale de 195 kw, capable d'une vitesse maximale de 3000 tr/min. - un multiplicateur de vitesse de rapport 1 B 3 muni de son syst6me de lubrification et dont l'arbre de sortie etait accoupl6 celui du regulateur. L'alimentation du moteur et le reglage de sa vitesse de rotation etaient assures par un generateur electrique statique 2 thyristors. 4 - INSTRUMENTATION L'instrumentation permettait de verifier B tout moment des essais les conditions de simulation et de dkterminer le comportement des materiels. Hornis les parametres nhcessaires B la conduite des essais, le systi-me permettait l'acquisition d'une cinquantaine de valeurs dont certaines visualisees sur &ran et table tracante en temps rkel. Citons les plus importantes : - les debits de carburant, d'eau injectee, d'huile - les pressions d'injection, sortie pompe B.P. et H.P., entree et sortie rechauffeur 23-11

- les pertes de charge du filtre carburant, des echangeurs - les temperatures de fluide : * carburant : entree --DOmDC B.P., entree et sortie Bchanueurs. sortie regulateur * huile : entree du rechauffeur de carburant, sortie du refroidisseur d'huile * B l'interieur du caisson climatique - les temperatures de masse : pompe carburant, regulateur, Bchangeurs. 5 - PROCEDURES D'ESSAI Des procedures de mise en route, d'arrst et de reconditionnement de l'installation avaient &e etablies et affinees lors des essais prgliminaires qui s'etaient deroul6.s du 26 OCTOBRE au 21 DECEMBRE 1977 pour optimiser et figer les nombreuses manipulations afin de renare les divers essais parfaitement comparatifs entre-eux.

6 - CRITERES DE REUSSITE Ces essais qui Gtaient effectues dans le cadre de la certification du moteur avaient pour but : - de demontrer que le circuit carburant etait capable de fonctionner sans interruption - de requler correctement le debit carburant dans les conditions de givrage susceptibles d'stre rencontrees dans le domaine de vol prevu - de determiner la durCe pendant laquelle ces fonctions etaient assurees correctement. En consequence, le comportement du circuit carburant devait etre analyse, en regime permanent, en terme de la stabilite de certains parametres de regulation : en particulier , debit carburant inject6 . position des vannes de decharge (VBV) et des stators variables IVSV), parametres qui servaient de reference pour justifier la rCussite de l'essai et la conformite des circuits vis B vis des rsglements. Bien que n'intervenant pas directement sur le fonctionnement du moteur lui-meme, le comportement de l'echangeur principal et du rechauffeur etait note pour chaque point d'essai, tant du point de vue thermique qu'hydraulique. 7 - RBSULTATS DES ESSAIS ET ANALYSE

7.1 CONFIGURATION SANS RECHAUFFEUR DE CARBURANT DES SERVOMECANISHES Le montage Btait donc celui du circuit carburant de base ccmprenant essentiellement en suivant le sens de l'ecoulement : . la pompe basse pression recevant le carburant frcid B la temperature TCa et B la pression PCa . l'echangeur huile/carburant . le filtre B carburant . la pompe haute pression . le regulateur et ses servomecanismes . le circuit d'injection . la boucle de recirculation 4 essais ont et6 effectues en condition decollage avec des temperatures de carburant B l'entree de la pompe basse pression suivantes : - 25 "C, - 30 OC, -35 OC, - 40 OC pendant des durees de 35 min pour les deux premieres temperatures et 60 min pour les deux autres. Comwrtement des circuits Deux comportements sont apparus suivant la temperature de carburant B

1'entri.e- .~~~__. des Couioaments :~ . temperature superieure B 0 OC (TCa 2 - 30 'C) : tous les circuits fonction- naient correctement, les parametres restaient stables au niveau des valeurs initiales temperatures negatives B l'entree des Gquipements (TCa < -30 OC) : le filtre B carburant etait le premier equipement B subir les effets du givrage. La perte de charge de l'element filtrant commen~antB croitre 1 B 2 minutes apres le debut de l'essai d'injection d'eau et le colmatage total entrainant l'ouverture du by-pass, intervenant aprk 8 B 10 minutes. A noter cependant, que plus la temperature etait basse, plus l'augmentation de la perte de charge Btait lente. Ceci peut s'expliquer par la consistance et la structure de la glace : des cristaux mous s'agglutinant plus facilement que des cristaux durs. En ce qui concerne l'injection du debit carburant dans la chambre et les systemes d'asservissement (VBV, VSV), on notait un fonctionnement ccrrect pendant au moins huit minutes. Au deli, apparaissaient des variations de debit du carburant inject6 de i 10 % environ autour de la valeur nominale, pouvant devenir au bout d'une dizaine de minutes suffisamment significatives pour affecter le fonctionnement du moteur. Les cristaux s'accumulant, il s'ensuivait 23-12

des instabilites de plus en plus importantes affec.tant les pressions d'asservissement des stators variables (VSV) et des vannes de decharge (VBV). Par exemple, l'essai B - 40 OC a 6te arrSt6 38 minutes aprk l'injection d'eau: la variation des parametres Btait telle qu'il n'etait plus possible d'exploiter les enregistrements. Des battements de 40 bars ont et6 releves sur la pression d'injection du carburant dans la chambre. A noter que dans tous les cas, 1'6changeur principal a toujours fonctionne correctement, sa perte de charge n'ayant d'ailleurs que tres peu varie : + 15 % au maximum. Ces essais ont montri: clairement que la pilotabilite du moteur est affectee tres rapidement dPs lors que les temperatures du carburant B l'entree du regulateur et des servomecanismes deviennent negatives. 1.2 CONFIGURATION AVEC RECHAUFFEUR DES SERVONECANISM?.!~ Dans cette configuration, un echangeur de chaleur rechauffe le carburant des systemes d'asservissement prelevg B la sortie de la pompe BP et dont le debit est environ 114 du debit consomi.; ce rechauffage etant effect116 par la totaliti! du debit d'huile a la temperature de sortie du moteur. Pour le reste, le circuit est le m6me que le circuit de base. 4 essais ant egalement &ti. effectues en condition decollage, soit pour des temperatures de - 25 OC, - 30 OC, - 35 OC, - 45 OC et des durees de 90 minutes. Comportment des circuits Mise B part l'ouverture du clapet de derivation du filtre pour les points B - 30 OC, - 35 OC, - 45 OC, quelques minutes aprk l*injection d'eau, ce qui traduit un comportement identique B 7.1, les syst6mes de regulation sont peu affectes par le presence de glace dans le carburant. Le debit carburant inject6 et la position des servomecanismes (VBV + VSV) sont rest& stables pendant les 90 minutes de l'essai, except6 pour TCa = - 45 OC oi~B 88 minutes exactement un accroissement de 10 % du debit carburant inject6,est a,pparu. Puis le debit est rest6 stable B sa nouvelle valeur jusqu'a la 94 em minute, moment ofi est appqru un deuxikme accroissement de pres de 40 %; l'essai ayant et6 arrSt8 a la 96 minute par suite de l'epuisement de la reserve destine B l'essai. Ceci montre que : des temps de fonctiorinement forts longs peuvent 6tre necessaires avant que n'apparaissent les premiers symptcmes dus au givrage et qu'il convient en consequence d'6tre prudent en la matiere quant aux extrapolations. . le systeme propose a ses limites come nous le pensions au cours de l'analyse puisque le carburant le long du circuit d'injection est toujours rest6 i temperature negative.

A noter qu'en ce qui concerne l'echangeur principal, il a et6 relev6 des colmatages de la matrice, conduisant B l'ouverture du clapet de derivation, puis 2 decolmatage pour TCa = -45 OC come attendu par l'analyse, suite au rechauffage des plaques frontales par l'huile. Mais ce decolmatage ne s'est PES produit B des temperatures moins froides; ce qui traduit encore l'effet de la structure de la glace sur les ph6nomPnes d'a.gglutination des cristaux et d'obstruction qui s'ensuivent.

Pendant~~ toute la duree~ de ~~ ces essais. les~~~ Dertes~ ~ de ~ ~ charoe ~~~~~~ ~ du rechauffeur ~ ~ ~ ~~~~~ ~ ~ ~ ~. du carburanc dcs scrvon6canismcs Sone resc6cs scxbles aux vslcurs d'0rigir.c craduisan-.qu'il n'y a jamais cu d'accumulacion de glace; lcs CenperaLurcs de paroi &ant toujours B temperature positive. 8 - CONCLUSION DES ESSAIS Les DrinciDaux enseisnements (ou confirmations) sui decoulent des essais effectues en conditions givrantes di 18 Janvier au 30 Mars 1978; sur des circuits carburant typiques de la famille des moteurs CFM56, fonctionnant avec du carburant contenant de l'eau dans les proportions definies par les rsglements, soit 300 PPM en volume (300 B 380 PPM lors des essais), sont les suivants : * Le fonctionnement correct est assure sans restriction de duree due B la presence de l'eau, des lors que la temperature du carburant h l'entree du regulateur est superieure B 0 OC, ceci malgre le colmatage 4ventuel du filtre principal. * Pour des temperatures de carburant inferieures B 0 OC au regulateur, le fonctionnement correct du circuit carburant n'a pu etre assure que pendant des durees limitees dont les valeurs dependent des niveaux de temperature B l'entree de la pompe BP et bgalement du taux de poussee delivr6 par le moteur. * Le premier ph6nomene anormal rencontre aprh 8 B 10 minutes de fonctionnement est une variation periodique du debit carburant, delivre B la chambre, autour de sa valeur nominale, ce phenomene restant stable dans le temps. 23-13

* Le deuxieme phenomene est une obstruction du circuit des servom6canismes du regulateur conduisant B une derive des parametres essentiels au fonctionnement du moteur : position des vannes de decharge et des stators variables, contrBle du debit carburant consome ; ce second phenomene apparaissait 10 B 35 minutes apres fonctionnement en conditions givrantes. Cette limite de fonctionnement correct, sans limitation d'aucune sorte sur les conditions moteur, est atteinte avec le circuit de base du CFM56, non equip6 de rechauffeur du carburant des servom6canismes, lorsque la temperature du carburant B l'entree de la pompe BP descend B - 30 OC. Cette limite est repoussee B - 45 OC par l'adjonction d'un petit Bchangeur huilelcarburant dont le r6le est de rechauffer le carburant qui circule dans les servomecanismes plus sensibles aux effets du givrage que le circuit principal du regulateur; la puissance calorifique extraite du circuit d'huile &ant sensiblement la mime dans les deux cas. Ainsi que nous l'avons deja vu, des fonctionnements au deSsOUS de - 30 "C sans rechauffeur conduiraient assez rapidement a une impilotabilite du moteur; de m&me, dans les m@mes conditions moteur, des fonctionnements au dessous de - 45 OC conduiraient aux m&nes anomalies avec des circuits equip& de rechauffeur, et des systemes annexes seraient necessaires pour "descendre" plus bas. Cependant, jusqu,B ce jour, aucun avionneur, aucune compagnie aerienne n'est demandeur, une temperature de - 45 Qc n'etant vraisemblablement pas de pratique courante !

IjQme Pi4RTIE - ACTUALISATION

Les etudes presentees precedement datent de l'epoque de la definition et du developpement des premiers CFM56. Elles sont donc vieilles de 15 ans aujourd'hui. Les essais decrits en deuxieme partie datent de 1977-1978. Qu'en est-il aujourd'hui ? * les syst&nes definis B 1'Bpoque se sont-ils revel& satisfaisant au cours de ces presque 10 annees d'exploitation des CFM56 ?

* Quelles sont les tendances actuelles ? 1 - EXPERIENCE ACQUISE 1.1 EN ESSAI EN VRAIE GRANDEUR SIR AVION

Au~~~~ debut ~~~~~ ~ de ~~~ ~l'utilisation ~~ ~ ~~ ~ des CFM56-2B sur les avions ravitailleurs KC135R ~~USAFa effectue B EGLIN [FLORIDE) des eskais climatiques sur un avion equip6 de ses 4 CFM56-2B. L'avion Btait install6 dans un hangar B l'intkrieur duquel la temperature avait et6 abaissee et maintenue B des niveaux compris entre - 29 et - 51 OC, pendant des durees suffisament longues, de facon que l'ensemble avion + moteurs ait eu le temps de se mettre B la temperature ambiante du hangar. TOUS les essais effectues montrerent un fonctionnement correct des moteurs lors des demarrages et des quelques minutes effectuees au ralenti; les durees de fonctionnement ne pouvant qu'itre courtes puisque l'ejection des gaz se faisait dans le hangar. On ne connait pas la quantiti: d'eau que pouvait contenir le carburant. Cet essai est relate car c'est le seul cas, B notre connaissance, oii des CFM56 ont fonctionne i si basse temperature au sol. 1.2 EN EXPLOITATION Contrairement i ce que l'on peut penser, il n'est pas ais6 pour un motoriste de connaitre quelles ont et& les conditions les plus severes que ses moteurs ont pu rencontrer en exploitation ... tout du moins quand ils ont fonctionne sans faire parler d'eux ! Apres 20 millions d'heures de vol cumulBes par tous les moteurs, nous n'avons jamais eu de plaintes concernant un moteur qui aurait eu des troubles de fonctionnement suite B un givrage presume du circuit carburant; certains de ces moteurs oo6rant Dourtant en zones r6Dut6es tres froides B certaines DCriodes de l'annee 0; ayant'rencontre en altituhe des couches plus froides que ies - 56,5 OC de l'atmosphere standard valables au dessus de 11 km. Mais il faut dire aussi que du carburant B - 45 "C au sol doit Stre fort peu frequent et qu'en vol, il faudrait voler plus de 5.5 heures B M = 0.8. au regime de decollage en ISA - 15 OC [Tamb = - 71 "C) et avec 300 PPM d'eau dans le carburant pour qu'il y ait une probabilite que les premiers signes de givrage apparaissent (cf 2eae partie). On peut donc en conclure que la solution couvrant des conditions jusqu's - 45 OC est satisfaisante. 23-14

2 - NOWEAUX CONCEPTS DE CIRCUI'I! CARBURANT Sur certains moteurs, pour ameliorer la consommation specifique du moteur install&, le refroidissement du circuit d'huile des alternateurs de l'avion est assure par le circuit carburant. Ce concept permet d'eviter l'echangeur situe derriere la soufflante, Bchangcur volumineux qui Cree des pertes de charge et des h6terogeneitGs de debit dans l'ecoulcment secondaire; ce qui se paie par une surconsommation. sur les CEM56-SA qui propulsent les AIRBUS A320, un echangeur huile-IDG/carburant est install6 dans la boucle carburant du moteur. Une vanne thermo-commandbe associee B un circuit de retour carburant aux reservoirs de l'avion permet d'obtenir un debit de refroidissement superieur au debit de carburant consomme par le moteur, particulierement aux :bas regimes pour lesquels la consommation est insuffisante pour assurer le refroidissement et du moteur et de son alternateur avion sans entrainer des sur-temperatures sur les circuits. Aux regimes de d6collage, cette vanne est en position fermee pour des questions de securite mais aussi parce que la consommation &levee le permet, meme aux conditions les plus chaudes. La Puissance calorifique, qui Provient du refroidissement d'un alternateur delivrant une puissance 6lectrique de 60 kva, represente environ la moitib de celle gener6e au plein gaz par le circuit de lubrification d'un CFM56. l'alternateur est alors un merveilleux rechauffeur de carburant ... tout du moins pour les conditions froides. Dans les mhes conditions que celles des essais decrits en 2 partie, le rechauffage de la totalit6 du debit ConsommG est de l'ordre de 10 OC , au regime de decollage en ISA - 124 OF (Tamb = - 54 "C). Ce qui signifie que sans rechauffeur des servomecanismes les phenomenes de givrage ne se manifesteraient que vers - 40 OC et vers - 55 OC avec un rechauffeur des servornecanismes. Par ailleurs, si l'on tient compte que la consommation specifique des moteurs d'aujourd'hui est plus faible que ceux conGus il y a 15 ans, on est en droit de penser B la suppression du rechauffeur des servomecanismes. La seule difficult6 est que si une panne intervient sur l'alternateur et que le pilote soit contraint de le debrayer mecaniquement, on ne saurait plus tenir que - 30 OC environ puisque l'on est ramen6 au circuit de base ... pour lequel il avait fallu un rechauffeur. Seules des etudes fines permettent de statuer. 3 - NOWEAUX MODELES MATUEMATIQLIES AU fil des annees, le modele decrit en premiere partie a eu plusieurs descendants. D'abord pour satisfaire la demande des avionneurs (BOEING, AIRBUS INDUSTRIE) qui avaient besoin de modeles mathematiques moteur pour : * aider i la certification de l'aeronef (extrapolation aux conditions chaudes et froides) * mieux restituer dans les simulateurs de vol les ind.ications pilotes provenant des circuits huile et carburant. Mais aussi pour nous, afin de mieux approcher la realite des phenomenes physiques. Le modele, qui au depart ne calculait gue des temperatures en regime stationnaire, a pu assez rapidement effectuer des calculs en regime thermique instationnaire.

Auiourd'hui~.~~~. .~ our les etudes~~~ sur ~ CFM56-5C~ ~ ~ ~ destinG ~~~~~~ B~~ l'avion ~ ...~~,A140. le~~ mod&le ~~~~ _._. eat__~ capable-de calculer les temperatures des circuits huile/carburant du moteur mais aussi du circuit d'huile de l'alternateur le long d'une mission complete, tout en evaluant la temperature du carburant delivre par l'avion aux moteurs, avec et sans recirculation de carburant dans ces r8servoirs. En un mot, la simulation est complete. Ainsi, on peut suivre par le calcul l'evolution des temperatures du carburant lors de vol en conditions extrgmes froides au niveau de 1'entri.e du regulateur et des servomecanismes, comparer les valeurs mais aussi les durees par rapport aux essais de certification decrits en 2enepartie et determiner les marges de securite en terme de givrage . Les etudes effectuees avec de tels modeles permettent d'optimiser encore plus finement les solutions technologiques et conduisent B des gains en masse et en coiit. 4 - PIEGES A GIVRE

~~~~~ ~ ~~~~~~~~ ~~~ .;e NOUS avons resolu les vroblemes de aivraoe- ~~ des ~~~~ circuits carburant moteur alii -~ sont poses B nous en rechauifant suffisamment le carburant pour faire fondre les cristaux de glace. Mais nous ne serions pas complets si nous ne parlions pas des systsmes bases sur la captation des cristaux. De tels systeines sont utilises avec succes sur certains moteurs d'hglicopteres tels que le "MAKILA" uui 6quipe'le SA331 SUPER PUMA. Comme les essais l'ont montre (cf ZkRepartie) le plus simple des systemes passifs pourrait Etre un filtre B carburant. Mais pour assurer des durees de fonctionnement suffisamment longues, l'encombrement risquerait d'etre prohibitif car la forme plissee 23-15

slav&re inadequate, la glace faisant des pontets entre les plis; tres rapidement le filtre ne travaille plus que sur sa peripherie. En consequence, les syst&mes sont plus compliqu6s et ressemblent plut8t. pour certains, B un Bchangeur de chaleur sans tubes, dans lequel ne subsistent que les chicanes. La captation se faisant d'abord sur les premieres chicanes, puis sur la deuxieme et ainsi de suite. Ces systkmes simples dans leur principe, demandent une longue mise au point faite B base d'essais pour determiner l'agencement interne gui conferent l'autonomie la plus longue. Si ces systemes sont si.di.isants dan; leur principe puisqu'ils ne necessitent pas d'energie, ils ont cependant quelques inconvenients : - leur autonomie est limitee - les debits de carburant i traiter doivent rester faibles sous peine d'encombrement important - la captation peut ne pas itre totale - enfin, % l'arrit du moteur, il faut degivrer le systeme et drainer l'eau de fusion de la glace. Tous ces inconvenients font que de tels dispositifs sont difficilement utilisables sur les turboreacteurs des avions commerciaux. 5 - CONCLUSION Pour les moteurs de la classe des 100 Mewtons de poussee et au del%, il semble que la meilleurs solution, pour &iter les consequences dues au givrage, soit celle qui consiste i rechauffer suffisamment la partie du carburant qui circule dans les organes les plus sensibles par un petit rechauffeur huilelcarburant. L'energie disponible sur le circuit d'huile doit permettre de couvrir de facon satisfaisante les conditions les plus sevires. Sur les moteurs tr6s ¢s, compte tenu de l'am8lioration de leur consommation specifique, des vitesses de rotation de plus en plus elevees, de l'environnement de plus en plus chaud autour des enceintes paliers de par l'augmentation des taux de compression, voire de la prise en charge du refroidissement des alternateurs par le circuit carburant moteur, un "circuit de base" consistant B rechauffer tout le debit consomme pourrait itre envisageable jusqu'i des temperatures de - 40 OC environ. L'experience montre que, pour &iter les phenomenes de givrage, il suffit que la temperature du carburant i l'entree des organes sensibles soit tres 14g6rement au dessus du point de congelation de l'eau; les cristaux de glace ayant peu d'inertie thermique d'une part, &ant donne leur faible masse, et, d'autre part, la temperature de peau des circuits en contact avec le carburant &ant tres proche de la temperature du fluide en raison du rapport des coefficients d'gchange entre le carburant et l'air qui environne les Cquipements. Neanmoins, si, au sein d'un mime equipement sensible, une circulation de carburant froid jouxte une circulation de carburant chaud, les coefficients d'echange par convection risquant d'6tre du mime ordre de grandeur, une analyse thermique devra &re effectuee pour s'assurer que la temperature des parois n'est pas negative sur le circuit chaud. Cependant, &ant donne les situations extrhement graves que peut entrainer le givrage des circuits de carburant, des essais recreant au plus pres les conditions de fonctionnement sur moteur en situation extrsme de givrage doivent itre effectues sur les circuits de la definition finale pendant des durees suffisamment longues pour s'affranchir de tout risque.

Discussion Author: 1. R. Jacques, Ecole Royale Militaire Pour une condition de fonctionnement donnie, la courbe La puissance calorifique degagiaestuniquementfonction de est unique la puissance calonfique recueillie par I'huile ne la tempiralure d'entrer de I'huile. Nya f-il des lors anciens depend plus en fait que de la viscositi de l'huile,soit de la autre parametre que la viscositi d'huile fonction de la temphature pour une huile donnie,ce qui Btait le cas dans temp&ature, qui influence ce transfert de chaleur? l'exposi. 23-16

Fig. 1 STOCKAGE TYPE DU CARBURANT EN EXPLOITATION CIVILE TETE : preambule

CENTRAL

CUVEDESTOCKAGE Abriennes (civil) ou INTERNE Enterrees (militaire) EXTERNE

Fig. 2 SYS'TEME CARBURANU DE BASE TEXTE : paragraphe 1 lkre partie

PARTIE HOTtUR

REGULATEUR POHPL Dt 23-17

Fig. 3 LE MODELE DE CALCUL TEXTE : paragraphe 4 - Ere partie

Parametres du cycle thermodynamique THERMODY NAMlQUE - V 1

Donnees definissant la I TECHNOLOGE I geomBtrie du moteur.

Calcul de la puissance CIRCUIT D'HUILE calorifique recueillie par I'huile. Les Differents I V Modules ECHANGEURS V CIRCUIT CARBURANT

RESULTATS I I

Fig. 4 ORIGINE DE LA PUISSANCE CALORIFIQUE RECUEILLIE PAR L'HUILE TEXTE : paragraphe 4.1 - lire partie

EVOLUTIONS DE LA PUISSANCE HUILE AVEC LE REGIME ET LA TEMPERATURE 23-18

MODULE 'I CIIRCUIT D'HUILE 'I TEXTE : paragraphe 4.3.1 lere partie

80 A ~ REGMEMOTEUR

50 .

30 .

20 .

10 Fig. 7

r I Pvissanoe re~usllliepar I'hulle WH (kW1 Fig. 8

PUUSSANCE CALORIFIQUE DECOLLAGE DISPONIBLE SUR - LE CIRCUIT D'HUILE MAXI-MONTEE __ - 30 TFRE : paragrapbe 6 - lire partie 20 CONDITIONS : Z=O M=0.6 ISA-124°F 10

~C"1-Y 0 10 20 30 40 50 BO TO 80 90

4 slvraps tortemen1 probable DECOLLAGE - I MAXI-MONTEE - Fig. 9 ....MAW-CONTlNll .... - -...... _ -.- .. . TEMPERATURES DU CARBURANT A L'ENTREE temp6rsture du carburant DES SERVOMECANISMES TEXT! : paragraphe 6.1 - lire partie

CONDITIONS : Z=O M=0.6 ?I ISA-124'P (Tamb=-54'C) I I I 1 -BO -50 -40 - 23-19

Fig.10 TEMPERATURES DU CARBURANT A L'ENTREE DES SERVOS ET DU REGULATEUR CONDITIONS : Z=O

temperatUre carburant (%c) DECOLLAGE - I I T MAXI-MONTEE - - -

TEXTE : paragraphe 6.2 ler partie I I I I I -60 -50 -40 -30 -20 -10 0

INJECTION D'EAU

Fig. 11 ...- DlSPOSlTlF DES ESSAIS DE GIVRAGE

CIRCULATION DU CARBURANT FROID

TEXTE : paragraphe 3.1.1 RESERVOIR . 11 m3 hepartie

POWE Y.P. 23-20

'EUR NEBIT

INJECTEUR DEGIVRE \ 1 I I

I I

Fig. 13 DISPOSPTIF DES ESSAIS DE CIVRAGE

I 1 I RINCAGE DES CIRCUITS I

TEXTE : paragraphe 3.1.4 Am 2he partie 2ARBURANT FILTRE 23-21

Fig. 14 DlSPOSlTlF DES ESSAIS DE GIVRAGE

SECHAGE" DU CARBURANT D'ESSAI I I Im 18 I i I-' I TEXTE : paragraphe 3.1.5 Pi COALESCEUR Zhe partie +*hjCIIR8"P.INT

POW< n.s

Fig. 15

TWTE : paragraphe 3.2 Zhe partie REGULATEUR DE TEMPERATURE

VERS LE REGULATEU DE TEMPERATURE Fig.16 I I EVAPORATEUR DlSPOSlTl F DES ES§AIS DE GIVRAGE ELECTROVAHNE

ENCEINTE CLIMATIQUE ENCEINTE r- CL.lMATlQUE

TEgTE : paragraphe 3.4 D'AZOTE LIqUIDE

2h partie (MOTEUR ELECTRIQUE)

VENTILATEUR DE BRASSAGE

Fig. 17 NOUVEAUX MODELES MATHEMATIQUES EXEMPLE DE CALCULS EFFECTUES POUR CFM56-5C TEKTE : paragraphe 3 - %me partie

RALENTI 10 montbe CROlSlERE DESCENTE I- I- I* ' RALENTI temperature Carburan1 I entree serYo-mec*"ism..

temperature d'huile entree moteur I

IempBratere carburant ~YXinjeoteurs

I L 24-1

THE INFLUENCE OF FUEL CIIARA(:TERISTI(.'S ON HETEROGENEOUS FI.,\ME PROP,\(;ATION M.F. Bardon, J.E.D. Gauthier and V.K. Rao Jkpamnent of Mechanical Engineering Royal Military College of Canada Kingston, Ontario K7K 5LO Canada

P = liquid fuel ABSTRACT 8 =gas phase This paper describes a theoretical study of flame pre L = laminar condition pagation through mixtures of fuel vapour, droplets and air mean =mean value (between reactants and products of under conditions representative of cold starting in gas turbines. combustion) It combines two previously developed models - one for mf =multicomponent fuel heterogeneous flame propagation and the other for describing 0 = condition in the reactants just ahead of the flame the complex evaporative behaviour of real fuel blends. Both front models have been validated against experimental data. and the r = (of all the) reactants combined model incorporates the effects of pressure, SOI =saturated conditions temperature, droplet diameter, turbulence intensity, delivered surf =droplet surface condition equivalence ratio, fuel prevaporization, and fuel type on flame T = turbulent condition propagation. Differences in the combustion performance of Jet V = fully prevaporized, is. gaseous condition Al, JP 4 and two single component reference fuels are compared. Conclusions are drawn regarding the use of pure INTRODUCTION compounds to represent real fuel blends, and the relative Flame propagation into heterogeneous mixtures of fuel importance of various engine conditions and spray parameters droplets, fuel vapour and air has received only limited on combustion. attention and is of great interest in many practical applications, including gas turbine main combustors. Mathematical models LIST OF SYMBOLS which predict the effect on combustion of changes in chemical, physical or aerodynamic propenies of a a.b. c. d. -coefficients for volatility equations (Appendix heterogeneous mixNre provide valuable insight into the ij '' " ' - relative importance of various parameters and can thereby aid B =mass transfer number in the design of future combustion systems. In particular, the need for a better understanding of both the qualitative and cp.G =specific heat and average specific heat at quantitative effects of fuel properties on flame propagation constant pressure motivates this work. Appropriate design strategiesrequire a Cl,Cz = Clausius-Clapeyron constants thorough understanding of the interaction between Cl,Cz,C3 =droplet size distribution constants (Appendix C) fundamental fuel characteristics and combustion behaviour. c, = shape factor in equation (4) Ultimately, a complete understanding of the many 4 =characteristic dimension in equation (4) subprccesses (fuel droplet trajectories, evaporation, mixing, D =droplet diameter reaction rate, turbulence, etc.) involved in heterogeneous E =activation energy flame propagation would allow a comprehensive computer f,f2 = temperature independent functions of fi model to be developed. Preliminary versions of such "exact" H" =latent heat of vaporization of the fuel at boiling models have been proposed (e.g. Aggarwal and Sirignano, point 111 and Swithenbank et al, [2]). The most common approach AH =enthalpy of vaporization in this type of numerical model is to solve a set of goveming k =thermal conductivity differential equations using some finite difference scheme. K =proportionality constant (in equation (A-7)) This approach is clearly necessary in the long Nn, since it is M =molecular weight the only means to completely model the physics of the MDD =mean droplet diameter problem. However, existing work is still rather embryonic, MSD =mean surface diameter and arguably, is still tw complex and uncertain for use in MMD = mean mass diameter most practical analyses. There is therefore an ongoing P =pressure requirement for simpler more tractable phenomenological pi = Legendre polynomials (Appendix B) models. Such models are developed using basic thermal R =universal gas constant considerations [3,4, ,5,6,7], considering flame propagation Re, = Reynolds number based on droplet diameter to be controlled by the interaction between droplet evaporation S = shifted variable in polynomials (Appendix B) rate and heat transfer ahead of the flame. The work reported S = burning velocity or heterogeneous buming in this paper uses this latter approach. velocity A flame propagation model has been developed which SMD = Sauter mean diameter uses the general style of Ballal and Lefebvre [61 but is f =characteristic tiefor droolet vawrization inside applicable to both rich and lean mixtures and includes a more the flame front realistic treatment of the augmentation of vapour in the gas T,TF = temperature and flame temperature respectively phase by evaporation and its contribution to flame propagation tl? =turbulence intensity [SI. It is applicable to monosized and polysized droplet s = flame front thickness distributions and accounts for the effects of changes in ab =vaporization (buming) constant pressure, tempera-, droplet diameter, overall equivalence @ =equivalenceratio ratio and fuel type on the heterogeneous buming velocity. P =density The basis of the model is the hypothesis that flame n =mass fraction of fuel vapour, it.. ratio of fuel propagation through heterogeneous mixtures is controlled by vapour to total fuel the total amount of fuel which becomes available in vapour form as droplets pass through the flame front. This is the Subscripts aggregate of vapour initially present in the mixture plus that a =air formed by partial or complete droplet vaporization during d = droplet condition inside the flame front or just flame uassaee. Relations were develoDed to evaluate~~ the after the flame front effx& eqhlence ratio of the gas phase inside the flnnic 6 =representativecondition inside the flame front front. Predicted flame cpdsage srisfs;rorily uirh eff =effective condition inside the flame front due to publiihrd rcsultr 1x1. The model is thcrcforc considcrcJ to be evaporated fuel useful in helping to elucidate the influence of fundamental fuel ev = at droplet evaporating temperature condition propenies on combustion in reciprocating engines, gas 24-2

turbines and other devices employing fuel sprays. A summary of the details of the model is included as In its initial form, the model, like other existing flame Appendix A. Figure 2 shows a comparison between the propagation models, was limited to single component fuels or predictions of the model for iso-octane, a single component at best, rather unrealistic binary mixtures. On the other hand. fuel, and measured flame speeds. The agreement is real fuels are composed of hundreds of components. This considered to be satisfactory. The uends predicted by the makes their volatility behaviour quite complex. Unlike a model have also shown to be correct under a wide range of single component fuel, the vapour pressure of a mixture otherconditions. [8] depends not only upon its temperature but also lipon how much has already evalmrated. For example, the vapour pressure of JP-4 decreases by an order of magnitude as it evaporates at constant temperature. Light volatile components produce high vapour pressures initially when most of the fuel is still liquid, but less volatile compounds remain behind, and the vapour pressure drops continuously as evaporation nrocePrl< Rw'cnt uork by tu0 of thc prcW1t authors [Y,IO,I1.121 h3s rejultd in thedevelopment of siniplc hathematical models for multicomponent hydrocarbon fuels. These allow real fuel volatility variations to be predicted with reasonable accuracy and the computahond procedures are simple enough to be included as subcomponents of larger analyses. m The aim of the work presented in this paper was to incorporate the multicomponent fuel equations into the heterogeneous flame propagation computatlonal procedures, then use the refined model to investigate the effect that these real fuel characteristics would have on ?ame propagation compared to single component idealizahons. Of particular interest aethe effects of real fuel propnties on combustion behaviour during cold stating and idling in a gas turbine. 0 0.2 0.4 0.6 0.8 i HETEROGENEOUS FLAME PROPAGATION VAFOUR FRACTION 0 MODEL Consider a reference frame fixed with respect to a Figure 2. Conipanson of Predicted and Mc~urrdLaminar plane flame of thickness 6 into which a heterogeneous mixture Flame Sped for Iso.Octane SprJy5181 flows at velocity S, as shown in Figure 1. The heterogeneous buming velocity is assumed to be equal to that of a MULTICONIPONENT FUEL VOLATILITY MODEL homogeneous mixture of equivalence mho equal to that of the The simplest functional relationship for the vapour total gaseous mixture available for chemical reaction inside the pressure of a pure component is the Clausius-Clapeyron flame front. To apply this model, it is therefore necessary to equation: calculate the effective equivalence ratio of the gas phase 4 including both the vapour initially present, plus the fuel w%% P,t = CI exp(-Cz / T) vaporizes from the droplets passing Frough the flame front. where: The heterogeneous buming velocity 1s then equal to the Psat = equilibrium saturation pressure burning velocity of a homogeneous airlfuel mxture having T = absolute temperature this overall equivalence ratio. C1, C2 = constants fora given pure substance The basis of the method outlined in references [9] - [I21 is the assumption that real multicomponent behaviourcan be described by eating C1 and C2 as functions of the extent of vaporization, defined by the vapour fraction a. This latter is the mass fraction of fuel in the form of vapour,

DO =-mass of fuel in the vapour phase total fuel mass (2) 00 S The ha4c vdpour pressure qustion of the multicomponent -.-:..,00 fucl is therrfort written 3( Pmfsat = ft exp(-fz/T) (3)

where f, and f2 are temperature independent functions of C2 Figure 1: Diagram of plane flame front propsgation alone, for any Iiarticular fuel. The technique uses the ASTM distillation data [I31 and specific gravity to predict the The principal assumptions inherent in the calculation variations of fi and f2 as well as those of the mean vapour procedures used in the model are as follows: phase molecular weight M,,f and enthalpy of vaporization LW with the vapour fractionh. The relationship between a) Transient droplet heat-up is not included - the quasi-steady each of these four quantities and the vapour fraction is treament of the well-known "dz law" is used; expressed by a polynomial equation. Appendix B gives funher details. b) individual droplet envelope flames or relay mnsfer are not Figure 3 shows a comparison between vapour explicitly included, in effect, all mixing and non-unifonn pressures predicted by the model and those measured flame spread effects are lumped by the averaging inherent experimentally for JF-4. The results are satisfactory; the in the concept of effective equivalence ratio; volatility model is therefore considered adequate for use in analysing real fuel behaviour in combustion systems c) buoyancy and droplet acceleration through the flame mne In this study, the turbulent burning velocity was deter- are neglected; mined for a wide variety of conditions using different fuels.

Results for.~ the~ ~ case ~~ ~ of ~ verv rich cold starting conditions in a ~ ~ ~~~~~ ~ ~ ~~ d) turbulence effects are accounted for using empirically spdk ignition engine havsbcen rcpo&d &uherr- 114.. In derived relations. this pspcr, thc. modcl is used to study Luxlitions relevant to cold starting of turhincs. The fuels stidied are JP-4. JET AI, and two single component reference fuels, n-nonane and 24-3 n-decane. The coefficients used for the volatility model for approximations, dmplet intemal mixing was assumed to be these fuels are included in Appendix B. sufficientto maintain the liquid composition uniform throughout. Thus vapour pressure at the surface drops in the 2o E manner shown in Figure 3 as evaporation proceeds. t Figure 5 compares JF-4 and n-nonane evaporating at an ambient temperature of 25"C, whereas Figure 6 shows the same fuels when buming surrounded by the spherical flame shell of the simple classical model. The results have been plotted as the ratio of diameter squared so as to better illustrate the multicomponent effects. The slope of this curve is the su called evaporation or combustion constant of the d2 law. Therefore, any single component fuel would give a straight line. 1

0.8

~ 'I

0.6

Nn VAPOUR FRACTION R 0.4

Figure 3. Measured and Predicted Vapour Pressure of JF 4 JP 4 0.2 GENERAL INFLUENCE OF MULTICOMPONENT FUELS Annex C gives details of the conditions assumed to be representative for the calculationspresented in this paper. Figure 4 shows the flame propagation speed calculated for conditions typical of idle conditions in the combustor. These (S) are the same as given in Appendix C except that pressure ratio at idle is taken to be 2. In order to compare combustion Figure 5 Evaporation Histories of JF 4 and Nonane Droplets behaviour independent of prevaporization, these results (Initial Diameter = 100 p,Ambient Tempera- = assume a fmed quantity of vapour (0 or 25%) for all fuels. 25°C) 5.5 1

5 r 0.8 9 4.5 0.6 i4 "$ Nn

E! 3.5 0.4 \

0.2 ,,,,,, ,,,,,,,, nome+ 2 0 250 300 350 400 450 500 550 2 4 6 8 10 -TEMPERATURE (K) TIME (ms) Figure 4. The Influence of Fuel Type and Prevaporization on Figure 6 Evaporation Histories of Buming JF 4 and Nonane Flame Speed for Fixed Reactant Conditions. Dropleu (Initial Droplet Diameter = 100 wm) The effect of temperature on flame propagation is clear in Figure 4. Flame speed increases some 60% over the Figure 5 shows that the initial JF-4 evaporation rate is temperature range shown. Likewise, the extent of faster than that of nonane and its final rate is slower. This prevaporization is very important as seen by comparing flame reflects its high vapour pressure when vapour fraction is low speeds with no pre-existing vapour (Q+) and those when and the decreasing vapour pressure as evaporation proceeds. 25% of the fuel is vaporized before entering the flame zone. The difference at 25OC is considerable. On the other hand, However, there isvery litde difference between the fuels at Figure 6 shows that under combustion conditions, when any given temperature and value of Q. This is perhaps surrounding temperamre is very high, the differences are surprising since the differences in volatility characteristics virtually negligible. Less than complete internal mixing might be expected to lead to significantly different combustion within the droplet would decrease the multicomponent effect behaviour. The reasons for this become easier to understand still further. This explains why flame propagation is but little if droplet evaporation and combustion are considered more influenced by variable volatility compared to a representative closely. Using the classical quai-steady droplet equation pure fuel of similar average volatility. On the other hand, [15] for a single 100 pm droplet in an infinite stagnant vaporization prior to combustion would be significantly medium, computations were performed to show the effect of influenced by the volatility characteristics of real variable fuel volatility. In addition to the usual multicomponent fuels. As seen in Figure 4, prevaporization 24-4 exem a very significant influence on flame propagation. It is here rather than in the combustion itself that the effects are important. The conclusion is thus that it is in the conditions leading up to combustion that differences in fuel 350 characteristics are signifcant for middle distillates such as the fuels studied here. The similarity of the flame speeds for the - different fuels seen in Figure 4 is thus due to the fact that 2 3W prevaporization was taken to be the same for all fuels. € RESULTS FOR GAS TURBINE STARTING CONDITIONS i Appendix C lists the conditions taken to be representative of the combustor under cold starting 9sm conditions. Droplets are injected at an initial diameter, 150 partially evaporate during a short period before entering the combustion wne and then bum at a rate determined by, ;il among other things, the reduced droplet size and the vapour 1W present due to prevaporization. The model predicts flame speed. The question is then M how to relate this to the starting performance of a real engine. The models developed here could be used as subcomponents Y" of larger cold starting models of v+ng degrees of 0.5 1 1.5 2 2.5 3 3.5 4 complexity. For purposes of comparison in this paper, a simple and accepted mans of relating fundamental flame PRESSURE RATIO speed to some recognized operational parameter was needed. For this puqmse, the blow-out velocity was selected because Figure 8 Effect of Fuel Type and Pressure Ratio on Blow-Out there is a straightforward way of relating it to flame speed Velocity (Initial Diameter = 1M) pn, Ambient without entraining other moot considerations which would Temperature = 2oOC) cloud the issue. The relation of Ballal and Lefebvre [161 was Used: In addition to the obvious major impact of pressure ratio (hence cranking speed) on combustion, there are several (4) other notewonhy observations which can be made from this figure. The fmt striking feature is how different the single component fuels khave compared to the real blends. It is In order to ensure that the model gave physically clear that, under some conditions at least, the use of pure realistic predictions of blow-out velocity, comparisons with compounds to represent JF-4 or Jet-A1 leads to very published experimental data were made. Figure 7 shows a unrepresentative results. The disparity can be attributed to the comparison between the measured blow-out velocity of different prevaporization occurring with the fuels during the nonane as presented by DeZubay [17] and the predictions of flight of the droplets from injector to flame wne. The poor the model. These values are for complete prevaporization, simulation of the real fuels by pure compounds of similar and are seen to agree satisfactorily. No data were found in the overall properties is in marked conat to the earlier open literature to permit a direct comparison for a observation that, all other conditions heterogeneous case, but the present model compares prevauorization king equal, flame propagation is virtually favourably to the theoretical prediction of Ballal and Lefebvre identical for all four fuels. [16]. In fact, the model presented here offers an advantage A second observation from Figure 8 is that there is a over that of Ballal and Lefebm in that it correctly predicts the maximum blow-out velocity occurring at some pressure ratio. well-known observation that droplets below a c& size Futher increases in pressure ratio (and thus in chamber temp- behave as if the fuel were completely evaporated. Having erature) reduce blow-out speed. This is due to the rich overall validated the model to the extent possible, it was then used to mixture (equivalence ratio of 3.3) and relatively small droplet investigate the relative im"nce of various parameters. size (100 pm) used in these calculations. For such conditions, excessive pressure rise results in a chamber temperature so high that evaporation to local values well beyond stoichiometric cccurs and the flame is slowed or extinguished altogether. The practical likelihood of such an 102 wa Occurrence seems small under cold starling conditions, but designers might need to take the possibility into account when systems designed for good cold starting are stand at warm ambient temperatures with high- crankhg-. speeds and better atomization. A final observation from Figure 8 is that the differences between JF-4 and JET AI are insimificant in the range of pressure mios up to the p3ks The& I3tter ire slightly different and occur at somen hat different pressure 0 r.itios, but arc again close enough that differences are probably not of prpractical significance. This apparent mdifierence to the volatility advantage of the wide-cut JP-4 A; cm probably be atmbuted to the combination of favourable . . spray size. very rich mixture and high ambient tempcrdtun 0,;. 0,;. " " " " 1' " " " " "1.5 ' " ,I ,d 2 (20'C) used in this particular calculation. Under cold stming FUEL-AIR EQWVALENCE RATTO conditions, coarser sprays and lower volatility would accenrunre rhe differences ktween widecut md high flash Figure 7 Comparison of Blow-Out Velocity Predictions with point fuels, leading to significantly bener vapiniwrion with Data ofDeZubay [171 the latter. Figure 9 represents similar conditions to thoae of Figure 8. cxcept that droplet size is assumed to be 350 pm Figure 8 shows the blow-out velocity for the assumed instead of 100 pn and ambient tempcrdrure is reduced to cold starting conditions listed in Appendix C. The effect of a -40°C. It again shows hat the pure fuels are poor represen- higher pressure ratio is to increase combustion chamber rstivcs for he blends: however. this ume, there are no peaks temperature, with concomitant effects upon evaporation and within this range cfprcssure ratio. For rhese less favourable flame speed conditions, the JP-4 shows its better volatility; for a given 24-5

7r is quite substantial. Figure 12 shows the imporlance of the droplet size distribution on blow-out velocity. The spray having the greatest range of sizes, and hence many small droplets, shows the highest blow-out velocity at low temperatures. By companson, a monosized spray has greatly inferior performance. The effect again, is due to the change in prevaporization. A relatively small mass fraction of small droplets can markedly improve evaporation rate. E .z J 110 !2 $$4 1W

2 1 1.5 2 2.5 3 3.5 $4 90 PRESSURE RATIO 2@ 80 Figure 9 Effect of Fuel Type and Pressure Ratio on Blow-Out Velocity (Initial Diameter = 350 pm, Ambient Temperature = -40°C) 70 blow-out velocity (Le. a given combustion performance) the 06 pressure ratio must be higher for the lower volatility JET A. @I This would correspond to somewhat poorer cold and statling 4 -30 -20 -10 0 10 20 30 40 iding perfomwnce in the case of 3 real engine. In view of the unruit3biltty of the single componcnt fuels -AMBIENT IEMPERA- eC) to 3dqumly represcnt thc rul fuel blends when vapiriwtion IS involved,the rcnuining discussion will concenrntc on the two Figure 11 Effect of Ambient Temperature on Optimum Droplet res1 fuel,. kigure 10 shows thc effca of temperamre and nicm Size droplet si2 (ShlD) on ,.ombustton performance. The pmicullr conditions uxd in the calsuladon agin lead to cwes which show w3k combustion performance at some optimum mean droplet si7e. Lower amhient tcmpzratun: decreases blow-out IZ0 E vclccity cubstdntidly, and also quires smsller droplets to artlin best results. At a mcan diameter of 125 dm. blow-out velocity ai 2WC is some higher lhan at 0°C. In reality, dn)plet si22 would be expected to become comer at lower tempcraturcs due to increved liquid viscosity, 3ggravating he diipsrily sUU funhcr.

,-. 80 3

g60

Y -TAMB=-4CK s 0 9 40 -- -TAME3 = +2C€ 4 -30 -20 -10 0 10 20 30 40 5m '. \ AMBIENT TEMPERA- fC) 20 ttgure 12 Effect of Droplet S17e Diimbution on let AI . ConlbJsuon (S.Mfl=lW CtnJ I I, CONCLUSIONS 0 110 160 210 2M) 310 360 1. The flame propagation model with multicomponent fuel SALTER MEAN DIAMETER OF DROPLETS @m) properties accounted for can shed useful light on com- Figure 10 Effect of Temperature and Droplet Sue on let A1 bustion behaviour. A simple example has ken Combustion presented here; but the multicomponent fuel model and the flame propagation model could also be utilized Figure 11 shows the change in optirrnun droplet size as individually or in combination as components of more ambient temperature is varied. The advantage of IF-4 at low complex gas turbine combustion analyses. For temperatures is clear in that droplets need not be as small as for instance, they could be used to help evaluate the let Al. However, the difference is not great for the rather short evaporation occurring from individual droplets and to evaporation time arbitrarily selected for these calculations. The calculate the combustion rate at diffemt locations within two cwes cross at higher temperature because JP-4 droplets evaporate tca quickly for the rich local equivalence ratio used for the combustor. the calculations. For either fuel, the effect of ambient temperature 24-6

2. Single component idealizations are quite satisfactory for 15. Kanury, A.M., Introduction to Combustion Phenomena, many combustion and flame propagation calculations; Gordon and Breach, New York. 1977. however, when prevaporization is imponaut, such as 16. Ballal, D.R. and Lefebvre, A.H., "Some Fundamental under cold starting conditions, real fuel behaviour is Asmts of Flame Stabilization." Fifth Internattonal badly represented by single component approximations. Syhposiwn on Airbreathing Engiks, Feb. 1981. Multicomponent volatility behaviour can critically 17. DeZubay, E.A., "Characteristics of Disk-Contolled influencehe conditions ieading up to combustion. Flame," Aero. Digest, Vo1.61, Nov.V, Ju1.1950, pp.94-

3. The extent of vrevawrization can exert~~ a dominant~ effect ~~~~ 96,102-104. on flm pro&3tiOn: therefore 3ccmte re?resentauon 18. Williams, F.A., Combustion Theory. 2nd ed., of this process is essential 10 any modrl. BenjamidCummings, Menlo Park, Califomia, 1985. Multicomponent effects must be accounted for. 19. Hubbard, G.L., Denny, V.E. and Mills, A.F., "Droplet 4. For the particular conditions studied here, the Evaporation: Effects of Transients and Variable differences between P-4 and Jet A1 are relatively Properties," 1ntJ.Hear Mass Trader. Vo1.18, 1975, slizht. However. detailed modelline of articular cases 0". inw-innx. wddbe expected to show much &t& impact of the 20. Gordon, S. and McBride, B.J., "Computer Program for more volatile PA. The equations presented here fox Calculation of Complex Chemical Equilibrium vapourpressure can readily be used for such studies. Compositions, Rocket Performance, Incident and 5. The effect of mean droplet size, size disuibution, Reflected Shocks, and Chapman-Jouget Detonations," Wngspeed (via pressure ratio), ambient temperature NASA, SP-273, March 1976. and fuel type have been studied for an'arbitrary set of 21. Fenn, J.B. and Calcote, H.F., "Activation Energies in conditions intended to represent cold starting in a gas High Temperature Combustion," Fourth Symp. (Int.) turbine. Particular conclusions relevant to this case have Combust., Williams and Wilkins, Baltimore, 1953, been drawn and may offer some qualitative guidance pp.231-239. However, actual behaviour is very sensitive to the 22. Gauthier, J.E.D., Flame Propagation in Mixtures of Fuel values of parameters assumed (eg. turbulence intensity). Droplets, Fuel Vupowand Air, thesis submitted to the These results should therefore be considered as merely Royal Military College of Canada, Kingston, Ontario an illusuation of the type of computations which can be 1987. performed with the flame propagation and volatility 23. Raju, M.S. and Sirignano, W.A., "Multicomponent models. Spray Computations in a Modified Centerbody Combustor," J.Prop.and Power, Vo1.6, N0.2, March- REFERENCES --r---,~nrii.~-1990. - -. 24. Lefebvre, A.H., Gas Turbine Combustion, McGraw- 1. Agganval, S.K. and Sirignano, W.A., "Numerical Hill, New York, 1983.

Modeline of One-Dimensional Enclosed~ ~~~~ Homoeeneous ~~~~~~ -~~~~-~~- 25. Andrews, G.E., Bradley, D. and Lwakabamba, S.B., and HetGogeneous Deflagrations," Comput.Fluids, "Turbulence and Turbulent Flame Propagation - A Vol.12, N0.2, 1984, pp.145-158. Critical A~vraisal."Combust.Flame. Vo1.24. 1975. 2. Swithenbank, J., Turan, A., Felton, P.G., "Three- p p .2 85 - 3 I% Dimensional TwWPhase Mathematical Modelling of Gas Turbine Combustors," Cas Turbine Combustor Design Problems, Lefebvre, A.H., ed., Hemisphere, APPENDIX A Washington, D.C.,, 1980, pp.249-314. 3. Polymeropoulous, C.E., "Flame Propagation in Aerosols FLAME PROPAGATION MODEL

of~~ Fuel ~ ~~~ Dmnlets.Aiel ~ ~~ ~-~ Vamr.~r-~ and Air." ComburrSci Technol., Vol 9, 1974. pp.197-207. A.1 Overall Model 4. Polymeropoulos, C.E.. "Flame Propagation in Acrosols Since the amount of fuel vapour in the fresh mixture is of Fuel Droplels , Fuel Vapor 2nd A& ' known, the main (:askis to find a procedure to calculate the Combust.Sci.Teg:hnol., Vo1.4 amount of fuel which vaporizes from the liquid droplets 5. Mizutani, Y. and Ogasawm, M., "MarFlame passing through the flame front. The time t for any mixture to Propagation in Droplet Suspension of Liquid Fuel," pass through the flame front is given by the ratio of the flame 1ntJ.Hear Mass Trader, Vo1.8, 1965, pp.921-935 front thickness to the buming velocity:

, ~~~~~~~~~~ 6. Ballal. D.R. and Lefebvre. A.H.. "Flame Promeationc in Heterugmeous hlixtu~&of Fuel Droplets. Fuel Vapor 6 and Air," Eighteenth Synp. (Inr., C".sr., The t=s Combustion Institute, Pittsburgh. 1981, pp.321-328. 7. Myers, G.D. and Lefebvre, A.H., "Flame Propagation in As stated above, the droplet diameterDd, while passing Heterogeneous Mixtures of Fuel Drr?s and Air," Combusr.Flme, Vo1.66, 1986, pp.lY3-210. through the flame front, decreases according to the "&law" 8. Gauthier. J.E.D. and Bardon. M.F.. "Laminx Flame Propaganon In Mmtures of Fuel Droplets. Fuel Vapour and Air,'Jlnstnf Energy. Vol LXII, No.4: I, Jun 1989, pp.83-88. Under steady conditions, the rate of heat uansfer to the 9. Bardon, M.F. and Rao, V.K., "Calculation of Gasoline reactants from the flame is balanced by the rate of hat Volatility," JJnst.of Energy, Vo1.57, 1984, pp.343-348. liberation in the flame [15,18]. From this, an expression can 10. Rao, V.K. and Bardon, M.F., "Estimating tho Molecular be found [SI to derive the effective equivalence ratio available Weight of Petroleum Fractions," Ind.Eng.Chem., in the flame front. This is simply the sum of original fuel Process DesDev., Vo1.27, N0.2, April 1985. vapour present plus that produced by the partial evaporation 11. Bardon, M.F., Rao, V.K., Vaivads, R. and Evans, occurring as droplets pass through the flame front. M.J.B.. "Measured and predicted Effect of the Extent of the Evaporation on Gasoline Vapour Pressure," JJnst.of Energy, Vol.65, No.441, December 1986. 12. Bardon, M.F., Nicks, G.W., Rao, V.K. and Vaivads, R., "A Vapour Pressure Model for MethanoVGasoline M85 Blends," SAE Paper #870366 presented at the International Congress and Exhibition, Demit, Feb.23- In accordance with the model hypothesis, the heterogeneous 27, 1987. buming velocity is equal to that of a homogeneous mixture of 13. "Distillation of gasoline, naptha, kerosine and similar equivalence ratio equal to $&,# Note that the laminar petroleum products," Test D86, American Society for heterogeneous buming velocity is a function of but Testing and Materials, Philadelphia, 1978. S, $gef 14. Bardon, M.F., Gauthier, J.E.D. and Rao, V.K., "Flame that $&,&s itself a function of SLvia equation (A-3). This Propagation through Sprays of Multi-Component Fuel," open form solution means that the value of (hence of JJnst.Energy, Vol.LXIII, No.405, Jun 1990, pp.53-60. SL must be found using an iterative method. 24-7

Gas phase properties are calculated using conventional APPENDIX B mixture procedures and component properties. The calculation of specific heat as functions of temperature is done FUEL VOLATILITY MODEL with third degree polynomials. Following the approach used by other workers [4,61, the representative gas properties In the present program, the functionsfl and fz in inside the flame front are assumed to be those of air at the equation (3) were derived in accordance with an improved arithmetic mean of the reactant and flame temperatures. The version of the procedure described in references [91-[12]. mean speciiic heat of the gas phase in the reactants is also The functionsf,, f2 and M, and AH are expressed by approximated by the value at the arithmetic mean of the shifted Legendre polynomiag: reactants and flame temperature. The In mle of Hubbard and Mills [19] is used to fl = alpl + %P2 +.- + a,P, calculate the appropriate reference temperature for single droplet evaporation. The flame temperature is taken as the adiabatic flame temperature calculated using the NASA fi blPl + bzPz + ... +bat', (B-1) Chemical Equilibrium program [20]. Mmfv = clPl + c2Pz + .... A.2 Burning Constant Ab AH = dlPl + dZP2 + ... The buming constant for use in the "& law" is found from The degrees of the least-squarespolynomials required to adequately approximate the functions fl. fz, Mm& and AH vary from case to case; they are determined largely by the shape of the ASTM distillation curve of the fuel. A curve whose slope changes considerably as the distillation where the droplet uansfer number is approximated by: progresses, results in functions that require a larger number of terms in the polynomial approximations, particularly the two functionsfl, and f2 In the present case, a fourth-order B= cpoev(L~- Tw~) polynomial was used to represent the functions for. The first (A-5) five Legendre polynomials are: fk + C~ATW-Tj7)

The droplet surface temperature is assumed equal to the P, = 1 boiling temperature of the liquid at the instantaneous value of P, =s Q. The effect of turbulence on droplet evaporation rate iS P, = (39 - 1)/2 accounted for using a conventionalcorrelation of the form P4 = (SS3 - 3S)/2 (B-2) a, a, = a,( 1 + 0.3~~8.5) (A-6) P5 = (35S4 - 30,f +3)/8 where the Reynolds nurdber is badon droplet diameter and where S = 1 - 2Q turbulence intensity. Reference [SI provides further details as well as a description of how droplet size distribution is Tables B-1 and B-2 give the values of the coefficientsused accounted for. form-4 and Jet AI respectively. A.3 Homogeneous Burning Velocity Table B-1: JF-4 Coefficients for use with Equation (B-I) A.3.1 Laminar The calculation of the laminar homogeneous buming velocity Sh is done using the modified Semenov equation [211:

Values in this table give pressure in kPa and enthalpy in J/mol. The value of K is calculated using the known conditions of maximum homogeneous laminar buming velocity at nmal Table B-2: Jet AI Coefficients for use with Equation (B-I) ambient conditions. The effects of temperature and pressure are accounted for with power law expressions using the method proposed by Gauthier and Bardon [XI to evaluate the exponents. A.3.2 Turbulent Reference [22] describes the technique used to compute turbulent buming velocity given the laminarflame speed. The equation used was of the form: For the single component fuels, the Clausius-Clapeyron Equation (1) was used. Table B-3: Coefficients for Equation (1) for Pure Fuels

This latter equation is based on empirical data for turbulent combustion under conditions similar to those in gas turbines [22]. n-nonane 1.573E+OX 5166.5 24-8

Figures B.l to B.3 compare Jet AI and n-decane at three temperatures. It is clear that the single component fuel does not reflect the hue vapour pressure khaviour at any given temperature; furthermore, it does not even icpresent a reasonable mean value except over a very limited temperature 2.52 range. E

0.02 let AI

t O.OI5 0.5 w3 y1

8 0.01 0 0 0.2 0.4 0.6 0.8 1 8 VAFQLR FRACTION R > 0.ws Figure B3 Vapour Pressure of Jet AI and Decane at +WC

0 APPENDIX C VmURFRACIlON R Data Used to Represent Typical Cold Starting Figure B1 Vapour Pressure of Jet AI and Decane at -2PC Coinditions in a Gas Turbine Ambient Pressure: 101 kPa Ambient Temperature: 2WC except where explicitly varied. Ressure ratio at sm-up: 1.2 Sauter Mean Diameter of Drop lets duced by the fuel injector: 100 pm [23,24] Tdulence intensity: 35Ocds [251 Local fuel air equivalence ratio in flame zone: 3.3 [23] Blow-out velccity coefficients C, = 1.33 [I61 (equation (4)): D, = 0.01 m [I61

Time available foi droplet evapon- uon before reaching thc flame: 0.007 s 1231 The coefficients C,-C, describing the droplet size distribution are as described by Lefebvre 1241: C1= MSDISMD C, = MDDISMD C3 = MMDISMD Default values used 0 Ud = 0.31 C2 = 0.21 = 0.46 [61 0 0.2 0.4 0.6 0.8 r Cl C3 VAFQUR FRACTION R Furcompariwn piirposs, a second polydisperx \pray is also used in Figure 12, referred to as "Polydisperse 6''using the Figure B2 Vapour Pressure of Jet AI and Decane at +20°C following values: C, = 0.84 C2 = 0.73 C3 = 0.89 [23] 24-9 Discussion Author: Our work so far has been exclusively on the case of droplets which are well mixed internally. This is due to the use of equilibrium vapor pressure, and implies that droplet 1. C. Moses, Southwest Research Institute evaporation and diffusion outwards is slow with respect to There are two major theories in the mechanism of droplet internal diffusion. This is one of the two limiting cases, the evaporation with multicomponent fuels. At one extreme, the other being the case with no internal mixing or onion-skin interior of the drop is well mixed. This allows for light evaporation. The latter is more nearly the case for heavy components to be continuously brought to the surface, and viscous fuels and very short residence times. The case in the thus the fuel effectively destills as you have discussed. At the cold combustor is probably closer to the equilibrium limit other extreme, there is no mixing and the droplet evaporates but we have not verified it experimentally. For the present much as an onion skin with all components, light and heavy, circumstances many of the conclusions would in any case be effectively evaporating at the same time, Have you unchanged. The relatively similar performance of JP4 and investigated these two possibilities to see which provides the JET AI for example, would be unaffected by which limiting better agreement? model we used in the calculations presented here.

25-1

THE DEVELOPMENT OF A COMPUTATIONAL MODEL TO PREDICT LOW TEMPERATURE FUEL FLOW PHENOMENA

R. A. Kamin and C. J. Nowack Naval Air Propulaion Center Trenton, NJ 08628 USA

B. A. Olmstead Boring Military Airplanes Seattle. Wa 88124 USA

SUMMARY

Fuel availability studies indicated that the relaxation of the F-44 freeze point specification could greatly increaae the yield of F-44 per barrel of crude. A thorough analysis was initiated to ensure that the higher freeze paint fuel would not form solid wax precipitates during low temperature Operations that could impact airCFaft mission performance. In order to evaluate the effects of a potential change in the freeze point specification OV~Fthe entire inventory of United States Naval aircFaft, a general three dimensional computational fluid dynamics code, PHOENICS 84. was modified for use. Inputs into the code include tank geometry (cylindrical. rectangular OF body fitted), mission profile (outside air temperatures or altitude, airspeed and time on station), and fuel properties (specific heat, thermal conductivity, viscosity. freeze paint, etc.). Outputs from the model include fuel cooldown and holdup (unpumpable frozen fuel1 as a function of time and position in the tank. The accuracy of the code was verified by experimental data obtained during flight and ~imulato~testing of instrumented tanka.

NOMENCLATURE

Area Numerical Constant Empirical Turbulence Model Coefficient Empirical Turbulence Model Coefficient Empirical Turbulence Model Coefficient Specific Heat Empirical Turbulence Model Coefficient Accelepation of Gravity Sensible Heat Enthalpy Subscript= Denoting Cartesian Coordinate Dire~tions Turbulent Kinetic Energy Permeability Total Latent Heat Pr e e 8uFe Numerical Constant Turbulent Reynolds Number (K2p/pcl Buoyancy Source Term Latent Heat Source Term Darcy Source Term Time Temperature Velocity Vector Cartesian Space Velocity Velocity In x,y,zDirections Cartesian Space Coordinates Coefficient of Thermal Expaneion Rate of Turbulent Kinetic Enepgy Porosity Thermal Conductivity Density Turbulent Prandtl Number Prandtl-Schmidt Number for Turbulent Dissipation Rate Prandtl-Schmidt Number for Turbulent Xinetic Energy Dynamic Viscosity Turbulent Vi BCOS i ty (C p K2 / c ) Kinematic Viscosity Angular Coordinate 25-2

INTRODUCTION

F-44, due to its unique requirements of a high flash point (60'C/140'F) and low freeze point [-46'C/-51*F), ia a specialty product which accounts for only a small fraction, less than 1%, of total refinery production. Therefore. to make the production of F-44 profitable to the refinery, it ie produced at a premium coet to the United States Navy. In addition, the incpeasing commercial demand for the refinery distillate stream combined with the increased use of heavier. high sulfur, lower yielding crudes CPeatez the potential for ahortfalla in the pFoduction af F-44. The U.S. Navy, in reeponse to these potential fuel shortages, has given a high prio ity to investigating methods of increasing the availability of F-44. Various Studies ','have shown that the two most c~iticalproperties affecting the yield of F-44 per barrel of CFude are the flash point and the freeze point. Since the flash point specification ia set mope ae a shipboard eafrty than an aviation Fequirement and cannot he lowered, the relaxation of the freeze pnint specification is the moat lagical and cost effective method to increasing F-44 yield. By i'elaxlng the current -46'C(-51"Fl freeze paint specification to the commercial Jet A specification of -40"C(-4OoF1, it is possible to obtain a 10% to 75% increase2. depending an ,crude ~ource,in the maximum theoretical yield of F-44 per barpel of crude.

However, before the specification change can be made, its potential impact on aircraft performance must be evaluated. This investigation is necessary because of the concern that the higher freeze point fuel may form solid wax pvecipitates during extreme low tempepature operations. These precipitates may cause plugging of filters or blockage of fuel tank transfer lines that could severely restrict flow to the engine.

The ideal method for assessing the impact of higher freeze point fuels would be to meaeure the fuel temperatures within the tank af each aircraft during extreme law tempevatupe operations. This, however, would have required an extensive number of tests that would be extremely costly rind time consuming. Therefore. the only practical cast effective method by which to approach this problem was to develop a mathematical computer model, verified by experimental results, to predict fuel temperature and holdup (portion of unusable fuel. due to feeezing) during a miaaion.

MODEL DEVELOPmNT

The turbulent, unsteady boundary conditionE and the extremely high Rayleigh numbers (10'5 encountered during a flight make the modeling of fuel cooldam and holdup witpin a tank a difficult and complex analytical problem. Although a one dimensional model has been developed to predict fuel temperatures in thin wing tanka. many aircraft fuel tanke "met be modeled in two OF three dimeneions in order ta give accurate results. Prior to initiating the development of the two and three dimensional model, a litefature araFch was performed tu determine if any existing multidimensional computational fluid dynamics(CF111 codes were available that could be mudif ied to handle this problem. After an extensive review of CFD cadee. the eenr~alpurpoue PHOENICS 84 (Parabolic, Hyperbolic OF Elliptical Numerical-Integration Code Serleal code waa determined as the best available code to solve this problem. The PHOENICS 84 code is a centralized, versatile system cs.pable of performing a large number of fluid-flow. heat transfer and chemical reaction calculations simultaneously *. The code consists of an inaccessible proprietary core program. called Earth, which contains the theoretical, problem solving algopithms. The Earth code obtains the Specific information necessary to define and solve each problem from two user accessible programs, called Satellite and Ground respectively.

An extensive effort was initiated to develop the specific use? input routines for Satellite and Ground that would be necessary to allow the source code, Earth, to both accurately and cost effectively predict fuel temperature and holdup. Major areas of development included 1. Optimum grid selection for both rectangular and cylindrical geometries. 2. Turbulence modeling. 3. Phase change modeling, and 4. Expert system development .

Qrid Selection

The optimum computational gPids were developed based on the criteria that they ppoduce well-converged. stable and accurate solutions at a= low a computational cost as poeaible. For bath rectangular and cylindrical geometriee, the grids developed were finely divided at the walls and the bottom of the tank whei-e the steepest temperature gradients OCCUF and became increasingly coarBer he they approached the center of the tank. The computational costs were further reduced for the twa dimensional simulations by assuming symmetry within the tank, i.e. at the vertical centerline av,/Je = 0 and JTlJe = 0. and only modeling half the tank. This assumption. which Cut the computational time in half, was verified by experimental testing to be accurate5. Finally, a grid refinement sensitivity study was performed to determine the in0 t coet effective grids, Figure 1, which would give accurate, well-converged solutionsB .

Turbulence Model

The convective flow in aipcraft fuel tanks. caused by the temperature gradient between the bulk fuel and outside air temperature. has been calculated to be in the transitional 01. turbulent regime. Rayleigh number >lo8 The! failure to account for the increased mixing of the fuel due to these convective flows could cause significant 25-3

error in the calculation of the local heat transfer coefficients. This error could reault in the prediction of fuel cooling and holdup rates which WeFe low. The initial approach to account far these turbulence effects wae tt calculate the local turbulent viscosity by using the K-E 2-equation turbulence model . The K-e model was choaen because of its demonatrated succes~ n calculating two dimenaional steady state heat transfer in turbulent buoyant flows7’$. The equations were modified to include buoyancy terms to account for turbulent kinetic energy generation and destruotion due to buoyancyQ. The resulting K-c equations formed were

The coefficients in equations (1) and (2) ape set to constant values in the high Reynolds number turbulence model. However, to account for the low Reynolds number occurring in the tupbulent natural convection, two of the coefficients, C2 and Cc, were set instead to functions of the turbulent Reynolds number, Ret. The coefficientd’ used in the modified K-E model were C! C! = O.OOexp[-2.5/(1+ Ret)] (3) 50 C] = 1.44

”,= 1.92 L1.0 - 0.3exp(-Re?)l (4) $= 1.44

OK’ 1.0

mc = 1.3 . These modifications were found to further enhance the accupacy of the model. Fuel Freezing/Thawing Model A fixed grid porous cell model was Selected to simulate the freezing/thawing of the fuel within the tank’‘. This model assumes that the crystalline-liquid, two phase region will behave as a porous medium and that Darcy’s law 2.u - - grad P P (5) will govern the flow. Thepefore. when the fuel in a specific cell is entirely in the liquid phase, K’ = a, and when the fuel is completely in the solid phase, K’ = 0.

The Darcy source terms, S,: Syand S,, used to simulate the flow behavior in the two phase flow regions are defined as

S,= -Au, 3 = -Av and SI= -Aw where,

A--C(l-A)a/(),3+$) and increases in a nonlinear manner from zero to a large value as the local, solid fraction increases from 0 (liquid) to 1 (Solid). The values of C and E’ were set to 1.6 X 103 and .001 respectively. Theae values, based on previous work , were set to ensure that 1. C would he sufficiently small enough to allow flow in the two phase region which contained only small percentages of solid fractions and 2. The limiting value of A (-C/q’) would be large enough to suppress the velocities in the cells containing large percentages of solid.

The Darcy source terms are added to the right hand side of the momentum equations as follows: u - momentum: 25-4

v - momentum: 2 -+ at div ( (pc)v) - div (p grad V) - ay + S,, +S,

w - momentum:

'-+ '-+ at div ((p:c)w)- div (p grad w)- + S, + S,

whet-e u = ui t vi t wk

In the totally liquid cell. these source terms equal zero and, therefore, make no contribution. AS a greater portion of the cell becomes solid, the size of the source terms increases until it becomes very large and suppresses all the velocities.

Another essential portion of the freezing model fop convection/diffusion heat transfe? phase change ie the energy equation. This. equation includes the latent heat 8ou~ceterm. SL, a8 followe! + aliv (PA)-div K/aIahh at CP - SL

In this model the phase change is accomplished through enthalpy, defined a8 the sum of se'nsible and latent heat=

H-h+AH (11)

The main advantage of this method is that there ape no explicit heat transfer conditions to be satisfied as the phaee change OCCUFB, and the tracking of the phase change is therefore not required. With this method, the source term in the energy equation becomes

SL + grad (pGAH) -. at (12)

For a mixture of a large number of different components, as is found in aviation fuel. AH is a function of tem~erature'~constrained by the limitn at

OLAHLL

where L, the total latent heat, is defined below the freeze point as

L-I '' C, (T)dt (13) t Although this temperatuFe dependence ie nonlinear, a simpiified linear relationship was used as a fiFSt approximation. Therefore, at any temperature. the value of AH has the following physical interpretation :

Liquid fraction = AWL Solid fraction = 1 -AWL .

Additional details an solving the phase change equations using the PHOENICS 84 code are reported in Fefrrence 11.

A 2-D simulation with and without the use of the fuel freezing model was performed to determine the effects of the freezing model on the prediction of fuel temperatures. The Fesults showed that the use of the model had little or no effect on the bulk fuel temperatures predicted. However. the simulation using the fuel freezing model predicted temperatures for the fuel located in the bottom of the tank 1.C to 3'C (2'F to 6'F) lower than the Simulation not using the model. Upon comparison with the expepimental results, Figure 2. the temperatures predicted by the simulation using the fuel freezing modal agreed more closely than those predicted by the simulation that did not use the model''.

The most important piece of data needed to determine whether a higher freeze point aviation fuel can be used is fuel holdup. However, this pFoperty is extremely difficult to model accurately. The difficulty in this prediction ia due to a number of reasons. First, small differences in temperature around the freeze point can result in a large 25-5 difference in fuel holdup. Therefore. if the temperatures predicted are only a few degrees Centigrade off from the actual values, they may have a large effect on the. predicted holdup. Secondly, since less than 10% of the fuel actually solidifies and this wax-like matrix can prevent the flaw of the remaining bulk liquid fuel, hdldup reaulta tend to be somewhat rhndom and may vary irom teat to teat. This variability, shown in duplicate tests to differ by as much as 30%15,cannot currently be theoretically modeled. The fuel freezing model uses an empirical holdup correlation. derived from results obtained using the Shell Thorntan cold flow tester, to determine the fraction of-holdup at any given temperature. The cold flow tester consists of two distinct chambers separated by a poppet valve. One hundred milliliters of fuel is placed in the upper chamber. and the fuel is cooled to the specific test temperature. Upon reaching the specified temperature, the poppet valve is opened, and the liquid fuel is allowed to flow to the lower chamber. The valve is then closed and the amount of fuel remaining in the upper chamber is measured. The percentage of fuel remaining in the upper chamber is the holdup fraotion of the fuel. Each fuel is tested Over a wide range of temperatures below its freeze point and a graph of holdup versus temperature is developed (Figure 3). This information. accessed through the Ground Poutine of the PHOENICS 84 code, is used to predict the percentage of solid and liquid fuel for each grid cell. ExDert System Development Due to the size and the complexity of the programming involved, the development of a user-friendly. menu driven format was necessary to avoid as much potential operator’ error as possible. The subroutines of the model were assembled to operate in a menu driven step-wise format. graphically depicted in Figure 4. After the user selects the type of aircraft and mission to be modeled, the external temperature exposures for the desired mission are determined by either of two methods. The temperature data may be selected by 1. Tracking the mission through an extensive database of recopded atmospheric temperatures obtained from approximately fifteen years of worldwide atmospheric measurements or 2. User defined requirements such a8 thoze meaaured during a test OP defined in literature (e.g. MI1-STD-PlOB, Climatic Extremes For Military Equipment). Next, the geometry of the tank to be modeled is selected. Since optimum grid aizee far both rectangular and cylindrical geometries have been developed and installed in the program, the user simply has to input the dimensions oi the tank and the amount ai fuel desired. The user then selects how the external baundapy conditions should be defined. The two choice8 available are 1. Actually defining the conditione from experimental tank skin measurements or 2. Calculating the temperatures baaed on environmental conditions (e.g. air temperatures and airspeed). The user is then asked to input the specific fuel properties (e.g. Cp , K, 6, IL, p, AH and Shell holdup tester data) for the fuel to be simulated. These properties are mostly temperature sensitive and, therefore, must be defined as a function of temperature. A subroutine was written far the Ground portion of the PHOENICS 84 code to input the property data into the Earth code. Although the Same properties were used for eveFy simulation, the user does have the option to input specific property values.

Finally, the user selects the type of data, e.g. temperature data, holdup data, etc., and location in the tank from which the data is desired. MODEL VERIFICATION

The computational model was verified in both two and three dimensions by a number of experimental tests. Model simulations were run for rectangular and cylindrical geometries using data obtained from both laboratory simulator and flight tests. TWO DIMENSIONAL MODEL Thick Wing Rectangular Tank

The results from a 1981 NASA flight test of an L-1011 aircraft16were used to verify the model. This research aircFaft contained an instrumented vertical thermocouple rake in the center of the approximately 22’ deep outboard tank. For the simulation, the tank was assumed to be full of Jet A and the recorded top and bottom skin temperatures were used a8 the boundary conditions. The predicted temperatures for the bulk fuel located in the CC~~PPof the tsnk were within 1’C to 3’C (Z’F to 6°F) of the meavured values (Figure 5). The predicted values for the fuel located in the bottom of the tank, thermocouple located approximately 1/4‘ from the bottom. Were consistently 2.C to 5°C (4°F to 10‘F) colde? than the experimental values. Since the temperatures for the entire simulation remained above the fuel’s freeze point, no fpozen fuel was predicted. Thin Wine, Rectaneular Tank The thin wing rectangular tank verification was performed utilizing test data recorded in a wing tank ~imulator‘~.The wing tank simulator consisted a1 a 20 gallon rectangular tank instrumented with a vertical thermocouple rake. Jet A fuel having a -46’C(-51’F) freeze point temperature Was Used. Initial fuel temperature for the simulation was -42.Ct-44-F). The top and bottom skin temperatures were used a8 the 25-6

boundary test conditions and held at apppoximately -5OeC(-58'F1 for the entire 280 minute test.

The predicted temperatures fa7 the bulk fuel we~ewithin approximately l'C(2'F) of the experimental results (Figwe 61. However, the predicted value of 30% fuel holdup did not agree with the 70% holdup value that was measured experimentally. Potential reasons for the large difference between the calculated and measured values were discussed in a prior section.

Cy1 indrical GeometrE

The cylindrical geometry model was verified using the experimental results obtained in a NAPC wind tunnel test of a.n instrumented 150 gallon FPU-3A external tank". The tank, Figure 7. contained two rakes of thermocouples. one placed in the front (Station 321 and the other in the Tear (Station 108) of the tank. The tank contained approximately 122.5 gallons of JP-5 leaving approximately 2.5 inches of ullage space at the top. Two separate simulations were performed. The first simulation utilized the measured wall temperatures recorded during the test as the boundary conditions, while the second calculated the tank wall temperatures from the external air temperatures and velocities measured during the teat.

The predicted temperatures for the calculated wall temperature boundary condition case showed excellent agreement (Figure 8) with the experimental results meaaured by the thermocouple located in the center of the tank (bulk fuel). Thr predicted temperatures fop the measured wall temperature boundary condition case, however, were apppaximately 3'C to 5°C (6°F to 10'F) cooler than the bulk fuel. This may he attpibuted to the fact that temperature measurements were taken at only four points on the tank's skin (at 0, 90, 180. 270 degree intervals) and the remaining temperatures along the circumference of the skin were determined by linear interpolation. If the interpolated tempepatures were in error then the boundary conditions used in the simulation would be in error, thepeby causing the difference in the ppedicted bulk fuel temperaturee. However, the measured wall temperature boundary condition simulation predicted fuel temperatures for the fuel located .25 inches from the bottom of the tank in excellent agpeement with the experimental values. while the calculated wall temperature boundary condition case predicted values approximately 2°C to 5°C (4°F to 10'F) above the experimental Fesults I' (Figure 01.

The holdup measured upon dvaining the tank at the completion of the test was 15%. This compared to the predicted values of 11% for the measwed wall temperature boundary condition case and 5% for the calculated wall temperature boundary condition case. Considering the inherent uncertainty in both the measurement and calculation of holdup, the agreement between the calculated value obtained in the simulation using the measured wall temperatures as the boundary condition and the actual holdup value was quite good.

THREE DIMENSIONAL MODEL

Flicht Test

A flight test of the 150 gallon instrumented FPU-3A tank mounted on an A-6 aircraf was used to vevify the model. The 3D model was needed for this simulation because the inclination of the tank during flight caused the formation of three dimensional temperature gradienta which could not be accounted far by two dimensional modeling. The

tank skin temneratures measured durineI the flight~ were used ae the baundarv conditions for the simulation. The tank contained 125 gallons of F-44 and had an initial fuel tempepature of ll'C(52'Fl. The predicted temperatures for the fuel located at the bottom of the tank were approximately 2'C to 5°C (4°F to 10°F) above the measured tempeaatures for the front thermaco ple and 1°C to 3°C (2°F to 3°F) below the measurements taken at the rear of the tankIY (Figure 10). Howeve=, 'the ability of the 3-D model to accurately predict the temperature difference between the front and the rear of the tank demonatpated ita importance in accurately modeling actual fuel behavior occ!urring during low temperature missions.

CONCLUSION

The model was demonstrated to accurately predict. in both two and three dimensions. fuel temperature profiles at temperatures both above and below the fuel's freeze point. This model. verified for both rectangular and cylindrical geometries, is capable of predicting fuel cooling for any anticipated aircraft mission. In addition, the model can also be used, with minor modifications, to predict the heating of the fuel within the tank, making it a valuable tool in investigating fuel behavior during supersonic flight.

Although the accuracy of the fuel holdup prediction was not as good as that of fuel temperature, general approximations were able to be made. This prediction wa8 based on an empirical corpelation derived from Shell Thopnton Cold Flow TesteF data because holdup was shown to be a eomewhat random phenomena containing an inherent degree of variability that could not be accounted for theoretically. Since holdup is such an extremely important piece of infopmation, additional simulator studiee are being 25-7 performed to develop the improved empirical correlations needed to allow the model to make better holdkp calculations.

The model is currently being used to predict the effects of higher freeze point fuels on naval aircraft mission performance. The resulta of this study will be ueed to determine if the F-44 freeze paint specification can be eafely pelaxed to the commercial Jet A specification, thereby increasing the potential yield Of F-44 per baprel of crude. REFERENCES 1. Liebevman. W., and Taylor, W.F., 'Effect of Refinery Variables on the Properties and Composition of JP-5,' RL.ZPE80, June 1980. 2. Varga, O.M., Lieberman. W., and Avella Jr., A.J., 'The Effects of Crude Oil and Processing on JP-5 Composition and Properties'. NAPC-PE-121C. July 1985. 3. McConnell. P.M., Desmarais, L.A.. and Tolle. F.F., 'Heat TransfeF in Airplane Fuel Tanks at Low Temperatures,' ASME paper 83-HT-102. 4. Spalding. D.B., 'A General Purpose Computer Program for Multi-Dimensional One and Two Phase Flow, 'Mathematics and Computers in Simulations, North Holland Press, Vol XXIII, 1981. 5. McConnell, P.M., 'Development and Use of a Fuel Tank Fluids Characteristics Mathematics Model - Phase I,' NAPC-PE-142C. January 1987. 6. Launder, B.E. and Spalding. D.B.. 'The Numerical Computation of Tupbulent Flows.' Computer Methods in Applied Mechanics and Engineering, Vol 3, 1974, pp. 269-289. 7. Markatas, N.C. and Pericleous, K.A., 'Laminar and Turbulent Natural Convection In An Enclosed Cavity.' International Journal of Heat and Mass Transfer, Vol. 27, No. 5, 1984, pp.755-772. 8. Humphrey, J.A.C., Sherman, F.S. and To, W.M., 'Numerical Simulation of Buoyant Turbulent Flow.' SAND85-8180. Aueust 1985. 9. R0di.W.. Turbuience Modela and Their ADDlicationa In Hvdraulica - A State-of-the-AFt, Institut Fur Hydromechanik, 2nd Ed, February 1584, pp. 27-31. 10. Olmetead, B.A. and Kamin, R.A., 'Numerical Analysipl of Turbulent Natural Convection in Airplane Fuel Tanks,' To be published. 11. Vollar, V. and Prakash, P., 'A Fixed Orid Numerical Methodology For Phase Change Problem Involving- Mushy Retian. and Convection In the Melt'. PDR/CHAM/NAO, July 1986. 12. Domanue. H.M. and Sha, W.T., 'Solidification with Natural Convection,' April 1588 13. Mehta. H.K., and Armstrong, R.S., 'Detailed Studies of Aviation Fuel Flowability, NASA CR-174938, June 1985. 14. McConnell, P.M., 'Development and Use of a Fuel Tanks Fluids Charactesiatica Mathematics Model, NAPC-PE-177C. October 1988. 15. Desmarais, L.A., and Tolle, F.F., 'Fuel Freeze Point Investigations,' AFWAL-TR-84-2405, July 1584. 16. Svehla. R., 'In Flight Atmospheric and Fuel Tank Temperature Measurements,' NASA Conference Publication CP2307, 1984. 17. Kamin. R.A. and McConnell. P.M.. 'Impact of Higher Freeze Point Fuels on Naval Aircraft Operations.' ASME paper 86-GT-262, June 1586.

ACXNOWLEDOMENT This program was sponsored by the Energy and Natural Resources Office of the United States Office of Naval Research. The authors wish to acknowledge P.M. McConnell and A.F. Grenich of Boeing Military Airplanea and L.L Byrd of the Naval Weapons Center for theip support and efforts in this program. CYLINDRICAL (18 X '17)

RECTANGULAR (20 x 22) FIGURE 1 OPTIMUM GRIDS FOR TWO DIMENSIONAL SIMULATIONS

10 0 f.. -10 2 -20 2 E -30 -40 -50 -60 -70 .a" Freezing 61ans

a- a- 0 10 20 30 40 M 60 70 80 90 100 110 120130 140 1% 160 170 180 190

EFFECT OF FREEZING MODEL. ON FUEL TEMPERATURE CALCULATIONS FIGURE 3 FUEL HOLDUP CURVE BASED ON SHELL THORNTON TESTER DATA

route temperature analyela *16-war data base (determine ' ambient and recovery temperaturse during IllghlJ 4Jae MIL 210-8 Tables

Select Fuel 6nk .Specily Tank Dlmensione .Specify Tank Wall Characterletbs L Pylon Tank Model Wing Tan\ Model Complex Geometry (Cylindrical Coords.) (CarterIan Coorda.) (Body Fitted 2-D; 3-0 2-0; 3-D Coords.) avsllable 1 4 Input Time Varylng Boundary Conditlona *S~eclIytank akin lemperilures It available *Speclly external heat trawler coellIcien1 and recowry temperature protlle

4nltlaI conditions .Temperature dependent luel propertie8 *Freezing propertlee

*QUIck look printout -Plots of temperature DIOflle9 *Plots of troren tuel reglone *Calculation 01 total amount 01 hold-up YO. time

FIGURE 4 FLOW DIAGRAM ON APPROACH TO USE COMPUTER MODEL Lagend: + Terldala 7060 I o Calculaled

-10 c /'>

.30LT -40 I I I I I 11 10 20 30 40 50 60 70 80 90 100 110 120- 130 140 19 3 Tim. mi"

FIGURE 5 COMPARISON OF CALCULATED AND1 MEASURED FUEL TEiMPERATURES FOR THICK WING RECTANGULAR TANK TEST

100

MsBurecl lop mm cabla -62 '- and 30 ~unday)1ompOmturea bw.uP -66 20 -66 .a 10 -70 0 0 40 80 120 1 BO 200 240 280 Time. min FIGURE 6 COMPARISON OF CALCULATED AND MEASURED FUEL TEMPERATURES FOR THIN WING RECTANGULAR TANK TEST 25-11

FIGURE 7 150 GALLON FPU-3A INSTRUMENTED FUEL TANK

L*l.nd: . TL 321,TL 322 led data

TR 322 lest data - Piedinion based on speclied wall temperatures 0- Piedinion based on heat lransler kundary -10 - -20 - -30 - -40 - **...._ '+. -50 - -60 - -70 - -80 - -90 - OO ,b ;o io a0 ;o $0 ;o io do ,bo ll0 1;o lA0 ld 1;o 1ko 1;o 1;o 1bo Time. Mn

FIGURE 8 COMPARISON OF CALCULATED AND MEASURED BULK FUEL TEMPERATURES FOR FPU-3A WIND TUNNEL TEST 25-12

L.(l.nd: TB 327 1061 data - CakuIat1.d (BC - wall 1.w) - Cakulatgl (BC - H.T. menkisnt + enema1 tew) I;/, , , ,

-60 -70 -60 -90

'0 '0 10 20 30 40 & $0 ;O i0 i0 1bO 1;O 1;O lkl lkl 1$ lA0 I;O 1bO 190 Tim. mi"

FIGURE 9 COMPARISON OF CALCULATED AND MEASURED FUEL TEMPERATURES (BOTTOM OF THE TANK) FOR FPU-3A WIND TUNNEL TEST

60.0 r 60.0 r- Legend: Legend: 40.0 0 Data 40.0 0 Data - Prediction p:\ - Prediction x 20.0

a k- -20.0

-40.0

-60.0 0.0 60.0 120.0 180.0 240.0 0.0 60.0 120.0 180.0 240.0 Time, min Time, min STATION 32 STATION 108

FIGURE 10

COMPARISON OF THREE DIMENSIONAL MODEL CALCULATED FUEL TEMPERATURES WITH VALUES MEASURED DURING FPU-3A FLIGHT TEST 25-13

3. R. Jacques, Ecole Rnyale Militaire Discussion Your calculations are done for a steady flight and given temperature gradients in the fuel tank. Does air turbulencc or aircraft manocuvring (roll or steep hank angles) not 1. C. Scott Bartlett, Sverdrup Technologies disturb completely the temperature distribution in the tank? Do you foresee any instances where the additional cost ofa 3-D solution over a 2-D one will be justified? Author: Although tank vibration, air turbulence and flight Author: manoenvres will cause fuel sloshing and movement in the The 3-D model, although much more costly to operate, tank, previous work performed by the Boeing airplane other than the 2-D model would be justified if the benefits of company has shown that these effects are relatively minor its results would be of a higher value than the associated and have little effect on thc fuel cooling profilcs or hold-up costs of the model. within the tank.

4. L. Richards, BAe Were you able to check that the position at the four hold-up was in agreement with the predicted position? 2. D. Hennecke, Technische Hnchschule Darmstadt How did you determine the internal heat transfer Author: coefficients for the wind tunnel and flight conditions? No, because the cylindrical tank was bill-of-material hardware. Access to view the internal freezing patterns was Author: impossible. However, the total amount of fuel hold-up, The external heat transfer coefficients were calculated by unpumpable frozen fuel, was measured during wind tunnel thc computer model using a general flat plate correlation. testing and compared to predicted results. The work is going Additional information regarding the specifics of the on in the NAPC low temperature fuel flow facility, which has correlation can be found in reference 5 of the paper. optical access to determine the frozen fuel locations.

28-1

ENVIXONmNTAL ICINQ TESTINQ AT THE NAVAL AIR PROPULSION CENTER by William H. Reardon and Vito J. Truglio Naval Air Propulsion Center P. 0. Box 7176 Trenton, New Jersey 08628-0176 U.S.A.

SUWRY

A compreheneive Environmental Icing Simulation System has been developed by the Naval Air Propulsion Center. This syetem has accommodated the testing of duoted and free stream mounted engines and free stream mounted aircraft inlets. Three areas are discussed in detail in this paper: 1. Navy specification icing test procedures. ~uccesso~iteria and rationale for the requirements : 2. The ovepall Eapabilities of the NAPC icing facilities in terms of mitical icing cloud parameteps such as liquid water content, mean effective droplet diameter. humidity and inlet air temperature; and how the icing environment is established, calibrated prior to test and verified during testing along with historical experience and lessons learned; 3. NAPC teat experience in Navy qualification programs for the T406, T700 and F404, as well as demonstration and development test programs performed with the TOMAHAWK Cruise Missile inlet and the F-14A aircraft inlet duct. Navy Smcification Reauirament# The Navy Qeneral Specifications MIL-E-005007E (turbofan/turhoJet)and MIL-E-008593E (turboshaft/turboprop) (reference 1) are the baseline documents from which Navy engine model specifications are built. They require that turbine engine anti-icing systems allow the engine to operate satisfactorily under the meteorological conditions as defined in Table I, with not mare than 5.0% total loss in thrust OF shaft horsepower (Slip) available and no more than 5.0% total increase in specific fuel consumption (SFC) at all operating Eonditions above 50% maximum continuous power (MCP) setting. Far operation at leaa than 50% MCP, 95% ai the desired power above 50% MCP must be obtained within specification acceleration times. No permanent performance deterioration is pepmitted after the meteaPological conditions have been removed.The engine anti-icing system munt also prevent accumulation of ice on any engine part exposed to the gaa path while operating in icing canditiana.The total lose in performance of 5% includes the effect of operation in the icing environment plus the effects of operation of the engine anti-icing ayatem. The baseline for determining engine perlormance loss is established by operating the engine with no customer bleed air or power extraction under the inlet temperature oanditions of Table I with air between 80 - 100% relative humidity (RH) and zePo liquid water content (LWC). Delivered power losses and fuel consumption increaaes are determined by comparison of engine performance when operating in the icing conditions defined in Table I with the baseline values. The test consists of two parts: a. Part A contains two runs at each of six power settinga under the three conditions of Table I, Part 1. At each power setting the engine is operated for a minimum of 10 minutes. During each period the engine is rapidly accelerated to intermediate power to demonstrate transient response. b. Part B contains a one-hour run at idle, followed by an acceleration to maximum power, under the conditions of Table I, Part 2.

Rationale for Smcification Reauirementa (Reference 2) Genepally. some form of anti-icing system is incorporated into the engine design to provide WaFm compressor bleed air to those surfaces susceptible to ice build up. Shedding of i@e into the gas path can result in flameout, stall, surge, and engine failure. Ice build up that chokes off the engine inlet can result in performance deterioration and/or turbine overtemperature due to air starvation, and control achedule changes caused by engine sensor malfunction due to icing.

Measupement of performance degradation is intended to be based an power loss at a particular throttle setting. This has also been interpreted to mean that the 3% power loss should be based on the intermediate power level setting, hecauae this ie the highest available power condition. Another interpretation is commonly used in the turboshaft engine community, where most of the engines tested are operating an an external torque limiter at the Table I cold inlet Conditions. This interpretation contends that shaft power which degrades with the initiation of anti-icing and the icing cloud can usually be recovered by increasing the throttle (power lever, collective pitch, etc.) until a temperature, torque, fuel flaw or speed limitel. is reached. This iB Justified since power is Tecovered if power is still available, up to 28-2

operational limiters, with the protection against excessive degradation reflected in the associated values of SFC. One alraumptian hepe is that. the correlation between calculated and measured turbine temperature does not shift during test. Deviations in the 5% power loss requirement nave been granted in the pant, based on the engine bleed requirements necessary to ppevant ice accumulation. The transient power requirements for power levels below 50% MCP were incorporated to confirm that and power loss at low power is not only recoverable, Ibut does not prevent the engine from accelerating to high power levels. Transient performance within 125% of Specification allowable transient times for poweF chan&es is required while operating in the icing environment. Part I testing is pepfopmed to determine the engine anti-icing capability throughout its entire operating range. The specification icing test requirements which were derived from National Advisory Committee on Aeronautics (NACA) Technical Notes 1855 and 2568 are based on data gathered from stratiform clouds (usually prevalent at 3,000-6.000feet) which have a Low to moderate LWC, and cumuliform clouds which occur from 4,000-24.000 feet and have a moderate to high LWC. In the case of stratiform clouds, depicted in Figure lA, i,he levels of LWC in the cloud form ape normally less than the 1 gram per cubic meter (gmlm3) teat requirement, while icing encounters in Eumuliform clouds (Figure 18) tsnd to be relatively short in duration (appFoximately 1 minute). The icing test conditione of the specification are fairly consistent with the enaounteps in cumuliform clouds. hut the engine is pequired to operate there for a much gpeater period of time. Although conaervative, the Navy ape@ifi@ation test requirement seems to adequately verify an enginea anti-icing system capability, a8 service experience shows a relatively good record with a low number of icing incidents Felated to the engine anti-icing system.

Part I1 testing of one houp rt ground idle in the Table I, Pavt 2 icing environment ia very simila~to the Federal Aviation Regulations (FAR) Part 33.68, CategoFy B teat requirement for civil ~e~tificatinnwhich requires a eimilarl condition run for 30 minutes. It ia intended to simulate engine opeFation, on the ground, in a freezing rain 01- icing envivonment while waiting fOF takeoff clearance. The ensuing acceleration to maximum power simulates the takeoff roll. Earlier versions of the Navy specification required altitude in addition to sea level icing tests. Altitude tests Were deleted from the specification, in part due to the high cost associated with performance of this test in an altitude chamber. Technically, the sea level icing conditions Pepresent the mast severe requirements for the engine anti-icing system. This was justified by comparing various engine anti-icing system enepgy outputa. to icing cloud energy input, at combinations of engine power level, and inlet flight and meteoFologica1 conditions. Low engine power, low Mach number, cold inlet conditions (idle. static) prod'uce the lowest compressor bleed anti-icing system energy. Sea level icing conditions provide the highest engine tatal massflow yielding greater total liquid flow. Combining high liquid flow with ;.i&nger dwell times encounteped st low airspeed. the greatest difference in energy or most severe requirement for the engine anti-icing system is created.

Allison Used specification sea level icing tests in the development of the model 501 and 250 engines based on studie5 made using data from flight teste at 3,700 to 19,000 feet. During the FAA Aipcraft Icing Symposium of April 1QI38, Allison indicated that an anti-icing system designed to meet the military specification requipements at sea level would result in a configuration which waz satisfactory at altitude, and in the meteorological conditions encountered in flight operations (reference 3). At the same symposium, airline operational axperien@e in icing conditions showed that the most potentially sepioua icing pFoblem in the jet engine airplane was engine icing an the ground, standing in line, waiting for take off clearance. In addition, experience showed that jets had somewhat le88 of a problem once airboPne, ae the high rate of climb allowed them to exit the laing envipanment quickly (r,eferen@e 4). It should be recognized that icing conditions can be encountered at high altitudes, although encounters above 22,000 feet are Fare.

Test Facilities Engine icing tests have been conducted at NAPC for the past 30 years. During this period the facilities which have been developed to perform these tests have grown in number as has the expertise to provide a reliable environmental icing system for testing engines.

There ape four small engine t(?St cells which have Sea level as well as altitude capability. The maximum total airflow far icing tests in these cells is 50 pounds/seeond (lb/sec). There alle 5 large engine cells, 2 sea level and 3 altitude, with a maximum air flow of 300 lh/sec. AS mentioned. altitude icing requirements have not been part of the Navy military specification in Fecent years; however, altitude icing tests have been performed at NAPC in the past. NAPC icing test history includes: (1) Turbojets: 552, 579, (2) Turbofans: TF41, TF30. TF34, F404. (3) Turbaahafts: T55, T700, T56., T406 and (4) the F107 enginelTOMAHAWK Cruise Missile inlet combination and the F-14A aircraft inlet. Currently, icing calibration installations are being prepared for the T800. FllO and F412 engines. There are two basic icing inlot configurations utilized at NAPC which are depicted in Figme 2, the direct connect and fFee Jet. Large engines with mass flows greater 28-3 than 40 lb/sec are directly connected and small engines with mass flows less than 40 lb/eec ape generally configured as a free jet. The F-14A inlet test required a free jet configuration despite Fequiring mass flows greater than 300 lb/aec. Cloud Simulation

Simulation of cloud conditions is a@@ompliahed with a spray system consisting ai horizontally mounted bars shown in Figure 3. Each bar has openings which accept either a plug 01- nozzle, allowing for pattern adjustment. Metered and controlled air and water to the nozzles is introduced inta individual manifolds to which each spray bar is connected. The nozzle. an aspirating type similar to a design developed by the National Aviation and Space Administration (NASA) is shown isometrically in Figure 4. Water is forced through a hypodermic Size tube and aspirated at its exit by mixing with a pressurized air source. Liquid water droplets are generated when the mixture exits an orifice. The air is heated to prevent icing of the nozzles. Flow of both water and air are remotely controlled, independently. which provides for the setting of LWC and drop diameter. In the direct connect configuration, nozzle water and air flow must be adjusted as engine air flow changes. In the free jet configuration, a constant duct airflow can be maintained, therefore a constant nozzle setting can be held for a given LWC and drop diameter regardless of engine power changes. The nozzle water is potable water which is demineralized and filtered. Nozzle air is also filtered as well as dried. It is important to minimize the entrained particulates which can seed the supercooled liquid water droplets causing them to freeze and form ice crystals. Saturation of inlet air is accomplished independently by injecting steam through a nozzle located in a plenum upstream of the water spray system.

Calibration Techniqu.6 Full scale calibrations are conducted prior to icing tests primarily because of variations that exiat from test to test. such as engine air flow range and inlet configuration diffepences. test @ell dissimilarities or modified icing requirements. Figure 5 shows the typical inlet arrangements for a full scale calibration and engine terrt. The primary objective of a calibration is to set up the entire icing simulation installation without the test article. Qeneration of LWC and drop diameter are verified quantitatively and qualitatively as required by the specification prior to the actual test. Inlet air temperature is measured using thermocouples looated upstream of the spray bars. Pressure is measured with a steam-heated free stream total pressure rake and wall static taps in the duct, down stream of the spray bars. Generally this probe is only used during calibrations. It is correlated with a total pressure probe located up stream of the spray bars for setting flow conditions during test. In the direct connect arrangement there is a reducer duct Which can be heated to prevent ice build up, however nozzle patterns are adjusted to minimize the amount of accretion on the Feducer surface. Humidity is measured during both the calibration and test with a thermally-cycled photocell type dew point sensor. Saturation is verified visually. Droplet size is controlled by proportioning measured water flow and air flow to the nozzles. During pre test calibrations this relationship is confirmed using a forward scattering spectrometer prohe (FSSP) which is traversed across the exit of the inlet duct where the test article interface would be located. This probe is a laser optical instrument which measures and counts droplets according to the amount of energy scattered when the droplets pass through a focal point in the laser beam. LWC is generated by introducing a metered amount of water flow proportioned to a given amount of measured inlet air flow. A single rotating cylinder introduced into the duct near its exit provides an independent means to measure and calculate LWC which can be compared to the metered quantity. The cylinder is housed outside the duct in a supercooled chamber. It is remotely immersed in the air stream for a set time interval and retracted back into the chamber. The ice weight is measured after the cylinder ia removed from the chamber. From the ice weight, erpoaure time and duct velocity, LWC can be calculated and compared to the metened quantity using the following equation (referenae 5): r I .

Where: Y = Specific Volume of Ice 1( = Catch Efficiency of Cylinder T = Exposure Time w= Weight of Ice = Length of Cylinder F = Radius of Cylinder This cylinder is used both during calibrations and tests to verify that the droplets are supercooled liquid and that evaporation or crystallization has not occurred. Ice distribution is checked during calibrations by positioning a 2 inch by 2 inch mesh screen at the test article inlet plane. The ice accretion is measured and the nozzle pattern adjusted as required. 28-4

CaDabiliti.9 The icing capabilities at NAPC can best be summarized as shown in Figures 6A and 6B. Total inlet temperature and LWC are shown as a function 0.f inlet airflow. The minimum temperatupe obtainable is -50°F at 40 lblsec and t3Z°F at 310 lblsec. The icing cloud simulation capacity is comprised of two envelopes which differentiate between free Jet and direct connect LWC capacitles. Nazzle water and air flow limitations are the constraining parameters that e5tabliah the LWC and droplet diameter boundaries.

NAPC ExDerience Producing a consistent icing cloud simulation in terms of LWC and droplet diameter ie the most important consideration in the development of an environmental simulation system for the performance of icing teste. This is due to the faat that the stability of the wateF droplets produced for such a system is highly sensitive to the surroundings. The supercooled droplets are susceptible to evaporation when the inlet air is not satupated and f?eezr-out (ice cryatallization) when particulates are present. Fu~thermorethere are uncertainties involved with any method available to mea8uFe and validate LW rendering it difficult to detect freeze-out or evaporation. Considerable effort has been made at NAPC to provide the best possible icing simulation system (refevence 6). To minimize freeze-out the nozzle water supply is demineFalized and filtered to one mimon. Nozzle air is e.180 filtered and dried. Inlet air is not filtered; however, there have been tests in the past substantiating that the amoun't of particulates in the supply are minimal. There have been many attempti3 made to measure LWC accupately. Currently the FSSP has been used to accurately count and measure droplet diameter reliably. Also in order to increase its usefulness, a program Was developed to calculate LWC from droplet count data acquired. However, the calculations have proven to be unrepeatable and unreliable. A major shortcoming is that the probe can not distinguish discretely between watei- droplets and ice crystals.

The method found to be most 1,eliable in terms of verifying LWC has been the single rotating cylinder. There ape inaccuracies inherent in this method but it is the only way presently known whepe freeze-out can be detected sin@e fpozen droplets will not accrete on the cyllndex In the near term, efforts are on-going to investigate the feasibility of an alternative to the FSSP, the phaseldoppler particle analyzer. This ia a non-intrusive instrument which is capable of measuring droplet size and velocity through a volume defined by the intersection of two laseF beams. The probe could poasibly be installed external to the inlet, thereby allowing for use during engine testing. However, it ie not anticipated that freeze-out detection could be possible using this instrument, NAPC Icinf Test ExDeFience 1. T406-AD-400 Turboshaft Engine

The Allison Qas Turbines Division (AQTD) T406-AD-400 turboshaft engine, designed to power the V-22 Tiltrotor aircraft, wa8 teated to the Allison Model Specification 937 Environmental Icing Test requirements duping its Low Production Qualification (LPQ) phase. Model Specification 937 was derived from Mil-E-8593E and was approved by the Navy. It contains success criteria of no mope than 5Z SHP loss and 6.5% SFC increase in the icing environment.

In the first test attempt (reference 7). the engine anti-icing system was found to be deficient at law power conditions: At Table I. Part 1 condition; -4OF, lgm/m3. below approx 50% MCP (1100 pound-feet (lb-ft)) unacceptable ice formed on the engine inlet guide vanes (IGV's). At Table I. Part 1 condition: t'23OF. 2gmlm3. unacceptable ice formed on the IQV's below approximately 400 lb-ft torque. At the Table I, Part 2 condition, the lowest torque at which the engine had adequate anti-icing capability for a one hour period of operation was 450 lb-ft. Below that, unacceptable ice formed On the IQV's. The engine did however, meet performance degradation requirements of 5% SHP and 6.5% SFC when measmed at constant Power Demand Signal (PDS). Allison in-house investigation of the engine anti-icing system post test (reference 8). revealed that the engine IQV anti-icing airflow supply was deficient. This flow is supplied via external plumbing from the fourteenth stage compressor discharge to the engine inlet housing, where it iz internally manifolded to the individual guide vanes through vane tip porte. Flow checks by Allison confirmed t'hat only 1/2 the design flow rate was passing through the vanes. In addition several areas of system ovepboard leakage were identified at anti-icing supply tube to enginf! Ease mating point8 and around guide vane bushings. Heat transfer modeling with these inputs predicted local areas of the IGV at the hub and blade tip which could be Buaceptible to icing. These results we~econeietent with the ice accretiona seen during the initial teat at NAPC, where on video monitoFe the ice accPetions Were seen acaumulating fFam the hub area outward Fadially along the leading edge of the IQV's.

A retest of the T406-AD-400 was performed with a modified IQV anti-icing flow 28-5 scheme, incorporating reworked IGV hub and tip exit holes and better sealing of the anti-icing flow path. In the retest, the engine anti-icing syatem prevented ice accumulation in the engine gas path at all three specification icing conditions. Losses in shaft horsepower related to operation of the anti-icing syatem in the icing environment exceeded the specification limit of 5% at the i23OF and -40F meteorological conditions of Table I when compaped at constant power demand signal (PDS). constant gas generator speed (NG) or constant measured gas temperature (MQT) above 50% MCP. In all cases, at and above 50% MCP, power loss was recoverable throughout the range of environmental icing test conditions with an increase in PDS while remaining within engine temperature and speed limits. This was attributable to engine torque limiting being in effect. The V-22 propeller gearbox torque limit is 2150 pound-feet, which is below the engine maximum power output capability at the Table I test conditions. The incpeases in SFC in the icing environment when compared at constant SHP Were within the Model Spec 937 limit of 6.5% at all conditions. For this test program the NAPC recommendation was to accept the engine as having met the icing requirements of Model Spec 037 for the V-22 installation. This would require granting a waiver for SHP loss associated with operation of the engine anti-icing sy~temin the icing environment. Presented in Tables IIA and IIB, ia the T406-AD-400 engine performance as run in the retest of the envimanmental icing test.

2. T700-QE-700 Turboshnft Engine (reference 9)

The anti-icing aystem of the T700-QE-700 engine contains an integral inlet particle separatop (IPS), whiah adds extensive flowpath surfaces beyond conventional engine configurations. Anti-icing of the T700 IPS, IGV's and engine front frame is accomplished with a combined system of continuou8 hot engine ail and modulated axial compressor bleed air. A valve variea the bleed air with corrected compressor speed through a variable compressor vane a@tuator linkage.

Testing was conducted in a freestream environment with a specially fabricated engine bellmauth and bulletnose assembly. The bellmouth and bulletnose were designed to he anti-iced by a fa@ility hot air BUpplY. In addition, engine power absorption was accomplished by means of a high speed waterbrake enclosed in the bulletnose.

No ice accumulations were observed on the engine inlet surfaces (IGV'S, struts, front frame) during the testing. Post test observations revealed that small amounts of ice had accumulated on the inside diameter of the IPS scroll cover assembly. This served to explain light puffs of smoke which were observed exiting the IPS blower during icing testing. This ice did not damage the IPS blower. Engine performance changes due to the operation of the anti-icing system and operation in the icing environment are shown in Tables IIIA and 1118. The T700-GE-700 Engine Model Specification DARCOM-CP-2222-02000C which is an Army approved derivative of MIL-E- 0085933, allowed up to 9% SHP loss and 7X increase in SFC under Table I conditions. Where levels of SHP exceeded these values during test, it was shown that up to intermediate rated power (IRP), a percentage of the lost power Was recoverable. For testing at the -49 icing conditions. additional power loss was observed once the icing spray was introduced, which was greater than the ~OEESeen with only engine anti-icing systems on. These power lbases were suspect due to inconsistency with the +239 results. Cycle analysis and other QE turboshaft engine icing test experience indicated that loaaes in periormance at icing temperatures for wet operation should be no worse than for dry operation. The T406 test ahowed a similar inconsistency at the -4% icing conditions. One explanation could be the difference in the energy required to vaporize the super cooled droplets at -49 versus the energy Fequired to vaponize the droplets at +239, combined with the difference in the axial location in the compre88or where that ohange of state occura, could effect overall engine performance depending on the contribution of each stage of compression to the overall efficiency of the compressnr. Transient performance between ground idle and IRP with customer bleed and anti-icing activated showed no indications of stall or instability. Standard procedure for documenting transient times in a turboshaft installation with a free turbine, where a waterbrake is used for power absorption, is based an the change in gas generator speed (NQ) equivalent tb that required to achieve 95% of the overall power change. This method most nearly reflects the transient response capability of the engine independent of the power absorber inertia and dynamics. The NO speed associated with the starting and finishing points of the transient excursion are defined by steady state operation at these points.

3. F404-QE-400 TurboIan Engin. (reference 10) The F404-QE-400 Engine Environmental Icing Test was conducted at NAPC in accordance with GE Model Specification CP45K0006. which allowed losses of up to 10% in thnust and up to a 10% increase in SFC while operating in the icing environment. The test requirements Were modified by the Navy to expedite the program. Only the flight idle and IRP power settings were retained. It was felt that if the engine could satisfactorily anti-ice at these conditions, it could anti-ice at any power in between

The installation was a direct connect inlet arrangement, where the 60-inch diametep facility air supply duct transitioned to 28-inch diameter at the engine inlet. The transition section was jacketed so it could be heated to prevent ice formation on the internal surfaces. The initial setup contained a labyrinth seal to allow a floating 28-6

thrust bed and associated force measurement, however, it Vias removed when it was found to be a BOUFC~of ice buildup. Thie neceasitated locking of the thrust stand becauae the axial displacement allowed by the labyrinth seal had been eliminated. In order to pvevent ice formiltian in the engine inlet duct from the inlet rakes, they were Femoved prior to the :Icing tezt. Without the inlet rakes and the labyrinth seal, the engine thpust could not be meaaured directly during ioing Funs, and was obtained by Eomparinon of parameters to precalibration values. In this Eaee, the relationship of gross thrust to engine preaaure ratio (EPR) obtained during the precalibration was used to determine engine net thruat, a8 a function of the 'engine pumping pressure Patio' (low prsssure turbine discharge pressure / engine inlet pressure). The engine inlet total pressure and total temperature during the precalibrations were based on temperature and pressure measurement rakes mounted in the inlet duct. After these rakes were removed for the icing tests, the engine inlet temperature was assumed equal to the average of four thermocouples (TC's) located upstream of the water spray bars. These TC's had agreed within 1 F of the engine inlet temperature based on the engine inlet rakes. After the pressure rakes in the inlet duct were removed, the engine inlet pressure was assumed equal to the pressure in the 60- inch plenum located upstream of the spraybars. By using this pressure together with the wall static pressures in the 28-inch duct, a corrected airflow was calculated which was within 2% of the airflow during precalibration.

The results of the test showed no significant ice buildup on the engine at any of the three environmental icing conditions tested. As shown in Table IV, the anti-icing bleed air caused th7ust losses of approximately 4-7% at constant measured turbine temperatupes. No additional significant performance loss was noticed while operating in the icing conditions. Increases in SFC at constant thrust at IRP were well within the 10% allowable limit.

4. F107-WR-400 Turbofan Engine with a TOMAHAW Missile Inlet The Navy Specification pequirements have also been tailored to address specific installations. The F107-WR-400 engine waa designed to power the TOMAHAWK Cruise Missile. Neither the engine nor the missile inlet was anti-iced in this inatallation As a demonstration, the specification teet wan tailored to provide a representative icing environment for the engine/miaeile inlet combination (Teference 11). An F107-!NF-400 engine installed behind the TOMAHAWK missile inlet was mounted in the freestream configuration. All testing was perfapmed at Sea Level, Mach 0.65 freeatream conditions with an inlet air total temperature of t23Q. Three ioing Funs were performed, consisting of two one-minute Puna and one five-minute run. The deaired liquid wate? content was 1.0 gmlm3 and the mean effective droplet diametep (MEDD) was 20 microns. The engine was run at cruise power fo= one minute with half the desired liquid water content (0.5 gm/m3). The only ice accumulation on the engine hardware was a 1/16-inch buildup on the center of the spinner located forward of the 1st stage fan. The TOMAHAWK inlet duct had a rough accretion of ice on the inside lo we^ half of the fopward 3/4 of its length at an average thickness of 1/16 inch. The inlet lip had a 1/8-inch build up on its circumference. The engine had ingHsted an 8-inch long Section without damage. A second one-mi:nute run at 1.0 gmlm3 yielded build up at similar locations at about double the thickness, indicating the rai:e of buildup was proportional to LWC. A five-minute run was attempted, during which, ice continuously built up and broke off the inlet lip. Engine Stall occurred at 2:50 into the run, after a section of ice from the outboard side of the inlet lip was ingested. Engine inlet housing vibvationa we~ehigh during the final two minutes of the Fun. Inspection of the engine on shutdown, revealed the only ice formation on engine hardware was a 1/16- inch buildup on the center of the spinner. On closer inspection, five let stage fan blade tips we~efound bent over. Review of the transient data inevealed that during each of the icing runs. the first minute was charactepized by an increase in fan speed (Nl). thrust and compressor discharge pressure (CDP). This was due to the maas flow increase introduced by the icing environment through the engine. Following, waa a drop in fuel flow, thrust and CDP due to inlet blockage from ice buildup and corresponding duct preesuve drop. These parameters cycled a8 ice broke off the inlet. The inlet pressure loas also affected speed match of the engine. After 2:50 and 4:20 of the icing run, N1 increased for a given gae geneTator speed (N2) at constant inlet temperature. Thoae stated timee appear to be when the two incidents of ice ingestion and damage occurred. N1 is higher for a given N2 due to ?educed efiicienw of the damaged fan blades. Engine thrust also decreased continuously during the test from inlet pressure loss. Because of the accumulation of ice on the engine installation hardware, engine thrust could not be determined from mete? readings during the icing Funs due to the increased drag force. Engine thpust loss was determined by comparison of engine parameters during the icing run with those obtained in the pretest performance calibrations. Figure 7 is a rep~esentationof the thrust loss! seen throughout the five minute run.

The severity of the icing environment in this demonstration may be evaluated as 28-7 follows: The missile traveling at Mach 0.65 will travel about 7.5 miles in one minute. To maintain the missile inlet below freezing (+23OF) at Mach 0.65 flight condition. the equivalent ambient temperature is -15OF. Cumuliform clouds at -159 do not exist below 12,000 feet altitude. In addition, this cloud form is typically leae than an average of 3 miles long if its LWC is 1 gm/m3. Which would pelate to a cloud dwell time for the missile of less than 30 seconds. Stratiform clouds exist at very low altitudes and at -ISOF, their LWC is typically less than 0.2 gm/m3. Stratiform clouds with this LWC have an average horizontal extent of 20 miles which would equate to apprnximately 3 minutee: duration in that environment. 5. F-14A Aircraft Inlet Icing Test In another developmental effort, NAPC utilized the specifieatian icing conditions to investigate a fleet ppoblem encountered by the F-14A aircraft. The program was initiated because of several reports of F-14A aircpaft engines experiencing foreign abject damage (FOD) from ice ingestion (reference 121. Frnm information available after one of these icing Incidents, the aircraft had been operating for approximately 30 minutes in what was perceived as icing conditions at 8,000-10,000feet at 250 knots (0.4Mn). The aircraft descended and made a normal carrier approach and landing. F-14A engine icing FOD was reported to have occurred in some incidents during the carrier approach and in others on carrier arrestment. Ice was found on the rear wall of the bleed/bypass cavity in an F-14A aircraft that did not experience ice damage.

Ppior to the NAPC test thepe had been essentially no previous testing conducted to determine how the bleed/bypass system functions in flight. Questions regarding the percentage of total inlet duct airflow that is bled, and whether the bleedlbypass system functions as an auxiliary inlet in flight were unanswered. Grumman Aircraft Covp. estimated that approximately 10% of the air entering tha inlet duct exits through the bleed/bypass door.

The F14-A aircraft engine inlet duct (Figure 8) is a variable geometry inlet used to decelerate free st~eamair in flight to provide even Subsonic airflow to the engine throughout the aircraft flight envelope. The inlet has a two dimensional. four-shock, external comppessian system with horizontally oriented ramps. It has a three degree initial fixed ramp and three automatically controlled variable compression rampa whiah are programmed to vary with Mach number. The compression Famp boundary layer is bled through a throat bleed slot and dumped overboard through the bleed/bypass door. The objective of this program was to attempt to duplicate the ice formations that had been observed in flight aircraft and recommend possible EOlutions to the icing problem. The program was later expanded tb include actual teating of the proposed heating blanket fixes. With limited facilities available for the icing inEtallation, the teat was Fun at sea level and reduced Mach number conditione. Due to airflow limitations, air was permitted to flow into and acrose the top and bottom of the inlet duct only. Air was allowed to flow acFoBs the top of the inlet duct in an attempt to simulate the inflight airflow conditions around the bleed/bypass door, which affects the airflow rate and direction of airflow through the bleed /bypass cavity. Simulating the aimflow patterns in and around the F-14A inlet duct in the area of the bleed/bypass system is critical, since the rate and pattepn of ice accretion in that area is dependent an this airflow pattern. A controlling factop in the ice build up in the bleed/hypass system cavity and on the leading edge of the f3 Famp is the amount of air that is bled by the bleedlbypass system; the ice build up is directly related to this airflow rate. Conditioned plant air was allowed to flow over the bottom of the duct to prevent distortion of the airflow to the engine from the inlet lower lip. Testing was conducted using plant air and a TF30-P-408 engine as a ‘workhorse pump’ behind the inlet in an attempt to cpeate a realistic simulation of the airflow patterns of the inlet duct. The port engine inlet duct from an F-14A aircraft was installed in the test cell and partially insevted into a 72-inch diameter plant facility duct. Aluminum closure plates installed between the facility duct and the F-14A duct were used to block airflow past the sides of the inlet. Aerodynamic surveys showed that the bleed/bypass functioned solely as a bleed for testing with the ramps in the stowed Position. The bleedlbypass system with the Number (#)I and f2 ramps lowered functioned a* a bleed for the lower engine power settings, but as engine power incpeased, the bleQdlbypass airflow deCFeased until the flow reversed. After the point of flow reversal was reached. the bleedlbypass functioned as an auxiliary inlet.

At the icing condition of t23”F inlet temperature, 2 gm/m3 LWC and 15 microns droplet diameter, with the f1 and f2 ramps stowed (low Mach condition), heavy ice accumulated on the f3 ramp leading edge and the inlet duct tie rad. Testing with the ramps lowered (high Mach condition), showed essentially no ic* formation in the bleed/bypaas cavity. Some ice formed on the lower corner of the f3 Famp leading edge and lower (engine gas path exposed) surface. In bnth cases there was alaa significant ice bulld up an the lower lip of the F-14A inlet. Testing was then performed with the rampa in the stowed position to determine where shed ice from the lt3 ramp leading edge would col1,ect. After achieving the ice buildup, the facility air was warmed rapidly, after which the engine was shut down. Examination of video tape of this event showed that lt3 ramp leading edge ice had broken away and lodged in the cavity above the lt3 ramp. Also found in the cavity was ice which bad accumulated on the hydpaulic ramp 28-8

actuator. This test demonstrated that the ice found in the bleed/bypass cavity of flight aircraft was mast likely caused by ice breaking from the 113 ramp leading edge. The large amounts of ice repopted in the bleed/bypass cavity of flight aircraft could be the result of flying through a series of icing and non-icing conditions. whereby ice may repeatedly form on the 113 Famp and break off collecting in the bleedlbypass cavity. This theory is supported by some of the icing incidents reported which stated that the ice found in the bleed/bypass cavity appeared to be composed of several layers.

References:

1. Naval Air Propulsion Center, Military Specifications Mil-E-O08593A, 15 Oct 1975 and Mil-E-O05007E, Apr 1983. 2. Naval Air Propulsion Center, NAPC-E-79003B/1, Mil-E-005007D Requirements Rationale, 10 Fcb 1981.

3. Q. Bianchini., Allison Division of General Motors,'Small Qas Turbine engines and Inlet Icing Prote@tion', Airc!raft Icing Protection Symposium, FAA, April 1969. 4. P.A. Soderlind. Northwest Airlines, .Airline Operational Experience in Icing Environment', Aircvaft Icing Protection Symposium, FAA, April 1969.

5. C. K. Rush and R. L. Wardlaw, National Aeronautical Establishment Canada Report, 'Icing Measurements With a Slngle Rotating Cylinder', LR-206.1957.

6. V. 3. Truglio. Naval Air Propulsion Center, NAPTC-OP-94. 'Sea Level Icing Rig Calibrations for Small Engine Tests,' October 1976.

7. Allison Qas Turbines Division, T406-AD-400 Quarterly Progress Report, 27 April 1989.

8. W. H. Reardon, Naval Air Propulsion Center, NAPC Ltr Ser PE21/E419, T406-AD-400 Environmental Icing Test Results, 2 June 1989.

9. S. H. Qreen, Qieneral Electpic Ca., Report R76AEQ031, T700-QE-700 Engine Qualification-Anti-icing Test, 30 April 1976.

10. R. R. Thaler. Naval Air PFopulsion Centep, NAPC-LR-78-23, F107-WR-400 Ice Ingestion and Icing Environment Tests of 15 Nov 1978, PE62:RRT:cJs, Ser E850, 30 Nov 1978.

11. R. L. Magnussen and K. P. Halbert, Qeneral Electpic Co., Report R79AEQ1026, F404-QE-400 Qualification Test Phase - Environmental Icing Test, Oct 1970. 12. M. K. Fall, Naval Air Propulsion Center, NAPC-PE-83. Icing Tests of the'F-14A Aipcraft Engine Inlet Duct snd Evaluation of Proposed Icing Fixes. Jan 1983. TABLE I

SEA LEVEL ANTI-ICINQ COUDITIOPS (MIL-E-005007K (AS))

PART 1 PART 2

Engine Inlet Total -4.FI-20'Cltl' t15PF(-100Cltl*(1) +23°F(-5eC)f1' 23'F(-5'Cltla Tempelature t I 1 I ~~ ~~ ~ II I ve1oci ty 0 to 60 knots 0 to 60 knots 0 to 60 knots 0 to 60 knots

Altitude 0 to 500 ft 0 to 500 ft 0 to 500 ft 0 to 500 ft

Mean Effective orop Diameter 20 microns 20 microns 20 microns 30 microns I I I I I Liquid Water content 1 gm/m3 2 gm/d 2 gm/m3 0.4 gm/m3 (Continuous) t0.25 gm/m3 f0.25 gm/m3 t0.25 gm/d f0.1 gn/m)

11) This condition is deleted for .nowfan engines.

TABLE IIA T4OB-AD-400 EBVIEOYIIBUTAL ICINQ TEST I1 TEST RESULTS SEA LEVEL I +23OF I2 LW

POWER ENGINE ENGINE ICING SaP SFC y3T SHP SFC SETTING AI1 BLO SP~Y (HP) (LBIBBII~P) (OF) x LOSS x Inc

~~~ ~~ 1RP OFF OFF OFF 6070 ,435 1305 ---- .... ON OFF OFF 5875 ,440 1432 -3.2 t3.7 ON ON OFF 5810 ,450 1445 -4.3 t4.2 ON ON ON 5743 ,452 143-1 = -5.4 +4.6

MC P OFF OFF OFF 5120 .453 ______ON OFF OFF 4060 ,487 -3.0 +3.1 ON ON OFF 4044 ,471 -5.4 t4.0 ON ON ON 4048 ,470 -5.4 t3.0 PART OFF POWER ON -3.1 +3.0 ON -4.3 r3.7 ON -4.3 t3.7 PART OFF OFF OFF 3852 ,480 POWER ON OFF OFF 3731 ,408 -3.1 t3.3 ON ON OFF 3728 .400 -3.2 r3.8 ON ON ON 3720 .4B0 -3.2 r3.8 PART OFF OFF OFF 3112 ,500 _-______POWER ON OFF OFF 3023 ,517 -2.0 t3.4 ON ON OFF 3000 ,520 -3.3 r4.0 ON ON ON 2880 .525 -3.7 t5.0

' ALL POWKR SETTINGS ARE CONSTANT POWER DEMAND SIGUAL (PDS) = MEASURED QAS TEMPERATURE (MGTI LIMIT = 1540-F TABLE IIB

T406-AD-400 EUVIBOUMKUTAL ICIU5 TEST I1 TEST EESULTS SEA LEVEL I -4-F I 1 gm/m= 11113

POWER ENGINE EN5IUE ICIW SEP ‘SFC YOT SEP SFC __SETTING ___A/I BLD SPRAY IEP) (LBIBBIEP)- -(OF1 X LOSS X IUC

I RP OFF OFF OFF 6070 ,429 1335 ---- ..__ ON OFF OFF 5778 ,444 1376 -4.8 r3.5 ON ON OFF 5742 ,447 1385 -5.4 *4.2 ON ON ON 5513 ,453 1372 = -9.1 +5.6

MCP OFF OFF OFF 5237 ,449 _.._ .-__ ON OFF OFF 5059 ,464 -3.4 t3.3

ON ON OFF 5003~~ ,467 -4.5 r4.0 ON ON ON 4780 .475 -8.7 +5.8

PART OFF OFF OFF 4852 ,454 ______.. POWER ON OFF OFF 4535 ,467 -2.5 +2.9 ON ON OFF 4473 .470 -3.8 r3.5 ON ON ON 4353 ,477 -6.4 r5.1

PART OFF OFF OFF 4188 .463 ____ .-._ POWER ON OFF OFF 4103 ,416 -2.1 t2.8 ON ON OFF 4045 ,478 -3.4 t3.2 ON ON ON 3047 .483 -5.8 *4.3

PART OFF OFF OFF 3685 ,485 ____ _... POWER ON OFF OFF 3607 ,488 -1.8 t2.7 ON ON OFF 3550 ,500 -2.9 t3.1 - ON ON ON 3482 ,503 -5.0 +3.7

‘ ALL POWER SETTINGS ARE CONSTANT POWER DEMAND SIGNAL (PDS) MEASURED GAS TEMPERATURE (MGT) LIMIT 1540-F

TABLE IIIA

T700-56-700 ENVIBONMEUT& ICIN5 TEST TEST RESULTS (5.E. B78AE5031) SEA LEVEL I t23OP I 2 gdm- LRK CUSTOW2R BLEED = OX

I- AS RUN DATA I

POVER A1I ICIN5 T4.56 SHP SFC SIiliI’ SEP SFC RATING VALVE SPBAY (OF) (HP) (LBIBBIUP) LOSS X LOSS X 111%

IRP OFF OFF 1544 1550 .487 __.__.._ ON OFF 1544 1410 ,508 140 9.0 4.3 ON ON 1544 1410 ,506 140 9.0 3.9

MC P OFF OFF 1432 1380 .481 _..__.-_ ON OFF 1432 1168 .525 212 15.4 6.0 ON ON 1432 1168 .525 212 15.4 6.0

PART OFF OFF 1350 1200 ,501 ---- __._.--. POWER ON OFF 1355 1200 ,521 4.0 ON ON 1445 1200 ,521 4.0

PART OFF OFF 1269 1000 ,526 ---- .--_---_ POWER ON OFF 1377 1000 ,551 4.7 ON ON 1371 1000 .551 4.1

PART OFF OFF 1197 800 ,572 ---- ____ POWER ON OFF 1315 800 .600 4.9 ON ON 1310 800 .eo0 4.9

PART OFF OFF 1130 600 ,640 ------_ --_- POWER ON OFF 1251 600 .687 7.3 ON ON 1233 800 ,687 7.3 TABLE IIIB

T700-GE-700 EUVIBOWNTAL ICIUG TEST TEST RESULTS (G.B. R76BB(i0311 SEA LEVEL I -4-F I 1 gr"* Lslc CDSTOYER RLEED = OX AS RUN DATA

POWER A/ I ICING T4.5H SAP SFC SW SW SFC RATIUG VALVE SPBAY (OF1 (BPI (LBIMRIBPI LOSS X LOSS X IUC

IRP OFF OFF 1501 1641 ,485 ---- __._.__. ON OFF 1530 1527 ,497 114 6.9 2.5 ON ON 1530 1547 ,502 114 6.9 3.5

MC P OFF OFF 1432 1548 ,486 ---- _.._____ ON OFF 1432 1363 .503 177 11.5 3.5 ON ON 1432 = 1363 ,505 177 11;5 3.9

PART OFF OFF 1270 1200 ,502 ---- ______POWER ON OFF 1359 1200 ,518 2.8 ON ON 1359 = 1200 ,520 3.6

PART OFF OFF 1195 1000 ,525 ---- ____ ..__ POWER ON OFF 1288 1000 ,543 3.4 ON ON 1288 ,. 1000 ,548 4.4 .___ PART OFF OFF ' 1123 800 .566 ---- ____ POWER ON OFF 1225 800 ,593 4.6 ON ON 1225 800 ,598 5.7 I PART OFF OFF 1047 600 ,642 ---- __-- _.__ POWER ON OFF * 1155 600 ,670 4.4 ON ON 1155 a 600 .670 4.4

AT IRP (DRY) THE ENGINE WAS RUN TO A CORRECTED SPEED LIMIT NQl = 103% AS MEASURED. DATA IS SUSPECT. DATA SHOWN IS ESTIMATED.

TABLE IV

F404-GE-400 ENVIEOUllENTBL ICING TEST TEST RESULTS (G.E. B79AEQ10161 PERFOFXAUCE LOSSES DWIW IcIm CONDITIOUS

ENGINE INLET TEMPERATURE 23-F 15-F -4-F

LIQUID WATER CONTENT 2 gmlm' 2 gm/m' 1 gm/m'

POWER SETTINQ IRP IRP FN.7500 LBS. (XNH=14.500 RPM)

THRUST LOSS (LBS.1 480 TO 770 500 560 AT CONSTANT T5H

ALLOWABLE (10%) 1150 1160 750 THRUST LOSS (LBS.)

SFC LOSS (%) 1.5 TO 3.7 2.6 1.9 TO 2.3 AT CONSTANT THRUST

ALLOWABLE SFC LOSS (%) 10 10 10 28-12

ALTITUDE: SEh LEVEL TO 22.000 FEET HAXIHUH VERTICAL EXTENT: 6500 FEET HORIZONTAL EXTENT: 20 MILES n x ,M

,$ 1.0 c 0.8 *f. 8 0.6 z 0.4 B fi 0.2 3 ZO

MEDIAN DROPLeT DIMCTER UICRONS GEOPOTENTIAL ALTITUDE - FEET i ID-’

FIGURE LA : CONTINDOUS WLXIMlM ICING CONDITIONS (MIL-E-005007E (AS); MIL-E-008593E (AS); FAR 25. APP. C)

ALTITUDE: 4,000 TO 22.000 FEET n HORIZONTAL EXTENT: 3 MILES

MEDIAN DROPLET DIWETER - MICRONS GEOPOTENTIAL ALTITUDE - FEET x 10-3

FIGURE 1B : INTERMITTElUT MAXIMLIM ICING CONDITIONS. (MIL-E-005007E (AS); MIL-E-008593E (AS): FAB 25. APP. Cl

FIGURE 1: .-ICIN6 CLOUDS 28-13

FREE JET

60" DIA.

DIRECT CONNECT

FIGURE 2: KIN6 INLET CONFI6URATIONS SPRAY NOZZLES IOR PLUGS1 WATER r\

FIGURE 3: ICIN6 SPRAY BAR SCHEMATIC 28-14

FIGURE 4: -ICING SPRAY BAR AND MOZZLE

TEMPERATURE DEW POINT SENSOR /DISTRIBUTION SCREEN RAKE

-< X

\ SPRAY BARS HUMIDIFICATION CALIBRATION TV AND/OR HI SPEED MOVIE CAMERA Q

1 ENGINE ,I 4, x TEST \PLEXIGLAS OUCT F I6 U R E 5: TYPIC A I. C A L I B R A TI0 N/T E S’r A R R A NG EM EN T S - LL W w -n a E W c c W

0 100 200 300 INLET AIR FLOW lIb/secl

a. INLET TEMPERATURE CAPACITY

Mean Effective Drop'

P - Eng. - , , -=0 1.0 \ I $12-25) I.',,, ,,,,,, (1 2-40): --I L/,,, , , ,, I I I,,,' ( 10-25) I 0.0 1I J 0 100 200 300

INLET AIR FLOW lIb/secl b. ICING CLOUD SIMULATION CAPACITY

FIGURE 6: NAPC ICING CAPABILITIES 28-16

- CI -0- I- M 10 I I- 20 I- W = 30 -I 40 a= 50 u

0 1 2 3 4 5 MINUTES INTO ICING

FIGURE 7: NET THRUST LOSS DURING ICING F107-WR-400 ICING TEST WITH TOMAHAWK INLE BLEEDlBYPASS DOOR

NO. 2 RAMP NO. 3 RAMP RAMPS LOWERED IHlGH MACH CONDlTlONl

B&EED/BYPASS DOOR

NO. 2 RAMP NO. 3 RAMP RAMPS STOWED [LOW MACH CONDlTlONl

BLEED/BYPASS DOOR

I I I I iI I 11 6 ,I I I I ::

BLEEDIBYPASS DOOR NO. 3 RAMP

t i -1 \ .------. I*=-( ( / --\-_.I -1

FIGURE 8: F14A AIRCRAFT ENGINE INLET DUCT 28-18

both a traversing pitot and static pressure probe and Discussion anemometers as well as a profile pressure rake 1-2 diameters upstream of the engine interface plane to quantify I. G. Bianchini, Allison Gas Turbine the aerodynamic flow field fidelity during the calibrations. During the T 700 icing test, you used a heated bullet nose connected to the engine inlet. Have you performed any 4. M. Holmes, RAE evaluation to investigate the heat conduction effects of the Does your organisation advise on anti-icing methods as well bullet nose on the engine anti-iced surfaces? as performing icing evaluation tests? How do you control humidity? Please comment on any Have you any evidence which shows that ice accretion on problems encountered with the humidity control system. engines& measured in ground-based facilities agrees with flight behaviour? Author: Prior to the T 700 icing test an extensive icing consideration Author: was performed with the facility, whenever there were The Naval Air Propulsion Center serves the US naval temperature profiles at the plane of the engine inlet. No heat aviator through the Naval Air Systems Command. We transfer calculations were performed to determine provide the navy with technical advice pertaining to all conduction heating. However, during the total engine and component systems. Examination of engine anti- environmental icing test, a similar facility heated inlet icing systems represents one facet of the support. system was used. During this test, the FAA ice ingestion TheF 14Ainletconetest wasanexampleofafleet problem condition was performed at both specification kind which was effectively simulated in our icing facility. conditions, throughout the engine power range. This test However, we normally test according to the requirements of required the operation of the engine anti-icing system for the military specification as highlighted in the paper. one minute to determine the worst accretion. The facility system was on for the entire test period. The results show that ice accreted on the engine gas path surface uniformly. 5. V. Garratt, RAE From these results we concluded that the facility system had Is the median droplet size mentioned the same as the mean no adverse effect on the meteorological conditions exposed effective diameter also mentioned? If it is so, how is it to the engine. defined? Steam was the medium used to humidify the :air supply. It What is the droplet size spectrum about the median? was controlled via a close loop electro/pneuniatic control system which used line pressure as a feedback signal. Author: Problems encountered with the control of humidity had For all intents and purposes the median droplet diameter is essentially been related to consistency of the of the the same as the mean effective droplet diameter because thermally heated photo cell sensors' ability to provide generally the distribution obtained from the FSSP laser accurate point temperature measurements. probe is a normal one. However, it is the mean that is calculated. The probe counts the droplets in 3 pdiameter 2. W. Alwang, Pratt and Whitney increments and the diameters averaged. How did you calibrate the collection efficiency of the For a 20 pm average droplet diameter, the droplet measured rotating cylinder? ranged from 3-SO pm.

Author: 6. I? Derouet, SWECMA Reference 5 of this paper is a report in which NAPC Est-ce que les appareils que vous utilisez aujourd'hui pour measured LWC usingasingle rotatingcylinder. In this report mesurer et rigler les parameter de guirage (diamitre goutte, there is a plot of collection efficiency versus velocity for quantiti deau) donnent les mimes risultats que aux utilises average drop diameters and given cylinder diameters. The dans les annLes 1950 pour Ltablir les normes des nuages average drop diameter is determined from the FSSP laser guirants? probe applied to this plot along with measured duct velocities to determine the collection efficiency. The probe which is used by NAPC today to measure the icing parameters of droplet size and relative moisture 3. C. Scott Bartlett, SverdNp Technologies content is designed to be an airborne cloud measurement Would you please comment on~thework performed to instrument. The results which have been gained from this ensure that the aerodynamic flow field fidelity is maintained probe through in-flight measurements are representative of for your subsonic free jet test installations? the engine requirements for specification of intermittent and continuous maximum icing conditions. Airborne test results Author: for droplet diameters show that they are within the average For the subsonic free jet icing installations, NAPC had used level of data scatter. 29-1

ICING RESEARCH RELATED TO ENGINE ICING CHARACTERISTICS

by SJ.Riley BTech MSc EO. Box 31, Rous-Royceplc Moor Lane,Derby DE2 8BJ United Kingdom

SUMMARY Physical properties and characteristics of ice formed by accretion have been investigated experimentally to provide a database relevant to civil turbofan engine and powerplant surfaces. This paper Key Nose cowl summarises part of that work, relating to 1nt*e probe unheated surfaces, including observations Spinner of ice accretion on various bodies over a Fan blades range of conditions and measurement of the Inner barrel Engine seCiion stator9 adhesive strength of ice samples. Guidevanes Bypas dud inlrumenlation NONENCLATURE Core engine dudng

g acceleration due to gravity FIGURE 1 Engine and nacelle icing h height of manometer liquid p static pressure considerations P dynamic (total) pressure v air velocity pa air density o ice accretion on static vane sets pl manometer liquid density o ice accretion on an engine intake leading edge o icerwater formatigns at a total ImoDlJ~IoN temperature of f2 C Ice accretion on airframe leading edges and forward-facing surfaces has been of o adhesive strength of ice samples increasing concern during recent years, on a moving collector. since it can degrade aerodynamic Further research on ice shedding performance and represent a potential source of ingestion damage, thus causing a characteristics, sublimation, insulation hazard to the safe progress of the effects and flowfield changes (drag) due aircraft. The solution of the problem as to the presence of ice would be useful to a safety issue is not straightforward, develop understanding of the physics. since any anti-icing or de-icing provision represents an engine performance penalty CALIBRATION OF ICING WIND TU"J%L which must be balanced against powerplant performance deficit that may occur without For all the studies covered in this paper, icing protection. Such performance a blown diameter sea level wind tunnel based at Rolls-Royce Hucknall was used. deficits could be generated by drag of The tunnel is fitted with water spray iced surfaces or by mechanical deformation nozzles which can produce a supercooled of compressor blading caused by major ice water cloud of controllable droplet size sheddina. It is desirable to determine and total liquid water input to simulate firstly-which surfaces require protection an atmospheric icing cloud. The tunnel and secondly the level of protection airflow and water spray beyond the tunnel necessary to optimise the balance between exit plane, in the region of the test real and potential Performance loss. In piece working section, were investigated order to acquire a database to enable such by recording the air temperature, air assessments to be efficiently and pressure (velocity) and liquid water accurately carfied out, Rolls-RoYCe has content distributions at four planes investigated icing phenomena and ice downstream of the tunnel fitted with a 280 protection systems appropriate to civil mm (11 inch) reducing nozzle. Figure 2 is turbofan engine and Powerplant Surfaces a diagram of the test configuration and (Figure 1) by means of Several measurement planes. The pressure and experimental research packages. This temperature profiles were investigated by paper summarises part of that work, means of pitot and thermocouple rakes, relating to ice properties and formation Figure 3. The pressure rake comprised 25 on unheated surfaces. The following pitot rakes spaced one inch apart centre experimental investigations are discussed: to centre interspersed by six probe type static tappings along a horizontal bar. o calibration of che icing wind The 25 thermocouples on the temperature tunnel used, for a range of rake were also spaced one inch apart along simulatcd ambient parametcrs a suDDortina bar. The rakes were moved in the verticai plane by means of a motor, o macroscopic and microscopic and their position indicated on a scale on observations of ice formed by the test cell viewing window by a pointer accretion at various conditions on the end of the pressure rake. 29-2

FIGURE 2 Icing tunnel calibration measurement planes

effect of varying ambient conditions.

Renentabilitv was~~ oood in the~~~~ hioh- velocity core. The outermost Lemperacures varied wich rime according to test cell conditions, buc variacion during a givcn scan was negligible. Sufficient time was allowed for stabilisation at each test point. A contour pl-otting program was used to plot results as contour maps for each test condition at each measurement plane. Figure 4 shgws the resultant pressure contours at 0 C in terms of pressure heads (inches of water) to 12 inches either side of the centreline. The geometric centrepoint of the tunnel is marked +. The pressures plotted on Figure 4 are related to velocities via a simplified Bernoulli's equation:- 2 P - p = 1/2 pa V and P-p=plgh hence V = 2 g h pa Figure 5 shows sample temperature contours in degrees centigrade to 12 inches either side of the centreline. Inspection of FIGURE 3 Instrumentation rakes mounted Figures 4 and 5 shows that the flow downstream of wind tunnel remains in a core formation up to at least 1.2 m from the nozzle exit plane, and the temperature of this core flow approaches Readings were taken at one inch intervals the nominal air temperature. The size of from below the tunnel upwards to avoid the.... core~~~~ diminishes with~ ~~~ downstream ~~~ ~~~ back-lash in the traverse screw thread and distance. The circular nature of the advancing mechanism.The vertical extent of pressure contours indicates that the flow measurements was chosen to encompass produced is axisymmetric, but the flow is significant pressure and temperature distributed about a point offset from the gradients, with outermost readings geometric centrepoint. This is due to the approaching ambient conditions in the test proximity of the test cell wall (to the cell. left of plots Figures 4 and 5 as viewed) and floor. Calculations suggest that the Readings were taken at nominal air actual velocity of the core flow exceeds velocities of 61 and 122 m/s (200 and 400 the nominal tunnel velocity. As distance fps) at the four measurement planes $,or from the exit plane increases the size of tots1 tunnel air temperatures of 0 C, this 'high velocity' core diminishes. For -15 C and -3OOC (pressure readings were a given nominal velocity, the local air taken at OOC only sinceo the diffgrences velocity more than 150 mm (6 inches) observed at plane A for 0 C and -15 C were radially from the centre of flow is not negligible. Sample rake positions were significantly changed with downstream retested at intervals to check distance, i.e. velocity and temperature experimental repeatability and note the gradients are greatest near to the nozzle. 29-3

PLANE A PLANE B PLANE C

v = 122 m/s FIGURE 4 Pressure profiles

PLANE A PLANE A PLANE C PLANE C v = 61 m/s r-v = 61 m/s v = 122 m/s p ..-

: -15OC

T = -3OOC FIGURE 5 Temperature profiles

Research was carried out in the Hucknall which shed more easily and was generally facility to determine the effects 8f more opaque. The contact area with the certain variables on ice accreted at -10 C grid was less for the higher speed and the on a steel tubing grid. Air velocities classic forward-facing horned formation tested were 61 m/s and 122 m/s at the wind was better defined, this being more tunnel exit plane. For corresponding test apparent with increased water input. conditions, it was found that the higher Figure 6 shows sample photographs of ice tunnel airspeed produced more brittle ice, accretions, and Figure 7 shows the difference in cross section at various test conditions. 29-4

4 gph, >20pm 4 gph, <20 um 4 gph, >20 um 4 gph, (20 pm 61 m/s 122 m/s

FIGURE 7 Profiles of gri.d ice accretions after 5 minutes at -1OOC was more opaque and featured a reduced As water input increased, i.e. simulated cloud liquid water content increased, the contact area and better defined horned amount of accreted ice increased formation. The ice formed on the proportionately. However, a reduction in collection grid was removed and weighed liquid water content associated with an after each test, and it was found that for increased exposure time to the icing a given total water input, a lower water condition had a similar effect to flow rate resulted in a greater quantity increasing velocity in that the resultant of ice due to the different transportation and heat transfer phenomena at the surface ice was more brittle, shed more easily, of the accreting ice. 29-5

Experimental evidence showed that a collecting surface and surroundings, and reduction in water droplet size had a any external forces prevailing. The similar effect on ice type as an increase variables contributing to a given icing in airspeed or reduction in liquid water condition include air temperature (static content, producing more brittle, opaque and total), air velocity, cloud liquid ice which shed more easily. water content, water droplet size and distribution, cloud extent (time in icing) Ice formed at different temperatures and altitude. The pertinent collecting displays different characteristics. There surface characteristics may be considered are two categories of ice normally quoted; to be rotational speed, heat input, shape, rime ice and glaze ice. Rime ice forms at material and surface finish. AS an low temperatures, is smooth, opaque and initial step towards understanding the tends to follow the shaDe of the surface physics and mechanics of the aforementioned phenomena, ice was formed on a stainless steel collector mounted downstream of the Rolls-Royce Hucknall 15 inch icing wind tunnel for a range of horned shape. In the case of a point airspeeds, air temperatures, liquid water attachment this type of ice forms a contents and droplet sizes. The samples rosette growing forwards and sideways. were observed manually, photographed in Examination of results discussed above close-up using a Hasselblad camera plus shows a tendency towards glaze ice bellows. and DhotoaraDhed throuah a characteristics rather than rime for an microscope using -a LeCtz camera attachment increase in airspeed, reduction in droplet plus light box. size or reduction in liquid water content. At the intermedi3te test total air The collector was cleaned prior to each temperature of -10 C, it will be noted test, and the tunnel run prior to the that either type of ice or a combination introduction of supercooled water to allow may form, dependant upon ambient temperature and airspeed to stabilise. conditions. Glaze ice is generally a Ice was collected for one minute at each greater problem in respect of aero-engine intakes since cloud liquid water content is greater at higher temperatures and the glaze ice shape will have a larger effect 'C and wate on drag and damage potential if the ice is (2 to 8 gph). Most tests were at a released. However, glaze ice tends to be nominal water droplet size of 20pm. but more brittle with a smaller contact area some tests were repeated at smaller and and is more readily shed. The only larger droplet sizes. deviation from transition between rime and glaze ice noticed during the experimental Following manual and close-up photographic investigations was an increase in opacity observation, each sample was removed for rather than an increase in clarity. The microscopic observation at 100 times physical formation characteristics magnification. It was not possible to distinguishing the two ice types are as make microscopic observations of ice follows:~ ~~ the ~~~~ maior ~ influencina~~~~~~~ factor is formed at 0 C due to rapid melting. The che freczing fraction: che proportyon OE wind tunnel was completely run down to che impinging supercooled waccr droplets avoid vibration which could be transmitted which freeze instantaneously on impact. to the microscope. There would have been For pure rime ice the freezing fraction is little advantage in setting the microscope 1.0, the resultant ice being opaque. At arrangement up in the test cell, since the low freezing fractions there is some water main contributory factor to increased flow along the surface prior to freezing, melting was the light source, and forming clear glaze ice. The post-impact condensation would have been a problem. water flux gives rise to the horned or rosette type growth of glaze ice. The Sample close-up photographs are given as associated increased frontal area and Figures 8 and 9. transient local heat and mass transfer coefficient distributions escalates the Figures 10 and 11 show microscopic photo- deviation from the original collector graphs corresponding to Figures 8 and 9. surface shape. As temDerature increased. the surface~~~~ The size and location of a test piece struccu;c of the ice became more uniform. should be chosen to lie within the No indicacion of che infrastructure could reasonably uniform core flow. However, be gained due parcly cc melcing and parcly the presence of a model will deviate the to inadequate magnification. Surface flow thus effectively increasing the size observations only could be made with of this core, and heat transfer phenomena equipment available. As airspeed at the model surface will influence the increases, water droplet Reypolds number

~~~ effective local air temperature e.g. and inertia Darameter (Bowden~~~~~ ,1 increases. viscous heating. These changes have opposing effects on collection efficiency, limiting the net MACROSCOPIC AND MICROSCOPIC OBSERVATIONS effect on ice accretion. From results it OF ACCRETED ICE was seen that as airspeed increased the ice became more opaque and brittle. This The appearance, net quantity, shape, observation was consistent with those made distribution, relative location, physical during the calibration research described properties and shedding propensity of ice earlier. As water input increased the formed by accretion are a function of amount of accreted ice increased ambient conditions at the time of proportionately. However, as previous formation, characteristics of the research has shown, a reduction in liquid -30°C -20°C -1OOC ooc FIGURE 8 Ice formations at 30 m/s. 20pm droplets, 5 q/ s water

at 91 m,

0 -2OOC -IS? -5 c OOC ~-FIGURE 9 Ice formations IIs, 20 um droplets, 5 g/s water

I ..I. ' .. .i

-. . .. , ... '.,......

FIGURE 10 Microscope results at 30 m/s, 20 um droplets, 5 g/s

FIGURE 11 Microscope results at 91 m/s, 20 Um droplets, 5 g/s 29-7 water content associated with increased vanes was contained in wooden ducting with exposure to give an equivalent total water a perspex viewing panel. The vanes were input haii a similar effect to increasing of constant profile, representing approx- velocity in that the resultant ice was imately the mid section of typical inlet more brittle, more easily shed, more guide vanes. The trailing edge of each opaque and features a reduced contact area vane was thickened to enable fitment of and better defined horned formation. For locating pins. Tests were conducted: a given total water input, reduced liquid water content resulted in a greater (A) with the 25 mm vane set in-line quantity of ice. Similarly, a given at Oo vane incidence quantity of water input at a higher velocity gave a greater ice accretion; (B) with gll vane sets at 23O to (A) this was also an effective reduction in and 0 vane incidence, liquid water content since volumetric flow to simulate fan swirl through the wind tunnel was increased. Water droplet size affected ice collection (C) as configuration (B) wgth the efficiency (quantity) and the type of ice 38 mm vanes skewed i20 . formed, although the effect was muc than that of other parameters. At low Prior to icing tests, the velocity temperatures, rime ice is gormed I profiles upstream of the vane sets were temperatures approaching 0 C glaze ice checked using a traversing pitot rake forms. The horned formation is less comprising 21 tappings connected to water apparent for a flat collector, peripheral manometers. The profiles were found to be ice growth providing the best indicator. uniform over the majority of the test Results clearly showed the transition from section, with negligible wall effects. rime to glaze ice characteristics as The rake was removed during icing tests temperature increased. At intermediate and replaced by a sealing plug in the temperatures, either type of ice or a ducting lower wall. Pre-test ice distri- combination formed on the unheated bution checks indicated satisfactory water collector. Examination of experimental (ice) distributions, even with one spray results also showed that the tendency nozzle operating only for low water input towards rime or glaze-type ice is affected conditions.~ ~~~~~~ ~ ~~~~ ~ Each test build~~ ~ was also by airspeed, droplet size and liquid water checked for leaks. The tunnel air content. Microscopic observations temperature was allowed to stabilise at indicated an associated surface area and the desired velocity before supercooled roughness change consistent with the water was introduced at each test foregoing discussion, i.e. as temperature condition. Results were recorded at increases freezing fraction is reduced and one-minute intervals or as appropriate via the droplets spread laterally on impact static pressure tappings connected to before freezing, giving the more open water or mercury manometers according to texture. the magnitude of pressure drops across the model. Still photographs were taken These investigations concluded that: remotely during each test and video monitoring was used, selected tests being The physical and mechanical recorded. Times of major ice sheds were properties of ice formed by noted. accretion are a function of formation conditions. Tests were 30 minutes long unless either ice blockage rendered it impossible to The effects of altitude, maintain conditions or an ice accretion rotational speed, vibration, and shedding cycle had been established. heat input, material type and The highest scheduled airspeed of 137 m/s surface finish may also (450 fps) could not be achieved due to influence ice properties, and these phenomena should be blockage by the vanes themselves, and 119 investigated. m/s I392 fps) was used as the practical maximum. For most builds temperatures of Alternative techniques fox -15OC. -6OC and -2OC at 61 m/s.. (200 fnsl~- . microscopic observations and 913m/s (300 -fy) air velocity and should be investigated. 0.15g/m and 0.3 g/m liquid water content were tested, although intermediate ICE ACCRETION ON STATIC VANE SETS temperatures, the higher velocity and a wider rang5 of liquid water content (0.1 Excessive core engine blockage due to to 0.6 g/m ) were tested for 25 mm and 38 icing of engine section stators can be mm angled vane sets. The swivelling vanes detrimental to engine performance, giving were tested at incidences of -20° Lalmosg a reduced flow area, increased pressure Darallel to the airflow). -10 . 0 loss and consequent power loss and must be idatuml, loo and~-2Oo~.-The 'iatter-daused avoided. However, it is desirable to maximum blockage by the vanes, limiting maximise the number of fan outlet guide achievable tunnel air velocity. In all vanes and compressor inlet guide vanes for tests the nominal water droplet size was optimised noise suppression. Therefore 20pm. model static vane sets of 19.1, 25.4, 38.1 and 55.9 mm 10.75, 1.0, 1.5 and 2.2 in) To make comparisons easier, results have spacing were subjected to icing conditions been plotted non-dimensionally in the form in the Hucknall 15 inch icing wind tunnel to investiaate the Dhenomena involved. NONDIMP = AP - APo where AP = measured The effects of fan -swirl and relative APo pressure drop angle of the vanes to the flow were also and APo = initial AP investigated. Each set of stainless steel 29-R

Figure 12 shows results at 0.3 g/m3 for from which it had been shed and an all spacings angled to the flow and Figure adjacent vane, providing a further 13 shows results for in-line and accretion surface to exacerbate the swivelling vane sets. build-up. Initial pressure drop across the vane sets varied considerably with Ob8ervations showed that ice formed at vane spacing and angle relative to the -2 C grew forwards and laterally (glaze airflow, but it must be assumed that this ice) and as temperature of formation 'blockage' due to the vanes is acceptable reduced, frontal area reduced (rime-type for the particular application. In many ice). In general AP increased with tests, for closely spaced vanes and low temperature, reflecting the different ice temperatures, the pressure drop across the types, but blockage was not necessarily vane set decreased slightly before greater at higher temperatures since increasing. This was apparently due to sheddina__ DroDensitv - also increased. For slight streamlining of the blade, caused similar reasons it did not necessarily by the initial (rime) ice formation. From follow that an increase in airspeed (and inspection of Figure 12 it may be seen associated increase in water input for a that blockages of 19 mm, 25 mm and 38 mm given liquid water content) caused spaced vanes were of the same order and increased blockage. Having noted that that 56 mrn vanes were much less shedding was a major contributory factor susceptible to blockage. From a in net ice accretion/blockage, shed ice comparison of Figures 12 and 13, no must pass between the blades to be conclusion can be drawn concerning the completely removed. For closely spaced benefit or otherwise of the presence of vanes, ice often wedged between the vane swirl due to the fan. Figure 13 shows that the amount of skew on the vanes 29-9 affected blockage greatly. This was because oblique vanes relative to the airflow themselves represent a large blockage, so additional blockage due to

~ice is ~~ low as a nercentaae of the initial value. In all cdses it was noted that ice on the leadlng edges tended to 'shield' pressure surfaces, and any ice located there shed and did not rebuild unless the leading edge ice shed. This is not apparent of fan blades, where the relative

-Dressme ~~ ~ surface area is larae. leadina edges thin and the blades are 'presented obliquely to the flow at their hubs. The pressure levels at which shedding occurred increased throughout a given test. This is thought to be due to initial sheds FIGURE 14 Typical runback ice removina the weaker ice. leavina the stronge; ice as a base for. new accretion. ICEJWATER FORMATIONS AT +2OC Once the initial more frequent shedding was complete, the shedding appeared to The purpose of these tests was to observe occur in a cyclic fashion unless build-up the interactive phenomena of ice and water was such that the ice could not shed on a typical turbofan intake surface and because it had nowhere to move to, e:g. to determine the effect of controlled ice bridging adjacent vanes formed a firm parameter changes on ice formation. At accretion base and blocked the corres- temperatures around O°C there is an ponding passageway. intermediate situation whereby both ice and running water exist on the surface. The investigations concluded that: Prediction of the physics, mechanics and hydrodynamics of this phase is a compli- o Closely spaced vanes are more cated heat and mass transfer balance. prone to blockage, and ice bridging, but initial pressure A full-scale two-dimensional section of a drops were similar. typical aero engine intake was mounted downstream of the Rolls-Royce Hucknall 15 o Blockage is highly dcpendant on inch icing wind tunnel and enclosed in shedding propensiry. AL ducting, the walls of which were shaped to conditions tested, significant simulated the airflow distribution for a blockages occurred. tvpical descent flight condition. The model was fitted with-a mock-up section of o Vane angle relative to the flow anti-icing system, comprising a length of was significant mainly 'piccolo' pipe featuring three rows of due to Fan had holes such that air jets impinged on the no quantifiable effect. intake model internal surfacs. A tunnel total air temperature of f2"C and water The data acquired must be considered in droplet volume median diameter of 20um conjunction with icing tolerance were used throughout testing. Ice and characteristics of a particular engine, water formations and behaviour on the e.a. surae marain. imDact resistance. intake.section highlight region and inner ingestion- oapab

compared with no flow. An increase ir quantities are dominated by evaporation airspeed at a given water flow rate and surface tension effects, but the (reducing liquid water content) affected presence of ice modifies water flow the type of ice formed but not the volume patterns. These phenomena produce a accreted prior to major sheddhg i.e. a transient sit:uation at the surface, but critical mass was reached. This is alsc the process is cyclic. The observations true up to a point of increasing water made were considered in the development of input, beyond which the increased water analytical models of icing behaviour, and flow limits ice formation, i.e. washing assisted in the interpretation of full away ice formed. Theoretically, more ice scale aircraft and engine tests in ice would be expected to form as airspeed forming conditions. increased since this effectively reduces the static air temperature, hence the ADHESIVE STRENGTH OF ICE FORMED BY reason for no ice formation at the lowest ACCRETION airspeed, when the static air temperature was positive. For higher airspeeds static Net ice accretion, as discussed air temperature was negative. previously, is a function of ice collection and shedding propensity, which Water rivulets observed were a result of in turn is related to ice mass, forces surface tension effects. Rivulets were acting upon it, shear strength of the ice

seen uDstream of~~ the ~~~~ anti-icina ~~~~~~~~~~~ ~ air and the adhesive bond strength at the exhaust- slot^ and were therefore not the ice/surface interface. Experimental result of effects afforded by the slot investigations were carried out to itself or the air issuing from it determine the bond strength of ice built up on rotating collectors mounted down- The presence of ice on the highlight of stream of an annular nozzle attached to the intake affected airflow around the the Hucknall 15 inch icing wind tunnel. model, confirmed by the almost concurrent The ice mass resulted in a centrifugal shedding of nose ice and surface ice. The shedding force which was recorded by ice acts as an insulator as indicated by a strain gauges on the collector arm front thermocouple inside the model. In some and back faces. Figure 15 shows the rig instances surface ice slid in a downstream schematically, direction, indicating that the interface had melted. In all cases ice build up and Two 25 nun (1.0 in1 square titanium

sheddina- followed a well defined~ ~~ ~~~ cvcle.~_~~~, collectors were mounted at a radius of 275 the cycling times being fairly constant nun (10.8 in), adjustable from normal to and a function of airspeed and water the tunnel airflow to 30° rotation. The input, and possibly partly du.e to the collector surfaces were degreased and grit effective air velocity change arising from blasted to a consistent and recognised increased ducting blockage with ice standard. The tunnel airspeed, drive present relative to the bare model. motor rotational sDeed and collector anale were arranged such that the impingement This aualitative testina- indica.ted that. velocity was always normal to the near the freezing point, the net coverage collector face. of an aerofoil by ice and/or water is the result of the interaction of several When ice formed on the collector with the effects. Additionally, ice accretion is rig spinning, the forces on the arm were: self limiting owing to insulation effects and flow pattern alterations due to shape (i) centrifugal force (CFI due to changes. Water flow patterns and the collector mass

-FIGURE 15 Ice adhesion rig 29-11

(ii) centrifugal force due to the greased) and coatings (PTFE and GRP). ice mass After the first ice shed of a given test, (iii) moment due to the collector/ the surface nature of the base on which ice CF offset from the arm subsequent ice built would not be as centreline pre-test, so most tests comprised more (iv) moment due to aerodynamic than one shed, until ice shed from the forces on the collector tunnel walls affected the results. (probably small compared with (i) to (iii)). Numerical values of average and maximum shear strength were calculated from strain This combination of forces resulted in a gauge readings. Figure 16 shows these net tensile force on the forward face of results against temperature for the bulk the collector and a compressive force on of tests. The mean values shown may be the rear face. When ice shed, the lower than the true mean in some cases resolved force was reduced and the owing to the inclusion of results from resultant change in strain was used to partial ice sheds not recognised as such. calculate the force which caused the The maximum shear strength values are shedding and hence the shear strength of dependant on the number of tests at a the bond and ice. It should be noted that condition. results would be affected to some degree by the presence of the strain gauges It has been suggested that, because the themselves. The strain gauge data and centre of mass of the ice sample was thermocouple readings were continuously offset from the shear face, the ice would monitored via a U/V chart recorder. A tend to peel rather than shed in pure slip ring unit was used to transmit shear. This was limited by not allowing measurements from the rotating shaft to the ice sample thickness to exceed 12 mm the recorder. Stobe-triggered video (0.5 in), but there is no evidence to cameras were used for observation of the support or dismiss the hypothesis that the icing process, and a still camera was used shedding mechanism was either entirely to record the condition of the ice peeling or initiated by peeling. fracture surface immediately after shedding. The size and shape of the ice collectors were not representative of a fan blade Pre-test checks on velocity (pressure) and section or other rotating engine water distributions showed that velocity component. There are two scaling effects; was even over the area swept by the firstly the aerodynamic forces increase as collectors, and water distribution was the size of the accretion surface is even after modifications to the grid. A reduced owing to the relative proportions deflector plate was added to avoid ice of the ice and 'clean' surrounding surface accretion on the arms and bridging to and secondly ice gains strength from provide possible strengthening. A large adjacent ice. The shape of the surface number of tests were carried out at each affects the aerodynamic characteristics condition to establish statistical and ice collection efficiency. credibility because of high scatter due to variations in accretion and shedding times The mechanism by which ice sheds (and hence ice mass) and experimental (adhesive/cohesive failure) is determined errors, although accretion and shedding by the fracture mechanics involved, and no were essentiylly cyclic. Most tests were attempt was made to study the nature of for 1.0 g/m liquid water content, 20um the adhesive bond between the ice and the droplets and a clean surface with impinge- metal surface. It was noted, however, ment velocities normal to the collectors that failure tended to be cohesive, i.e. of 61 to 213 m/s (200 $0 700 fps) and failure occurred within the ice, at lower temp%ratures of -2 to -20 c (note that for total air temperatures. At higher a 30 collector angle the normal impinge- temperatures adhesive failure occurred, ment velocities were twice the tunnel i.e. the bond between the ice and the airspeeds). Additional tests were carried metal surface was broken. The trpsition out to investigate the effect3 of liquid occurred between -6 and -10 C air water content (0.5 to 2.5 g/m ), droplet temperature, which also corresponds with size (A5 to 30um), impingement angle (30 the transition between rime and glaze ice. and 35 ), surface preparation (clean and The ice type and failure mechanism did not

4 -.- (e ... m......

I. -. -......

.-.. . _", .,.. .". .I.. 4. I. .+. L .... --c ...

Average shear strength. Maximum shear strength. FIGURE 16 Adhesive strength results 29-12

appear to significantly affect results o The research summarised in this paper cxceut that an increase in aradient of the has provided a significant quantity of strength-temperature curves-was noticeable data which has improved knowledge and at lower temperatures. Figure 16 indi- understanding of phenomena associated cates that shear strength increased as air with ice formed by accretion. temperature at formation reduced, approximately linearly. The impact speed had a 0 Further investigations are required in significant effect on shear strength, order to provide a more comprehensive higher accretion velocities producing understanding of the physics involved. lower strength ice, probably due to increased aerodynamic forces, reduced o The data available has been used in the capability for trapped air to escape, design, development and in-service increased droplet spread prior to freezing phases of turbofan installations. and changes in water flow characteristics at the ice surface. A reduction in RE!?ERENCES droulet diameter auuarentlv increased

shek strength, which- 1s in opposition to 1 Bowden et a1 'Enainecrina summarv. of~~~ the findings of calibration tests airframe icing ccchnical described earlier. This has been daca', TR ADS-4, ?larch attributed to the flat collector versu 1964. the smaller contact area grid. A 58 change in impingement angle did not have Discussion any quantifiable effect on shear strength, A but this is a relatively small change. 1. V. Naval 'greased' surface gave lower adhesion. A Truglio, Air Propulsion Center PTFE coating reduced shear strength and Regarding your icing test rig, is there any compensation GRP reduced it further. Liquid water necessary in the generation of liquid droplets due to what content variation had no significant seems like relatively short time between the nozzles and effect on strength. tunnel exit plane? Also is there any consideration to humidity effects'? The results of investigations were used to validate a fan blade ice accretion and Author: shedding model. For predictive purposes, Investigations (theoretical and experimental) were carried the averaaed data showed that strength to of decreases with total air ggmperature out determine a minimum distance downstream the increase at a rate of -30 kN/m C and the spray grid. A model had to be placed to ensure that the maximum Shear strength falls at a rate of droplets were as required. A limitation is the physical length -85 kN/m C. of the test cell, but the rig is being refitted with this criterion as a design feature. Humidity has been measured but is not a problem due to the MNCLUSIONS tunnel configuration, with heat exchangers drained The following conclusions may be drawn upstream of the control room and the fact that the entire from the work covered. room soaks down to the test temperature. MODELISATION NUMERIQUE DE L'EVOLUTION D'UN NUAGE DE GOUTTELETTES D'EAU EN SURFUSION DAN8 UN CAISSON GIVRANT

CREISMEAS paul et COURQUET Joel D.G.A. O.N.E.R.A./C.E.R.T. C.E.Pr. Saclay 2 Avenue Edouard BELIN b.p. 4025 91481 ORSAY-CEDEX 31055 TOULOUSE-CEDEX

~ Dans un caisson de givrage, l'injection se fait en premelange Depuis peu, le C.E.Pr. a et en amont de la maquette dans un developpe pour ses besoins soucis d'homogeneite.Cependant, la d'essais en qivraqe un outil de distance separant la naissance du calcul qui lui permet de predire nuage et l'impact sur la maquette l'evolution de qoutelettes d'eau &ant de l'ordre de 4 a 5 m, il en surfusion dans un ecoulement peut se produire une modification d'air froid. du D.V.M. du nuaqe par effet d'evaporation. Ce code, baptise M.A.GI.C. (Modelisation et Analyse du I1 est alors utile de GIvrage en Caisson) a des qualite disposer d'un outil predictif qui de robustesse et de souplesse puisse quantifier une telle d'emploi que l'on attend d'un code 6volution.Pour ce faire, il industriel. devient necessaire de modeliser le plus completement possible les Afin de confronter les &changes thermiques entre la phase resultats numeriques obtenus par liquide et la phase qazeuse, en M.A.GI.C. avec des mesures prenant en compte l'hygrometrie. physiques, une analyse a ete menee en prenant pour reference des Le code numerique M.A.GI.C. resultats de granulometrie sur un (Modelisation et Analyse du spray de qouttelettes dans une GIvrage en Caisson) a ete conp soufflerie de laboratoire. dans ce but au C.E.Pr.. Le resultat de la Etant donnees les confrontation s'est avere tres incertitudes qui pesent sur les satisfaisant. differents coefficients d'echange liquide/vapeur, il s'est aver6 ABSTRACT necessaire de confronter les a Since a few time C.E.Pr. has resultats numeriques des developed a computation tool resultats experimentaux. for its own icing tests needs. a notre connaissance, il n'existe pas dans la litterature This tool allowed the de resultats de mesure prediction of supercooled droplets experimentaux d' evolution de evolution in a cold air flow. diametre de gouttelettes d'eau , This program is named faisant intervenir l'hygrometrie. M.A.GI.C. and has qualities of Cette constatation nous a stability ans adaptability amene a concevoir un montage expected for an industrial experimental et realiser une program. serie de mesures portant sur le spectre de diametre d'un nuaqe et In oder to compare numerical sur le diametre volumique median. results from M.A.GI.C. to physical measurements, an analysis based on Les resultats de ces mesures qranulometry measurement ant pu Ctre directement compares a concerning droplets inside a ceux du calculs numeriques du laboratory wind tunnel is codes M.A.GI.C., et la comparaison performed. s'est avere tres satisfaisante. The results of the comparison apres avoir presente les are very acceptable. principales caracteristiques du code M.A.GI.C.,nous confrontons 1-INTRODUCTION les resultats de calcul sur la mise en vitesse de gouttelettes a Lors de la certification quelques resultats issus de la aeronautique de materiels volants biblioqraphie. en vol qivrant simule, il est indispensable de respecter les Dans le paragraphe suivant, normes de simulations editees par nous introduisons l'environnement l'organisme certificateur [I]. experimental qui a permis les mesures, et nous comparons ces Ces normes permettent de resultats a ceux de M.A.GI.C.. calculer les differents parametres necessaires l'exdcution des 2-PRESENTATION DO CODE l4.A.GI.C. essais de qivraqe [2], et en (Modelisation et Analyse du GIvraue en particulier le diametre volumique Caisson) median du nuaqe givrant. 30-2

Le code M.A.GI.C. a et6 conqu f = 1.56 + 0.61 ( sc 1/3 Re '/* pour predire l'evolution des pour (Sc 'p3 Re 'I2)> 1.4259 differentes classes de diametre qui composent un nuage de Sc : nombre de Schmidt defini par: gouttelettes. Les champs aerodynamique, thermique et hyqr0metriqu.e etant donnes, on utilise une approche : Lagranqienne pour determiner la Dif diffusivite massique de la vapeur dans l'air entre -40 C et position de chaque classe de s-l gouttelettes.Celles-ci peuvent 40 C en m2 3 &re SOUS &tats possible!;: Bilan de mantite de mouvement : . liquide (eau) Le bil.an de quantite de . Solide ( glace) . liquide+Solide (melange mouvement s'ecrit: eau/qlace) . Seule la phase liquj.de fait l'objet d'une etude comparative Dt dans cette publication. Au cours de leur parcours, les gouttes subissent l'influence A chaque pas de temps, on de differents efforts qui calcule par interpolation lineaire s'exercent sur elle. les caracteristiques aerodynamiques, thermiques et Ces efforts sont dbs: hyqrometriques au voisinage des qouttelettes. . a la trainee . a la pression engendree par on suppose que la phase le champ d'acceleration de liquide, de par son evaporation l'aerodynamique ne modifie pas ces . a ].a poussee d'Archimede caracteristiques de l'air. . a la force de Basset qui tient compte de l'histoire du on resout alors explicitement mouvement de la goutte. les trois bilans suivant: En general, seules les forces Bilan de masse: de trainee sont prises en compte: l'ensemble des autres efforts etant negligeable dans le cas ou la masse volumique de la particule est plus de 100 fois superieure B celle de l'air. avec Le bilan de quantite de .Kx: coefficient d'echange de mouvement se met sous la forme: matiere eau/air .Mv : Masse molaire de vapeur .pi D2 : surface d'echange eau/air .xvs : fraction de vapeur saturante Le coefficient de trainee est .xv : fraction mol.aire de calcule par: vapeur dans le flux cx = (24/Re) (1+0.15 La fraction molaire de vapeur est donnee par l'expression Re : Reynolds associe la suivante: qoutte w=HrPvs/P .Hr : hyqrombtrie au voisinaqe de la qoutte .ws : presslon de vapeur bilan d'enerqie saturante au voisinage de la qoutte .P : pression statique au voisinaqe de la qoutte : Le coefficient d'echanqe KX se . m masse de la qoutte calcule de la manibre suivante: . D : Diametre de la goutte . TS : Temperature de surface de la qoutte . H : Enthaloie massiaue . Cpl: Capacite calorifique de la goutte liquide avec . Tg : Temperature du qaz porteur . Hv-H1: chaleur latente de f = 21 0.216 ( SC 'I3 'I2)2 vaporisation pour (Sc 1/3 Re '?')< 1.4259 30-3

3- LE caLcuL DES LOIS DE TRAINEE DANS NOUS presentons dans les LE CODE l4.A.GI.C. figure 3-2-2 et 3-2-3 une comparaison entre la mise en 3-1 emression du coefficient de vitesse des gouttes calculee par trainee M.a.GI.C., avec celles donnees Par les references [71. Une expression generale du coefficient de trainee peut se Les conditions initiales sont mettre sous la forme [2]: presentees dans le tableau suivant: Cx=f (Re,We, Sc,Pr, 8)

Re : nombre de reynolds qui vitesse de l'air caracterise le regime de la goutte . we : nombre de weber qui identifie l'influence de la deformation de la qoutte. on observe que l'allure des sc et pr : respectivement nombre courbes est conservee, mais de Schmidt et nombre de Prandt qui M.A.GI.C . a tendance a sous caracterise les echanges massiques estimer nettement la mise en par convection ou diffusion, ainsi vitesse des gouttes par rapport que les dchanges thermiques au aux resultats de [7]. voisinage de la gouttelette, dans le film de vapeur qui l'entoure. I1 reste delicat d'avancer une cause, la loi de mise en B : nombre de Spaldinq qui vitesse des gouttes dans [7] caracterise l'intensite du flux n'etant pas precisee. massique. La discussion precedente Compte tenu de la difficulte s'est portees sur des resultats de de trouver de telles correlations, calculs: la reference [9] va nous on suppose que Cx=f(Re), et malgre permettre une comparaison avec des cette simplification, de mesures experimentales par nombreuses lois existent [31, [4], anemometrie laser. 151, [GI. Les conditions initales sont Nous en donnons quelque les suivantes: exemples: cx = 24/Re loi de Stokes vitesse de l'air (m/s) Cx= i24/Re) (1+Re2/3/6) [5] vitesse initiale cx = (2S/Re0'85)+0.48 [a] des gouttes (m/s) cx = 24/Re pour Re < 0.48 [6] temperature de Cx = 27/Re0.84 Re [0.48:78] [6] l'air ( K ) CX = 0.271 pour Re Z 78 [61 taille des gouttes CX = 0.32 + 24/Re + 4.4/ Re [3] ( w )

Les differentes formulation NOUS presentons les sont representees sous formes de comparaisons entre l'experience et courbes sur la figure 3-2-1. le calcul par M.A.GI.C. sur les figures 3-2-4 pour chaque diametre On constate la bonne de goutte enonce dans la tableau concordance existant entre les ci-dessus. courbes, exception faite de celle correspondant la loi de Stokes, La comparaison s'avere qui se detache nettement des satisfaisante, l'allure des autres et qui sous estime la force courbes obtenues par le calcul de tralnee. suit assez bien les resultats experimentaux. ___-3-2 confrontation avec des resultats de 1amioUraDhie Cependant, on notera que dans ce cas M.A.GI.C. a une tendance a NOUS avons selectionne deux surestimer la mise en vitesse des publications [,7],[9] quf nous gouttelettes par rapport a permettent de situer les resultats l'experience. obtenus par M.A.GI.C. par rapport B des resultats de calculs et a 3-3 svnthese des rbsultats de mesure. La loi de calcul de la 30-4

trainee aerodynamique des qouttes le spray ant termine leur regime utilisee dans M.A.GI.C:. nous de mise en vitesse. permet de soutenir une comparaison avec des resultats isus de * Une deuxieme mesure de mesures [9]. qranulometrie est realisee a une distance suffisante pour qu'il y Mais cette confrontation ait evol.ution notable du diametre semble montrer que l'on peut des qouttes, mais que l'effet de s'attendre a une surestimation de la pesanteur puisse Stre considere la mise en vitesse des comme negligeable. qouttelettes de la part de t4.A.GI.C.. 4-1-1 conditions -aerodvnamigues 4- CONFRONTATION PREDICTIONS/MEm SUR L'EVOLUTION D'UN NUAGE DE GOUTTELETTES. les conditions aerodynamique sont .4-1 .- conditions.. .. . emarimentaw resumees dans le tableau ci- mesure,.d.egr_anulometrie dessous. Pour disposer d'un outil numerique capable de predire vitesse de ! 13.5 ! l'evolution des differentes l'ecoulement ! I classes de diametres de qouttes m/ 5 1 I composant un nuaqe, il est ----- , necessaire de formuler une hygrometrie ! 33 ! modelisation des echanqes % !a I thermiques entre les phases ! 38.2 ! liquides et qazeuses qui soit satisfaisante. taux I I de turbulence ! 1 ! L'evolution du diametre d'une % I 4 qouttelette composant le nuage est I fonction de nombreux facteurs.Nous distance , I pouvons citer par exemple, la le l'injecteur ! I temperature de la qouttelette, !m I , celle du fluide porteur ou encore ! mesure 1 ! 0.25 ! la quantite de vapeur d'eau dans ! mesure 2 ! 1.95 ! l'ecoulement, quantite qui est I I I mesuree par l'hyqrometrie. L'experience qualitative Les mesures ont eu lieu dans acquise au C.E.Pr. dans les bancs la souffl.erie 52 de l'I.A.T. de St de simulation de vols qivrant a CYR qui n'est pas equipee d'un tendance a montrer que sur les systeme de contrBle de distances d'utilisation clu nuaqe l'hygrometrie. qivrant, ( 4 B 5 m), l'hygrometrie a une influence preponderante sur I1 s'ensuit que l'hygrometrie l'evaporation des qouttes. mesuree est celle qui a et6 imposees par les conditions Notre recherche dans la meteoroloqiques pendant toute la litterature pour collecter des duree de la campaqne de mesures. resultats de mesures portant sur l'evolution de diametres de 4-1-3 conditions d'iniection gouttes en fonction de et sranulometrie 6 lrabscisse 1'hyqromMrie s'est avere vaine. __x=O.25 m Une campaqne de mesures a Le spray sur lequel a ete donc et6 realisbe a 1'I.A.T. de St realisee les mesures de CYR, sous contrat C.E.Pr. [lo], qranulometrie a ete obtenu B dont l'un des buts etait de mettre l'aide d'un injecteur pneumatique en evidence l'aspect echanqe du C.E.Pr. [lo] et figure 4-1-3-1. thermique qouttelettes/air par l'intermediare de mesures Les mesures de qranulometrie granulometriques. proprement dites - repartition massique des qoutelettes en Afin de mettre en evidence classes de diametres et Diametre cet aspect transfert thermique Volumique Median - sont issues de sans que d'autre parametres mesures par "MALVERN". viennent perturber les mesures, celle-ci ont ete realisees dans Les differents D.V.M. mesures les conditions suivantes: dans le nuaqe sont realises par variation de la perte de charge du * Le champ de vitesse de la circuit d'air d'alimentation de phase qazeuse a ete choisi l'injecteur pneumatique. uniforme et tres peu turbulent. Le tableau suivant resume les * Une premiere mesure de differentes valeurs de la perte de granulometrie est realisee a une charge - A P - et les D.V.M. distance de l'injecteur correspondant. pneumatique ou les qouttes formant Durarit toute la campaqne de mesure, le debit d'eau de classes de diametre superieur a 50 l'injecteur est constant a une ~m . valeur de 4 l/h, valeur qul correspond a un point de 4-3-2 comuaraison entre fonctionnement courant en essai de D.V.M. givraqe. Les resultats sont portes sur la figure 4-3-2-1 . On note une bonne coiencidence pour des D.V.M. superieur a 20 fim. En revanche M.A.GI.C. semble sous estimer l'evolution des tres faibles D.V.M..

4-4 etude ..numerime_de lLinfluene __~hyqro.in&&x& Du point de vue experimental, Ces quatre valeurs de D.V.M. la mesure de l'hygrometrie reste couvrent la qamme de diametres qui toujours soumise 6 une qrande est demandbe en essai de qivraqe incertitude. [ll. C'est pourquoi il nous a paru 4-2 conditions du calcul Dour interressant d'evaluer, d'une M. A. G1.C. fapon purement numerique l'influence de ce parametre sur Les conditions aerodynamiques l'evolution du D.V.M. d'un spray. prises pour M.A.GI.C sont celles Les conditions du paraqraphe reproduites au cours de la 4-2 ont ete reprises a l'exception campagne de mesure dans la de l'hyqrometrie. soufflerie de 1'1.A.T. [lo]. Les resultats sont presentes On supqose qu'a l'abscisse de sur les courbes de la figure 4-4-1 la premiere mesure de pour les valeurs d'hyqrometrie granumometrie ( x=0.25 m ) , les suivantes : 20%. 30%, 40%, 50%. gouttelettes ont fini leur phase 60%. de mise en vitesse: en consequence, la vitesse initiale Cela souligne l'importance de des gouttelettes dans M.A.GI.C. l'hyqrometrie dans la prediction est celle de l'aerodynamique. de l'evolution d'un nuaqe qivrant. Les conditions aerodynamiques Ce phenomene est sont prises constantes au cours du particulierement sensible dans le temps une valeur de 37%. cas de faibles D.V.M.. La repartition massique en 5- CONCLUSION classe de diametres correspond a celle mesuree dans la souffleries L'ensemble des lois a l'abscisse x=o.25 m, et est d'echanges thermiques et donnee dans les tableaux de la dynamiques de M.A.GI.C. permet figure 4-2-1. d'obtenir sur des melanges eau/air des resultats realistes. Lors de campaqne de certification de materiel La confrontation des mesures aeronautique en vo1 simule et du calcul se fait pour une givrant, le C.E.Pr. est tenu de abscisse de x=1.95 m. respecter des conditions de D.V.M. de plus en plus precises. Nous presentons une synthese portant d'une part sur une I1 est desormais possible, a comparaison classes par classes et l'aide de M.A.GI.C. d'evaluer d'autre part sur le comportement l'ecart de D.V.M. entre global du spray, identifie par son l'injection et la maquette en D.V.M.. essai, donc d'apporter d'&ventuelles corrections sur les 4-3-1 comparaison classes par parametres d'injection pour c1 ass e s obtenir le D.V.M. nominal . NOUS presentons sur les D'autre part, par cette figures 4-3-1-1 a 4-3-1-4 une approche laqranqienne, il est comparaison graphique entre les envisaqeable de determiner le resultats experimentaux et le comportement dynamique des calcul. differentes classes de qouttes au voisinage d'un obstacle. on constate une qrande coherence entre calcul et Cela permet d'evaluer le experience.Cependant, M.A.GI.C. D.v.M. reel impactant sur la sous estime l'evolution des maquette. 30-6

I1 est cependant necessaire 28-29 March 1988 INSA ROUEN dans l'avenir de pouvoir Paper 4-4 confronter M.A.GI.C. a d'autre experiences en diphasiques, [lo] BALLEREAU P. notament des mesures faisant Institut Aerotechnique de etat de melanges eau/glace, et a Saint Cyr des mesures realisees dans des Granulometrie d'un brouillard conditions reelles d'un banc forme par un injecteur et soumis a d'essai. un ecoulement d'air Contrat n- 89-02299 BIBLIOGRAPHIE C.E.Pr./I.A.T. note I.A.T. n' 198/90 du 13 111 reglement selon la P.A.R. avril 1990 25 <& y.n,U. CX [21 BOMMELAER AND 0 N .W p CREISMEAS .- .L" 000000 Icing Test Techniques for Air 000000 Intake Screens on Helicopters Functioning in Temperature Around O~C. I.S.A.B.E. Athene Septembre 1989

131 K1:RMA" Clement Rend

souffleries cryogeniques . These 11-43 1981 O.N.E.R.A. I\: c c [4] FEUILLEBOIS, IASEK r Laboratoire d'Aerothermique In 0 C.N.R.S. a- Modelisation du m comportement de gouttelettes dans z une soufflerie a givrage-Etude pour le C.E.Pr. Rapport final de contrat du Centre d'Essai des Propulseurs n' RC88-11 W [5] CAO MING-HUA 0 Further study on the 0 prediction of liquid fuel spray trace du Cx en fonction du nombre BEISING INSTITUTE de Reynolds AIAA paper 1983 figure 3-2-1 161 A.D. BAYKA i .droplet evaporation rate controlles combustion model AGARD N 353 (1983)

[7] S. WITTING M. AIGNER Kh. SAKHANI and Th. SATTELMAYER On the Parameter Definition for Fuel Atomisers Institut fur thermische stromungmashinen Universtat Karlsruhe (T.H.) Kaiser Str 12 d7500 Karlsruhe W. GERMANY

[8] Gabriel P. PITA Droplet size distribution and liquid volume concentration in a loi de mise en vitesse d'apras [7] water spray : predictions ans -figure 3-2-2 measurements. Instito Superior Technic0 dep. Engenhira Mecanica 1096 Lisbon Portugal.

[9] WECH Measurements of droplet size and velocity in the spray cone of an injection nozzle for combustion chambers : comparaison with numerically calculated data. Proceedings of Institute for Liquid Atomisation and Spray System (I.L.A.S.S.) conference on spray combustion 30-5

W 2 < d 1.00 LI: W ul 0.75 ul +W 2 0.50

0.25

0. 00 ~ 0. 050. 100, 150. 200. 250. 300, 35 ABSCISSE EN mm

loi de mise en vitesse draprits P. A.GI .C. fiwre 3-2-3

figures 3-2-2 et 3-2-3: mise en vitesse pour des gouttes de diametre 5, lo, 20, 50, loo, 500 pm.

0 d

m ow 6'"VI u VI wm 04 d * 0 d

N 0 d

00000 .o vmr\lZdddd

S/W N3 3SS31IA

loi de mise en vitesse comparaison experience [9] / calcul fiuure 3-2-4

schema d,un injeoteur pneumatiwe figure 4-1-3-1 30

% 15

0 0 50 100 150 200 250 300 dismeves Wml

fisure 4-3-1-1 cas 41

15 x

0.-

0 50 100 150 200 250 300 diemeve8 lvml fisure 4-3-1-2 cas 42

0 0 M 100 150 2w 250 3w dlam&lrsr (vml mure 4-3-1-3 cas 43 30-9

15 x

0 100 200 0 50 250 300

fiqureI 4-3-1-4 cas t44

tableau 4-3-2-1

35 30 hg = 30% -D.V.M. 25 20 + hg = 40% 15 0 hg = 50% io -A hg = 60% 0.1 0.2 0.2 0.3 0.3 0.4 0.4 0.5 0.5 0.6 0.6 0.7 5 5 5 5 5 5 pelte de charae & Viniecteur

figure 4-4-1 30-10

2. T. Sutton, BAe Discussion Is there any typical explanation in your table 4.3.2.1 why the error is least at a medium diameter? 1. M. Mulero, INTA Did you compare the results of your model with other Author: models for similar application? I cannot give you a very clear answer. There are two possibilities. First, small diameter droplets do not evaporate Author: hut behave like solids. MAGIC would assume that they do Typical for the MAGIC code is the use of hygromatic evaporate. Second, I do not know what the validity and parameters. In the literature we did not find any code of this precision of the measurements done by Moven with small kind. MVDs was. 31-1

ICING TEST CAPABILITIES FOR AIRCRAFT PROPULSION SYSTEMS AT THE ARNOLD ENGINEERING DEVELOPMENT CENTER

C. Scott Bartlett. J. Richard Moore, and Norman S. Weinberg Sverdrup Technology, Inc., AEDC Group and Ted D. Garretson, U. S. Air Force Arnold Engineering Development Center Arnold Air Force Base, TN 37389 (USA)

SUMMARY clouds of supercooled water droplets. Ice accretion on these surfaces always results in a degra- Icing test capabilities for full-scale turbine dation of performance and operational safety. engine propulsion systems have been developed Icing tests can be conducted in an altitude facility in the Engine Test Facility (ETF) at the Arnold and cover a wide range of Mach numbers, Engineering Development Center (AEDC). The pressure altitudes, and inlet conditions without current capability to simulate natural in-flight icing regard to the prevailing outside weather condi- environments includes knowledge and control of tions. Further, the variables which define an icing the principal factors that govern the mechanics cloud can be accurately controlled and monitored and thermodynamics of icing, namely, water during tests in altitude test facilities. The use of droplet size distribution, liquid water content, the altitude icing test facility has become an cloud temperature, pressure, and propulsion acceptable approach for evaluation and quali- system airflow rates. The AEDC facilities provide fication of aerospace systems for flight into icing the capability to conduct evaluation tests in either conditions. Icing test facilities have been the direct-connect or free-jet mode. developed at the AEDCIETF which provide the capability to simulate icing environments for a The methods and hardware used to inject wide range of aircraft propulsion systems and liquid spray into a cold airstream to simulate in- components. flight icing conditions are discussed. The spray manifold systems and spray injection nozzles INTRODUCTION currently in use at AEDC are described. The methods used to operate and control the cloud Aircraft engine icing test facilities have been generation systems are given. Pulsed-laser and developed at AEDC/ETF to simulate natural pinhole viewing techniques used to view rotating atmospheric icing conditions. The facilities have engine hardware in freeze-frame, and low-light been used to conduct icing tests for military and cameras coupled with fiber-optics used to pene- commercial and domestic and international trate to the inner recess of inlet system to customers. The current capability to provide icing observe real-time transient icing processes on environments for engine testing includes the surfaces susceptible to icing are described. Test kdowledge and control of the principal factors that experiences in both direct and free-jet connect govern the mechanics and thermodynamics of icing testing are addressed. icing, namely, water droplet size distribution, liquid water content, cloud temperature, humidity, Recent ice accretion scaling techniques and pressure, and airflow. The AEDC facilities provide test results, and developments and observations the capability of simulating the free-stream icing in cloud liquid water content and droplet sizing are conditions in either the free-jet or direct-connect briefly discussed. Uses of real-time ice accretion testing modes as illustrated in Figs. 1 and 2, detectors for facility calibration and test article ice respectively. The direct-connect testing mode is accretion rate monitoring are addressed. typically used to conduct engine icing tests. The free-jet testing mode is used to conduct engine, BACKGROUND inletlengine combination, component, sensor, and external surface (wing, empanage, etc.) icing The formation of ice on aircraft propulsion tests. system surfaces occurs during flight through

* The research reported herein was performed by the Arnold Engineering Development Center (AEDC), Air Force Systems Command. Work and analysis for this research were done by personnel of the Air,Force and personnel of Sverdrup Technology, Inc.;AEDC Group, operating contractor of the AEDC propulsion test facilities. Further reproduction is authorized to satisfy needs of the U. S. Government. 31-2

I HEATED IOTAL-TEMPERATURE AND SPRAYBARS (CONFIGURATION 1, FIG. 7) i ' TOTAL.PREIsURE PROBES

r BELLMOUTH c THRUST STAND

i-7ENTERING

ENGINE INLET DU(T// TEST CELL (4.9-m DIAMEIER) FLOW STRAIGHTENER -HOLOGRAM SYSTEM FOR WATER DROPLET SIZE OETERMINATlON

Fig. 1. A direct-connect propulsion engine altitude icing lest cell configuration.

HOLOGRAM SYSTEM FOR WATER AIR ATOMIZING WATER SPRAY DROPLET SIZE DETERMINATION 7 NOULE (CONFIGURATION 2, FIG. 7)- ?TEST ARTICLE ,-TEST (ELL FIIEE-JET NOZZLE I DIAMETER) 1 /ii 1 (3.75-111 i ,/'~~.I-m DIAMETER),,/ ~ ---==== ~ ~ ~~ ~ .- .

TO FACILITY EXHAUSTERS

/ 8.- ~ ~~ r ~-~3..-- i .''I I kl6*~ Ta 37 ,," / . ~ i-~~~ u/ r.. INLET DUCf FREE41 NOULE EXIT (1.8-m DIMETER)-HEATED TOTAL-TEMPERATURE AND TOTAL-PRESSURE PKOBES Fig. 2. A free-jet propulsion engine altitude icing test cell configuration. The natural icing conditions that are typically tions that have been used for the purpose of simulated during testing are well known and testing turbojet and turbofan engines for military documented in the "Airworthiness Standards," applications are given in Table 1, Ref. 2. Ref. 1. A summary of the icing conditions is presented graphically in Fig. 3. These airworthiness The icing conditions that exist at the engine standards pertain specifically to icing on external face during actual flight are seldom identical to aerodynamic surfaces and not to engine perform- those that exist it1 the freestream. The air entering ance under icing conditions. Standard icing condi-

I. PRESSURE ALTITUDE RANGE, SEA LEVEL TO 6.100 m 2. MhXlMUM VERTICAL EXTENT. 2,000 m 3. HORIZONlAL EXTENT, SlANOARO DISTANCE OF 32km 0.9,

PRESSURE ALTITUDE, m x IO3 MEAM EFFEGIVE DROP DIAMETER, Fm a. Ambient temperature versus b. Liquid water content versus mean effective ambient pressufe, stratiform drop diameter, stratiform clouds clouds Fig. 3. Cloud icing conditions (from Ref. 1). 31-3

I. PRESSURE ALTITUOE RANGE, 1,200 to 6,700 2. HORllONTAL EXTENI, STANDARD OF 4.8 kin 3.0

.-, 2.5 CUSS Ill-M CONTINUO E 2.0 i

sb 2: S 1.5 z55 3 1.0 =s 0 - 0.5

0 IS '20 25 30 35 40 45 SO MEAN EFFECTIVE DROP DIAMITLR, pin c. Ambient temperature versus ambient d. Liquid water content versus mean effective pressure, cumuliform clouds drop diameter, cumuliform clouds Fig. 3. Concluded.

Table 1. Icing Conditions Specififed in the Military Specifications for Turbojet and Turbofan Testing (MIL-E-5007D) I. Sea-Level Anti-Icing Conditions Attribute Condition I Condition I1 Liquid Water Content 1 gmim3 2 gmim3 Atmospheric Air Temperature - 20°C ( - 4'F) - 5°C ( + 23°F) Flight Velocity Static Static Altitude Sea Level Sea Level Mean Effective Drop Diameter 15 um 25 pm II. Altitude Anti-Icing Condtions Attribute Condition I Condition I1 Liquid Water Content 0.5 gmim3 0.5 gmim3 Atmospheric Air Temperature -2PC (-4°F) - 20°C ( - 4°F) Flight Velocity (Mach No.) 0.32 0.71 Altitutde 6,100 m (20,000 ft) 6,100 m (20,000 ft) Mean Effective Drop Diameter 15 um 15 urn the inleffengine has generally either been liquid water content required to simulate different accelerated or decelerated from freestream inlet capture ratios can be estimated (Ref. 4). A according to the flight and inlet Mach number typical plot of the required direct-connect engine match or capture, as illustrated in Fig. 4, Ref. 3. face liquid water content ratio to the free stream The level of airflow acceleration or deceleration is liquid water content is shown in Fig. 5 (Ref. 5). a function of the inlet capture ratio as determined from both the flight Mach number and the inlet FACILITY PERFORMANCE CAPABILITIES Mach number. Therefore, the inletlengine icing conditions can be quite different from the free- The facility performance capabilities currently stream conditions. For free-jet icing testing, the available for icing environment simulation are free-stream stagnation conditions can be set in summarized in Fig. 6. The meteorological icing the plenum chamber and the cloud will adjust conditions, Fig. 3, are completely within the direct- itself to enter the propulsion system in a manner connect testing capabilities of the AEDCETF for closely duplicating natural flight. In the direct- engines requiring up to approximately 340 kgisec connect mode of testing, the proper stagnation maximum sea-level airflow. Engines requiring sea- conditions must be set to account for any losses level airflows greater than 340 kgisec can be in the installed configuration, and the liquid water tested at increasingly higher altitudes. Many icing content must be adjusted according to the inlet problems occur at reduced power settings, hence capture condition to be simulated. The change in meaningful icing tests can be conducted on 31-4

M - MACH UUMRtR io6 A - [ROSS IE(1lON AREA LWC - LlOUlO WAIER CONTENT r FREE-SIREAM CONDITIONS [---A, MAXIMUM CONTINUOUS INLET PRESSURE

. AEDUErF - IGO I (ELLS

IHF AIRFLOW INEFITEO BY IHE ENGINE a. Inlet Mach number less than fliyht Mach number, LWCm c LWCCF

fREE-SIREMI (ONDIIIONS

(OMPRESSOR EWES

AIRfLOW, kg/lEC a. Air supply and exhauster capability

AfDC/fTF - T CELL b. Inlet Mach number greater than flight Mach 40 AEDUETF - 1 CELL number, LWCCF c LWCm MINIMUM [ONTINUOUS INLEII TEMPERATURE Fig. 4. Schematic showing possible stream tube configuration for turbofan icing conditions (from Ref. 3). INLET TEMPERATURE 5 WITH 5.PERCfNT LlQUlO AIR INJECTION SUBSCRIPTS: tF - [OMPRESSOR FAtE LL 8 & . . m . FREESTREAM 0 100 300 SO0 700 AIRFLOW, kg/SE(

b. Estimated inlet air temperature capability Fig. 6. Facility performance capability.

Spray nozzle manifolds are used to provide the water droplets for the icing cloud. Typical

Y DROPLET DIAMETER (MASS spray manifolds used at AEDCiETF are shown in MEDIAN) = 20 pm Fig. 7. The water spray nozzles are commercially CL 0 available, two-fluid atomizers. AEDC currently has an inventory of ten different air atomizing spray nozzles, each offering a unique range of droplet FLIGHT MACH NUMB€R M, size and water flow rate. The number and position of the spray nozzles can be varied within the Fig. 5. Compressor liquid water content manifold to provide a uniform icing spray distri- loading factor for two compressor bution at the test section. As illustrated in Figs. 1 face Mach numbers. and 2, the icing spray manifold is typically located in the test cell plenum chamber upstream of the engines with maximum airflow requirements test section. This arrangement allows the spray greater than the facility capability. As indicated in manifold the flexibility to be adapted to different Fig. 6, the low-temperature portion of the propulsion system installations without major meteorological conditions can be achieved with modifications. Both the air and demineralized the addition of liquid air to the inlet air; however, water, delivered through an all stainless steel this technique has not been required at AEDC for plumbing, pump, and storage system, and furnished icing tests. The facility performance capabilities to the spray manifold, are filtered and heated to for free-jet testing are identical to those shown in temperatures near 1OO°C to prevent freezing in Fig. 6. However, the size of the engines that can the spray nozzles during testing (Ref. 6). be tested in a free-jet mode is significantly reduced since substantial airflow must spill The liquid water content of the cloud repre- around the engine to ensure adequate flight flow- sents the amount of liquid water in the atmo- field simulation. The future addition of an icing sphere and is typically expressed in terms of capability to the new ETF C-side test cells, with grams of liquid watevper cubic meter of air. The larger conditioned airflow capacity, will expand the liquid water content of clouds varies greatly with AEDC icing test capability. cloud type and atmospheric conditions. The 31-5

BEllMOUIH DIAMETER 7 r NOZZLE PLUG LOCAIIOH SPRAY BAR

SPUY NOZZLE LOCATION J (ONFIGURATION 1 CONFIGURATION 1

Fig. 7. Water spray manifold systems. droplet size spectrum is generally characterized The amount of spray water injected into the by the droplet mass median diameter. The mass flow field to obtain the specified liquid water median diameter represents the droplet size for content is analytically determined. The liquid water which half of the total water mass is contained content is specified as a test condition and the above and below that diameter. The liquid water amount of water evaporated as the cloud travels content and droplet size are controlled by varying from injection to the test section of the test cell is the water flow rate and atomizing air pressure to determined analytically from a mathematical model the spray manifold. The size and number of the (Ref. 3). By measuring the airflow and injected spray nozzles selected can be tailored to a water flow rate precisely, the desired liquid water particular engine installation. Liquid water contents content at the test section is accurately set and up to above 3.9 grams per cubic meter have been maintained. obtained at AEDC with spray nozzles calibrated for mass median droplet diameter from 15 to 35 The uniformity of the icing cloud is gauged for microns. demonstrating the ability to cover a each unique installation by evaluating the broad range of expected natural icing conditions. uniformity of ice accreted on a calibration grid positioned in the test section prior to test article An extensive computational fluid dynamics installation. This time-honored technique has been (CFD) capability exists at AEDCETF (Ref. 7). A automated at AEDC by application of piezo- full three-dimensional Navier-Stokes flow-field electric transducers operating in the ultrasonic solver is used to evaluate the aerodynamic fidelity frequency range (Ref. 8). An array of the of icing test installations, and its use is especially "ultrasonic" transducers is integrated into a important to ensure the natural flight flow field is calibration grid, and the thickness of ice accreted adequately simulated during free-jet icing tests. A on the transducer is calculated from the signal full three-dimensional, two-phase flow code is received by a dedicated signal processing system available to help predict droplet trajectories during remotely located from the test cell. This icing testing. This code is used in spray nozzle automation of the calibration grid technique for positioning to ensure the most uniform spray cloud uniformly assessment greatly reduces time cloud possible is available during testing. expended for icing test preparation.

Icing Test Techniques at AEDC/ETF The spray nozzles used in the spray manifold system are calibrated to determine the water The technique used at AEDCiETF to produce droplet size production and the control parameters an icing cloud is to inject a continuous stream of necessary to obtain the specified droplet size. The water droplets into a cold airstream directed at the calibration data are obtained in the AEDCETF engine. Injection of the water spray is accom- icing research test cell, R1 D. Researchers collect plished in the low-velocity region of the test cell droplet size and size spectrum data with a laser- upstream of the test cell bellmouth. The injected based imaging technique. Calibration data water droplets are accelerated to near the air currently exist for ten different air atomizing spray speed by aerodynamic forces, and through heat nozzles capable of meeting a wide range of and mass transfer, the droplets reach the air specified conditions. temperature in a supercooled liquid state (Ref. 3). 31-6

The operation of the water spray system is rapidly and uniformly. A sequence of controlled essentially the same for all icing tests. The test events occurs to properly establish the desired apparatus used to make the clouds for either icing cloud target set points in a reasonably short

direct-connect or free-jet testing , has been time. During the duration of the icing cloud, the standardized. A standard icing cloud simulation cloud simulation program maintains the desired computer program exists to provide the necessary liquid water content and mass median droplet control measurements and system stability to diameter by aLitomatic control of the water and meet icing cloud test requirements. The control atomizing airflow rates through flow control valves. program is designed to: establish defined clouds After the specified cloud duration, the cloud is in the test cell free-stream airflow; rapidly control terminated rapidly by closing the water spray bar the cloud liquid water content, altitude (pressure), supply valves, opening the water system drain Mach number, and duration: execute controlled valves, and allowing water to recirculate to the changes for multi-level cloud requirements; and water tank. The icing simulation program, for terminate clouds as rapidly as possible. control flexibility, can extend or abort an icing cloud on command. The cloud simulation program provides automated sequence of required events that Many state-of-the-art video techniques have include pre-cloud system preparation, cloud been successfully used to view ice accumula- initiation, cloud duration control, cloud termination, tions during icing tests of aeropropulsion systems and post-cloud system shutdown/ preparation for at the AEDCIETF. In most cases, low-light-level the follow-on cloud. For the pre-cloud system black-and-white charge-coupled device (CCD) preparation, during the time the prescribed test cameras adapted to a pinhole lens arrangement cell free-stream air test conditions are being have been used to view ice accumulations on established, water is recirculated to a water engine front frame struts, inlet guide vanes, reservoir. The required conditions (water and spinners, and first-stage fan blades. high-pressure atomizing air flow rates, temperatures, and manifold pressures) to admit flow into Viewing of stationary structures such as the spray bars are set. Once the free-stream air engine inlet fronl: frames and structures buried out and spray system conditions are set, water of normal view within the inlet has been recirculation is terminated, the water spray bar accomplished by installing a pinhole lens in the supply valves are opened, and water flows into inlet ducting. The typical lens used will provide a the spray bars. The water system drain valves 50-deg field of view. Two such lens installed 180- (located at the end of the water spray bars deg opposed will provide near full coverage of the opposite the water supply valves) remain open for inlet face. The pinhole lens is installed as shown the first 10 sec of water flow to provide a more in Fig. 8. A hole of approximately 4 cm in diameter rapid purge of air from the spray bars. During this in the inlet duct is required for the installation. purge, there is essentially no flow of water out of This small hole minimizes the effect of the the spray nozzles. When the water drain valves viewing system on the inlet flow field. The actual are closed, water is injected into the airstream assembly is flush with the duct wall. The lens assembly is designed to accept a dry gaseous nitrogen purge to /!j keep it clean and free of moisture. SlUIGHT PINHOLE LENS The engine inlet is lighted by fiber-optic cables coupled to vari- * ~ PURGE GAS able intensity light sources installed external to the test cell. Fiber-optic cables are installed, as shown in Fig. 8, using a fiber- optic light guide threaded into the inlet duct. A hole of approx- TEST ARIKLE imately 2 cm in diameter in the duct is required. Four light PURGE GAS PURGE GAS sources normally pro- Fig. 8. Typical pinhole lens and fiber optics lighting system vide sufficient coverage 31-7 for inlet ducts in the 1-m-diam class. The variable conducted at AEDC to provide the technology intensity light source is useful in adjusting the light advancements necessary to improve the ability to level for maximum performance from the low-light- perform aerospace propulsion system icing tests. level cameras. A dry gaseous nitrogen purge As a part of the icing test technology development keeps the optics clean and moisture free. program, an icing research test cell was constructed in 1975. The cell, Test Cell R-lD, was Successful viewing of ice accumulation on modified in 1986 to provide a larger free-jet nozzle rotating fan blades has been accomplished by area. The cell can be used in either a 30.5-cm or using a laser system as a light source (Fig. 9). 46-cm-diam free-jet mode. The test cell, also Laser light is routed to the engine inlet through a capable of direct-connect icing testing, matches single coherent fiber-optic cable which can be many of the features of the larger AEDC icing test attached to the inlet duct in the same manner as cells and provides a capability to simulate icing described above. Coherent fibers are used to environments similar to the larger test cells. minimize the light loss within the cable itself. The fiber-optic cable ends in a lens assembly designed The ultrasonic ice thickness measurement to spread the beam to the proper dimensions at technique is one of the recent developments the inlet face. The laser light system provides a resulting from the continuing icing test technique frequency-controlled light source which can be technology development program at AEDC. Other synchronized with the rotational speed of the fan current areas under technology development to provide stop-action viewing of ice accumulation include ice scaling methods, real-time nonintrusive and evidence of shedding on individual fan blades. cloud liquid water content uniformity measure- With proper camera alignment, focusing, and ment, and icing cloud simulation requirements. careful mismatch of light source and fan speed synchronization, views through the rotating fan The work in icing scaling is being conducted, blades to the accumulations on the stator in concert with other US. Government agencies, assemblies just downstream of the fan have been to ease the restrictive nature of the scaling laws obtained. currently used at AEDC. Figure 10 illustrates successful scaling results obtained at AEDC (Ref. 9).

The cloud liquid water content uniformity measurement effort is aimed at developing electro-optical diagnostics to replace the piezo- electric transducer-equipped calibration grids. The current approach under consideration is to use a fluorescent dye in the water, excite the droplets with a light source, and correlate the liquid water content uniformity with the uniformity of the AND DURATION CONTROLLED intensity of light scattered by the water droplets.

Fig. 9. Typical strobed laser viewing system The work in icing cloud simulation is directed schematic. towards gaining an understanding of the influence The piezo-electric ultrasonic ice thickness transducers mentioned earlier can be flush mounted to provide an ice accretion thickness measuring system for many test articles. The measuring system provides the capability to measure ice thickness and growth rate in near real time. A hot wax technique is routinely used to produce wax molds to document complete ice accretion shape and surface characteristics. This documentation is beneficial for post-test analysis of the ice accretions.

A continuing icing test tech- Fig. 10. Ice shape scaling results, test article size, full nology development program is and 113-scale (from Ref. 9). 31-8

of icing parameter uncertainties in the validity of The AEDCIETF icing test facilities allow test, icing test results. The efforts to date are reported evaluation, and qualification of full flight systems in Ref. 10. The work indicates which icing condi- as well as subsystems or components. The capa- tion uncertainties are most influential in altering bility to expose subsystems or components to test results. A typical result of the work, shown in icing conditions is particularly important in the Fig. 11, illustrates the impact of droplet sizing system design process, since it allows design uncertainty on ice accretion test results. concepts to be evaluated prior to committing resources to full system development. The AEDC icing research test cell (R-1D) described earlier is particularly well suited for subsystem or component design evaluation. Because of its small size, it can be operated more economically than the larger test cells. Once component or subsystem icing evaluations are complete, AEDCiETF icing facilities are capable of conducting full-scale enginelinlet compatibility HAU 0012. 21.IN. (MOR0 AlRHIIL ANGLE OF ATTACK = 40 icing tests. AMBIENT ItMPtRATURE = -10s MASS MEDIAN OROPLEI DIIAtElER = 20 pm LlOUlD WATER (ONIBI - I .3 gmh" Altitude test facilities such as those at AEDC YELCiIW - I8 dSB offer a safe and economical method for environ- IESI DURATION = 480 SEC mental icing testing. In these facilities, control [bHllRMll = 0.008 over the pertinent flow variables such as temperature, pressure, liquid water content, droplet size, Dm = 20 f 3 pm: dJSUMtD UN(tR1AlNM and Mach number can readily be achieved. The icing clouds produced in the icing test cells closely simulate those found in nature. Many IO 20 30 40 30 successful icing evaluations have been conducted MASS MfOlAN OROfLET OIMIETER, Om, pm at AEDC. and an experience base exists to con- Fig. 11. Effect of uncertainty of mass median duct accurate and closely controlled icing tests. droplet diameter on the iced airfoil drag coefficient uncertainty. REFERENCES SUMMARY 1. "Airworthiness Standards: Transport Category The icing test capabilities available at Airplanes: Appendix C. '' Federal Aviation AEDClETF have been successfully used during Reaulations, Part 25, Federal Aviation the evaluation of many aircraft propulsion systems Administration. and components. Many tests have been conducted in both the free-jet and direct-connect test 2. Military Specifications for Turbojet and modes. The testing has been conducted for inlet Turbofan Testing. MIL-E-5007D, 1973. systems, engines, and external flow surfaces such as full-scale cockpit, empamage, and wing sec- 3. Willbanks, C. E. and Schulz, R. J., "Analytical tions. Icing evaluations have been conducted at Study of Icing Simulation for Turbine Engines AEDC during various stages of aircraft propulsion in Altitude Test Cells." AEDC-TR-73-144 (AD- system developmental programs over a wide 770069), November 1973. range of conditions as shown below: 4. Pfeifer, G. D. and Maier, G. P., "Engineering Free jet Summary of Power Plant Icing Technical Data. " FAA-RD-77-76, July 1977. - Bell helicopter inlet/engine Range Tested at - ALCM AEDC 5. Bartlett, C. S., "Icing Scaling Considerations - GLCM Mach: 0.1 --f 0.8 for Aircraft Engine Testing. " AIM-88-0202, - Various full scale inlets Alt: SL --t 9 km January 1988. - Inlet guide vane sectors iTemp: -29 + 5°C - Icing sensors Size: 0.5 + 2.4 m 6. Marek, C. J. and Bartlett, C. S., "Stability Relationship for Water Droplet Crystallization - Full scale components diam with the NASA Lewis Icing Spray Nozzle.'' Direct Connect DM: 15 + 35 um AIM-88.0289, January 1988. - FllO LWC: 0.2 -+ 3.9 - F101 gmim3 7. Phares, W. J., Cooper, G. K.. Swafford, T. - TF39 W.. and Jones, R. R., "Application of Compu- - F109 I tational Fluid Dynamics to Test Facility and 31-9

Experiment Design." AIM-86-1733, June 3. V. Garratt, RAE 1986. Do you apply any corrections for the modification of a) the LWC, and b) the droplet spectrum by impingement of the droplets on the bell mouth entry? The larger droplets will 8. Hansman, R. J. and Kirby, M. S., "Real Time tend to hit and coalesce. The incidents of these larger Measurement of ice Growth During Simulated droplets would slightly affect the ice accretion. and Natural Icing Conditions Using Ultrasonic Pulse-Echo Techniques." AIM-86-0910 Author: 9. Ruff, G. A,, "Analysis and Verification of the During the test cell preparations for icing testing, efforts are Icing Scaling Equations, Volume I." AEDC-TR- taken to position the water spray nozzles so that impingement of water on the bell mouth is eliminated. There 85-30 (AD-A162226), November 1985. are two reasons for it. One is the unknown effect impingement would have on the liquid water content and 10. Bartlett. C. S. and Foster, R. G., "The Effect droplet diameter spectrum. The other is that droplet of Experimental Uncertainties on Icing Test impingement might form ice accretions on the hell mouth Results. " AIM-90-0665. that are potentially damaging to the the test article. Since the amount of impingement and its effect on liquid water content and droplet diameter spectrum are unknown and Discussion efforts are taken to eliminate droplet impingement, no corrections are applied during testing.

1. W. Grabe, NRC Ottawa To what atomizer air pressure do you work? 4. K. Piel, BMW - RR Aeroengines How do you determine atomizer air temperatures and water Are there any results available of comparisons between temperatures required? testing at AECD and actual aircraft icing conditions, e.g. on slave engines etc? Author: We do not currently have a water spray injection system necessary to conduct icing tests in the largest of the Author: propulsion test cells at AECD, the C1 and the C2 Comparisons between icing tests results obtained in ground (Aeropropulsion System Test Facility, ASTF) test cells. test facililties and results obtained in other ground test Some planning has already been done in consideration of facilities or results obtained in natural or artificial flight have adding an icing test capability to these test cells. The facility been carried out. The comparisons range from close is designed to test the air breathing systems capable of agreement to poor to unacceptable agreement. The generating 90.000+ pounds of thrust. The ability to test agreement issue is a continuing problem within the icing specific engines at specific icing conditions would require an community and is currently best explained as a result of analysis and a feasibility assessment. uncertainty associated with the absolute values of the icing conditions. There are no published comparisons between 2. M. Holmes, RAE AEDC and flight icing tests. Have you any experiemce of conducting icing tests in the ASTF at Tullahoma and if these facilities have the capability to test the new generation of civil fan engines, producing up 5. I? Derouet, SNECMA to 90.000 Ib thrust at take off, which demand a very high air At the station of the spray bars, the water is heated to 100 C. flow at low altitude conditions? Taking into account the distance between spray bars and the test engine, do you think that the droplets have sufficient Author: time to reach the temperature of the air flow? Have you done The water spray nozzles at AEDC are operated at typical calculations? atomizing air pressures of 100 and up to 150 psia. Studies have been conducted to determine the lower level of temperature of the atomizing air necessary to prevent Author: freezing of the droplet as they exit the spray nozzle. These The spray droplets do have sufficient time to reach thermal studies are documented and referenced (ref 6, Marek and equilibrium with the air. Calculations are carried out for Bartlett) in the paper. Typically at AEDC, the water is each icing test installation to ensure that there is sufficient heated to about 180-200 E This is as warm as we can droplet residence time. Details of the calculation procedure comfortably heat the water to prevent boiling in the spray and mathematical model are described in a report by Schulz system and yet keep the spray nozzles warm enough to and Willbanks, ref. 3 of this paper. prevent ice deposits from forming at the water discharge.

ICING TEST PROGRAMMESANDTECHNIQUES E Carr D Woodhouse Project Chief Engineer Senior Project Engineer AeroGasTurblnea and Fuel Injectors Combwlion Teohnalogy Centre CombustionTechnologyCentre Aero and InduslrialTechnology Ud Aero mdIndunrial Technology Ud Burnley. Lancashire, England Burnley, Lmcashire, England

ABSTRAC'I

An Altitude Test Facility with a main chamber 4m diameter x 12m long and capable of providing air flows up to 5kg/s and simulating altitudes up to 15km, is operated at Bumley in the UK by Aero and Industrial Technology Limited (AIT)

This paper details the capability of the facility, describes the type of work carried out and reviews the experience obtained on icing programmes since the plant was commissioned in 1953.

Examples of the procedures used to establish the susceptibility of equipment to icing are given and cover the use of scale models, the evaluation of probes and the testing of complete helicopter engine intakes. FIG 1 AIR FLOW PATH THROUGH FACILITY

INTRODUCTION access door removed and an engine being installed In the early 1950's Joseph Lucas (Gas Turbine Equipment) Limited built a high altitude test plant at Burnley in the UK on behalf the then British Ministry of of Air is drawn into the plant by a seven stage compressor Supply. Over the years this plant has had a high level of fined with interstage and after cooling. It then passes utilisation and continuing programmes of refurbishment through refrigeration and absorption driers to reduce the and updating and is now wholly owned by AIT Limited and dew point. At this stage the air is essentially dry at 480 operated on a normal commercial basis. At its inception kPa and +IO C and it then flows through two loaded the plant was primarily intended for the examination of expansion turbines to reduce the temperature to -80 C. aero gas turbine combustion stability and relight problems The air then passes through a trim heater to one of the but in practice it has been used for a much wider range of two test areas Le. the engine test cell or the rig test cell. programmes.

The main advantages of the plant are its low running costs, its flexibility and the rapidity with which test points On leaving the test area in use the air passes through a can be achieved. As a result the plant is very cost effective large gas cooler to the anitude control valves and it is the for the evaluation of the light-up performance of large aero setting of the latter which chiefly controls the test pressure. engine combustion systems, the full range performance The air is then drawn from the plant by two exhausters. As assessment of smaller gas turbine engines and also the will be seen from Figure 1 various bypass lines, bleed testing of ancillary equipment required to operate at lines and antisurge valves are fined so that it is possible to altitude or cold day conditions.The facility is a CAA use or bypass one or both of the expansion turbines and approved test house. use one or both of the exhausters as required. The second exhaust blower has a substantially larger capacity PLANT DESIGN AND OPERATION than the first, hence it the test pressure is in the range 9OkPa to 38kPa it is possible to induce additional air The configuration and operation of the plant is described directly from atmosphere i.e. bypassing the input below, following the path taken by the air flow through the compressor and the expansion turbines. main plant equipment. Figure 1 shows the main plant tems schematically and Figure 2 an isometric cut away PLANT CAPABILITY drawing, also showing the main plant items. Figure 3 shows a view of the end of the main test chamber with the Air Flow, Pressure and Temperature 32-2

4 Dehydrators. 5 Extra air dryer. 6 Water cooling towers. 7 Air inlet filter. 8 Turbo compressor. 9 Second extraction blower. 10 First extraction blower. 1 1 Extraction blower intercooler. 12 Refrigeration dryer. 13 Main gas cooler. 14 Diffuser. 15 Engine test chamber.

FIG2 'CUT AWAY'VIEW SHOWllNG MAIN PLANT COMPONENTS

FIG 3 INSTALLATION OF ENGINE INTO TEST CELL 32-3

An air flow of up to 5,7 kgls can be supplied at pressures ENGINE TEST CELL down to 38 kPa (i.e. corresponding to an altitude of 7620m). Over the pressure range 38 kPa to 13 Engine Calibration e,g, kPa(7620m to 15240m) the maximum air flow is 3,4 kds. Altitude tight-up characteristics. The basic plant will supply the air at temperatures in the Altitude performance measurement (bhp or thrust). range -65 C to +75 C over the air flow/pressure range Engine starting assessment following prolonged quoted above. Higher air supply temperatures are temperature soak. provided if required by installing additional heaters. Intake Icing Tests. Probe Icing. Rapid changes of air pressure and temperature can be Foreign body ingestion. achieved. Water ingestion. Intake filter evaluation. Typically Decompression Testing. 100 kPa to 46 kPa in 165 seconds Evaluation of anti-icing equipment. 46 kPa to 100 kPa in 60 seconds Auxiliary power unit (APU) performance and starting. C to -54 C in 13,3 minutes t40 RIG TEST CELL -33 C to +38 C in 8,5 minutes Combustion Chamber Evaluation e.g. Rapid decompression tests can be simulated by specially constructed test rigs. Determination of ignition and stability limits. Evaluation of light round characteristics. ENGINE TEST CELL DIMENSIONS Combustion efficiency testing. Reheat system evaluation (Scale models). The engine test cell is 12,2 m long by 4m diameter giving a normal working space 7,9m long by 1,8m diameter. Entry is through a 2,6m diameter door with a 3000 kg Portable Altitude Chamber e.g capacity crane runway. Engine and APU starting assessment following prolonged RIG TEST CELL DIMENSIONS temperature soak. Combustion chamber ignition and stability testing with An open test cell is available which will accommodate test ability to view flame propagation. rigs requiring low pressure air supplies (e.g. combustion system test rigs). This cell is 8,23m long x 3,66m wide x REQUIREMENTS FOR ICING TESTING 6,lm high. An additional small sub-atmospherlc test chamber is available which can be installed In the rig test To carry out effective icing tests it is necessary to cell and this chamber is 2,4m long by 1,5m diameter. reproduce the correct conditions, i.e., air velocity, air pressure, air temperature, water content and water droplet SERVICES size and to test for predetermined times. The requirements are set out in the Joint Airworthiness A comprehensive range of services is available and Requirements (JAR) and the Federal Aviation Regulations includes (FAR) which specify the conditions applying under various flight regimes and the minimum satisfactory performance Provision for the supply of a range of fuel types at requirements. Copies of graphs taken from FAR 25 temperatures controlled between -50 C and t100 C. Appendix C showing the requirements for intermittent and continuous maximum icing conditions are reproduced in Electrical supplies at all of the normal industrial and Figures 4,5,6 and 7. aircraft voltages and frequencies.

Air humidity control. hTERM TrENT MAX MLM C.MvL FORM CLOUDS ATMOSPHEH C C hG COkD T OhS D NATER COhTEKT , MEAh EFFECT,. E DROP-ET DlA Equipment for ice accretion and water ingestion testing. 0-

Computerised data logging.

Colour video observation and recording.

Gas analysis to EPA standards.

Fuel injector cleaning and or calibration.

Workshop facilities for equipment manufacture and repair.

An in-house probe build and calibration department referenced to NPL standards. MEAN EFFECTIVE DROPLET 0IA:MICROMETRES TYPICAL WORK PROGRAMMES

Typical work programmes carried out in the two test areas FIG 4 are:. 32-4

CONTINUOUS MAXIMUM (STRATIFORM CLOUDS) In addition there can be a requirement for testing to ATMOSPHERIC ICING CONDITIONS establish the effect of ice formations becoming dislodged. LIQUID WATER CONTENT Y MEAN EFFECTIVE DROPLET DIA. This can be determined by either allowing the ice to form "" and shed by the process applying in service or by forming "F.- ",a NGE: S.L.-6.71 km blocks of ice in refrigerators and injecting pieces of a 0.8 k known size into the item to be tested. Hailstones can also 5 0.7 be produced and injected into the test item by means of a ? 0.6 specially designed gun. 5 0.5 5 0.4 0.3 rEMPERATURE 3 0.2 5 0.1 ICING TEST EQUIPMENT $0 The parameters specific to or of particular importance MEAN EFFECTIVE DROPLET DIA. - MICROMETRES when conducting icing tests are air humidity, water droplet size and airhvater approach conditions. FIG5 It is essential that the humidity of the air supplied is closely controlled, if it is too low any ice formation can be ablated INTERMITENT MAXIMUM [CUMULIFORM CLOUDS) and if it is too high additional ice will form. It is normal ATMOSPHERIC ICING CONDITIONS practice to control the humidity in the range 90% to 100%. AMBIENT TEMPERATURE v PRESSURE ALTITUDE In the AIT plant the required humidity is obtained by injecting steam into the upstream air supply ducting. +5 The required water droplet size is obtained by injecting 0 water through air blast atomisers into the air fed to the test u -5 equipment. Seven atomiser nozzles are available (see W Figure 8) and all of the nozzles or a selected few can be a 3 -10 used as required. It is possible to control the water flow to each nozzle separately and to adjust the atomising air wz -15 a pressure to control both the water distribution and the quality of atomisation. Until recently the droplet size -20 2t- achieved was established using an oiled slide technique t- for the measurement and counting of the droplets. 5 -25 z Recently Laser Equipment (Malvern Type 2SOOC) has 5-30 been installed for water droplet measurement. This 6 equipment determines the droplet size by establishing the -35 extent of the light scatlered from a laser ligM beam.

-40 0123456789 PRESSURE ALTITUDE km

FIG 6

CONTINUOUS MAXIMUM (STRATIFORM CLOUDS) ATMOSPHERIC ICING CONDITIOIUS AMBIENT TEMPERATURE v PRESSURE ALT.

II 01234567 PRESSURE ALTITUDE km FIG8 WATER SPPAY INJECTORS

FIG 7 32-5

The correci air flow conditions at approach to the equipment on test are obtained by using specially shaped and sized ducts and controlling the flow and condition of the air supplied from the main plant. The final check before commencing a new test programme is usually to fit a grid into the aidwater supply duct immediately upstream of the equipment on test and to check that the required ice distribution is obtained i.e. either uniform or biased as required.

ICING TECHNIQUES

It follows from the foregoing that the techniques for icing testing are as follows:.

1. Build a suitable test configuration. 2. Set up the required operating conditions. 3. Test for the time required to prove compliance with the specification whilst observing any ice build and taking TYPICAL CONFIGURATION INTAKE ICING measurements to determine any change of air flow or FIG 9 FOR TESTS pressure loss. 4. Shut down quickb following the test and record the extent of any ice formed.

ICING TEST PROGRAMMES

Examples of the icing test programmes carried out are given in the following subsections.

HELICOPTER INTAKE TESTING

Tests have been carried out to determine if ice formed and became dislodged in a helicopter engine intake. The programme was carried out in two parts (a) using the intake alone and (b) with the intake feeding an operating engine. FIG 10 NPICAL CONFIGURATION FOR ENGINE AND INTAKE ICING TESTS

In the first series of tests it was possible to explore a wide range of conditions without risk of damaging the engine and also with the facility to collect and examine any ice obtained. In the second series of tests it was possible to confirm that a Satisfactory configuration had been obtained and to prove that it fuwilled the type test reauirements.

Figure 9 shows the configuration of the set-up used for intake testing and Figure 10 shows the arrangement used for engine testing with the correct aircraft intake fitted. The main features of the test rigs are(a) the air supply ducting provided specifically for the test programme to achieve the correct velocity conditions at approach to the intake, (b) the facility for catching any ice becoming detached during the tests on the intake only, (c) design features permitting quick access to the parts subject to icing and observation of ice build-up during testing.

Figure 11 shows icing on a wire mesh grid fitted across an aircraft intake to check the water distribution. In this case the highest water concentration was required towards the outboard side of the passage and it was confirmed by the test. Figure 12 shows icing on the outer lip of the same aircraft intake after the subsequent performance test with FIG 11 DISTRIBUTION TEST the intake grid removed. 32-6

FIG 13 INSTALLATION OF AN AIRCRAFT INTAKE

FIG 12 INTAKE ICINGTEST

Figure 13 shows in more detail the ducting used on an aircraft intake to achieve the required air flow conditions and permit observation of ice build-up.

PROBES

The probes subdivide into two catagories, fixed items and items required to move under operating conditions. FIG 14 DROPLET SIZING PROBE ICING TESl

With the former all that is required is to ensure that the correct flow condlions are achieved, check the reading obtained from the probe and record the extent of ice accretion. With the latter it is also necessary to prove that the actuation mechanism still operates under maximum ice conditions.

Figure 14 shows ice accretions on a droplet sizing probe evaluated in the AIT tesi chamber and Figure 15 shows a flight refuelling probe which was required to extend when "iced-up".

DEVELOPMENT OF HEATED MATS

Another aspect of the work programme has been the development of heated anti-icing mats. In this case it is possible to determine the minimum heat input required for various sections of the heating mat by separately controlling the heat input to sections of the mat. The FIG 15 REFUELLING PROBE INSTALIATION results of the development programme are then normally confirmed by tests on the production version. 32-5

ENGINE INTAKE DUST SEPARATORS SCALE MODELS

When flown under dusty conditions helicopter engines are As stated earlier one of the main advantages of the facility fitted with separators at the engine inlet and 1 is is its relatively low operating cost. The lower cost results necessary to prove that they do not block with ice. Again partly from the smaller size and air flows of the facility and the approach adopted is to manufacture ducting to give of course this can be in some cases a restricting factor. the required approach conditions, bearing in mind any pitch or yaw requirements, to test as demanded by the One way to overcome this problem is to use scale models JAR etc and measure any change of performance and and this approach has proved very successful in research make a detailed record of the ice formed. and development programmes. The use of half linear scale models with appropriate changes to operating One configuration used for helicopter engine separator conditions where necessary, can reduce the required air assessment is shown in Figure 16 and a photograph of mass flow by a factor of four thus bringing many the ice build-up is shown in Figure 17. programmes within the scope of the facility.

Figure 18 shows a 50% scale model of an aircraft intake. As the equipment was used for a researchidevelopment exercise it was also made with adjustable or replaceable comDonents.

FIG 16 PARTICLE SEPARATOR, TYPICAL CONFIGURATION

FIG 18 INSTALLATION OF A HALF LINEAR SCALE MODEL INTAKE

CONCLUDING REMARKS

The facility is eminently suitable for either development or certification testing of equipment which requires air flows of 5,7 kgls or less.

It is possible to simulate accurately the conditions specified in the FAR and JAR and to quickly obtain repeat tests to either evaluate development modifications or check performance repeatability.

FIG 17 PARTICLE SEPARATOR ICING TEST 32-8

Discussion systems. Tests carried out by AIT confirm that the laser technique indicates smaller droplet diameters than the oiled 1. S. Riley, Rolls Royce slide method. Calibrations in the altitude test facility ai The maximum airspeed of your tunnel as a relatively large Burnley showed reasonable correlation with the data given facility is low. There are many small high speed facilities by R. G. Keller of the General Electric Company, published available. Are there plans to increase its capacity? in AGARD CP 236, i.e. a spray measured at a volume median diameter of 25 pm by the oil slide technique was Author: shown to haveamediandiameterofapproximately 15 pm by The procedure normally adopted in the AI” altitude test the laser technique. facility is toinstall airsupply ductssized to give the approach It follows that larger water droplets are being used than was velocity required for the system being evaluated. The main the case prior to adoption of the laser system for spray plant limitation is therefore the maximum aimow capability. characterisation. The accuracy of the oiled slide technique is Whilst there have been several studies into the feasibility of open to question because of the intrusive nature of the test increasing the plant capabililty, to date they have all being and the sensitivity of the result to the characteristics of the rejected as they would all have negated the main advantage oil used on the slides. Therefore, since first the laser of the present facility, namely, the relatively low operating technique is considered to be more accurate and second the cost. There is, of course, an ongoing programme for test obtained with the largerdroplets would he more rather updating the instrumentation and control systems available than less stringent and third as our enquiries showed that the in the facility. majority of the test facilities in the Western world had already adopted thelasersystem,it wasdecided to change to 2. E. Brook, Rolls Royce this technique. It is apparent that the test conditions You have recently changed from measuring water droplet specified by the various authorities for icing Certification distributions by oil slides to a laser technique. Have the calibration, should he redefined using modern laser appearant calibration of your spray nozzle charged? systems. Could you comment on the charge seen, as the certification icing envelopes used by the authorities are based on measurements carried out before modern laser techniques 3. H. Hoffmann, DLR were available? Your instrument for thc measurcment of particle size scattering, is it an own development or the FSSP of PMS? Author: We had substantial reservations concerning the change to Author: the laser technique and only decided to recommend the No, this is the Malvern equipment. It is commercially change to customers after a protracted evaluation of the two available. 33-1

A DOCUMENTATION OF VERTICAL AND HORIZONTAL AIRCRAFT SOUNDINGS OF ICING RELEVANT CLOUDPHYSICAL PARAMETERS

by H.-E.Haffmann German Aerospace Research Establishment (DLR) Institute for Atmospheric Physics D-8031 Oberpfaffeahofen Post Wessling/Obb. Germany

Summary icing flights to get intormations about the dependence of the icing relevant cloudphysical parameters on cloud In a homogeneous st-cloud (in a high pressure area) porameters. Here, cloud parameters imply the type of the tatol water content TWC is about linearly increasing clouds and the height above cloud bose. We have with increasing distance from the cloud base and ob- always got these informations when we could fly tains its largest value beneath the top (0.39 respec- through the whole cloud from its top to its base re- tively 0.49 g/m3 ). The median volume diameter MVD spectively vice versa or when we could fly during one is nearly remaining constant and has predominantly icing flight definite flight pathes in different heights. small values (between 15 and 23 pm). The phase of The data which we hove collected during such vertical the particles in a11 st-clouds, evaluated up to now. ond horizontal sounding of clouds. together with infor- was fluid. Such a regularity was not found in any of mations on the meteorological synoptic situation, are the other types of inhomogeneous clouds of a warm serving os input data for a climatology of icing rele- front. Apart from temperature T, which is decreasing vant cloud physical parameters producing aircraft relo- nearly linearly in these clouds, too, the course of TWC ted icing degree severe (7,8.9.10,1 1,12). and MVO is very irregularly. Both the parameters can After a short description of the measuring procedure in have several maxima in different distance from the section 2. in section 3 the vertical structure of the bose. The maxima values of TWC can be up to cloud physical parameters total water content TWC, 0.45 g/m3 and those of the MVD up to 460 pm. The temperature T and median volume diameter and in sec- phose of the particles could vary between fluid and tion 4 the horizontol structure of these cloud physicol solid. Not only the vertical structures but also the ho- parameters ore shown. There is differentiated between rizontal structures show great differences of the par- Sc, St-clouds in a high pressure area and Sc, As, ticle distributions: In the clouds of a high pressure Ac, Ns-clouds in the range of o wormfront. For each area more than 90% of the particles had diameters be- case two example are shown. In section 5. the vertical tween 2 and 32 pm and in the clouds of a warm front and the horizontal structures of one of the examples more than 49% respectively 99% diometers between 33 for clouds in a high pressure area and in the range of and 600 pm. o warmfront ore compared. In the sections 3.4.5, for selected points the particle size distribution is shown, too. The results given in this paper are taken from 10 Notations and 11.

D Ipml Diameter of cloud particles Fp [kml Flight path H [ml Height above cloud base MVO [pml Median volume diameter of cloud particles T [OCl Static air temperature TWc[dm 1 Totalwater content (water content from 2. Measurinq instruments employed and performonce of fluid and solid particles) meosurements

For the vertical and horizontal structures, the volues for the TWC were measured by a Johnson-Williams hot wire instrument. the values for the temperature by 1. Introduction a Rosemount plotinum-wire instrument, and the volues for the MVD by Knollenberg PMS instruments FSSP Since 1983 the "Icing of Aircraft" is investigated in the and OAP. The TWC values used for the particle size di- DLR - Institute for Atmospheric Physics. For this pur- stributions are derived from the FSSP and OAP meo- pose, an aircraft has been equipped as an icing re- surements, too. search aircraft (1.2). By the results of icing flights The measurements were carried out in the course of which were conducted with this aircraft. it was tried, icing flights which were conducted in a region between before all, to answer the question in which manner the the north edge of the Alps and a distance of obout normalized and the aircraft related icing degree is de- 200 km north of it. Each value of the points in the pending on cloudphysical parameters (3.4,5,6). For im- figures is the mean value formed on a flight time of proving the forecast of icing we tried also during these 20 sec. That means a flight distance of about 1.2 km. 33-2

3. Results of vertical soundinqs

3.1 Two examples for Sc. St-clouds in a high pressure areo

o-4j-.-.--I o-4j-.-.--I 25I$ 25 um td 18 10'

1. Example

8w m wo H HXI

0 D Fig. 2. Total water content TWC in dependence on diameter r- 7 -,- r- -r 7 - r- -r- -r -,- r-T 7 -20 -15 -10 -5 0 t5 "c t10 T of particle D for the paints (1) to (4) of the vertikol struc- ,...... ,...... ,...... ture of fig. 1. I 2 5 10 20 50 IW 2W WpmlOOOllYD

Fig. 1. Total water content TWC, temperature T. and median volume diameter MVD in dependence on height above cloudbase H. Phose of cloud particles: Fluid Mea- Height Number o Height of the cloud base: Between 400 and 500 m - wring above TWC T MM D cloud mint No. cloudbase particles

(m) (g/m3) (OC) (pun) (w) pro cm3

1 243 0.16 -2.4 15.0 10.3 313.0 2 426 0.35 -3.4 21.0 13.8 266.0 3 704 0.49 -4.5 23.0 14.9 186.0 4 775 0.01 -1.7 280.5 14.1 0.2

Airmass: xPs

Table 1. Some data for the selected measuring points of fig. 1. 33-3

2. Example

'"1 m

r -7 -I- r 7 7 - r- r -r -,- rT 7 -20 -15 -10 -5 0 t5 "c t10 7 ,...... * ...... ,...... 1 2 5 10 20 M 100 200 5WpmlWO UVD

Fig. 3. Total water content TWC, temperature T. and median volume diameter MVO in dependence on height above cloudbase H. Phase of cloud particles: Fluid Height of the cloud base: -400 m

Fig. 5. Total water content TWC in dependence on diameter of particle D for the points (1) to (4) of the vertical structure of fig. 3.

Mea- Height - Number 01 suring above TWC T MY0 0 cloud point No. cloudbase particles

(m) (g/4 (PI pro cm3 r-7-i-T-r--rrT--rTy (OC) (w) -20 -15 -10 -5 0 t5 "c t10 I 1 ...... ,...... ,...... 1 71 0.04 -5.3 63.7 4.8 30.0 1 2 5 10 20 M 1W 2W MOpmlOOO MyD 2 406 0.23 -5.7 19.0 7.2 293.0 3 591 0.39 -4.9 23.0 9.8 179.0 0.01 +1.0 63.7 67.4 0.03 Fig, 4. Total water content TWC. temperature T. and 4 833 median volume diameter MVO in dependence on height above cloudbase H. Airmass: xSp Phase of cloud particles: Fluid ( On the same flight as fig. 3, but about 7 min later Some doto for the selected measuring and about 5 km dislocated ) Table 2. Height of the cloud base: -600 m points of fig. 3. 33-4

3.2 Two examples for Sc. Ac. As, Ns-clouds in the range of a warm front

1. Example-

2w OL,t"" I I I I Fig. 6. Total water content TWC in dependence on diameter 0 0.2 0.4 0.6 0.; &3 1lO Tm: of particle D for the points (1) to (4) of the vertical struc- r -Y -8-T- -r 7 - r r 7 -I- r T 7 -20 -15 -10 -5 0 t5 "c I ture of fig. 4. ,...... ,...... ,...... t10 1 2 5 10 20 50 1W 2W 5Wpm10W UVD

Fig. 7. Total water content TWC, temperature T. and median volume diameter MVD in dependence on height obove cloudbase H. Mea- Height Phase of cloud particles: Fluid -. Number of suring obove TWC T MVD [I cloud Height of the cloud base: * 1400 m point No. cloudbase particles

(m) (9/m3) (OC) (pm) (pi) pro cm3

1 126 0.07 -5.8 11.0 6.1 268.0 2 269 0.23 -6.1 19.0 7.3 289.0 3 443 0.42 -4.5 43.9 9.4 34.0 4 522 0.04 -2.2 63.7 57.2 0.2

Airmass: xSp

Table 3. Some data for the selected measuring points of fig. 4. 33-5

2. Example t solid i

fluid

Fig. 8. Total water content TWC in dependence on diameter of particle D for the paints (1) to (3) of the vertical structure of fig. 7.

L 1 0 0 0.2 0.4 0.6 ' 0.i gim' 1l0 TWC r -7 -,- r T 7 - r l-7 -I- r T -7 -20 -15 -10 -5 0 t5 "c t10 I ,_____...... ,...... ,...... r.. .. ,. . . . . , ...... ,. . ..., 1 2 5 IO 20 50 1W ZW WOpmlOw UW

Fig. 9. Total water content TWC. temperature T. and median volume diameter MVD in dependence on height Mea- Height Number of above cloudbase H. suring above TWC T UM) 5 cloud Phase of cloud particles: Fluid / solid point NO. cloudbase particles Height of cloud base: *lo00 m

(m) (g/m3) (OC) (pm) (pm) pro cm3

1 225 0.03 +1.7 142.6 105.0 0.4 2 976 0.33 -1.7 123.0 19.5 87.0 3 1316 0.29 -3.6 63.7 30.0 21.0

Airmoss: mP

Table 4. Some data for the selected measuring points of fig. 7. 33-6

10-1

TWC 10-2 dm310-3

25 2 5 -. 18 1 0' IC? pm I$

25 25 18 1 0' IC? um 1d

Fig. 10. Total water content TWC in dependence on diameter of particle D for the points (1) to (8) of the vertical structure of fig. 9.

Mea- Height - Number of luring above lWC T MM) D cloud oint No. cloudbose porticles ,i

Table 5. Some data for the selected measuring points of fig. 9. 33-7

4. Results of horizontal soundings

4.1 One example for Sc. St-cloud in a high pressure area

Fig. 11. Total water content TWC. temperature T, and median volume diameter MVD in dependence on flight path Fp in constant flight altitude. Phase of cloud particles: Fluid Height above cloudbase: *450 to 650m Distance from cloudtop: *150m Flight altitude: 1050m

Fig. 12. Total water content TWC in dependence on diameter of particle D for the points (1) to (4) of the horizontal structure of fig. 11.

iWC g/m3

Number of Mea- Height - suring above TWC T MM D cloud paint No. cloudbase particles 0.6 (m) (s/m3) (pm) pro cm3 (w) 0.4 1 451 0.16 -1.4 19.0 10.3 89.3 2 449 0.27 -1.6 19.0 12.9 86.5 3 449 0.25 -1.5 19.0 9.0 178.3 0.31 -1.7 21.0 9.4 168.5 4 447 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 Fo km Airmass: xSp

Fig, 13. Total water content TWC in dependence on Some data for the selected measuring Table 6. flight path Fp for two different heights above cloudbase points of fig. 11. Phase of cloud particles: Fluid

Height above cloudbase: 375 to 575m .__--__--Height above cloudbase: 450 to 650m 33-8

T 0 C 105l -I:/ -15

-201I , , , . . . li-20: OL, , . , . , . , . , . , . , , . I. r 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 0 5 10 15 20 25 XI 35 40 45 50 55 fp km fp km

Fig. 14. Temperature T in dependence on flight path Fig. 16. Total water content TWC, temperature T. and Fp for two different heights above cloudbase. median volume diameter MVD in dependence an flight Phase of cloud particles: Fluid path Fp in constant flight altitude but deeper penetration into the cloud, caused by continuous ascent of Height above cloudbase: 375 to 575m the cloud surface in flight direction. .____- ___ Height above cloudbase: 450 to 650m Phase of cloud Iparticles: Fluid Flight altitude: 11 15m Height above cloudbase: ? Distance from cloudtap: ?

4.2 One example for Sc, Ac, As. Ns-cloud in the range of a warm front

MVD Pm mi i, 2wi 'i roo; oi Mi I

._+-_---r-+- 20j -? 10:-10- I 10 ,I 5: ' :-151 21 I - 12-20, 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 fp km fp km

Fig. 15. Median volume diameter MVD in dependence Fig. 17. Total woter content TWC. temperature T, and on flight path Fp for two different heights above median volume diameter MVO in dependence an flight cloudbase. path Fp. Phose of cloud particles: Fluid Phase of cloud particles: Fluid Height above cloudbase: *435m Height above cloudbase: 375 to 575m Distance from cloudtop: ? ____ _-_ __ Height above cloudbase: 450 to 650m Flight altitude: 1835m 33-9

., TWC 9/m3

ll 10-44 , 0.8 25 25 I 18 10‘ 12 um ID’ 0.4o,61

0 10 20 30 40 50 60 70 80 90 Fp km

D Fig. 19. Total woter content TWC in dependence on flight poth Fp for four different heights above cloudbase. Phase of particles: Fluid

Height above cloudbase: * 435m - -__-----Height above cloudbase: * 725m ...... Height above cloudbase: * 1035m 25 25 -.- - -. - Height above cloudbase: * 1325m 18 10’ 12 um 10’ wc3::f-J=qD g/m 10-3

1 o-~ 25 25 1 0 18 1 0‘ 12 pm ID’ C 0 10- Fig. 18. Toto1 voter content TWC in dependence on diameter of particle D for the points (1) to (4) of the horizontol 5- structure of fig. 17. 7---- 0-...... __---___-- - -5.

-10-

-15-

-207 . I . I ’ I ’ I 0 10 20 30 40 50 60 70 80 90 Mea- Height - Number c FP km uring above TWC T MYD 0 cloud lint NO. cloudbase particles Fig. 20. Temperature T in dependence on flight path (m) (9/m3) (OC) (w) (pm) pro cm‘ Fp for four different heights above cloudbose. Phase of particles: Fluid 1 440 0.20 +0.3 103.3 73.2 1.8 2 438 0.33 -0.6 23.0 26.3 20.9 Height above cloudbase: 435m 3 435 0.08 -0.3 202.7 26.9 8.3 - _-__ ---- Height above cloudbase: * 725m 4 434 0.08 -0.7 17.0 16.7 14.0 ...... Height above cloudbose: 4 1035m - - - - -. - Height above cloudbose: 4 1325m 4irmass: mP

Table 7. Some data for the selected measuring points of fig. 17. 33-10

5. Comporison between representative results of cloud:! in o hiqh pressure ore0 and clouds in the ronqe ot a worm front

5.1 Comporison between results of vertical soundings

Fig. 21. Median volume diameter MVD in dependence on flight path Fp for four different heights above cloudbase. Phase of particles: Fluid

Height obove cloudbase: * 435m ______Height above cloudbase: * 725m ...... _. .. Height above cloudbase: * 103% -.-. - - Height above cloudbase: * 1325m

Fig. 22. Total water content TWC in dependence on height above cloudbase H in o high pressure are0 and in the range of a warmfront. Phase of cloud particles: Fluid - respectively fluid / solid - In o high pressure area ______In the range of o warmfront 33-11

2wo m m 18w

ISM) H H 16w 16w 1402 lux) 12w 1ZW 1wO 1oW 8W

600

ux)

ux) 2w 2w 0 0 -20 -15 -10 -5 0 t5 OC *lo I Fig. 24. Median volume diameter MVD in dependence on height above cloudbase H in a high pressure ore0 Fig. 23. Temperature T in dependence on height and in the range of o warmfront. above cloudbase H in a high pressure orea and in the Phase of cloud particles: Fluid - respectively range of a wormfront. fluid / solid Phase of cloud particles: Fluid - respectively - In o high pressure orea fluid / solid In the range of 0 warmfront - In o high pressure orea _-_--_In the range of 0 warmfront

25 25 18 10' I$ pm 13 D In the range of a warmfront In the range of a warmfront

In a high pressure area In a high pressure area Fig. 26. Total water content TWC far the particle range 2 to 32 pm, 33 to 310 pm and 311 to 600 pm, and its part Fig. 25. Total water content NIC in dependence on particle on the total water content of a11 the particles in X for one diameter D for one point of the vertikal structure in a high point of the vertical structure in a high pressure area and pressure area and in the range of o warmfront in the range of a warmfront. 33-12

5.2 Comparison between results of horizontal soundings

MW vm 1000 500 1 lWC 9/m3 I-

0.8

0.6 2Lr1 . , . , . , . , . ,' . , . I 0.4 0 10 20 30 40 50 60 70 80 90 Fp km 0.2

\,., \,., ,J'\ - 0 Fig. 29. Median volume diameter MVD in dependence 0 10 20 30 40 50 60 70 80 90 on flight path Fp in constant flight altitude in a high Fp km pressure area and in the range of a warmfront. Phase of particles: Fluid

Fig. 27. Total water content TWC in dependence on In a high pressure area flight path Fp in constant flight altitude in a high .______In the range of a warmfront pressure area and in the range of a warmfront. Phase of particles: Fluid

In a high pressure area .______In the range of a warmfront

T 0 C

D In the range of a warmfront 0 - .- , , , , , , , , I, I,

'" I I -10:I-, -15

-20 __1 0 10 20 30 40 50 KO 70 80 90 FD km

Fig. 28. Temperature T in dependence on flight path Fp in constant flight altitude in a high pressure area In a high pressure area and in the range of a warmfront. Phase of particles: Fluid Fig. 30. Total water content TWC in dependence on particle diameter D far one point of the horizontal structure in a In a high pressure area high pressure area and in the range of a wormfront. In the range of a warmfront 33-13

Median volume diameter MVO of clouds of a high pressure area had predominantly small values (between 11 and 23 pm), Only close to cloud base respectively .- i __(99 %) close to cloud top it could get larger values (see Fig. 1.3.4.24). Up to 90% of the TWC was formed by particles with diameter between 2 and 32 pm (see Fig. 2.5.6.25.26). MVO of clouds in the range of a warm front was strongly fluctuating between small and large values. I! ., 2 5 The moximum volues were between 100 ond 460 pm (see Fig. 7.9.24). Up to 50% of the TWC was formed D by particles with diameter between 33 and 600 pm (see Fig. 8.10,25,26). In the range of a wormfront The phase of the particles in clouds of a high pressure area was always fluid (see Fig. 1.3.4). The phase of the particles in clouds in the range of a warm front could vary between fluid and solid (see Fig. 9).

In a high pressure area

Fig. 31. Total water content lWC for the particle range 2 to 32 pm, 33 to 310 pm and 311 to 600 pm, and its part --__-_---___Results ascertained from the horkontal---- saundinqs- on the total woter content of a11 the particles in % of one WC in the clouds of a high pressure area was vorying point of the horizontal structure in a high pressure area and between values of 0.17 and 0.31 g/m3 on a flight in the range of a warmfront. path of 70km (see Fig. 11.13.27). In the clouds in the range of o worm front the volues of TWC were varying between 0 and 0.32 g/m3 on a flight path of 90 km (see Fig. 17.19.27). In a spezial case, when the cloud surface was ascending in flight direction the TWC in the clouds of o high pressure area - when flying in constant altitude - was increasing respectively decreasing in dependence on flight path length (see Fig. 16). Become of ascending surface of cloud - in spite of constant flight altitude - 6. Discussion of the results the airplane had flown deeper and deeper into the cloud. Before Fp: 30km the oirplane was above the The results discussed in this paper were oscertained on maximum TWC (see Fig. 1). on Fp: 30km the airplane icing flights in winter months in altitudes up to-4000m was in the moximum TWC and behind Fp: 30km the in a region between the north edge of the Alps ond o oirplane wos below the moximum TWC. distance of -200 km in the north of it. T in the clouds of a high pressure area and in the -___-______Results oscertoined from the vertical soundisqs range of a warm front was constant in constant Total woter content TWC of clouds of a high pressure heights above cloud base (see Fig. 11,14,20,28). area was about linearly growing with growing distance Only in a special case, when the cloud surface was as- from cloud base. Its maximum was located directly be- cending in flight direction T was decreasing when flying low the cloud top and attained values between 0.40 in a constant altitude (see Fig. 16). and 0.50 g/m3 (see Fig. 1,3,4.22). TWC of clouds in the range of a warm front was stron- MVO of the clouds of a high pressure kept constant on gly fluctuating between cloud base and cloud top and smoll volues (* 20 pm) (see Fig. 11.29). Up to 93% of had maximum values in different heights . The moxi- the TWC was formed by particles between 2 and 32 pm mum values were between 0.20 and 0.45 9/m3 (see (see Fig. 30.31). Fig. 7,9,22). MVD of the clouds in the range of a warm front was strongly fluctuating between small and large values. Temperature T in clouds of o high pressure area was The maximum values were between 100 and 200 pm decreasing linearly with increasing distance from (see Fig. 17.21 29). cloud base. Only near the base or near the top, Up to 99% of the TWC was farmed by particles with T could increase (see Fig. 1.3.4.23). diameters between 33 and 310 pm (see Fig. 18.30.31). T in cluods in the range of a warm front was decreo- sing with increasing distance from cloud base (see The phase of the particles, on the flights discussed Fig. 7.9.23). here, was always fluid. 33-14

References rameters on Horizontal Soundings of Stratiform Clouds. (DLR - Report in preparation) 1. Hoffmann. H.-E.; Demmel, J.: First Stage of Equip- ping a Type Do 28 as a Research Aircraft for Icing. 12. Hoffmann, H.-E.: A Climatologiy for Cloud Physical and First Research Results. ESA-TT-855, 1984. Parameters Causing Aircraft Related Icing Degree Sever,= (DLR - Report in preparatioi,) 2. Hoffmann. H.-E.; Demmel, J.: DFVLR's Research Aircraft Do 28, D-IFMP, and its Meosurim) Equipment. ESA-TT-972. 1986.

3. Forecasters' Giude on Aircraft Icing. Air Weather Discussion Service, AWS/TR-80/001, 1980. 1. M. Holmes, RAE 4. Hoffmann. H.-E.; Roth, R.; Demmel, J.: Standardi- As a result of your investigations, do you believe that the zed Ice Accretion Thickness as a Function of Cloud FAA standard clouds are still representative of what occur!; Physics Parameters. ESA-I'-1080. 1988. in the real environment?

5. Hoffmann. H.-E.: Icing Degree Modercite to Severe: Author: If and Where in Clouds. CAS Proceedings, Jerusalem The FAA standards for icing clouds are not representative August 28 - September 2, 1988. for the real atmosphere.

6. Hoffmann. H.-E.: The Analysis of Three Icing Flights 2. R. Tnogood, Pratt and Whitney Canada with Various Ice Accretion Structures when Reaching You mentioned that your measurements indicate variance Icing Degree Severe. CAS Proceedings. Stockholm Sep- with the FAA 25, App C atmospheric conditions. Woulcl tember 9-14, 1990. you agree then that your measurements are also at variance with the JAR atmosperic definitions? 7. Roach, W.T.; Forrester. D.A.; Crewe, M.E.; Watt, K.F.: Are the results of your research being coordinated with the An Icing Climatology for Helicopters. Special Investi- new effort in the United States to re-examine the definition. gation Memorandum 12. Meteorologocal Office, Brack- of the icing atmosphere? nell, 1984. Author: 8. Hoffmann. H.-E.; Roth. R.: Cloudphysical Porame- Our measurements are at variance with the FAA and aka ters in Dependence on Height Above Cloud Base in with the JAR atmospheric icing conditions. Different Clouds. Meteorol. Atmos. Phys. 41, 247-254, After each yearly DLR Icing Flight Season we are sending 1989. special lists with our icing results to theuniversity of Dayton respectively to the Naval Research Lab. in Washington. 9. Hoffmann, H.-E.: The Horizontal and Vertical Struc- They are collecting new icing data in a contract from FAA to tures of Cloud Physical Parameters. 5th WMO Scientific re-examine thedefinitions of the icing atmosphere. Conference on Weather Modification and Applied Cloud Physics, Beijing. China. 8-12 May 1989. WMP Report 3. V. Garratt, RAE No. 12. 1 am not sure how you differentiate between solid and liquid particles. 10. Hoffmann. H.-E.; Demmel, J.; Horst, H.; LGbel, H.: A Documentation of Icing-Relevant Cloud Physical Pa- Author: rameters on Vertical Soundings of Stratiform Clouds. This is our most important problem. We apply two solutions: DLR - Mitt. 90-07, 1990. (An ESA-Translation in pre- At first observations by the pilots or flight engineers, paration). secondly with a back scatter device to determine the normalized visibility. The results of both are only qualitative. 11. Hoffmann. H.-E.; Demmel, J.; Horst, H.; Stingl. J.: The percentage between solid and liquid particles cannot be A Documentation of Icing-Relevant Cloud Physical Pa- determined. 34-1

DEVELOPMENTS IN ICING TEST TECHNIQUES FOR EROSPACE APPLICATIONS IN THE RAE PYESTOCK ALTITUDE TE r FACILITY M Holmes, V E W Garratt, R G T Drage Propulsion Department, Royal Aerospace Establishment Pyestock, Hants. GU14 OLS, England ABSTRACT equipment is a continuing process and significant improvements have recently The altitude test facilities at RAE been introduced. These are centred on one Pyestock are used in support of clearance of the most important elements of the of aero-engines, intakes and helicopter whole process, the simulation of the rotors to operate under severe icing defined cloud. conditions. An important aspect of the work is the simulation of the wet icing Some icing clouds consist of a mixture of cloud in terms of water concentration, supercooled water droplets and particles mean droplet size and spectrum. Water of dry ice and completely different spray rakes or booms have been developed techniques are used for producing these for this activity and individual nozzles two components of the cloud. Ice parti- calibrated in a purpose built wind tunnel. cles are produced by milling chips off A laser particle sizer has been used to prepared ice blocks and introducing them calibrate typical spray nozzles and into the main air flow upstream of the attempts made to establish a traceable liquid water injection point. However, standard. Although this paper mainly customers seldom specify a requirement for deals with the development of cloud dry ice and the subject has therefore not simulation, it also includes a short attracted quite the same development description of the facilities and the effort as the production of water capability for monitoring ice formation droplets. It is thus not given much and shedding. prominence in this paper. Water droplets are produced using an array of spray 1 INTRODUCTION nozzles placed in the cold inlet air stream. The droplets rapidly lose heat to Safe operation of civil and military their cold surroundings to become super- aircraft and helicopters in weather cooled, as in the natural cloud, retaining conditions which can cause ice build-up on their liquid state until impact with an engine intakes, engine fan, compressor and object triggers solidification. helicopter rotor blades is a prime concern of the air worthiness authorities. Apart from the difficulty of producing a Regulations have therefore been produced consistent uniformly distributed spray in Europe and the USA which identify the pattern across the duct, which may be up test conditions with which aerospace to 2.6m in diameter, there is the problem vehicles in ground-based facilities must of measuring the droplet size distribution comply before clearance to fly in icing itself. Although this should ideally be conditions is given. The advantages of done whilst the test is in progress, the gaining test evidence on ground-based practice at RAE has been to calibrate facilities prior to flight testing is that individual nozzles in a controlled they can provide specified, consistent and environment in a low speed wind tunnel repeatable test conditions irrespective of especially adapted for this purpose. A the prevailing weather. They also permit laser particle sizer has been used to good the use of non-flightworthy test vehicles, effect in this work. all of which contribute towards a considerable reduction in time and cost to This paper begins by reviewing the icing obtain clearance for flight in ice-forming certification regulations, noting any conditions. differences between the various regulatory authorities and the effect this can have Such test facilities exist at the Royal on setting up test conditions. The test Aerospace Establishment, England, in the facilities at Pyestock are then described form of two of the altitude test cells at together with a brief description of the Pyestock. These were primarily intended capability for monitoring ice formation for steady-state and transient performance and shedding. Finally, the development of evaluation of air breathing missile and water spray rakes is discussed, with aero engines, but a capability for icing particular attention paid to the tests was recognised and incorporated at calibration of spray nozzles and the the design stage. Although there has been development of special equipment for that long experience of icing tests extending purpose. over twenty years, development of techniques, instrumentation and monitoring 34-2

2 REVIEW OF ICING REGULATIONS of actual meteorological data by the National Advisory Committee in Aeronautics Icing tests at Pyestock form part of an (NACA) contai-ned in Refs, 6,7 and 8. Ref overall icing certification programme 6 gives two sets of curves depicting cloud agreed between the manufacturer and one or liquid water content (LWC) versus mean more of the three regulatory authorities effective droplet diameter or volume empowered to grant the relevant median diamet:er, (VMD) , over an ambient operational clearance for aerospace temperature range of O'C to -30°C for vehicles manufactured and/or tested in the continuous maximum icing and 0°C to -40°C United Kingdom (UK). These authorities for intermittent maximum icing. Ref 7 are:- gives two curves of LWC factor versus cloud horizontal extent to allow for (a) The UK Ministry of Defence duration of flight in icing conditions and Procurement Executive (MOD(PE)) which Ref 8 is the source for plots of ambient deals with military equipment covered by temperature versus pressure altitude. the appropriate Defence Standard (Ref 1). The spectrum of droplet sizes to be (b) The British Civil Aviation employed for demonstrating compliance with Authority (CAA) which administers the icing regulations is not strictly defined. Joint (European) Aircraft Requirements Probably beca.use it would be very (JAR) applicable to modern transport difficult, if not impossible, to simulate aircraft and propulsion systems (Ref 2 and the range of sizes which occur in real 3) and also the old British Civil Aircraft clouds. The droplet spectra tabled in Def Requirements which still apply EO types Stan 00-970 a.nd AC 20-73 all plot as. originally certified to these regulations. normal distributions and are aimed at the theoretical assessment of the icing hazard (c) The Department of Transportation of in terms of water catch rate (20 micron the United States of America (USA) Federal droplets) and water impingement limits Aviation Administration which issues (50 micron droplets). Federal Aviation Regulations (FAR) (Ref 4) and associated Advisory Circulars which At Pyestock, the practical operating are applicable to UK manufactured conditions recommended for the air blast aerospace equipment which is required to spray nozzles are those which produce a operate in America. near normal size/volume distribution; that is, as far as possible, bimodal The nature of the regulations, which distributions are avoided as giving applies to all authorities, demands that unrepresentative icing conditions. clearance to operate in icing conditions is dependant on a three-part assessment. In general, the icing test requirements The first requirement is for an in-depth for the civil aircraft authorities quoted theoretical analysis of the susceptible show very close agreement and it would icing areas on the flight vehic1.e at the appear that there is a gradual convergence most severe atmospheric conditions which towards a common policy covering all may produce ice accretion and their affect aspects of clearance to operate in natural on the vehicle as a whole. This is icing conditions. followed, wherever posssible, by full- scale rig tests at these critical condi- The main differences at the moment lie in tions and, finally, flight tests are the requirements for the simulated performed in real icing environments. The conditions for testing rotorcraft and total programme should be agreed between those for testing turbine engines. In the the aerospace manufacturer and the airwor- former, the CAA has reduced the severity thiness authorities, but the demonstration of the icing atmosphere below 3000 m from of satisfactory operation under simulated that quoted in Refs 3 to 6. In the conditions is the sole responsibility of latter, whereas the CAA/JAR requires the manufacturer under the scrutiny of the simulated altitudes, forward speeds and appropriate authority's official:; or their engine powers for the flight regime under delegates. scrutiny, the FAA states that compliance with regulations can be adequately All three regulatory authorities refer to demonstrated over a range of engine powers the same two standard wet icing at ground level static conditions but with atmospheres detailed in Refs 5 to 8. considerably increased liquid water These are the Maximum Continuous icing concentrations. conditions relating to long tract:s of stratiform cloud and the Maximum 3 DESCRIPTION OF ICING FACILITIES Intermittent icing typical of short span cumuliform cloud, illuscrated in Fig 1. The altitude test facilities at Pyestock, Both are based on a statistical treatment pictured from overhead in Fig 2, consist 34-3

of five test cells which are provided with cross-flow heat exchanger containing pipes air from a central compressor house at the through which an aqueous ammonia solution pressure and temperature corresponding to at temperatures down to -52°C is the required altitude and Mach Number circulated. The ammonia is stored in two conditions. Exhauster-compressors in the 500 tonne capacity insulated tanks which same building extract the exhaust gases provide a reservoir to maintain cold and depress the pressure in the test conditions for the duration of a test. chamber to the required altitude. The The tanks are recharged by a refrigeration test vehicle, be it aero engine or test plant over a 24-hour period. Thus, rig, is mounted in the test cell and whereas Cell 3 can be provided with cold measurements of pressures, temperatures, air for periods of 3 to 4 hours, limited fuel flow, thrust, etc are taken by a only by air drying capacity, the duration computer-controlled data-gathering and of tests in Cell 3 West is limited by the analysis system. rate of circulation of the stored coolant to between half an hour and three hours. Of the four test cells currently in The actual time 'on condition' is commission, two have the capability of determined by the required air mass flow, conditioning their inlet air to the sub it's temperature and the quantity of water zero temperatures required for simulating in the ambient atmosphere. icing conditions. These are Cells 3 and 3 West, the former being used primarily for In Cell 3 West, the consequence of not testing military engines and the latter drying the inlet air prior to entering the for testing large civil fan engines. For cooler is that some of the entrained water wet icing tests a water spray rake is vapour is deposited on the cooler tubes in mounted in the cell air inlet duct which the form of frost which progressively is used to inject a cloud of water builds up until its effect on the natural droplets into the airstream ahead of the frequency of the tubes is to cause them to test vehicle. An installation diagram of resonate in the air stream leading to such an arrangement is shown in Fig 3. curtailment of the test. This phenomenon Both facilities have large test chambers: is a function of the atmospheric water Cell 3 is 6m diameter and Cell 3 West, vapour content, air flow, air inlet which has a diameter of 7.6m, is big temperature and duration but is enough to accommodate a helicopter restrictive in practice only on the fuselage (less rotors) as well as the largest civil fan engines at lower latest civil fan engines. A majority of altitudes, ie where air flow demand is the aero engine testing is done in the greatest. The frost particles also have a connected mode, in which all of the inlet tendency to shed suddenly when the air air flows through a duct into the front of flow is increased, such as during a slam the engine. Whenever plant capacity acceleration of the engine under test, allows, however, it is preferred to test unintentionally producing a dense cloud of engines in conjunction with their intakes ice particles which is unrepresentative in the free jet mode, a method which is for the scheduled wet icing test. always applied to helicopter rigs. For this technique, air is discharged from a A further difference between these two subsonic nozzle to envelope the test inlet conditioning systems is the effect vehicle thus giving a better representa- on humidity. Ideally, the relative tion of the free-stream flow field humidity of the air should be between 85 presented to it. to 95% at the point where the droplets are injected so as to minimise the evaporation There are significant differences between of droplets. Cell 3 has an advantage in the two cells in the way the inlet air is this respect as the air is initially dried conditioned, which affects the test and brought up to the desired humidity envelopes that can be covered and the using a steam injection system. Cell 3 relationship between the simulated and West, on the other hand, has to accept natural icing conditions. In Cell 3, air whatever humidity levels occur following is drawn from atmosphere and dried by partial freeze-drying of ambient air passing through silica gel beds into the during its passage through the cooler. facility compressors, which then feed a Experience shows, however, that by proportion of the high pressure air after avoiding icing tests on particularly humid a further drying process into a cold air or very dry days., and even choosing most turbine (CAT) thereby reducing its favourable times of the day, that this is temperature to a minimum value of -70°C. not a too serious drawback. Humidity is This is then mixed with warm air in a monitored using a Michell cooled mirror chamber adjacent to the test cell to dew point probe placed in the low velocity produce the required inlet air air upstream of the spray rake which temperature. In Cell 3 West, air is provides dew point temperatures from which induced from atmosphere into a 3-stage 34-4

the relative humidity at the wat:er spray or entry spinner of an engine. Remotely rake is then automatically calculated. operated high definition still camera shots are also taken for record purposes, The icing cloud is produced by a number of typical of which is Fig 5, whilst high- air blast water spray nozzles placed in speed cine up to 1200 frames per sec is horizontal rows about 127 nun apart on an available for recording the ice shedding array of arms forming a sturdy structure, and resulting particle trajectories which contain the water feed and air associated wi.th the post-icing engine supply to individual nozzles. Fig 4 shows accelerations. For connected tests, a typical installation of a spray rake. camera viewing windows are mounted in the Three rake assemblies are available inlet duct. These are kept frost and mist ranging from one with 37 nozzles for a free by electro-thermal heating. 0.9 m dia duct to one with 242 nozzles for a 2.4 m dia duct. A new rake for the 4 PRODUCTION AND MEASUREMENT OF WATER future family of large civil engine will DROPLETS require 310 nozzles for a 2.64 m dia duct. The water supply is demineralised and held 4.1 SDrav nozzle in a tank at about 20°C pressurised to 690 kPa (100 1b.in’ abs), flow being Airblast atomising nozzles are used to controlled by remotely-operated valves. produce a cloud of water droplets, a The water injection system is deceptively typical nozzle being shown in Fig 6 complicated and has been the subject of comprising a central water nozzle extensive development at Pyestock over surrounded by an annular air passage. many years, justifying a separate Three different sizes of water nozzles are description of it and its calibration in currently available depending on the LWC the following sections. range required. The water jet emerges at a low velocity compared with its The overall spray pattern of the various enveloping air jet, this high relative water injection rakes is checked in the velocity promoting the break-up of the test chamber. This is achieved by water into fine droplets. A noteworthy mounting a target grid of rods downstream feature of this nozzle is the protrusion of the spray rake and blowing air at a of the water nozzle from the air cap face, temperature no higher than -15°C with both which allows the water flow to be set high and low water flow rates. The low largely independently of the airflow temperature ensures that all droplets although there is some increase in water impacting on the grid will freeze and the flow with atomising air pressure due to resulting ice accretion pattern when the ‘ejector’ effect. If, as in some examined after 3 to 5 minutes of water nozzle designs the water nozzle is injection gives an excellent indication of withdrawn into the air cap, then the water the uniformity of the spray. flow can be strongly influenced by the atomising air pressure. Dry ice particles are introduced into the test cell in a totally different way, The other main parameter in producing the upstream of the water spray rake. They required droplet VMD is the atomising air are produced by milling ice blocks which pressure. This is measured by means of have been moulded and then cured for hypodermic tubes inserted into two of the several days at -20°C. The required control spray arms connected to separate number of cured blocks are loaded onto a pressure transducers, ensuring that the variable speed conveyor inside a pre- pressure is measured as near as possible cooled chamber and are moved into contact to the actual nozzles. Because of the low with a variable speed multi-blade airflow rates, the arms act as plenum cutter. The resulting ice particles, chambers. The mean of these measurements which are approximately 1 mm in size, are is taken as representative of the whole collected and fed into the main airstream rake and the atomising air pressure is set through 26 distribution tubes of varying using the difference between this and the lengths. rake duct static pressure. Before dealing with the subject of 4.2 sflowcontrol producing and measuring water droplets, mention must first be made of the The control and accurate measurement of extensive facilities for viewing ice the water flow through the spray rake is accretion and its subsequent shedding from important both. for the realisation of the the test vehicle. Five channels of closed specified LWC and the production of circuit TV are available for remote correct droplet VMD. The water flowmeters viewing and recording on tape. One of used for this purpose are either of the these channels can provide a strobed view Pelton wheel or turbine type and are of rotating parts, such as the fan blades calibrated before each icing installation 34-5

using a dedicated traceable gravimetric thousands of droplets in a few seconds and calibration system. On the large rake can be operated remotely. there is a fundamental problem in that a head difference of up to 2.4 m between the Various lenses can be used with the top and bottom arms causes unequal water instrument covering different particle flows if not corrected. The current size ranges and working distances. The method of solution uses constant level latter is defined as that distance from weir chambers on each arm in which the the lens which must contain all the spray. water levels are maintained by optical Experience has shown that the best working sensors controlling valves in the water arrangement is with a 300m lens giving a feed line, shown in Fig 7. The water is working distance of 400m and a particle displaced from the chambers by (droplet) size range of 5.8 to pressurising them with nitrogen gas. 564 microns. This working distance Whilst very successful in producing equal accords well with the 610m tunnel width. flows to the spray nozzles, this system is The lower droplet size limit of subject to pulsations in the water supply 5.8 microns is not a disadvantage in to the weir chambers caused by the practice as any volume of liquid contained continual opening and closing of the below this size would generally be a very valves in the feed lines. This has made small proportion of the total in the the on-line measurement of total water distributions typically used for icing flow difficult. New improved methods of cloud simulations. Additionally, the flow control are currently under Malvern software takes some account of the development. undersize droplets. 4.3 SDray calibration facility The Malvern particle sizer has become a widely accepted method of particulate The water droplets should preferably be measurement and is generally self checking measured in the test cell close to the in that mis-alignment or dirty lenses are test vehicle, but in practice this has not immediately obvious. However, a been found to be feasible in a large-scale Verification Reticule is also used as an engineering environment not conducive to additional functional check. This the use of delicate instruments. The consists of an optical glass flat which alternative solution is to calibrate can be attached to the receiving lens of individual spray nozzles in a separate the particle sizer on which about 10000 facility. At Pyestock, this comprises an chrome dots of known sizes are deposited open circuit wind tunnel having a 0.37 m* randomly within an 8mm diameter circular working section connected to exhauster area. The VMD of this array is initially compressors with capacity to generate up determined by the manufacturers within a to 80 m/s air velocity over a twin nozzle specified tolerance of +2 microns. If the spray mast, as shown in Fig 8. A laser particle sizer repeats this measurement particle-sizer produced by Malvern within the above tolerance then it can be Instruments is mounted at the side of the concluded that the system is functioning working section with its beam directed correctly. With such a complex opto- through the droplet cloud shown in Fig 9. electronic system this alone is of great This instrument's principle of operation worth. However, initially it was thought is the measurement of the diffracted laser that the reticule might also serve as a light pattern generated by the droplet much needed transfer standard for forward cloud intercepting the laser beam, Ref 9. diffraction particle sizers. The pattern is then converted into a Unfortunately, a thorough investigation by droplet volume spectrum by Malvern the National Physical Laboratory (UK) Instruments proprietary software using an showed that this was not feasible for Olivetti computer. A typical computer various reasons, eg the random overlapping output showing the spray analysis is shown of the dots make the accurate independent in Fig 10. As can be appreciated, this calculation of the VMD impossible. widely accepted instrument, Refs 10 and 11, is a major advance on earlier 5 TYPICAL SPRAY NOZZLE CALIBRATIONS techniques based on the use of oiled or coated glass slides onto which droplets 5.1 Nozzle characteristic curves were captured and photographed for later analysis. The laborious task of sizing a The presentation of spray calibration few hundred droplets under a microscope facility results in the form of nozzle against a standard graticule did not lend characteristics (Fig 11) at constant water itself to accurate and repeatable flow rates has marked advantages. The measurements of a large number of samples. data gathering is expedited by only The laser instrument, on the other hand, needing to change the air pressure is non-intrusive, on-line, counts many significantly at each test point. Generally, the whole curve can be covered 34-6

in about thirty minutes of testihg with -5°C air stream in under a metre or so, a about fifteen test points. The form of 30 micron droplet with 27 times more mass, the line is well defined with little might require 5 metres, although it would scatter making for excellent overall still be supercooled within about a metre. accuracy. The asymptotic curves clearly show the operational limits of t:he From this point of view, the separation nozzles. For instance, a required VMD between the spray rake and the target becomes impossible to achieve above a needs to be greater than 5 m to ensure certain water flow for a given size nozzle representative ice accretion, especially irrespective of atomising air pressure. with a broad droplet spectrum or a VMD If the LWC is too high, a larger (water) greater than 20 micron. nozzle becomes necessary to produce a 20 micron VMD, say. An intrinsic problem with this theoretical analysis is that the initial heat This is further illustrated by the transfer is strongly dependant on the 20 micron working curves shown in Fig 12 droplet to air stream relative horizontal comparing four different size nozzles. velocity. As the emergent water jet, These show that the limit of water flow before break.-up,is surrounded by the which can produce a 20 micron VMD varies (warm) atomising air it cannot be known from 10.4 litre/% for a 0.41 mm dia nozzle precisely at what velocity the drop is to 16.3 litre/h for a 0.91 mm dia nozzle. actually formed and exposed to the cold air stream. The way round this has been 5.2 Effect of main stream air velocity to consider the two possible extremes of initial droplet velocity, ie zero and main As well as the main parameters of water air stream velocity, and produce two flow rate and atomising air pressure, it cooling curves where the true state of has been observed that the main air stream affairs lies somewhere in-between. For velocity over the spray rake also the purposes of this paper and to influences droplet size, which reduces as illustrate simply the comparative thermal air velocity increases. The reason for history of three different drop sizes, this is not fully understood, but may be only the zero velocity case has been due to the increased turbulence causing presented. secondary atomisation. An investigation of this effect was thus undertaken and the 6 CONCLUDING REMARKS results of VMD vs tunnel air speed over a range of nozzle water flows are shown in Although a solid data base on the Pyestock Fig 13. It can be seen that this effect spray nozzles has been achieved there is is quite appreciable which justifies the considerable scope for further work mainly use of a tunnel for such work rather than through extending the capability of the operating in still air. One interesting spray calibration facility. feature of these results is that the higher the water flow the earlier, in The present air velocity limit of 76 m/s terms of air speed, does the effect is not fully representative of the commence. A factor in this may be the velocities used in the actual icing tests, increasing momentum of the water jet and as previously mentioned, this allowing it to persist, so giving the parameter can affect the droplet VMD. It secondary turbulence longer in which to is intended to increase this velocity by have an effect. reducing the tunnel area by means of an insert. This will consist of a tube of 5.3 Effect of main stream air 0.4 m dia supported and sealed centrally temperature in the existhg tunnel. Provision will be made for the Malvern transmitting and As the demineralised water supply fed to receiving lenses to be sealed into this the spray rake is maintained at 20°C to tube. In thi.s way the tunnel air velocity prevent it freezing before or on reaching can be increased to 150 m/s. the nozzle, it may be questioned if the droplets have become supercooled and/or All the measurements reported on here have reached air stream temperature by the time been made wit:h the spray bar 0.7 m from they arrive at the target, as they would the laser beam. This distance is often in a natural cloud. This is important exceeded in i.cing tests and the effect of because Ref 12, for instance, shows that this is not knovm and needs to be ice accretion takes different forms investigated. Increased evaporation or depending on the state of the droplets at coalescence may occur, changing the VMD. impact. Theoretical studies at Pyestock, The present tunnel arrangement only allows some results from which are given in fairly small changes in this sampling Fig 14, suggest that while a 10 micron distance and therefore a scheme has been droplet comes to thermal equilibrium in a suggested whereby a group of nozzles is 34-7

supported on a 'sting'. This sting is 2 Joint Airworthiness Requirements: then introduced at the mouth of the tunnel JAR-2s large aircraft and JAR-E and can be inserted varying distances engines, 1986. downstream into the working section of the tunnel, giving a maximum separation of 3 British Civil Airworthiness 4.8 m. Requirements: BCAR 29 rotorcraft (post 17.12.86), BCAR-G rotorcraft As the tunnel air velocity and the nozzle (pre 17.12.186) and Paper G610 to laser beam distance increases so does Issue 2 18.9.81 rotorcraft. the spray become more diffuse. On the standard Malvern instrument it is 4 Federal Aviation Regulations: necessary for the spray to obscure 10 to Airworthiness Standards Part 25 30% of the laser beam to give sufficient transport category airplanes, scattering signal. Because of the very Part 27 normal category rotorcraft, low water flows sometimes specified in the Part 29 transport category rotor- icing tests, even the use of two spray craft and Part 33 aircraft engines. nozzles has not always allowed this criterion to be strictly satisfied. For 5 Aircraft ice protection. FAA this reason the Pyestock particle sizer advisory circular AC 20.73 1971 has had an extra amplification stage fitted allowing the obscuration to go as 6 Continuous maximum and intermittent low as 2%. maximum atmosphere icing conditions: liquid water content versus mean However, it is envisaged that with the effective drop diameter. increased tunnel speed and separation even NACA TN 1855, 1949. this may not be sufficient. Therefore the enhancement scheme will provide two 7 Continuous maximum and intermittent representative groups of five and seven maximum atmospheric icing nozzles. These will be sting mounted conditions: liquid water content complete with their air and individual versus cloud horizontal distance. water supplies, and will afford the option NACA TN 2738, 1952. of choosing any number of active nozzles within each group thereby permitting the 8 Continuous maximum and intermittent obscuration to be adjusted depending on maximum atmospheric icing water flow rate, target distance and conditions: ambient temDerature tunnel air velocity. versus pressure altitude. NACA TN 2569, 1951. Future work at RAE will thus continue to be aimed at meeting customer requirements 9 Swithenbank, et al, "A laser and satisfying the evolving demands of the diagnostic technique for the international regulatory authorities. In measurement of droDlet and Darticle this respect, a document about to be size distribution": Progress in published by the FAA entitled "The Astronautics and Aeronautics 53 Aircraft Icing Technology Handbook" should (1977) 421. stimulate international efforts towards producing a single comprehensive set of 10 Dodge, L.G. and Cerwin, W.A., icing requirements aimed at worldwide "Liquid particle size measurement application. International effort might techniques". ASTM STP 848 edited by also be appropriate to update the existing Tishkoff, Ingebo and Kennedy, 1984. icing cloud characteristics bearing in mind that these are based on data gathered 11 Ugur Tuzun, Farhad A Farhadpour, nearly forty years ago using flight "Comparison of Light Scattering with instrumentation far less precise than other Techniques for Particle Size modern equipment. Perhaps this should be Measurement". Particle extended to other parts of the globe not Characterization 2 (1985) p104-112. previously surveyed. The test facilities at Pyestock could play a role in this work 12 Marck, C J and Bartlett, C S., by evaluating the latest flight-standard "Stability relationship for instruments. waterdroplet crystallization within NASA Lewis icing spray nozzle". REFERENCES AIM-88-0289 26th Aerospace Sciences 1 Defence Standards: 00-970 Vol 1 Meeting, Nevada, 1988. Aircraft, Vol 2 Rotorcraft and 00-971 Engines, 1983 Copyright @, Controller HMO London (1990) 0.5 1. PRESSURE ALTITUDE RANGE,SL.TO 6700m (22,000fl) 2 MAXIMUM VERTICAL EXTENT, 200Om (6,500ft) 0.e 3. HORIZONTAL EXTENT, STANDARD DISTANCE OF 17.4 NAUTICAL MILES mE 0.7 .m 0.6 z wc " 05 E 0.4 s 3 0.3 - 3 : 0.2

0.1

0 " 15 20 25 30 35 40 MEAN EFFECTIVE DROPLET DIAMETER - MICRONS

STRATIFORM CLOUDS

I. PRESSURE ALTITUDE RANGE,1200m TO 6700m (4,000ft TO 22,000fl) 2. HORIZONTAL EXTENT, STANDARD DISTANCE OF 2.8 NAUTICAL MILES

NOTE:- DOTTED LINES INDICATE POSSIBLE EXTENT OF LIMITS.

15 20 25 30 35 40 45 50 MEAN EFFECTIVE DROPLET DIAMETER -MICRONS

CUMULIFORM CLOUDS

FIG 1 CLOUD CHARACTERISTICS FIG 2 AERIAL VIEW OF TEST FACILITY

FREE JET TED, MODE

FIG 3 TYPICAL ENGINE INSTALLATION 34-10

I FIG 4 SPRAY RAKE INSTALLATION

I FIG 5 ICE BUILD-UP ON ENGINE INLET AIR CAP \

WATER NOZ

FIG 6 TYPICAL SPRAY NOZZLE

ATMISING MANIFOLD

WTER PRESSURE CONTROL TUBE

LEVEL SENSORS

WATER lNLEl

FIG 7 SPRAY RAKE AND CONTROL SYSTEM 34-12

FIG 8 SPRAY CALIBRATION TUNNEL

DROPLET DIAMETER [ microns 1

Malvern Instruments MASTER Particle Sizer M6.10

9.1 14.4 7.2 9.8 5.8 5.3 System number 1973 Diode jp335x

FIG 10 SPRAY DROPLET SIZE DISTRIBUTION 061mm NOZZLE - - - - 041mm NOZZLE t L.5L litres lhr 0 908 litresl hr

ATOMISING AIR PRESSURE (~POI

FIG 11 SPRAY NOZZLE CHARACTERISTICS

ATOMISING AIR O/.lmm 0.76" PRESSURE IkPoI I 0.2" i' I LIMIT OF FLOW WHICH CAN I 1 PRODUCE 20um VMD

20 MICRON VMD CURVES FOR FOUR DIFFERENT SPRAY NOZZLES 100

50 10 0 15 0 0 20 0 WATER FLOW RATE llitreslhr 1

FIG 12 EFFECT OF SPRAY NOZZLE SIZE 25-

2L - 0 - 2.27 lilresl hr x .L.5L 23 - A - 948 M.M.P - MEAN ATOMISING - AIR PRESSURE E 22- 2. I ir W M.AAP-148.8kPa 21- 2. \, 5 ’A, 0 z 20- 5 0 !i!19- W f g 18-

17 -

16 -

%- 20 3u 40 50 60 70 80 MAIN AIR STREAM VELOCITY im Is 1

FIG 13 EFFECT OF TUNNEL VIiLOCITY

CONDITIONS AIR STREAM TEMPERATURE = -5% AIR STREAM VELOCITY - 914ml* DROPLET INITIAL VELOCITY = 0 \ DROPLET INITIAL TEMPERATURE = 10°C

SYMBOL DROP DIAipmI ____- 10 - 20 30

I 1 2 3 4 5 DISTANCE FROM NOZZLE iml

FIG 14 DROPLET TEMPERATURE REDUCTION 34-15 Discussion droplet size during the tests at high altitude? As you have not, are you satisfied that the parameters you use for setting up the droplet size give correct droplet sizes 1. W. Grabe, NRC Ottawa during the tests at high altitudes? Are the nozzles on a horizontal bar individually controllable and have you found flow graduations from the inlet to the Author: outlet side? It is the practice at RAE to allow for the effect of low After calibration of a nozzle in a calibration tunnel, do you prsssure at altitude an atomization performance by testing check it in a test situation and do you then adjust for nozzles in a low pressure chamber. However, by setting air differences? pressure by the pressure difference between ambient and nozzle air supply we are confident that calibration in the Author: wind tunnel is applicable to altitude. The water flow to individual nozzles is not controlled during a test, but the nozzles could be blanked off prior to a test if 3. C. Scott Bartlett, Sverdrup Technology some bias to the cloud pattern is required. A metal grid is In reference to your data showing a strong influence of used to collect ice downstream of the spray nozzles to check dioplet injection station air velocity on test section droplet wether the distribution meets the requirements. size, could you comment if this result could be attributed to We do not check the calibration done in the wind tunnel instrument error and, if not, can you offer an idea of what against mneasurements taken in the test chamber, as the process might cause the effect? latter presents a harsh environment for precision measuring equiment. The most influential parameters (air velocity, Author: atomizing air pressure) are controlled in the calibration RAE are confident that instrument error is not responsible tunnel, and we have no evidence that additional droplet size for the influence of air velocity or mean droplet size when measurements in the test chamber are justified. using the Malvern instrument, although some other instruments are known to be sensitive to droplet velocity. 2. E. Brook, Rolls Royce This is a result of the basic principle of operation of the You have looked at the variation of droplet size with air Malvern instrument that produces a stationary diffraction velocity in your calibration tunnel, but your calibration is image of a moving droplet. The effect could result from the carried out at sea level. Have you looked at the variation of reasons mentioned in paragraph 5.2 of the paper.

35-1

EXTREE D'AIK DHEI.IC0FTEK.S PROTECTION POUR LE VOL EN CONDITIONS SEICEUSES 011 CIML\STES

par Xavier de la SEKYETTE et Philippe CABRIT

AEROSPATIALE Division H6licopt&es 13725 Marignane Cedex France

ABSTRACT 2) CONCEPTION DUNE ENTREE D'AIR la fonction principale d'une entree d'air est d'assurer 2.1 Fonctions d'une entree d'air I'alimentation correcte du moteur ; de plus, celle-ci doit assurer la protection du moteur eontre les effets de I'humidite La fonetion principale d'une entree d'air est atmospherique et des objets extgrieurs, afin de garantir la d'assurer I'alimentation coi-recte du moteur en air securite du vol quelies que soient les conditions exterieures. dans twii le domaine et les attitudes de vol reven- Le eomportement des different$ types d'entree #air utilises diques pour I'helicoptere. Un certain nomhre de sur heiicoptere en conditions de neige et de givre est pr6- criteres doit Btre respect6 afin de garantir le bon sent6 en insistant sur les conditions critiques lKes a cha- fonctionnement du moteur (indices de distonion cune des solutions. L'experimentation &ant indispensable en pression et en temperature en particulier) et pour I'etude de ce type de phenomene, des moyens d'essais d'autres doivent etre optimises pour ameiiorer les specifiques ont et6 devveloppes ; ils permettent maintenant performances de I'appnreil (perte de charge entree de simuler toutes les conditions critiques de neige ou de d'air, trainee y eompris trainee y compris trainee givre identifiees ; la qualite de simulation a et6 prouvee par de captation etc... ). des essais de recoupement en conditions naturelles. Par ailleum, I'entr6e d'air helicopt&re doit assurer la protection du moteur eontre les corps etrangers, 1) INTRODUCTION contre le sable (dans certains cas d'utilisation specifiques oh les degradations des moteun sont tres Le vol tout temps sans restriction operationnelle est un rapides), et contre ies effets de l'humidite atmo- objeetif ineluctable pour tout vehicule a&mnautique. Les spherique. helieopteres n'echappent pas B cette regle, et les mnstructeurs ont certifie des appareils en condition de II convient aussi de prendre en compte I'aspect vol I.F.R., et, plus recemment, sans restriction en bruit exteme dans la definition de Yentree d'air. conditions givrantes. Enfin, comme pour tow les elements de Aujourd'hui, I'utiiisation operationnelle d'helicopteres I'h6licoptere. I'ensemhle d'entree d'air doit etre avec des conditions meteorologiques advenes devient de realisable en production : 2 un coat et une masse plus en plus courante au fur et imesure que les opera- maximum, presenter une bonne fiabilite. et ne pas teum prennent confiance dans les quaiites de ieur penaliser ies operations de maintenance. materiel. II apparait done qui la fonetion protection contre L'ensemhle des travaux effect"& pour assurer la eertifi- la neige et le givre de I'entree d'air est certes une cation des appareils en vd tout temps (plusieurs annees fonction preponderante pour ia skurite de dam le cas du Puma et du Super Puma), associe S I'appareii et indispensable i ce titre, mais qu'elle I'experienee acquise en service a permis d'ameiiorer de ne peut en aueun cas etre ie seul critere de dimen- maniere tres significative la connaissance du comporte- sionnement ; on obtiendrait aiors "ne "mauvaise" ment des entrees d'air en conditions de neige et de entree d'air peu performante. et pouvant peut &re givrage et des risques assoeies ces conditions. Un bon mettre en cause la &wit& du vol au travers exemple de cette tendance est donne par I'evolution du d'autres criteres, tels que des instabilites de fonc- reglement F.A.R. en ce qui concerne la protection des tionnement moteur. par exemple. entrees d'air en vol sous ia neige : initialement, aucune mention de cette condition n'etait faite dans le regie- 2.2 Principaux tvws d'entrhe d'air ment. et des appareils tels que le Super Freion ant et6 certifies, sans aucun essai i la neige. Puis, la notion de Les differentes fonetions d'une entree d'air &ant neige a 6th introduite dans les paragraphes de la FAR tres variees, et sa forme tres fortement dependante 27-1093 et 29-1093,et enfin, I'AC 29-2 change 2 propose de I'architecture generale de I'helicoptere (position des essais de demonstration particulierement diffilciles a et nombre des moteurs, forme fuselage et capo- realiser (1 h 30 d'essai sous la neige tombante avec une tages, recirculation gaz d'echappement dam le flux visihiiite inferieure B 114 mile ...) rotor, etc...), et du moteur, il'n'est pas possible d'envisager une forme univeneile d'entree d'air. On Le but de ce document est de synthetiser I'experience peut cependant degager 2 familles : aequise par Aerospatiale en ce qui concerne ia protection des entrees d'air en conditions de neige et de givre. En particulier, ies conditions critiques de neige et de givre determinees sur les differentes entrees d'air recemment developpees par Ahrospatiale seront citees. De plus, les moyrns specifiques d'essais d~vveioppesrecemment pour etudier et justifier ies entrees d'air seront dewits. 35-2

1) Entrees d'air statiwe Les entrees d'air dynamique permettent de recuPe-- rer la pression dynamique en voi d'avancement et Leur pian d'entree est grossierement parallele a d'abtenir une tres faibie distorsion de pression i'ecoulement d'air en voi d'avancemtmt (Cf Fig. 1). dam ie pian d'entrde du compresseur.

"OLUlE Le dimensionnement optimal est determine par :

-la position frontale des entrees d'air par rapport aux capots. ce qui implique parfois I'installation de longs conduits (plus diffirciles installer et plu:; ICOMPRESSEUR iourds). La perte de charge propre aux conduits est aiors compensee en voi stationnaire par EN'"EE I'ubsence de recyeiage de gaz chauds, en voI d'avancement par la recuperation de pression dy- ~ - - - LRI I- --I-- namique et ia non-absorption de la couche limite, -le profii des levres d'entree #air doit Stn? suffisamment +pais et d'une forme qui Bvite tow: decollement des filets dair que ce soit en vol sta-- tionnaire ou en voi d'avancement.

Les entrees d'air dynamique sont bien adaptees ir des moteurs entree d'air axiale, et 2. des heiieop-. teres dont ia vitesse de croisiere est elevee.

2.3 -on de I'entrhe d'air

La fonetion "protection du moteur par I'entree est souvent giobale. et ii est difficiie de separer ia protection contre I'humidite atmospherique de:; GRILLE autres cas (corps &rangers en particuiier). Pouir cette raison, ies diverses protections sont mention.. Figure ? : ENTREE UAIR STATIQUE nees, mais seuies celies concernant I'humiditti Les entrees d'air statique presentenl. des avantages atmospherique sont d6tailEes. qui sont : Le chapitre 3 decrira les conditions precises de - simplieit6 neige et de givre qui doivent &e demontr6es. mai:; - faibie masse ii convient des maintenant de rappeier ies diffe-. - moins de risques d'ingestion de corps etrangers rentes formes d'humidite atmospherique qu'un - trainee de i'appareil reduite par une meiileure helicoptere peut reneontrer : integration au orofii du fuselage. Bien cancue, et apres optimisation, une telle entree ])La piuie. qui est de I'eau sous forme iiquide. d'air est capable de recuperer une partie IJusqu'a presente dam I'atmosphere en gouttes de gro:; 50 %) de la pression dynamique du voi diametre, et avec une temperature ambiante d'avaneement. Par contre, I'indice de distorsion des positive. La reaiisation de I'entree d'air d'un pressions dans le plan du compresseur et ie risque heiicoptere doit etre teile qu'il n'y ait pas de d'absorptian d'air chaud restent en general plus risque de drainage vers ie moteur d'eau collect& importants que dam ie cas d'une entree d'air sur ie fuselage. Dans ce cas ies demonstrations: dynamique. effectuees par ies motoristes pour la certification. moteur permettent de dhontrer le bon La configuration optimaie est aims defini par : fonctionnement.

-un rayon de courbure de la levre amont de I'en- 2)La piuie vergiagante, qui est de la pluie tree d'air tres important afin d'eviter ie decoile- survenant par une temperature exterieure ment des fiiets d'air et de recuperer une negative. Cette configuration entmine des partie de la pression dynamique a grande captations tres rapides sur I'ensemble de la vitesse. Structure de I'heiicopt8re. et aujourd'hui, aucun helicoptere ne peut etre justifie dans ces Cette configuration. dans la mesure 00 ia concep- conditions. Ce cas est done cite pour memoire et tion generaie de I'appareil permet d'eviter ia il. ne sera pas prls en compte dans ia suite de reingestion d'air chaud, constitue un excellent cet expose. eompromis du point de vue performances de i'appareii. 3)Le givrage, qui est ia presence d'eau iiquide dans I'atmosphere B une temperature inferieure B 0% 2) Entrees dair dvnamiaue sous la forme de gouttelettes de pent diametre (ae 10 a 40 microns). L'etat "eau iiquide" avec Leur plan d'entree fait face a I'ecoulemellt d'air en une temperature exterieure negative &ant voi d'avaneement (Cf Fig. 2). instable. les goutteiettees se transforment en glace VOLUIE des qu'eiles entrent en contact avec un objet quelconque. Ce phenamene peut entrainer des captations tres importantes sur la structure hbiicoptere et dans ies entrees d'air moteur, susceptibies d'induire des endommagements ou des arras moteur. 35-3

4) La neige, qui est la presence d'eau solide (glace) dans I'air temperature inferieure a 0% sous is LZ-7 EXlRAll forme de cristaux. La forme des eristaux peut varier de manisre tres importante en fonction de la temperature exterieure et de la transforma- tion de la neige (cas de neige soulevee par le rotor). On peut grossierement dire que lorsque la temperature est basse, les cristaux sont petits et peu eollants. et que iorsqu'elle est proche de 0°C. la tendance inverse peut atre observe, pouvant entrainer des accumulations importantes sur ia structure ou les entrees d'air. AXE DE SYMETRIE J-,_- Differents svstemes sont couramment utiIis6s DOUI' assurer la Droteetion dobale des entrees d'air : Figsure 4 : SEPAPATEUR DE PARTICULES JNTEGRE

1) Systeme de chauffage 4) Systeme a grille Le principe est simple et permet de s'affranchir des probkmes lies 2 I'humidite atmospherique pour des hegrille placee sur I'entree d'air permet de prate- temperatures inferieures B 0°C en evitant la forma- ger efficacement le moteur contre les elements tion de captation de glace. I1 presente &rangers, sauf lorsque les conditions exterieures I'inconvenient de necessiter des apports d'energie imposent I'utiiisation d'un fiitre anti-sable speci- importants, qui peuvent Stre reduits s'il est fique (cas relativement limit&). De plus, cette grille possible de seulement degivrer periodiquement assure une protection efficace en condition de gi- I'entree d'air (degivrage au lieu d'antigivrage). I1 wage puisqu'elle capte les gouttelettes d'eau fat alms que le moteur presente une eapacite surfondues presentes dam ]'air. Pour assurer d'ingestion de morceaux de glace suffisante sans I'alimentation du moteur en cas de colmatage de la degradation du moteur et sans perte de puissance. grille, il est necessaire dans certains cas de prevoir Un moyen complementaire de protection contre un by-pass en arriere de la grille. I'ingestion de corps &rangers doit &re associe au simple chauffage. Pour le vol sous la neige. la grille protege le moteur contre I'ingestion de captations accumulees sur un 2) Systeme 2. impulsion point de la cellule et pouvant se detacher, par contre, eiie n'assure aucune protection intrinseque Cest un systeme de degivrage qui, au lieu de de- dans cette condition car ia neige n'accroche pas sur coiler les captations par ehauffage. assure cette ia grille ; le bon fonctionnement depend alors de ia fonction par choc. Ce systeme n'est "raiment efii- forme du conduit d'entree d'air. caee que lorsque I'epaisseur de captation atteint I1 apparait done qu'une grille externe, piacee sur plusieurs millimetres, et la majorite des petits tur- I'entree d'air. permet de proteger efficacement le bomoteurs d'helicoptere n'ont pas de capacite moteur dans la majorite des cas, il existe d'ingestion suffisante. cependant des conditions pour lesquelles cette protection peut Ctre incomplete (neige par 3) Systeme a inertie exemple), et des captations peuvent se produire dans I'entree d'air. II est alms possible de placer Le prineipe consiste a donner a I'air "ne compo- une seconde grille a proximite du moteur, qui sante de rotation, afin de eentrifuger les particules evitera ]'ingestion par celui-ei de captations (eau, sable, corps etrangers, glace) contenues dans formees dans les conduits d'entree d'air : diffhrents celui-ei. Cette mise en rotation peut Ctre obtenue a essais ont demontre que la protection appokee par partir dune serie de vortex montes sur un panneau la grille externe etait suffisante pour eviter tout (Cf Fig. 3) (tres bonne efficacite au sable, mais colmatage dans I'entree d'air ou sur la grille perte de charge importante), ou bien dam I'entree interne. Un exemple de protection d'entree d'air d'air moteur (I.P.S.) (Cf Fig. 4) (efficacite de I'uide de 2 grilles est present6 sur la figure 5. separation moins bonne, mais perte de charge mains importante). Dans certains cas, des deflec- teurs sont places dans I'entree d'air helicoptere pour assurer eette fonction. Dam tous ies cas, il est necessaire d'y assoeier un ensemble d'extraction des particules centrifugees soit par ventilateur, Soit par effet de trompe. Ces systemes sont efiieaces en ce qui concerne la protection contre les objets &angers (y compris ies morceaux de glace), mais GRlLLE DENTREL il est souvent necessaire d'y associer un antigivrage DAIR de certaines Darties DOW assurer une protection complete. k+ GRILLE 4 AXISYMETRIQUE

Figuse 5 , EXEMPLE DE SYSTEME DE PROTECTIONA DOUELE GRILLE L'avantage principal des protections a base de grille est que la perte de charge apportee par ces dernieres est relativement faible (des ant de plus tendance & ameliorer les indices de distorsian), et que c'est un systeme passif ne necessitant aucun apport d'energie et d'une fiabiiite quasi infinie. De plus, la penalit6 en masse est negligeable comparee aux autres solutions. Ces differents elements ont conduit Aemspatiale a retenir ce principe de protection sur tow ies helicopteres recemment concus, bien que leur mise au point soit plus delicate que les autres Systemes et qu'un certain nombre de conditions critiques doit Ctre demontre. 35-4

3)CONDITIONS DE NEIGE ET DE GNRE PL JUSTIFIER - Deux cas sont & consid6rer : POINTS CRITIQUE9 Vol & mande vitesse (V > 70 kts) 3.1 Conditions civrantes Le tube de courant alimentant I'entree #air est L'atmosphere givrante est definie traditionnelle- concentre ee qui entraine ie passage d'une grande ment par la FAR 25 annexe C ou la FAR 29 annexe quantite d'air donc d'eau B travers Une surface de C auxquelles s'ajoutent des exigences de fonction- grille faible. La zone interessee se colmate aiors nement a" sol en brouiiiard givrant (FAR 29-1093 rapidement et I'aiimentation du moteur s'effectue b2). alors par la zone rest& propre.

La FAR definie les combinaisons de temperature Pour traverser cette zone les filers d'air sont de- exterieure. concentration d'eau, diametre de gout- flechis el cela entraine "ne separation des telettes, et taille de nuage qui doivent Btre gouttelettes avant ie passage au travers de la grille, demontrees. celle-ci reste Claire de captation dans certaines zones (Cf Fig. 6). La formation de ee bouciier est Dune maniere gCn6rale on peut dire que les capta- rapide, done peu de gouttes surfondues pen(rtrent .% tions sont d'autant pius importantes que la I'interieur du conduit #entree d'air. concentration est Clevk, la temperature faible, et le temps d'exposition au nuage givrant important. II convient cependant de rappeier que la concentration maxi diminue avec la temperature.

Les points critiques B couvrir, pius particulierement au cmrs de la justification de I'entree d'air dependent largement du type de protection retenu : 1) Si la protection est un anti-givrage par chauffage, il est necessaire de demontrer son bon fonctionnement dans les conditions de temperature minimale revendiquks, avee la concentration d'eau maximale ; de plus, la source d'energie de Figsure 6 : PROiECJlON ANJlGlVRE PAR GRILLE ehauffage doit Btre ajustk au niveau minimum FN VOL D.'AVANCEMENI que I'm pat rencontrer en vol (regime moteur minimum pour chauffage h 1'2, tension minimum pour chauffage electrique). II peut Ctre neces- Vol & faibl- (V < 30 kts) saire, dam eertains cas, de eoiivrir certains defaits de fonctionnement du syst.&mede chauf- Les lignes de courant sont beaucoup pius espacees fage si une fiabilite sumsante ne peut pas Ctre et on observe un colmatage beaucoup plus lent de d6montree. la grille mais plus uniforme (Fig. 7). Si le debit d'air par unite de surface n'est pas trop eieve, la On mentionnera enfin ia verification du non-en- grille peut rester poreuse, les captations sur les fils dommagement de I'ensemble entree d'air en du grillage assurant une defiexion sufisante pour fonctionnement a la temperature maximale &iter le colmatage du passage entre les depats de d'utilisation du degivrage, et wee line alimenta- glace. Un dimensionnement adCquat de la grille ou tion en Cnergie maximale. I'installation d'un by-pass permet d'assurer l'alimentation correete du moteur. Un des points devant Btre verifie pendant les essais est I'absence de point froid en aval d'une La formation de la protection naturelle par les zone chauffee, car la probabilite de recongelation captations de glace est plus lente que dans le cas est elwee. de ia grande vitesse ce qui entraine la penetration dans le conduit d'entree d'air d'une plus grande 2) Dam le cas de degivrage par chauffage. les points quantite de givre. Les accumulations eventuelles exposh ei-dessus sont applicables. mais ii faut y doivent pouvoir Btre integrees par le moteur sans associer la recherche des conditions oh le taux trouble de fonctionnement. de captation est le plus important.

3)Pour un degivrage pa!. impulsion, deux conditions peuvent Ctm cnt1ques : celle correspondant au taux de captation maximum, et le cas de temperature proehe de 0°C. avec les- quells les captations sont relativement molles et mains sensibles aux accel6rations g6neree.s par les impulseurs.

4) Contrairement au cas des protections par chauffage. le grilles ne presentent pas de problemes d'une maniere generale B faible temperature : les captations sur les fils de la grille solit dures et se detachent lorsqu'elles atteignent une certaine epaisseur. ee qui permet de demontrer une durCe de fonctionnement sans limitation. On doit cependant verifier dans ces conditions qu'il n'y a pas de risque de colmatage de I'ensemble de I'entrbe d'air par les captations se produisant sur la grille. Figure 7 : PROJECJION AMiGiVRE PAR GRILLE A BASE VITESE 35-5

Lorsque la temperature est superieure i -5% la forme des captations sur la grille evvolue : ceiies-ci deviennent beaucoup plus larges et accrochees autour des fils de griiie, alors qu'a faible temperature, la captations se forme a partir du point #arret et en amont de ceiui-ci ; le colmatage de la grille est souvent rapide car les vides de maille sont remplis par la glace. Ce phBnom6ne s'explique par un retard a la congelation di) i un trop faible &art de temperature entre I'ambiante et 0°C pour obtenir Une congelation instantan& (voir travaux de simulation de ce point dans le paragraphe 5). Pour des'temperatures tres proches de 0°C le retard peut &re suffisamment important pour que ia grille capte partiellement les gouttelettes d'eau surfondue et que par contre des captations se produisent dans le conduit d'entrBe d'air et puissent induire des instabilites de fonctionnement du moteur. Ce phenomene ne se produit que dam un domaine de temperature tres etroit (de I'ordre de 1,5"C), et il est largement dipendant de la forme de la griiie et de sa position relativement B Yentree d'air. La figure 8 presente une eourbe experi- mentale donnant des captations relevees en aval de la grille en fnnetinn de la temperature.

EiPiT DE LA ItMPIRAWRE

figure 8. WOLUTDN DES DEPOiS DE GLACE DAN3 It CONDUll EN FONCllON O€ LA l€M?€RATUR€

Une attention particuliere doit donc etre apportee i la verification du bon fonctionnement de YentrBe d'air i temperature tr6s proche de 0°C.

3.2 b& Les conditions d'homologation des helicopteres en vol sous la neige ont et6 prkishes par I'indice 2 de I'Advisory Circular 29.2.

Outre la verification du bon fonctionnement de Yentree d'ah pour differents types de neige et de temperature, il est necessaire de faire un voi d'une heure et demie dans des conditions de precipitation extremes (visibilite inferieure a 400m sans brouillard representant environ 1 g/m3), et comportant 5 mm de stationnaire en neige recirculante, ainsi que 1 heure de vol de croisiere, le reste du temps Btant des operations de roulage ou d'attente au sol.

II est tvis diffieile de trouver dam la nature des conditions de concentration requises par ce texte ; de plus, la condition de vol en stationnaire neige recirculante ne correspond pas B un cas de vol ope- rationnel ear ia visibilite dans ees conditior nuile et aueun pilote ne restera dans de conditions. Ceci montre donc la severit6 de cc genees vis B vis des conditions reelles d'utilisatian. 35-6

Les systemes a inertie, de par leur forme, neces- Tout d'abord le domaine de vitesse a et6 abaisse a sitent une protection contre la neige (chauffage 10 kts ee qui permet "ne simulation satisfaisante pour I.P.S.). II est possible de $'en affranchir du stationnaire. Ceci necessite, outre I'etablissement dam le cas des filtres type vortex en montant d'un ecoulement stable dans la soufflerie givrante, ces derniers parallelement a I'ecoulement d'air. II de realiser un nuage givrant homoghe avec des y a alms effet de separation de la neige en vol quantites d'eau iry'ectees faibles (utilisation d'un d'avancement. petit nombre d'iry'ecteun adapt& B ees conditions particuli(.res). Afin de garantir une bonne simula- Comme eeci a deja et6 mentionne ei-dessus, les tion, il est apparu necessaire de verifier grilles sont totaiement transparences BUX nocons I'ecoulement d'air autour de la maquette et de le de neige : elks n'assurent une protection qu'en comparer & celui relev6 en stationnaire sur helico- arrctant les captations qui peuvent se former en ptere (vi:;ualisation par fils de kine). D'autre part amont de la grille. les bancs d'essais ont 6th instrumentes et calibres pour effectuer des essais pour une temperature Cest une des raisons pour lesquelles une proche de 0°C. seconde grille est montee dans certains cas au plus pres de I'entree d'air moteur ; elk evite Dans les deux cas. il a ete mis en evidence la ne- I'ingestion par celui-ei de captations qui peuvent cessite de disposer dun modele theorique permet- se former dans les conduits #entree d'air (voir tant d'evaluer I'influence des parametres d'essai sur paragraphe 2.3 et figure 5). la qualit4 de la simulation au niveau de la maquette en essai.

En effet le nuage givrant est r6alise par la pulverisation de I'eau necessaire i temperature positive dans un courant d'air dont la saturation en eau 4.1. Movens d'essais au give : peut ne pas &re totale.

Les essais de mise au point et de certification en Pour aboutir une simulation satisfaisante d'un givrage des entrees d'air d'helicopthre s'effectuent nuage givrant il est necessaire que les gouttelettes depuis de nombreuses annees dans des souffleries arrivent dans la section d'essai en @tatde surfusion Permettant de reproduire les conditions givrantes (temperature inferieure a 0°C) i une vitesse voisine naturelles par pulverisation d'eau dam un courant de celle de I'ecoulement et que I'evaporation dans #air froid. I'ecoulement non sature ne remette pas en cause la concentration d'essai. Le madele mis au point par En France, les essais sont effectues par le CEPR de le CEPR montre : Saclay dam des installations (Cf Fig. 9) permettant de reproduire la vitesse de "01, la temperature, - L'importance du temps de transit entre grille et L'altitude, l'hygrom0trie. le diametre des gouttelettes maquette pour les essais proches de 0% en effet et la concentration d'eau. une distance grillefmaquette trop faible conduit a grande vitesse un nuage ne contenant qu'une partie des gouttelettes en surfusion.

- L'effet de I'hygrometrie sur la concentration pour les essais & base vitesse et a temperature proche de 0°C. la concentration peut etre reduite par des facteurs de I'ordre de 50 % par I'evaporation des gouttelettes entre grille et maquette. Dans ce cas la temperature des gouttelettes est inferieure a celle de I'eeoulement.

- L'hygrometrie insuffisante, done I'evaporation, n'a pas un effet determinant sur les diametres de gouttes.

Les rCsultats obtenus permettent de mettre en *vi- dence les parametres de rbglage impokants pour chaque type d'essai et d'apporter pour chaque point d'essai effectue les faeteun correctifs comes- pondant aux conditions d'essai reellement FiQW'a 9 : EANC DE GIWAGE DU CEPQ DE SACMY rencontrees.

Ces travaux effectues par le CEP a la fois sur le domaine d'utilisation des bancs d P.essam et I'analyse L'essai est effectue a hide d'un biti d'essai repro- des conditions d'essai permettent de disposer d'un duisant I'installation motrice (moteur, capotages, moyen d'essai en givrage artificiel bien adapt6 aux entree d'air et systeme de protection). Les condi- conditions particulieres des entrees d'air tions qui etaient utilisees jusqu'il y a quelques d'helicopt&re. annees pour eouvrir les exigences de certification d'helicoptsre ne volant pas en conditions givrantes Des recoupements effectues sur differentes entrees prevues, Ctaient : d'air entre le comportement observe en simulation au banc et en conditions naturelles ou sur d'autres Vitesse 50 a 130 kts moyens d'essai, demontrent une bonne represents- temperature -5 -30°C tivite du banc CEPE y compris a base vitesse a une temperature proche de 0°C. Ce moyen d'essai Avec I'evolution des conditions d'utilismtion des he- peut etre exeluslvement utili& en certification. licopteres, le CEPR a fait evoluer les plages d'essais possibles pour entrees d'air helicoptere. 35-1

Le fonctionnement prolong6 dans ces conditions 4.2 Movens d'essai la neiee (au-del& de 10 secondes) ne constitue pas une condition d'utiiisation normale de I'helicopt6re. 4.2.1 Essai en conditions naturelies : En effet il s'agit d'evoluer tres pres du sol en i'absence totale de visibilite, ce qui n'est possible sed Le moyen d'esai actuellement reconnu pour la que sur un terrain bien degage avee des reperes certification des helicopteres B la neige est I'essai au sol qui ne sont pas caches par ia neige. La en condition de neige naturelie. Les exigences de phase de 5 minutes prevue dans I'essai type re- certification necessitent d'obtenir ies conditions commande par la FAA permet de couvrir avec suivantes : des marges tres importantes les conditions normaies d'utilisation dans ces conditions cor- averse de neige importante (visibiiite reduite a respondant & I'atterrissage et au decollage sur 400 m sans brouiiiard) un terrain enneigh ; duree de ces conditions : 1 h B 1 h 30 Les conditions de voi B faibie vitesse ne sont pas neeessairement, malgrC la concentration plus - conditions critiques de temperature (en ghn4ral eievee, parmi ies pius severes pour ie fonction- autour de 0°C). nement de Yentree d'air. En effet le champ aero- Afin d'effectuer ces essais avec un niveau de sea- dynamique different, pour certains types rite satisfaisant. ces conditions meteorologiques #entree d'air. de celui existant en voi de croi- doivent etre reneontrees dam des zones 00 il est siere n'entraine pas une concentration de neige possible de voler & grande vitesse avec une visibi- dans ies memes zones des conduits. D'autre part lite reduite (ce qui exelut pratiquement les zones la faibie vitesse d'avancement ne favorise pas de haute montagne). I'aecumuiation de neige sur les memes zones de ia structure qu'en voi rapide et ne favorise pas La recherche de ces conditions obligatoirement en ie detaehement de ces eventuelles accumulations periode hivernaie peut nCcessiter plusieurs se- qui pourraient Ctre ingerees par les moteurs. maines avec des deplacements sur de grandes distances afin de s'adapter aux Cvolutions de la A ce jour, ies essais de certification en voi sous la meteorologic. Pour cela, ce type d'essai ne peut Ctre neige doivent etre effectues sur un hCiicopte5re en reserve qu'm parcours final de certification et ii est vol en conditions naturelies. L'experience montre donc necessaire pour les construfteurs de disposer que le spectre recommand6 par ia FAA est rm ob- de moyens de deveioppement permettant de jectif tres difiiciie & r6aliser. II est donc necessaire s'assurer de la validit6 des protections choisies. de developper des moyens compl6mmentaires de demonstration : ceux-ci peuvent comprendre des La condition de neige naturelle qui peut etre trou- essais au banc (voir paragraphe suivant) et des vee le pius aisement est ie val stationnaire dans ie essais specifiques aux appareils. Le voi prolong6 en nuage de neige soufflee par ie rotor de I'helicoptere stationnaire neige recirculante est un moyen eom- (neige recircuiante). En effet ii suffit de trouver piementaire ti-& efiicace car ii permet relativement une zone recouverte de neige frakhe et un essai est facilement de mettre en evidence les points possible avec une epaisseur tres faibie au-dessus de "sensibies" d'une entree d'air : des visualisations laqueiie I'helicoptere peut se depiacer afin d'Btre par endoscope sont me aide appreciable pour ce toujours au-dessus de la eouche de neige fniche (le type d'essai. La justification d'un temps prolong6 souffle du rotor ayant pour effet de "baiayer" la (superieur B 10 minutes) en vol en neige recircu- neige autour de I'appareii). lante apporte une grande confiance sur la qualife de la protection, mais celie-ci doit &tre confirmhe Des conditions favorabies pour effectuer des essais en voi d'avancement. Par eonti-e, ia non justifica- en neige recircuiante peuvent Stre trouvees aise- tion d'un temps prolong6 ne signifie pas ment pendant la saison hivernale meme apres Une necessairement que la protection de I'entree d'air averse de neige de faible importance. Toutefois n'est pas certifiable : il peut en effet exister dans I'obtention des conditions les pius critiques pour ce certain cas des effets de seuil avec des concentra- type d'essai correspond en general a une tempera- tions elevees, au-deia desquels des captations ture de Yair legerement positive, ia neige ne peut peuvent se developper dans I'ent- d'air. aiors etre soulevee par le rotor que si ceiie-ci est Nous considerom qu'un autre moyen complemen- restbe en surface sufiisamment froide pour &iter taire de demonstration est le voi en conditions i'agglomeration des flocons. Dans ce cas il est reeiles de neige, teiies qu'elies peuvent etre ren- necessaire que la chute de neige ait lieu & temp& eontrek lorsque I'kquipage recherche les rature negative et soit suivie d'une legere elevation "conditions FAA". II y a alors accumulation de temperature de rair ambiant : cette condition d'exp&ience en voi sous ia neige dans des condi- n'est rencontree que rarement au cows d'un hiver, tions de neige ties diverses, avee des durees meme dam une zone od ies chutes de neige sont d'exposition variables, suivies ou non de rechauf- frequentes. fage, des vitesses de vol adapt& & la visibiiite, ete... Le bon fonctionnement de ?installation mo- Les points d'essai en neige recircuiante, bien que ne trice pendant ces vois est un Clement important & permettant pas ia demonstration CGmpi6te neces- prendre en compte pour la justification de I'entree mire a la certification d'une nouveile entree d'air. d'air. pewent Stre integr6s dans le programme de certifi- 4.2.2 Essais artifieiels cation de I'helicoptPre. II convient toutefois de faire les remarques suivantes sur ce type d'essai: II est tres vite apparu indispensable de disposer de moyens d'essai artiiiciels pour Btudier ie compor- - La Concentration de neige pendant un essai en tement B la neige des entrees d'air, afin de reduire neige recirculante est tr6s eievee (de I'ordre de 3 ies caits de deveioppement, mais aussi pour per- a 4 fois celle des %verses de neige les pius mettre de faire des essais comparables entre eUX severes) (temperature ambiante. concentration, quaiit6 de neige) et donc de juger de I'influence de differents parametres. 35-8

Les essais peuvent etre effectues sur un heiicoptere 2) Neige utili& complet ou sur un bSti d'essai, eomprenant I'entree d'air et son environnement (y compris environne- La neige utiiisee qui est collect& avant I'essai peut ment thermique (Cf Fig, 10). avec un ventilateur Ctre de la neige naturelle fraiche ou de la neige assurant le debit d'air moteur. Cette derniPre soiu- artificielle. Cette derniere solution permet est tion de loin la plus soupie, et, avec quelques d'effectuer ies essais en chambre~ ~ ~~~ climatiaue et ~.~~~~ precautions. permet de garantir me bonne simula- donne ainsi ie maximum de soupiesse dans le choix tion. Les essais banc permettent de fair.? un grand des conditions d'essai. nombre d'observations (forme et position des eap- tations 6ventuelles), et de mesures (temperature de La neige artificielle est realisee i I'aide de canons paroi en particuiier). de neige utilises dans les stations de ski (canon B eau pulverisee et air comprime, Canon B eau puiv6- risee sur un ventilateur). La meilleure qualit4 de neige est obtenue pour les temperatures d'air infb- rieures a -10°C. La neige fabriquee est stock& dans la ehambre climatique pour Ctre utilisee dans ies heures qui mivent lorsque la temperature de I'essai est obtenue (en general les conditions critiques sont situees pres de 0°C). Cette methode est preferable I'utiiisation directe du canon a neige sur le materiel en essai ; en effet :

1)L'essai est effectue a la temperature designee (y compris BU voisinage de 0" et mCme temperature ieg6rement positive par rechauffage de la chambre climatique juste want I'essai)

FWiC l0 : ESSAI NCICC AU OANC 2) La concentration de neige est bien maitride,

La methodologie d'essai suivante a et6 mise au 3)11 n'y a pas de superposition de phenomenes point: parasites comme le givrage des grilles d'entree d'air par I'eau surfondue contenue dans le nuage 1) Methode d'ingestion praduit par ie canon a neige.

11 s'agit de pulveriser dans I'air aspire par le II est A noter que ia neige artifieieile ne forme pas moteur de la neige a la concentration voulue. de flocons et presente done des differences par La neige est pulverisee par brossage devant l'entree rapport a la neige naturelle tombante. Par exempie du moteur a I'entree du ventilateur projetant la la relation traineelmasse des particuies est diffe- neige dans la zone de i'entree d'air (Cf Fig. 11). La rente (vitesse de chute dam i'air 2 a 3 fois plus pulverisation de neige s'effectue par brossage importante), cet el6ment est donc a prendre en manuei, cette solution qui peut paraititre tres compte dans des dispositifs separateun i inertie. artisanale permet le maximum de regularit6 et s'adapte aisement a tous les types de neige. Elle a Des essais de comparaison ont 6th effectues P iso- donc et@ retenue pour tous les essais i base conditions d'essai avec de la neige artificielle et de vitesse. Dans le cas 00 I'on veut integgrer I'effet de la neige naturelie. Ces essais ont montre un com- la vitesse de I'appareil, il est necessaire de soufiler portement qualitatif identique (captations aux de ia neige a I'amont des capotages avec Un memes endroits) mais quantitativement legerement ventilateur, mais ii est quasiment impossible de different (captations plus s6veres avec de la neige garantir une concentration precise au droit de artificielle). Ceei peut s'expliquer par : I'entree d'air. 1)Des impacts plus nombreux sur les parois des conduits coudPs (deflexions mains importantes des particules).

2) Un accrochage plus facile des particules sur ies asperit& des parois.

3) Echange thermique plus faci:? entre cristaux et NEIGE parois favorisant la fusion par:ielle de I'eau et donc I'adhesion de la partieule.

D'autres essais ant et4 effechk afin de eomparer , les resultats obtenus au banc avec ceux de vol. On t _. retrouve globalement le mCme &art qu'entre essai en neige naturelle et en neige artificielle.

L'experience accumulee au cours des ann& re. centes nous amhe dorenavant a mettre au point ia i protection la neige de ]'entree d'air au banc, puis VENlltAlEUR de verifier en voi le bon fonctionnement de la protection. Cette demarche a et6 suivie sur 3 installations motriees differentes et aucun Figure 'I?: INGESilONARiIFlCiELLE DE NEIGE DANS comportement particuiier non decele au banc n'a UNE ENJREE LYAIR et6 note au corm des essais en VOI. 35-9

5) CONCLUSION Discussion La part des travaux de developpement d'une entree d'air liee B la protection contre la neige et le glvre a tres net- 1. A. Spirkl, MTU tement augment6 au cours des dernieres annees. Initia- Have you - in the case of partial accretion of ice on the inlet lement. ies etudes ont et6 menees de maniere. tres screen - observed pressure distortion at the position of the empirique, B partir dessais effectues directement sur helicopt&-. Ce moyen

Les travaux effectues par Aerospatiale dans les 4 der- 2. ?? nieres annees sur la protection des entrees d'air ont et6 Can you give me an indication about the pressure loss that men& sur 4 installations motrice differentes et repre- occurs at fully iced inlets at high mass flows? sentent environ 60 heures d'essai en souflerie givrante, 40 heures en chambre ciimatique neige, et 4 campagnes Author: d'essais en vol la neige. 11s ont permis d'optimiser le B It should not be more than 5% of the atmospheric pressure. concept de protection par grille qui est i notre Sens le plus performant, tant au niveau masse. perte de puis- But in most of the test points the loss remains below this sance associee. que pour la fiabilite de la protection. value. REPORT DOCUMENTATION PAGE 4. Security Classiticai

ISBN 92-835-0618-9

5. Originator Advisory Group for Aerospace Research and Development North Atlantic Treaty Organization 7 rue Ancelle. 92200 Neuillv sur Seine. France 6. Title LOW TEMPERATURE ENVIRONMENT OPERATIONS OF TURBOENGINES (DESIGN AND USERS PROBLEMS)

7. Presented at the Propulsion and Energetics Panel 76th Symposium held in Brussels, Belgium, 8th-12th October 1990.

8. Author(s)/Editor(s) 9. Date Various May 1991

0. AuthorWEditor’s Address 1 1. Pages Various

2. DistributionStatement This document is distributed in accordance with AGARD policies and regulations, which are outlined on the back covers of all AGARD publications. 3. Keywords/Descriptors

Turboengine design Low temperature operations Turboengine operations Icing tests Icing Ice tolerant design 1 Cold weather starting Ice protection Anti-icing systems Low temperature ignition

4. Abstract

The Confercnce Proceedings contain the 33 papers presented at the Propulsion and Energetics Panel 76th Symposium on “Low Temperature Environment Operations of Turboengines (Desig and User’s Problems)”, which was held 8tl-12th October 1990 in Brussels, Belgium.

The Symposium was composed of the following sessions: Cold Weather Operational Experience and Requirements (6); System Design Considerations (14); Fuel Effects and Lubricants Behaviour (5); Icing Conditions and Testing (8). Questions and answers of the discussions folloa each paper in the Proceedings.

Both the engine designers and the users made ample use of this Symposium to exchange their views of the state of their art and their experience within and between their respective communities. The papers presented and the discussions represent a significant contribution to improved cold weather tolerant and anti-icing design and to safer aircraft operation in a low temperature environment. A case was also made for advanced and larger dedicated testing facilities, and some specialized software developments were indicated. ~ 4GARD Conference Proceedings 480 AGARD-CP-480 AGARD Conference Proceedings 480 AGARD-CP-480 4dvisorv Grouu for Aerosuace Reserch and Advisory Group for Aerospace Reserch and Development, NATO Development, NATO LOW TEMPERATURE ENVIRONMENT Turboengine design LOW TEMPERATURE ENVIRONMENT Turboengine design OPERATIONS OF TURBOENGINES Turboengine operations OPERATIONS OF TURBOENGINES Turboengine operations ‘DESIGN AND USER’S PROBLEMS Icing (DESIGN AND USERS PROBLEMS Icing Cold weather starting buhlished May 1991 Cold weather starting Published May 199 1 ~ 386 pages Anti-icing systems 386 pages Anti-icing systems Low temperature Low temperature operations operations The Conference Proceedings contain the 33 paper! The Conference Proceedings contain the 33 papers Icing tests Icing tests presented at the Propulsion and Energetics Panel 76tl presented at the Propulsion and Energetics Panel 76th Ice tolerant design Ice tolerant design Symposium on ‘2ow Temperature Environmen Symposium on “Low Temperature Environment Ice protection Ice protection Operations of Turboengines (Design and User’! Operations of Turboengines (Design and User’s Low temperature ignition Low temperamre ignition Problems)”, which was held 8th-12th October 1990 in Problems)”, which was held 8th-12th October 1990 ir Brussels, Belgium. Brussels, Belgium.

P.T.O. P.T.0

AGARD Conference Proceedings 480 AGARD-CP-480 AGARD Conference Proceedings 480 AGARD-CP-480 Advisory Group for Aerospace Reserch and Advisory Group for Aerospace Reserch and Development, NATO Development, NATO Turboengine design ~LOW TEMPERATURE ENVIRONMENT Turboengine design LOW TEMPERATURE ENVIRONMENT OPERATIONS OF TURBOENGINES Turboengine operations OPERATIONS OF TURBOENGINES Turboengine operations (DESIGN AND USERS PROBLEMS Icing (DESIGN AND USERS PROBLEMS Icing PublishedMay 1991 Cold weather starting Published May 1991 Cold weather starting 386 pages Anti-icing systems 386 pages Anti-icing systems Low temperature Low temperature aperations operations The Conference Proceedings contain the 33 paper The Conference Proceedings contain the 33 papers Icing tests Icing tests presented at the Propulsion and Energetics Panel 76tl presented at the Propulsion and Energetics Panel 76th Ice tolerant design Ice tolerant design Symposium on “Low Temperature Environment Symposium on “Low Temperature Environmen Ice protection Ice protection Operations of Turhoengines (Design and User’ Operations of Turboengines (Design and User‘s Low temperature ignition Low temperature ignition Problems)”, which was held 8th-IZth October 1990 in Problems)”, which was held 8th-12th October 1990 ii Brussels, Belgium. Bmssels, Belgium.

P.T.O. P.T.C The Symposium was composed of the following sessions: Cold Weather Operation; The Symposium was composed of the following sessions: Cold Weather Operational Experience and Requirements (6); System Design Considerations (14); Fuel Effects an Experience and Requirements (6); System Design Considerations (14); Fuel Effects and Lubricants Behaviour (5); Icing Conditions and Testing (8). Questions and answers of tt Lubricants Behaviour (5); Icing Conditions and Testing (8). Questions and answers of the jiscussions follow each paper in the Proceedings. discussions follow each paper in the Proceedings.

Both the engine designers and the users made ample use of this Symposium to exchang Both the engine designers and the users made ample use of this Symposium to exchange :heir views of the state of their art and their experience within and between their respecti! their views of the state of their art and their experience within and between their respective :ommunities. The papers presented and the discussions represent a significai communities. The papers presented and the discussions represent a significant zontrihution to improved cold weather tolerant and anti-icing design and to safer aircra contribution to improved cold weather tolerant and anti-icing design and to safer aircraft aperation in a low temperature environment. A case was also made for advanced and largt operation in a low temperature environment. A case was also madc for advanced and iarger jcdicated testing facilities, and some specialized software developments were indicated, dedicated testing facilities, and some specialized software developments were indicated.

ISBN 92-835-0618-9 ISBN 92-835-0618-9

The Symposium was composed of the following sessions: Cold Weather Operation The Symposium was composed of the following sessions: Cold Weather Operational Experience and Requirements (6);System Design Considerations (14); Fuel Effects ar Experience and Requirements (6);System Design Considerations (14); Fuel Effects and Lubricants Behaviour (5);Icing Conditions and Testing (8). Questions and answers of tl Lubricants Behaviour (5); Icing Conditions and Testing (8). Questions and answers of the discussions follow each paper in the Proceedings. discussions follow each paper in the Proceedings.

Bolh the engine designers and the users made ample use of this Symposium to exchanl Both the engine designers and the users made amplc use of this Symposium to exchange their vicws of the state of their art and their experience within and between their respecti thcir views of the state of their art and their cxperience within and between their respective communities. The papers presented and thc discussions represent a significa communities. The papers presented and the discussions represent a significant Contribution to improved cold wcather tolerant and anti-icing design and to safer aircri contribution to improved cold weather tolerant and anti-icing dcsign and to safer aircrafc operation in a low tcmperature environmcnt. A case was also made for advanced and larg operation in a low temperaturcenvironment. Acasewasalso madefor advancedandlargcr dedicatcd testing facilities, and some specializcd software developments were indicated. dedicated testing facilities, and some spccialized software developmcnts wcre indicated.

ISBN 92-835-0618-9 ISBN Y2-835.06 18-9 7 RUE ANCELLE .92200 NEUILLY-SUR-SEINE DIFFUSION DES PUBLICATIONS FRANCE AGARD NON CLASSIFIEES Telephone (1)47.38.57.00 . Telex 610 176 Telecaoie I I \47.38.57.99 L'AGARD nc ddtient pas de stacks de ses publications, dans un but de distribution genirale B I'adresse ci-dessus. La diffusion initialc des publications de I'AGARD est effectuee auprks des pays mcmbres de cettc organisation par I'intcrmediaire des Centres Nationaux de Distribution suivants.A I'exception des Etats-Unis, ces centres disposent parfois d'exemplaircs additionnels; dans lcs cascontraire, on peut se procurer ces cxcmplaircs sous forme de microfiches ou de microcopies auprb des Agcnces de Vente dont la liste suite. CENTRES DE DIFFUSION NATIONAUX ALLEMAGNE ISLANDE Fachinformationszentrum, Director of Aviation Karlsruhe c/o Flugrad D-75 14 Eggenstein-Leopoldshafcn 2 Reykjavik BELGIQUE ITALIE Coordonnatcur AGARD-VSL Acronautica Militaire Etat-Major de la Force Aericnne Ufficio del Dclegato Nazionale all'AGARD Quarticr Reine Elisabeth 3 PiazzaJc Adenauer Rue dEvere, 1140 Bruxelles 00144 RomaEUR CANADA LUXEMBOURG Dirccteur du Service des Renseignemcnts Scientifiques Voir Belgique Ministire de la Defense Nationale NORVFGF. . ___ Ottawa, OntarioKlAOK2 Norwegian Defence Research Establishment DANEMARK Attn: Bibliotckct Danish Defence Research Board P.O. Box 25 Vcd Idractsparkcn 4 N-2007 Kjeller 2100 Copenhagen0 ESPAGNE INTA AGARD Publications) Pintor aosales 34 28008 Madrid PORTUGAL ETATS-UNIS Portuguese National Coordinator to AGARD National Aeronautics and Space Administration Gabinete de Estudos e Programas Langley Research Center CLAFA MIS 180 Base de Alfragide Hampton, Virginia 23665 Alfragide 2700 Amadora FRANCE O.N.E.R.A. (Direction) ROYAUME UNI 29, Avenue de la Division Leclerc Defence Research Information Centre 92320, Chitillon sous Bagncux Kentigern House 65 Brown Street GRECE Glasgow G2 XEX Hellenic Air Force Air War College TURQUIE Scientific and Technical Library Mill! Savunma Ba~kanligi(MSB) Dckclia Air Force Base ARGE Dairc BaSkanhii (ARGE) Dckclia,AthensTGA 1010 Ankara LE ('EUTRE NhTlOUAL DE DISTRIBUTIOU DES ETATS-IJNIS (\ASA) UL I>L I llU'l PA> l>L S IOCKS DES PI.'BLIC:\lIONS ,\Ci.\RD ET LEI DEhlANDES D'EYE>lPLAIK€S 1>01\ tNI I: 1 KI. .Al)Kl..S$l..l 3 1)1111 c'l'l~.\llN I AL. SER\'I('E NATIONAL TECIIKIQIIC I>C L'IKrOR\l.ATION [K'l IS, I)ON 1 I.A1)111 SSI: SL I I AGENCES DE VENTE National Technical Information Service ESAhformation Retrieval Service The British Library (NTIS) European Space Agency Document Supply Division 5285 Port Royal Road 10, rue Mario Nikis Boston Spa, Wetherby Springfield, Virginia 22161 75015Paris West Yorkshire LS23 7BQ Etats-Unis France Royaume Uni Les demandes demicrofiches ou dephotocopicsdcdocumcnts AGARD(y compris lesdemandcsfaitesaupresdu NT1S)doivcntcomportcr la denomination AGARD, ainsi que le numero de strie de I'AGARD (par exemple AGARD-AG-315). Des informations analogues, telles queletitreet ladatcdepublicationsont souhaitables.Vcuillcrnotcrqu il yalieudcspCcificr AGARD-R-nnnct AGARD-AR-nnnlorsdela commandede rapports AGARDct des rapportsconsultatifs AGARDrespectivement. Des references bibliographiqucscompletesainsiquc des resumes des publications AGARD figurent dans les journaux suivants: Scientifique and Technical Aerospace Rcpons (STAR) Government Reports Announcements and Index (GRAM) publie par la NASA Scientific and Technical publie par le National Technical Information Service Information Division Springfield NASA Headquarters NlT) Virginia 2216 I Washington D.C. 205 !I 6 Etats-Unis Etats-Unis (accessible kgalemcnt en mode interactif dam la base de donnecs bibliographiques en lignc du NTIS, et sur CD-ROM)

Imprimk par Speciulised Printing Services Limited 40 Chigwell Lane, Loughton, Essex IGIO 3Tz NATO ~@~OTAN 7 RUE ANCELLE . 92200 NEUILLY-SUR-SEINE DISTRIBUTION OF UNCLASSIFIED FRANCE AGARD PUBLICATIONS

Telephone (l)47.38.57.00 ' Telex 610 176 Tclefax (1)47.38.57.99 AGARD does NOT hold stocks of AGARD publications at the above address for general distribution. Initial distribution of AGARD publications is made to AGARD Member Nations through the following National Distribution Centres.Further copies are sometimes available from these Centres (except in the United LStates), but if not may be purchased in Microfiche or Photocopy form from the Sales Agencies listed below. NATIONAL DISTRIBUTION CENTRES BELGIUM LUXEMBOURG Coordonnateur AGARD - VSL See Belgium Etat-Major de la Force Airienne Quartier Reine Elisabeth NETHERLANDS Rue &Evere, 1140 Bnrxelles Netherlands Delegation to AGARD National Aerospace Laboratory, NLR CANADA Kluyvenveg 1 Director Scientific Information Services 2629 HS Delfi Dept of National Defence Ottawa,Ontario KIA OK2 NORWAY Norwegian Defence Research Establishment DENMARK Attn: Biblioteket Danish Defence Research Board P.O. BOX 25 Ved ldraetsparken 4 N-2007 Kjellei 2100 Copenhagen 0 PORTUGAL FRANCE Portuguese National Coordinator to AGARD O.N.E.R.A. (Direction) Gabinete de Estudos e Programas CLAFA 29 Avenue de la Division Leclerc Base de Alfragide 92320 Chitillon Alfragide GERMANY 2700 Amadora Fachinformationszentrum SPAIN Karlsruhe INTA AGARD Publications) D-7514 Eggenstein-Leopoldshafen 2 Pintor kosales 34 GREECE 28008 Madrid Hellenic Air Force TURKEY Air War College Milli Savunma BaSkanltgi (MSB Scientific and Technical Library ARGE Daire BaSkanhB (ARC2 ) Dekelia Air Force Base Ankara Dekelia, AthensTGA 1010 UNITED KINGDOM ICELAND Defence Research Information Centre Director of Aviation Kentigern House c/o Flugrad 65 Brown Street Reykjavik Glasgow G2 8EX ITALY UNITED STATES Aeronautica Militare National Aeronautics and Space Administration (NASA) Ufficio del Delegato Nazionale all'AGARD Langley Research Center 3 Piazzale Adenauer M/S 180 00144 Roma/EUR Hampton, Virginia 23665 'I HI: I 'NII lil>ST.A'l't.3 NATIONl\I. I>ISTRIHl~IION t'1.N I RE (NASA DOES NOT HOI .U SI'OCKS OF .AGARD l'UI%I K'ATIONS AND ,\l'PLlCA'l IOSS FOR COIJldSSHOIII.D BE \IAVE I)IKECT'10 TIE SXrIONAI.'I LCllNlCAL INFOHMVIATION SKKVICE (N I 15) ATTHI: ADDRI:SS BELOW SALES AGENCIES National Technical ESAhformation Retrieval Service rhc Briti>h LIlxar! Information Service (NTIS) European Space Agency D.,cumr.ni Suppl) Ccntrr. 5285 PoR Royal Road 10, rue Mario Nikis Bosun Spa. Htthcrh) Springfield,Virginia 22161 75015Pans \Ve\t York\hirc IX37110 Unitt:d States France United KingJom Requests for microfiches or photocopies of AGARD documents (including requests to NTIS) should include the word'AGARD'and the AGARD serial number (for example AGARD-AG-3 15 Collateral information such as title and publication date is desirable. Note that AGARV Reports and Advisory Reports should be specifked asAGARD-R-nnn and AGARD-AR-nnn, respectively. Full bibliographical references and abstracts of AGARD publications are given in the following journals:

Printed by 5pecralfsed rrrntfng3eTvfces umircu 40 ChigweN Lane, Loughton, Essex IG103TZ