The Pennsylvania State University
The Graduate School
College of Engineering
DESIGN AND TESTING OF AN IMPROVED CENTRIFUGALLY
POWERED PNEUMATIC DE-ICING SYTEM FOR HELICOPTER
ROTOR BLADES
A Thesis in
Aerospace Engineering
by
Matthew D. Drury
© 2016 Matthew D. Drury
Submitted in Partial Fulfillment
of the Requirements
for the Degree of
Master of Science
May 2016 The thesis of Matthew D. Drury was reviewed and approved∗ by the following:
Jose L. Palacios
Assistant Professor of Aerospace Engineering
Thesis Advisor
Sven Schmitz
Assistant Professor of Aerospace Engineering
George A. Lesieutre
Professor of Aerospace Engineering
Head of the Department of Aerospace Engineering
∗Signatures are on file in the Graduate School.
ii Abstract
A full-scale centrifugally powered pneumatic de-icing system for helicopter blades was developed. The designed system makes use of the pressure differential created within a spinning rotor blade to deform a 0.03 in. thick metallic leading edge cap, producing the transverse shear stresses necessary to delaminate accreted ice. Two prototype designs were fabricated and parametrically tested. Both designs consist of a stainless steel leading edge cap tied to the blade surface via flexible stainless steel ribs. The leading edge section is sealed to the blade surface such that it can be pressurized, inflating and deflating as a microvalve is cycled between high and low pressure lines that run along the blade span. Both designs were fabricated and installed on a truncated 12 in. span K-MAX blade section. Rotor ice testing of the prototypes was conducted at The Pennsylvania State University Adverse Environment
Rotor Test Stand. Input pressures representative of those produced by centrifugal pumping along a full-scale 24 ft. radius blade rotating at 270 RPM were provided to the prototypes via a pneumatic slip ring. Parametric testing of the two configurations demonstrated the superiority of one of the designs, which was capable of removing accreted ice thicknesses as small as 0.078 in. at -14◦C and 0.112 in. at -5◦C. The selected configuration was reproduced and installed on the outboard 8 ft. section of
iii a full-scale 24 ft. K-MAX blade. Full-scale icing testing was conducted at Kaman’s outdoor whirl tower during the month of February. A portable icing cloud generator capable of providing 40 minutes of continuous icing through 12 NASA Standard icing nozzles was developed to provide representative icing conditions. A second cloud generator providing uncontrolled water droplet sizes was also implemented due to power availability limitations to operate the portable icing cloud generation system. The full-scale system was tested at static air temperatures within the Federal
Aviation Regulation (FAR) Part 25/29 icing envelope and was able to delaminate ice thicknesses as small as 0.08 in. at 270 RPM. The full-scale de-icing system as installed on the K-MAX blade draws a peak instantaneous power from the rotor of 4.15 HP during system deflation and a maximum of 0.02 HP during system pressurization, or a maximum power density of 4 W/in2 compared to the 25 W/in2 required by electrothermal de-icing systems. Weight of the centrifugally powered pneumatic de-icing system is comparable to that of existent erosion caps installed on in-service helicopters, since the de-icing system requires only slight modification to the erosion caps and the addition of lightweight pneumatic tubes to the rotor blade. A method to predict de-icing system delamination performance using cohesive zone failure methods in the Abaqus finite element software was investigated. Using geometry representative of the physical prototypes tested, the pneumatic pressure required to delaminate representative ice shapes as predicted by the model was compared to experimental results. It was found that the de-icing system response predicted by the model consistently over predicted the pneumatic pressure required for ice shape delamination. At -14◦C, the delamination model was found to be
iv more accurate, producing a 46% error compared to experimental data. Warmer temperatures showed larger errors in model results compared to experimental data.
Supported by observations made during prototype ice testing, the exclusion of brittle ice fracture from the two dimensional, plane-strain model is believed to be the root cause of the discrepancy between predicted and experimental results.
v Table of Contents
List of Figures ix
List of Tables xiv
Acknowledgments xv
Chapter 1 Introduction 1 1.1 Icing Overview ...... 1 1.1.1 Aircraft Icing Research History ...... 1 1.1.2 Aircraft Icing ...... 3 1.1.3 Implications of Icing to Rotorcraft Operations ...... 7 1.2 Ice Protection Systems ...... 10 1.2.1 Electro-Thermal ...... 10 1.2.2 Ultrasonic ...... 12 1.2.3 Low Adhesion Strength Coatings ...... 13 1.2.4 Fluid Anti-Icing ...... 15 1.2.5 Electro-Impulse ...... 17 1.2.6 Electrovibratory ...... 19 1.2.7 Pneumatic ...... 21 1.3 Methods of Ice Testing ...... 25 1.3.1 In Fight Testing ...... 25 1.3.2 Icing Wind Tunnels ...... 28 1.3.3 Hover Test Stands ...... 29 1.4 Review of Prior Centrifugally Powered Pneumatic De-Icing Systems . 32 1.5 Objectives ...... 36 1.6 Thesis Overview ...... 38 1.6.1 Chapter 2: Design and Analysis of a Pneumatic De-Icing System 38 1.6.2 Chapter 3: Prototype Rotor Ice Testing and Finite Element Model Prediction Validation ...... 38
vi 1.6.3 Chapter 4: Portable Icing Cloud Generator Design, Construc- tion and Testing ...... 39 1.6.4 Chapter 5: Full Scale Pneumatic De-Icing System Testing . . 39 1.6.5 Chapter 6: Conclusions and Recommendations for Future Work 39
Chapter 2 Design and Analysis of a Pneumatic De-Icing System 40 2.1 Introduction ...... 40 2.2 Prototype I Design ...... 41 2.3 Prototype II Design ...... 44 2.4 Finite Element Modeling of De-Icing Systems ...... 47 2.4.1 Blade ...... 47 2.4.2 Leading Edge Cap ...... 48 2.4.3 Elastomer ...... 49 2.4.4 Ice ...... 50 2.4.5 Mesh ...... 51 2.5 Prediction of De-Icing System Delamination Performance ...... 53 2.5.1 Review of Cohesive Failure Theory ...... 54 2.5.2 Application of Cohesive Zone Methods to the Finite Element Method ...... 57 2.5.3 De-icing Performance Modeling Process Overview ...... 60 2.5.3.1 LEWICE ...... 61 2.5.4 Shedding Model ...... 62 2.5.4.1 Abaqus Implementation and Results ...... 65 2.6 Effects of Aerodynamic Pressures ...... 69 2.6.1 XFOIL ...... 69 2.6.2 Mapping Surface Pressures in Abaqus ...... 69 2.6.3 Results ...... 72 2.7 Mitigation of Leading Edge Cap Deformation Due to Aerodynamic Pressures ...... 73
Chapter 3 Prototype Pneumatic De-Icing System Testing and Finite Ele- ment Model Prediction Validation 82 3.1 Small-Scale Hover Icing Testing ...... 82 3.1.1 Description of the Test Facility ...... 82 3.1.2 Description of Test Blades and Prototype De-Icing Systems . . 84 3.1.3 Test Method ...... 86 3.1.4 Results ...... 88
vii 3.2 Comparison of Pneumatic De-icing system Model and Experimental Test Results ...... 95
Chapter 4 Portable Icing Cloud Generator Design, Construction and Testing102 4.1 Background and Motivation ...... 102 4.2 Icing Cloud Generator Design ...... 103 4.2.1 Air System ...... 105 4.2.2 Water System ...... 106 4.2.3 Icing Nozzles ...... 108 4.2.4 Icing Cloud Controller ...... 112 4.3 Icing Cloud Generator Testing ...... 114
Chapter 5 Full Scale Pneumatic De-Icing System Testing 118 5.1 Full-Scale Blade Modifications ...... 118 5.2 Full-Scale Hover Icing Tests ...... 124 5.2.1 Test Method ...... 125 5.2.2 Test Results ...... 130 5.3 System Power Consumption ...... 133
Chapter 6 Conclusions and Recommendations for Future Work 140 6.1 Conclusions ...... 140 6.2 Future Work ...... 143 6.2.1 Pneumatic De-icing System Design ...... 143 6.2.2 Pneumatic De-icing System Modeling ...... 145
References 147
viii List of Figures
1.1 Examples of glaze (a), mixed (b), and rime (c) ice [1]...... 5 1.2 Dependence of icing conditions on ambient temperature, LWC and MVD [1]...... 6 1.3 Typical ice accretion distribution along a rotor blade [2]...... 6 1.4 FAR Part 25/29 Appendix C Icing Envelope [2], [3]...... 7 1.5 Degradation of airfoil performance due to ice accretion [4]...... 9 1.6 Distribution of terminating occurrences from in-flight icing encoun- ters. [5]...... 10 1.7 Schematic of electrothermal de-icing system installed on a rotor blade [6]. 11 1.8 Schematic of unltrasonic de-icing actuator concept installed on rotor blade erosion cap [7]...... 13 1.9 Ice adhesion strength of various commercial ice phobic coatings [8]. . 14 1.10 Fluid anti-icing system diagram [9]...... 16 1.11 Cross sectional view of fluid anti-icing system installed on a rotor blade [9]...... 16 1.12 Electro impulse coil installed on the leading edge of a wing [9]. . . . . 18 1.13 Electro-impulse de-icing system diagram [9]...... 18 1.14 Electrovibratory de-icing system actuator configurations [9]...... 20 1.15 Drawings of the first pneumatic de-icing boot as developed and tested by B.F. Goodrich [10]...... 22 1.16 Operation of the PIIP [11]...... 23 1.17 Schematic of pneumatic deicing system as installed on the UH-1 rotor [9]. 24 1.18 Schematic of pneumatic deicing boot as installed on the UH-1 rotor [12]. 25 1.19 Schematic of the Canadian National Research Council (NRC) Low- Temperature Laboratory in Ottawa, Ontario [13]...... 26 1.20 Schematic of the U.S. Army Helicopter Icing Spray System (HISS) [13]. 27 1.21 Photograph of the Lockheed F-117 in the McKinley Climatic Labora- tory [14]...... 28 1.22 Typical set-up of an icing wind tunnel [13]...... 29 1.23 AERTS facility layout [4]...... 30
ix 1.24 Photograph of OH-50 rotor head in the AERTS facility [4]...... 31 1.25 (a) Simple schematic of centrifugal pumping concept, (b) experimen- tally measured pressures created by centrifugal pumping on a 24 ft. blade at 280 RPM [15]...... 33 1.26 Photograph of the second pneumatic de-icing system prototype [16]. . 35 1.27 Measured lift and drag coefficients of the pneumatic de-icing system compared to an iced airfoil [17]...... 35 1.28 Initial centrifugally powered pneumatic de-icing system developed by Penn State and Invercon in the undeformed and deformed states [16]. 36
2.1 Schematic of the first pneumatic de-icing system prototype in the un-pressurized (top) and pressurized (bottom) configurations. . . . . 42 2.2 Schematic of lap-joint shear test set-up...... 43 2.3 Sample stress/strain data from lap joint shear adhesive tests...... 44 2.4 Schematic of the second pneumatic de-icing system prototype in the un-inflated (top) and inflated (bottom) configurations...... 45 2.5 Schematic of Prototype II alternating bond pattern for flexible metallic ribs...... 46 2.6 Uni-axial test data for EPDM elastomer from Ref. [18]...... 50 2.7 Representative ice shape attached to Prototype II de-icing system. . . 51 2.8 Mesh convergence of leading edge cap for Prototype I...... 53 2.9 Mesh convergence of leading edge cap for Prototype II...... 53 2.10 Example of a Mode I delamination event [19]...... 55 2.11 Dependence of cohesive zone method results on mesh size from Ref [19]. 59 2.12 Process followed to predict de-icing system performance for a given pneumatic pressure...... 61 2.13 Forces acting on ice shape along a blade section [2]...... 63 2.14 Maximum percent bond-line remaining allowable for total ice shape shedding...... 64 2.15 Path defined in Abaqus to gather cohesive surface separation data. . . 66 2.16 Sample delamination results from Abaqus model...... 67 2.17 Effect of increasing pneumatic pressure on ice bond-line damage prop- agation...... 68 2.18 Predicted delamination performance of Prototype II de-icing system for three ice shape thicknesses...... 68 ft 2.19 Pressure distributions for a NACA 23012 airfoil at 700 s ...... 70 2.20 Example of pressures applied to the Abqaus model for the aerodynamic pressure study. The pictured pressure distribution is for α = 10◦. . . 71 2.21 Leading edge cap deformation profile caused by aerodynamic pressures at α = 10◦. Note that units are in meters...... 72
x 2.22 Magnitude of maximum leading edge cap deformation due to aerody- namic pressures...... 73 2.23 Effects of aerodynamic pressures on leading edge cap deflection. . . . 75 2.24 Comparison of pneumatic de-icing system performance for increased cap thickness...... 76 2.25 Uniaxial test data used to model elastomer in Abaqus finite element model (original data adopted from Ref. [18])...... 77 2.26 Effect of increase elastomer stiffness on leading edge cap deflection due to aerodynamic pressures...... 78 2.27 Leading edge cap deformation due to aerodynamic pressure for several elastomer stiffnesses and materials...... 79 2.28 Comparison of the effect of various parameters on the ice delamination performance of the de-icing system for Case 1 ice...... 80
3.1 Photograph of the AERTS facility...... 83 3.2 Ceiling view of the AERTS facility...... 84 3.3 Photograph of Prototype I installed on the test blade with pneumatic suction (top) and pressure (bottom) applied...... 85 3.4 Pressures produced by centrifugal pumping on a full-scale rotor blade at 280 RPM...... 86 3.5 Rotor ice testing condition location in FAR intermittent icing envelope. 87 3.6 AERTS rotor ice testing procedure...... 89 3.7 Prototype I before and after de-icing system actuation...... 90 3.8 Prototype II before and after de-icing system actuation ...... 92 3.9 Effects of temperature on Prototype II de-icing capabilities...... 93 3.10 Comparison of Prototype I and II minimum ice thickness required for delamination at -14◦C...... 94 3.11 Comparison of Abaqus predicted and prototype measured de-icing system deformation with 4 psi pneumatic pressure input...... 95 3.12 Abaqus predicted delamination for Prototype II for ice shapes produced at -14◦C and -5◦C during AERTS rotor ice testing...... 97 3.13 Maximum percent bond-line remaining allowable for total ice shape shedding...... 98 3.14 Comparison of experimental and Abaqus predicted minimum ice thick- ness required for shedding for ice shapes produced during AERTS rotor ice testing...... 99 3.15 Ice left behind after de-icing system pressurization...... 100 3.16 Cracks formed in accreted ice by de-icing system pressurization. . . . 100
xi 4.1 Schematic of the portable icing cloud generator. Note that red lines represent air and blue represents water...... 104 4.2 Photograph of the icing cloud system air pressure control hardware. . 105 4.3 Schematic of air and water distribution manifolds...... 106 4.4 Photograph of the icing cloud generator water sub-system...... 108 4.5 NASA Standard and Mod-1 icing nozzles. Standard nozzles were used in the portable cloud generator. Note that flow is from left to right [20]. 109 4.6 Cross sectional view of nozzle housing...... 109 4.7 NASA Standard icing nozzle calibration curve [20]...... 110 4.8 Photograph showing location of the icing nozzles mounted on the fan. 111 4.9 Portable icing cloud system control panel...... 112 4.10 Initial testing of the portable icing cloud conducted in the AERTS facility...... 115 4.11 Portable icing cloud test envelope for successful continuous icing cloud. 116 4.12 Schematic and photograph showing portable icing cloud in-use during full-scale rotor icing tests...... 117
5.1 Schematic showing location of de-icing system as installed on a full scale rotor blade...... 119 5.2 Photograph of modified de-icing blade mounted on the Kaman whirl stand for full-scale icing tests...... 120 5.3 Schematic of Prototype II de-icing system...... 121 5.4 Close up view of de-icing system installed on a full-scale K-MAX blade. 122 5.5 Schematic of pneumatic pressure lines as installed on the full-scale K-MAX blade...... 123 5.6 Installation of pnematic pressure lines on full-scale blade...... 124 5.7 Photograph of the Kaman Whirl Tower...... 125 5.8 Final icing cloud configuration produced large, uncontrolled size water droplets...... 127 5.9 Full-scale rotor ice testing procedure...... 128 5.10 Sample hall sensor output voltage showing de-icing system inflation and deflation during testing...... 129 5.11 Comparison of protected (right) and unprotected (left) blades after de-icing system test...... 130 5.12 Example of ice shape measured on unprotected blade...... 131 5.13 Severe ice shapes produced by large water droplets during de-icing system testing...... 132 5.14 Ice patch left behind after de-icing system test...... 133 5.15 Equalization of pressure differential between high and low pressure lines during de-icing system de-pressurization...... 135
xii 5.16 Rate of air depletion from de-icing system leading edge cap during system de-pressurization...... 136 5.17 Power draw from rotor induced by Coriolis Forces during system de-pressurization...... 138
6.1 Experimentally measured pressures produced by centrifugal pumping along a 24 ft. radius rotor blade at 270 RPM...... 144 6.2 Schematic of pneumatic de-icing system configuration for inboard and outboard blade regions...... 145
xiii List of Tables
2.1 Material properties of metals used in de-icing systems...... 49 2.2 Material properties of ice...... 51 2.3 Cohesive interface properties of ice to leading edge cap as measured in Ref. [16] ...... 59 2.4 Mesh element size used for calibration in Ref. [16] ...... 65 2.5 Ice shape thicknesses used for parametric study...... 76
3.1 Prototype I rotor ice testing summary...... 89 3.2 Prototype II rotor ice testing summary...... 91 3.3 Thickness of ice shapes from Prototype 2 rotor ice testing at -14◦C and -5◦C used for model validation...... 96
5.1 Sample single-shot de-icing test results...... 131
xiv Acknowledgments
First and foremost I would like to thank my advisor, Dr. Jose Palacios, for giving me the opportunity to return to Penn State and earn my Master of Science degree.
Without his support, guidance, and assistance, no matter the time of day or night, this research would not have been a success. I would also like to thank Dr. Joseph
Szefi of Invercon for his support throughout my research. His expertise of pneumatic systems and actuation were invaluable in the design and fabrication of the de-icing systems. Without his support, this research would not have been a success. I would also like to thank Dr. Sven Schmitz and Dr. George Lesieutre for taking time to review this thesis.
I would like to express my gratitude to my fellow graduate students Miguel Alvarez,
Belen Veras-Alba, Ahmad Haidar, Yiqiang Han, Sihong Yan, and Edward Rocco for their support and recommendations throughout my research. In addition to those mentioned, I would like to sincerely thank my fellow graduate students, Jared Soltis and Talyor Knuth, for their support during the long hours of full-scale testing. Their dedication along with that of Dr. Palacios and Dr. Szefi, was essential to the success of full-scale testing.
Finally, I would like to thank my family for their unwavering support and for
xv teaching me the importance of hard work. My grandfather, Dr. George Lenyo, gave me the gift of flight, which inspired a lifelong passion for aviation that has driven my career path. For this, I am deeply grateful. Lastly, I would like to thank my finance,
Brittany, for her support and for keeping me motivated throughout my graduate studies. This research was funded by NASA LEARN Fund Number NNX14AF54A.
xvi Chapter 1 | Introduction
1.1 Icing Overview
1.1.1 Aircraft Icing Research History
The dangers of in-flight icing conditions to aviation had not been realized until the mid-
1920’s when the United States Air Mail Service began its daily service between New
York and Chicago. This route often times required flight in instrument meteorological conditions (IMC) in which icing conditions were regularly encountered [21]. The mountainous terrain of the airway required pilots to climb to higher altitudes where icing clouds were usually found. Typically, during the time spanning from fall until early spring, the frequency of flying by the Air Mail Service was greatly reduced due to icing. In fact, as one aviator noted of the route during winter, "the greatest of all our problems is ice [22]."
Before any laboratory icing experiments were attempted, flights into icing condi- tions were carried out to gather data concerning the effect of various surface coatings such as oil or grease on ice accretion [23] as well as identifying types of ice (rime and glaze) and the conditions in which they formed [23]. Subsequently, the National
1 Advisory Committee for Aeronautics (NACA) constructed the first icing research wind tunnel in 1928 at Langley Research Center. The wind tunnel was a small one (6 inches in diameter) constructed of an insulated shell and was cooled by a commercial refrigeration unit. Four commercial spray nozzles were used to produce a crude icing cloud by controlling air and water pressure [21]. Tests in this tunnel studied the effect of surface coatings on ice adhesion. Some of the peculiar coatings that were investigated included oil, grease, Vaseline, paraffin wax, molasses, corn syrup, honey, commercial paint, goose grease, and White Karo syrup [24]. Following these tests, recommendations from both pilots and engineers concerning ice protection was to simply avoid flight into icing conditions stating, "there appears little likelihood of successful prevention of the formation of ice on the airplane in flight by the applica- tion of any preventative means [23]." Despite these conclusions, icing research efforts continued and shortly thereafter yielded the pneumatic de-icing boot [10] as well as the use of engine bleed air to melt accreted ice [25].
In 1942 construction of the NACA’s Icing Research Tunnel (IRT) began in
Cleveland, Ohio. The wind tunnel was to make use of the state-of-the-art refrigeration system of the Altitude Wind Tunnel being constructed to test engines at high altitudes and high speeds. The IRT was designed to have a 6 ft. x 9 ft. test section, produce velocities of up to 400 miles per hour and it would be capable of conducting both icing and aerodynamic research. Ice testing of various military and commercial
fixed-wing aircraft components in the tunnel commenced in 1944 following completion of icing cloud calibration. In 1960 the IRT was placed on standby status and NASA considered deactivating the facility completely due to the lack of demand for its use by industry. It was Boeing’s need to develop an anti-icing system for the CH-47
2 Chinkook’s engine inlets and their ability to convince NASA that rotorcraft icing was a unique issue that quite possibly saved the IRT from being shut down [21].
Since the establishment of NASA’s IRT in 1942, several other icing wind tunnels have been constructed. Goodrich Corporation in Ohio [26], the Italian Aerospace Re- search Centre (CIRA) [27], the Cox Icing Research Facility in New York City [28], and the Canadian National Research Council (NRC) Low Temperature Laboratory [29], to name some of the larger facilities, all house icing wind tunnels similar to that of NASA’s IRT. For in-flight icing research on helicopters, several methods have been developed and used. Examples of these include the Helicopter Icing Spray
System (HISS) operated by the U.S. Army [30], the Canadian NRC Ottawa Spray
Rig [29], and the McKinley Climatic Laboratory of the U.S. Air Force [?]. In addition to in-flight and wind tunnel ice testing, hover icing tests provide another useful and more economical option. Currently, there is only one hover test stand that is maintained and operated for testing of truncated rotors under icing condistions. The facility, the Adverse Environment Rotor Test Stand (AERTS), is located at Penn
State University [31].
1.1.2 Aircraft Icing
Atmospheric icing conditions occur when the ambient temperature drops below 0 ◦C.
When an aircraft flies through an icing cloud, droplets of supercooled water suspended within the cloud impact and freeze to the aircraft surface. Accretion of ice to a body traveling through the air is dictated by the heat transfer from the surface of the body.
The latent heat released from supercooled droplets upon freezing to the surface is counteracted by the convective cooling of the air passing over the body. It is the
3 balance of these thermodynamic processes that determine percent of the droplet that freezes on impact, referred to as the freezing fraction. Freezing fraction depends upon the ambient air temperature, liquid water content (LWC) of the icing cloud, the droplet size and the velocity of the body. Since the diameter of the droplets within a cloud varies, the median volumetric diameter (MVD) is used to define the droplet size distribution [32].
Colder ambient temperatures, lower LWC, a small MVD and low velocities result in the droplets freezing immediately on impact to the surface. This forms a relatively streamlined, but rough, ice shape which has an opaque milky appearance created by air pockets trapped within the ice. These are known as rime icing conditions. On rotorcraft, rime ice is formed on the low speed root region of the rotor blades. For ambient temperatures closer to 0 ◦C, a high LWC, large MVD and higher velocities, the entire impacted droplet does not freeze immediately. Instead, unfrozen portions of the impacted droplets run back and freeze further aft on the surface, creating a region of thicker ice build-up called "horns" that resemble a fish tail. This ice is clear in color and is known as glaze ice [32] and is found along the outboard region of helicopter rotor blades.
4 There is no clearly defined transition point from rime to glaze ice, instead there exists a transition region in which ice accreted can have the clarity of glaze ice but the conforming shape of rime ice. This transition region is referred to as mixed icing conditions. An example photograph of ice shapes formed in the three icing regimes is shown in Figure 1.1 and the effect of temperature, MVD and LWC on the regimes is illustrated in Figure 1.2. At warm temperatures and/or high velocities, kinetic heating of the surface of the body due to air friction prevents ice from forming. For a helicopter rotor, this occurs at the tip region of the blade [33]. A schematic of the ice formation along a rotor blade is given in Figure 1.3.
Figure 1.1: Examples of glaze (a), mixed (b), and rime (c) ice [1].
Two atmospheric icing envelopes are defined by the Federal Aviation Regulations
(FAR) in Appendix C of 14 CFR Part 25/29, continuous maximum and intermittent maximum. These envelopes outline a region of MVD, ambient temperatures, and
LWC in which transport aircraft ice protection systems must operate. Intermittent maximum icing conditions are more severe, i.e. larger LWC, than continuous maximum conditions and are designed to represent conditions within stratus cloud formations
17.4 nautical miles long while continuous maximum conditions represent conditions within cumulus cloud formations 2.6 nautical miles long. These conditions are
5 summarized in Figure 1.4. Typically, the continuous maximum conditions have been used for certification of airframe ice protection systems (IPS) and intermittent maximum conditions are used for engine ice protection [3].
Figure 1.2: Dependence of icing conditions on ambient temperature, LWC and MVD [1].
Figure 1.3: Typical ice accretion distribution along a rotor blade [2].
6 Figure 1.4: FAR Part 25/29 Appendix C Icing Envelope [2], [3].
1.1.3 Implications of Icing to Rotorcraft Operations
When an aircraft flies into a cloud of supercooled water, ice begins to build up on the structure of the vehicle. Ice accretion poses several problems, the most prominent being degradation of aerodynamic performance of the lifting surfaces such as wings or rotor blades. As ice builds up on a lifting surface its aerodynamic shape is lost and its surface roughness is increased dramatically, thus reducing lift and increasing drag as demonstrated in Figure 1.5. Although ice protection systems have been in use on
fixed wing aircraft since the 1930’s, numerous icing incidents are reported every year and occur on aircraft with and without ice protection systems. From 1978 to 2002 the Federal Aviation Administration (FAA), National Transportation Safety Board
(NTSB) and NASA have created 2,033 reports related to aircraft icing incidents and
7 accidents. Of these reported events, aerodynamic stall followed by loss of control was the most common outcome of an in-flight icing encounter [5] (see Figure 1.6).
More recently, from 2009 to 2014, the European Aviation Safety Agency (EASA) has reported 65 in-flight incidents concerning loss of control. Of the loss of control incidents, 20% of the most severe cases (accidents) and 8% of the less severe cases
(serious incidents) were caused by atmospheric icing conditions [34]. The dangers posed by in-flight icing conditions are so great that the National Transportation
Safety Board (NTSB) has had airframe icing listed on its most wanted list for safety improvements since 1997 [35].
Helicopters frequently perform time sensitive operations, such as search and rescue and emergency medical evacuations, which makes them more likely to encounter in-flight icing conditions. Ice build-up on rotor blades decreases their aerodynamic performance, leading to potentially dangerous changes in the maneuvering charac- teristics of the vehicle, increases in power required to maintain flight, increases in vibratory loads transferred to the airframe, and decreases in autorotation performance.
By increasing vibration levels, crew and passenger comfort is reduced along with a reduction of the life of components in the vehicle. If ice accretion continues without use of an IPS, required power levels can exceed installed engine power capabilities, resulting in an uncontrolled loss of altitude. In the event that ice is shed by centrifugal forces from the rotor, the ice becomes a projectile and poses a ballistic concern to the airframe and its components. In most cases, ice does not shed from rotor blades in a symmetrical fashion, introducing a large imbalance that could cause severe damage to the engine and transmission of the vehicle [36].
8 (a) Drag coefficient
‘ (b) Lift coefficient
Figure 1.5: Degradation of airfoil performance due to ice accretion [4].
9 Figure 1.6: Distribution of terminating occurrences from in-flight icing encounters. [5].
1.2 Ice Protection Systems
1.2.1 Electro-Thermal
Electro-thermal de-icing systems rely on heating the blade surface to melt the ice- leading edge interface using heater elements embedded within the blade or mounted to the inner surface of the erosion cap. Once a sufficient percentage of the bond line has melted, the ice is shed under a combination of aerodynamic and centrifugal forces. An example rotor blade with installed heater elements is shown in Figure 1.7.
Although effective, these systems are heavy (mostly due to power supply requirements) and consume large amounts of electrical power. According to Ref. [37], a typical electrothermal de-icing system require power densities as high as 25 W/in2. The addition of redundant electrical alternators to meet FAA certification requirements can introduce weight gains of up to 245 lbs. as discussed in [9], thus limiting the
10 de-icing system application to helicopters with higher payload capacity. It is not possible, given the available electrical power, installed to activate the entire de-icing system simultaneously, for this reason the heaters are broken up into zones which are powered one at a time [37]. During operation, ice is allowed to build up until thickness reaches approximately 0.3 in. at which point each zone is activated individually.
Doing this cycling procedure is necessary to limit power consumption and improve shedding performance (assisted by loads acting on the accreted ice thickness). Use of this system creates the issue of runback, in which melted ice runs aft along the blade surface and refreezes in unprotected regions. This ice buildup can degrade the aerodynamics of the blade even further, without the possibility to actively remove the accretion. As of now, this ice protection system is the only one that is certified by both the FAA and Department of Defense (DoD) for use in production helicopters [9].
Figure 1.7: Schematic of electrothermal de-icing system installed on a rotor blade [6].
11 1.2.2 Ultrasonic
Palacios et al., suggested the use of ultrasonic waves as a method of non-thermal ice protection for rotor blades in Ref. [38]. Piezoelectric actuators bonded to the inner surface of erosion caps are driven at high frequencies (20-100 kHz) to create transverse shear stresses at the ice-leading edge interface. If the shear stresses created exceed the shear adhesion strength of the ice to the metallic cap, the ice is shed under a combination of aerodynamic and centrifugal forces. A schematic depicting an ultrasonic de-icing system is provided in Figure 1.8. Initial research and testing found that excitation of the actuators tended to fracture and/or debond from their host structure. Early systems were incapable of producing transverse shear stresses necessary when power input to the system was less than 100 Watts. Overmeyer was able to optimize the actuator bond line in Ref. [7], thus maximizing the interfacial shear stress and decreasing the likelihood of actuator debonding. It was found that a transient excitation of the actuators was necessary to promote successful shedding of accreted impact ice during rotor ice testing, since using a single tone excitation did not promote ice shedding. The average electrical power required by the system was found to be 4 W/in2. A few years later in Ref. [1], Soltis investigated the effect on ultrasonic de-icing systems from adding tailored stress concentrators (addition of material in strategic locations to increase localized stresses) to the erosion resistant cap. The new system performance was compared to that of an electrothermal de-icing system and evaluated based on the ability of the system to shed ice during rotor testing along with the electrical power consumed. It was found that the ultrasonic system required 289 W to shed accreted ice and performed as well as the electrothermal system. It was found however, that while power to the thermal system could be
12 increased to improve its performance further, the maximum power supplied to the ultrasonic system was limited by the fracture of the actuators.
Figure 1.8: Schematic of unltrasonic de-icing actuator concept installed on rotor blade erosion cap [7].
1.2.3 Low Adhesion Strength Coatings
Low adhesion strength coatings were the first form of ice protection investigated, dating back to the 1920’s when coatings such as wax, oil and grease were tested for their anti-icing capabilities. These coatings are the only form of passive ice protection, and as their name implies, rely on a low adhesion strength between themselves and accreted ice to assist with shedding of ice under centrifugal loads. Since their conception, many researchers have developed low-adhesion strength coatings with varying levels of success. Several of the developed coatings are claimed to be "ice phobic", meaning that these coatings not only have a very low ice adhesion strength, but also prevent ice from building up on the surface in the first place. Ice phobic
13 coatings tend to be made of silicon or Teflon. While a number of the developed coatings exhibit very low ice adhesion strengths, in particular ice phobic coatings such as NuSil, discusussed in Ref. [8], has an adhesion strength of 5 psi compared to stainless steel, a commonly used material for rotor blade erosion caps, which has an adhesion strength of 80 psi. A sample comparison of adhesion strengths for ice phobic coatings developed for Pratt and Whitney is provided Figure 1.9.
Figure 1.9: Ice adhesion strength of various commercial ice phobic coatings [8].
An important feature of a coating that dictates its ability to shed accreted ice is its surface roughness. In Ref. [1], Soltis showed that the adhesion strength of a material is heavily dependent on surface roughness and small increases in roughness can significantly increase adhesion strength. Over time this poses a problem for any coating used on rotorcraft since ice phobic coatings do not have the erosion resistance capabilities needed for implementation to rotor blades. Ice phobic coatings tend to be very soft and do not survive on rotor blade environments. Rotor ice testing of
14 these coatings conducted at Penn State’s AERTS has shown that, in some cases after just one test, the material can become eroded by the super cooled water droplets impacting the surface, causing it to lose its low adhesion strength characteristics. In summary, no coating able to prevent ice accretion has been found to date. Coatings that reduce ice adhesion strength have not been implemented to rotor systems since they do not have the erosion resistance required.
1.2.4 Fluid Anti-Icing
The use of fluid anti-icing systems, also referred to as ”weeping wing", is an accepted practice among the fixed wing community for in-flight ice protection and its application to rotorcraft has been investigated. In fixed-wing aircraft, an alcohol/glycerin mixture is pumped from reservoirs within the wings. Metallic leading edge caps are perforated with thousands of small holes which the fluid mixture flows from and is distributed by the airflow over the wing. The chemical mixture prevents the formation of ice on the leading edge surface because they have a much lower freezing point than water
(−43 ◦C) [39].
In the 1960’s Bell adopted this concept to be used on a Bell HU-1 helicopter. To transfer the fluid mixture from reservoirs contained within the fuselage to the rotating frame of the blades, a "slinger-ring" was used. This configuration uses a flexible tube that is run through the shaft to the blade root. Fluid is pumped to the blades from the reservoirs and centrifugal forces then carry the mixture through the interior of the blade. Small holes along the leading edge are milled into the blade to allow the
fluid to escape where it is distributed by the airflow. Schematics depicting a fluid anti-icing set-up are provided in Figures 1.10 and 1.11 [9].
15 Figure 1.10: Fluid anti-icing system diagram [9].
Figure 1.11: Cross sectional view of fluid anti-icing system installed on a rotor blade [9].
The system as tested by Bell added a total weight of 194 lbs to the vehicle, carried
11 gallons of anti-icing fluid, and was capable of providing ice protection for 1 hr.
24 min. No aerodynamic penalty is introduced by these systems because ice is not
16 required to build up, which is not the case with other mechanical or thermal de-icing systems. Results from in-flight icing tests were promising and it was recommended that the system be developed further and put into production. However, funding was not available at the time and the project ended [9]. A second issue was the clogging of the orifices during take off and landing.
1.2.5 Electro-Impulse
The use of electro-magnetic impulses for aircraft de-icing was first suggested in 1937 and had been investigated in limited capacity in Europe before NASA Lewis and
Wichita State University began testing the concept in the 1980’s [40]. Copper coils mounted inside the leading edge of the wing of an aircraft, and separated from the surface by a small gap are used to create a large impulse when powered by discharging a bank of high voltage capacitors. When the capacitors are discharged, eddy currents are introduced to the leading edge. Instantaneous forces on the magnitude of hundreds of pounds are produced, creating interfacial stresses between the ice and metallic surface that act to fracture the ice. A schematic of this system is provided in Figures
1.12 and 1.13.
17 Figure 1.12: Electro impulse coil installed on the leading edge of a wing [9].
Figure 1.13: Electro-impulse de-icing system diagram [9].
18 The described system was tested in icing conditions on several platforms including both fixed and rotary wing aircraft as well as engine inlets in Ref. [41]. Electrical power required for successful de-icing was estimated to be 1 kW for general aviation sized aircraft and 3 kW for a helicopter with gross weight of 15,000 lbs, roughly
10% of the electrical power required for an electrothermal system. Two issues arise with implementation on helicopter rotor blades however. The first simply being that of sufficient space available within the blade for the coils and other required components. Material must be removed from the blade to make room for the coils and wire, reducing fatigue life of the blade. Second, erosion caps on rotor blades are much less compliant than skin on the wing of an airplane and therefore make the installation of the system difficult [9].
1.2.6 Electrovibratory
Another method of mechanical de-icing that has been tested on helicopter rotor blades is electrovibratory de-icing systems. These systems essentially shake the blades to remove ice that has formed. By mounting mechanical shakers on the blade root and driving them at frequencies matching the natural frequencies of the blade, large accelerations are created which shed accreted ice. In 1978, Bell Helicopter and the
U.S. Army investigated the de-icing performance of several shaker mounting locations on a 21 in. chord UH-1 blade in Ref. [42]. The shakers used 1.25 lbs. eccentric weights driven by a 0.5 HP motor. Mounting locations were within the blade spar, on the blade trailing edge, on the blade grip at the root, and on the pitch arm. These mounting locations are illustrated in Figure 1.14.
19 Figure 1.14: Electrovibratory de-icing system actuator configurations [9].
It was found that driving the system in a frequency range from 0 Hz to 47 Hz, the blade was able to be de-iced in all locations except for the tip region.To achieve full blade de-icing capability, an auxiliary ice protection system (such as electrothermal) would need to be installed in this area. While the tested system was low weight (67 lbs) and required only 1.3 kW, fatigue on the blade was a concern [9].
20 1.2.7 Pneumatic
In 1930 the first pneumatic de-icing boot was invented and tested by B.F. Goodrich and the NACA in Cornell University’s icing wind tunnel. The boot was constructed of a rubber sheet reinforced with fabric fastened to the leading edge of the surface to be protected and was inflated by compressed air from a pump [10]. Inflation of the"’protective overshoe"’ as it was referred to early on in Ref. [10], created transverse shear stresses at the ice-leading edge interface. Stresses exceeding the adhesion strength of the ice to the surface would cause the ice to debond from the surface and be carried away by airflow over the wing. Cycling the system several times at a frequency that allowed the boot to fully inflate and deflate removed accreted ice
(Ref. [21] notes that early systems were cycled 3 times per minute). A schematic of the early system is illustrated in Figure 1.15. Promising results from wind tunnel and flight tests led to the funding and construction of Goodrich’s first icing wind tunnel [21]. Not long thereafter the pneumatic de-icing boot was adopted to most airliners to protect wing, tail and vertical stabilizer leading edge surfaces.
21 Figure 1.15: Drawings of the first pneumatic de-icing boot as developed and tested by B.F. Goodrich [10].
Still today, the pneumatic de-icing system is the most prominent form of ice protection for fixed-wing aircraft and their design and operation remains mostly unchanged. Instead of using compressed air from a pump, modern systems use engine bleed air is used to inflate the boot and have optimized inflatable zones, making the systems more efficient. Elastomeric boot material has also been improved, allowing for more aerodynamically smooth surfaces and improved UV and erosion resistance [43].
An example of an improved design is the pneumatic impulse ice protection system
(PIIP), developed by B.F. Goodrich in the late 1940’s. This new system was comprised of several small inflatable spanwise tubes constructed of reinforced fabric and covered with a metallic erosion resistant material, illustrated in Figure 1.16. When pressurized, the tubes rapidly expand, in about 50 µs, to dislodge the ice. Unlike traditional
22 pneumatic de-icing systems which require less than 20 psig to operate, the PIIP required very high pressures, ranging from 400-1500 psig. The high pressures required the addition of a dedicated on-board air compressor to run the de-icing system.
Figure 1.16: Operation of the PIIP [11].
Pneumatic de-icing systems are low power, low weight solutions to in-flight icing.
Total system weight for a twin engine business aircraft is only 50 lbs., and consumes
1 HP from the engine along with a very small amount of power to run inflation solenoid valves [43]. These characteristics make pneumatic ice protection appealing to medium sized helicopters that are limited in available electrical power and available payload. In 1979, NASA, B.F. Goodrich and the U.S. Army began development of a pneumatic de-icing system for helicopter rotor blades. A newly developed erosion
23 resistant polyurethane elastomer (ESTANE) material produced by B.F. Goodrich was used to construct the pneumatic boot that was otherwise very similar in design to those used on fixed wing aircraft and is depicted in Figure 1.18 [12]. The de-icing boots were fitted to production UH-1H helicopter rotor blades to be used for in-flight icing tests. Air pressures of 25 psig was provided to the system via engine bleed air through a series of valves and pressure regulators and a pneumatic slip-ring to transfer the pressurized air from the rotor hub to the rotating blades. A diagram of the system is provided in Figure 1.17. A full inflation of the boot took 2 seconds and a full inflation/deflation cycle was completed in 30 seconds. Total installed weight of the system was measured to be 40 lbs and power consumption was negligible.
Figure 1.17: Schematic of pneumatic deicing system as installed on the UH-1 rotor [9].
Multiple in-flight ice tests were conducted, both under natural and artificial icing conditions. Testing revealed the system was capable of successfully de-icing the blades
24 when ice had accumulated to roughly 0.3 in. in thickness. Inflation of the boots in forward flight was shown increase power required to the rotor and increased torque by 27% along with changing flight dynamic qualities but these deviations in handling were deemed to be acceptable by test pilots [9]. It was discovered that forward flight above 90 KCAS with the system deflated would induce moderate vertical vibrations to the airframe and inflation of the boots at this airspeed produced severe vertical vibrations. Erosion of the boots along with several punctures were discovered during testing that effected the ability of the system to inflate and deflate [12]. These set-backs and a lack of funding led to the project ultimately being abandoned.
Figure 1.18: Schematic of pneumatic deicing boot as installed on the UH-1 rotor [12].
1.3 Methods of Ice Testing
1.3.1 In Fight Testing
Certification of full-scale aircraft components that are too large for wind tunnel testing must be subjected to in-flight icing tests. Two methods for in-flight ice testing are available. The two methods are natural in-flight ice testing,("chasing the weather") and artificial in-flight ice testing. Natural icing trials, as one may imagine, are expensive and time consuming. Reliance on the weather and the ability
25 of the test aircraft to reach the altitudes and distances required to find proper icing conditions makes this type of testing difficult for smaller vehicles [13]. Artificial spray systems provide an easier means to evaluate ice protection systems as the cloud can be controlled to produce a desired icing condition and are often used for ice protection system certification. Icing tests obviously require that ambient temperature be cold enough to produce ice, so tests are only able to be run during winter months. The
Canadian National Research Council (NRC) Low Temperature Laboratory in Ottawa,
Ontario was the home for the NRC Ottawa Spray Rig pictured in Figure 1.19. This facility allowed a helicopter to hover 30 meters downstream from an array of icing spray nozzles. Icing clouds produced by this system were measured to have cloud densities as high as 0.9 g/m3 and droplet sizes ranging from 20-60 microns [29]. The facility is no longer operational.
Figure 1.19: Schematic of the Canadian National Research Council (NRC) Low- Temperature Laboratory in Ottawa, Ontario [13].
Artificial forward flight icing tests can be conducted using the U.S. Army Helicopter
Icing Spray System (HISS) as described in Ref. [30]. A modified Boeing CH-47C is used as an airborne spray tanker. An external spray boom towed behind the aircraft at forward speeds up to 150 KIAS, produces an artificial icing cloud which helicopters or slow flying fixed wing aircraft can fly in. The spray rig is capable of producing a
26 continuous cloud for 30 minutes with a cross section of 15 ft x 55 ft measured 200 ft aft of the boom. A schematic of the set-up is provided in Figure 1.20. Engine bleed air and an auxiliary power unit (APU) provide regulated air pressure that is combined with water via atomizing spray nozzles to produce a representative icing cloud. A chase aircraft is used both as a safety precaution during in-flight icing tests with the HISS and also provides high speed video of ice protection systems during operation on the test aircraft.
Figure 1.20: Schematic of the U.S. Army Helicopter Icing Spray System (HISS) [13].
The ability of the HISS tanker to climb and descend to locate desired ambient temperatures for testing makes it an attractive option compared to in-flight testing using fixed altitude spray rigs such as NRC. However, issues with the HISS have been discovered and pointed out in Ref. [13] and Ref. [44]. The spray system produces unrealistically large droplet sizes when LWC exceeds 0.5 g/m3 (100 microns for an
LWC of 1 g/m3) and the cloud density tends to be nonuniform requiring the pilots of the test aircraft to weave throughout the cloud.
The McKinley Climatic Laboratory, located at Eglin Air Force Base, provides the ability to subject full-scale vehicles and components to harsh environments, see Figure
1.21. An insulated hanger with refrigeration and heating systems allow vehicles as
27 large as the Lockheed C-5A transport aircraft with engines running to be subjected to temperatures ranging from (−62 ◦C) to (77 ◦C). The ability to produce up to 25 inches of rain per hour, snow and, 50 mph winds, create a unique testing option compared to chasing the weather.
Figure 1.21: Photograph of the Lockheed F-117 in the McKinley Climatic Labora- tory [14].
1.3.2 Icing Wind Tunnels
When full-scale testing is not required or aircraft components to be tested are relatively small, icing wind tunnels provide an economical and convenient alternative to in-flight ice trials. Typically, icing wind tunnels are used to study the effects of ice accretion to 2-D airfoil performance and sub-scale models, for development of ice protection systems, and to validate ice accretion simulation computer codes. Currently, there
28 are several icing wind tunnels in operation, NASA Glenn Icing Research Tunnel
(IRT) [45], the Goodrich Icing Wind Tunnel [26], the Cox Icing Research Facility [28], the Italian Aerospace Research Centre (CIRA) Icing Wind Tunnel [27], and Boeing’s
Research Aerodynamic Icing Tunnel [46], to name some of the larger facilities. All of the mentioned facilities are closed loop wind tunnels and have refrigeration systems that allow them to be run all year at temperatures spanning the FAR icing envelope.
NASA’s IRT is one of the largest of the facilities mentioned, with a test section measuring 9 ft by 6 ft and is capable of reaching speeds of 350 kts. Spray bars containing atomizing nozzles are located upstream of the test section and provide a controllable icing cloud. A schematic of a typical test section is shown in Figure 1.22.
Figure 1.22: Typical set-up of an icing wind tunnel [13].
1.3.3 Hover Test Stands
Icing wind tunnels have limited test capacity in the size of the test specimen that can be evaluated. Also, in particular for rotorcraft, sub-scale rotors tested in icing wind tunnels are not able to reproduce centrifugal forces representative of a full-scale blade [31] on a full-scale chord blade. To address this need, in 2009 Penn State
University constructed its Adverse Environment Rotor Test Stand (AERTS). The
29 facility uses a QH-50 helicopter rotor hub capable of spinning 5 ft. radius blades up to 1200 RPM. Ballistic walls surround the test stand to protect operators from any ice that is shed or structural failures. Consequently, full-scale centrifugal blade forces can be reproduced in the AERTS facility on full-scale chord, span-truncated rotors, expanding the range of components and conditions that can be evaluated. If needed, electrical and pneumatic slip-rings are available to provide power and/or pressurized air to the blades. Schematics of the AERTS are provided in Figures 1.23 and 1.24.
Figure 1.23: AERTS facility layout [4].
30 Figure 1.24: Photograph of OH-50 rotor head in the AERTS facility [4].
The facility is capable of reproducing atmospheric icing conditions conforming to the FAR icing envelope. NASA standard icing nozzles are arranged in two concentric rings and mounted in the ceiling of a 20 ft x 20 ft x 11.5 ft commercial freezer unit.
A uniform icing cloud is produced by maintaining specific air and water pressures inputs to the nozzles. Each nozzle can be turned on and off separately allowing for full control of cloud MVD and LWC. In Ref. [31], Palacios showed that ice shapes created in the facility match, in very good agreement, with those produced by the
NASA Glenn IRT.
31 1.4 Review of Prior Centrifugally Powered Pneumatic
De-Icing Systems
In recent years, Penn State and Invercon have addressed two of the main concerns that led to the abandonment of the application of pneumatic ice protection boots for helicopter rotor blades. Namely, erosion of the ESTANE material and the pneumatic slip-ring required to route engine bleed air to the boots.
The possibility of using pressure differentials created inside of a rotating helicopter blade for actuation was investigated as an alternative to pneumatic slip rings and other actuation systems that require electrical power. In Ref. [15] Szefi et. al experimentally verified that the pressures generated along a blade due to rotation are governed by the hydrostatic equation dp = ρ(r)rΩ2 (1.1) dr where ρ is the density of air, r is the blade radial location and Ω is the rotational speed. A 24 ft. Kaman K-MAX blade was modified to allow two pneumatic hoses to be run down the center of the blade (internally) from root to tip and instrumented with pressure sensors at each end. One hose was open to atmosphere at the root and sealed at the tip to create a high pressure during rotation. The second hose was of the opposite configuration, sealed at the root and open to atmosphere at the tip, creating a low pressure distribution during rotation.
Full scale whirl tower testing of the concept showed that an almost constant 7.5 psig pressure differential between the high and low pressure lines is available along the blade when spun at 280 RPM. The experimentally measured pressures along
32 the blade in both high and low pressure lines and a simple schematic illustrating the concept are provided in Figure 1.25. By cycling a valve between the low and high pressure lines, the pressure differential can be utilized for actuation purposes as discussed in Ref. [15].
(a) Centrifugal pumping schematic
(b) Experimentally measured pressures
Figure 1.25: (a) Simple schematic of centrifugal pumping concept, (b) experimentally measured pressures created by centrifugal pumping on a 24 ft. blade at 280 RPM [15].
33 Following these findings, a collaborative effort between Penn State and Invercon developed a pneumatic de-icing system using only the measured centrifugally generated pressures and thus eliminating the need for a pneumatic slip-ring as described in
Ref. [16] The first prototype developed consisted of a metallic erosion cap cut into spanwise strips. The strips were then bonded to inflatable elastomeric bags which were bonded to a NACA 23012 blade surface. A schematic of the prototype is provided in
Figure 1.28. During non-icing conditions, the low pressure line was used to deflate the pneumatic bags and hold the metallic strips to the aerodynamic shape of the blade. When the system is to be inflated, the valve is switched to the high pressure line and the bags inflate, creating transverse shear stresses at the ice-leading edge interface that act to debond the ice.
Wind tunnel testing of this early system showed undesirable aerodynamic per- formance in both the deployed and undeployed states. An improved system was developed to reduce the aerodynamic penalty created by the erosion cap when unin-
flated. This second prototype replaced the segmented erosion cap with a continuous one, and it is shown in Figure 1.26. Wind tunnel testing showed the new design eliminated the aerodynamic penalty created by the erosion cap of the first design.
Additionally, the effect of inflation of the erosion cap on airfoil drag was investigated and compared to airfoil data from literature that had accreated ice. It was found that an airfoil with ice accreation had a more detrimental effect on aerodynamic performance than that created when inflating the ice protection system to remove the ice. Sample results from these tests are given in Figure 1.27.
34 Figure 1.26: Photograph of the second pneumatic de-icing system prototype [16].
(a) Wind tunnel lift measurements (b) Wind tunnel drag measurements
Figure 1.27: Measured lift and drag coefficients of the pneumatic de-icing system compared to an iced airfoil [17].
Rotor ice testing of the second design at Penn State’s AERTS facility was conducted across a test matrix spanning the FAR Part 25/29 Icing Envelope. Since the AERTS is only capable of accepting 5 ft. radius blades, the pressures produced by centrifugal pumping there were not large enough to actuate the system (since the pressure generation is proportional to r2). Therefore, a pneumatic slip-ring was used to transfer air pressures representative of those generated on a full scale blade to the test blades. Results from the ice testing showed the pneumatic de-icing system ability to delaminate accreted ice thicknesses as small as 0.06 in. It was observed, however, that at colder temperatures in the icing envelope (less than −15 ◦C) the inflatable
35 elastomeric bags became stiff. Once this happened, higher input air pressures to the system were required to delaminate accreted ices from the blades. The potential for the pneumatic bags to rupture also posed a concern during testing.
Figure 1.28: Initial centrifugally powered pneumatic de-icing system developed by Penn State and Invercon in the undeformed and deformed states [16].
1.5 Objectives
The objective of this research is to improve upon the design of an existing centrifugally powered pneumatic de-icing system and to evaluate the performance of the system in full-scale hover icing tests. A successful design will be capable of operating with the pressures created by centrifugal pumping inside a spinning full-scale rotor blade. To achieve this objective, the following steps must be taken:
1. Design and analyze an improved pneumatic de-icing system, making use of
finite element methods and cohesive material constitutive laws. The weight and
36 size of the new design should be comparable to that of existing erosion caps on
in-service rotary-wing vehicles.
2. Experimentally evaluate the designed ice protection system(s) under represen-
tative atmospheric icing conditions in Penn State’s Adverse Environment Rotor
Test Stand (AERTS). The tested designs will be evaluated based on their ability
to shed accreted ice, and the top performing configuration will be installed on
a full-scale blade for further testing. Experimental results will also be used for
model validation.
3. Install the selected design onto a full-scale rotor blade, making as few modifica-
tions to the original blade as possible, and maintaining the clean aerodynamic
shape of the blade.
4. Design and construct a portable icing cloud generator capable of creating
icing conditions typical of those encountered in atmospheric flight. The cloud
generator will provide a means to conduct full-scale hover icing tests of the
de-icing system.
5. During full-scale hover icing trials, evaluate the ability of the de-icing system
to shed accreted ice relying solely on centrifugally generated pressures and
centrifugal forces.
37 1.6 Thesis Overview
1.6.1 Chapter 2: Design and Analysis of a Pneumatic De-Icing
System
The design and performance of a benchmark centrifugally powered pneumatic deicing system is summarized. From this benchmark design, an improved ice protection system is developed for both the inboard and outboard sections of a 24 ft. (7.32 m.) blade using Abaqus finite element software. The ability of the new system to delaminate accreted ice is evaluated using Abaqus cohesive surface method to predict ice-leading edge interfacial failure. Blade leading edge cap deformation due to aerodynamic pressures produced by the rotating blade are examined. As the cap must be compliant enough to deform under applied system pressure yet stiff enough to resist being pulled away from the blade during non-icing conditions, the parametric design of the leading edge cap thickness is of utmost importance.
1.6.2 Chapter 3: Prototype Rotor Ice Testing and Finite
Element Model Prediction Validation
Prototype de-icing systems were fabricated and tested in Penn State’s Adverse
Environment Rotor Test Stand (AERTS). Results from these small-scale experiments were used to refine the ice protection system designs further and arrive at a configuration suitable for full-scale testing. A finite element method using cohesive zones methods for predicting pneumatic de-icing system ability to delaminate ice is reviewed. The described method is validated using results from AERTS testing of the prototype
38 de-icing systems.
1.6.3 Chapter 4: Portable Icing Cloud Generator Design,
Construction and Testing
A portable icing spray system capable of producing a cloud controllable within the
Federal Aviation Regulations (FAR) Part 25/29 icing envelope has been developed.
This generator was used to conduct full-scale ice testing of a centrifugally powered pneumatic de-icing system. Details of the design and testing of the cloud generation system are presented in Chapter 3.
1.6.4 Chapter 5: Full Scale Pneumatic De-Icing System Testing
The process of integrating the final pneumatic de-icing system design onto an existing full-scale 24 ft. (7.32 m.) K-MAX rotor blade is presented in Chapter 5. Full-scale testing of the outboard section design was conducted at Kaman Aerospace Corp. and showed the de-icing system ability to delaminate accreted ice with only centrifugally generated pressures. Results from the testing are discussed in this chapter.
1.6.5 Chapter 6: Conclusions and Recommendations for Future
Work
The final chapter gives an overview of the research performed. Recommendations for future work dealing with mobile full-scale icing testing as well as improvements to the developed centrifugally powered pneumatic de-icing system are provided.
39 Chapter 2 | Design and Analysis of a Pneumatic De-Icing System
2.1 Introduction
The principle of operation for a pneumatic de-icing system is to create transverse shear stresses between accreted ice and the leading edge cap by inflating a flexible structure. If the interfacial stresses created by system inflation exceed the cohesive properties of the ice to the substrate, the ice debonds from the surface and is carried away by the air flow over the surface or centrifugal forces in the case of rotor blade or propeller applications. Fixed-wing aircraft typically use flexible polymer boots to achieve this effect. However for rotorcraft, erosion of the material becomes an issue due to the harsh erosion environments in which they operate. To address the issue of erosion, erosion resistant materials must replace the polymer boots of fixed-wing vehicles. Unfortunately, erosion resistant materials are generally metallic and less
flexible, so special measures must be taken to achieve the desired inflation effect.
For this research, two novel pneumatic de-icing system designs were developed based on the systems presented in Ref. [16]. The focus of the first design (Prototype
40 I) was to eliminate the stiffening of the inflatable rubber bags at lower temperatures described in Ref. [16]. Lessons learned during fabrication and bench top testing of
Prototype I drove the decisions made for initial designs while developing Prototype
II. This chapter discusses the two de-icing systems developed during this research and the methods used to model them.
2.2 Prototype I Design
The pneumatic de-icing system tested by Bailey in Ref. [16] proved its ability to delaminate accreted ice, however it was noticed that the inflatable rubber bags became stiff at colder temperatures, thus degrading the system performance. Using Bailey’s design as a baseline, a new design which replaced the inflatable rubber bags with metallic ribs was developed. The new design, Prototype 1, was constructed of a
0.03 inch thick 304 Stainless Steel erosion cap pressed with a die into the shape of a
NACA 23012 airfoil. The selected thickness and material are typical of erosion caps installed on in-service helicopters. To eliminate the issue of the rubber bags stiffening at colder temperatures, the bags were removed and the inner surface of the leading edge cap was used as the pressure sealing surface. A single pressurized zone was separated by span-wise running 0.005 inch thick strips of 1095 Carbon Spring Steel that was bonded to the blade and leading edge cap, and located on both the upper and lower blade surfaces. Starting from the leading edge moving aft, the widths of the metallic ribs were 0.25 in, 0.75 in., and 1 in. For ease of fabrication, each rib used a bond-line length of 0.25 in. The perimeter of the de-icing system was sealed with 0.0625 in. thick ethylene propylene diene (EPDM) elastomer, which was bonded to the inner cap surface and the outer blade surface. The elastomer selected for the
41 de-icing design was rated to −40 ◦C and had a shore hardness of 60A. A schematic of
Prototype 1 is given in Figure 2.1.
Figure 2.1: Schematic of the first pneumatic de-icing system prototype in the un- pressurized (top) and pressurized (bottom) configurations.
During operation, the bonds and metallic ribs are the primary load carrying structure for the de-icing system, and are subjected to high shear stresses created by centrifugal loads. For this reason, a high shear strength methacrylate adhesive
(Loctite H4500) was selected to bond the spring steel ribs to the blade and cap surfaces. Adhesion strength testing was conducted on the selected adhesive to verify manufactures quoted strength values, as well as to confirm the surface preparation techniques of the substrates to be bonded. A standard lap joint shear method was used as described in Ref. [47]. A schematic of the experimental set-up is provided in
42 Figure 2.2 and sample results from the shear testing are presented in Figure 2.3. The quoted shear strength for the selected adhesive is 3000 psi., while the experimentally measured average strength was 2483 psi. This represents a difference of 18%.
Figure 2.2: Schematic of lap-joint shear test set-up.
43 Figure 2.3: Sample stress/strain data from lap joint shear adhesive tests.
2.3 Prototype II Design
The design of Prototype II aimed to address short comings in Prototype I that were noticed during bench-top tests. Most noticeable of these were the location of the maximum leading edge cap deformation and the large discontinuity created at the aft location of the cap during actuation. In Figure 2.1 b, it is clear that the maximum deflection of the Prototype I leading edge cap is a maximum at the aft region. However, it is desirable that the leading edge cap of a pneumatic de-icing system create its maximum deformation closer to the leading edge, since this is where ice accretes. For this reason, it was decided that the aft region of the cap in Prototype
II would be tied to the blade surface using an EPDM elastomeric sheet. Using this flexible material would prevent the cap from departing the blade surface as in
Prototype I, thus requiring the cap to deform closer to the leading edge. In addition to the de-icing benefits, constraining the aft cap region removes the discontinuity
44 created during system actuation, thus greatly improving aerodynamic performance during actuation. The result is a deformed shape that represents more closely the
’ballooning’ shape typical of fixed-wing pneumatic de-icing boots. A schematic of
Prototype II in the deflated and inflated states is provided in Figure 2.4.
Figure 2.4: Schematic of the second pneumatic de-icing system prototype in the un-inflated (top) and inflated (bottom) configurations.
Prototype II did not use the flexible 1095 Carbon Spring Steel ribs as in Prototype
I. Instead, continuous spring steel ribs were wrapped around the blade leading edge and bonded to the inner surface of the cap and outer surface of the blade to provide the load path for centrifugal forces acting on the system. The metallic ribs were 4 in. in width and 0.005 in. thick and were bonded to the blade and cap in an alternating fashion as illustrated in Figure 2.5.
45 (a) Prototype II alternating bond pattern.
(b) 3-D model of Prototype II of alternating bond pattern.
Figure 2.5: Schematic of Prototype II alternating bond pattern for flexible metallic ribs.
46 Like Prototype I, the leading edge cap of Prototype II was constructed of 0.03 in. thick 304 Stainless Steel. However, for Prototype II, the inner 0.015 in. portion of the cap thickness was segmented in the span-wise direction at the leading edge and upper and lower airfoil surfaces. By introducing these segments, the local stiffness of the cap is reduced in these areas, thus allowing the cap to flex in critical ice accretion locations, creating higher stresses to assist with ice delamination. These segments can be seen in Figure 2.4.
2.4 Finite Element Modeling of De-Icing Systems
Finite element modeling of both prototype designs was conducted, using Abaqus finite element software, to guide the system geometry selections. Initial modeling aimed at simply maximizing leading edge cap deflection so as to create high transverse shear stresses along the outer surface of the leading edge cap to promote delamination of accreted ice. Later modeling studied the effects of aerodynamic pressures on cap deformation during non-icing conditions and methods of minimizing this unwanted effect. All models created were 2-D to reduce computational cost.
2.4.1 Blade
In the Abaqus finite element model, only the leading edge portion of the NACA
23012 blade geometry that contained the de-icing system was included to reduce computational time. The blade material was modeled as 304 Stainless Steel for both
Prototype I and II. In reality, the rotor blade is not constructed of this material, however for this analysis the material is unimportant as the blade served only as a platform for mounting the de-icing system. In Abaqus, the clamped boundary
47 condition is used to fix all rotational and translational degrees of freedom of a body.
To ensure the blade remained fixed, clamped boundary conditions were applied to the aft portion of the blade.
2.4.2 Leading Edge Cap
For Prototype I, the leading edge cap was modeled as a continuous extruded geometry that was created in Solidworks using NACA 23012 airfoil coordinates and subsequently imported into Abaqus. The leading edge cap was constructed from 0.03 in. thick 304
Stainless Steel and was held to the rotor blade via the 1095 Carbon Spring Steel ribs on both top and bottom surfaces. Tie constraints were used in Abaqus to bond the
Carbon Spring Steel ribs to the blade and leading edge cap surfaces and can be seen in Figure 2.1. The bond-line length and rib sizes were identical to those discussed in Section 2.3. Material properties used to define the 304 Stainless Steel and 1095
Carbon Spring Steel in the models are listed in Table 2.1.
The leading edge cap of Prototype II was constructed from 304 Stainless Steel, again using the NACA 23012 airfoil data in Soldworks and importing the part into
Abaqus. Unlike Prototype I, two separate 0.015 in. thick caps were constructed and tied together to create one 0.03 in. thick part. This method allowed the inner 0.015 in. cap to be segmented as per the design discussed in Section 2.4. Recall that Prototype
II did not use the array of Carbon Spring Steel ribs bonded to the top and bottom blade surfaces. Instead, the Prototype II design used continuous Carbon Spring Steel sheets that wrapped around the blade as shown in Figure 2.5. In Prototype I, the function of the ribs were to hold the leading edge cap to the blade during de-icing system actuation. Conversely, in the design of Prototype II, the continuous metallic
48 sheets were introduced to ensure that the elastomer would not carry centrifugal loads during rotation, and provided little stiffness normal to the rotor blade surface. For this reason, the Carbon Spring Steel sheets were not included in the Abaqus model of Prototype II and the leading edge cap in the model was tied to the rotor blade upper and lower surfaces with the elastomer only. The elastomer bond-lines can be seen in Figure 2.4.
Table 2.1: Material properties of metals used in de-icing systems.
Material Young’s Modulus (GPa) Poisson’s Ratio Density (kg/m3)
1095 Spring Steel 200 0.30 7861
304 Stainless Steel 193 0.27 8027
2.4.3 Elastomer
EPDM elastomer was used in the design of Prototype II for the purpose of constraining the aft portion of the leading edge cap to the rotor blade surface. Surface-to-surface ties were used to bond the elastomer to the blade and leading edge cap surfaces in the
Abaqus model. Introduction of the elastomer into the design for the second prototype required the selection of a hyperelastic strain energy density model. Most rubber materials can be assumed to be isotropic and incompressible as described in Ref. [48] and therefore a Poissons ratio of 0.5 was used. The Neo-Hookian strain energy density is derived from statistical mechanics and thermodynamic principles and is the simplest of the hyperelastic models, requiring only one coefficient determined from experimental test data. It is sufficiently accurate for strains of less than 40% [49]. For
49 these reasons it was decided that the Neo-Hookian model would be used to model the elastomer. Abaqus accepts either strain energy material coefficients or material test data into its hyperelastic models. Uniaxial tension material test data for an
EPDM elastomer with a shore hardness of 55A adopted from [18] is shown in Figure
2.6 and was used to define the model material. It should be noted that the design of
Prototype I did used elastomer, but its purpose was solely to provide a pressure seal and was therefore not included in the finite element model.
Figure 2.6: Uni-axial test data for EPDM elastomer from Ref. [18].
2.4.4 Ice
To study the de-icing systems’ ability to delaminate accreted ice, Abaqus parts resembling realistic ice shapes were created. The geometry of the ice shapes were
50 calculated using LEWICE, a NASA developed software capable of predicting ice shapes representative of those encountered in-flight. The ice shapes were constrained to the surface of the leading edge cap using experimentally measured cohesive properties.
An example ice shape tied to the de-icing system Abaqus model is shown in Figure
2.7. Basic material properties of ice used for the models were taken from Ref. [16] and are summarized in Table 2.2.
Figure 2.7: Representative ice shape attached to Prototype II de-icing system.
Table 2.2: Material properties of ice.
Young’s Modulus (GPa) 9
Poisson’s Ratio 0.27
Density (kg/m3) 917
2.4.5 Mesh
A free mesh technique was used on all parts in the Abaqus finite element model assembly. 2-D, plane strain, reduced integration, quadratic elements (CPE8R) were used for all parts except the elastomer in Prototype II which used 2-D plane stress
51 elements (CPS4R). Element size within the rotor blade was not critical, therefore elements were relatively large. Care was taken with all parts that would undergo bending to have at least three quadratic elements through the thickness so as to capture bending effects accurately. Examples of these parts are the leading edge cap, spring steel ribs, and elastomer. Element sizes along the outer edge of the leading edge cap and inner edge of the ice shape (i.e., the interface between the ice and cap) was controlled to accurately capture cohesive failure effects. This topic is discussed in more detail in Section 2.5.1.1.
To ensure accuracy in a finite element model solution, a mesh convergence study must be conducted. A mesh convergence study is the process of running a model with a certain element diameter and monitoring the convergence of a solution as the mesh diameter is decreased. This process is continued until the solution has converged within a defined tolerance. By running and re-running the model with increasingly smaller element diameters, the smallest element size required by the model for solution accuracy can be obtained. Using element sizes any smaller than that determined by the convergence study only increases computational cost with little to no change in solution accuracy.
Required element sizes for both Prototype I and II models were determined using a mesh convergence study. The study monitored the convergence of the maximum leading edge cap deformation value when the de-icing system was pressurized with
8 psi. A convergence tolerance of 5% was chosen to determine mesh convergence.
Once this tolerance was met the mesh size was decreased further to ensure that the solution had indeed converged. Results from convergence studies for Prototype I and
II are presented in Figure 2.8 and Figure 2.9 respectively.
52 Figure 2.8: Mesh convergence of leading edge cap for Prototype I.
Figure 2.9: Mesh convergence of leading edge cap for Prototype II.
2.5 Prediction of De-Icing System Delamination
Performance
In Ref. [16], a process was developed that is capable of quantifying the effectiveness of a pnuematic de-icing system at removing of accreted ice from its surface. This 53 process adopted the proven technique of modeling composite delamination, which occurs when micro-cracks are formed between two layers of composite laminates and propagate under a combination of interfacial peel and shear stresses. The cohesive zone method is the most recently developed technique used for modeling composite delamination and crack growth, and is the method adopted in Ref. [16] to predict pneumatic de-icing system performance. The process of using the cohesive zone method for modeling the ice/leading edge interface of a pneumatic de-icing system is discussed in the subsequent sections.
2.5.1 Review of Cohesive Failure Theory
The ice/leading edge interface of the pneumatic de-icing system can be modeled using the cohesive zone method, a technique applied in predicting composite delamination and crack growth [17]. This method is governed by traction-separation laws, as opposed to typical stress-strain laws, which assume a decrease in the load carrying capability of the interfacial cohesive layer as the two adhered substrates are separated.
Once the adherents reach a critical separation distance as defined by material cohesive and damage properties, that region can no longer carry any load and is said to be failed [19]. This process is illustrated graphically in Figure 2.10 for a Mode I delamination for a double cantilever beam.
54 Figure 2.10: Example of a Mode I delamination event [19].
In Figure 2.10 the traction separation curve (bottom) describes the delamination process of a double cantilever beam under-going pure Mode I loading (top). From points 1-2 the cohesive zone behavior is linear elastic and the slope of the curve is called the penalty stiffness, Kp. Damage of the cohesive layer begins once separation
0 reaches a critical value,δI , which occurs at point 2 and is given by
0 T δI = (2.1) Kp where T is the Mode I ultimate tensile strength of the cohesive layer. After damage initiation, the load carrying ability of the cohesive zone is reduced until a critical
F failure separation,δI , is reached, denoted by point 4. The critical Mode I failure separation is given by 2G δF = IC (2.2) I T where GIC is the fracture energy of the interface. At this point the cohesive zone is completely failed and can no longer carry any load. The described process is exactly
55 the same for Mode II and Mode III delamination, only using ultimate strength and fracture energies corresponding to the proper failure mode.
Pressurization of the pneumatic de-icing system does not produce a pure Mode
I, II or III loading condition, but rather a combination of the three. To model this effect, the power law, a popular mixed-mode failure criterion, can be applied. The critical mixed-mode failure separation is given by
v u 2 u 1 + β 0 0 0 t δ = δI δII 0 2 0 2 (2.3) (δII ) + (βδI ) where β is called the mode mixity and is given by
δ β = II (2.4) δz
and δz is the relative Mode I separation. The power law assumes that the delamination mechanisms in Mode II and Mode III are the same and are therefore combined into a total tangential displacement given by
q 2 2 δII = δx + δy (2.5)
where δx and δy are the relative tangential separations in the x and y directions respectively.
Using the described method, the critical mixed-mode failure separation for the ice/leading edge interface of the pneumatic de-icing system was calculated to be 6.88
µm using experimentally measured properties listed in Table 2.3 from Ref. [16].
56 2.5.2 Application of Cohesive Zone Methods to the Finite
Element Method
In the finite element method used, a cohesive interaction between bodies can be defined using cohesive material properties along with damage initiation and propagation criterion. This interaction acts as a rigid bond until the specified damage initiation criteria is reached, at which point the cohesive zone begins to soften and its ability to carry load is decreased as dictated by the damage propagation criteria. There are two types of elements that can be used to model the cohesive zone: cohesive elements and cohesive surfaces. Cohesive elements are used when the cohesive layer to be modeled has a finite thickness, while cohesive surfaces are best suited for bond-lines that have essentially zero thickness. The latter, cohesive surfaces, are therefore chosen to model the ice/leading edge interface of the pneumatic de-icing system.
Often, the cohesive material properties that dictate the onset of damage (ulti- mate failure strengths in Mode I and Mode II/III) between the bonded substrates are unknown and must be experimentally determined. To determine the ultimate interfacial strengths to be used in finite element models, physical experiments capable of capturing Mode I, Mode II/III cohesive failure response were conducted. Typically to acquire Mode I failure data, a double cantilever beam test (see Figure 2.10) is used where the two bonded substrates are "peeled" apart until cohesive failure is achieved. To measure ultimate Mode II/III failure behavior, a lap joint shear test
(see Figure 2.2) can be used. Once both Mode I and Mode II/III failure data is gathered, a finite element model matching each physical experiment is built and run in an iterative process, changing the ultimate failure strength of the model until the
57 predicted response matches that of the experiment. This process is known as model calibration and is essential to ensuring accuracy of finite element models using the cohesive zone method. It is important to note that a mesh convergence study to de- termine element size required for accurate stress/strain response of the structure must be conducted prior to beginning model calibration, as failing to do so will produce inaccurate results. The mesh size that is determined by the convergence study and that is ultimately used for model calibration must remain constant throughout the calibration process, as the response of the cohesive zone is highly dependent on mesh density as discussed in Ref. [19]. In other words, different mesh element sizes will yield different cohesive delamination results. The dependence of cohesive zone finite element models is illustrated in Figure 2.11 from Ref [19]. Therefore, once a model containing cohesive surfaces calibrated, the mesh density used in that model must be applied to all models containing the calibrated cohesive properties thereafter. The described calibration process was conducted in Ref. [16] and the determined cohesive properties are listed in Table 2.3. The mesh element sizes used for the calibration process are listed in Table 2.4.
58 Figure 2.11: Dependence of cohesive zone method results on mesh size from Ref [19].
Table 2.3: Cohesive interface properties of ice to leading edge cap as measured in Ref. [16]
Abaqus Option Interface Properties Values Elastic Cohesive surface modulus (penalty stiffness) 1x1012 (N/m3) type=traction in all three directions Knn,Kss,Ktt
Ultimate strength in mode I, T 1.4x106 (N/m2) Damage initiation, Ultimate strength in Mode II, S 1.95x106 (N/m2) criterion=quads Ultimate strength in Mode III, N 1.95x106 (N/m2)
Mode I fracture, G 1 (N/m) Damage evolution, IC Mode II fracture, G 2 (N/m) type=energy IIC Mode III fracture, GIIIC 2 (N/m)
Mixed mode Power law, α 2 behavior
59 2.5.3 De-icing Performance Modeling Process Overview
The process used to predict a pneumatic de-icing systems ability to remove ice from its surface when pressurized is summarized as follows. Firstly, a representative ice shape is created using LEWICE and imported to the Abaqus finite element model of the de-icing system. The ice shape is then constrained to the leading edge surface using cohesive failure properties. Pressures created by centrifugal pumping are applied to the model and the percentage ice/leading edge interface bond that has failed due to system actuation are outputs of the model. This percentage is then compared to the percent bond-line required to be failed by an ice shedding model developed for this research. If the percent bond-line delaminated as predicted by the Abaqus model is greater than that required by the shedding model, then the de-icing system has shed the entire accreted ice shape. This process is summarized in Figure 2.12 and is covered in greater detail in the following sections.
60 Figure 2.12: Process followed to predict de-icing system performance for a given pneumatic pressure.
2.5.3.1 LEWICE
LEWICE is a computer code developed by former NASA Lewis (now NASA Glenn) for the purpose of predicting ice accretion to 2-D airfoils and other bodies [50]. The ice accretion model in LEWICE uses a time stepping potential flow field method to predict incoming super-cooled water droplet trajectories and impingement limits on the body. An analytical ice accretion model that calculates the thermodynamic processes between impinging droplets and the body is used to determine ice shape growth. At the beginning of each time step, the body geometry is updated to contain
61 the ice that has accreted in the previous step, allowing the ice shape to "grow".
By specifying icing parameters such as LWC, MVD, icing time, and ambient air temperature, the user can easily create ice shapes representative of those created during in-flight icing encounters. In this research, LEWICE was used to generate predicted ice shapes representative of those accreted during prototype rotor ice testing
(see Chapter 3). The predicted ice shapes were used in the de-icing system modeling approach. An example of an ice shape created in LEWICE for de-icing system modeling can be seen in Figure 2.7.
2.5.4 Shedding Model
A model capable of predicting ice shedding events due to centrifugal forces acting on an ice shape was created to assist with pneumatic de-icing system analysis. The model uses ice shape geometry values that are read in from LEWICE output files to calculate the volume, mass, and adhesion area of the ice shape. Rotor speed, location of the ice shape along the blade span, and rotor radius are then used to calculate the centrifugal forces acting on the ice. If the shear adhesion strength of the ice to the blade is less than the shear pressure created by centrifugal forces acting on the ice, then the ice sheds. A scehmatic illustrating the forces acting on an ice shape is provided in Figure 2.13.
62 Figure 2.13: Forces acting on ice shape along a blade section [2].
To account for ice delamination created by the pneumatic de-icing system, the ice adhesion area contains a modification parameter that represents a percent reduction in ice/leading edge bond-line length.If the reduction in area was large enough to allow the centrifugal forces acting on the ice to overcome the now reduced shear adhesion forces of the ice to the blade, then the ice shape was shed. By varying the percent bond-line failure parameter, the percent bond-line remaining after pneumatic de-icing system actuation to result in total ice shape shedding is calculated. The calculated bond-line failure is then compared to the delamination predicted by the Abaqus model. In Figure 2.14, results from the shedding model for ice shapes representative
63 of those produced in prototype rotor ice tests (discussed in Chapter 3) are presented.
The results in Figure 2.14 represent the maximum allowable percent bond-line that can remain after pneumatic de-icing system actuation for the entire ice shape to shed.
Note that the model predicts that a very large (up to 99% bond-line length) must fail for the ice shape to shed under centrifugal loads, however these ice shapes are quite small (less than 0.1 in. thick), so this result is expected due to their low mass.
Figure 2.14: Maximum percent bond-line remaining allowable for total ice shape shedding.
64 Table 2.4: Mesh element size used for calibration in Ref. [16]
Part Element Length (mm) Leading Edge Cap 0.25 Ice 0.86
2.5.4.1 Abaqus Implementation and Results
The cohesive zone method was applied to the Abaqus finite element models of the de-icing system discussed in Section 2.4. Ice shapes created in LEWICE were tied to the leading edge cap surface using the cohesive properties measured by Bailey in
Ref. [16] which are listed in Table 2.3. Eight noded, quadratic, plane stress elements were used to mesh the leading edge cap and ice shape. The size of the mesh elements used in the model were the same as those used in Ref. [16] to determine the cohesive failure properties during model calibration. These element sizes are listed in Table
2.4. Note that was important to verify that the element sizes in Table 2.4 be less than or equal to the element size required by the mesh convergence study discussed in Section 2.4.5 (see Figure 2.9).
Pneumatic pressures applied to the inner surface of the leading edge cap of the de-icing system were ramped from 0 psi to 8 psi (recall, a full-scale rotor can generate
±4 psi for a net pressure differential). Ramping the pressures in this fashion allowed the entire range of pressures available via centrifugal pumping on a full scale rotor blade to be covered in one analysis run. During model post-processing, a path was created along the ice/leading edge interface for the purpose of gathering cohesive surface separation data (see Figure 2.15).
65 Figure 2.15: Path defined in Abaqus to gather cohesive surface separation data.
In Abaqus, Mode I separation between nodes of a cohesive surface is captured by the variable COPEN, while Mode II/III nodal separation is captured by the CSLIP variable. The COPEN and CSLIP variables for each desired pneumatic pressure were output and plotted over the normalized path wrap distance (i.e., the normalized length ice/leading edge bond-line length). Figure 2.16 illustrates the nodal separations created when Prototype II was pressurized to 5 psi. To illustrate the manner in which the delamination grows as pneumatic pressure is increases, Figure 2.17 is presented.
The results shown in Figure 2.17 are as expected, as the length of the failed bond-line increases as pressure is increased. In Figure 2.16, the solid horizontal line represents the minimum nodal separation required for mixed-mode cohesive element failure which was calculated to be 6.88 µm using the process described in Section 2.5.1. Any elements within the region labeled "Delamination Region" have completely failed.
Since the pneumatic de-icing system does not impart single mode failure, but rather mixed-mode failure, it is appropriate to use the magnitude of the Mode I and Mode
II/III separations to determine whether or not delamination has occurred. The
66 magnitude of the Mode I and Mode II/III separations is represented as the total separation in Figure 2.16. It should also be noted that the absolute value of Mode
II/III failure is presented, since only the magnitude of this parameter is of important.
Figure 2.16: Sample delamination results from Abaqus model.
To better visualize the effect of pneumatic pressure on ice bond-line failure, Figure
2.18 is provided. These particular curves show the effects of pneumatic pressure on
Prototype II delamination performance for three ice thicknesses (Case 1 = 0.080 in.,
Case 2 = 0.097 in., Case 3 = 0.133 in.). The thicknesses of the ice shapes used for modeling purposes were created to match those produced during prototype rotor ice testing and can be found in Table 2.5. In Figure 2.18, the knees in the curves represent the critical pressure which produces a rapid bond-line failure, and is taken as the pressure required for total ice shape delamination. For the case presented, the predicted pneumatic pressure required to produce total ice shape shedding is
67 about 9 psi for Case 1 (0.08 in. max ice thickness), 5 psi for Case 2 (0.097 in. max ice thickness), and 4 psi for Case 3 (0.133 in. max ice thickness) or a decrease in pneumatic pressure required of 55% from Case 1 to Case 3.
Figure 2.17: Effect of increasing pneumatic pressure on ice bond-line damage propa- gation.
Figure 2.18: Predicted delamination performance of Prototype II de-icing system for three ice shape thicknesses.
68 2.6 Effects of Aerodynamic Pressures
The centrifugally powered pneumatic de-icing systems presented in this chapter rely on deformation of the metallic leading edge cap to promote delamination of accreted ice. The fact that low pressures (less than 8 psig) are able to produce sufficient deformation to debond ice raises a concern that aerodynamically produced pressures that act on the blade surface may deform the leading edge cap an unacceptable amount during non-icing conditions. A method to predict the effect of aerodynamic pressures on leading edge cap movement was developed using a combination of Abaqus
finite element software and XFOIL, a development and analysis software for 2-D subsonic airfoils.
2.6.1 XFOIL
XFOIL is an open-source, interactive software created at MIT in the 1980’s for the purpose of developing and analyzing 2-D, subsonic airfoils using panel methods. The user specifies a text file containing 2-D airfoil x-y coordinates, or can chose one of several pre-loaded NACA airfoils already contained within the program. By specifying
Reynolds number, Mach number, and angle of attack, pressure distributions over the airfoil surface are calculated. XFOIL has many additional features that were not used in this research and are therefore not mentioned here, but are covered in Ref. [51].
2.6.2 Mapping Surface Pressures in Abaqus
To study the effects of aerodynamic pressures on leading edge cap deformation during non-icing conditions, pressure distributions representative of those encountered at
69 the tip of a full-scale rotor blade were obtained using XFOIL. Dimensions for the full-scale rotor used were consistent with those of a Kaman K-MAX blade, whose radius is 24 ft. and rotates at 280 RPM. Angles of attack, α, ranging from 0 degrees to 10 degrees were studied for a 16 in. chord NACA 23012 airfoil. For this study,
ft the tip speed of the full-scale K-MAX rotor (700 s ) was used to calculate pressure distributions, since this blade location will produced the highest pressures. Pressure coefficient, Cp, distributions produced by XFOIL for each case were converted to dimensional form using 1 P = ρv2C (2.6) 2 p where v is the rotor tip speed and ρ is the density of air at sea level standard conditions.
Pressure coefficient distributions produced by XFOIL for are shown in Figure 2.19.
ft Figure 2.19: Pressure distributions for a NACA 23012 airfoil at 700 s .
70 The pressure distributions produced by XFOIL were then mapped to the surface of the de-icing systems metallic leading edge cap in the Abaqus model. To do this in Abaqus, an analytical field was created for each α which contained the geometric location along the cap surface for each data point, and the dimensional pressure value at that location. These analytical fields were then used to create a load distribution on the leading edge cap. An example of the pressure distribution when applied to the model is shown in Figure 2.20. In addition to the aerodynamic pressure acting on the outer surface of the leading edge cap, a suction pressure equal to that created by centrifugal pumping at the blade tip was applied to the inner cap surface. This set-up represents the configuration of the de-icing system at the blade tip during non-icing conditions when the cap is held to the blade to maintain its aerodynamic shape.
Figure 2.20: Example of pressures applied to the Abqaus model for the aerodynamic pressure study. The pictured pressure distribution is for α = 10◦.
71 2.6.3 Results
Results from the aerodynamic pressure study revealed that the negative pressures produced by centrifugal pumping at the blade tip are not sufficient enough to counteract the leading edge cap deformation induced by aerodynamic pressures. As expected, the deformation of the cap increases with angle of attack as illustrated in
Figure 2.22. For α=0◦, the maximum cap deformation is 1.175 mm and increases to
1.9 mm at α = 10◦. Note that the deformation plotted in Figure 2.22 is the magnitude of the maximum cap deflection and therefore the cap could be either pulled away from the blade or compressed into the blade as seen in Figure 2.22. In Figure 2.22 the maximum deformation occurs on the cap upper surface, however the cap is also compressed towards the blade on the lower surface.
Figure 2.21: Leading edge cap deformation profile caused by aerodynamic pressures at α = 10◦. Note that units are in meters.
72 Figure 2.22: Magnitude of maximum leading edge cap deformation due to aerodynamic pressures.
2.7 Mitigation of Leading Edge Cap Deformation
Due to Aerodynamic Pressures
Prediction of the effects of aerodynamic pressures on leading edge cap movement was discussed in Section 2.6. Results from this modeling showed an un-desirable amount of deformation in the metallic leading edge cap at angles of attack above 5 degrees.
A parametric study was conducted to determine whether or not the initially chosen cap thickness (0.03 in) was optimal in terms of stiffness to resist deformation from aerodynamic pressures yet flexible enough to maintain the de-icing systems ability to promote ice delamination.
Leading edge cap thicknesses ranging from 0.03 inches to 0.05 inches were examined for angles of attack from 0 to 10 degrees. New part geometries were created for each
73 leading edge cap in Abaqus. Using the method described in Section 2.6.2, pressure distributions created in Xfoil for the NACA 23012 were mapped to the blade surface.
The maximum negative pressure supplied by centrifugal pumping at the blade tip (-3 psig) was applied to the model to oppose the aerodynamic pressures, reproducing the conditions that would be encountered in flight. For each case the magnitude of the maximum deflection for each cap thickness and pressure distribution were calculated.
Results from this study are shown in Figure 2.23. As expected, increasing angle of attack creates larger maximum cap deflections for all thicknesses. For the 0.03 inch thick cap, a maximum deflection of 1.91 mm is created at 10 degrees angle of attack, while the remaining two thicker caps have very similar deflections of 1.57 mm and
1.60 mm for the 0.04 inch cap and 0.05 inch cap respectively. The most significant reduction in leading edge cap deformation (1.16 mm.) occurs at 0◦ angle of attack when the leading edge cap thickness is increased from 0.03 in. to 0.05 in. It is clear from Figure 2.23 that there is a noticeable gain in cap stiffness from 0.03 inches to
0.04 inches, but very little from 0.04 inches to 0.05 inches. For this reason, the 0.04 inch thick cap was chosen to continue with in the parametric study.
74 Figure 2.23: Effects of aerodynamic pressures on leading edge cap deflection.
To study the ability of the pneumatic de-icing system to delaminate accreted ice with a 0.04 inch thick leading edge cap, two Abaqus ice shape parts representative of those formed in prototype rotor ice testing (discussed in Chapter 3) were created using LEWICE. Each ice shape part was constrained to the leading edge cap in the
Abaqus finite element model using cohesive surface interactions. For convenience, the ice thicknesses used in this analysis are given in Table 2.5. Using the modeling process described in Section 2.5, the performance of the de-icing system with the increased thickness cap was studied for each ice thickness and compared to the de-icing system with a 0.03 inch thick cap. Results are shown in Figure 2.24. In Figure 2.24, only one of the configurations delaminates completely within the examined pressure range
(Case 2: 0.03” Cap). For the Case 1 ice condition, increasing the leading edge cap thickness from 0.03 inches to 0.04 inches results in a reduction of percent bond-line failed at 4 psig of 13% (note that the pressures used for comparison were arbitrarily
75 selected for demonstration purposes). For the Case 2 ice condition, the leading edge cap thickness increase results in a reduction of 16% in percent ice bond-line failed at
4 psig.
Table 2.5: Ice shape thicknesses used for parametric study.
Ice Thickness (in) Case 1 0.078 Case 2 0.097
Figure 2.24: Comparison of pneumatic de-icing system performance for increased cap thickness.
It was noticed during the aerodynamic pressure modeling, that not all of the leading edge cap deflection was due to deformation of the metallic cap itself, some of the deflection was actually caused by translation of the cap due to deformation of the elastomer in the aft portion of the de-icing system. In light of this observation, the effects of increasing the stiffness of this elastomer was investigated. To do this, the
76 stiffness of the uniaxial stress-strain test data used to define the elastomer material in the Abaqus finite element model was increased by both 20% and 40%. The original and modified test data used are shown in Figure 2.25. Finite element models for leading edge cap thicknesses of 0.03 inches and 0.04 inches were studied for angles of attack ranging from 0 to 10 degrees.
Figure 2.25: Uniaxial test data used to model elastomer in Abaqus finite element model (original data adopted from Ref. [18]).
The effect of increasing elastomer stiffness on leading edge cap deflection due to aerodynamic pressures are shown in Figure 2.26. These results show that while increasing the stiffness of the elastomer does help mitigate the adverse effects of aerodynamic pressures on the deflection of the cap, these gains are small. At α = 0◦, the largest decrease in cap movement (6.9%) is observed when increasing elastomer stiffness by 40%. At α = 10◦, a decrease of 3.8% is noticed when stiffness is increase by 40%. There is no benefit at α = 5◦.
77 Figure 2.26: Effect of increase elastomer stiffness on leading edge cap deflection due to aerodynamic pressures.
Observing that increased elastomer stiffness did not result in a significant reduction of leading edge cap deformation created by aerodynamic pressures, it was decided to examine the effects of replacing the elastomer with a thin span-wise metallic rib.
Two metallic rib thicknesses were examined, 0.005 in. and 0.01 in., and were modeled as 304 Stainless Steel. The ribs were tied to the top and bottom blade surfaces and to the inner surface of a 0.03 in. thick cap in exactly the same fashion as the elastomer. Aerodynamic pressures representative of those created at α = 0, 5, and
10◦ were mapped to the leading edge cap surface while suction pressures produced by centrifugal pumping was applied to the inner cap surface. Results from this study are compared to earlier model configurations for comparison in Figure 2.27.
78 Figure 2.27: Leading edge cap deformation due to aerodynamic pressure for several elastomer stiffnesses and materials.
Referring to Figure 2.27, the largest reduction in leading edge cap movement is achieved at α = 0◦ when the elastomeric material is replaced with a less compliant 304
Stainless Steel material. As expected, by increasing the Stainless Steel rib thickness, the amount of cap movement decreases. By replacing the original elastomer with
0.01 in. thick 304 Stainless Steel, the leading edge cap deformation is reduced by
86%. However, at α = 5◦, the amount of cap deformation is increased for this same configuration by 13%. For α = 10◦, the switch to Stainless Steel from elastomer again provides a decrease in cap movement, but only by 12%.
Since the ability of the de-icing system to remove accreted ice is reliant on leading edge cap deformation, the effect of changing materials and leading edge cap thickness on the performance of the de-icing system must be studied. To do this, the Abaqus ice delamination model was run for each system configuration examined in Figure
79 2.27 using the Case 1 ice shape. To be clear, only the parameter listed in the legend of Figure 2.27 were changed for that configuration, all other parameters were as in the original configuration, i.e, elastomer stiffness, leading edge cap thickness, etc.
Results from this study are presented in Figure 2.28.
Figure 2.28: Comparison of the effect of various parameters on the ice delamination performance of the de-icing system for Case 1 ice.
The configuration that replaces the elastomer by a 0.01 in. thick 304 Stainless
Steel rib, which provided the best ability to resist aerodynamic cap deformation, is the worst performing de-icing system of the group. Results from this study show that all forms of modification to the system sealing material (i.e., replacing the elastomer with thin stainless steel) along with increasing leading edge cap thickness degrades the de-icing performance of the system. The configuration which increases
80 elastomer stiffness by 40% from baseline stiffness follows the baseline configuration
(0.03” Cap) trend until 5 psig, where the rate of delamination increases. At 8 psig, the baseline configuration outperforms the configuration with increased elastomer stiffness slightly (by about 4%), making it the most desirable configuration in terms of de-icing performance.
Results from this parametric study suggest that replacing the elastomeric material with a metallic material, while providing significant increases in preventing unwanted leading edge cap deformation due to aerodynamic pressures, restricts movement of the de-icing system preventing accreted ice from being removed with the available system pressures. Increasing leading edge cap thickness has a similar effect on the performance of the de-icing system. The most promising configuration in terms of de-icing performance is provided by the baseline configuration which was the configuration fabricated and tested in Chapter 3. In terms of overall performance, increased elastomer stiffness reduces leading edge cap deformation due to aerodynamic pressures by 4% compared to the baseline configuration while providing identical delamination performance up to 5 psig and small penalties above this pressure. For this reason, increasing elastomer stiffness and using 0.03 in. thick Stainless Steel to construct the leading edge cap is recommended as a suitable compromise to improve the de-icing system. Potential configurations that could assist reducing the shape deformation introduced by aerodynamic forces will be discussed in the Future Work section of this thesis.
81 Chapter 3 | Prototype Pneumatic De-Icing System Testing and Finite Element Model Prediction Validation
3.1 Small-Scale Hover Icing Testing
The ultimate goal of this research was to design and test a centrifugally powered pneumatic de-icing system on a full-scale rotor. Once the two prototype de-icing sys- tems were designed and preliminary analyses were run, the next step was to fabricate and test the systems on a small scale to then be tested under representative icing conditions at the Penn State AERTS facility. Results from prototype testing would then serve as a means to select one of the two de-icing systems for implementation onto a full-scale rotor blade.
3.1.1 Description of the Test Facility
Penn States AERTS is a one-of-a-kind test facility that provides the capability to evaluate a wide range of de-icing and other rotorcraft systems [31]. A 20 x 20 x 11.5
82 ft industrial walk-in freezer creates the outer walls of the facility which can maintain temperatures ranging from 0◦C to -20◦C. Contained within the freezer is a 125 HP electric motor that drives up to 5 ft. radius rotor blades at rotational speeds up to
1500 RPM. A photograph of the facility is provided in Figure 3.1. The hub assembly from a QH-50 unmanned helicopter provides cyclic and collective pitch control to the blades, and a 6-axis load cell provides a means for real-time hub load monitoring during testing. The rotor is surrounded by a ballistic wall to provide protection in the case of blade structural failure and other ballistic concerns.
Figure 3.1: Photograph of the AERTS facility.
Two concentric rings of NASA Standard icing nozzles mounted in the freezer ceiling produce an icing cloud and can be seen in Figure 3.2. The 15 icing nozzles operate by aerosolizing pressurized air and water to produce super-cooled water droplets. By controlling the air and water pressure inputs to the nozzles to specific values as per nozzle calibration curves, water droplet size (MVD) can be controlled to create an icing cloud representative of atmospheric icing conditions. By turning
83 individual nozzles on or off, the icing cloud LWC can be controlled.
Figure 3.2: Ceiling view of the AERTS facility.
An electric slip rings provides 48 signal channels along with 24 electrical power channels to the rotating frame, while a pneumatic slip ring provides shop air pressure to the rotating frame to allow pneumatic de-icing systems to be tested.
3.1.2 Description of Test Blades and Prototype De-Icing Systems
Both prototype de-icing systems were installed on 12 in. span truncated, 16 in. chord, paddle sections of a Kaman K-MAX blade and mounted at the tip of a 36 in. radius carrier blade, as seen in Figure 3.1 and Figure 3.3. The design of both
Prototypes I and II matched those discussed in Chapter 2. Only one operational de- icing system of each design was constructed and each prototype was tested separately
(i.e., one blade held an operational de-icing system while the opposite blade held a
"dummy" paddle to balance the blades). Since the span truncated prototype blades
84 were not capable of producing the necessary pressures to match those produced by a full-scale blade (recall that centrifugally generated pressures increase with r2) a pneumatic slip ring was used to deliver the desired pressures to the de-icing system.
Figure 3.3: Photograph of Prototype I installed on the test blade with pneumatic suction (top) and pressure (bottom) applied.
85 3.1.3 Test Method
To evaluate the systems deicing capability, a test matrix was developed. The matrix reproduced centrifugally generated pressures experienced along the outboard half
(0.5 r/R to R) of a 24 ft. radius blade operating at 280 RPM. The radial pressures along the same blade (Kaman K-MAX rotor blade) were measured by Szefi et al. in
Ref. [15] and are shown in Figure 3.4. These pressures were used as the basis for testing, as the application of the pneumatic de-icing system to the full-scale K-MAX blade was the ultimate goal of this research.
Figure 3.4: Pressures produced by centrifugal pumping on a full-scale rotor blade at 280 RPM.
Icing conditions were selected to represent a severe atmospheric icing condition within the FAR Part 25/29 intermittent maximum icing envelope. A typical test condition in the AERTS facility, MVD = 20 µm. and LWC = 1.9 g/cm3, was selected
86 for both designs. The location of this point within the FAR icing envelope is shown for convenience in Figure 3.5.
Due to blade manufacturing limitations and blade imbalances upon shedding, the rotor speed tested was 450 RPM for Prototype I, producing a centrifugal acceleration equivalent to that seen at 0.5 r/R of the 24 ft. blade. Due to even larger blade imbalances, rotor speed for Prototype II testing was 385 RPM. Note that testing at reduces RPM is a conservative approach, since there will be reduced forces assisting with the ice removal process.
Figure 3.5: Rotor ice testing condition location in FAR intermittent icing envelope.
For prototype icing tests, blade spool up to the desired RPM (450 for Prototype I and 385 for Prototype II) was done with negative air pressure supplied to the de-icing system. The negative air pressure values used for each test condition can be found in
Table 3.1 and Table 3.2. The negative air pressures acted as a means to maintain the aerodynamic shape of the blade by holding the leading edge cap to the blade surface until the de-icing system was activated. When the desired rotor speed was reached,
87 the icing cloud was turned on for a pre-determined length of time which would accrete ice thicknesses that were believed the de-icing system could shed. At the end of icing time, the cloud was turned off and the deicing system was then cycled between the desired negative and positive air pressures to delaminate accreted ice. The frequency that the de-icing system was cycled at was about 0.2 Hz., this equated to the time it took for the system to fully inflate or deflate. Once ice shedding occurred, or it was decided it would not occur for a current test current condition, the rotor was spooled down and the ice shed area was examined. Initially, only colder temperatures within the icing envelope were to be tested, however a warmer temperature was added to the test matrix later on to study how temperature affects de-icing system performance.
A flow chart summarizing the testing process is provided in Figure 3.6.
3.1.4 Results
In Table 3.1, a testing summary for Prototype I is given and lists the average ice thicknesses required for shedding at input pressures representative of those produced by centrifugal pumping at various radial locations along a 24 ft. K-MAX blade. The pressures and suctions applied were determined from Figure 3.4 to reproduce pressure conditions at outboard radial locations along a full-scale blade. As expected, the minimum ice thickness required for shedding increases with decreasing radial location along the blade due to a decrease in available pressure and centrifugal force. A sample photograph of the first prototype before and after system actuation is shown in Figure
3.7.
88 Figure 3.6: AERTS rotor ice testing procedure.
Table 3.1: Prototype I rotor ice testing summary.
Temp Pressure Suction Ice Thick. K-MAX r/R (◦C) (psig) (psig) (in) 1 -14 4.5 -3 0.25 0.9 -14 4 -3 0.28 0.7 -14 2 -5 0.32 0.5 -14 1 -5 0.39
89 (a) Ice accretion along blade span prior to de-icing system actuation.
(b) Ice shedding induced by de-icing system actuation.
Figure 3.7: Prototype I before and after de-icing system actuation.
90 Results from testing of Prototype II are summarized in Table 3.2. Glaze ice regimes can be recreated at temperatures warmer than -10◦C as noted by Han in
Ref. [52]. In this regime, ice has a lower adhesion strength [1]. The failure mode of the pneumatic deicing system relies on transverse shear stresses to debond and fracture accreted ice. Therefore, it was expected that the system would perform better at colder temperatures where ice is less compliant and will more readily fracture when deformed. Rotor ice testing results from Prototype II are shown in Table 3.2. At low temperatures, Prototype II was able to remove ice 0.078 in. thick with centrifugal pressures corresponding to those seen at the tip of a 24 ft. radius blade, and with centrifugal forces acting on the ice representative of only 50% radius. It should be noted that Prototype II was not tested at centrifugal pressures representing a K-MAX r/R of 0.5 and -14◦C. However, the required ice thickness for shedding could be estimated from the linear trend of the preceding data points.
Table 3.2: Prototype II rotor ice testing summary.
Temp Pressure Suction Ice Thick. K-MAX r/R (◦C) (psig) (psig) (in) 1 -14 4.5 -3 0.078 0.9 -14 4 -3 0.097 0.7 -14 2 -5 0.133 0.5 -14 1 -5 – 1 -5 4.5 -3 0.11 0.9 -5 4 -3 0.15 0.7 -5 2 -5 0.17 0.5 -5 1 -5 0.2
91 (a) Ice accretion along blade span prior to de-icing system actuation.
(b) Ice shedding induced by de-icing system actuation.
Figure 3.8: Prototype II before and after de-icing system actuation .
92 Figure 3.9: Effects of temperature on Prototype II de-icing capabilities.
The effects of temperature on the ability of Prototype II to delaminate accreted ice can be seen in Figure 3.9. Prototype II was test at -14◦C and -5◦C, and counterin- tuitively, the de-icing system performed better at lower temperatures. The fact that the minimum ice thickness required for ice shedding increased by 43% on average from
-14◦C to -5◦C was attributed to the fact that the de-icing system relies partially on the accreted ice fracturing caused by leading edge cap pressurization. As temperature increases, ice stiffness decreases, and therefore the accreted ice is able to flex with the leading edge cap as it deforms without fracturing.
To compare the performance of the two prototype deicing systems, the minimum ice thickness required for delamination was normalized by g’s of centrifugal acceleration experienced at 90% span of the 3 ft. carrier blade. The normalization was done
93 because Prototypes II and II were not tested at the same rotor speed. The minimum ice thickness required per g is illustrated in Figure 3.10. Comparing results from
Prototype I and II, the ice thickness required for ice delamination for Prototype II is
58% lower than that of Prototype I, for pneumatic system pressures equivalent to those generated at tip of a 24 ft. rotor blade. A sample photograph of Prototype
II before and after system actuation is shown in Figure 3.8. Due to the significant increase in de-icing system performance from Prototype I to Prototype II, Prototype
II was chosen as the final design that would be integrated onto a full-scale rotor blade for hover ice testing.
Figure 3.10: Comparison of Prototype I and II minimum ice thickness required for delamination at -14◦C.
94 3.2 Comparison of Pneumatic De-icing system Model and Experimental Test Results
In Chapter 2, a process to predict the ability of a pneumatic de-icing system to delaminate and shed accreted ice was presented. The described prediction process applied cohesive zone methods commonly used to predict composite delamination to an Abaqus finite element model of the de-icing system ice/leading edge interface.
Outputs from the Abaqus analysis provided the percentage of the ice/leading edge bond-line that had failed due to system pressurization. Using the ice shedding model, along with the calculated bond-line failure from Abaqus, whether or not the ice shape was shed could be predicted. As a first check, the deformed shape of the prototype de-icing system when pressurized was compared to the Abaqus predicted shape and is shown in Figure 3.11.
Figure 3.11: Comparison of Abaqus predicted and prototype measured de-icing system deformation with 4 psi pneumatic pressure input.
95 To compare the described modeling process to experimental results, ice shapes with thicknesses that matched those encountered during Prototype II rotor ice testing in the AERTS facility were created using LEWICE. Ice shape thicknesses for each validation case are listed in Table 3.3. Using each ice shape case, the Abaqus delamination model was run by ramping system pneumatic pressure until the required percent bond-line failure was achieved, or the ice delaminated completely. Results from the Abaqus delamination model are presented in Figure 3.12.
Table 3.3: Thickness of ice shapes from Prototype 2 rotor ice testing at -14◦C and -5◦C used for model validation.
Temp. (◦C) Ice Thick (in) -5 0.112 -5 0.150 -5 0.177 -5 0.20 -14 0.078 -14 0.097 -14 0.133
In Figure 3.12 it can be seen that the bond-line delaminates in a quadratic manner until a critical system pneumatic pressure is reached, upon which the bond-line then undergoes a rapid delamination to failure. Recall that the proposed modeling process described in Chapter 2 took advantage of the reduction in ice adhesion area created by delamination to assist in shedding the ice shape. In Figure 3.13, the required percentage bond-line failure induced by pneumatic de-icing system actuation for each ice shape examined is presented.
96 (a) Ice shapes from -14◦C tests.
(b) Ice shapes from -5◦C tests.
Figure 3.12: Abaqus predicted delamination for Prototype II for ice shapes produced at -14◦C and -5◦C during AERTS rotor ice testing.
97 Notice that only a very small portion of the bond-line (as low as 1% for ice thicknesses less than 0.1 in.) was predicted by the shedding model to be allowed to remain after de-icing system pressurization for entire ice shape to shed. This is as expected since centrifugal forces acting on the thin ice shapes are low. Also notice that when comparing the results in Figure 3.13 to the Abaqus delamination results in Figure 3.12, the ice shape begins a rapid bond-line failure long before the required percent bond-line failure for ice shedding is reached. Therefore, it is concluded that the inclusion of shedding in the modeling process is not necessary for the thin ice shapes that the pneumatic de-icing system is capable of removing.
Figure 3.13: Maximum percent bond-line remaining allowable for total ice shape shedding.
Results from the prediction comparison runs are compared to AERTS rotor ice testing results in Figure 3.14. In both cases, the Abaqus delamination model over predicts the pneumatic pressure required to delaminate the same ice shapes that were successfully removed during experimental rotor ice tests. The linear trend of
98 the -14◦C case matches the experimental linear trend with a 46% error compared to experimental data while the -5◦C case linear trend yields a 300% error.
(a) Ice shapes from -14◦C tests.
(b) Ice shapes from -5◦C tests.
Figure 3.14: Comparison of experimental and Abaqus predicted minimum ice thickness required for shedding for ice shapes produced during AERTS rotor ice testing.
99 The large error between experimental results and Abaqus model results is at- tributed to the exclusion of brittle ice fracture from the model. It was noticed during prototype rotor ice testing that the entire ice shape rarely shed, but instead ice was left behind on the blade top and bottom surfaces as seen in Figure 3.15.
Figure 3.15: Ice left behind after de-icing system pressurization.
Figure 3.16: Cracks formed in accreted ice by de-icing system pressurization.
100 It is hypothesized that ice that accretes to the leading edge cap surface while the system was un-pressurized fractures when the de-icing system is pressurized. Evidence of crack formation be seen in Figure 3.16, where the ice de-icing system had been pressurized but the accreted ice did not shed due to its thickness being too small.
The ice fracture along with delamination and centrifugal forces act to shed the ice.
It is therefore recommended that future modeling efforts of the pneumatic de-icing system include brittle fracture of ice to fully capture the physics of the problem.
101 Chapter 4 | Portable Icing Cloud Generator Design, Construction and Testing
4.1 Background and Motivation
Currently, rotor ice testing of a full-scale blade requires the use of one of two facilities:
The Helicopter Icing Spray System (HISS) or McKinley Climactic Laboratory. A third option would be to test the vehicle in natural adverse weather conditions.
Mentioned approaches are cost prohibited for prototype development. In addition,
flight testing of de-icing system presented in this thesis would prove hazardous due to the early prototypical stage of the technology. For these reasons, a portable icing cloud generator was developed, as part of this research, to allow testing of full-scale blades on a rotor blade whirl tower at a reduced cost. This chapter describes the design and operation of the developed portable icing cloud system.
102 4.2 Icing Cloud Generator Design
The design of the portable icing cloud system was modeled closely from that of Penn
States AERTS facility, with the main differences being that it needed to be easily transportable and self-contained. Like the AERTS, the portable cloud system consists of water and air sub-systems that work together to provide air and water at specific pressures to NASA Standard icing nozzles. The icing nozzles produce the icing cloud via atomization. A 30 HP electric air compressor provides the required air pressure and flow rates to the nozzles, as well as pressurizing two 40 gal. hydro-pneumatic water tanks, which provide the pressurized water to the icing nozzles. The nozzles are mounted to the face of a 8 ft. diameter fan (in the downstream location) for full-scale rotor blade testing, allowing the air-flow from the fan to assist in carrying the cloud to the desired location. Selected nozzles instrumented with pressure sensors, along with electronic pressure regulators are used as inputs to a feedback control loop that maintains the desired MVD. Electronic air and water shutoff valves allow the icing cloud to be turned on and off remotely and are controlled by a custom created Lab
View code. A schematic of the cloud system is provided in Figure 4.1. The entire system, excluding the fan, was contained within a 4 x 10 x 8 ft. insulated, heated enclosure to prevent the water sub-system components from freezing during icing tests. The following sections discuss each sub-system of the cloud generator in further detail.
103 Figure 4.1: Schematic of the portable icing cloud generator. Note that red lines represent air and blue represents water.
104 4.2.1 Air System
A 30 HP air compressor provides the required air pressure and flow rates (15 CFM per nozzle at 50 psi) to the icing nozzles via a manifold that distributes air from the main airline to individual lines to each nozzle (see Figure 4.3). The main outlet from the compressor is split between the air and water sub-systems (see Figure 4.2), each side having its own electronic air pressure regulator (recall that the water sub-system uses air pressure to pressurize the water tanks). The air pressure regulator is used in a feedback control loop along with electronic pressure sensors on the nozzles and air manifold to maintain specific pressures as per the nozzle calibration curves. An electronic shut-off valve on the main airline allows the operator to turn air pressure to the nozzles on and off remotely. A bypass valve is used divert air through the water lines between icing tests to clear out excess water and prevent the nozzles from freezing.
Figure 4.2: Photograph of the icing cloud system air pressure control hardware.
105 Figure 4.3: Schematic of air and water distribution manifolds.
4.2.2 Water System
Water enters the system through a standard garden hose attachment and immediately is passed through a 20 micron filter. From there, water is passed through a water softener which removes 105 grains per gallon of hardness and impurities down to
9 ppm. A reverse osmosis water filtration system then purifies the water further, down to 1 ppm before it enters the hydro-pneumatic holding tanks. The mentioned components can be seen in Figure 4.4.
The water holding tanks are painted steel pressure vessels rated to 150 psi. Air pressure to the tanks is provided by the 30 HP air compressor which is also used to pressurize the air sub-system. A feedback control loop between water pressure sensors on the icing nozzles and an electronic air pressure regular at the tank inlet controls the tank pressure such that a specified water pressure, and therefore MVD, at the
106 nozzles is maintained. Each water holding tank is instrumented with an electronic liquid level sensor, which automatically turns off the reverse osmosis system pump when the tank is full, thus stopping the water flow through the purification system.
During operation, only one tank is pressurized, and the in-active tank is left open to atmosphere so it can be filled if needed. The exit of each holding tank is fed into a smaller tank containing a liquid level sensor which notifies the operator when the water tank currently in use is empty. From the holding tanks, the main water line is sent to a manifold which distributes water to each icing nozzle. The water manifold and two of the 12 icing nozzles (one on the outer and inner rings on the fan) are instrumented with electronic pressure sensors that are used in the pressure feedback control loop. An electronic water shut-off valve, which is activated via the
Lab View control software, allows the user to stop the flow of water from the tanks to the nozzles (i.e., turn the icing cloud on or off) remotely.
107 Figure 4.4: Photograph of the icing cloud generator water sub-system.
4.2.3 Icing Nozzles
Twelve NASA Standard icing nozzles are used to produce the icing cloud for the portable spray system. The nozzles operate by aerosolizing the air and water inputs, i.e., air pressure is used to break up the water stream to produce a spray. A cross- sectional schematic of a NASA Standard nozzle is provided in Figure 4.5. The nozzles are constructed from two concentric tubes. Water is input to the smaller inner tube
(from the left in Figure 4.5), while air pressure enters through holes around the perimeter of the outer tube wall to break up the water stream before the nozzle exit.
To deliver air and water pressure to each nozzle, cylindrical nozzle housings were designed and fabricated from 6061 Aluminum. The pressurized air chamber portion of the nozzle housing is formed by a small gap (0.01 in.) between the inner diameter
108 of the housing and the outer diameter of the nozzle, and is sealed on top and bottom by rubber o-rings. Air and water pressure are input to the nozzle through the housing walls via standard threaded pipe fittings as seen in Figure 4.6.
Figure 4.5: NASA Standard and Mod-1 icing nozzles. Standard nozzles were used in the portable cloud generator. Note that flow is from left to right [20].
Figure 4.6: Cross sectional view of nozzle housing.
109 To create a representative atmospheric icing cloud, nozzle air and water input pressures must be maintained to specific values as per the nozzle calibration curves provided by NASA. By controlling the air and water input pressures, water droplet size produced by the nozzles can be controlled to a desired size. The calibration curves for a NASA Standard nozzle is provided in Figure 4.7. In Figure 4.7, each curve represents line of constant air pressure, and the horizontal axis is the pressure differential between air and water. In practice, the operator selects the desired air pressure (15 psi was typically used for this research) and desired MVD, and the corresponding air/water pressure differential is then used to determine the required water pressure. The portable icing cloud system makes use of a feedback control loop to maintain the desired pressures.
Figure 4.7: NASA Standard icing nozzle calibration curve [20].
110 The nozzles were mounted in two concentric squares on the face of an 8 ft. diameter fan in the downstream location as seen in Figure 4.8. Each nozzle was individually selectable via shut-off valves located on the air and water manifolds, allowing the
LWC of the icing cloud to be adjusted to meet testing requirements.
Figure 4.8: Photograph showing location of the icing nozzles mounted on the fan.
111 4.2.4 Icing Cloud Controller
A custom Lab View code was developed to provide complete control of the icing cloud from a remote location. The controller allows the cloud to be turned on and off, water droplet size to be selected and maintained via a feedback control loop, and water holding tanks to be filled. In Figure 4.9 a screen shot of the icing cloud control panel is provided. Each section of the display is numbered and explained in this section.
Figure 4.9: Portable icing cloud system control panel.
112 1. Air pressure and droplet size selector. The top circular dial allows the user to
select the desired air pressure while the lower dial controls the MVD of the
water droplets. The required water pressure is determined via the calibration
curves in 2. and is displayed in the gage to the right of the MVD selection dial.
2. NASA Standard icing nozzle calibration curves. During operation, the selected
combination of air pressure, ∆P, and MVD is marked on the curve by a blue
square (see legend in bottom right of graph) and the actual is denoted by a
green circle. This output allows the user to know when cloud conditions have
stabilized and provides a means for ensuring the proper conditions have been
selected.
3. Main air and water pressure gages. These gages display the air and water
pressures as measured on the main lines by pressure transducers.
4. Icing cloud ON/OFF. These buttons control the electronic shut-off valves to
the main air and water lines as well as the bypass valve. A timer at the bottom
of this box controls a delay to opening the bypass after the main water valve is
commanded to close. The delay is set to the time required for the water valve
to close, ensuring water is not pushed through the air-lines.
5. Air and water pressure feedback control gains. Adjusting these values allows
the user to fine tune the behavior of the cloud.
6. Stop watch. Allows the user to easily monitor time the icing cloud has been
spraying.
7. Water tank status lights. These lights inform the user when the water holding
113 tank that is being used is empty. The tank full light and tank fill buttons are
used to begin filling an empty water holding tank and lets the user know when
the tank being filled is full. Once the tank fill light is illuminated, the water
pump is automatically shut-off.
8. De-icing system ON/OFF. This portion of the panel controls the pneumatic
de-icing system and the frequency that it will pressurize and de-pressurize.
9. Main air and water pressure regulator signals. These read-outs provide the user
a means of troubleshooting by displaying the voltage being sent to the air and
water pressure regulators by the feedback control loop.
4.3 Icing Cloud Generator Testing
Initial testing of the portable icing cloud system was carried out in the AERTS facility.
The goal of initial testing was to first tune the feedback control loop to provide proper cloud response and subsequently determine the operation limitations of the system.
For testing, the icing nozzles were mounted on horizontal bars and placed in the
AERTS freezer (see Figure 4.10).
Early in testing it was discovered that the airflow though the nozzles during non-icing conditions (when no water is being sprayed) was not sufficient to prevent the nozzles from freezing. This lead to the addition of electrothermal mat heaters being fitted to the outside of each nozzle housing and covered with metallic insulation.
Also added was a eletrcothermal rope heater and insulation along around the main water lines. Further testing proved that the mentioned modifications were successful in eliminating the freezing issue.
114 Figure 4.10: Initial testing of the portable icing cloud conducted in the AERTS facility.
The 30 HP air compressor was found to be more than capable of providing the necessary flow rates and air pressures (recall that each nozzle consumes 15 CFM of air at 50 psi). Icing cloud conditions successfully tested with all 12 nozzles operating are highlighted on the NASA Standard nozzle calibration curve in Figure 4.11. It should be noted that the pictured operation envelope was not exhaustively tested and does not necessarily define absolute system limits. However, for this research the envelope bounds tested were more than sufficient.
115 Figure 4.11: Portable icing cloud test envelope for successful continuous icing cloud.
For full-scale rotor ice testing, the fan with icing nozzles was raised to so that it was level with the rotor plane and positioned 2 ft. from the rotor blade tips.
This configuration took advantage of the rotor inflow to pull the icing cloud into the rotor plane and is depicted in Figure 4.12. The cloud system was proven to produce a representative icing cloud during full-scale testing, however insufficient electrical power to run the 30 HP air compressor necessary for all nozzles to operate was not available. Therefore, the number of nozzles that could be run at a time was reduced from 12 to 4. The reduction in nozzles significantly increased the required ice accretion time to an unacceptable level (LWC was greatly reduced), and the use of the portable spray system was abandoned for full-scale rotor ice testing. In place of the portable system, non-air assisted water pressure nozzles were adopted for the remaining tests. The spray nozzles produced large (200 µm), uncontrolled water
116 droplets, but cloud LWC was increased to an acceptable level (estimated 0.6 g/m3).
Full-scale rotor ice testing is discussed in greater detail in Chapter 5.
(a) Schematic of portable icing cloud during full-scale rotor icing tests.
(b) Photograph of portable icing cloud during full-scale rotor icing tests.
Figure 4.12: Schematic and photograph showing portable icing cloud in-use during full-scale rotor icing tests.
117 Chapter 5 | Full Scale Pneumatic De-Icing System Testing
In the previous chapters, the process of designing, analyzing and testing two centrifugally powered pneumatic de-icing prototypes was covered. The ultimate objective of these chapters was to yield a de-icing system design that would be scaled up to a full-scale rotor blade and tested under hover icing conditions. Unlike prototype testing at the Penn State AERTS, where pneumatic pressures were supplied to the de-icing system via a pneumatic slip-ring, pressures for the full-scale tests would be produced entirely by centrifugal pumping, thus demonstrating the capabilities of the newly developed technology. The process of modifying a full-scale Kaman K-MAX rotor blade to accept the de-icing system, along with full-scale icing testing results are presented in this chapter.
5.1 Full-Scale Blade Modifications
A retired, full-scale, 24 ft. Kaman K-MAX rotor blade was supplied to Penn State by Kaman to be fitted with the developed pneumatic de-icing system. Driven by promising prototype rotor ice testing results, the Prototype II design was chosen to
118 be installed on the outboard 8 ft. section (0.67R -R) of the full-scale blade. The location of the de-icing system is depicted in Figure 5.1 and Figure 5.2.
Figure 5.1: Schematic showing location of de-icing system as installed on a full scale rotor blade.
119 Figure 5.2: Photograph of modified de-icing blade mounted on the Kaman whirl stand for full-scale icing tests.
Standard K-MAX blades are fitted with a 0.03 in. thick metallic leading edge erosion cap above a 0.04 in. thick rubber layer. To maintain the original aerodynamic shape of the blade, the original erosion cap as well as the rubber layer were removed from the supplied blade and replaced with the de-icing system leading edge cap. As with the prototype design, the leading edge cap of the de-icing system was constructed of 0.03 in. 304 Stainless Steel, with the inner 0.015 in.of the cap segmented in the span-wise direction to create localized flexure points. Using a die, the leading edge cap was formed from one continuous metallic sheet into the shape of a NACA 23012 airfoil. Thirteen 6 in. x 4 in. x 0.005 in. 1095 Carbon Spring Steel sheets were bonded in an alternating fashion to the top and bottom surface of the blade in the span-wise direction as depicted in Figure 5.3. A close up photograph of the final full-scale de-icing system installed on the K-MAX blade is provided in Figure 5.4.
120 (a) Exploded, cross sectional view of Prototype II spring steel rib bonding pattern.
(b) Sectional view of Prototype II spring steel rib alternating bond pattern.
Figure 5.3: Schematic of Prototype II de-icing system.
121 Figure 5.4: Close up view of de-icing system installed on a full-scale K-MAX blade.
The spring steel ribs act as the centrifugal load carrying structure for the leading edge cap during blade rotation while not significantly increasing stiffness normal to the blade surface, which would impede leading edge cap movement during pressurization.
EPDM elastomer was bonded to the blade and inner cap surface to form a flexible seal on all edges. A total bond area of 282 in2 between the elastomer and spring steel ribs provided a factor of safety of 13 for the expected centrifugal loading condition.
Strips of 0.015 in. thick 304 Stainless Steel were bonded on top of the elastomer to reduce the likelihood of tearing as can be seen in Figure 5.1.
122 During bench-top testing, pressure was supplied to the system via a 0.165 in. diameter tube running along the blade leading edge and entered the system at the inboard cap edge as depicted in Figure 5.5. A 0.75 in. diameter low pressure line was run from the blade root to the blade mid-span via the inboard control rod channel
(see Figure 5.6). From here, a channel was milled in the blade top surface (and patched afterwards) from mid-span to the tip, allowing the 0.75 in. diameter hose to span the entire blade. The line that runs the entire span provides low pressure during rotation. A 24V microvalve located at the blade root was cycled at 0.07 Hz to open and close the 0.165 in. diameter line to atmospheric pressure. When open to atmosphere, centrifugal loads pressurize the air inside the line, thus inflating the deicing system. 10 mW of power and a 5V signal to the valve was supplied through an electrical slip-ring, and controlled by a data acquisition computer. A Hall sensor was installed near the inboard edge of the system as a method of monitoring the motion of the cap.
Figure 5.5: Schematic of pneumatic pressure lines as installed on the full-scale K-MAX blade.
123 (a) De-icing system low pressure line. (b) De-icing system micro valve.
Figure 5.6: Installation of pnematic pressure lines on full-scale blade.
5.2 Full-Scale Hover Icing Tests
As discussed earlier in this thesis, full-scale ice testing is cost prohibitive and currently must be conducted at either McKinley Climatic Laboratory, the Helicopter Icing Spray
System, or in flight natural icing conditions. A new approach to conduct full-scale hover ice testing was adopted for this research, which closely matched the method used at the Penn State AERTS. The modified full-scale K-MAX blade were tested at the Kaman Whirl Tower in Bloomfield, CT (see Figure 5.7) under representative icing conditions produced by the developed portable icing cloud generator(see Chapter
4). Unlike the Penn State AERTS, which makes use of a walk-in freezer to contain the rotor hover stand, the Kaman Whirl Tower is located outdoors, and for this reason it is only possible to conduct icing tests when ambient temperatures are below
124 freezing. The following sections describe the full-scale rotor ice tests of the developed centrifugally powered pneumatic de-icing system conducted at the Kaman Whirl tower.
Figure 5.7: Photograph of the Kaman Whirl Tower.
5.2.1 Test Method
Testing of the centrifugally powered pneumatic de-icing system was conducted at the Kaman Whirl Tower, located in Bloomfield, CT. during the month of February
(February 9 - 13, 2015). The portable icing cloud generator was also tested during this time, however due to power limitations, sufficient compressed air was not available at the whirl tower, thus reducing the number of nozzles that could be operated from twelve to four. For this reason the ability of the cloud generator to operate in a cold environment and produce a realistic icing cloud with measurable ice accretion
125 was demonstrated separately before testing the deicing system. In addition to using the portable spray system during initial testing, two water pressure lines producing
2400 psi each and equipped with a 45◦ angle non-air assisted aerosolizing spray nozzle at their tips were used to increase the liquid water concentration in the cloud, speeding-up the ice accretion process. The benefit of direct water pressure nozzles is that the large air flow rates needed for the NASA nozzles in the portable icing system is not required. For the purpose of this test, providing uncontrolled water droplet mixtures of super-large droplets (estimated to range between 40 µm to 400
µm) provides a conservative worst-case ice accretion scenario, since larger droplets at high liquid water concentrations will be generated. The larger particles size and water content will increase the ice accretion impingement limits. Large droplets were visually observed at the exit of the nozzle, and a reduction on droplet size was qualitatively observed at approximately 20 ft. from the water output. The water pressure lines were mounted on the ground providing sufficient distance for the reduction in size of the water droplets created. The cloud dissipation provided a more realistic (but unmeasured) droplet size to enter the rotor plane. The cloud
LWC was calculated to be 0.6 g/cm3 with an assumed MVD range of 40-200µm at the rotor plane. A photograph of the final setup is shown in Figure 5.8.
126 Figure 5.8: Final icing cloud configuration produced large, uncontrolled size water droplets.
Two different test methodologies for ice protection were explored with the pneu- matic de-icing system: continuous ice protection and single de-icing occurrence. For the continuous ice protection method, ice was allowed to accrete for a set amount of time with the de-icing system turned off. At set time intervals the de-icing system was turned on and cycled three times at 0.07 Hz., and subsequently turned off. This process was repeated throughout the entire test multiple times in an attempt to exer- cise the ability of the system to operate as it is envisioned during flight (intermittent operation post ice accretion for controlled time intervals). The single de-icing method tests consisted of accreting ice for a set time length with the de-icing system turned off followed by a 30 second period of cycling the system once the pre-determined icing time had elapsed. After the de-icing system was turned off, the rotor was also spun down. A schematic of the test process for single-shot deicing is provided in Figure
5.9.
127 Figure 5.9: Full-scale rotor ice testing procedure.
For both de-icing methods, system deployment was monitored using real-time time data provided by a Hall sensor. A sample of this output is displayed in Figure 5.10.
In the Hall sensor output, system inflation and deflation corresponds to motion of the of airfoil surface. An increase in voltage represents system inflation, while deflation causes a decrease in voltage. Selected icing times were determined by measuring
128 accreted ice thicknesses in early tests. Rotor speed during ice accretion was 250 RPM and was increased to 270 RPM for de-icing system deployment in the single shot de-icing tests. Rotor speed was held constant at 270 RPM (595 g’s at blade tip) for the continuous ice protection tests. Several tests were also conducted at a deicing rotor speed of 230 RPM (430 g’s at blade tip) to observe the de-icing systems ability to operate at lower input pressures and centrifugal forces acting on the ice.
Figure 5.10: Sample hall sensor output voltage showing de-icing system inflation and deflation during testing.
129 5.2.2 Test Results
Both test methods were successful at removing accreted ice and proved the ability to protect full-scale rotor blades with semi-passive centrifugally powered pneumatic de-icing systems. Selected successful de-icing test results are presented in Table 5.1.
Detached ice thicknesses measured as small as 0.08 in., and were successfully shed during the single shot deicing tests for a rotor speed of 270 RPM and at temperatures as low as -15◦C. Ice thicknesses for a rotor speed of 230 RPM were measured to be as small as 0.1 in. It is worth noting that the minimum thickness required for electrothermal deicing system operation is 0.3 in. A sample photograph of the protected result of semi-continuous ice protection are shown in Figure 5.11 and Figure
5.12.
Figure 5.11: Comparison of protected (right) and unprotected (left) blades after de-icing system test.
130 Table 5.1: Sample single-shot de-icing test results.
Temp. (◦C) Ice RPM De-Ice RPM Ice Thick (in) -10 250 230 0.1 -12 250 230 0.15 -14 250 270 0.08
Figure 5.12: Example of ice shape measured on unprotected blade.
The large droplets produced by the non-air assisted nozzles created ice shapes are more severe than what would be encountered in flight (FAR Part 25/29 icing envelopes specify a maximum intermittent droplet diameter of 55 µm). Impingement limits were noted to be further aft than those of standard icing conditions, creating a more difficult ice protection scenario for the deicing system. An example of the severe
131 ice shapes encountered during testing is illustrated in Figure 5.13. These severe ice shapes were successfully removed by the proposed pneumatic deicing system.
Figure 5.13: Severe ice shapes produced by large water droplets during de-icing system testing.
During icing tests, it was noticed that a small section of the protected blade consistently left residual ice after the deicing system had been inflated as seen in
Figure 5.14. It was determined that this portion of the cap was unable to deform completely and thus unable to create the stresses necessary for ice delamination. The restriction of cap movement in this area was attributed to unwanted adhesive bonding during system manufacturing.
132 Figure 5.14: Ice patch left behind after de-icing system test.
5.3 System Power Consumption
The final full-scale pneumatic deicing system provides significant weight and power savings compared to electrothermal deicing systems, which typically require 25 W/in2 for effective ice protection, while the tested centrifugally powered pneumatic de-icing system consumes 4 W/in2 at its peak power draw. Since the pneumatic system consists of only the modification to the leading edge cap and lightweight hoses, and does not require heavy non-rotating electrical components, its weight is comparable to that of the existing protective leading edge cap. The only electrical power draw of the pneumatic system is 10 mW to power the microvalve located at the blade root. Since the pneumatic de-icing system essentially acts as a centrifugal pump, a
133 mechanical power loss is introduced to the main rotor while compressing the air used to inflate the system. While de-pressurizing the system, the air volume within the leading edge cap is ejected from the blade in the radial direction from the rotor tip.
The flow rate of air from the blade tip creates a Coriolis force that acts as a drag on the rotor tip, which in turn produces an increase in required rotor power.
When the de-icing system begins its deflation process, a constant 5.5 psig pressure differential is present between the high and low pressure lines. Once the valve is cycled to deflate the system, the 5.5 psig pressure differential drives the flow out of the system until the leading edge cap is pulled to the blade surface and the volume of air in the cap is depleted. The change of the pressure within the system can be approximated using the differential form of the ideal gas law at constant temperature given by m˙ − pV˙ p˙ = M (5.1) V where V is the volume of air in the system, V˙ is the volume flow rate of air from the system, P is the air pressure, m˙ is the mass flow rate of air from the system, and M is the molar density of air.
The volume flow rate can be described using the Darcy-Weisbach equation of pipe
flow s πpd5 V˙ = (5.2) 8fDL where d is the diameter of the low pressure line (0.75 in.), L is the length of the pipe which the air must flow through, and fD is the Darcy friction factor which was taken to be 0.0189 for this calculation. Mass flow rate is given by the simple relation
m˙ = ρV˙ (5.3)
134 where ρ is the density of air at standard conditions. The decay of pressure from the de-icing system during deflation is presented in Figure 5.15. It is worth noting that the time required for the pressure in the lines to equalize (i.e., drop from 5.5 psig to
0 psig) is about 8 seconds, which is in agreement (with a 12% difference) with the measured time that was required by the de-icing system to fully deflate at during full-scale (recall that the system was cycled at 0.07 Hz. which equates to a period of
14 seconds and therefore 7 seconds for the system to deflate). Volume and volume
flow rates of the ejected air can be seen in Figure 5.16.
Figure 5.15: Equalization of pressure differential between high and low pressure lines during de-icing system de-pressurization.
135 (a) Volume of air within de-icing system leading edge cap.
(b) Volume flow rate of air from de-icing system.
Figure 5.16: Rate of air depletion from de-icing system leading edge cap during system de-pressurization.
136 The Coriolis force produced by the air flowing from the tip of the rotor blade acts as a drag at the blade tip and is given by
¯ ¯ Fc = −2mΩ × v¯ (5.4) where Ω¯ is the rotor rotational velocity of the rotor, v¯ is the velocity of the air being ejected, and m is the mass of the air. Power for a constant rotational rate is
Q = FcR (5.5)
P = ΩQ (5.6) where Q is the torque produced by the Coriolis force, and R is the moment arm (i.e., the rotor blade radius). Employing a numerical time integration of the presented relationships, the instantaneous power drawn from the rotor was calculated and is shown in Figure 5.17.
At its maximum flow rate during system de-pressurization, the de-icing system draws a maximum power from the rotor of 4.15 HP and decays to zero once the air volume in the leading edge cap has depleted. Note that this result is for only one blade fitted with the de-icing system, and therefore would be doubled if the de-icing system was installed on both blades. Using the same flow rates for inflation as were used for the deflation process (see Figure 5.16), the increase in power created by centrifugal pumping to create the 5.5 psig pressure differential can be calculated using
137 the power equation for centrifugal pumps
hgρV˙ P = (5.7) η where h is the pressure differential measured in feet of head, g is gravitational acceleration, ρ is the density of air, V˙ is volume flow rate, and η is the pump efficiency.
Using this relation for the maximum flow rate and assuming a conservative efficiency of 10%, the peak power draw from the rotor is 0.02 HP.
Figure 5.17: Power draw from rotor induced by Coriolis Forces during system de- pressurization.
Now, for comparison, the maximum power draw from a rotor fitted with the developed centrifugally powered pneumatic de-icing system is compared to the increase in hover power required for a typical helicopter by adding 200 lbs. of electrical equipment necessary for and electro thermal de-icing system. Assuming the vehicle
138 has a 4 bladed rotor of radius 24 ft. and a gross weight of 22,000 lbs.(similar to a
Sikorsky UH-60L Black Hawk), the increase in hover power required produced by the extra weight of the electrical equipment is 32 HP, while the maximum power draw imposed on the rotor for this vehicle by the pneumatic de-icing system is 16.5 HP, or about a 50% decrease. Also note that the power loss from the pneumatic de-icing system is only present while the system is in use (i.e., during icing conditions when air is being pumped in and out of the system), while the 32 HP hover power increase imposed by the electrical system weight is always present. Note that the increase in power required to add the 200 lbs. of electrical equipment to the vehicle only considers the power increase due to weight, and not the electrical power required to run the electrothermal system. If the electrothermal system is assumed to protect the same blade area as the pneumatic system, the electrical power required for just one protected blade is 19 kW (25.5 HP), recall that electrothermal systems require
25 W/in2, giving a total power of 57.5 HP to de-ice one blade at a time.
139 Chapter 6 | Conclusions and Recommendations for Future Work
6.1 Conclusions
A full-scale centrifugally powered pneumatic de-icing system for helicopter blades was developed and tested under representative icing conditions. The designed system makes use of the pressure differential created within a spinning rotor blade to deform a 0.03 in. thick metallic leading edge cap, producing the transverse shear stresses necessary to delaminate accreted ice. Conclusions that can are drawn from the development and testing of the pneumatic de-icing system are as follows:
• An improved centrifugally powered pneumatic de-icing system that replaced
the inflatable rubber structures with flexible metallic structures was designed.
• The new de-icing system design was installed on span-truncated rotors and was
tested under representative icing conditions at the Penn State AERTS.
• Of the two designs tested, Prototype II proved superior in its de-icing abilities.
Results from testing showed that Prototype II was capable of delaminating ice
140 shaped 69% thinner than Prototype I.
• Prototype II was able to delaminate ice thicknesses as small at 0.078 in. at
-14◦C and 0.112 in. at -5◦C when 4 psig of pneumatic pressure was applied to
the system (representative of centrifugally generated pressures at the tip of a
full-scale rotor blade). Compared to an electrothermal de-icing system that
requires about 0.3 in. of ice thickness for effective ice protection, the tested
pneumatic de-icing system provides a 74% decrease in required ice thickness.
• A previously developed method to predict the ability of the de-icing system
to delaminate accreted ice using the cohesive zone method and Abaqus finite
element software and a custom ice shedding code was used to predict the
developed de-icing system performance.
• Comparisons between delamination model predictions and experimental rotor
ice testing results revealed that for the cold temperature case (-14◦C) the Abaqus
delamination model produces a linear trend with a 46% error with respect to
experimental results. For ice shapes created at the warmer temperature case
(-5◦C), the linear trend produced by the Abaqus model has a 300% error with
respect to experimental data. The large discrepancy in experimental and model
results is believed to be caused by the exclusion of ice fracture, which was noted
during rotor ice tests, and 3-D effects from the 2-D delamination model.
• The Prototype II de-icing system was installed on the 8ft. outboard tip region
of a full-scale 24 ft. K-MAX blade, replacing the original erosion cap from the
blade.
• A portable icing cloud generator capable of producing representative atmospheric
141 icing conditions for full-scale icing tests was designed and constructed. The
cloud generator was tested successfully with all twelve nozzles for droplet sizes
up to 45 µm and air pressures up to 25 psi.
• Full-scale icing tests were conducted at Kaman’s whirl tower at static air
temperatures within the FAR Part 25/29 Standard Icing Envelope.
• It was demonstrated during full-scale icing test that the de-icing system is capa-
ble of delaminating ice thicknesses as small as 0.08 in. using only centrifugally
generated pressures.
• Electrical power requirements of the de-icing system are negligible (10 mW for
a microvalve to cycle between high and low pressure lines).
• A peak instantaneous mechanical power loss of 4 HP is imparted on the rotor
by the de-icing system during de-pressurization. The power draw, which is
produced by the flow of air from the blade tip creating a Coriolis force on the
blade, decays to zero once the volume of air is evacuated from the system (about
7 seconds). During pressurization the de-icing system acts as a centrifugal
pump, and draws a maximum of 0.02 HP from the rotor.
• The pneumatic system directly replaces existent leading edge caps and introduces
minimal weight related to plastic pressure lines spanning the length of the rotor.
The system does not require heavy non-rotating electrical components, and its
weight is negligibly more than that of the existing protective leading edge cap.
142 6.2 Future Work
6.2.1 Pneumatic De-icing System Design
The developed centrifugally powered pneumatic de-icing system was proven in both small-scale prototype and full-scale rotor icing tests. However, improvements to the system design can be made. Firstly, the de-icing system that was tested on a full-scale blade was only installed on the outboard 8 ft. section (0.65%R to R) to take advantage of the positive pressures available via centrifugal pumping at this radial location (see Figure 6.1). To protect the entire blade from root to tip, a new design must be developed for the inboard half of the blade which uses the negative pressures available to remove accreted ice.
A possible inboard configuration would consist of a metallic leading edge cap formed into a shape that is slightly larger than the blade contour such that when negative pressures are applied, is pulled flush to the blade surface. While ice is accretes to the blade, the metallic cap would be pulled and held to the blade surface to maintain its aerodynamic shape using the available negative pressures as with the designs discussed in this thesis. Once sufficient ice thickness has accreted, negative pressure is released from the system and residual stresses along with aerodynamic pressures act to deform the metallic cap and delaminate the accreted ice shape. Due to the operation of this configuration, the system design would be similar to that of
Prototype I (see Chapter 2). The described configuration is depicted in Figure 6.2.
Note that one span wise continuous leading edge cap could be used if a custom die to form the airfoil shape was used, however doing so would be cost prohibitive.
Instead, two separate leading edge caps could be used (one for inboard and one
143 Figure 6.1: Experimentally measured pressures produced by centrifugal pumping along a 24 ft. radius rotor blade at 270 RPM. for outboard) to simplify fabrication. Doing this would introduce a discontinuity where the two caps meet at the blade mid-span. The discontinuity, when the de-icing system is actuated, could create additional stresses in this area to assist with ice delamination.
In Chapter 2, a parametric study to reduce leading edge cap deformation due to aerodynamic pressures was conducted. Results from the parametric study showed that increasing elastomer stiffness by 40% from the baseline provided a 4% reduction in cap movement due to aerodynamic pressures while increasing the required pneumatic pressure for delamination from the baseline by only 4.5%. Future de-icing system designs should consider further investigation this design change.
144 Figure 6.2: Schematic of pneumatic de-icing system configuration for inboard and outboard blade regions.
6.2.2 Pneumatic De-icing System Modeling
The use of Abaqus finite element softeare to predict ice/leading edge interface delamination was used in this research. Interfacial cohesive properties experimentally measured in Ref. [16] were used for the analyses in this thesis. The cohesive properties were however measured only at -10◦C. To form a complete model that can be used to accurately model future pneumatic de-icing systems, the cohesive properties listed in
Chapter 3 should be experimentally measured for temperautures ranging from 0◦C to -20◦C.
To date, all finite element models used to predict ice/leading edge delamination in this thesis and in Ref. [16], brittle fracture of ice in the finite element models have been neglected. During rotor ice testing of the de-icing systems, it was noticed that acuation of the de-icing system would fracture the accreted ice in addition to
145 introducing delamination. It is expected that the inclusion of fracture to the models would greatly increase the accuracy of results based on observations of ice fracture made during rotor ice testing.
146 References
[1] Soltis, J., T., "Design and Testing of an Erosion Resistant Ultrasonic De-Icing System for Rotorcraft Blades," Masters Thesis, Aerospace Engineering Dept., The Pennsylvania State University, University Park, PA, 2013.
[2] Brouwers, E., W., "The Experimental Investigation of a Rotor Icing Model with Shedding," Masters Thesis, Aerospace Engineering Dept., The Pennsylvania State University, University Park, PA, 2010.
[3] Jeck, R., K., "Icing Design Envelopes (14 CFR Part 25 and 29, Appendix C) Converted to a Distance-Based Format," Federal Aviation Administration, Rept. DOT/FAA/AR-00/30, Washington D.C., April 2002.
[4] Han, Y., and Palacios, J., L., "Analytical and Experimental Determination of Airfoil Performance Degradation due to Ice Accretion," AIAA 2012-2794, 4th AIAA Atmospheric and Space Environment Conference, New Orleans, LA, 2012.
[5] Green, S., D., "A Study of U.S. Inflight Icing Accidents and Incidents, 1978 to 2002," AIAA 2006-82, 44th AIAA Aerospace Sciences Meeting and Exhibit, Reno, NV, 2006.
[6] Fibre Metal Laminates Center of Competence. Retrieved Nov. 30, 2015, URL: http://www.fmlc.nl/research-development/innovation.
[7] Overmeyer, A., D., "Actuator Bondline Optimization and Experimental Deicing of a Rotor Blade Ultrasonic Deicing System," Masters Thesis, Aerospace Engineering Dept., The Pennsylvania State University, University Park, PA, 2012.
[8] Burkett, B., "Ice Phobic Coatings on Controlled and Covered Surfaces," 68- 2012-000156, American Helicopter Society 68th Annual Forum, Fort Worth, TX, 2013.
[9] Coffman, H., J., J., "Helicopter Rotor Ice Protection Methods," Journal of the American Helicopter Society, Vol. 32, No. 2, 1987, pp. 34-39.
[10] Geer, W., C., and Scott, M., "The Prevention of the Ice Hazard on Airplanes," NACA Tech. Note No. 345, Washington D.C., 1930.
147 [11] Martin, C., A., and Putt, J., C., "Advanced Pneumatic Impulse Ice Protection System (PIIP) for Aircraft," Journal of Aircraft, Vol. 29, No. 4, 1991, pp. 714-716.
[12] Haworth, L., A., Oliver, R., G., and Woratschek, R., "JUH-1H Pneumatic Boot Deicing System Flight Test Evaluation," USAAEFA AD-A140 986, Edwards AFB, CA, 1983.
[13] Olsen, W., "Survey of Aircraft Icing Simulation Test Facilities in North America," NASA TM-81707, Cleveland, OH, 1981.
[14] Eglin AFB McKinley Climatic Laboratory. Retrieved Jan. 13, 2016. URL: http://www.eglin.af.mil/shared/media/photodb/photos/022808-F-0000P- 064.jpg.
[15] Szefi, J., T., Cormier, B., and Ionno, L., "Full Scale Experimental Validation of Dynamic Centrifugally Powered Pneumatic Actuators for Active Rotor Blade Sur- faces," 70-2014-0098, American Helicopter Society 70th Annual Forum, Montreal, Quebec, 2014.
[16] Bailey, M., J., "Finite Element Modeling of Ice Delamination from a Pneu- matic De-Icing System for Helicopter Rotor Blades," Masters Thesis, Aerospace Engineering Dept., The Pennsylvania State University, University Park, PA, 2014.
[17] Palacios, J., Wolfe, D., and Bailey, M., "Ice Testing of a Centrifugally Powered Pneumatic Deicing System for Helicopter Rotor Blades," 70-2014-0092, American Helicopter Society 70th Annual Forum, Montreal, Quebec, 2014.
[18] Cheremisinoff, N., P., Boyko, R., and Leidy, L., Elastomer Technology Handbook, CRC Press Inc., Boca Raton, FL, 1993, pp. 580.
[19] Turon, A., Davila, C., G., Camanho, P., P., and Costa, J., "An Engineering Solution for Solving Mesh Size Effects in the Simulation of Delamination with Cohesive Zone Models," NASA TM-2005-213547, Hampton, VA, 2005.
[20] Ide, F., "Liquid Water Content and Droplet Size Calibration of the NASA Lewis Icing Research Tunnel," NASA TM-102447, Cleveland, OH, 1990.
[21] Leary, W., M., "We Freeze to Please: A History of NASA’s Icing Research Tunnel and the Quest for Flight Safety," NASA SP 2002-4226, Washington D.C., 2002.
[22] Smith, W., L., "Weather Problems Peculiar to the New York-Chicago Airway," Monthly Weather Review, December, 1929, pp. 503-206.
[23] Carroll, T., and McAvoy, Wm., H., "The Formation of Ice Upon Exposed Parts of an Airplane in Flight," NACA Tech. Note No. 293, Washington D.C., 1928.
148 [24] Knight, M., and Clay, W., C., "Refrigerated Wind Tunnel Tests on Surface Coatings for Preventing Ice Formation," NACA Tech. Note No. 339, Washington D.C., 1930.
[25] Theodorsen, T., and Clay, W., C., "Ice Prevention on Aircraft by Means of Engine Exhaust and a Technical Study of Heat Transmission from a Clark Y Airfoil," NACA Rept. No.403, 1931.
[26] Herman, E., "Goodrich Icing Wind Tunnel Overview," AIAA 2006-862, 44th AIAA Aerospace Sciences Meeting and Exhibit, Reno, NV, 2006.
[27] Ragni, A., Esposito, B., Marrazzo, M., Bellucci, M., and Vecchione, L., "Calibra- tion of the CIRA IWT in the Low Speed Configuration," AIAA 2005-471, 43rd AIAA Aerospace Sciences Meeting and Exhibit, Reno, NV, 2005.
[28] Al-Khalil, K., Salamon, L., and Tenison, G., "Devlopment of the Cox Icing Research Facility," AIAA 98-0097, 36th AIAA Aerospace Sciences Meeting and Exhibit, Reno, NV, 1998.
[29] Oleskiw, M., M., "A Review of 65 Years of Aircraft In-Flight Icing Research at NRC," Canadian Aeronautics and Space Journal, Vol. 47, No. 3, 2001.
[30] Belte, D., and Woratschek, R., "Helicopter Icing Spray System (HISS) Evaluation and Improvements," USAAEFA AD-A170 732, Edwards AFB, CA, 1986.
[31] Palacios, J., Brouwers, E., and Smith, E., "Adverse Environment Rotor Test Stand Calibration Procedures and Ice Shape Correlation," 66-2010-000192, American Helicopter Society 66th Annual Forum, Phoenix, AZ, 2010.
[32] Kind, R., J., Potapczuk, M., G., Feo, A., Golia, C., and Shah, A., D., "Experi- mental and Computational Simulation of In-Flight Icing Phenomena," Progress in Aerospace Sciences, Vol. 34, No. 5-6, 1998, pp. 257-345.
[33] Gent, R., W., Dart, N., P., and Cansdale, J., T., "Aircraft Icing," Philosophical Transactions: Mathematical, Physical and Engineering Sciences, Vol. 358, No. 1776, 2000, pp. 2873-2911.
[34] European Aviation Safety Agency, Annual Safety Review, Cologne, Germany, 2015.
[35] Weener, E., "Lessons from Icing Accidents and Incidents," NTSB Presentation to Experimental Aircraft Association, 2011.
[36] Withington, T., "Bad Vibrations," Aerosafety World, February, 2010, pp. 24-28.
149 [37] Flemming, R., J., and Hemming, R., J., "The Past Twenty Years of Icing Research and Development at Sikorsky Aircraft," AIAA 2002-0238, 40th AIAA Aerospace Sciences Meeting and Exhibit, Reno, NV, 2002.
[38] Palacios, J., L., "Design, Fabrication, and Testing of an Ultrasonic De-Icing System for Helicopter Rotor Blades," Masters Thesis, Aerospace Engineering Dept., The Pennsylvania State University, University Park, PA, 2008.
[39] DOW Chemical Company, DOW UCAR Aircraft De-Icing Fluids, Form No. 183-00021-0709, 2009.
[40] Schrag, R., and Zumwalt, G., "Electro-Impulse Deicing - Concept and Electro- dynamic Studies," AIAA 84-0021, 22nd AIAA Aerospace Sciences Meeting and Exhibit, Reno, NV, 1984.
[41] Zumwalt, G., "Icing Tunnel Tests of Electro-Impulse De-Icing of an Engine Inlet and High-Speed Wings," AIAA 85-0466, 23rd AIAA Aerospace Sciences Meeting and Exhibit, Reno, NV, 1985.
[42] Lemont, H., E., and Upton, H., "Vibratory Ice Protection for Helicopter Rotor Blades," USAAMRDL-TR-77-29, Fort Eustis, VA, 1978.
[43] Sweet, D., "Giving Ice the Boot: Understanding Pneumatic De-icing," UTC Aerospace Systems Tech. Bulletin 101, 2015.
[44] Coffman, H., J., J., "Review of Helicopter Icing Protection Systems," AIAA 83-2529, AIAA Aircraft Design, Systems and Technology Meeting, Fort Worth, TX, 1983.
[45] Ide, R., F., and Oldenburg, J., R., "Icing Cloud Calibration of the NASA Glenn Icing Research Tunnel," NASA TM-2001-210689, Cleveland, OH, 2001.
[46] Angers, S., "Nice Ice, Baby," Boeing Frontiers, March, 2008, pp. 30-31.
[47] ASTM International, Standard Test Method for Apparent Shear Strength of Single-Lap-Joint Adhesively Bonded Metal Specimens by Tension Loading (Metal- to-Metal), ASTM Standard-D1002-10, 2010.
[48] Ali, A., Hosseini, M., and Sahari, B., B., "A Review of Constitutive Models for Rubber-Like Materials," American Journal of Engineering and Applied Sciences, Vol. 3, No. 1, 2010, pp. 232-239.
[49] MSC Software, Nonlinear Finite Element Analysis of Elastomers, MSC Software White Paper, 2010.
[50] Wright, B., "User Manual for the NASA Glenn Ice Accretion Code LEWICE," NASA CR-2002-211793, Cleveland, OH, 2002.
150 [51] Drela, M., and Youngren, H., XFOIL 6.9 Users Manual, 2001.
[52] Han, Y., "Theoretical and Experimental Study of Scaling Methods for Rotor Blade Ice Accretion Testing," Masters Thesis, Aerospace Engineering Dept., The Pennsylvania State University, University Park, PA, 2011.
151