Linear Aeroelastic Stability of Beams and Plates in
Three-Dimensional Flow by Samuel Chad Gibbs IV
Department of Mechanical Engineering and Materials Science Duke University
Date:
Approved:
Earl H. Dowell, Supervisor
Kenneth C. Hall
Donald B. Bliss
Thomas P. Witelski
Thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in the Department of Mechanical Engineering and Materials Science in the Graduate School of Duke University 2012 Abstract Linear Aeroelastic Stability of Beams and Plates in Three-Dimensional Flow by Samuel Chad Gibbs IV
Department of Mechanical Engineering and Materials Science Duke University
Date:
Approved:
Earl H. Dowell, Supervisor
Kenneth C. Hall
Donald B. Bliss
Thomas P. Witelski
An abstract of a thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in the Department of Mechanical Engineering and Materials Science in the Graduate School of Duke University 2012 Copyright c 2012 by Samuel Chad Gibbs IV All rights reserved except the rights granted by the Creative Commons Attribution-Noncommercial Licence Abstract
The aeroelastic stability of beams and plates in three-dimensional flows is explored as the elastic and aerodynamic parameters are varied. First principal energy meth- ods are used to derive the structural equations of motion. The structural models are coupled with a three-dimensional linear vortex lattice model of the aerodynam- ics. An aeroelastic model with the beam structural model is used to explore the transition between different fixed boundary conditions and the effect of varying two non-dimensional parameters, the mass ratio µ and aspect ratio H∗, for a beam with a fixed edge normal to the flow. The trends matched previously published theoreti- cal and experimental data, validating the current aeroelastic model. The transition in flutter velocity between the clamped free and pinned free configuration is a non- monotomic transition, with the lowest flutter velocity coming with a finite size spring stiffness. Next a plate-membrane model is used to explore the instability dynamics for different combinations of boundary conditions. For the specific configuration of the trailing edge free and all other edges clamped, the sensitivity to the physical parameters shows that decreasing the streamwise length and increasing the tension in the direction normal to the flow can increase the onset instability velocity. Finally the transition in aeroelastic instabilities for non-axially aligned flows is explored for the cantilevered beam and three sides clamped plate. The cantilevered beam con- figuration transitions from an entirely bending motion when the clamped edge is normal to the flow to a typical bending/torsional wing flutter when the clamped
iv edge is aligned with the flow. As the flow is rotated the transition to the wing flutter occurs when the flow angle is only 10 deg from the perfectly normal configuration. With three edges clamped, the motion goes from a divergence instability when the free edge is aligned with the flow to a flutter instability when the free edge is normal to the flow. The transition occurs at an intermediate angle. Experiments are carried out to validate the beam and plate elastic models. The beam aeroelastic results are also confirmed experimentally. Experimental values consistently match well with the theoretical predictions for both the aeroelastic and structural models.
v Contents
Abstract iv
List of Tablesx
List of Figures xi
List of Abbreviations and Symbols xv
1 Introduction and Literature Review1
2 Structural Model 10
2.1 Beam Structural Model Derivation...... 11
2.1.1 Boundary Conditions...... 14
2.1.2 Equations of Motion...... 15
2.1.3 Normalized Equations of Motion...... 15
2.1.4 Bending Separation of Variables...... 16
2.1.5 Torsional Separation of Variables...... 17
2.2 Specific Beam Mode Shapes...... 19
2.2.1 Clamped-Free...... 19
2.2.2 Pinned-Free...... 22
2.2.3 Clamped-Clamped...... 25
2.2.4 Free-Free...... 28
2.3 Pinned Edge Torsional Spring Model...... 29
2.4 Plate Structural Model...... 33
vi 2.4.1 Plate Structural Analysis Typical Results...... 38
2.5 Forced System Modification...... 44
3 Aerodynamic Model 47
3.1 Aerodynamic Theory Introduction...... 47
3.2 Vortex Lattice Aeroelastic Model...... 55
3.2.1 Downwash State Relations...... 56
3.2.2 Non-dimensional Generalized Force...... 58
3.2.3 Governing Aeroelastic Matrix Equations...... 60
3.3 Code Development...... 61
3.3.1 Matrix Definition...... 62
3.3.2 Flutter Speed and Eigenvalue Determination...... 62
3.3.3 Generating Time Histories from Eigenanalysis...... 67
3.4 Inclusion of Fixed Support Structures...... 68
3.5 Mirroring to Simulate Wind Tunnel Walls...... 69
3.6 Using ANSYS Structural Modes...... 70
3.7 Rotated Wing Analysis...... 72
3.7.1 Generalized Force Calculation...... 75
3.7.2 Downwash Calculation...... 75
4 Results from Aeroelastic Simulations 77
4.1 Dimensional Beam Simulations...... 77
4.1.1 Time History Analysis vs. Eigenanalysis...... 78
4.1.2 Fixed Leading Airfoil Effect...... 79
4.2 Wind Tunnel Wall Confinement Effects...... 79
4.2.1 Out-of-Plane Normal to Flow Confinement...... 80
4.2.2 In-Plane Normal to Flow Wind Tunnel Wall Confinement... 81
vii 4.3 Non-dimensional Simulations (Modified from Journal of Fluids and Structures Journal Submission)...... 82
4.3.1 Leading Edge Spring Simulations...... 83
4.3.2 Aspect Ratio Variation Simulations...... 85
4.3.3 Mass Ratio Variation Simulations...... 86
4.4 Plate Simulations...... 90
4.4.1 NASA Simulations (Configuration 6)...... 90
4.4.2 Increasing the Flutter Velocity...... 94
4.4.3 Additional Plate Boundary Configurations...... 98
4.4.4 Discussion...... 107
4.5 Axially Misaligned Analysis...... 108
4.5.1 Axially Misaligned Beam Simulations...... 108
4.5.2 Axially Misaligned Plate Simulations...... 115
5 Experiments 120
5.1 Experiments to Validate Beam Model...... 120
5.1.1 Beam Structural Experiments...... 122
5.1.2 Beam Aeroelastic Experiments...... 123
5.2 Experiments to Validate Plate Model...... 125
5.2.1 Design of Experimental Setup...... 126
5.2.2 Static Structural Experiments...... 128
5.2.3 Dynamic Structural Experiments...... 130
5.2.4 Plate Aeroelastic Experiments...... 138
5.3 Configuration 1 Aeroelastic Experiments...... 139
5.4 Configuration 6 Aeroelastic Experiments...... 142
6 Conclusion and Future Work 145
6.1 Conclusions...... 145
viii 6.2 Future Work...... 147
6.2.1 Theoretical...... 147
6.2.2 Experimental...... 148
6.2.3 Applications...... 148
A Beam Aeroelastic Experimental Data Points 150
B Configuration 2 Raw Data 152
Bibliography 154
ix List of Tables
2.1 Non-Dimensional Natural Frequencies for a Single Edge Fixed Beam. 30
2.2 NASA Membrane Properties...... 38
4.1 Plate Aeroelastic Simulation Summary (ζs = 0.01)...... 98
4.2 Plate Aeroelastic Simulation Summary (ζs = 0.05)...... 99 4.3 Rotated Wing Properties...... 108
5.1 Beam Experimental Parameters...... 121
5.2 Equipment Used in the Ground Vibration Experiment...... 131
5.3 First Three Modal Damping Ratios with No Tension...... 137
5.4 Equipment Used in the Flutter Experiment...... 139
5.5 Plate Aeroelastic Experimental Results...... 141
5.6 Configuration 1 Experimental Results...... 141
5.7 Flutter Speed and Frequency for the Un-Tensioned Specimen: Theory and Experiment...... 143
A.1 Experimental Datapoints for a Clamped-Free Plate...... 150
A.2 Experimental vs Theoretical Error...... 151
x List of Figures
1.1 Continuous Mold-Line Link...... 7
1.2 Plate Configurations to Explore...... 7
2.1 Clamped Free Schematic...... 20
2.2 Clamped Free Frequencies...... 20
2.3 Clamped Free Bending Mode Shapes...... 21
2.4 Pinned Free Schematic...... 23
2.5 Pinned Free Frequencies...... 23
2.6 Pinned Free Mode Shapes...... 24
2.7 Clamped Clamped Schematic...... 25
2.8 Clamped Clamped Frequencies...... 26
2.9 Clamped Clamped Mode Shapes...... 27
2.10 Free Free Frequencies...... 29
2.11 Free Free Mode Shapes...... 30 ˜ 2.12 Pinned-Free, Clamped-Free and Large Kα Mode Shapes...... 31 2.13 Structural Frequency Evolution with Leading Edge Torsional Spring. 32
2.14 Configuration 1 Natural Frequencies and Mode Shapes...... 40
2.15 Configuration 2 Natural Frequencies and Mode Shapes...... 40
2.16 Configuration 3 Natural Frequencies and Mode Shapes...... 41
2.17 Configuration 4 Natural Frequencies and Mode Shapes...... 41
2.18 Configuration 5 Natural Frequencies and Mode Shapes...... 42
xi 2.19 Configuration 6 Natural Frequencies and Mode Shapes...... 42
2.20 Configuration 2 Natural Frequency Evolution for Chord Variation.. 43
2.21 Configuration 2 Natural Frequency Evolution for Tension Variation. 44
3.1 Visualization of Structural Mode Shapes with Vortex Lattice Wake. 48
3.2 Expanded Schematic of Vortex Lattice Mesh...... 50
3.3 Aeroelastic Simulation Model...... 60
3.4 Typical Near Flutter Time History...... 63
3.5 Near Flutter Time History Modal FFT...... 64
3.6 Near Flutter Time History Modal Damping...... 65
3.7 Typical Velocity Sweep...... 67
3.8 Typical Root Locus...... 67
3.9 Mirrored Wall Schematic...... 70
3.10 Cantilevered Wing Configuration Schematic...... 71
3.11 Aerodynamic and Elastic Coordinate Systems...... 73
3.12 Rotated Wing Mesh Visualization...... 74
4.1 Eigenanalyis vs Time History Analyis Root Locus...... 78
4.2 Eigenanalyis vs Time History Analysis Damping vs. Velocity..... 78
4.3 Leading Airfoil Root Locus...... 79
4.4 Leading Airfoil Damping vs. Velocity...... 79
4.5 Impact of Out-of-Plane Confinement on Flutter Frequency Prediction 80
4.6 Impact of In-Plane Confinement on Flutter Velocity Prediction.... 81
4.7 Impact of In-Plane Confinement on Flutter Frequency Prediction.. 82 ˜ 4.8 Flutter Frequency and Velocity vs. Kα ...... 84 4.9 Flutter Velocity as a function of the Aspect Ratio...... 86
4.10 Flutter Velocity and Frequency vs. µ ...... 87
xii 4.11 Growing Mass Ratio Simulation...... 88
4.12 Configuration 6 Aeroelastic Results...... 92
4.13 Plate Structural Model Convergence Plots...... 93
4.14 Plate Structural Model: Support Structure Influence Plots...... 95
4.15 Configuration 2 Aspect Ratio Variation Flutter Boundary...... 96
4.16 Configuration 2 Aspect Ratio Variation Flutter Boundary Mode Shapes 96
4.17 Configuration 2 Aspect Tension Variation Flutter Boundary..... 97
4.18 Configuration 2 Tension Variation Flutter Boundary Mode Shapes.. 97
4.19 Configuration 1 Aeroelastic Results...... 101
4.20 Configuration 2 Aeroelastic Results...... 102
4.21 Configuration 3 Aeroelastic Results...... 103
4.22 Configuration 4 Aeroelastic Results...... 105
4.23 Configuration 5 Aeroelastic Results...... 106
4.24 Rotating Beam Flutter Boundary...... 110
4.25 Rotation Angle=0, One Period Flutter Motion...... 111
4.26 Rotation Angle=6.92, One Period Flutter Motion...... 112
4.27 Rotation Angle=11.53, One Period Flutter Motion...... 113
4.28 Rotation Angle=90, One Period Flutter Motion...... 114
4.29 Rotated Plate Aeroelastic Boundary...... 115
4.30 Rotated Plate Aeroelastic Boundary Mode Shapes...... 115
4.31 Rotation Angle=0, One Period Flutter Motion...... 117
4.32 Rotation Angle=45 deg, One Period Flutter Motion...... 118
4.33 Rotation Angle=60 deg, One Period Flutter Motion...... 119
5.1 Experiment Apparatus...... 120
5.2 Natural Frequency Experimental Results...... 122
xiii 5.3 Mass Ratio Variation with Experiment...... 124
5.4 CAD Rendering of Baffle...... 126
5.5 Close up of the Connector...... 127
5.6 Different Strain Settings Allowing for Varying Span-wise Tension... 128
5.7 Stress Strain Curve and Estimation of Elastic Modulus...... 129
5.8 Estimated Poisson’s Ratio...... 130
5.9 Photographs of the Experimental Setups...... 131
5.10 Configuration 1 Dynamic Experimental Results...... 132
5.11 Configuration 2 Dynamic Experimental Results...... 133
5.12 The (1,2) Mode Visualization for Configuration 2...... 133
5.13 Configuration 4 Dynamic Experimental Results...... 134
5.14 Ground Vibration Test Setup for Configuration 4...... 135
5.15 Laser Readout and Shaker Excitation Locations...... 136
5.16 Natural Frequency Results for 4 Levels of Tension: Theory and Ex- periment...... 137
5.17 Photograph of Baffle Inside the Wind Tunnel...... 139
5.18 Configuration 1 Aeroelastic Experimental Results...... 140
5.19 Example Waterfall Plot for the Un-Tensioned Specimen...... 142
B.1 Configuration 2 Sample Spectrum Analyzer Output...... 153
xiv List of Abbreviations and Symbols
Symbols
As Number of airfoil elements in normal to flow(~y) direction
Ac Number of airfoil elements in chordwise(~x) direction
At Total number of airfoil elements E Young’s modulus of the structure
G Shear modulous of the structure
h Structure thickness
I Area moment of inertia of the structure
Iea Moment of inertial around the elastic axis
Kk,l Kernel function for the influence of the k’th discrete Γ on the l’th panel
K¯ Stiffness matrix
L, Lx Structure streamwise length
M¯ Mass matrix
m Mass per unit length of the structure
p(x, y, t) Aerodynamic pressure at the panel location (x,y) at time (t)
Q~ Generalized aerodynamic force
~rt Distance from circulation element to point in space (t) S, Ly Structure normal to flow length
Ss Number of structure elements in normal to flow(~y) direction
xv Sc Number of structure elements in streamwise(~x) direction
St Total number of structure elements T Structure kinetic energy
Ty,Tx Elastic tension in the subscript direction U Free stream fluid velocity
V Structure potential energy
Vd Vertical velocity of the elastic structure at collocation points
Ws Number of wake elements in normal to flow(~y) direction
Wc Number of wake elements in streamwise(~x) direction
Wt Total Number of structure elements w(x, y, t) Displacement at Structure location (x,y) at time (t)
x , y Streamwise and span wise direction respectively
(xζ , yη) (x,y) location of the (ζ, η) panel α vortex lattice relaxation factor
δW Virtual work
~Γ Discrete circulation values
γ(x, y) Continuous circulation at (x,y)
λ Natural frequency of the structure
η Index of column in vortex mesh
Φ Vector of position and velocity coordinates of natural modes
ρ Density
Θ~ Vector with Γ and Φ
ζ Index of row in vortex mesh or aerodynamic damping ratio de- pending on context item[ζs] Structural damping ratio
xvi Superscripts
˜ Non-dimensional
˙ Time derivative
0 Spatial derivative
~ Vector quantity
¯ Matrix quantity
Abbreviations
VLM Vortex Lattice Method
xvii 1
Introduction and Literature Review
This thesis is to outlines a technique to predict the aeroelastic instability bound- ary for one-dimensional beams and two-dimensional rectangular plates due to three- dimensional aerodynamic forces. Specifically the linear aeroelastic instability bound- ary for a wide variety of configurations and parameters is explored. The most com- mon aeroelastic instability encountered is a flutter instability. Flutter is the dynamic instability of a structure in a moving fluid that exhibits unsteady oscillations due to the interaction between the structure and the fluid. Such systems tend to exhibit limit cycle oscillations (LCO) that persist even if the free stream velocity falls below the flutter onset velocity creating what is called a hysteresis band, the possibility of multiple states at a given velocity. However, because all of the analysis conducted in this paper is linear, the origins of this hysteresis behavior is not explicitly discussed. Historically, the majority of flutter research has been focused on suppressing flut- ter because it is catastrophic in many structures including aircraft, bridges, and turbomachinery. Recently, attention has been refocused to gaining a better un- derstanding of flutter, especially for the cantilevered beam configuration, due to a growing interest in small scale energy harvesting systems. In addition to energy
1 harvesting applications, the configurations explored throughout this thesis can also be used to understand the dynamics of human snoring [19] and to reduce the noise generated during landing by subsonic fixed wing aircraft [26]. For this thesis, the aeroelastic models are specifically used to:
• Analyze the aeroelastic instabilities for a cantilevered beam in the transition between pinned and clamped leading fixed edge
• Analyze the aeroelastic instabilities for a beam with a clamped leading edge as the governing non-dimensional parameters are varied
• Analyze the aeroelastic instabilities for a plate with three sides fixed, a proposed configuration to reduce airframe noise on low subsonic aircraft during landing
• Analyze the aeroelastic instabilities that occur for plates as the boundary con- ditions are varied
• Analyze the aeroelastic instabilities for axially mis-aligned flows for a one-side clamped beam and three-sides clamped plate
Generally the motivation for the research stems from a desire to continue to advance the understanding of the aeroelastic instabilities that occur in rectangular structures. Developing an aeroelastic model requires developing models for both the structural dynamics and the aerodynamics. Once an aeroelastic model is created the model is used to analyze configurations of interest. The first problem explored is the interaction between a cantilevered flexible elastic beam and a uniform axial flow, a canonical fluid-structure interaction problem. It is well known that this system exhibits a flutter instability in low subsonic flow as the free stream velocity is increased above a critical velocity. The structure then enters a large and violent limit cycle oscillation (LCO). Since the experimental observations
2 of the flapping flag by Taneda [30] in 1968, many scholars have explored the stability of this system experimentally and theoretically. Although extensively explored in the literature, a full understanding of the dynamics of this relatively simple fluid- structure interaction remains elusive. In addition to the problem’s inherent physical significance, Doar´eand Michelin [7], Dunnmon et al. [11] and Giacomello and Porfiri [16] have recently proposed using the phenomena for energy harvesting applications and Eloy and Schouveiler [12] and Hellum et al. [18] have explored the potential of using this flutter for propulsion. Furthermore, Balint and Lucey [3], Huang [20] and Howell et al. [19] have shown that flutter in the human soft palette can explain snoring and Watanabe et al. [38] has explored flutter in the printing industry. Many structural and aerodynamic models have been developed or applied to improve the understanding of the dynamics of this system. The initial models looked at the limiting cases where either the streamwise or normal to flow dimension of the elastic member is assumed to be infinite. For the first case, the problem approaches a two-dimensional limit. In the two-dimensional limit the potential flow equations have been solved to determine the aerodynamic forces using the continuous equation with the appropriate boundary conditions [20, 22, 17, 39] and or discrete approximations. The discrete approximations can be split into the discrete vortex models [31, 34, 35, 25,1, 19] or numerical simulations solving the Navier-Stokes equations [3, 39]. In the latter limit, where the length is much larger than the span, a slender body approximation has been used by Lemaitre et al. [23] to explore the dynamics. For the two-dimensional case, Howell et al. [19] explored the influence of spatial confinement and Michelin and Smith [24] and Tang and Pa¯ıdoussis[36] have modeled the influence of cascades. In addition to these two-dimensional aerodynamic models, researchers have cou- pled different structural models when exploring the response of the system. The structural models have largely consisted of linear and non-linear models of beams
3 with simple out of plane displacements. In general linear structural models are used to explore the stability boundary as parameters are varied. Non-linear models have been used by Michelin et al. [25], Tang and Pa¯ıdoussis[35], Tang et al. [32], Tang and Pa¯ıdoussis[34] and Dunnmon et al. [11] to explore the post critical behaviors such as LCO amplitude and hysteresis loops which are observed experimentally. Recently interest in piezoelectric energy harvesting has motivated detailed exploration of the non-linear post critical behavior because predicting the amplitude and frequency of the limit cycle is vital to optimizing the energy harvested from the system [11, 16,7]. The critical velocities predicted by the two-dimensional models are remarkably similar to each other regardless of the solution technique used. Unfortunately their collective agreement does not match published experimental results reported by Taneda [30], Kornecki et al. [22], Watanabe et al. [38], Yamaguchi et al. [40], Tang et al. [32], Eloy et al. [14] and Dunnmon et al. [11]. In fact, across the range of parameters tested the two-dimensional model predicted flutter boundaries are signif- icantly below the experimentally observed values. Even when Huang [20] attempted to create a two-dimensional experimental model by having test pieces span the wind tunnel, the experimentally observed critical velocities are still much higher than the theoretical predictions. This discrepancy has motivated the application of three-dimensional aerodynamic models. Many of the initial three-dimensional aerodynamic models were used to ex- plore the flutter characteristics of a single configuration. For example Tang et al. [32] used an unsteady three-dimensional vortex lattice model(VLM) and a non-linear structural model to explore the flutter boundary and post critical behavior of a sin- gle aluminum plate. The success of initial three-dimensional simulations to match the flutter boundary between theory and experiment has prompted the most recent explorations of the stability boundary in parameter space with three-dimensional aerodynamic models by Eloy et al. [13] and Eloy et al. [14]. In general these simula-
4 tions have shown much better agreement with the experimental results. Furthermore an exploration of the three-dimensional effects of in-plane normal to the flow confine- ment by Doar´eet al. [8] demonstrates that the small distance between wind tunnel walls and experimental specimen required to produce the two-dimensional limit ex- perimentally would be prohibitively difficult to achieve. Three-dimensional effects are believed to explain the systemic discrepancies between strictly two-dimensional theoretical predictions and experimental observations for the critical flutter velocity. With the new understanding of the importance of three-dimensional effects on the quantitative behavior of this fluid-structure system there is a need to analyze the im- pact of different influences such as structural boundary conditions, confinement and experimental support structure with a three-dimensional aerodynamic model. The three-dimensional unsteady vortex lattice model remains a versatile means to ex- plore the aforementioned influences. Numerical simulations have the benefit of being able to model the effect of different configurations without changing the framework of the analysis. The work presented for this configuration is a continuation of the work done by Tang et al. [32]. The VLM aerodynamic model is generalized and used to explore the stability boundary for the cantilevered beam in the non-dimensional parameter space. Specifically the critical flow velocity as a function of mass ratio and aspect ratio is explored and compared with new experimental results as well as experimental and theoretical results found in the literature. In general the qualita- tive trends and quantitative values match the existing three-dimensional theoretical and experimental results. Additionally the analysis of this configuration explores the effect of the leading edge boundary condition on the critical flutter velocity. Using a leading edge tor- sional spring the transition between the two limiting cases is presented, including a surprising, non-monotonic transition in the critical flutter velocity. Finally normal to the flow confinement in both the in plane and out of plane directions are presented.
5 Next, an aeroelastic model is created to analyze the aeroelastic stability of two- dimensional rectangular plates. The project was initially motivated by a desire to analyze a plate configuration similar to one created by NASA’s proposed aircraft noise reduction effort is explored. NASA, as a part of its strategic plan in 2000, defined goals for designing the next generation of commercial transport aircraft with several performance requirements, one of which is noise reduction.[26] Experimental and numerical studies have shown that a large portion of aircraft noise during landing is generated by the interaction of shed vortices and wing structure at the discontinuity between the wing and the trailing edge flap.[6, 28] The noise reduction potential of several geometries and mechanisms have been studied, but experiments showed that the most effective method for significant noise reduction is to introduce a continuous mold-line link (CML), a fairing surface that smoothly connects the edge of the flap to the wing.[29] This is shown in Fig. 1.1. The experiments are performed using a rigid fairing, but to actually implement this method on an aircraft the fairing must be deformable. Therefore, a flexible plate, or a plate-membrane structure, is an ideal material for the fairing structure because it can be hidden for most of the time and extended when the trailing edge flaps are deployed. A plate has stiffness in bending, while a plate-membrane has both bending stiffness and stiffness due to applied tension. Both types of structures will herein be referred to as ”plates” for simplicity. Despite significant progress in reducing noise from other sources, such as airframe and propulsive devices, an assessment of the overall progress toward the next gen- eration of aircraft showed that additional research in CML’s may be necessary for meeting the noise reduction goal.[4] Because these structures are flexible and would be designed to be light-weight, it is important to analyze their aeroelastic behavior to prevent structural failure due to divergence or flutter. Rectangular panel prob- lems have been studied extensively in the past, specifically the aircraft structural
6 Figure 1.1: Continuous Mold-Line Link panel problem with all edges clamped[9], and the flag flutter problem described ear- lier. However, there is less existing research on the aeroelastic behavior of panels for non-traditional applications, where the more physically correct boundary conditions are not necessarily those that have been extensively studied. NASA’s CML project is just one of many problems that may require the use of novel plate structure de- signs. As the design of aerospace structures focuses more on lighter materials and novel configurations, analytical and experimental results for unexplored boundary conditions and different materials will important in determining viable designs.
1 2 3 X X
X XXX X
X 4 5 6 X X X
X X X XX X
X X X Figure 1.2: Combinations of boundary conditions and flow directions explored in this paper. The diagonal marks indicate a clamped boundary and other boundaries are free with no restraint. The arrows indicate different fluid flow directions that are considered. The ’x’ symbols indicate the presence of a baffle next to the plate boundary instead of free space. Each configuration considers a single fluid flow direction.
This section analyzes the structural dynamics and linear aeroelastic instabilities of a plate using five different sets of boundary conditions in addition to the NASA
7 CML configuration. The boundary conditions are shown schematically in Figure 1.2, in which the diagonal marks indicate clamped boundary, the absence of marks indicate free boundary, the ’x’ symbols indicate the presence of a baffle near the plate boundary instead of free space, and the flow direction is from left to right. The baffle is necessary in the experimental set up - all clamped boundaries are baffled because there must be a structure with which the clamping is applied. However, some free boundaries are also baffled to provide structural support to the entire experimental set up. The theory models the structural dynamics using a plate-membrane model that accounts for flexural rigidity of the material (fourth order derivative) and tension applied to the material (second order derivative). The structural model is coupled to an unsteady vortex lattice aerodynamic model that accounts for the plate as well as any baffle structure surrounding the plate. A modular baffle system is designed around the plate and is able to apply either clamped or free boundary conditions at any of the four edges of the plate. The baffle design and experimental data are presented. Next, the transition between configurations is explored as the axial alignment of the flow is varied. This exploration is motivated by the quantitative and qualitative transition in flutter boundary and motion as the orientation of boundary conditions relative to the flow is changed. For example, for a plate with three sides free, if the trailing edge is free the system becomes unstable in a flutter instability, but if the system is rotated 90 deg so the free edge is aligned with the flow then the dynamic flutter becomes a static divergence. For this section the appropriate mesh and coordinate transformations are presented to analyze structures which are not aligned with the flow. The aeroelastic stability is then solved for as the flow angle is varied. Experimental results are then presented to validate the theoretical models. Fi- nally there are concluding remarks about the research conducted to this point as well
8 a brief discussion of future work.
9 2
Structural Model
In this thesis both a one-dimensional beam and a two-dimensional plate structural model will be derived and discussed. The first structural model developed is that of a beam in bending and torsion. Although the derivation of the governing structural equations and natural mode shapes is straight forward, finding a single source that contains the equations of motion derivation as well as the natural mode shapes for all boundary conditions is difficult. Because the natural modes for a uniform property beam are used for the analysis of the plate, it is convenient to have a complete reference for a beam with all possible combinations of boundary conditions required for the analysis in this thesis. The following section outlines the steps, starting with the energies of a beam, using these energies to derive the unforced equations of motion and the associated natural boundary conditions for a beam, applying a separation of variables technique to determe the spatial mode shapes.
10 2.1 Beam Structural Model Derivation
In order to derive the equations of motion for this structure, the first step is to define the potential and kinetic energy equations for the system. Assuming that the motion of the beam can be described as the linear combination of an out of plane displacement w(x, t) and a rotation around the elastic axis of the beam θ(x, t) the expression for the potential energy of the beam can be written as, where x is the axis which runs along the length of the beam:
1 Z L ∂2w2 1 Z L ∂θ 2 V = EI 2 dx + GJ dx (2.1) 2 0 ∂x 2 0 ∂x
Similarly, the kinetic energy for this system can be written as:
1 Z L ∂w2 1 Z L ∂θ2 T = m dx + Iea dx (2.2) 2 0 ∂t 2 0 ∂t
with m being the mass per unit length and Iea the moment of inertial around the elastic axis per unit length. Now that the kinetic and potential energy expressions have been written, the next step is to apply Hamilton’s Principal. The principle as stated in Dowell and Tang [10] for a conservative system, is that the time integral of the virtual change in kinetic energy minus the virtual change in potential energy must equal zero. This can be expressed mathematically as:
Z t2 [δT − δV ] dt = 0 (2.3) t1
The next step is to rewrite the virtual changes in kinetic and potential energy in terms of a virtual change in w(x, t), (δ(w)) and θ(x, t), (δ(θ)). Starting with the equation for potential energy and applying the virtual change δ operator:
( 2 ) Z t2 Z t2 1 Z L ∂2w 1 Z L ∂θ 2 δV dt = δ EI 2 dx + GJ dx dt (2.4) t1 t1 2 0 ∂x 2 0 ∂x 11 The δ operator may be treated like the differential operation:
" # Z t2 1 Z L ∂2w ∂2w 1 Z L ∂θ ∂θ 2EI 2 δ 2 dx + 2GJ δ dx dt (2.5) t1 2 0 ∂x ∂x 2 0 ∂x ∂x
Knowing that the final result must end up multiplying δw and δθ it is clear that the next step is to integrate by parts. For this equation integrate by parts with respect to x for the EI term. ∂2w ∂2w Let: u = EI and ∂v = δ dx (2.6a) ∂x2 ∂x2
∂ ∂2w ∂w ∂u = EI dx and v = δ (2.6b) ∂x ∂x2 ∂x
Using the integration by parts relationship
Z Z udv = vu − vdu (2.7) and the transformations given in Equations 2.6, The EI portion of Equation 2.5 can be rewritten as:
L Z t2 ∂2w ∂w Z L ∂ ∂2w ∂w EI δ − EI δ dx dt (2.8) 2 2 t ∂x ∂x 0 ∂x ∂x ∂x 1 0
Integrating by parts once more: L L Z t2 ∂2w ∂w ∂ ∂2w EI δ − EI δ (w) dt ∂x2 ∂x ∂x ∂x2 t1 0 0 (2.9) Z t2 Z L ∂2 ∂2w + 2 EI 2 δ (w) dx dt t1 0 ∂x ∂x
Equation 2.9 is in a form that can be directly included into Equation 2.3. A similar exercise can be conducted for the GJ portion of the equation. This yields
" L # Z t2 ∂θ Z L ∂2θ GJ δθ − GJ δθ dx dt (2.10) 2 t ∂x 0 ∂x 1 0 12 Next a similar analysis must be done for the kinetic energy. Again integration by parts is used until there is an integral statement which multiplies δw and another statement which multiplies δθ. Substituting the kinetic energy (T) from Equation 2.2 into Equation 2.3.
( ) Z t2 1 Z L ∂w2 1 Z L ∂θ2 δ m dx + Iea dx dt (2.11) t1 2 0 ∂t 2 0 ∂t
Now applying the δ operator:
Z t2 1 Z L ∂w ∂w ∂θ ∂θ 2m δ + 2Iea δ dx dt (2.12) t1 2 0 ∂t ∂t ∂t ∂t