Design optimization of the A320 engine inlet cowl
Bruno Santos da Conceição
Thesis to obtain the Master of Science Degree in Aerospace Engineering
Supervisors: Prof. Luís Filipe Galrão dos Reis Prof. Vítor Manuel Rodrigues Anes
Examination Committee Chairperson: Prof. Filipe Szolnoky Ramos Pinto Cunha Supervisor: Prof. Luís Filipe Galrão dos Reis Member of the Committee: Prof. Aurélio Lima Araújo
November 2016 ii Acknowledgments
I would like to express my gratitude to Prof. Lu´ıs Reis and V´ıtor Anes for their support and recom- mendations through the development of the Thesis. I would like to thank TAP Engines Engineering department for their availability and for the opportunity of having direct contact with the components at their facilities.
I would like to express my gratitude to my parents and grandparents for their support and for their moti- vational speeches when most needed. I would like to thank my sister for her wise advises when difficult decisions had to be made. Finally, I would like to thank my girlfriend for having been so supportive, patient and comprehensive when late night work had to be done.
iii iv Resumo
As aeronaves operam em meios nos quais os seus componentes estao˜ sujeitos a grandes variac¸oes˜ de pressao˜ e temperatura. Em estruturas como as nacelles dos motores, que sao˜ compostas por varios´ componentes e materiais, tornam-se vis´ıveis alguns sinais de desgaste e corrosao,˜ originados pela sua operac¸ao˜ em ambientes como o acima descrito. Nestes casos, devem de ser tomadas medidas correctivas. Os paineis´ acusticos´ da entrada de ar do Airbus A320/A321, apresentam alguns problemas de desgaste e corrosao˜ nas doublers de alum´ınio das juntas.
Por forma a poder desenvolver-se uma acc¸ao˜ correctiva ao n´ıvel das juntas dos paineis´ acusticos,´ deve ser realizada uma analise´ do comportamento mecanicoˆ e das forc¸as actuantes nas mesmas.
Nesta tese de mestrado, foi desenvolvida uma metodologia para a analise´ do comportamento mecanicoˆ das juntas dos paineis´ acusticos´ com base em ferramentas, tais como, Dinamicaˆ de Fluidos Com- putacional (CFD), Metodo´ dos Elementos Finitos (FEM) e Desenho Assistido por Computador (CAD). A modelac¸ao˜ da entrada de ar e dos seus componentes e´ feita atraves´ da utilizac¸ao˜ do programa Solidworks. A utilizac¸ao˜ do programa de CFD, STAR CCM+, permitiu a determinac¸ao˜ do carrega- mento aerodinamicoˆ a que a entrada de ar esta´ sujeita. A analise´ estrutural dos componentes da junta do painel acustico´ e´ realizada com uso das ferramentas de FEM, dispon´ıveis no programa ANSYS Workbench. Todos os passos envolvidos nestas analises´ sao˜ explicados e os resultados obtidos sao˜ apresentados.
Palavras-chave: Aeronave, entrada de ar, paineis´ acusticos,´ carregamento aerodinamico,ˆ com- portamento mecanico,ˆ juntas.
v vi Abstract
Aircraft operate in environments in which the components are subject to large temperature and pressure variations. In structures as the engine nacelles, composed by several components and materials, the presence of wear and corrosion becomes noticeable, due to the their operation in environments as the one foremost described. Corrective actions must be employed to the components which present this kind of problems. The acoustic panels of the inlet cowl of the Airbus A320/A321, present corrosion problems on the aluminium doublers of the joints.
In order to develop a corrective action to the joint of the acoustic panels, the analysis of the mechanical behaviour and forces acting on the joint must be made.
In this master’s thesis, a methodology involving Computational Fluid Dynamics (CFD), Finite Element Method (FEM) and Computer Aided Design (CAD) tools is developed in order to analyse the mechanical behaviour of the acoustic panel’s joint. The geometry of the inlet cowl and of its components is modelled with the use of Solidworks software. The determination of the aerodynamic loads acting on the inlet cowl is made with the use of CFD tools, with STAR CCM+ software. The structural analysis of the members of the joint of the acoustic panels is made with the use of FEM tools in ANSYS Workbench software. The steps involved in the analysis are explained and the results are presented.
Keywords: Aircraft, inlet cowl, acoustic panels, aerodynamic loads, mechanical behaviour, joints.
vii viii Contents
Acknowledgments...... iii Resumo...... v Abstract...... vii List of Contents...... ix List of Tables...... xiii List of Figures...... xv Acronyms and Nomenclature...... xix
1 Introduction 1 1.1 Motivation...... 1 1.2 Objectives...... 3 1.3 Thesis Outline...... 4
2 Background 5 2.1 Turbofan Engine...... 5 2.2 The Nacelle...... 7 2.3 Aircraft noise...... 9 2.3.1 Aircraft Noise Sources...... 9 2.3.2 Noise Reduction Technology...... 10 2.3.3 The Acoustic Liner...... 12 2.4 Computational Fluid Dynamics - Overview...... 13 2.4.1 Governing Equations of Fluid Flow...... 14 2.4.2 RANS Equations and Turbulence Models...... 18 2.5 Joints and Fasteners...... 19 2.5.1 Comparison between mechanically fastened and adhesive bonded joints..... 19 2.5.2 Fasteners...... 20 2.5.3 Fastened Joints Failure and Prevention...... 21 2.5.4 Honeycomb Sandwich Tensile and Shear Properties...... 24 2.6 Bolted Joints in Ansys Workbench...... 24 2.6.1 Modelling Bolts...... 25 2.6.2 Modelling Contact in Joints...... 25
ix 3 Aerodynamic Loads Determination - CFD Methodology 27 3.1 CAD Modelling of the A320/321 Nacelle Inlet...... 27 3.2 Simulated Flight Conditions...... 29 3.3 CFD Approaches...... 30 3.3.1 Methodology 1...... 30 3.3.2 Methodology 2...... 30 3.4 The CFD Methodology...... 32 3.4.1 Computational Domain...... 32 3.4.2 Generated Mesh...... 32 3.4.3 The Problem Physics...... 34 3.5 Boundary Conditions...... 36 3.5.1 Setting The Boundary Conditions...... 36 3.5.2 The FAN Boundary Condition Calculation...... 37 3.6 Methodology Suitability...... 38 3.7 CFD Results and Discussion...... 39
4 FEM Methodology - Analysis of the Joints of the Acoustic Panel 51 4.1 Inlet Cowl Components...... 51 4.1.1 The Acoustic Panel Joint...... 52 4.2 Preparing the FEM Simulations...... 52 4.2.1 Aerodynamic Load...... 53 4.2.2 Geometry and Geometry Importation...... 54 4.2.3 Materials and Properties...... 55 4.2.4 Contact Between Components...... 58 4.2.5 Defining the Mesh...... 59 4.3 Determination of the Critical Load Condition...... 60 4.3.1 Model Considerations...... 60 4.4 Approach to Analyse the Acoustic Panel Joint...... 61 4.4.1 Model Simplifications...... 62 4.5 Analysis of the fasteners of the Joint...... 64 4.5.1 Components Contact and Mesh Refinement...... 64 4.5.2 Convergence Analysis...... 65 4.5.3 Result and Discussion of the Fasteners Analysis...... 66 4.6 Analysis of the Interface between Internal Honeycomb Core and Internal Doubler..... 70 4.6.1 Components Contact and Mesh Refinement...... 70 4.6.2 Convergence Analysis...... 71 4.7 Results and Discussion of the Joints Analysis...... 72 4.7.1 The Flatwise Tensile Strength...... 73 4.7.2 The Shear Stress...... 74
x 5 Conclusions 75 5.1 Future Work...... 76
6 Bibliography 77
A Appendix - Noise Certification Points 81
B Appendix - T.P.S. Experimental Properties 83
C Appendix - LOCTITE EA 9658 AERO - Technical Datasheets 85
xi xii List of Tables
2.1 CFM56-5B engine data [13]...... 7 2.2 Quota Count system classification [22]...... 10 2.3 Torque Coefficient for different bolt conditions [34]...... 22
3.1 Simulated flight conditions...... 30 3.2 Main physical properties of the simulation...... 35 3.3 Boundary condition summary...... 37 3.4 T.P.S Pressure Coefficient comparison between numerical and experimental results.... 41 3.5 Half and complete model comparison...... 41 3.6 Pressure variation for the simulated flight conditions...... 43 3.7 Wall Shear Stress variation for the simulated flight conditions...... 43
4.1 Isotropic material properties from CES2015 software...... 55 4.2 Honeycomb core orthotropic properties...... 56 4.3 Manufacturer honeycomb properties and approximations[55]...... 58 4.4 Moment and and reaction forces for different loading conditions...... 62 4.5 Force and moment reaction convergence analysis...... 62 4.6 Mesh parameters of the fasteners...... 65 4.7 Bolts convergence analysis...... 66 4.8 Hi-Lok fasteners analysis...... 68 4.9 Comparison between the theoretical preload and resultant tension force on the fasteners. 69 4.10 Mesh element size at each refinement...... 71 4.11 Joints reaction forces as function of the mesh refinement...... 72 4.12 Simulation result used for the Joints properties determination...... 73
B.1 T.P.S properties of experimental essay [53]...... 83
xiii xiv List of Figures
1.1 Three different views of the A320/A321 Inlet Cowl...... 2 1.2 Acoustic Panel side section and main Joint. (a) Acoustic Panel side section view;(b) Joint main components...... 3
2.1 Turbofan scheme [11]...... 6 2.2 Different engine/nacelle configurations. (a) - A320 Under wing pod-mounted engine, [15]; (b) - F117 fuselage buried engines, [16]...... 7 2.3 A320 representative nacelle components, modified from [17]...... 8 2.4 QC levels for night regime at Heathrow airport [22]...... 10 2.5 QC levels for Rolls-Royce powered aircraft at approach [19]...... 11 2.6 Engine noise sources - 1960s vs Modern desing [19]...... 11 2.7 Turbofan relative power levels of noise sources at takeoff an approach, modified from [19]. 12 2.8 Single and Double layer acoustic liner configuration, modified from [25]...... 13 2.9 Three splices vs zero-splice technology; (a) - A320 inlet with three splices; (b) - A380 zero-splice inlet, from [27]...... 13 2.10 Fluid element for conservation laws [28]...... 15 2.11 Hi-Lok fastener mechanism [37]...... 21 2.12 Different failure modes in shear loading [34]...... 23 2.13 Bolts solicitation modes [42]. (a) - Flange under Separation; (b) - Flange under Compres- sion...... 26
3.1 CFM56-5B engine, modified from [45]...... 28 3.2 Acoustic panels configuration. (a) - The inlet cowl three acoustic panels configuration; (b) - Mapped lower acoustic panel...... 28 3.3 Circumferential length of the inlet duct. (a) - Front view; (b) - Side view...... 29 3.4 Engine CAD Model...... 29 3.5 Engine configuration for CFD simulations...... 31 3.6 Computational domain and boundaries...... 32 3.7 The Mesh and its different levels of refinement...... 34 3.8 T.P.S geometry and generated mesh for the computational simulation...... 39 3.9 T.P.S Cp distribution for computational and experimental results...... 40
xv 3.10 Model configurations. (a) - Half Model Configuration; (b) - Complete Model Configuration. 40 3.11 Wall y+ for the four flight conditions. (a) - Takeoff 0o; (b) - Takeoff 9o; (c) - Takeoff 16o; (d) - Cruise...... 44 3.12 Residuals evolution for the four flight conditions. (a) - Takeoff 0o; (b) - Takeoff 9o; (c) - Takeoff 16o; (d) - Cruise...... 45 3.13 Mass flow convergence for the four flight conditions. (a) - Takeoff 0o; (b) - Takeoff 9o; (c) - Takeoff 16o; (d) - Cruise...... 46 3.14 Velocity field for the four flight conditions. (a) - Takeoff 0o; (b) - Takeoff 9o; (c) - Takeoff 16o; (d) - Cruise...... 47 3.15 Mach number at the fan for the four flight conditions. (a) - Takeoff 0o; (b) - Takeoff 9o; (c) - Takeoff 16o; (d) - Cruise...... 48 3.16 Nacelle pressure distribution for the four flight conditions. (a) - Takeoff 0o; (b) - Takeoff 9o; (c) - Takeoff 16o; (d) - Cruise...... 49 3.17 Nacelle wall shear stress distribution for the four flight conditions. (a) - Takeoff 0o; (b) - Takeoff 9o; (c) - Takeoff 16o; (d) - Cruise...... 50
4.1 Inlet Cowl Components. (a) - Inlet Cowl External View; (b) - Inlet Cowl Internal Section, modified from [59]...... 52 4.2 Acoustic panel. (a) - Acoustic Panel side section view; (b) - Joint main components.... 53 4.3 Inlet Cowl geometry. (a) - Inlet Cowl isometric view; (b) - Inlet Cowl rear view; (c) - Inlet Cowl internal view; (d) - Inlet Cowl internal view with Attachment Ring...... 54 4.4 Honeycomb Cell Configuration and Coordinate System, from [55]...... 56 4.5 Cell Nomenclature [56]...... 57 4.6 Referential of the honeycomb properties...... 57 4.7 Several pinball spheres in the joint...... 59 4.8 Different simulation setting for the determination of the critical load. (a) - Mesh configura- tion used in the model; (b) - Geometry constrain - Fixed Support; (c) - Pressure Distribu- tion for Takeoff 16 condition; (d) - Wall Shear Stress distribution for Takeoff 16 condition.. 61 4.9 Force and Moments relative error as function of the mesh refinement...... 63 4.10 Joint’s Fasteners Pattern. (a) - Fasteners pattern on acoustic panel and Attachment Ring; (b) - Main dimensions of the fasteners pattern...... 63 4.11 Model simplifications. (a) - Inlet cowl most stressed region; (b) - Spring model and lateral faces of the section...... 64 4.12 Geometry displacement due to aerodynamic load...... 65 4.13 Fasteners mesh refinements. (a) - Fasteners Refinement 1; (b) - Fasteners Refinement 5. 66 4.14 Hi-Lok nomenclature and position...... 67 4.15 Hi-Lok configuration in the Joint and Hi-Lok loading. (a) - Complete Joint view; (b) - Internal Joint view - No Honeycomb; (c) - Stress on the fasteners; (d) - Deformed and Non deformed fasteners configuration...... 69
xvi 4.16 Hi-Lok torque measurement. (a) - Hi-Lok fixed in a vise; (b) - Torquing of the Collar.... 70 4.17 Joint Connection referential. (a) - Referential for lower and side joints; (b) - Referential zoom...... 71 4.18 Lower and higher joint mesh refinements. (a) - Model lower refinement; (b) - Model higher refinement...... 72 4.19 Reaction force convergence for the Joints...... 73 4.20 Coordinate systems of the Joints and resultant forces. (a) - Joints referential; (b) - Resul- tant forces on the joints...... 74
A.1 Noise certification points for ICAO Annex 16 and FAA FAR36 [19]...... 81
xvii xviii Acronyms and Nomenclature
Acronyms
AoA = Angle of Attack ASTM = American Society for Testing and Materials BC = Boundary Condition CAD = Computer Aided Design CFD = Cmmputational Fluid Dynamics Cp = Pressure Coefficient DNS = Direct Numerical Simulation DOA = Design Organization Approval EASA = European Aviation Safety Agency EPNL = Effective Perceived Noise Level FAR36 = Federal Air Regulation, part 36 FEM = Finite Element Method ICAO = International Civil Aviation Organization LES = Large Eddy Simulation PWL = Power Level of sound QC = Quota Count RANS = Reynolds Average Navier-Stokes SST = Shear Stress Transport
SFσ = Safety Factor for tension stress
SFτ = Safety Factor for shear stress TAP = Transportadora Aerea´ Portuguesa T.P.S. = Turbine Powered Simulator UK = United Kingdom US = United States
xix Nomenclature
A = Area
Afan = Area of the fan a = Static speed of sound a0 = Stagnation speed of sound a∞ = Free stream speed of sound B = Bottom face b = Width of the specimen
Cv = Specific heat at constant volume c = Maximum distance from the neutral axis
Dfan = Fan diameter d = Fastener diameter E = East face, Specific energy
Es = Young’s modulus of the original solid material ∗ EX = Equivalent Young’s Modulus in X direction ∗ EY = Equivalent Young’s Modulus in Y direction ∗ EZ = Equivalent Young’s Modulus in Z direction E˙ = Rate of consumption of chemical enerfy of the fuel F = Shear force
Fi = Preloa
Gs = Shear modulus of an isotropic material ∗ GXY = Shear modulus in plane XY ∗ GYX = Shear modulus in plane YX ∗ GXZ = Shear modulus in plane XZ ∗ GYZ = Shear modulus in plane YZ h = Length of the honeycomb cell wall, joint thinnest plate thickness I = Second area moment i = Internal Energy K = Torque coefficient k = Spring Stiffness, thermal conductivity, turbulent kinetic energy k0 = Reference thermal conductivity L = Length of the specimen
Lsting = Length of the sting l = Length of the cell wall M = Mach number, bending moment
M∞ = Free stream Mach number m˙ = Mass flow m˙ a = Engine mass flow
xx m˙ fan = Mass flow at the fan m˙ 0m = Mas flow at 0 m altitude m˙ 11000m = Mass flow at 11000m altitude N = North Face P = Load p = Pressure pamb = Static pressure p0 = Stagnation pressure R = Universal gas constant
Re = Reynolds number
Recrit = Critical Reynolds number S = Sutherland constant, south face
SE = Energy source term
Si = Internal energy source term
SMx = Energy source term in the x component of the moment equation
SMy = Energy source term in the y component of the moment equation
SMz = Energy source term in the z component of the moment equation ∗ Sij = Mean rate of deformation T = Top face, temperature, torque
Tamb = Static temperature
T0 = Reference temperature t = Time, fastener grip, honeycomb cell wall thickness u = Flight speed, velocity in the x direction u = Velocity vector uτ = Friction velocity v = Velocity component in y direction v∞ = Free stream velocity W = West face w = velocity component in the z direction X = Referential component x = Referential component, displacement Y = Referential component y = Referential component, distance from the wall y+ = Non-dimensional wall distance Z = Referential component z = Referential component γ = Ration of specific heats
δij = Kronecker delta operator = Rate of viscous dissipation
xxi εX = Strain in x direction
εY = Strain in y direction η = Engine thermal efficiency µ = Dynamic viscosity
µt = Turbulent viscosity
µ0 = Reference dynamic viscosity ν = kinematic viscosity
νs = Poisson’s ration for isotropic material ∗ νXY = Equivalent Poisson’s ration in plane in XY with X loading direction ∗ νYX = Equivalent Poisson’s ration in plane in XY with Y loading direction ∗ νZX = Equivalent Poisson’s ration in plane in ZX with Z loading direction ∗ νZY = Equivalent Poisson’s ration in plane in ZY with Z loading direction ρ = Density
ρs = Density for solid isotropic material ρ∗ = Equivalent density
ρ0 = Stagnation density
ρ0m = Density at 0 m altitude
ρ11000 = Density at 11000 m altitude σ = Tensile stress, bending stress, bearing stress
σmax = Maximum tensile stress τ = Thrust, viscous stress, shear stress
τ xymax = Maximum shear stress
τw = Wall shear stress
τij = Viscous stress components, i and j can assume x,y,z components φ = Flow property φ0 = Time varying fluctuation with zero mean value of a flow property ϕ = Mean flow property ω = Tubulence frequency ∆p = Pressure difference Φ = Dissipation function
xxii Chapter 1
Introduction
1.1 Motivation
Commercial aircraft engines are invariably external pod-mounted and they are usually attached to the wing. The engines are enclosed in a structural housing, called the nacelle. The main goal of a nacelle is to reduce the drag associated to airflow passing around the engine, minimize engine noise propagation and to provide a smooth airflow to the engine [1]. The nacelle configuration varies with the engine type [2].
TAP, the Portuguese airline has a fleet of 80 aircraft. From the 80 aircraft fleet, 19 aircraft are Airbus A320 and 3 are Airbus A321 [3]. At TAP, both the A320 and A321 aircraft share the same engine and the same inlet cowl. The engine used in these aircraft is the CFM56-5B. Despite the use of the same engine, the engine is installed on each aircraft type with different rates. The inlet cowl of the A320/A321 is composed of three acoustic panels. The acoustic panels form a diffuser that provide an uniform airflow to the engine and also have acoustic properties, that partially cancel the engine’s noise [4,5]. The inlet cowl of the A320/A321 is presented in Fig. 1.1.
During the aircraft operation, the inlet cowl is exposed to large pressures and temperature ranges. Atmospheric temperatures can range from -65oC to 50oC, and some components of the inlet cowl can reach even higher temperatures due to the anti-icing system. In flight operation the inlet is exposed to rain, hail and birds. On ground operation debris can be suctioned with the airflow into the inlet. All these factors contribute to the inlet degradation [6].
When aircraft components start to degrade, corrective actions must be employed. The corrective actions depend on many factors, such as the severity of the problem and its location on the component. The corrective actions can be found in the repair manuals provided by the component manufacturer.
With regard to the acoustic panels, there are several repairs that are foreseen by the manufacturer. From time to time, aluminium corrosion appears on the doublers of the joint of the acoustic panels. In Fig. 1.2,
1 the acoustic panel and its components are presented. A repair for the aft doubler of the panel (please see Fig. 1.2) is foreseen by the manufacturer, however, no repair for the internal doubler is present in the repair manual.
TAP is certified with a Design Organization Approval (DOA). The certification is granted by EASA (Eu- ropean Aviation Safety Agency). This certificate grants to TAP the authorization to design changes for repairs, for some aircraft areas. Changes can be done to areas as Avionics, Structures, Nacelles, and others [7]. Repairs are classified as minor or major. TAP is certified to classify a design or repair as major or minor. Minor changes can be directly approved by TAP, however, major changes must be approved by EASA. By definition, a minor change is one that has no appreciable effect on the mass, balance, structural strength, reliability, operational characteristics, noise, fuel venting, exhaust emission, or other characteristics affecting the airworthiness of the product [8]. All other changes are classified as major.
Since the repair of the corroded internal doubler of the acoustic panel is not foreseen in the repair manuals of the aircraft, making use of the certification to propose a repair for this component can be beneficial for when such problems arise.
Hereupon, the work presented in this master’s thesis aims to support the approval of the repair with the analysis the mechanical behaviour of the joint where the doublers are applied.
Figure 1.1: Three different views of the A320/A321 Inlet Cowl.
2 1.2 Objectives
The main goal of the thesis is to analyse the mechanical behaviour of the joint of the acoustic panel. In order to accomplish that, the following objectives are proposed:
• Create a CAD (Computer Aided Design) representation of the engine’s nacelle of the A320/A321 aircraft, to be used in CFD (Computational Fluid Dynamics) and FEM (Finite Elements Method) simulations.
• Develop a methodology to determine the aerodynamic loads acting on the Inlet Cowl, for different flight conditions.
• Develop a methodology to analyse the mechanical behaviour of the joint of the acoustic panel.
– Determine the critical loading condition.
– Determine the stresses on the fasteners and analyse their safety.
– Analyse the stresses on the fasteners and relate them with the preload of the fasteners.
– Determine and analyse the stresses on the interface between the internal honeycomb core and the internal doubler.
(a)
(b)
Figure 1.2: Acoustic Panel side section and main Joint. (a) Acoustic Panel side section view;(b) Joint main components.
3 1.3 Thesis Outline
The structure of the master’s thesis is briefly described in the present section. This thesis comprises 5 Chapters.
The first one is the Introduction. The introduction corresponds to the present chapter. The framework and motivation for this work are presented.
The second one is the Literature Review. In this chapter theoretical concepts are presented in order to support the following chapters.
The third chapter corresponds to the CFD Methodology. In this chapter a methodology to determine the aerodynamic loading using CFD is presented. The CFD analysis were performed using STAR CCM+. The modelling procedures of the inlet cowl, for CFD purposes, is presented. The different approaches used to simulate the aircraft engine operation are explained. The simulation parameters are presented and explained. The suitability of the methodology is analysed, by applying the developed method- ology to an existing model with experimental data results. The results obtained are presented and discussed.
The fourth chapter corresponds to FEM Methodology. In this chapter a methodology, using the Finite Element Method, is develop in order to analyse the mechanical behaviour of the joints of the acoustic panels. A CAD model of the inlet cowl, with the discriminated components, is created. The importation of the loads, determined in the CFD analysis, into ANSYS Workbench is made. The critical loading condition is determined. Simplifications made to the complete model are presented, in order the reduce the computational effort and allow the analysis of the joint. A structural analysis is made to the fasteners of the Joint and the interface between the Internal Honeycomb Core and the Internal Doubler is analysed. The results obtained in each analysis are presented and discussed.
The fifth chapter contains the conclusion of the developed Thesis and some suggestions for future work.
4 Chapter 2
Background
2.1 Turbofan Engine
In the decades of 1920-1930 the need for faster aircraft with higher maximum flight altitude was recog- nised. In order to achieve these goals, effort was made in that sense, both in Germany and in England. At the end of the 30’s decade the age of the jet aircraft began. These goals were obtained with the development of the Turbojet engine, which was based on the compressor-turbine gas generator. With this new engine, flight speed and thrust highly increase when compared to piston engines, supersonic flight was now possible. This allowed the construction of larger passenger aircraft with long-range capa- bilities. After the development of turbojet engines, turboprop and turbofan engines were developed, also based on the gas generator [5].
The Turboprop engines are used in moderate-sized aircraft with medium-speed, 135m/s (300mph). They have high efficiency, however, they have a huge disadvantage which is the maximum flight speed. An- other disadvantage is the need of a massive gearbox to connect the propeller to the engine shaft. The speed limitation is associated to the propeller blades. That is, flight speed must be slow enough to avoid flow disruption due to shock waves.
The higher efficiency of a turboprop when compared to a turbojet, for a given thrust and flight speed, can be expressed by equation (2.1),
T u T E˙ = + 2 (2.1) 2ηt m˙ au where E˙ is the rate of consumption of chemical energy of the fuel, m˙ a is the engine flow rate, T the given thrust, u the flight speed and ηt the engine thermal efficiency. That is, the minimum energy consumption is associated with the largest possible airflow through the propulsion unit. For a turboprop the airflow through the propeller can be up to 50 times the rate of airflow through the engine core. In a turbojet the flow that passes through the core is limited by the fuel-air ratio needed to not compromise the thermal
5 efficiency and thrust per unit engine mass [5].
Turbojet engines have noise and fuel consumption problems in the speed range that airliners fly, M=0.8. This is the reason why, nowadays, most of the airliner aircraft are powered by turbofan engines. Turbofan engines have lower noise emissions and fuel consumptions for this speed range [9]. The turbofan engine share the same basic core as the turbojet, but it has an additional turbine that powers a large fan in the front of the engine [10]. The fan produces a large amount of thrust with little additional fuel consumption, up to 80 percent of the engine total thrust is generated by the fan [9]. A part of the mass flow that passes through the fan enters the engine core, the other part bypasses it. The ratio of the amount of flow that bypasses the engine to the amount that enters the engine defines the bypass raito [9]. As the fan is enclosed in a duct, its aerodynamics can be controlled up to high subsonic speeds, M=0.85. The incoming flow speed is reduced as it passes through inlet diffuser [5]. A turbofan configuration is presented in Fig. 2.1.
Figure 2.1: Turbofan scheme [11].
At TAP, three different aircraft models from the Airbus A320 family are used. The Airbus A319, the A320 and the A321. All of them equipped with CFM56-5B turbofan engine’s model. The CFM56-5B is one of the two engine’s models available for the Airbus A320 family. The other engine available is the V2500-A5 from Pratt&Whitney [12]. About 60 percent of the Airbus A320 family are equipped with CFM56-5B. The engines are installed with different rates for each type of aircraft. Both the A320 and A321 aircraft share the same engine’s inlet cowl (explained in the next section) and engine. The simulations presented in this work were made using the data of the CFM56-5B for the A321 rate, since the A321 works with higher rates than the A320. The main data about the CFM56-5B is presented in Table.2.1.
6 Table 2.1: CFM56-5B engine data [13].
Engine Data
Thrust 120kN (27000lbs) Mass Flow 400kg/s (≈900lbs/s) Bypass Ratio 5.7 Fan Diameter 1.73m (68.3in) Number of Blades 36
2.2 The Nacelle
The present work aims to analyse the aerodynamic flow that interacts with the inlet cowl of an A320/321 engine and determine its influence on the joints of its acoustic panels. Therefore, a description of the nacelle functions, components and application is done in the following.
Generally speaking, a nacelle is a structural housing for the engine of an aircraft. For each engine type there are different nacelles. Aircraft can be divided in two categories, civil and military. Civil nacelles, for aircraft with multiple engines, are invariably externally pod-mounted. They are mounted on the wing or attached to the fuselage. Civil aircraft with single engine, have the engine centred with the aircraft and buried in the fuselage. Military aircraft’s engines are also usually buried in the fuselage [1]. The engine integration on the aircraft must be made in order to maximize the engine efficiency, avoid the interaction of hot exhaust gases with the aircraft structure and to avoid the ingestion of foreign objects. Over wing mounted engines or fuselage mounted engines are usually used to ensure some ground clearance. Some military aircraft have engines buried in the fuselage in order to protect the engines from enemy fire [14]. In Fig. 2.2, different engine/nacelle configurations are presented.
(a) (b)
Figure 2.2: Different engine/nacelle configurations. (a) - A320 Under wing pod-mounted engine, [15]; (b) - F117 fuselage buried engines, [16].
Engine’s inlets can be classified as subsonic and supersonic. Supersonic inlets aren’t addressed in this thesis. Information about supersonic inlets can be found in [5]. In subsonic turbofan engines the nacelle’s main goal is to provide a uniform airflow to the engine in all flight conditions, in order to preserve the engine’s good performance. The nacelle also intends to reduce drag and noise propagation from
7 the engine. The nacelle inlet behaves as a diffuser, it decelerates the incoming flow to Mach numbers varying between 0.4 - 0.7 at the fan face. Depending on the flight condition the mass flow demanded by the engine varies. For level cruise the incoming fluid can suffer some deceleration when entering the nacelle inlet plane. For low speed high-thrust operations (takeoff and climb) the mass flow demand will increase, accelerating the flow next to the inlet plane. Flow separation can occur internally or externally to the nacelle. Flow separation is to be avoided for all flight conditions, because it lead to an increase in drag and reduces the engine’s performance. External flow separation can occur due to flow acceleration around the lip. For completely subsonic flow, separation can occur due to the pressure drop associated to the flow acceleration, this is followed by a rising pressure region, causing the boundary-layer to separate; For high subsonic flows the flow can become locally supersonic and separation can occur due to shock-wall interactions [5].
In Fig. 2.3, a representative scheme of the A320/321 nacelle main components is presented. The nacelle of commercial aircraft as the Airbus A320 family is composed by: Inlet cowl, fan cowl, thrust reverser, exhaust nozzle and centerbody. The inlet cowl is composed by lip, inner barrel, outer Barrel, forward bulkhead and aft bulkhead. The inner barrel is composed by three acoustic panels that create the inlet diffuser. The acoustic panels have acoustic properties and intend to reduce the emitted noise produced by the engine. Fan cowls protects the engine and creates a continuity between the inlet cowl and the thrust reverser. It also provides access to the power plant at maintenance on ground [2]. Thrust reverser are devices that allow the reduction of the aircraft ground stopping distance. This is accomplished by devices that deflect engine flow generating forward thrust [1]. The exhaust nozzle and centerbody creates an exhaust.
Figure 2.3: A320 representative nacelle components, modified from [17].
8 2.3 Aircraft noise
Nowadays, it would be unthinkable to live in a world without aircraft. Aviation became a fast, economical and flexible way to connect communities, individuals and businesses over long distances [18]. It became a critical factor in the development of the global economy. United Kingdom (UK) is an example of the development in the aviation sector, where in the last 30 years air-traffic has increased five-fold and nowadays half the population fly at least once a year [19]. The above brings us to a delicate subject, the aircraft noise and its environmental impact. Aircraft noise and emissions are and have been a major concern to residents around airports. Aircraft noise can be considered a local issue since population are mostly affected by takeoff and landing. Still, aircraft emissions pose a global issue [19].
Noise or undesirable sound, has several adverse effects on humans. Communication interference, sleep interference, physiological responses and annoyance [20].
In the 1960s commercial jet aircraft became widespread, this was when certifications and regulations to characterize aircraft noise became indispensable. At the end of the 1960s the US Federal Aviation Administration published a noise certifications regulation, Federal Air Regulation, part 36 (FAR36). Sub- sequently, the International Civil Aviation Organization (ICAO) issued the Annex 16 with similar lines. Both regulations intend to measure the public annoyance response to aircraft noise using ”Effective Per- ceived Noise Level” (EPNL) in dB. For an aircraft entering service the EPNL must be measured for three certification points, sideline condition, cutback condition and approach condition (please see Appendix A). In addition to the referred regulations, local airports have the authority to implement some additional requirements, that can be more restrictive than the above referred. This is the case of the Quota Count system (QC) used in London airports in order to manage night aircraft movement noise [21]. This sys- tem is one of the most rigorous and is surely the most critical since the classification doesn’t take in to account the aircraft takeoff weight [19]. The night period has the duration of 6.5h, from 23h30 to 6h00, local time. Depending on the night time an aircraft may not be schedule to land or takeoff depending on its QC classification. In Fig. 2.4, the QC classifications for the Heathrow Airport night regime is pre- sented. In Fig. 2.5, the QC classification is presented for Rolls-Royce powered aicraft at approach. More information about the QC system can be found in [22]. The aircraft classification is presented on Table 2.2. This type of regulations generates challenges for engine and airframe manufacturers, contributing to the development and research of solutions for more silent aircraft.
2.3.1 Aircraft Noise Sources
Commercial aircraft noise can be divided in two main sources, airframe noise and engine noise. Airframe noise is important at takeoff and landing due to flow interaction with both landing gears and the high lift devices, flaps and slats. The main noise sources in a turbofan engine are: the fan, compressor, combustion chamber, turbine and jet noise [19]. These sources have different propagation paths, i.e., sound generated by the fan can propagate through the engine intake to the forward arc and through
9 Table 2.2: Quota Count system classification [22].
Noise Classification (EPNdB) Quota Count
More than 101.9 16 99 - 101.9 8 96 -98.9 4 93 - 95.9 2 90 - 92.9 1 87 - 89.9 0.5 87 - 86.9 0.25
Figure 2.4: QC levels for night regime at Heathrow airport [22]. the bypass duct to the rear arc. Compressor noise propagates through the forward arc. Combustion chamber, turbine and jet noise propagates in the rear arc [23]. In Fig. 2.6, a comparison between 1960s aircraft engines (Turbojet Engine) and modern aircraft engines noise sources is made. As it can be seen the main differences between models are the magnitude of the sound sources and the presence of a fan in the modern design. In 1960s turbojet engine the main noise source was the jet. Jet noise magnitude started decreasing with the introduction of turbofan engines with higher and higher bypass ratios. The presence of coaxial streams reduced the jet noise due to an overall jet velocity reduction [19]. Nowadays, the main noise source of turbofan engines is the fan.
Fig. 2.7, represents the turbofan engine’s relative sound power levels (PWL) for the main noise sources at takeoff and approach. At takeoff fan noise and jet noise are dominant, airframe noise is small. At approach both fan and airframe noise are dominant.
2.3.2 Noise Reduction Technology
To achieve the regulation imposed noise levels, noise reduction technology is used. As fan noise is a predominant noise source, in both takeoff and approach, some technology used to reduce its influence
10 Figure 2.5: QC levels for Rolls-Royce powered aircraft at approach [19].
Figure 2.6: Engine noise sources - 1960s vs Modern desing [19]. is presented.
• Active Noise Control - Actuators used to cancel fan tones are mounted inside the ducted walls of the fan, both in the inlet and exhaust ducts. Actuators cancel target modes that have more impact on the community during takeoff and landing [24].
• Negatively Scarfed Inlet - Consists in an inlet with an extended lower lip, this allows to reflect noise upwards away from the observer at the ground. However it introduces large aerodynamic penalties [24].
• Fan Trailing Edge Blowing - consists in reducing fan noise by filling the fan’s wake trough the injection of air at the trailing edge of the fan’s blades. Doing so reduces both the wake defect and the turbulence intensity [24].
• Acoustic Liners - Acoustic liners are one of the mostly used acoustic treatments to reduce radi- ated fan noise. Acoustic liners are applied on the internal walls of the engine inlet, bypass ducts, thrust reverser and nozzle [25].
11 Figure 2.7: Turbofan relative power levels of noise sources at takeoff an approach, modified from [19].
2.3.3 The Acoustic Liner
As already referred, acoustic liners are one of the mostly used acoustic solution to reduce radiated fan noise. In [6], acoustic liners are defined as sheets of perforated materials backed by acoustical cavities.
The most common acoustic liner configuration is the single layer honeycomb sandwich. It is com- posed by the perforated sheet, honeycomb cavities and backing sheet. Each component has specific functions, when all combined it gives the desired acoustic properties. The cavities work as dead-end labyrinths which are designed to trap sounds of specific wavelengths. The liners are tuned to absorb specific ranges of frequencies, this is achieved by defining the flow resistance of the perforated sheet and the volume of the acoustical cavities. The backing sheet works as the pressure wall of the structure [6].
Material selection for acoustic liners must be carefully made. In the material selection acoustic properties must be taken into account. As acoustic liners are applied on the internal walls of the engine inlet, bypass ducts, thrust reversers and nozzle, care must be taken due to the environmental properties existing in each one of these sections. Inlet duct liners must support a wide range of atmospheric temperatures (-65oC to 50oC), erosion, corrosion and contamination [6].
Typical liners used in turbofan application are either Single or Double layered, with typical lengths be- tween 1 to 2 inches (25.4 mm to 50.8 mm) [26]. Both configurations are represented in Fig. 2.8. Double layer liners have an acoustical septum that separates both layers. Comparison between both types of liners showed that maximum noise reduction is higher for the single layer liner. However, in terms of effective perceived noise, used in certifications, double layer liner proved to be more effective over the frequency range of 1 - 6.3 kHz. This is the range in which the human ear experiences more irritation. There exist other types of liners other than the single and double layer liners. In [26], the behaviour of folded liners is investigated.
Liners are usually manufactured in segments and their installation is made with the use of axial and circumferential splices. Splices are hard and have poor acoustical properties, these liner discontinuities reduce the effectiveness of noise reduction. Studies revealed that the reduction of the spliced and
12 Figure 2.8: Single and Double layer acoustic liner configuration, modified from [25]. patched areas in the inlet allow a significant noise reduction. It also showed that noise attenuation can differ by 30dB from a spliced liner to a zero-splice liner [19].
Airbus is an example of the improvements in this technology. The Airbus A320 equipped with CFM56-5B has an inlet with 3 splices, the A340 may have an inlet with 2 splices, if the installed engine is the Rolls- Royce Trent500 and the A380 has a zero-splice inlet. The A320 and A380 are represented in Fig. 2.9. In 2006 Airbus received the ”Decibel´ d’Or” environmental award for its Zero-Splice technology. Despite the size and the 4 engines of the A380, it can reach QC1 level [25].
(a) (b)
Figure 2.9: Three splices vs zero-splice technology; (a) - A320 inlet with three splices; (b) - A380 zero-splice inlet, from [27].
2.4 Computational Fluid Dynamics - Overview
In the present section an overview of Computational Fluid Dynamic (CFD) is presented in order to theoretically support the CFD Methodology chapter. CFD has an important role in the present work, as it is the set of tools that allows the determination of the aerodynamic loads applied to the nacelle.
Computational Fluid Dynamics can be defined as the set of tools and methodologies that allows the analysis of systems involving fluid flow, heat transfer and associated phenomena, based on computer
13 simulations [28]. When the analysis field is aerodynamics, the phenomena are governed by the Navier- Stokes equations, introducing the need of solving numerically partial differential equations [29].
CFD commercial programs are based in numerical algorithms organized in order to solve the defined problem. To obtain solutions all commercial CFD codes have to pass through three main processes, Pre-processing, Solving and Post-processing.
• The Pre-processor works as an interface that allows the operator to specify the problem informa- tion. This is made through the definition of the computational domain, grid generation, definition of the physical models, fluid properties and boundary conditions.
• The Solver solves the problem introduced by the user through the application of numerical meth- ods. There are three numerical solution techniques that can be employed: Finite Difference, Finite Element and Spectral methods. Most commercial programs use the Finite Volume method which is a special finite difference formulation. The numerical algorithm consists of the following steps: Integration of the governing equations of fluid flow over the control volume of the domain; dis- cretization - process of converting the integral equations in algebraic equations; solution of the algebraic equations by an iterative method [28].
• The post-processor organizes the results and present them in a graphical form so they can be analysed.
2.4.1 Governing Equations of Fluid Flow
The fluid mechanics laws are based on the conservation law and are governed by the conservation of Mass, momentum and energy. The equation of mass conservation is also known as the continuity equation. The momentum conservation reflects the Newton’s second law. The energy conservation is known as the first law of Thermodynamics. The conservation law for a quantity U is described as:
”The variation of the total amount of a quantity U inside a given domain is equal to the bal- ance between the amount of that quantity entering and leaving the considered domain, plus the contributions from eventual sources generating that quantity.” Charles Hirch [30].
In order to derive the governing equations of fluid flow a small fluid element, as represented in Fig. 2.10, is considered.
δx, δy, δy represent the element sides. N, S, E, W , T and B, represent the element faces which stands for North, South, East, West, Top and Bottom.
The complete derivation of the equations presented in this section can be found in [28].
14 Figure 2.10: Fluid element for conservation laws [28].
2.4.1.1 Continuity Equation
The mass conservation expresses the fact that in a fluid system, mass cannot be created or disappear. The mass balance for a fluid element:
The rate of increase of mass in fluid element = The net rate of flow of mass into the fluid ele- ment
Performing the mass balance to the fluid element results the three-dimensional mass conservation or continuity equation for a compressible fluid. Eq. (2.2).
∂ρ ∂(ρu) ∂(ρv) ∂(ρw) + + + = 0 (2.2) ∂t ∂x ∂y ∂z
Where x, y and z are the spatial components, u, v, and w are the velocity components, ρ and t are density and time, respectively.
In the vector notation, Eq. (2.3).
∂ρ + div(ρu) = 0 (2.3) ∂t
Where u, is the velocity vector.
2.4.1.2 Momentum Equation
Newton’s second Law states:
The rate of increase of momentum of a fluid particle = The sum of the forces on the fluid parti- cle
The forces applied on a fluid particle can be divided into surface forces (pressure force and viscous force) and body forces (centrifugal force, Coriolis force, electromagnetic force). The last ones are accounted
15 in the source terms of the momentum equations.
Du ∂(−p + τ ) ∂(τ ) ∂(τ ) ρ = xx + yx + zx + S (2.4) Dt ∂x ∂y ∂z Mx
Dv ∂(τ ) ∂(−p + τ ) ∂(τ ) ρ = xy + yy + zx + S (2.5) Dt ∂x ∂y ∂z My
Dw ∂(τ ) ∂(τ ) ∂(−p + τ ) ρ = xz + yz + zz + S (2.6) Dt ∂x ∂y ∂z Mz
Equations (2.4), (2.5) and (2.6), represent the x, y and z components of the momentum equation, respectively. p and τij represents respectively, the pressure and the viscous stress components. The
SM terms represent the source terms. The suffix i and j, in the viscous stress represent respectively, the stress that acts in the j direction on a surface normal to i.
2.4.1.3 Energy Equation
The first law of Thermodynamics states:
The rate of increase of energy of fluid particle = Net rate of heat added to fluid the particle + The net rate of work done on the fluid particle
The energy equation:
DE ∂(uτ ) ∂(uτ ) ∂(uτ ) ∂(vτ ) ρ = − div(pu) + xx + yx + zx + xy (2.7) Dt ∂x ∂y ∂z ∂x ∂(vτ ) ∂(vτ ) ∂(wτ ) ∂(wτ ) ∂(wτ ) + yy + zy + xz + yz + zz ∂y ∂z ∂x ∂y ∂z
+ div(kgradT ) + SE
In Eq. (2.7), E, k, T and SE represent the specific energy, thermal conductivity, local temperature and energy source term respectively.
2.4.1.4 State Equations
Assuming that fluid is in thermodynamic equilibrium it is possible to describe the state of a substance by the mean of two states variables. Among the above five equations unknowns (Eq. (2.2), (2.4), (2.5), (2.6) and (2.7)) there are four variables that represent thermodynamic variables: p-pressure, ρ-density, i- internal energy and T -temperature. Using ρ and T as state variable it is possible to create the state
16 equations for pressure and internal energy for a perfect gas (Eq. (2.8) and (2.9)).
p = ρRT (2.8)
i = CvT (2.9)
In Eq. (2.8) and (2.9), Cv and R, represent respectively the the speficic heat a constant volume and the universal gas constant. This assumption creates a link between the mass, momentum and energy equations and cut out two thermodynamic variable unknowns.
2.4.1.5 Navier-Stokes equations for a Newtonian fluid
A Newtonian fluid is a fluid in which the viscous stresses are proportional to the rates of deformation. Constants of proportionality are used to relate stresses to the deformation. We have the dynamic vis- cosity, µ, which relates stresses to linear deformations and the second viscosity, λ, to relate stresses to volumetric deformations. By introducing the viscous stresses in the momentum equations (Eq. (2.4), (2.5), (2.6)) and after some rearrangements in the energy equation (Eq. (2.7)) we obtain the Navier- Stokes equations. The conservative form of the Navier-Stokes equations are presented in Eq. (2.10) to (2.15).
∂ρ + div(ρu) = 0 (2.10) ∂t ∂(ρu) ∂p + div(ρuu) = − + div(µ grad u) + S (2.11) ∂t ∂x Mx ∂(ρv) ∂p + div(ρvu) = − + div(µ grad v) + S (2.12) ∂t ∂y My ∂(ρw) ∂p + div(ρwu) = − + div(µ grad w) + S (2.13) ∂t ∂z Mz ∂(ρi) + div(ρiu) = −p div u + div(k grad T ) + S + Φ (2.14) ∂t i
p = ρRT and i = CvT (2.15)
The above equations create a closed mathematical system of seven equations with seven unknowns. Allowing the system to be solved with the addition of initial and boundary conditions. Eq.(2.10), repre- sents the continuity equation. From Eq. (2.11) to (2.13), the momentum equations. In Eq. (2.14), the energy equation and in Eq. (2.15), the state equations. In Eq. (2.14), Φ represents the effects of the viscous stresses, dissipation function. Si represents the internal energy source.
The complete derivation of the equations presented in this section can be found in [29].
2.4.1.6 Tubulence Modeling
A turbulent flow is characterized by an unsteady, random and chaotic behaviour. A flow becomes tur- bulent for values of Reynolds number , Re, above a critical value, critical Reynolds number Recrit . This
17 is, when the inertial forces becomes important when compared to the viscous forces. For Re bellow the
Recrit the flow is laminar, i.e, the flow is smooth and behaves in an orderly fashion. The turbulent flow is three-dimensional and rotational, having turbulent eddies of a wide range of length scales associated to it [28].
2.4.1.7 Turbulence Modeling Methods
In order to simulate real life turbulent flow different mathematical models are presented. The methods are grouped in the following three categories. Direct Numerical Simulation (DNS) of turbulent flows is a numerical model with the ability to simulate the whole range of the turbulent statistical fluctuations at all relevant physical scales. However, DNS has very high computational effort associated to it, being unaffordable to use in industrial applications. Large Eddy Simulation (LES) as DNS computes directly the turbulent fluctuations in space and time, but only above a certain length scale. For lower scales semi-empirical laws are used, reducing the computer effort. Reynolds Average Navier-Stokes (RANS) model only calculates the turbulent average flow properties, it ignores the turbulent fluctuations. Mean flow properties are enough for most engineering purposes, together with the lower computational effort makes the RANS model the most widely applied approximation [30].
2.4.2 RANS Equations and Turbulence Models
RANS equations are obtained by the application of Reynolds decompositions to the Navier-Stokes Equa- tions. Reynolds decompositions (Eq. (2.16)) defines a flow properties φ as the sum of a steady mean component ϕ (Eq. (2.17)) and a time varying fluctuation with zero mean values φ0(t) [28].
φ(t) = ϕ + φ0(t) (2.16) 1 Z ∆t ϕ = φ(t) dt (2.17) ∆t 0
With application of the Reynolds decomposition to the Navier-Stokes equations six new terms appear in the equations. These terms correspond to turbulent stresses, known as Reynolds stresses. New terms also appear for scalar equations. For the calculation of RANS equations turbulence models need to be introduced in order to predict the extra terms and close the system of equations.
2.4.2.1 Turbulence Models
Turbulence models are defined as function of the number of additional transport equation used to close the RANS system of equations. There are models with zero, one, two or seven extra transport equations. In the present work a two-equations model is used, the k − ωSST (Shear Stress Transport Model). The most widely used two-equation models are the k − and the k − ω. The k − ωSST is a model that combines the benefits of both k − and k − ω models. It uses blending functions to create a smooth
18 transition between the two models, as function of the distance from the nearest wall. It combines the good near wall results of the k − ω model with the good far field results of the k − .
More information about turbulence models can be found in [28, 31]
The above models are based on Boussinesq assumption. This is, Reynolds stresses, τij, are propor- ∗ tional to mean rates of deformation Sij, please see Eq.(2.18) [32].
2 τ = 2µ S∗ − ρkδ (2.18) ij t ij 3 ij
k, µt and δij represents the turbulent kinetic energy, the turbulent viscosity and the Kronecker delta operator, respectively. ω and ε, represent respectively the turbulence frequency and the rate of viscous dissipation.
2.5 Joints and Fasteners
Ideally an aircraft structure would be composed by a single unit, made of a single material and involving only one manufacturing process. Although, general requirements dictate that due to maintenance, re- pair and accessibility requirements, aircraft must be made of several components. These components are joined together, through the use of several types of connectors, creating assemblies. In turn, the assembly of these sub-assemblies creates the aircraft. Joints create a load transmission path between components. Joints can be divided into mechanically fastened joint, adhesive bonded joint and welded joints. The latter type is not analysed here, more information can be found in [33, 34]. In fastened joints, bolts or rivet are used to connect components. In adhesive bonded joints the connection is made through the use of adhesives [35].
2.5.1 Comparison between mechanically fastened and adhesive bonded joints
Both adhesive bonded and fastened joint have some advantages and disadvantages. Adhesive bonded joints are mostly used as secondary bonding or co-cured joining. They are mostly used to connect skins and in composite structures constructions. Adhesive bonded joints have great fatigue properties, and grant sealing properties to the joint. However, bonded joints also have some drawbacks, a bonded joint can’t be disassembled, can break if materials with different thermal properties are used, and there are difficulties associated to the joint inspections. Bolted or riveted joints, are joints with a lower risk associated to them. They can be disassembled, and connection can be made for any thickness joint. The inspection of fastened joints is easier than that of the adhesive bonded ones and have a relatively easier manufacturing process. However, fastened joints aren’t watertight. Additional cost may be introduced in the sealing process. Stress concentrations appear in the drilled holes section, and the joints are also prone to crack propagation. In fastened joints, loads are transferred between the joints elements by
19 compression of the internal faces of the fastener holes, a small component of shear is present on the outer faces of the elements due to friction. In bonded joints the loads are mainly transferred by shear on the element surfaces. [33; 35]
2.5.2 Fasteners
Fasteners are largely used in all kinds of constructions. Fasteners can be found in several types of sizes and materials, making the correct fastener selection very important in the joint design. Fasteners can be divided into Permanent Fasteners and Removable Fasteners.
2.5.2.1 Removable Fasteners
Bolts are externally threaded fasteners that are intended to be tightened or released by torquing a nut. Bolts and nut are usually used to clamp two or more parts together. By tightening the bolt, it will stretch, producing a clamping force called pretension or preload. If the bolt and nut are correctly installed the preload will remain whether an external load is exerted or not. Bolts can be found with a variety of strengths and sizes and head shapes. Threads used in aircraft fasteners design follow the 60o American National Form of thread specifications. The thread specifications will influence the fastener fatigue strength. Both the thread and the head-to-shank radius influence the bolt fatigue strength. Nut’s materials are usually more ductile than the bolt’s one. When tightening the nut, the nut threads will deflect and seat on the bolt threads. Washers are used both under the bolt’s head and under the nut. The function of the washer is to reduce stress concentrations, due to the presence of burrs and sharp edges after the hole drilling. Additional information about removable fasteners can be found in [36].
2.5.2.2 Permanent Fasteners
Rivets are considered permanent fasteners and are widely used in aeronautics. Rivets are used to fasten parts that have parallel surfaces. Any cold working material can make a suitable rivet. Rivets can be found with various types of finishes, such as, plating, parkerizing, or paint. However, rivets have lower tensile and fatigue strength when compared to bolts or screws, and can’t be disassembled. Rivets are neither watertight nor airtight, requiring the applications of an additions sealing compound to achieve that. There are several types of rivets and rivet’s materials, each one with specific applications. Semi-tubular rivets; Blind rivets; Hi-Shear fasteners; Hi-Lok fasteners; Taper-Lok Fasteners;
Special attention is paid to Hi-Lok fasteners, since the fasteners used in the acoustic panel joints are of this type. More information about other permanent fasteners can be found in [36]. Hi-Lok fasteners are high strength fasteners that combine the best features of a rivet with those of a bolt and nut. The Hi-Lok consists of two components, a precision threaded pin and a threaded collar. The Hi-Lok is
20 usually installed in components where high shear strength is required, on application in which the shank expansion would cause undesirable effect and where high clamp is desired due to sealing requirements. The Hi-Lok application requires the access to both sides of the joint. The pin must be installed in the joint’s hole. The pin has hexagonal recess that allow to hold it in place with the use of a wrench, while the collar is tightened. The collar has a hexagonal portion that is sheared when a specific torque is attained, leaving the joint with the correct preload associated to the size of the fastener. In Fig. 2.11, a representative scheme of the Hi-Lok mechanism is presented. Information about permanent fasteners can be found in [36].
Figure 2.11: Hi-Lok fastener mechanism [37].
2.5.2.3 Fastener Preload
As already referred the preload is the clamping force associated to the stretching of the bolt due to its tightening. The correct preload can be obtained if the elongation of the bolt can be measured. However the elongation measurement can’t always be measured in an easy way, or it is even impossible to measure it. To correctly apply the preload it is usually used a torque wrench. The preload can be related to the torque with the following equation, Eq. (2.19):
T = KFid (2.19)
In Eq.(2.19), T represents the torque [Nm], K the torque coefficient, Fi the preload [N] and d the major diameter [m] of the fastener. In Table 2.3, torque coefficients are presented for different bolt conditions [34].
2.5.3 Fastened Joints Failure and Prevention
Joint are one of the most common source of failure in structures. There are several parameters in joint design that can affect both the static strength and the fatigue life of a joint. The main joint failure modes and some general guidelines for joint design are presented next.
21 Table 2.3: Torque Coefficient for different bolt conditions [34].
Bolt Condition K
Nonplated, black finish 0.30 Zinc-plated 0.20 Lubrificated 0.18 Cadmium-plated 0.16 With Bowman Anti-Seize 0.12 With Bowman-Grip nuts 0.09
2.5.3.1 Joint Failure Modes
Fastened joints usually fail in one of the following modes, failure due to bending of the fastener or member (Fig. 2.12 (b)), failure of the rivet by pure shear (Fig. 2.12 (c)), failure due to rupture of the connected members or plates due to pure tension (Fig. 2.12 (d)), failure due to bearing stress of the fastener or of the plate (Fig. 2.12 (e)), failure by shearing and tearing (Fig. 2.12 (f) ang (g), respectively). The last two modes of failure can usually be prevented if specific design guidelines are followed. The guidelines are presented in the following section.
The expression to calculate the bending stress in the members is expressed as follow (Eq. (2.20)):
M σ = (2.20) I/c
The bending moment M can be calculated as follows M = F t/2, where F is the shearing force, t is the grip of the fastener. σ is the bending stress and I/c is the section modulus of the weakest member.
The stress of a rivet in pure shear can be calculated through the following expression (Eq. (2.21)):
F τ = (2.21) A
In Eq.(2.21), τ represents the shear stress, F represents the shear force, and A represents the cross- sectional area of the rivet.
For rupture of connected members by pure tension, the tensile stress is calculated as in Eq. (2.22)
F σ = (2.22) A
In Eq. (2.22), A is the net area of the plate.
The bearing stress can be calculated as in Eq. (2.23)
F σ = − (2.23) A
22 In Eq. (2.23), A is the projected contact area of the rivet. A = hd, h is the thickness of the thinnest plate, d is the fastener diameter.
Figure 2.12: Different failure modes in shear loading [34].
2.5.3.2 Guidelines for Joint Design
In order to avoid joint failure there are some general guides that should be followed in joint design [33,36].
• A minimum distance of 2d (d - fastener diameter) from the edge of the joint member should be used.
• A minimum spacing of 4d and a maximum spacing of 8d should be used, to improve tension and shear efficiency and to avoid inter-rivet compression bucking respectively.
• Minimum Hi-Lok diameter should be d=3/16.
• Mixed fasteners should be avoided in a joints. As rivets have a better hole fit than bolts, rivets will pick up most of the load. Rivets tend to overload.
• In fatigue critical areas the thickness of countersunk sheet should be equal or greater than 1.5 times the depth of the countersunk head of the fastener.
• There are different rows patterns usually used. Single-row pattern, Double-row pattern, Staggered- row pattern and Triple-row pattern. Double-row pattern have the most efficient load transfer, if a minimum 4d spacing is used no tension failure should occur. Staggered-row joint are used in pressure tight joint or due to space limitations. Triple-row only used in special requirements due to assembly costs. It also reduces eccentric effects.
In [38], analysis about the influence some design parameter on riveted joints are presented. The effect of the number of rivet rows, rivet row spacing, rivet pitch in row and rivet pattern are analysed.
23 • Number of rivet rows: Results showed that the joint fatigue strength is significantly increased with the increasing number of rows. This is through up to 3 rivet rows. For four and five rows the results fall into the scatter of the three rows configuration. The increase in number of rows results in a reduction of the secondary bending due to a longer overlap.
• Rivet row spacing: The increase of the row spacing is beneficial to joint fatigue properties. The increase in row spacing implies a reduction in secondary bending. Although, the spacing increase also increases the joint weight due to the large overlap.
• Rivet pitch in Row: The optimum ratio between pitch in the row and rivet diameter is favorable in the range of 2.5 to 3.75
• Rivet pattern: No difference in fatigue strength whether using in-line or staggered double-row joints.
2.5.4 Honeycomb Sandwich Tensile and Shear Properties
The determination of tensile and shear properties of an honeycomb core sandwich are very important. These properties allow to understand the range of applicability of such kind of materials. The standards ASTM C297-94 [39], and C273-00ε1 [40], present respectively, the standard test method for flatwise tensile strength of sandwich construction and standard test method for shear properties of sandwich core material.
The flatwise tensile strength can be calculated as follows, Eq.(2.24),
P σ = (2.24) A
In Eq.(2.24), σ represents the flatwise tensile strength in MPa, P the load in N, and A the cross-sectional area in mm2.
The shear stress can be calculated as follows, Eq.(2.25).
P τ = (2.25) Lb
In Eq.(2.25), τ represents de core shear stress in MPa, P the load specimen in N, L the length of specimen in mm, and b the width of specimen in mm.
2.6 Bolted Joints in Ansys Workbench
Ansys Workbench has several tools that allows the analysis of joints. Several contact options are avail- able as well as different methods for bolt analysis.
24 2.6.1 Modelling Bolts
Ansys Workbench has some methods that allows the simulation of bolts. The methods vary largely in complexity, computational effort to obtain results and representation of the reality. The following methods are commonly used to model bolted connections. Method 1 - no bolts with bonded contact; Method 2 – use of beam or line elements; Method 3 – Solid Element Bolts; Method 4 – Modelling the Bolt Threads. The methods are ordered in the complexity increase sense.
• Method 1 - No bolts - bonded contact, is the simplest approach to simulate bolted joints. No bolts are modelled, instead a bonded contact is used between interfaces of the joint. To introduce more reality into the analysis the interface should correspond to the pressure cone area around the fastener hole. This process is the simplest one to simulate fastened joints. Although, it doesn’t allows to capture joint separation and introduces unrealistic stiffness.
• Method 2 - Beam elements. Beam elements can be models in to ways, by creating a beam connec- tion in Ansys Workbench, or by creating a line body. The latter must be created in DesignModeler. The beam elements can be attached to edges or faces. Once again creating the area correspond- ing to the pressure cone creates a more realistic simulation. In this method both preload and contact can be used.
• Method 3 - Solid Element Bolts. The solid element bolts method is the most realistic of the above ones. Contact and pretension can be used. Since the element is modelled, forces and moments can be directly obtained, as well as other results of interest. The method is more complex to set up.
• Method 4 - Modelling the Bolt Threads. These model is used to obtain informations about the threads behaviour. It is extremely computationally expensive, since a very fine mesh must be used. The method is usually used with frictional contact on the threads.
Information about Bolts modelling in Ansys Workbench can be found in [41, 42].
2.6.2 Modelling Contact in Joints
When modelling bolted connections using the Finite Element Method several approximations can be used. Each model has a specific level of approximation and limitations associated to them. Depending on the loading conditions different types of contact can be used to model the joint behaviour. If a load is applied in such way that the joint element tend to separate, a preload higher than the applied force must be used. The preload must ensure that joint members are kept together. From the contact point of view, when simulating a bolt under flange separation, the contact between the bolt head/flange and nut/flange can be assumed to be glued. This is, a bonded connection can be used. Since the joint is being separated the elements will move together. When analysing a joint in which the flange is under compression, the bolt head and nut can separate from the flange. The contact in this areas can be
25 modelled as a frictionless contact, allowing the bolt to separate from the joint. As the flange is under compression, bonded contact can be used between the flange members. When loads are exerted in the transverse direction, the loads are sustained due to the frictional force between bolt head to flange and nut to flange contact. Frictional contact can be defined in Ansys Workbench. When contacts other than the bonded type are used, non-linear effecst are introduced. This introduces an higher computational effort to solve the problem. In Fig. 2.13 (a) and (b), a scheme of the joint flange under separation and compression are respectively presented. Information presented in this section can be found in ANSYS Workbench Help software and in [42].
(a) (b)
Figure 2.13: Bolts solicitation modes [42]. (a) - Flange under Separation; (b) - Flange under Compression.
26 Chapter 3
Aerodynamic Loads Determination - CFD Methodology
In the present section a CFD methodology is presented in order to determine the aerodynamic loads acting on the nacelle inlet of the Airbus A320/321. The different approaches used in order to correctly simulate flow physics are addressed and the methodology suitability analysed. The simulation results and discussion are presented.
3.1 CAD Modelling of the A320/321 Nacelle Inlet
Both A320 and A321 aircraft models share the same engine inlet cowl and engine model, CFM56-5B. A CAD model of the engine inlet had to be created in order to be used in the CFD simulations. A correct modulation of the inlet is of capital importance to correctly simulate the flow behaviour at the engine inlet. The CAD was created using SOLIDWORKS software.
The challenge involved in modelling the engine inlet and nacelle is mostly due to the lack of informa- tion available about the geometric specifications and due to the complexity associated to the curvature changes of the inlet. Most of the reliable information is manufacturer property and isn’t available. The main geometry data sources used in the design modelling are presented bellow.
• Airbus website - in Airbus website three views drawings of the aircraft are available [44]. Overall measures of the aircraft can be used but attention must be taken since measures may have a deviation of ±100 mm from the real ones.
• Airbus A320 Manuals - Airbus A320/A320Neo Aircraft Characteristics Airport and Maintenance Planning manual was used. In this manual more detailed information about the engine and nacelle is presented [43].
27 • TAP Measurements - At TAP facilities some component measurements were performed. The spinner main dimensions, the fan forward acoustic panel length, the inlet acoustic panels circum- ferential length and a mapped acoustic panel geometry.
Both the spinner and the forward acoustic panel are represented in Fig.3.1.
Figure 3.1: CFM56-5B engine, modified from [45].
The aircraft inlet is composed by three acoustic panels, the upper, the lower and the side panel. The provided mapped acoustic panel corresponds to the lower one, please see Fig. 3.2. From this panel it was possible to extract the diffuser curvature.
(a) (b)
Figure 3.2: Acoustic panels configuration. (a) - The inlet cowl three acoustic panels configuration; (b) - Mapped lower acoustic panel.
The circumferential length of the acoustic panels allowed to determine the inlet scarf angle. The scarf angle is +5.4o, which is a typical value for subsonic conventional aircraft (about +6o). The lower acoustic panel was the only panel we had information about. Being so, it was assumed that the diffuser would have the same length and curvature of the lower panel all around the inlet. It was also assumed that from the lower part of the inlet to the upper part the panel length would linearly increase with 5.4o. This can be seen in Fig.3.3. The external geometry was modelled based on information available in the manuals and Airbus website. Several sketches were created in order to get a smooth and continuous external geometry variation. The final configuration is presented in Fig. 3.4.
28 (a) (b)
Figure 3.3: Circumferential length of the inlet duct. (a) - Front view; (b) - Side view.
Figure 3.4: Engine CAD Model.
3.2 Simulated Flight Conditions
In each different flight phase the aerodynamic loads vary. In order to structurally analyse the acoustic panel joints the maximum aerodynamic load must be determined. The takeoff and cruise phases were considered to be the most critical flight phases. In the takeoff phase engines are pushed close to their maximum power, in turn, the maximum air mass flow is suctioned by the engine. In the cruise phase maximum aircraft speed is attained. In both these phases strong interactions between fluid and structure occurs. The landing phase wasn’t considered critical due to the low speeds of the aircraft and low power employed.
For the takeoff phase, three different conditions were analysed, Angle of attack (AoA) =0o, AoA=9o and AoA=16o. All simulations were performed at sea level. A takeoff speed of 149 m/s (488.72 ft/s) was considered and the maximum engine mass flow was used, 400 kg/s ( 882 lbs/s).
For the cruise phase some approximations had to be made. An altitude of 11000 m (36000 ft) and a speed of 250 m/s (820.2 ft/s) were considered. No information about the engine mass flow was available for the cruise phase. In order to rectify the engine mass flow for the cruise condition the relation presented in Eq. (3.1), was used. Despite the non-linearity of the engine operation, the only correction made was to the air density. With this approach it was assumed that the engine’s rotational speed is maintained constant between the takeoff and cruise. The equivalent mass flow resulted in 110 kg/s (242.5 lb/s).
29 ρ11000m m˙ 11000m =m ˙ 0m (3.1) ρ0m
In Eq. (3.1), m˙ 11000m and m˙ 0m represent both the mass flow at 11000 m and 0 m. ρ represents the density for the same conditions.
A total of four flight conditions were simulated. The conditions used are summarized in Table 3.1.
Table 3.1: Simulated flight conditions.
Flight Conditions Altitude [m] Mach Number Mass Flow [kg/s]
Takeoff AoA = 0o 0 0.438 400 Takeoff AoA = 9o 0 0.438 400 Takeoff AoA = 16o 0 0.438 400 Cruise 11000 (36000 ft) 0.847 110
3.3 CFD Approaches
Two different methodologies were implemented in order to obtain the aerodynamic load on the engine nacelle. All CFD simulations were performed using STAR-CCM+ software. As the engine fan blades are constantly rotating, real boundary conditions are very complicated to reproduce. Some simplifications to the real model had to be done. Both implemented methodologies intended to reproduce the engine’s operation.
3.3.1 Methodology 1
The first approach intended to simulate the engine fan operation. In order to simulate the fan a STAR- CCM+ inbuilt fan simulator was used. Some simulations were performed in order to understand the fan simulator’s reaction to different inputs. A porous media was used at the engine core in order to simulate flow dissipation and in order to generate a uniform flow at the engine exit. To correctly simulate the fan and to obtain the correct mass flow, data as the fan performance curves, fan blade geometry and rotational speed was needed. Due to poor results, unrealistic flow behaviour and lack of information about the engine’s fan, this methodology was dropped.
3.3.2 Methodology 2
In order to simplify the complexity of the real engine model and to avoid the problems associated to the fan modelling, an alternative methodology was introduced. This methodology was the adopted one to
30 perform all CFD calculations. In order to simplify the real engine model the following simplifications were made:
• Fan Simplification - in this model, the fan geometry was replaced by a section with the same diameter of the real fan. Boundary conditions were applied to the fan section to simulate the fan flow suction.
• Engine Core Simplification - due to the amount of components and their complexity, along with the complex flow behaviour in the engine core, the decision to not model the engine core compo- nents was made.
• Exhaust Simplification - as the focus of this work is to analyse the effect of the flow at the engine inlet, no exhaust flow simulation was performed. Instead, a ”Sting” was created at the end of the nacelle. With the introduction of a sting the interaction of both hot and cold jets was avoided, reducing the simulation complexity. The sting has a length 5 times the fan diameter to minimize the wake effect on the area closer to the engine, as made in [46]. In [47] it is shown that the presence of the ”sting” improves the simulation convergence when compared to a model without ”sting”.
The final engine configuration for the CFD simulations is presented in Fig. 3.5.
The complete description of the methodology is presented in the next section.
Figure 3.5: Engine configuration for CFD simulations.
31 3.4 The CFD Methodology
In the present section the various steps involved in the CFD methodology are explained. Methodology 2 was the methodology used to perform the CFD simulations. The main simplifications made to the engine/nacelle geometry were presented in the previous section. All the performed simulations were made using an isolated STAR-CCM+ engine model.
3.4.1 Computational Domain
In order to simulate the atmosphere surrounding the engine a computational domain was created. A rectangular prism with typical dimension was used. The referential origin was created at the engine fan plane, from that point the domain was created with the dimensions presented in Fig. 3.6. All dimensions were set as function of the fan diameter, Dfan. The domain dimensions are important since the distur- bances introduced in the flow field by the nacelle must be dissipated by the distance to boundary, other wise, reflection of the disturbance might occur and affect the accuracy of the results. The boundaries used in the simulations are also presented in Fig. 3.6.
Figure 3.6: Computational domain and boundaries.
3.4.2 Generated Mesh
The mesh is the result of the discretization process of the computational domain, known as space discretization. This is, the conversion of the computational domain into computer language (numbers). The real geometry is converted into a set of points that creates the mesh. At each mesh point the mathematical models are solved to obtain the computational solution. The mesh is very important because it has direct impact on the accuracy of the CFD results [31].
Meshes can be distinguished between structured and unstructured meshes. Structured meshes are composed by families of intersecting lines, one for each space dimension. Each mesh point is created by the intersection of only one line of each family. On the other side, unstructured meshes have arbitrary
32 mesh points distribution. When both compared structured grids are often more efficient in terms of accuracy, CPU time and memory requirement. However, unstructured meshes automatically generated by the software are increasingly being used over the structured ones, due to time saving in the generation process. Depending on the complexity of the problem, structured meshes can take weeks or months of engineering time to be created [31].
STAR-CCM+ has the ability to generate both structured and unstructured meshes. It uses different meshing models that allow the creation of the volume mesh.
Structured meshes can be created by using the Direct Meshing Model. This method creates high quality hexahedral meshes for swappable bodies. If the body isn’t completely sweeppable it can be divided into parts that can be sweepped.
Unstructured meshes can be created by the following five models, Tetrahedral, Polyhedral, Trimmed, Thin Mesh and Advancing Layer Mesh. Thin Mesh models are applied to thin structures (e.g., plates). All other methods are suitable to use in the analysis of the engine nacelle. However, some can bring advantages when compared to the others. The selected model affects the solutions accuracy, and CPU time to generate the mesh. Tetrahedral model is the fastest model in terms of mesh generation and is the one which uses less memory. However, it requires five to eight times the number of cells required in a polyhedral or trimmed mesh, to obtain a same accuracy solution. Trimmed mesher provided high-quality meshes, it generates hexahedral meshes with minimal cell skewness and trimmed cells next to the surface. Advancing Layer Mesh is the combination of wall surface polygonal elements with polyhedral core elements [48].
A Prism Layer Mesher is also available. This mesher generates orthogonal prismatic cells next to walls and boundaries. The prismatic layer is used in order to calculate accurately the near wall flow, the Boundary Layer. It allows the determination of wall forces, heat transfer as well as flow features, such as flow separation. Depending on the turbulence model and on the mesh refinement, these properties can be directly calculated (y+ <1) or calculated with wall functions for coarser meshes (y+ >30) [48]. In Eq. (3.2) and (3.3), the y+ and the friction velocity are represented respectively.
u y y+ = τ (3.2) ν
rτ u = w (3.3) τ ρ
In Eq. (3.2) and (3.3), y, τw, ν, ρ are respectively, distance from the wall, wall shear stress, kinematic viscosity and fluid density.
The generated mesh in the present work was composed by a trimmed mesh combined with a prismatic layer. The prismatic layer was only generated on the surfaces of interest, that is, on the surfaces where the wall shear stress was to be calculated. By only applying the prismatic layer to the desired surfaces
33 the computational effort was reduced. A few layers of prismatic cells are needed, typically 5 to 8 to determine the above properties [48]. Ten layers were used in the computational simulations. The number of layers was also taken into account in order to obtain a y+ <200. Due to computational limitations, a mesh with a maximum number of 1.61 Million cell was created. In order to obtain better accuracy in the regions of interest a block refinement was created. Sections of different sizes and with different levels of refinement were created in the regions next to the engine nacelle. The Mesh and its different levels of refinement can be seen in Fig. 3.7.
Figure 3.7: The Mesh and its different levels of refinement.
3.4.3 The Problem Physics
The models used to simulate the problem physics are explained in the present section.
In order to correctly represent the engine model, three-dimensional analysis were performed. In some analysis where the components under study are axisymmetric geometries, two-dimensional analysis can be performed. In the present case three-dimensional analysis were performed to account for the engine asymmetries and to be able to obtain the pressure and wall shear stress on the engine’s nacelle.
All simulations were performed in steady state. As the incoming flow into the computational domain remains the same during the simulations, this is, no perturbations are introduced in the simulations, it is not expected that time dependent phenomena will appear. Unsteady simulations are performed, for example, in problems with time-varying boundary conditions, transient heat transfer problem or when physical instabilities exist.
As one of the main objectives of the present work is to determine the aerodynamic loads on the engine’s nacelle, the fluid used in the analysis was a gas, more precisely air. To define both air Dynamic Viscosity and Thermal Conductivity Sutherland’s Law was used, respectively represented by Eq.(3.4) and (3.5). As all simulations were performed for Mach numbers above 0.3 compressible effects had to be taken into account. In order to model the air density the ideal gas model was used, expressing the density
34 variations as functions of the pressure and temperature.
µ T 3/2 T + S = 0 (3.4) µ0 T0 T + S
k T 3/2 T + S = 0 (3.5) k0 T0 T + S
In Eq. (3.4) and (3.5), µ and µ0, represented respectively the dynamic viscosity and reference dynamic viscosity, k and k0 the thermal conductivity and reference thermal conductivity, T and T0, the temperature and reference temperature and finally S, represents the Sutherland constant [48].
RANS equations were used along with SST K-Omega turbulence model to perform the simulations. SST K-Omega model utilization is recommended by the software for transonic simulations and simulations in which recirculation regions are present. Recirculation is expected for high angles of attack. This turbu- lence model can either use Low y+ or All y+ boundary treatments depending on the mesh refinement. Low y+ uses the direct calculations approach for the boundary layer (y+ < 1). The All y+ Wall treatment uses blended wall functions that allows the calculation of the buffer by appropriately blending the viscous sub-layer and the logarithmic regions [48].
The Coupled Flow Model is used. This model solves simultaneously the conservation for mass, mo- mentum and energy. The model uses an implicit spatial integration along with the multigrid method. A second order upwind discretization scheme was used. Information about multigrid method and dis- cretization schemes can be found in [31, 29, 48].
A summary of the physical properties is presented in Table 3.2.
Table 3.2: Main physical properties of the simulation.
Model Physical Properties
Space Dimension Three-Dimensional Time Dependence Steady-State Material Gas Compressibility Compressible - Ideal Gas Mathematical Model RANS equations Viscous Regime Turbulent Turbulence Model SST K-Omega
35 3.5 Boundary Conditions
3.5.1 Setting The Boundary Conditions
Boundary conditions (BC) have a very important role in CFD simulations as they have a direct impact on the flow behaviour and on its interactions with the components involved in the simulation. In the present work, boundary conditions can be divided into Domain BC and Engine BC.
The Domain boundary conditions intend to simulate the flow surrounding the engine. The boundary conditions were applied to each of the boundaries specified in section 3.4.1. Three types of BC were used to define the domain BC, free Stream, pressure outlet and symmetry.
• A Free Stream BC was used to simulate the incoming airflow into the domain, simulating the aircraft velocity. The free stream BC was assigned to CV INLET boundary in all flight conditions. In flight conditions with angles of attack other than zero, free stream B.C. were also assign to SYMMETRY 2 and SYMMETRY 4 boundaries in order to simulate the incoming airflow with angle of attack.
• A Pressure Outlet BC was assigned to the CV OUTLET boundary. Doing so, the pressure will adjust itself to a desired value, the reference pressure.
• The Symmetry Plane BC was used in all SYMMETRY boundaries, except for the mentioned case explained above, where free stream conditions are used. The symmetry plane creates an imaginary plane of symmetry with a mirror effect.
When selecting the BC for the engine model special care must be taken. As already reffered the engine fan representation and simulation is very complex. In [49-52] simulations of powered engines are per- formed and different types of boundary conditions are presented, both for the intake and exhaust of the engine. In the present work, the analysis of the exhaust jets weren’t taken into account, the main focus is on the intake, to specify the fan boundary conditions.
In order to simulate the flow entering the engine inlet, three types of boundary conditions can be specified at the fan. The mass flow, the pressure or the velocity can be defined for a specific engine working condition. In the present work a pressure boundary condition was used. This is, a pressure variation ∆p, relative to the reference pressure value, is applied to the fan face. The main engine property intended to be simulated was the engine mass flow absorbed by the engine for a specific flight condition. In order to determine the desired ∆p, the isentropic relations were used. By using the one-dimensional isentropic relations for a perfect gas, a starting value for the ∆p was defined. From this point the ∆p was corrected until the desired mass flow was achieved in the simulation. In [47], both mass flow and pressure boundary conditions are analysed. Results showed that pressure based boundary conditions have better convergence. Wall boundary conditions were used on the rigid sections.
• A Pressure Outlet BC was used at the fan face with a defined pressure difference relative to the reference pressure.
36 • Wall BC were used on rigid surfaces, LIP, SPINNER, ACOUSTIC PANEL, CORE COWL and JET. The wall BC set the impermeability condition. The walls were set to adiabatic walls, and on all components of interest, prismatic layer meshes were generated.
The summary of the boundary conditions used in all four simulations are presented in Table 3.3.
Table 3.3: Boundary condition summary.
Boundary Conditions Physical Boundaries Takeoff (0o) Takeoff (9o) Takeoff (16o) Cruise (0o)
CV INLET Free Stream Free Stream Free Stream Free Stream CV OUTLET Pressure Outlet Pressure Outlet Pressure Outlet Pressure Outlet FAN Pressure Outlet Pressure Outlet Pressure Outlet Pressure Outlet PANEL Wall Wall Wall Wall LIP Wall Wall Wall Wall CORE COWL Wall Wall Wall Wall SPINNER Wall Wall Wall Wall JET Wall Wall Wall Wall SYMMETRY 1 Symmetry Plane Symmetry Plane Symmetry Plane Symmetry Plane SYMMETRY 2 Symmetry Plane Free Stream Free Stream Symmetry Plane SYMMETRY 3 Symmetry Plane Symmetry Plane Symmetry Plane Symmetry Plane SYMMETRY 4 Symmetry Plane Free Stream Free Stream Symmetry Plane
3.5.2 The FAN Boundary Condition Calculation
As aforementioned, the FAN BC was set by applying a pressure difference relative to the reference pressure. The use of one-dimensional isentropic relations allowed to specify the starting ∆p value for the simulation. Next the procedure is explained.
The information available to perform these calculations is: the mass flow (m˙ ), the atmospheric static pressure (patm), the atmospheric static temperature (Tatm), the free stream velocity (V∞) and the free stream speed of sound (a∞). The free stream Mach number can be simply obtain with the expression represented in Eq. (3.6).
V∞ M∞ = (3.6) a∞
With the above information and equations (3.7)-(3.10) it is possible to calculate the stagnation proper- ties for the free stream.
T γ − 1 0 = 1 + M 2 (3.7) T 2
37 γ p γ − 1 γ−1 0 = 1 + M 2 (3.8) p 2
1 ρ γ − 1 γ−1 0 = 1 + M 2 (3.9) ρ 2
1 a γ − 1 2 0 = 1 + M 2 (3.10) a 2
In the above equations T0, p0, ρ0 and a0 represent the stagnation properties. T , p, ρ and a the static properties at a specific point, and M is the Mach number at that point. γ is the ratio of specific heats for a perfect gas.
Knowing that the stagnation properties remain the same through an isentropic flow, it is possible to relate the free stream properties with the properties at the fan. The isentropic relations can be applied since no shock waves are expected to occur in the inlet duct. Knowing the stagnation properties, the mass
flow at the fan (m˙ fan) and the fan area (Afan), it is possible to determine the static pressure at the fan. To do so, the equations (3.7)-(3.10) must be solved along with the mass flow function at the fan, Eq. (3.11).
(γ+1) √ ! 2(γ−1) m˙ p0 γ 1 fan = √ M (3.11) A an fan γ−1 2 f RT0 1 + 2 Mfan
The presented equations can be found in [5].
3.6 Methodology Suitability
When performing computational simulations it is very important to be able to analyse the suitability of the computational model by comparing the simulation results with experimental data. In the present case, no experimental data was available to enable such kind of analysis. In order to analyse the suitability of the implemented computational model, a different engine with experimental data was used.
The NAL-AERO-02-01 T.P.S. (Turbine Powered Simulation) wind tunnel experimental model was used. This model represents an axissymmetric turbofan engine from the Japanese Aerospace Technology Research Institute. This engine model was also used in [49, 50], to analysis the suitability of the com- putational model. Information about the engine and experimental data was available in [53]. From the available information it was possible to create a two-dimensional representation of the model. From that model a three-dimensional CAD model was created, enabling the preparation of a simulation with the same parameters used in the present work simulations. A closer inspection of T.P.S. model revealed that the engine had a fan diameter about four times smaller than the Airbus’s one. To solve the size
38 difference issue a proportionality relation was used to maintain the domain size and mesh cells density. As shown in section 3.4.1, the domain size was set as function of the fan diameter, doing so, the propor- tionality was automatically maintained. The mesh was refined with different levels for different sections. The cells size were made four times smaller to achieve the same cell density. In order to determine the fan ∆p for this model, similar calculations, as used in the Airbus engine model, were made.
The simulation was performed for cruise condition. Data from the T.P.S. test conditions is presented in Appendix B. All simulation parameters were maintained the same in order to evaluate the suitability of the computational model. Both the geometry and the generated mesh are presented in Fig. 3.8.
Figure 3.8: T.P.S geometry and generated mesh for the computational simulation.
In order to compare the computational results with experimental one’s a graphic with the pressure coef- ficient (Cp) distribution was made, please see Fig. 3.9. A small code was created in order to enable the comparison between the computational and the experimental results. By introducing the x coordinate of each experimental point, the closest numerical points were selected. The mean Cp value of these points was used to calculate the relative error between numerical and experimental results. By analysing both Fig. 3.9 and Table 3.4, the good agreement between numerical and experimental results can be seen. The calculated errors are very small, except for the coordinate closer to zero meters and those closer to one meter. The Cp of the points closer to x = 1m are far from the experimental value due to the mesh transition between the CORE COWL and the JET boundaries. In that region there is a transition from a region with prismatic layer to a region without prismatic layer. Since the computational settings used were the same in both simulated models it is expected that the results obtained in the Airbus model to be as suitable as the TPS one’s.
3.7 CFD Results and Discussion
Four flight conditions were simulated in order to determine the critical flight condition to which the maxi- mum aerodynamic load is associated. The results of these simulations are presented and discussed in the present section.
39 Figure 3.9: T.P.S Cp distribution for computational and experimental results.
It is important to refer that despite the fact of having a symmetry plane on the engine geometry, all computational simulations were performed using the complete geometrical model. Computational effort could have been employed to generate a finer mesh for half the geometry, rather than being used in a coarser mesh for the complete engine model. Although, the complete geometrical model brings an advantage for the following part of this work, the Finite Element Method (FEM) simulations. The use of the complete model allows to import both the complete pressure field and the wall shear stress distributions into the FEM analysis. To evaluate the differences between using the half engine model and the complete model with the same amount of mesh cells, a simulation was performed for the Takeoff conditions with 0o angle of attack. Both the pressure and wall shear stress were compared for the half and complete model. In Fig.3.10, both the half and complete models are presented.
(a) (b)
Figure 3.10: Model configurations. (a) - Half Model Configuration; (b) - Complete Model Configuration.
From the result it could be seen that the relative error was less than 3% when compared to the most refined model. It could also be seen, for both the pressure and wall shear stress, that the range of values of the half model (most refined) was within the range of values of the complete engine model. When
40 Table 3.4: T.P.S Pressure Coefficient comparison between numerical and experimental results.
x coordinate [m] Cp experimental Cp numeric Relative Error [%]
0 0.0466 0.0122 73.8095 0.0139 -0.9870 -0.9908 0.3738 0.0279 -0.8366 -0.8236 1.5541 0.0279 0.9343 0.9161 1.9473 0.0558 -0.6715 -0.6736 0.3154 0.0837 -0.62104 -0.6155 0.8920 0.0837 0.5234 0.5302 1.3001 0.1116 -0.6692 -0.6589 1.5437 0.1395 -0.6899 -0.6949 0.7243 0.1860 -0.7239 -0.7277 0.5208 0.1860 0.2941 0.3003 2.1054 0.2558 -0.7714 -0.7782 0.8893 0.3721 -0.7409 -0.7591 2.467 0.3721 0.1566 0.1542 1.4771 0.6976 0.1803 0.1642 8.9428 0.9302 0.0058 0.8958 79.1188 1 -0.2300 -0.1348 41.3902 analysing the wall shear stress values a larger difference is evident between both models. Although comparing the magnitude of the wall shear stress with that of the pressure it can easily be seen that the wall shear stress maximum value is about 1% of the pressure one. Using the complete model it can be assumed that FEM analysis are perform in a conservative way, since the range of values is larger than that of the half model. The results used in the comparisons above made are presented in Table 3.5.
Table 3.5: Half and complete model comparison.
Property Complete Model Half Model Relative error [%]
Maximum Pressure [Pa] 14815 14419 2.75 Minimum Pressure [Pa] -11757 -11444 2.74 Maximum Wall Shear Stress [Pa] 126.75 100.55 26.05 Minimum Wall Shear Stress [Pa] 0.15414 0.16821 9.98
All simulations performed ran until the stopping criteria was attained. The control parameters in the present work were the residuals, the mass flow and the y+. The y+ was used to control the mesh quality, in Fig.3.11, it can be seen that all mesh y+ are below 200, as desired. Both the residuals and
41 mass flow were used to control the solution convergence. The residuals are the measure of the absolute error of the discretized equations used in the iterative solution process. In the present work the solution was considered to be converged when all the residual had values bellow 10−4. Although, the mass flow variation was analysed in parallel with the residual to ensure that this important quantity had also converged. Since the mass flow is the main parameter used to set the engine working conditions it is important to have a stabilized mass flow value.
Fig. 3.12 and 3.13, show that the results of all simulations are within the defined control parameters. This is, in all flight conditions the residuals are below 10−4 and the mass flow converged to the desired values, approximately 400 kg/s for the takeoff conditions and approximately 110 kg/s for cruise.
From the result obtained from the four flight conditions, four main parameters are presented and dis- cussed. The velocity vectorial field, the mach number distribution at the fan, the pressure field and the wall shear stress, on the surfaces of interest.
The velocity field for the four simulations is presented in Fig.3.14. Fig.3.14, corresponds to the velocity field at the midsection plane of the domain. The two-dimensional representation of the nacelle allows to create a visual analogy between the engine inlet and an airfoil.
At takeoff with 0o it can be seen that the flow is well behaved, no phenomena like separations or shock waves are present. It can also be seen that flow is decelerated in the diffuser zone. The flow is ac- celerated both in the lower and upper LIP. This can be visualized around the LIP leading edge. As the angle of attack increases the velocity also increases around the leading edge, as expected to occur in an airfoil. At the angle of attack of 9o the flow still behaves uniformly. Increasing the angle of attack to 16o massive alterations appear. Both at the upper and lower part of the nacelle, flow separation occurs. On the upper part the separation only affects the external flow, however, on the lower part the flow di- rectly affects the inlet region. As it can be seen a recirculation section is present in the lower separated region, and propagates up to the fan. The flow distortion at the inlet usually has a negative impact on the engine performance [50]. For the three above conditions the stagnation point appears on the inner part of the upper lip and exterior part of the lower lip. These points migrate respectively to inner and outer parts of the lip as the angle of attack increases. For the three above conditions all points on the flow are subsonic, since the free stream Mach number is approximately M=0.438, far from the transonic regime. For the cruise condition it can be seen that the stagnation points are now in the interior region of both upper and lower lip. Although the flow has a uniform behaviour at the inlet, the exterior is affected by shock waves. The free stream mach number is of M=0.847, clearly in the transonic region where this type of phenomena are expected as the flow is accelerated around the lip.
In Fig.3.15 the Mach number at the fan is presented for the four flight conditions. Typical values for fan Mach number are between 0.4 - 0.7 [5]. It can be seen that for the three takeoff conditions the fan Mach number is within the typical values. For the takeoff at 0o a completely uniform Mach distribution on the fan is visible. This is expected by the fact that the flow is aligned with the engine. For the condition with angle of attack 9o the distributions is mostly uniform. Some small regions of higher velocity appear at the
42 bottom. At 16o the flow separation is clear, the Mach distribution is highly affected by this phenomena. This can be seen at the bottom where the Mach number is very close to zero. Flow distortions effect the engine’s performance, as referred to in [50]. The cruise condition shows a uniform Mach distribution at the fan. The Mach number is slightly below the typical values, maybe due the approximations made to the engine mass flow for this flight condition.
Analysing the pressure distribution, the effect of the diffuser can be seen, please see Fig. 3.16, for the 0o takeoff condition. As the velocity is reduced the pressure increases in the duct. For the three takeoff conditions the maximum pressure can be found around the lip leading edge, varying slightly with the angle of attack. With the increase of the angle of attack a higher pressure region appears at the upper part of the inlet as well as on the lower exterior region. Suction regions appear on the regions where the flow is accelerated. The maximum pressure has about the same magnitude for the three takeoff conditions, due to the fact of being used the same velocity and density in these simulations. The minimum pressure increases in magnitude as the angle of attack increases, due to a further acceleration of the fluid around the leading edge. For the cruise condition the maximum pressure can be visualized around the leading edge. On the external region the minimum pressure can be verified on the region where the flow is accelerated. Due to the appearance of shock waves on this flight condition a large pressure variation exists on the external region. The shock wave phenomena has a rising pressure effect after the shock, this can be identified on the external upper section. The pressure distribution for the four flight conditions is represented in Fig.3.16. In Table 3.6 the maximum and minimum pressure result are presented for the four flight conditions.
Table 3.6: Pressure variation for the simulated flight conditions.
Property Takeoff 0o Takeoff 9o Takeoff 16o Cruise
Maximum Pressure [Pa] 14815 14998 14798 13541 Minimum Pressure [Pa] -11757 -38840 -49922 -14071
Looking at Figure 3.17 it is visible that the wall shear stress is much smaller in the regions associated to small velocities, as the regions close to the stagnation points. It is visible that in regions of higher velocity the wall shear stress has also higher values. In the case of 16o takeoff, the separations regions are clearly characterized by the low velocity regions, with small wall shear stress represented in dark blue. In Table 3.7 the maximum and minimum pressure result are presented for the four flight conditions.
Table 3.7: Wall Shear Stress variation for the simulated flight conditions
Property Takeoff 0o Takeoff 9o Takeoff 16o Cruise
Maximum Wall Shear Stress [Pa] 94.41 153.14 166.69 66.53 Minimum Wall Shear Stress [Pa] 0.15 0.22 0.13 0.16
Comparing the pressure with the wall shear stress it is clear that the influence of the wall shear stress is much smaller than that of the pressure. Comparing both, the wall shear stress is about 1% of the
43 pressure magnitude. The wall shear stress has an almost irrelevant contribution to the total aerodynamic load applied on the engine inlet.
(a) (b)
(c) (d)
Figure 3.11: Wall y+ for the four flight conditions. (a) - Takeoff 0o; (b) - Takeoff 9o; (c) - Takeoff 16o; (d) - Cruise.
44 (a) (b)
(c) (d)
Figure 3.12: Residuals evolution for the four flight conditions. (a) - Takeoff 0o; (b) - Takeoff 9o; (c) - Takeoff 16o; (d) - Cruise.
45 (a) (b)
(c) (d)
Figure 3.13: Mass flow convergence for the four flight conditions. (a) - Takeoff 0o; (b) - Takeoff 9o; (c) - Takeoff 16o; (d) - Cruise.
46 (a) (b)
(c) (d)
Figure 3.14: Velocity field for the four flight conditions. (a) - Takeoff 0o; (b) - Takeoff 9o; (c) - Takeoff 16o; (d) - Cruise.
47 (a) (b)
(c) (d)
Figure 3.15: Mach number at the fan for the four flight conditions. (a) - Takeoff 0o; (b) - Takeoff 9o; (c) - Takeoff 16o; (d) - Cruise.
48 (a) (b)
(c) (d)
Figure 3.16: Nacelle pressure distribution for the four flight conditions. (a) - Takeoff 0o; (b) - Takeoff 9o; (c) - Takeoff 16o; (d) - Cruise.
49 (a) (b)
(c) (d)
Figure 3.17: Nacelle wall shear stress distribution for the four flight conditions. (a) - Takeoff 0o; (b) - Takeoff 9o; (c) - Takeoff 16o; (d) - Cruise.
50 Chapter 4
FEM Methodology - Analysis of the Joints of the Acoustic Panel
In the present chapter the methodology used to determine the mechanical behaviour of the joint of the acoustic panel and the fasteners of the joint is presented. The description of the components of panel is made and the different approaches to the problem are presented. All simulations presented in this section were performed using ANSYS Workbench software.
Results from the CFD methodology chapter are used here, in order to correctly simulate the aerodynamic loading on the structure.
4.1 Inlet Cowl Components
The inlet cowl is a structure composed by several components. Each component has a specific function. Different types of materials are used in their construction.
The inlet cowl of the A320/321, equipped with CFM56-5B engines, can be divided in four main sections, the Lip, the Bulkheads, Outer Barrel and Inner barrel. Both the Airbus A320 and A321 share the same Inlet Cowl model.
• The lip works in similar fashion to the leading edge of an airfoil. It is simultaneously connected to forward bulkhead, outer barrel and inner barrel. The lip is made of aluminium.
• Both the forward and aft bulkhead have the role of introducing stiffness to the structure. Bulk- heads are made of titanium.
• The outer barrel creates a geometric continuity between the lip and the fan cowls and it is attached to both the forward and aft bulkheads. The outer barrel is made of composite materials.
51 • The inner barrel is composed by three acoustic panels. The acoustic panels are made of several components and materials. The acoustic panel are sandwich components composed by an alu- minium honeycomb core, aluminium inner skin and by an outer skin made of perforated aluminium. The three acoustic panels have different sizes and shapes, due to the scarf angle of the inlet cowl.
The inlet cowl of the Airbus A320/A321 and its components are represented in Fig. 4.1.
The inlet cowl is attached to the aircraft’s engine by means of a titanium attachment ring. The acoustic panels are fastened to the attachment ring which in turn is connected to engine. The fasteners used in the connection are Hi-Lok fasteners.
The joint intended to be analysed, is the joint created by the interface between the acoustic panel and the attachment ring.
(a) (b)
Figure 4.1: Inlet Cowl Components. (a) - Inlet Cowl External View; (b) - Inlet Cowl Internal Section, modified from [59].
4.1.1 The Acoustic Panel Joint
The acoustic panel’s joint is one of the main components under analysis. In order to clarify what section of the panel corresponds to the that joint, a figure with two views of the acoustic panel is presented, Fig. 4.2. In Fig. 4.2 (b), the main components of the joint are presented. Both the aft doubler and the internal doubler are components that make contact with the attachment ring. One of the goals of this work is to analyse the joint behaviour at the interface between the internal doubler and internal honeycomb core, as well as the mechanical behaviour of the fasteners in the joint.
4.2 Preparing the FEM Simulations
In order to understand the effect of the aerodynamic load on the acoustic panel’s joint and fasteners, several FEM simulations were performed. Depending on the type of information intended to be obtained some modifications have to be done in the simulations.In all simulations, parameters such as the aerody- namic load, geometry importation, material selection, contacts, solvers, mesh, and bondary conditions
52 (a)
(b)
Figure 4.2: Acoustic panel. (a) - Acoustic Panel side section view; (b) - Joint main components. must be defined. In the present section the transversal steps used in the preparations of the simulations are presented.
4.2.1 Aerodynamic Load
In all the simulations of this work, the structure is disturbed by an aerodynamic load. Since the aerody- namic loads were obtained with Star CCM+ software, they had to be imported to ANSYS Workbench. The latter software has the capability to import external data. The external data can be imported to the software by mean of the component ”External Data” available in the software Toolbox. This tool has the capability to import data from external files. When defining the data to be imported the user has several option. The software allows to define the type of coordinate system, the type of data (Pressure, Temperature, Force, displacement, others) and to associate the unit type to it.
In the present analysis both the pressure and wall shear stress were imported to ANSYS. The files correspond to the four fight conditions, takeoff 0o, 9o, 16o and cruise. The pressure file is provided with the pressure magnitude and the coordinates corresponding to its application points. The wall shear stress file is provided with the three components of the wall shear stress and a coordinate system associated to each imported mesh point. The units of the system must be defined when importing the data. Attention must be made to verify if the imported coordinate system is coincident with the ANSYS one.
53 4.2.2 Geometry and Geometry Importation
In order to generate the structural analysis, a CAD representation of the model must be created and imported into ANSYS environment. ANSYS Workbench has its own inbuilt modelling software, the DesignModeler. Although, all the geometries used in the analysis were created using Solidworks. The CAD files were imported in Parasolid format. Some adjustment were made to the geometry in order to simplify the mesh generating process. The adjustments were made using DesignModeler. In order to reduce the complexity of the geometry some simplifications were made to model. Since the objective of the present work is to analyse the joint of the acoustic panel, special attention was made to correctly simulate the components in that region. In Fig. 4.1 (a) and (b), two different views of the inlet cowl are presented. The three sections corresponding to the acoustic panels are visible with three different colors. In order to reduce the number of contact regions, the outer barrel, the forward bulkhead and inner barrel were all made as a single component. The region of the joint is modelled with all the correct components. The aft bulkhead is not modelled, this decision reduces the rigidity of the joint, making the analysis more conservative. The modelled sections are presented in Fig. 4.3 (c) and (d), with and without attachment ring, respectively.
(a) (b)
(c) (d)
Figure 4.3: Inlet Cowl geometry. (a) - Inlet Cowl isometric view; (b) - Inlet Cowl rear view; (c) - Inlet Cowl internal view; (d) - Inlet Cowl internal view with Attachment Ring.
54 4.2.3 Materials and Properties
As referred in section 4.1, several materials are used in the construction of the inlet cowl and acous- tic panel. The material properties of the components used in the simulations are presented here. All material properties must be introduced in the software. As referred in the previous section some simpli- fications were made to the model. The section that results from the above combination of components is assumed to be made of aluminium 2024-T3, the doublers of the joint are made of aluminium 2024-T3. The Hi-lok pins are made of A-286, an Iron base superalloy. The collar and washer are made of AISI 303 stainless steel. The honeycomb core is made of aluminium 5052. Finally, the attachment ring is made of titanium Ti-6Al-4V. In Table 4.1, the isotropic material properties are presented. All proper- ties were obtained form CES EduPack 2015 Software, except those of the honeycomb core, which are calculated based on the geometrical parameters. Honeycomb properties are presented in Tab.4.2, for the referential used in ANSYS. The Honeycomb is modelled as an orthotropic material. The Honey- comb properties calculations are presented in the next section. When defining an orthotropic material in ANSYS Workbench, a new referential must be created and assigned to the material.
Table 4.1: Isotropic material properties from CES2015 software.
Material Al 2024-T3 A-286 AISI 303 Ti-6Al-4V Al 5052
Density [kg/m3] 2750 7900 7900 4410 2670 Young’s Modulus [GPa] 72 201 192 110 70 Poisson’s Ratio 0.33 0.33 0.25 0.31 0.33 Tensile Yield Strenght [MPa] 248 586 204 786 66 Tensile Ultimate Strength [MPa] 359 896 528 862 172
4.2.3.1 Defining the Honeycomb Properties
The honeycomb core used in the acoustic panels is made of aluminium and has hexagonal cells shape, Fig. 4.4. The cell size is 3.175 mm (1/8 in), the aluminium is the 5052 and the foil thickness is 0.0762 mm (0.003 in). Honeycomb can be found in a variety of material and cell shapes. The manufacturing process of honeycomb generates different in-plane mechanical properties. This is due to the different cell wall thickness. Some walls have twice the thickness of the others, in a specific direction. More information about the manufacturing process of honeycombs can be found in [54]. Honeycomb out-of-plane stiffness and strength are much larger than the in-plane ones. In-plane stresses tend to bend the cell walls. In the out-of-plane direction compression or extension collapse stresses are much larger.
Manufacturer properties for this specific honeycomb are presented in Table 4.3. Taking a closer look at these properties, it is clear that the approximations made to the Poisson ratio and Young’s modulus are too severe. In order to obtain more realistic and conservative properties, the calculation method presented in [56] is used. In this calculations, it was assumed that the the honeycomb has regular cells,
55 Table 4.2: Honeycomb core orthotropic properties.
Material Honeycomb Alloy 5052
Density [kg/m3] 192 Young’s Modulus X direction [GPa] 3.36 Young’s Modulus Y direction [GPa] 0.0161 Young’s Modulus Z direction [GPa] 0.0161 Poisson’s Ration XY 0.33 Poisson’s Ration YZ 0.99 Poisson’s Ration XZ 0.33 Shear Modulus XY [Pa] 6.476E8 Shear Modulus YZ [Pa] 2.903E6 Shear Modulus XZ [Pa] 6.476E8 Tensile Yield Strength [MPa] 66 Tensile Ultimate Strength [MPa] 172
(a) Hexagonal Honeycomb Configuration. (b) Honeycomb Coordinate System.
Figure 4.4: Honeycomb Cell Configuration and Coordinate System, from [55]
this is, cells with the same wall thickness, internal angles of θ=30o and l = h, please see Fig. 4.5. If the cells are regular, the in-plane properties are isotropic. In such cases, the in-plane properties are defined by E∗ and G∗, respectively the honeycomb equivalent Young’s modulus and equivalent Shear modulus. The simplified equations for the in-plane properties of regular cells are presented bellow from Eq. (4.1) to (4.5). The out-of-plane properties are calculated with equations from Eq. (4.6) to Eq. (4.8).
ρ∗ 2 t = √ (4.1) ρs 3 l
E∗ E∗ t3 X = Y = 2.3 (4.2) Es Es l
56 2 ∗ εY cos θ νXY = − = (4.3) εX (h/l + sinθ)sinθ
∗ εX (h/l + sinθ)sinθ νYX = − = 2 (4.4) εY cos θ
G∗ G∗ t3 1 E∗ XY = YX = 0.57 = (4.5) Es Es l 4 Es
E∗ h/l + 2 t ρ∗ Z = = (4.6) Es 2(h/l + sinθ)cosθ l ρs
∗ ∗ νZX = νZY = νs (4.7)
G∗ G∗ t XZ = YZ = 0.577 (4.8) Gs Gs l
Figure 4.5: Cell Nomenclature [56]
In the equations from Eq.(4.1) to (4.8), the indexes X, Y and Z, represent respectively, the three refer- ential components (Fig.4.4 (b)). The index s represent the properties of the solid material from which the Honeycomb is made. ρ, E, ν, ε and G represent respectively, the density, the Young’s modulus, Poisson’s ration, strain and shear modulus.
In Table 4.2 the properties used to model the Honeycomb are presented. The properties correspond to those associated to the Honeycomb referential used on Workbench, please see Fig.4.6.
Figure 4.6: Referential of the honeycomb properties.
57 Table 4.3: Manufacturer honeycomb properties and approximations[55]
Material Honeycomb Alloy 5052
Density [kg/m3] 192 Young’s Modulus X direction [GPa] ≈0 Young’s Modulus Y direction [GPa] ≈0 Young’s Modulus Z direction [GPa] 6.205 Poisson’s Ration XY ≈0 Poisson’s Ration YZ ≈0 Poisson’s Ration XZ ≈0 Shear Modulus XY [Pa] ≈0 Shear Modulus YZ or w [Pa] 537.8E6 Shear Modulus XZ or L [Pa] 1447.9E9
4.2.4 Contact Between Components
Whenever two or more components surfaces touch each other, these components are said to be in con- tact. When contact exists it is assumed that the components do not interpenetrate each other. To simu- late different contact situations, ANSYS Workbench has different types of contact available. Bonded, no separation, rough, frictional, frictionless and forced frictional sliding, are the available types of contact. In the simulations of the present work the two types of contact used were bonded and frictionless. These two types of contact are briefly explained, information about the other types of contact can be found in the software ANSYS Workbench help. When using bonded contact it is assumed that components in contact are glued together. No sliding or separation is allowed between components. Bonded contact was used in the acoustic panel components. The frictionless contact allows components to slide and to separate from each other. Unlike the bonded contact type, frictionless contact has nonlinear formu- lation that results in longer solutions time. This type of contact is used to simulate contact between the acoustic panel doublers and the attachments ring. It is also used to simulate the contact between the fasteners shank and the Joint’s holes.
When selecting a type of contact care must be taken to select a compatible formulation type. The recommended type of formulation for the frictionless contact is the Augmented Lagrange formulation. The interface treatment - ”Adjust to Touch” was also select in the simulations. This property creates an offset to close existing gaps between component and establish initial contact. This adjustment is based in the pinball region. The pinball region is a spherical boundary that surrounds each contact detection point. The sphere radius can be defined by the user. This radius sets the distance in which components are assumed to be in contact. In Fig. 4.7, several pinballs spheres of action are represented associated to multiple connections.
58 Figure 4.7: Several pinball spheres in the joint.
4.2.5 Defining the Mesh
When making structural analysis in ANSYS Workbench, five main meshing methods are available for solids. These methods are based on basic building blocks, tetrahedrons, hexahedrons, pyramids and prisms. Depending on the type of method selected, some building blocks will be used in the construction of the mesh. The five main methods are, the patch conforming method, the patch independent methos, the sweep method, multizone and hex-dominant. Both the patch conforming and patch independent methods use mainly tetrahedron elements. The sweep method generates an hexahedral mesh. A very clean CAD must be used in order to use this method. When comparing a tetrahedral with an hexahedral one, the hexahedral mesh is more efficient. It requires much less element than a tetrahedral mesh to obtain the same solution accuracy. Although an hexahedral mesh require more time to be generated. The multizone method consists in a method that automatically decomposes the geometry into sweepable regions and free mesh regions. Finally, the hex-dominant method creates a mainly hexahedral mesh. This method is used if the sweepble method can’t be applied. In the present work, the hex-dominant method was used.
As the computational resources are limited, local refinements were used to increase the mesh density in locals of interest. In all simulations, a global mesh size was defined. Then, depending of the analysis type, local refinements were created. The local refinements were done with the used of both edge sizing and body sizing functions. Edge sizing was used to refine the fastener’s mesh, by dividing the edges, of the fasteners and fasteners hole, in a desired number of sections. The body sizing was used to size the joint’s members. This was done by defining the element size in the desired component.
When defining the mesh in ANSYS Workbench it is possible to define the order of the elements used in the simulation. This is done by defining the midside nodes as kept or dropped. If kept is selected quadratic elements are used, otherwise linear elements will be used. For 2D and 3D solid bodies analysis the kept option is used as default. This option was maintained for the simulations.
The information presented in this section can be found in ANSYS Workbench software Help.
59 4.3 Determination of the Critical Load Condition
Four flight conditions were simulated using CFD. In order to determine the critical loading condition, the results of the four simulations were imported to ANSYS Workbench. Both the wall shear stress and pressure were imported. An analysis of the force reaction and moment reaction is made to determine the critical loading conditions.
4.3.1 Model Considerations
In the present analysis the complete model of the inlet cowl was used in the analysis. In order to reduce the simulation complexity and the computational effort, no fasteners were used in the simulation. In doing so, nonlinear formulation is avoided and simulation time is reduced. Bonded contact is used in all components. Since only reaction forces and moments are to be obtained, the use of Bonded contact doesn’t compromises the results.
Each material property was assigned to its correspondent component. A new coordinate system was created to allow the introduction of the orthotropic properties of the honeycomb.
All simulations were performed with the same geometry and mesh. Doing so, there is a base of com- parison between the simulations. The mesh used was of the hex dominant type, with 305136 elements. No local refinements were used. The typical element size was 1.5e-2 m. See Fig. 4.8 (a).
For each simulation, the aerodynamic loadings (wall shear stress and pressure), were imported into the model. See Fig. 4.8 (c) and (d).
To ensure that the geometry is held in place, a constraint is imposed at the attachment ring. The constraint is defined as a fixed support in Ansys Workbench. This simulates the real connection between the inlet cowl and the engine. See Fig.4.8 (b).
In order to understand if the weight of the structure would be a critical factor in the analysis, a simulation with only the structure weight was also created.
4.3.1.1 Results and Discussion
Reaction forces and moments corresponding to the analysed loading conditions are presented in Table 4.4. As it can be seen by inspection of Table 4.4, as the angle of attack increases, both forces and moments reaction increase. It is possible to conclude that the loading corresponding to the Takeoff 16o flight condition corresponds to the critical loading condition. When analysing the results corresponding to the structural weight condition (Table 4.4), it is visible that all components of the reaction force and moment reaction, have the opposite sign when compared to the other analysis made. This means that the addition of the structural weight reduces the global loading.
60 In Table 4.5, convergence analysis of the mesh is made. This was made in order to analyse the influence of the mesh element size in both the force and moment reactions. The limit of the refinement was set by the computational capability, this is, the memory limit.
From the analysis of Table 4.5, it is possible to verify that despite fluctuation of the total reaction force and moments, their values do not vary in order of magnitude. A maximum difference of 51N from the most refined solution, from the force reaction. A maximum difference of 61N.m for the moment reaction. These differences aren’t significant when compared to the total magnitude. In Fig. 4.9, the relative error between consecutive mesh refinements is presented, both for the force and moment reactions. From this figure it is possible to verify that the error between consecutive mesh refinement is below 1% and that the error is decreasing. The small error and its decreasing in value is a good indicator of the convergence of the results.
(a) (b)
(c) (d)
Figure 4.8: Different simulation setting for the determination of the critical load. (a) - Mesh configuration used in the model; (b) - Geometry constrain - Fixed Support; (c) - Pressure Dis- tribution for Takeoff 16 condition; (d) - Wall Shear Stress distribution for Takeoff 16 condition.
4.4 Approach to Analyse the Acoustic Panel Joint
In order to obtain enough resolution to capture the desired properties, both for the fastener of the joint and the interface between the internal honeycomb and internal doubler (see Fig. 4.2), some simplifica-
61 Table 4.4: Moment and and reaction forces for different loading conditions
Force Reaction [N]] Moment Reaction [N.m] Loading Condition X Y Z Total X Y Z Total
Takeoff 0o 52.8 -834.3 -446.1 947.5 2128 0.8 -4.8 2128 Takeoff 9o 157.8 -21035 -2237.5 21154 14476 132.2 -1.4 14476 Takeoff 16o 216.4 -34512 -3549.1 34695 20236 103.9 -8.5 20236 Cruise 18.9 -1690.6 -7175.2 7371.7 6168.4 1.5 -2.5 6168.4 Structural Weight -2.23E-5 1380 -6.50E-3 1380 -1490.4 -5.4E-3 3.1E-4 1490.4
Table 4.5: Force and moment reaction convergence analysis.
Force Reaction [N] Moment Reaction [N.m] Mesh No Elements X Y Z Total X Y Z Total
131933 215.6 -34437 -3333.6 34599 20172 114.7 -9.1 20172 305136 216.4 -34512 -3549.1 34695 20236 103.9 -8.5 20236 692027 239.9 -34440 -3611.2 34630 20199 106.7 -10.4 20199 895203 245.1 -34469 -3644.8 34662 20175 112.4 -10.2 20175 1163624 230.1 -34452 -3638.6 34645 20153 106.8 -9.2 20153 1589863 248.7 -34456 -3649.3 34650 20175 116.8 -10.1 20175 tions had to be done to the model.
The three acoustic panels, that compose the inner barrel, have different sizes. The joint of each panel has several fasteners holes. The number of hole varies from panel to panel due to their different sizes. The total of fasteners holes is 304. The holes create a pattern on the joint. Through a visual inspection it was possible to identify 14 patterns on the lower acoustic panel, 13 and 11, on the upper and side acoustic panels respectively. Some measurements were made in order to determine the pattern main dimensions. In Fig. 4.10, the pattern is presented as well as it dimensions.
Due to the high number of fasteners used to connect the inlet cowl to the attachment ring, it is not possible to simulate the complete geometry. The number of interface contacts and the nonlinearity of the formulation makes this analysis impossible with available the resources. Some simplifications had to be made to the model in order to analyse the Joint’s behaviour.
4.4.1 Model Simplifications
In order to simplify the model, the equivalent stress of the complete bonded model was analysed. The loading conditions used in the simulations were the critical ones, Takeoff 16o. From these analysis it was possible to identify the region with higher stress levels. This region corresponds to the interface region
62 Figure 4.9: Force and Moments relative error as function of the mesh refinement.
(a) (b)
Figure 4.10: Joint’s Fasteners Pattern. (a) - Fasteners pattern on acoustic panel and Attachment Ring; (b) - Main dimensions of the fasteners pattern. between the lower and side acoustic panels. As the most stressed region comprises these two panels, it was decided to create a section of the model that would include a fastener pattern on each side of the interface. Both, the most stressed region and sectioned model, are presented in Fig 4.11 (a).
As the inlet cowl was sectioned, new boundary conditions had to be implemented in order to simulate the complete structure behaviour. The approach used was to simulate the rigidity of the complete model with the use of springs. The model is explained next. The model is presented in Fig. 4.11 (b).
• The maximum displacement of the complete model was determined. The maximum displacement of the complete model is 7.12E-5m.
• The reaction force on both lateral faces of the section were calculated. The reaction forces corre- spond to the Takeoff 16o loading condition.
63 • knowing both the displacement and the force reaction on each face of the section, it is possible to determine a spring stiffness for each face. The spring stiffness is calculated with the use of the Hooke’s Law, Eq. (4.9).
F = −kx (4.9)
Where F is the applied force, K is the spring stiffness and x is the displacement. The stiffness of the spring used in the Lower Panel face is 1.186E8 N/m, the stiffness on the Side Panel face is 9.206 N/m.
• Knowing the stiffness of the two springs and the components of the reaction force, the springs were created and aligned to work in the line of action of the reaction forces.
(a) (b)
Figure 4.11: Model simplifications. (a) - Inlet cowl most stressed region; (b) - Spring model and lateral faces of the section.
4.5 Analysis of the fasteners of the Joint
In the present section the analysis of the fasteners of the joint is made. One of the main objectives of the present work is to understand the mechanical behaviour of the fasteners of the Joint, when subjected to the critical aerodynamic loading. The results obtained in the simulations are compared to theoretical results, in order to understand how far from the material limit the fasteners get when loaded. Results of the simulations are presented and discussed.
4.5.1 Components Contact and Mesh Refinement
Due to the aerodynamic loading conditions, this is, simulating the airflow approaching the inlet with an angle of attack of 16o, the inlet tends to bend. The sections bend in such way that the acoustic panels
64 tend to separate from the attachment ring, tensioning the Hi-Lok fasteners. The movement of the joint can be seen in Fig. 4.12. Since the Hi-Loks are under tension, bonded contact was used between the pin head and doubler, and between Hi-Lok collar and the attachment ring. The contact type used between the Hi-Lok shank and the holes of the Joint, was the Frictionless contact type. Frictionless contact was also used between the attachment ring and the doublers of the acoustic panel. Doing so, separation between the attachment ring and the panel is possible, tension and shear forces can be transferred to the fastener. Once again the refinement of the model is conditioned by the computer resources. Since in these simulations we intend to analyse the mechanical behaviour of the fastener, the mesh was locally refined. In all simulations, the mesh size of the different components of the acoustic panel were maintained the same. Only the the number of elements in the mesh of the fasteners varied between different refinements. The refinement was made by defining the element size of the fastener, which was kept the same between refinements, and then by increasing the number of elements around the edges of the fasteners and holes. In Fig. 4.13, the difference between the lowest and most refined meshes is presented. In Table 4.6, information about the fasteners refinement is presented.
Figure 4.12: Geometry displacement due to aerodynamic load.
Table 4.6: Mesh parameters of the fasteners.
Refinement 1 2 3 4 5 Edge Divisions 5 10 15 20 25 Mesh No Elements 132150 134838 143908 208108 274881
4.5.2 Convergence Analysis
In order to analyse the convergence of the results related to fasteners simulations, the maximum equiv- alent stress (von Mises) was analysed. The bolt with maximum equivalent stress was used to make the convergence analysis. Bolt65 5 was the bolt with maximum stress. The bolts nomenclature and position can be seen in Fig. 4.14. In the Hi-Lok reference the numbers 15, 32 and 65, represent the distance
65 (a) (b)
Figure 4.13: Fasteners mesh refinements. (a) - Fasteners Refinement 1; (b) - Fasteners Refinement 5. of the fastener hole form the end edge of the panel. As at each refinement the position of the nodes is changing, it was decided to create a referential at the maximum stress node. The coordinates of the referential were used to set the same referential in each refinement, allowing to be constantly analysing the same point on the bolt.
As it can be seen from the analysis of Table 4.7, there still exist a fluctuation on the stress values. Although, in the last three refinements the stress value are close and the error between consecutive refinements is below 1%. Refinement 5 was used to obtain the result for the bolts analysis.
Table 4.7: Bolts convergence analysis
Refinement No Elements von Mises [Pa] Error%
1 132150 1.206E7 1.074 2 134838 1.219E7 6.084 3 143908 1.298E7 0.038 4 208108 1.299E7 0.964 5 274881 1.287E7 -
4.5.3 Result and Discussion of the Fasteners Analysis
The result presented next are based on the behaviour of the fastener when subjected to the aerodynamic load. From the analysis of the displacement of the model it was possible to verify that the Hi-Lok are subjected to combined loads (see Fig. 4.15 (d)). Therefore, both tension and shear stresses must be analysed. In this section the tension and shear of the Hi-Loks is analysed and compared to the material yield strength. The tension stresses are also compared to the preload of the Hi-Loks.
66 Figure 4.14: Hi-Lok nomenclature and position.
Fasteners in the joint have different diameters. The Hi-Lok65 1 to Hi-Lok65 6, have a diameter of 4.76mm (3/16in). All the other Hi-Lok have a diameter of 6.35mm (1/4in). All results presented cor- respond to the simulations with refinement 5.
4.5.3.1 Analysis of the Tension Stress and Shear Stress
In Table 4.8, σmax represents the Hi-Lok maximum tensile stress, τxymax represents the maximum shear stress, S.F.σ and S.F.τ represent respectively the safety factor for tension stress and shear stress. The in tension yield strength can be found in Table 4.1, and the in shear yield strength can be related to the yield strength in tension with the von Mises relation, Eq. (4.10). More information about von Mises criteria can be found in [34].
τyield = 0.577σyield (4.10)
From the analysis of the results presented in Table 4.8, it is possible to verify that Hi-Lok65 4 and Hi- Lok32 1, are the fasteners that show higher normal stress, respectively for its diameter range. The fasteners with higher shear stress are the Hi-Lok65 5 and Hi-Lok32 1. When comparing these results to the yield strength limits it is visible that the stresses are much lower than those that could create the fastener failure. The minimum fastener factor of safety is 47 for the tension analysis and 246 for the shear analysis.
In Fig. 4.15(a) a model of the joint is presented. In Fig. 4.15(b) the honeycomb core was removed to allow the internal visualization of the joint. In Fig. 4.15 (c) a representation of the joint only with the doublers and attachment ring. This figure allows to verify that the regions of the fasteners with higher stress, correspond to the regions of contact of the doublers and perforated skin of the Panel, with the fasteners. As the attachment ring is a fixed structure, the presence of an external load creates a bending moment on the Panels. The bending moment tends to create a rotation on the panel, making the fasteners bend.
67 Table 4.8: Hi-Lok fasteners analysis
Hi Lok σmax τxy S.F.σ S.F.τ
Hi Lok65 1 6.88E+06 9.26E+05 85 365 Hi Lok65 2 6.87E+06 9.41E+05 85 359 Hi Lok65 3 7.56E+06 6.62E+05 78 511 Hi Lok65 4 1.25E+07 4.94E+05 47 684 Hi Lok65 5 1.22E+07 1.31E+06 48 259 Hi Lok65 6 1.07E+07 1.26E+06 55 268 Hi Lok32 1 1.18E+07 1.38E+06 50 246 Hi Lok32 2 4.49E+06 6.11E+05 131 553 Hi Lok32 3 7.83E+06 9.56E+05 75 354 Hi Lok32 4 8.54E+06 9.29E+05 69 364 Hi Lok15 1 1.03E+07 1.30E+06 57 260 Hi Lok15 2 6.36E+06 6.82E+05 92 496 Hi Lok15 3 6.94E+06 1.09E+06 84 309 Hi Lok15 4 5.82E+06 7.57E+05 101 447
4.5.3.2 Pretension Results
As already referred, Hi-Lok fastener with two different diameters are used in the main joint. When installing the Hi-Lok, a section of the collar shears when a particular torque value is obtained. This torque sets the preload on the fastener. The Hi-Loks used in the joint are composed by the pin and collar. Pin with references HL633-6 (diameter 3/16in) and HL633-8 (diameter 1/4in) were used. The collars used are HL585-6A and HL585-8A, that correspond respectively to the Pins. In the data sheet of the collars information about the shearing torque of the collar [57]. A range of toque of 3.4 - 4.5 Nm (30- 40 inch-pounds) is associated to the HL585-6 Collar, a torque range of 6.8 - 9 Nm (60-80inch-pounds) is associated to the HL585-8 collar. At TAP facilities it was possible to measure the tightening torque of one of each kind of fasteners, with the use of a torque wrench. As a torque wrench was used, the torque level on the wrench was increased a little at each time to capture the torque limit that would cause the shearing of the collar. Both collars were within the corresponding range. As only one fastener of each type was tested, the mean values of the torque range were used for the analysis. In fact, the torque of the two fasteners tested corresponded to the mean values, 35 and 70 respectively. In Fig. 4.16, the torque measurement process is presented.
Rearranging Eq.(2.19), it is possible to obtain the preload, Fi.
T F = (4.11) i Kd
68 (a) (b)
(c) (d)
Figure 4.15: Hi-Lok configuration in the Joint and Hi-Lok loading. (a) - Complete Joint view; (b) - Internal Joint view - No Honeycomb; (c) - Stress on the fasteners; (d) - Deformed and Non deformed fasteners configuration.
The theoretical preload is presented in Table 4.9. A Torque Coefficient, K, of 0.16 was used. It corre- sponds to Cadmium-plate fastener conditon. Hi-Lok65 4 and Hi-Lok32 1 were the fasteners with higher tension stress, of the two different fasteners diameters. The tension stress for each of these bolts, was converted into tension force in order to allow to make a comparison between the normal force (due to the aerodynamic loading) and the theoretical preload. As the maximum tension force on the fasteners is only about 5% of the theoretical preload, it is possible to conclude that the Joint is safe against flange separation. The above result for the fasteners mechanical behaviour are in accordance with the real
Table 4.9: Comparison between the theoretical preload and resultant tension force on the fasteners.
Hi-Lok Theoretical Preload [N] Tension Force [N] % of Preload
Diameter 4.76 [mm] 5148.44 227 (Hi-Lok65 4) 4.41 Diameter 6.35 [mm] 7784.04 375 (Hi-Lok32 1) 4.81
Acoustic Panel fasteners. The analysis of the acoustic panels removed from the aircraft do not present any kind of wear on the fasteners. This is a good indicator of the reliability of the simulation results. The non-existence of wear on the fasteners indicates that the loads on the fasteners are low.
69 (a) (b)
Figure 4.16: Hi-Lok torque measurement. (a) - Hi-Lok fixed in a vise; (b) - Torquing of the Collar.
4.6 Analysis of the Interface between Internal Honeycomb Core and Internal Doubler
The main goal of the present section is to determine the forces acting at the interface between the internal honeycomb core and the internal doubler. The doublers of the acoustic panels installed on the A320/A321 suffer deterioration due to aluminium corrosion. Repairs concerning the aft doubler exist and can be applied to the acoustic panel. Although, there are no repairs concerning the replacement of the internal doubler. In the present section the load acting between the above mentioned components is to be determined in order to enable a correct adhesive selection. The connection between the doubler and the honeycomb is made through the use of adhesives.
4.6.1 Components Contact and Mesh Refinement
The model used in the present section is the same used in the previous one, that is, bonded contact was used in all the acoustic panel components. The contact made between the attachment ring and the doubler was frictionless contact. Frictionless contact is also used between the fasteners shank and the holes of the fasteners.
In order to capture the forces acting between the doubler and the honeycomb, Joints Connections where used between the components. There are several types of joints available, fixed, cylindrical, universal, and others, that allow a correct joint behaviour simulation. In the present analysis the joint type used was the fixed one, since the components are adhesive bonded together. The joint connection define the interface between the bodies. One of the bodies is defined as the reference body and the other as the mobile body. When using the joint probe, the results correspond to the forces exercised by the mobile body on the reference body. A coordinate system as to be created and associated to the Joint. Forces are then reported on the joint coordinate system. More information about joints and joint probes can be found in ANSYS Workbench software Help.
70 In the present work, two joints are analysed. One joint correspond to the Lower Panel section and the other to the Side Panel section. The two joints and respective referential are presented in Fig. 4.17.
Since the main goal is to analyse components of the acoustic panels joints, the refinement focuses on these component and not on the fasteners. Once again, the approach used to refine the mesh was base on refining the mesh elements size on the components of interest. The mesh element size of the internal honeycomb core and internal doubler was changed at each mesh refinement. For all other components the mesh element size was kept the same at each refinement. Four levels of refinement were created, please see Table 4.10. The refinements limits were set by computational limitations. In Fig. 4.18, both the model lower and higher refinement are presented.
(a) (b)
Figure 4.17: Joint Connection referential. (a) - Referential for lower and side joints; (b) - Referential zoom.
Table 4.10: Mesh element size at each refinement.
Mesh Element Size [m] Refinement Internal Honeycomb Core Internal Doubler Mesh No Elements
1 3E-3 3E-3 173292 2 2.5E-3 2.5E-3 184449 3 2E-3 2E-3 203062 4 1.5E-3 1.5E-3 263761
4.6.2 Convergence Analysis
Since the main objective is to determine the forces acting on the interface between the internal honey- comb core and the internal doubler, the convergence analysis is made based on the reaction forces. The reaction force for both lower and side joints are presented in Table 4.11, as function of the mesh
71 (a) (b)
Figure 4.18: Lower and higher joint mesh refinements. (a) - Model lower refinement; (b) - Model higher refinement. refinement. In Fig. 4.19, the force reaction convergence for both the lower and side joints is presented. It is possible to verify that the result vary very little between refinements, demonstrating that the result can be assumed as converged. Refinement4 results were used in the following sections.
Table 4.11: Joints reaction forces as function of the mesh refinement.
Force Lower Joint [N] Force Side Joint [N] Refinement No Elements X Y Z Total X Y Z Total
1 173292 139.9 917.1 441.9 1027.6 -130.3 574.8 341.2 681.0 2 184449 145.1 915.9 443.6 1027.9 -137.5 574.1 348.3 685.4 3 203062 151.1 913.3 450.2 1029.3 -142.5 571.8 356.9 689.0 4 263761 160.2 912.9 450.2 1030.5 -150.6 571.3 365.37 694.7
4.7 Results and Discussion of the Joints Analysis
The result obtain from the simulation of the Joints are presented in Table 4.12, the result correspond to those the Refinement 4. The Joints can be considered to be a sandwich construction. In order to determine the Joints properties, calculations to determine the flatwise tensile strength and shear stress are made. Formulation presented in standards ASTM C297-94 and C273-00ε1 were used. In the simulations, the internal honeycomb core was assign as the reference body, the internal doubler was assign as the mobile body. With this convention, the results obtained in the simulations correspond to the forces that the internal doubler creates on the internal honeycomb core. In Fig. 4.20, the resultant forces on both the lower and side joints is presented on their respective referential. The coordinate system used on each joint is also presented. Looking at the result, it is visible that the forces aren’t equally distributed
72 Figure 4.19: Reaction force convergence for the Joints. on both Joints. This result was expected since the model section used do not coincide with a symmetry plane of the flow and neither of the geometry.
Table 4.12: Simulation result used for the Joints properties determination.
Force Lower Joint [N] Force Side Joint [N] Refinement No Elements X Y Z Total X Y Z Total
4 263761 160.2 912.9 450.2 1030.5 -150.6 571.3 365.4 694.7
4.7.1 The Flatwise Tensile Strength
From the analysis of the result and Fig. 4.20, it is visible that component of the force that creates the tensile load on the joint is the Y component of the force. In the present CAD model, both the lower and side cross-sectional areas are the same, A = 9946 mm2. Analysing the force results, it is visible that the Y component in the Lower Joint is larger than that of the Side Joints. The Y component of the Lower joint is used for the tensile strength calculation.
The flatwise tensile strength resulting from the Y component of the force can be calculated as follows, Eq. (4.12).
P 912.97 σ = = = 0.092MP a (4.12) A 9946
In Eq. (4.12), σ represents the flatwise tensile strength in MPa, P the load in N, and A the cross-sectional area in mm2.
The adhesive used to bond the aft doubler to the acoustic panel is the LOCTITE EA 9658 AERO, which
73 is an Epoxy film adhesive. It is here assumed that the adhesive used to bond the internal honeycomb core to the internal doubler is the same. Tensile Lap Shear Strength test result are presented in [58], for different conditions and and film weights. The results correspond to the adhesive application 2024- T3 and honeycomb core with 3/8 inch cell of 5052 aluminium. The lowest tension result is 2.1 MPa (please see Appendix C). Comparing both results, it appears that the result obtained from the simulation correspond to 4.38% of the reference value. A force 22.82 times the force obtained in the simulation would be needed to attain the reference value.
4.7.2 The Shear Stress
The shear stress resulting from the aerodynamic load on the Joint corresponds to the X and Z compo- nents of the joint force. Analysing the resultant force, it is visible that the Z component on the Lower joint is the largest component that produces shear in the Joint. This force is used to calculate the shear stress in the Joint. The calculation is made as follows, Eq.(4.13).
P 450.22 τ = = = 0.045MP a (4.13) Lb 9946
In Eq.(4.13), τ represents the shear stress in MPa, P the load in N, L the length of specimen in mm, and b the width of specimen in mm.
In [58] results for the Lap Shear Strength test are presented. In this case the, the test specimens are bare Aluminium 2024-T3 with phosphoric acid anodizing and corrosion inhibiting primer. This treatment are used to avoid corrosion and are also applied to the doublers. In this case, testing results from [loctite], present a minimum value of 12.1 MPa (please see Appendix C). Once again this result are much higher than those obtained in the simulations. The simulation result correspond to about 0.4% of the limit strength.
(a) (b)
Figure 4.20: Coordinate systems of the Joints and resultant forces. (a) - Joints referen- tial; (b) - Resultant forces on the joints.
74 Chapter 5
Conclusions
The overall results of this project allowed to reach the following conclusions:
The aim of this work was to analyse the joints of the acoustic panels of an Airbus A320/A321. In order to accomplish that, a CAD model had to be modelled. Two CAD models were created. The first one to be used in the CFD analysis and the other one for the FEM analysis, in which all components of the joint were discretized. A correct modelling of the nacelle can be very challenging since there is no symmetry planes in the geometry. All the three acoustic panels have different sizes and orientations. The geometry presents curvatures in all direction making the modelling of the inlet cowl a real challenge.
A methodology using Computational Fluid Dynamics to determine the aerodynamic loads acting on the inlet cowl of an aircraft’s engine was developed. The aerodynamics loads, pressure and wall shear stress, were obtained for four flight conditions. From this analysis it can be concluded that the increase of the angle of attack has a raising effect of the pressure magnitude. It could be concluded that the flight condition associated to the highest pressure magnitude is the Takeoff 16o. It was also verified that the effect of the wall shear stress has very little contribution for the loading magnitude, since it corresponds to about 1% of the Pressure magnitude. The effect of the Inlet duct diffuser could be verified, and typical values for the Mach number at the fan were obtained. The CFD methodology allowed to analyse the flow behaviour around the nacelle and the inlet duct. The suitability of the Methodology employed to determine the aerodynamic loads was verified. Good agreement between numerical results and experiment ones allows to reach that conclusion.
A methodology to analyse the joints of the acoustic panels using FEM was developed. Result from the CFD analysis were imported into the Finite Element Software. The loading conditions corresponding to the different flight conditions were analysed. Results showed that the flight condition corresponding to the critical loading is the Takeoff 16o.
With the development of this work, it was possible to verify that the analysis of fastened components require high computational effort, due to its nonlinear contact formulation. When having limited compu- tational resources, some simplifications to the models may have to be done to obtain results.
75 The analysis of the joint of the acoustic panel allowed to verify that the Joint is subjected to combined loads. The loads on the panels tend to bend the fasteners. The analysis of the fasteners allows to conclude that the forces transmitted to the fasteners are low. Stresses in the fasteners are far from the yield limits, both for tension and shear loading. It is also possible to conclude that the loads on the fastener are far from the preload ones, indicating that the forces applied on the acoustic panel aren’t enough to make it separate from the attachment ring. It can be concluded that the fasteners of the joint have a high safety factor and they do not present risk to the Joint safety.
Forces Acting on the interface between the internal honeycomb core and and internal doubler were determined. Flatwise tensile strength and shear stresses were calculated. Comparison of this result with with experimental results allows to conclude that the L0CTITE EA 9658 AERO can be used to connected both the internal honeycomb core and and internal doubler, with a large safety margin.
Finally, the development of this thesis allowed to create a methodology that allows the analysis of aero- dynamic and structural parameters of the inlet cowl of an aircraft engine. This methodology could be used and adapted to analyse other models involving aerodynamic loads and structural analysis.
5.1 Future Work
In the future it would be interesting to perform a similar analysis with more computational power.
It would be interesting to analyse larger sections of the inlet, or even the complete model.
An experimental analysis of the joint of the acoustic panel could be done in order to compare experi- mental and numerical results.
Finally, the developed model could be used to approach different problem.
76 Chapter 6
Bibliography
[1] Kundu, A. K. (2010). Aircraft design (Vol. 27). Cambridge University Press.
[2] Shmyrov,V.F., Other (2010). Airplane power plants systems designing. Kharkiv: National Aerospace University.
[3] TAP. TAP Fleet. From: http://www.tapportugal.com/Info/en/fleet-history/fleet-2016. Accessed: 2016- 10-01.
[4] Airbus. Component maintenance manual for the Airbus A320.
[5] Hill, P. G., Peterson, C. R. (1992). Mechanics and thermodynamics of propulsion. Reading, MA, Addison-Wesley Publishing Co., 1992, 764 p., 1.
[6] Watson Jr, H. A. (1970). Structural and Environmental Studies of Acoustical Duct-Lining Materials. The Journal of the Acoustical Society of America, 48(3C), 795-814.
[7] EASA. (2014). Approval certificate. From: https://www.tap-mro.com/Docs/Certifications/LIS%20EASA %20Part-21.pdf. Accessed :2016-10-01.
[8] EASA. Commission regulation (EU) No 748/2012. From: http://eur-lex.europa.eu/legal-content/EN/ TXT/?id=1473415871666&uri=CELEX:32012R0748. Accessed: 2016-10-01.
[9] Davis, M. (2003). The standard handbook for aeronautical and astronautical engineers. McGraw- Hill.
[10] Storm, R., Skor, M., Koch, L.D., Benson, T., Galica, C. (2007). Pushing the Envelope: A NASA Guide to Engines. National Aeronautics and Space Administration.
[11] Federal Aviation Administration and Safety Research Corporation of America (2012). Federal Avia- tion Administration (FAA),1.
[12] Airbus. A320 Family. From: http://www.airbus.com/aircraftfamilies/passengeraircraft/a320family/. Accessed: 2015-11-15.
77 [13] CFM. In the wings with the CFM56 engines. From: http://walter.bislins.ch/aviatik/media/cfm-technical- data.pdf. Accessed: 2015-11-15.
[14] Borges, T. (2014). Propulsao˜ Folhas de Apoio. Lisboa: Secc¸ao˜ de folhas.
[15] Eurospot. (2006). Airbus A320-111. From: http://www.airliners.net/photo/Airbus/Airbus-A320- 111/1027446. Accessed: 2016-10-05.
[16] (2002). Lockheed F-117 Nighthawk. From: https://en.wikipedia.org/wiki/Lockheed F-117 Nighthawk. Accessed: 2016-10-05.
[17] CFM . AMM - Aircraft Maintenance Manual - CFM56-5B. vol. 70-10-00 Nacelle Components.
[18] Azimi, M., Ommi, F., & Alashti, N. J. (2014). Using acoustic liner for fan noise reduction in modern turbofan engines. International Journal of Aeronautical and Space Sciences, 15(1), 97-101.
[19] Astley, R. J., Agarwal, A., Joseph, P. F., Self, R. H., Smith, M. G., Sugimoto, R., & Tester, B. J. (2007). Predicting and reducing aircraft noise.
[20] Airport Cooperative Research Program (2008). Effects of Aircraft Noise: Research Update on Selected Topics. Sunthesis 9.
[21] Leylekian, L., Lebrun, M., & Lempereur, P. (2014). An overview of aircraft noise reduction technolo- gies. AerospaceLab, (6), p-1.
[22] Department for Transport (2013). Night Flying Restrictions at Heathrow, Gatwick and Stansted Stage 1 Consultation. From: https://www.gov.uk/government/uploads/system/uploads/attachment data/ file/330354/night-noise-decision.pdf. Accessed: 2015-12-03.
[23] Astley, R. J. (2009). Numerical methods for noise propagation in moving flows, with application to turbofan engines. Acoustical science and technology, 30(4), 227-239.
[24] Huff, D. L. (2007). Noise reduction technologies for turbofan engines. Glenn Research Center, Cleveland, Ohio: NASA.
[25] Geng, D., Yi, K., Shang, C., Yang, J., & He, Y. (2015). Aplication status of composite acoustic liner in aero-engine. 10th International Conference on Composite Science and Technology ICCST/10.
[26] Sugimoto, R., Astley, R. J., & Murray, P. B. (2010). Low frequency liners for turbofan engines. In Proceedings of the 20th International Congress on Acoustics.
[27] Airbus A380 Engine. From: http://www.aerospaceweb.org/question/propulsion/q0233a.shtml. Ac- cessed: 2016-08-01.
[28] Versteeg, H. K., & Malalasekera, W. (2007). An introduction to computational fluid dynamics: the finite volume method. Pearson Education.
[29] Lomax, H., Pulliam, T. H., & Zingg, D. W. (1999). Fundamentals of computational fluid dynamics. Applied Mechanics Reviews, vol, 55, B61.
78 [30] Hirsch, C. (2007). Numerical computation of internal and external flows: The fundamentals of computational fluid dynamics. Butterworth-Heinemann.
[31] Gilmore, J. D. (2012). Computational Fluid Dynamics Analysis of Jet Engine Test Facilities. Univer- sity of Canterbury.
[32] Boussinesq eddy viscosity assumption. From: http://www.cfd-online.com/Wiki/Boussinesq eddy viscosity assumption. Accessed: 2016-08-03.
[33] Niu, M. C. Y. (1997). Airframe stress analysis and sizing (Vol. 2). Hong Kong: Conmilit Press.
[34] Richard G. Budynas, J. Keith Nisbett (2011). Shigley’s Mechanical Engineering Design, Ninth Edi- tion. McGraw-Hill.
[35] Baker, A. A. B. (2004). Composite materials for aircraft structures. AIAA.
[36] Niu, C. (1988). Airframe structural design: practical design information and data on aircraft struc- tures. Conmilit Press.
[37] lisi Aerospace. Hi-Lok Fastening system. From: http://www.lisi-aerospace.com/products/fasteners/ internally-threaded/collars/Pages/nut-hi-lok.aspx. Accessed: 2016-09-07
[38] Skorupa, A., & Skorupa, M. (2012). Riveted lap joints in aircraft fuselage: design, analysis and properties (Vol. 189). Springer Science & Business Media.
[39] ASTM International. (1999). Standard Test Method for Flatwise Tensile Strength of Sandwich Constructions. Designation: C 297 – 94.
[40] ASTM International. (2000). Standard Test Method for Shear Properties of Sandwich Core Materi- als. Designation: C273-00ε1.
[41] ”Modeling Bolted Connections” http://docslide.us/documents/whitepaper-bolted-connections.html, Accessed: 2016-10-05.
[42] Montgomery, J. (2002, April). Methods for modeling bolts in the bolted joint. In ANSYS User’s Conference (Vol. 5).
[43] Airbus, (2015). Aircraft characteristics airport and maintenance panning - Airbus A320.
[44] Airbus. Autocad 3-view aircraft drawings - A320. From: http://www.airbus.com/support/maintenance- engineering/technical-data/autocad-3-view-aircraft-drawings/. Accessed: 2015-11-07.
[45] CFM International. CFM56-5B Engine. From: https://www.cfmaeroengines.com/engines/cfm56/, Accessed: 2015-11-09.
[46] Hall, Z., Ahuja, V., Hartfield, R., Shelton, A., & Ahmed, A. (2009). Optimization of a Turbofan Inlet Duct using a Genetic Algorithm and CFD. AIAA Paper, 3775, 22-25.
[47] LUCIOLI, J. (2011). CFD analysis of a nacelle at high angle of incidence. Politecnico Di Mi- lano.
79 [48] STAR CCM+ 9.06 version. Help Manual.
[49] Zhaoguang, Tan and Yinchun, Chen and Wei, Huang (2012). Research for jet flow influences to wing-mounted civil aircraft aerodynamic characteristics. ICAS - International Congress of the Aeronau- tical Sciences.
[50] Liu, K., Sun, Y., Zhong, Y., Zhang, H., Zhang, K., & Yang, H.(2014). Numerical investigation on engine inlet distortion under crosswind for a commercial transport aircraft. ICAS - International Congress of the Aeronautical Sciences.
[51] Chen, H., Yu, N., Rubber, P., & Jameson, A. (1983). Flow simulations for general nacelle configura- tions using Euler equations. AIAA paper, 539, 1983.
[52] Li, J., Li, F., & Chen, H. (1998). 3-D flow simulations for general powered engine nacelles using Euler equations. In 36th AIAA Aerospace Sciences Meeting and Exhibit (p. 929).
[53] HIROSE, Naoki, et al. Transonic 3-D Euler Analysis of Flows around Fanjet Engine and TPS (Tur- bine Powered Simulator)-Comparison with Wind Tunnel Experiment, Evaluation of TPS Testing Method and 3-D Flow. 1989.
[54] Hexcel. (2000). HexWeb Honeycomb Attributes and Properties. From: http://www.hexcel.com/resour ces/datasheets/brochure-data-sheets/honeycomb attributes and properties.pdf. Accessed: 2016-05- 01.
[55] Hexcel. HexWeb HONEYCOMB SANDWICH DESIGN TECHNOLOGY. From: http://www.hexcel. com/resources/datasheets/brochure-data-she ets/honeycomb sandwich design technology.pdf, Accessed: 2016-05-01.
[56] Gibson, L. J., & Ashby, M. F. (1999). Cellular solids: structure and properties. Cambridge university press.
[57] ”Hi-Lok Collar Mechanical Lock Application Stainless Steel 1/16 grip variation, Shear Application” http://www.jet-tek.com/hi-lok-collars/hl585.pdf, Accessed: 2016-06-08.
[58] LOCTITE, ”LOCTITE EA 9658 AERO Epoxy Film Adhesive”, https://tds.us.henkel.com/NA/UT/ HNAUTTDS.nsf/web/22C510C43FA6A5B58525715C001BD519/$File/LOCTITE4%20EA%209658%20 AERO.PDF. Accessed: 2016-09-30.
[59] SRM - Structural Repair Manual CFM56-5B, vol. 54-00-00 General.
80 Appendix A
Appendix - Noise Certification Points
Figure A.1: Noise certification points for ICAO Annex 16 and FAA FAR36 [19].
81 82 Appendix B
Appendix - T.P.S. Experimental Properties
Table B.1: T.P.S properties of experimental essay [53].
T.P.S. Experimental Properties
M∞ 0.6024
p0 80406 Pa
T0 319.4 K
patm 62920 Pa MFR 0.49609
2 Afan 0.10808 m
83 84 Appendix C
Appendix - LOCTITE EA 9658 AERO - Technical Datasheets
Both the tensile lap shear strength and the flatwise tension strength experimental results of the adhesive LOCTITE EA 9658 AERO are presented. Data from [58].
85 Technical Process Bulletin LOCTITE EA 9658 AERO Epoxy Film Adhesive (KNOWN AS Hysol EA 9658)
Tensile Lap Shear Strength - Tensile lap shear strength tested per ASTM D1002 after curing 1 hour @ 350°F/177°C. Adherends are 2024-T3 bare aluminum treated with phosphoric acid anodizing per ASTM D3933 and primed with companion low VOC water based corrosion inhibiting primer LOCTITE EA 9258.1 AERO. The primer was cured 60 minutes at 350°F/177°C. Nominal primer thickness was 0.020-0.24 mils (5-6 microns).
LOCTITE EA 9658 AERO LOCTITE EA 9658 AERO Test Temp. Sample 0.060 psf (290 g/m2) UNS 0.10 psf (490 g/m2) NWG Conditioning °F °C psi MPa psi MPa -67 -55 4350 30.0 4370 30.2 77 25 5200 35.9 5290 36.5 Dry 250 121 3960 27.3 4030 27.8 350 177 3110 21.4 3000 20.7 Wet 77 25 3870 26.7 3790 26.1 750 hrs. at 250 121 3430 23.6 3390 23.4 158°F (70°C) & 95% R.H. 350 177 1760 12.1 1840 12.7 77 25 3910 27.0 3670 25.3 1000 hrs. at 350°F (177°C) 350 177 2760 19.0 2640 18.2 77 25 3230 22.3 3170 21.8 3000 hrs. at 350°F (177°C) 350 177 2710 18.7 2690 18.6 77 25 2780 19.2 2750 19.0 6000 hrs. at 350°F (177°C) 350 177 2530 17.4 2190 15.1
3 of 5 Technical Process Bulletin LOCTITE EA 9658 AERO Epoxy Film Adhesive (KNOWN AS Hysol EA 9658)
Flatwise Tension Strength - tested per ASTM C297 after curing 1 hour @ 350°F/175°C. Adherends are 2024-T3 bare aluminum treated with phosphoric acid anodizing per ASTM D3933 and primed with companion low VOC water based corrosion inhibiting primer LOCTITE EA 9258.1 AERO. The primer was cured 60 minutes at 350°F/177°C. Nominal primer thickness was 0.020-0.24 mils (5-6 microns). The honeycomb core was 3/8 inch (9.50 mm) cell 5052 non-perforated aluminum core.
The 0.060 psf (290 g/m2) unsupported film was reticulated onto the core.
Thermally aged samples were drilled through each cell wall with a 0.10 inch (2.5mm) diameter drill for thermal exposure.
LOCTITE EA 9658 AERO LOCTITE EA 9658 AERO Test Temp. Sample 0.060 psf (290 g/m2) UNS 0.10 psf (490 g/m2) NWG Conditioning °F °C psi MPa psi MPa -67 -55 1350 9.3 1310 9.1 77 25 1260 8.7 1220 8.4 Dry 250 121 1020 7.0 1010 7.0 350 177 700 4.8 640 4.4 Wet: 750 hrs. at 77 25 980 6.7 930 6.4 158°F (70°C) & 95% R.H. 250 121 770 5.3 690 4.8 77 25 1140 7.8 1130 7.8 1000 hrs. at 350°F (177°C) 350 177 450 3.1 420 2.9 77 25 1070 7.4 1060 7.3 3000 hrs. at 350°F (177°C) 350 177 420 2.9 410 2.8 77 25 1000 6.9 930 6.4 6000 hrs. at 350°F (177°C) 350 177 300 2.1 320 2.2
4 of 5 88