Whole

School of Aerospace, Civil and Mechanical Engineering
University College

Australian Defence Force Academy

University of New South Wales

ASTUDY OF SWEPT AND UNSWEPT NORMAL
SHOCK WAVE/TURBULENT BOUNDARY LAYER INTERACTION
AND CONTROL BY PIEZOELECTRIC FLAP ACTUATION
A Thesis Submitted for the Degree of
Doctor of Philosophy

Jonathan Stuart Couldrick

July, 2006

i
This piece of work is dedicated to
my Little Chickens.

DECLARATION

I hereby declare that this submission is my own work and to the best of my knowledge
it contains no material previously published or written by another person, nor material
which to a substantial extent has been accepted for the award of any other degree or
diploma at UNSW or any other educational institution, except where due
acknowledgement is made in the thesis. Any contribution made to the research by
colleagues, with whom I have worked at UNSW or elsewhere, during my candidature,
is fully acknowledged.

I also declare that the intellectual content of this thesis is the product of my own work,
except to the extent that assistance from others in the projects design and conception or
in style, presentation and linguistic expression is acknowledged.

Jonathan Stuart Couldrick
July 2006
ii
Abstract iii
Abstract

The interaction of a shock wave with a boundary layer is a classic viscous/inviscid
interaction problem that occurs over a wide range of high speed aerodynamic flows.
For example, on transonic wings, in supersonic air intakes, in propelling nozzles at off-
design conditions and on deflected controls at supersonic/transonic speeds, to name a
few. The transonic interaction takes place at Mach numbers typically between 1.1 and
1.5. On an aerofoil, its existence can cause problems that range from a mild increase in
section drag to flow separation and buffeting. In the absence of separation the drag
increase is predominantly due to wave drag, caused by a rise in entropy through the
interaction.

The control of the turbulent interaction as applied to a transonic aerofoil is addressed in
this thesis. However, the work can equally be applied to the control of interaction for
numerous other occurrences where a shock meets a turbulent boundary layer. It is
assumed that, for both swept normal shock and unswept normal shock interactions, as
long as the Mach number normal to the shock is the same, then the interaction, and
therefore its control, should be the same.

Numerous schemes have been suggested to control such interaction. However, they
have generally been marred by the drag reduction obtained being negated by the
additional drag due to the power requirements, for example the pumping power in the
case of mass transfer and the drag of the devices in the case of vortex generators. A
system of piezoelectrically controlled flaps is presented for the control of the
interaction. The flaps would aeroelastically deflect due to the pressure difference
created by the pressure rise across the shock and by piezoelectrically induced strains.
The amount of deflection, and hence the mass flow through the plenum chamber, would
control the interaction. It is proposed that the flaps will delay separation of the
boundary layer whilst reducing wave drag and overcome the disadvantages of previous
control methods. Active control can be utilised to optimise the effects of the boundary
layer shock wave interaction as it would allow the ability to control the position of the
control region around the original shock position, mass transfer rate and distribution.

Abstract iv
A number of design options were considered for the integration of the piezoelectric
ceramic into the flap structure. These included the use of unimorphs, bimorphs and
polymorphs, with the latter capable of being directly employed as the flap. Unimorphs,
with an aluminium substrate, produce less deflection than bimorphs and multimorphs.
However, they can withstand and overcome the pressure loads associated with SBLI
control.

For the current experiments, it was found that near optimal control of the swept and
unswept shock wave boundary layer interactions was attained with flap deflections
between 1mm and 3mm. However, to obtain the deflection required for optimal
performance in a full scale situation, a more powerful piezoelectric actuator material is
required than currently available.

A theoretical model is developed to predict the effect of unimorph flap deflection on the
displacement thickness growth angles, the leading shock angle and the triple point
height. It is shown that optimal deflection for SBLI control is a trade-off between
reducing the total pressure losses, which is implied with increasing the triple point
height, and minimising the frictional losses.
Acknowledgements

v
Acknowledgements

There are many who have guided me through this project. Some have given me amazing
advice and assistance and some have given me the occasional encouraging word or
friendly ear. In the idea of the butterfly effect, I would not have got to where I am today if
it were not from all of you.

First and foremost, I would like to thank Professor Sudhir Gai, whose critical eye has
raised the standard of my work. He, along with my other supervisors, Doctor John
Milthorpe and Doctor Krishna Shankar, has supported my work financially and
academically for five years. These three wise men have tolerated my style and made
available numerous resources for the completion of this thesis.

Also, to all the men in the ACME workshop who truly can make anything: to Tony Carthy
for giving my jobs priority at the right time; to Mark Dumbrell and Franco Foppoli for
making the impossible with continually finer tolerances; and to Rik Wearings timeless
advice about the stupidity of some of my requests. Also, I would like to thank Bob
Bleakley for the numerous hours of wind tunnel time. Last, but by no means least, to Bill
Doran for the 2am starts, the pearls of wisdom about life and for hopefully a nice cardigan
on graduation (perhaps this should be to Fin).

This thesis would definitely have not been possible if it were not for Ms. Danica Robinson,
who I am indebted to as I would not be here if it were not for her. She has tolerated my
endless phone calls in the pursuit of funds and answers. These funds have allowed the
provision of an amazing UCPRS completion scholarship.

I would also like to acknowledge support for this work through the ARC Linkage
Programme (2003), which enabled me to conduct part of my research in the Engineering
Department at the University of Cambridge, UK as part of a collaborative programme
between CUED and UNSW@ADFA. I am grateful to Dr. Holger Babinsky for his
generous advice and to Ms. Harriet Holden for help during the normal shock experiments.

Lastly, I would like to thank my family and friends for their support and encouragement.
To Antti, Jeff, Orio and Stephen I dont think I would have had as much fun with my work
if it werent for you guys; especially the Uruguay contingent.
vi
Contents

vii
CONTENTS

Declaration i
Abstract iii
Acknowledgements v
Table of Contents vii
Table of Figures xiv
Table of Tables xx
Nomenclature xxi

CHAPTER1 Shock Wave Boundary Layer Interaction Control 1-40
1.1 Introduction 1
1.1.1 Present Approach: 2
1.2 Literature Survey 5
1.2.1 The UNS Interaction 6
1.2.1.1 INCIPIENT SEPARATION 8
1.2.1.2 SEPARATION 9
1.2.2 The SNS Interaction 13
1.2.3 SBLI Control 18
1.2.3.1 MASS TRANSFER (INJECTION OR SUCTION) 18
1.2.4 Passive Control of Shock Wave/Boundary Layer Interaction (PCSBLI) 21
1.2.4.1 THE EFFECT OF PCSBLI ON THE FLOW FIELD 22
1.2.4.2 FACTORS AFFECTING PCSBLI 25
1.2.4.3 SLOT AND GROOVE CONTROL 28
1.2.4.4 HYBRID 29
1.2.5 Mesoflaps for Aeroelastic Recirculation Transpiration (MART) 31
1.2.6 Piezoelectric Actuators for Flow Control 37
1.3 Present Investigation 38
1.3.1 Active Control of Shock/Boundary Layer Interaction (ACSBLI) 38
CHAPTER 2 Experimental Arrangement 41-60
2.1 ADFA Experimental Facility and model 41
2.1.1 The Wind Tunnel Facility 41
2.1.2 Wedge/Shock Generator 43
2.1.3 Model design 44
2.1.4 Flow visualisation 45
2.1.5 Oil flow visualization 47
Contents

viii
2.1.6 Pressure Sensitive Paints Couldrick et al. (2004b) 47
2.1.7 Adiabatic Wall Temperature 49
2.1.8 Data Acquisition and Controller System 50
2.1.9 Experimental Accuracy 51
2.1.10 Voltage Generators/Power supplies Physik Instrumente (1996) 51
2.2 Cambridge experimental facility and model 53
2.2.1 Wind Tunnel arrangement 53
2.2.2 Co ordinate system. 54
2.2.3 Unimorph control device 56
2.2.4 Measurement locations 57
2.2.5 Flow visualization 58
2.2.6 Calculation of velocity Smith (2002) 58
2.2.7 Measurement apparatus calibration and checks 59
2.2.8 Experimental Accuracy 60
CHAPTER 3 Piezoelectric Flap Actuators 61-85
3.1 Introduction 61
3.2 Smart Materials 61
3.3 Piezoelectric Ceramic Properties 63
3.3.1 Aging 64
3.3.2 Temperature, Voltage and Stress limitations 64
3.3.2.1 TEMPERATURE LIMITATIONS 64
3.3.2.2 VOLTAGE LIMITATIONS 65
3.3.2.3 MECHANICAL STRESS LIMITATIONS 65
3.4 Experimentally obtained PZT properties 65
3.4.1 Youngs Modulus of Elasticity & Poissons Ratio 65
3.4.2 Piezoelectric Material Dielectric Constant 67
3.5 Bender Actuators 69
3.5.1 Unimorphs 69
3.5.2 Bimorphs 70
3.5.3 Multimorphs 70
3.6 Active Control Structural Configurations 71
3.6.1 Unimorph configuration 71
3.6.2 Bimorph/Multimorph Configuration 71
3.7 Alternate Active Control Configurations 74
3.7.1 Alternate Flap Configuration 74
3.7.2 Alternate Flapless Configuration 75
3.8 Multimorphs/Bimorphs Testing 75
3.8.1 Multimorph Electrical configuration 76
Contents

ix
3.8.2 Multimorph Electrical Loading No Point Load Applied 76
3.8.3 Multimorph Electrical Loading Point Load Applied 78
3.8.4 Bimorphs 80
3.8.5 Multimorph/Bimorph Summary 80
3.9 Unimorph testing 81
3.9.1 Unimorph Substrate 81
3.9.2 Unimorph configuration 81
3.9.3 Unimorph Summary 84
3.10 Piezoelectric Actuator Summary 84

CHAPTER 4 Unimorph Actuator Deflection 87-116
4.1 Introduction 87
4.2 Literature Survey 88
4.3 Classic Theory 92
4.4 Finite Element Modelling (FEM) 94
4.5 17.64kPa Results Line Optimisation (Couldrick et al. (2003)) 95
4.5.1 Theoretical deflection variation with substrate thickness for no load 95
4.5.2 Theoretical deflection variation with substrate Youngs Modulus for no load 96
4.5.3 17.64kPa Constant Pressure Load Predictions 97
4.5.4 Quadratic Pressure load predictions 98
4.5.5 17.64kPa Pressure Deflection Summary 100
4.6 30kPa Results Surface Optimisation 101
4.6.1 Deflection for no load 101
4.6.2 Deflection for 30kPa Constant Pressure Load 103
4.6.3 Quadratic Pressure load predictions 106
4.6.4 Classic Theory for Quadratic pressure loads 108
4.6.5 30kPa Pressure Load Summary 110
4.7 Improved FEM Prediction 111
4.8 Unimorph Tip Deflection Contour 112
4.9 Conclusion 115

CHAPTER 5 Swept Normal SBLI Control 117-154
5.1 Introduction 117
5.2 Uncontrolled Swept SBLI 117
5.2.1 Schlieren 117
5.2.2 Oil Flow visualisation 118
5.3 Piezoelectric Flap Actuator Controlled SBLI 120
5.3.2 Surface Pressures 122
5.3.2.1 BOUNDING STREAMLINE L1 124
Contents

x
5.3.2.2 BOUNDING STREAMLINE L3 126
5.3.2.3 LEADING LEG OF THE LAMBDA SHOCK 127
5.3.3 Drag Coefficient 128
5.3.4 Swept Normal SBLI Unimorph Control Summary 130
5.4 Pressure Sensitive Paint (PSP) Investigation of Swept SBLI Control 131
5.4.1 PSP Technique 131
5.4.2 Uncontrolled SBLI 131
5.4.3 Closed Unimorph (500V) 132
5.4.4 Uncontrolled Unimorph (0V) 134
5.4.5 Fully Deflected Unimorph ( 500V) 135
5.4.6 Flap deflection effects 137
5.4.7 PSP Summary 137
5.5 Mechanically deflected flap SBLI Control 138
5.5.1 Oil Flow Visualisation 138
5.5.1.1 0MM FLAP DEFLECTION 138
5.5.1.3 2MM AND 3MM FLAP DEFLECTION 140
5.5.1.4 LEADING LEG SHOCK ANGLE 142
5.5.2 PSP Results 143
5.5.2.1 THE UNCONTROLLED CASE 144
5.5.2.4 2MM AND 3MM FLAP DEFLECTION 146
5.5.2.5 COMPARISON WITH CONTROLLED PSP DATA 148
5.5.3 Discrete Pressure data 149
5.5.3.1 THE L1 STREAMLINE 149
5.5.3.2 THE L3 STREAMLINE 150
5.6 Swept Normal SBLI Summary 152

CHAPTER 6 Unswept Normal SBLI Control At M=1.3 155-188
6.1 Uncontrolled SBLI 155
6.2 Unimorph Controlled Flap Deflection 157
6.2.2 Schlieren and Total Pressure traverses 159
6.2.3 Surface Pressure Measurements 163
6.2.4 Unimorph Control Limitation 165
6.3 Mechanically Fixed Flap Deflections 0mm 165
6.3.1 Schlieren and Total Pressure Traverses 166
Contents

xi
6.4 Mechanically Set Flap Deflections 1mm to 3mm 170
6.4.1 Schlieren and Total Pressure Plots 170
6.4.2 Total Pressure Maps 176
6.4.5 Velocity Traverses 183
6.5 Summary 185

CHAPTER 7 Unswept Normal SBLI Control at M=1.5 189-220
7.1 Uncontrolled SBLI 189
7.2 Unimorph Controlled SBLI 191
7.2.1 Schlieren and Total Pressure traverses 191
7.2.3 Discrete Pressure Measurements 197
7.2.4 Unimorph Control Limitation 198
7.3 Mechanically Set Flap Deflections 0mm 199
7.4 Mechanically Set Flap Deflections 1mm to 3mm 203
7.4.2 Total Pressure Maps 208
7.4.5 Velocity Traverses 214
7.5 Normal SBLI Control at M=1.5 Summary 217
7.6 Normal SBLI Control Summary 219

CHAPTER 8 Theoretical Optimisation 221-256
8.1 Theoretical 2 D Aerodynamic Model 221
8.1.1 Downstream Mass Flow (suction) through a single hole/slot 222
8.1.2 Total Slot Length and Mass Flow Injection Angle 224
8.1.3 The displacement thickness growth angle 225
8.1.4 Leading Leg Shock Angle () 227
8.1.5 Triple Point Height 228
8.2 Analysis of the MART System for an UNS Interaction at M=1.4 230
8.2.1 The MART Six Flap System 230
8.2.2 The MART Four Flap System 233
Contents

xii
8.3 Analysis of the Unimorph System 236
8.3.1 The Unimorph System for an UNS Interaction at M=1.5 236
8.3.2 The Unimorph System for a SNS Interaction at M
n
=1.3 239
8.3.3 The Unimorph System for an UNS Interaction at M=1.3 241
8.3.3.1 MODIFIED THEORETICAL MODEL 242
8.4 Piezoelectric Actuated Optimised SBLI Control 246
8.4.1 Optimal Deflection for SBLI Control 246
8.4.2 Acting Pressure Distribution 246
8.4.3 Piezoelectric Requirements 246
8.4.3.1 SNS INTERACTION CONTROL (M
N
=1.3) 246
8.4.3.2 UNS INTERACTION CONTROL (M=1.3) 249
8.4.3.3 UNS INTERACTION CONTROL (M=1.5) 250
8.5 Optimal Shock Location 252
8.6 Further work 253
8.7 Conclusion 254

CHAPTER 9 Concluding Remarks 257

References 261-268

Appendix A Material Properties
A1 Piezoelectric Properties
A2 Aluminium Data Sheet
A3 Aluminium Three Point Test

Appendix B Tip Deflection Theory (Structural Modelling)
B1 Classic Plate Theory
B2 FEM Grid Resolution

Appendix C Programs
C1 Finite Element Modelling (Ansys)
C2 Experimental Data Acquisition (ASYST)
C3 Pressure Sensitive Paint (Matlab)

Appendix D Electronics
D1 OEM Amplifiers

Appendix E Boundary Layer Theory
Contents

xiii
E1 Mass Flow Within a Boundary Layer

Appendix F Published Papers
F1 Couldrick, J. S., Shankar, K., Gai, S. and Milthorpe, J. (2003) "Design of "Smart"
Flap Actuators for Swept Shock Wave/Turbulent Boundary Layer Interaction
Control", Structural Engineering and Mechanics: An International Journal
16(5): pp. 519-532.
F2 Couldrick, J. S., Gai, S., Milthorpe, J. and Shankar, K. (2004) "Active Control of
Swept Shock Wave/Turbulent Boundary Layer Interactions", The Aeronautical
Journal 108(2): pp. 93-102.
F3 Couldrick, J. S., Gai, S., Milthorpe, J. and Shankar, K. (2004) "Investigation of
Active Control of Swept Shock Wave/Turbulent Boundary Layer Interactions using
pressure sensitive paints", The Aeronautical Journal 108(9): pp. 483-490.
F4 Couldrick, J. S., Gai, S., Milthorpe, J. and Shankar, K. (2005) "Normal Shock
Wave/Turbulent Boundary Layer Interaction Control using "Smart" Piezoelectric
Flap Actuators", The Aeronautical Journal 109(11): pp. 577-583.
List of Figures xiv
LIST OF FIGURES

Figure 1.1 Active Control of Shock/Boundary Layer Interaction using piezoelectric actuators
Figure 1.2 - Weak UNS Interaction, Green (1969)
Figure 1.3 Holographic Interferometry of the uncontrolled UNS Interaction at M=1.1, Gibson et al.
(2000)
Figure 1.4 a) Wall Shear stress distribution through a SBLI from Green (1969) and b) The separation
bubble growth from Delery (1985) with increasing Mach number to demonstrate incipient
separation
Figure 1.5 Lambda structure of the separated UNS interaction, Green (1969)
Figure 1.6 Detailed lambda structure, Atkin and Squire (1992)
Figure 1.7 Supersonic Tongue, Seddon (1960)
Figure 1.8 Three dimensional Sharp Fin Induced SBLI with an unseparated flow structure, Kubota and
Stollery (1982)
Figure 1.9 Surface flow beneath swept normal shocks and the flow viewed parallel to the shocks for a)
unseparated and b) separated flow from Green (1969)
Figure 1.10 Three dimensional Sharp Fin Induced SBLI with a separated flow structure, Kubota and
Stollery (1982)
Figure 1.11 SNS Interaction showing secondary separation from Settles and Dolling (1992)
Figure 1.12 Shadowgraph and pressure distributions over an aerofoil at 6 with a) no boundary layer
control and b) blowing through an upstream slot, Pearcey (1961)
Figure 1.13 Active control on a supercritical transonic aerofoil by suction through a double slot at a) 4>
and b) 5>- Thiede et al. (1984)
Figure 1.14 Passive Control of Shock/Boundary Layer Interaction
Figure 1.15 Drag coefficient variation with freestream Mach number for solid and porous surface
aerofoils from Nagamatsu et al. (1987)
Figure 1.16 SBLI Structure with Slot Control, Smith et al. (2002)
Figure 1.17 Slot control positions on the side-wall to study SNS Interactions from Babinsky et al.
(1999)
Figure 1.18 Hybrid SBLI Control with PCSBLI and downstream suction, Delery and Bur (1999)
Figure 1.19 Mesoflaps for Aeroelastic Transpiration a) Four stream wise flap structure, Wood et al.
(1999) and b) SBLI structure, Jaiman et al. (2003)
Figure 1.20 The SBLI at M=1.37 a) uncontrolled and with b) MART Control of the SBLI with a four
flap 191 m array, Lee et al. (2002)
Figure 1.21 Active Control of Shock/Boundary Layer Interaction
Figure 2.1 ADFA Supersonic Blow-down Wind Tunnel Schematic.
Figure 2.2 ADFA Supersonic Blow-down Wind Tunnel a) photo of the Mach 2 liners with wedge on
the floor to create a swept shock and b) Schematic showing dimensions.
Figure 2.3 ADFA Test Section with Shock Generator.
List of Figures xv
Figure 2.4 Control Plate for Swept Normal SBLI Control showing a) Pressure Port and Unimorph
Layout, b) Streamlines and c) A Photograph of the Unimorph Control Plate.
Figure 2.5 Unimorph Flap Design for Swept Normal SBLI Control.
Figure 2.6 ADFA Schlieren System.
Figure 2.7 Camera/Test Surface/Light Source Set-up for PSP experiments.
Figure 2.8 High Voltage Amplifier with casing from Physik Instrumente
Figure 2.9 University of Cambridge Blow-down Supersonic Wind Tunnel #2.
Figure 2.10 Cambridge Supersonic Wind Tunnel Arrangement with interaction Control Plate.
Figure 2.11 University of Cambridge Control Plate showing pressure port layout for a) piezoelectric
actuation unimorph flap control and b) mechanically pre-set flap deflection control.
Figure 2.12 Unimorph Flap Design for Unswept Normal SBLI Control.
Figure 2.13 a) Unswept Normal SBLI Control Plate showing pressure port layout and b) the mechanical
deflection devices
Figure 2.14 a) Pitot tube set-up with b) a flat head pitot tube and c) a straight pitot tube - Smith (2002)
Figure 3.1 - Piezoelectric ceramic clamping configuration for tensile testing
Figure 3.2 Stress strain curve for piezoelectric ceramic PZT-5H.
Figure 3.3 Stress strain curve for piezoelectric ceramic PZT-5H.
Figure 3.4 A Basic Unimorph
Figure 3.5 A Basic Bimorph
Figure 3.6 A Multimorph
Figure 3.7 Original Unimorph Testing Configuration
Figure 3.8 The Final Unimorph Configuration
Figure 3.9 A Basic Bimorph Flap Configuration
Figure 3.10 Adapted Bimorph Flap Configuration
Figure 3.11 An Alternative Active Flap Control Configuration.
Figure 3.12 An Alternative Active Control without Flaps.
Figure 3.13 Multimorph Electrical Configurations a) Manufacturer Configuration, Morgan Matroc)
(2000) and b) Alternative #2 Configuration
Figure 3.14 Multimorph Performance for Datum Voltages of a) 80V, b) 90V and c) 100V
Figure 3.15 Multimorph Performance with equal Applied and Datum Voltages.
Figure 3.16 Applied Voltage Required to Obtain Zero Deflection for Variable Point Loads
Figure 3.17 Unimorph Tip Deflection variation with Applied Voltage.
Figure 3.18 Unimorph Tip Deflection variation with Point Load
Figure 3.19 Unimorph tip deflection for given applied voltages and various point loads
Figure 4.1 Modified Unimorph Flap configuration.
Figure 4.2 CLPT Model showing no deflection or slope at the start of the unimorph
Figure 4.3 Unimorph Finite Element Model in ANSYS 6.0.
Figure 4.4 Unimorph deflection vs. Normalised substrate thickness due to 500V applied voltage
(1000Vmm
-1
).
List of Figures xvi
Figure 4.5 Deflection vs. Normalised Substrate Youngs Modulus for a 500V applied voltage
(1000Vmm
-1
).
Figure 4.6 Tip deflection for a 17.64kPa uniform pressure load for varying substrate: a) thickness (E
s
=
70GPa); and b) Youngs Modulus (t
s
= 1.1mm)
Figure 4.7 Quadratic pressure loading.
Figure 4.8 FEM tip Deflection predictions for a quadratic load for varying substrate: a) thickness; and
b) Youngs Modulus.
Figure 4.9 Contours of unimorph deflection (mm), with a 500V applied field, for a) Classic Theory and
b) FEM.
Figure 4.10 Contours of unimorph deflection (mm) using Classic Theory, with a uniform pressure load,
for applied voltages of a) -500V open flaps and b) 500V closed flaps
Figure 4.11 Contours of unimorph deflection (mm) using FEM, with a uniform pressure load, for
applied voltages of a) -500V open flaps and b) 500V closed flaps
Figure 4.12 Quadratic pressure loading showing isobars
Figure 4.13 Contours of unimorph deflection (mm) using FEM, with a quadratic pressure load, for
applied voltages of a) -500V flaps open and b) 500V closed flaps
Figure 4.14 Unimorph deflection (mm), using Classic Theory for a quadratic load, whilst varying
normalised substrate thickness and Youngs Modulus for applied fields of a) -500V open
flaps and b) 500V closed flaps
Figure 4.15 Top Profile of the unimorph deflection, Z axis, with a uniform 17.64kPa pressure for
applied fields of a) -500V and b) +500V
Figure 4.16 Exaggerated Side profile of the unimorph deflection, Z axis, with a uniform 17.64kPa
pressure for applied fields of a) -500V and b) +500V
Figure 5.1 Schlieren photograph of the uncontrolled swept SBLI.
Figure 5.2 Oil Flow visualisation of the uncontrolled swept SBLI a) tunnel sidewall, b) control surface
and c) pen enhanced oil flow.
Figure 5.3 Pen enhanced oil flow visualisation of the swept SBLI with a 0V unimorph control.
Figure 5.4 a) Co-ordinate system and b) Variation of
1
[= (
fs
-
s
)] for front shock with normal Mach
number (Accuracy 1.5), after Squire (1996).
Figure 5.5 a) Control Plate showing Pressure Port and Unimorph Layout and b) the Surface Pressure
distribution along the normal to the wedge shock.
Figure 5.6 Surface Pressure distribution along the back edge of the upstream unimorph.
Figure 5.7 a) Bounding Streamlines L1 and L3 and b) the surface pressures along L1 for swept SBLI
Figure 5.8 Surface Pressures for swept SBLI control along L3.
Figure 5.9 Variation of
ll
with Normal Mach number, Squire (1996).
Figure 5.10 Drag Coefficient vs. Applied Voltage for unimorph controlled swept SBLI.
Figure 5.11 Pressure distribution map of the uncontrolled swept SBLI using PSP.
Figures 5.12 PSP Pressure distribution map of unimorph swept SBLI control with 500V applied voltage
(Closed).
List of Figures xvii
Figure 5.13 a) 3D Bubbles of Influence of slot control of the SBLI and b) the effect of slot strength,
Smith et al. (2002).
Figures 5.14 PSP Pressure distribution map of unimorph swept SBLI control with a 0V applied voltage.
(Open actuator inactive)
Figures 5.15 PSP Pressure distribution map of unimorph swept SBLI control with a -500V applied
voltage. (Open)
Figure 5.16 Oil flow visualisation of swept SBLI control with 0mm deflected flaps test section.
Figure 5.17 Oil flow visualisation of Swept SBLI Control at M=1.3 with a 1mm flap deflection
Figure 5.18 Oil flow visualisation of Swept SBLI Control at M=1.3 with flap deflections of a) 2mm and
b) 3mm.
Figure 5.19 Variation of
2
[= (
cl
-
s
+ 4.5)] for front shock with normal Mach number, after Squire
(1996).
Figure 5.20 Pressure distribution map of the uncontrolled swept SBLI using PSP
Figure 5.21 Pressure distribution map of the Swept Normal SBLI using PSP with flaps deflected to
0mm
Figure 5.22 Pressure distribution map of the Swept Normal SBLI using PSP with 1mm deflected flaps.
Figure 5.23 Pressure distribution map of the Swept SBLI using PSP with flaps deflected to a) 2mm and
b) 3mm.
Figure 5.24 Surface Pressures along L1 for swept SBLI control
Figure 5.25 Surface Pressures along L3 for swept SBLI.
Figure 6.1 a) Schlieren pictures and b) total pressure traverses of the uncontrolled SBLI at M=1.3,
Holden (2004).
Figure 6.2 Oil Flow visualisation of the uncontrolled SBLI at M=1.3, Holden (2004).
Figure 6.3 Oil Flow visualisation of SBLI with Unimorph Control at 0V at M=1.3 and Main Shock at
X/
o
=0.
Figure 6.4 Schlieren pictures of unimorph controlled SBLI at M=1.3 with a) closed flaps (500V), b)
inert flaps 0V & c) open flaps (-500V).
Figure 6.5 Discrete pressure data of unimorph control at M=1.3 with
Figure 6.6 a) Schlieren pictures and b) Total pressure traverses of flap control of the SBLI with 0mm
deflection at M=1.3.
Figure 6.7 Oil Flow visualisation of SBLI with 0mm Flap Deflection at M=1.3 and the Main Shock at
X/
o
=0.
Figure 6.8 Discrete pressure data of flap control with 0mm deflection at M=1.3.
Figure 6.9 a) Schlieren pictures and b) Total pressure traverses of the M=1.3 SBLI control with 1mm
flap deflection
flap deflection
flap deflection
Figure 6.12 Schlieren picture of Mesoflaps with Aeroelastic Transpiration Control of the SBLI at M=
1.4 , Lee et al. (2002).
List of Figures xviii
Figure 6.13 Total Pressure Maps with 0mm flap deflection at M=1.3.
Figure 6.14 Total Pressure Maps of the M=1.3 SBLI control flaps deflected to a) 1mm and b) 2mm
Figure 6.15 Total Pressure Maps of the M=1.3 SBLI control with 3mm flap deflection.
Figure 6.16 Oil Flow visualisation of flap control of the SBLI deflected to a) 1mm, b) 2mm and c)
3mm.
Figure 6.17 Discrete pressure data for SBLI control at M=1.3 with flaps deflected to a) 1mm and b)
2mm.
Figure 6.18 Discrete pressure data for SBLI control at M=1.3 with 3mm flaps deflection.
Figure 6.19 Velocity traverses for M=1.3 SBLI control with flaps deflected to a) 0mm and b) 1mm
Figure 6.20 Velocity traverses for M=1.3 SBLI control with flaps deflected to a) 2mm and b) 3mm.
Figure 7.1 a) Schlieren pictures and b) total pressure traverses of the M=1.5 SBLI uncontrolled, after
Holden (2004).
Figure 7.2 Oil Flow visualisation of the uncontrolled M=1.5 SBLI, Holden (2004).
Figure 7.3 a) Schlieren pictures and b) Total pressure traverses of unimorph control of the M=1.5 SBLI
with closed flaps (+500V).
Figure 7.4 a) Schlieren pictures and b) Total pressure traverses of unimorph control of the M=1.5 SBLI
with open flaps (-500V).
Figure 7.5 Oil Flow visualisation of unimorph controlled SBLI at M=1.5with 0V.
Figure 7.6 Discrete pressure data of unimorph control for the M=1.5 SBLI a) closed flaps (+500V), b)
inert flaps 0V & c) open flaps (-500V)
Figure 7.7 a) Schlieren pictures and b) Total pressure traverses of the M=1.5 SBLI with flap control at
0mm deflection.
Figure 7.8 Oil Flow visualisation of the M=1.5 SBLI with 0mm flap deflection control.
Figure 7.9 Discrete pressure data of flap control of the M=1.5 SBLI with 0mm flap deflection control.
flap deflection.
flap deflection.
flap deflection.
Figure 7.13 Total Pressure Maps for flap control of the SBLI deflected to a) 0mm and b) 1mm.
Figure 7.14 Total Pressure Maps for flap control of the SBLI deflected to a) 2mm and b) 3mm.
Figure 7.15 Oil Flow visualisation of the M=1.5 SBLI control with flaps deflected to a) 1mm, b) 2mm
and c) 3mm.
Figure 7.16 Discrete pressure data for the M=1.5 SBLI control with flaps deflected to a) 1mm, b) 2mm
and c) 3mm.
Figure 7.17 Velocity traverses for flap control of the SBLI deflected to a) 0mm and b) 1mm.
Figure 7.18 Velocity traverses for flap control of the SBLI deflected to a) 2mm and b) 3mm.
Figure 8.1 Theoretical SBLI Flap Control Model
Figure 8.2 a) Actual Downstream Flap Tip Deflection and b) Modelled Downstream Flap Tip Deflection
Figure 8.3 Flow configuration and definitions (after Doerffer and Bohning (2000))
List of Figures xix
Figure 8.4 The Slot Created by Upstream Flap Deflection a) dimensions and b) mass flow injection
angle
Figure 8.5 Mass injection a) without flap deflection and b) with flap deflection.
Fugure8.6 Calculated Angles (Not to scale)
Figure 8.7 Model of Mass Flow Angles
Figure 8.8 Model for calculating the shock angle from the total deflection angle for the upstream leg of
the lambda shock
Figure 8.9 The Modelled Lambda Structure including heights
Figure 8.10 The Modelled Lambda Structure including Lengths
Figure 8.11 Displacement Thickness growth angle variation for the MART six-flap deflection
Figure 8.12 Schlieren pictures of the MART Six-Flap System with mesoflap thicknesses of a)63.5m,
b)101.9m and c)150.6m, - Hafenrichter et al. (2003)
Figure 8.13 The leading leg shock angle variation for mesoflap deflection of the MART six-flap
deflection assuming no mass injection effect.
Figure 8.14 Displacement Thickness growth angle variation for deflection of the MART Four-flap
system
Figure 8.15 Schlieren pictures of the MART Four-Flap System with a mesoflap thickness of 190.5m
Hafenrichter et al. (2003)
Figure 8.16 The leading leg shock angle variation for mesoflap deflection of the MART four-flap
deflection
Figure 8.17 Displacement Thickness growth angle variation for Unimorph flap deflection at M
1
=1.5
Figure 8.18 The leading leg shock angle variation for unimorph deflection.
Figure 8.19 The Triple Point Height variation with unimorph deflection
n1
=1.3.
Figure 8.21 The leading leg shock angle variation for unimorph deflection
1
=1.3
Figure 8.26 Deflection Contours for a Quadratic 17.64kPa pressure load with a -500V and a
piezoelectric charge constant d
31
of -625x10
-12
mV
-1
.
31
of -750x10
-12
mV
-1
.
31
of a) -750x10
-12
mV
-1
and b) -1000x10
-12
mV
-1
.
Figure 8.29 Deflection Contours for a Quadratic 30kPa pressure load with a -500V and a piezoelectric
charge constant d
31
of a) -875x10
-12
mV
-1
and b) -1250x10
-12
mV
-1
.
Figure 8.30 Unswept Normal SBLI Control with a rear shock position at M=1.5 using flaps deflected to
a) 0mm, b) 1mm, c) 2mm and d) 3mm.
List of Tables xx
LIST OF TABLES

Table 1.1 Varying porosity effect on leading leg angles for a central shock position on a passive control
plate (100mm in length) - Gibson et al. (2000)
Table 1.2 Oblique Shock Angles for various shock positions on a 4% porosity passive control plate
(100mm in length) - Gibson et al. (2000)
Table 1.3 MART Flap Deflections for the six-flap and the four-flap arrays
Table 2.1 UNSW@ADFA Oncoming Boundary Layer Properties.
Table 2.2 University of Cambridge Oncoming Boundary Layer Properties.
Table 3.1 Electrical strain parallel (1) and normal (2) to the applied load for the front (F) and rear (R) of
piezoelectric ceramic.
Table 4.1 Tip deflection predictions for CLPT, Final FEM Grid Resolution and FEM with Increased
20x20x5 Grid Resolution for Uniform/Quadratic Pressure Load
Table 6.1 Summary of shock Angles for the uncontrolled and unimorph controlled SBLI at a nominal
Mach number of 1.3, (accuracy of 1).
Table 6.2 Summary of triple point height and boundary layer thickness for the uncontrolled and
unimorph controlled SBLI at M=1.3, (accuracy of 1mm).
Table 6.3 Shock Angles for the uncontrolled and flap controlled SBLI at M=1.3, (accuracy of angles
1).
Table 6.4 Summary of triple point heights and boundary layer thicknesses for flap controlled SBLI at
M=1.3 for various Z positions and flap deflections as determined from figs. 6.6 & 6.9,
(accuracy of 1mm).
Table 7.1 Summary of shock Angles for the uncontrolled and unimorph controlled SBLI at M=1.5.
Table 7.2 Summary of triple point heights and boundary layer thicknesses for flap controlled M=1.5
SBLI at various spanwise positions and flap deflections as determined from figs. 7.3a & 7.4b.
Table 7.3 Shock Angles for the uncontrolled and flap controlled SBLI at M=1.5, (accuracy of 1).
Table 7.4 Summary of triple point heights and boundary layer thicknesses for flap controlled SBLI at
M=1.5 for various Z positions and flap deflections as determined from figs. 7.6 & 7.9a to c,
(accuracy of 1mm).
Table 8.1 Theoretical Model Inputs
Table 8.2 Optimal deflections for unimorph controlled SBLI
Nomenclature xxi
NOMENCLATURE
Symbols
A Extensional Stiffness Matrix
B Coupling Stiffness Matrix
B
1
Hole Flow Variable
c
1
Hole Flow Constant
c
d
Point Drag Coefficient
C
d
Drag Coefficient
C
1
, C
2
, C
3
, C
4
, C
5
Integration Constants
d Diameter of pitot probe
d
1
Hole Flow Constant
d
11
1
st
Element of the Bending Element of the Compliance Matrix
d
31
Piezoelectric Charge Constant
D Bending Stiffness Matrix
dP Pressure Difference
E Youngs Modulus
(EI)
eff
Effective Bending Stiffness
G Gap
I Illuminescence Intensity
k Temperature Dependant Coefficient
l Test Section Length
L1, L3 Streamline
L Length
m& Mass Flow
M Moment
M
n
Normal Mach Number
N Load
P Structural Point Load
p, P Pressure
r Recovery Factor
R Reattachment Position
S Separation Position
S1, S2, S3 Shock Position
Nomenclature xxii
t Thickness
T Temperature
u Streamwise Velocity
V Applied Voltage
W Structural Pressure Load
x Normalised Substrate Thickness
X Streamwise Distance
Y Spanwise Distance
z Normalised Substrate Youngs Modulus
Z Vertical Height
| Oblique Shock Angle/Leading Leg Shock Angle
; Ratio of Specific Heats (1.4 for air)
o
o
Oncoming Boundary Layer Thickness
o
o
*
Oncoming Displacement Thickness
A Unimorph Deflection
A
1
Difference in Shock Angles (|
fs
- |
s
)
A
2
Difference in Shock Angles (|
cl
- |
s
+ 4.5)
c Strain
u Deflection Angle
u
o
Oncoming Momentum Thickness
A Lambda
k Curvature
v Poissons Ratio
o Stress
B
w
Wall Shear Stress
9 Unit Reynolds Number
Nomenclature xxiii
Superscript
o Mid Plane

Subscripts
aw Adiabatic Wall
cl Convergence Line
e Boundary Layer Edge
ef Effective
fs Front shock
G Gap
i Induced
inf Upstream Influence
inj, injection Mass Injection
I Initial
ll Leading Leg
0 Stagnation
P Piezoceramic Properties
ref PSP Reference Picture
S Substrate Properties
S Main Shock
s static
t Test Section
T Total
U Unimorph
V Applied Voltage
w Wall
W Pressure Load
X Streamwise Length
Y Vertical Height
1 Freestream (Upstream of Shock)
2 Freestream (Downstream of Shock)
Freestream
E Swept Shock
Nomenclature xxiv
Abbreviations
ACSBLI Active Control of SBLI
ARC Australian Research Council
EDM Electro Discharge Machining
FC Flap Controlled
FEM Finite Element Modelling
MS Main Shock
PCSBLI Passive Control of SBLI
PSP Pressure Sensitive Paint
PZT Lead Zirconate Titanate
SBLI Shock Wave/Boundary Layer Interaction
SC Slot Controlled
SNS Swept Normal Shock
ULS Uncontrolled Lambda Structure
UNS Unswept Normal Shock

Chapter 1. Shock Wave Boundary Layer Interaction Control

1-1
CHAPTER 1. Shock Wave Boundary Layer Interaction Control

1.1. - Introduction

The interaction of a shock wave with a boundary layer is a classic viscous/inviscid
interaction problem that occurs over a wide range of high speed aerodynamic flows;
such as on transonic wings, in supersonic air intakes, in propelling nozzles at off-design
conditions and on deflected controls at supersonic/transonic speeds, to name a few. The
transonic interaction takes place at Mach numbers typically between 1.1 and 1.5. On an
aerofoil its existence can cause problems that range from a mild increase in section drag
to flow separation and buffeting. In the absence of separation the drag increase is
predominantly due to wave drag, caused by a rise in the flow entropy through the
interaction.

In flows without boundary layers a shock wave would meet, be generated or reflected
by a solid surface. In such a flow the pressure at the surface would increase
discontinuously across the shock. However, the presence of a viscous boundary layer
does not allow this to happen, as the inner part of the boundary layer has subsonic
velocities and discontinuities are not possible.

In real flows, the interaction has a complicated structure due to the mixed flow regions
with adjacent subsonic and supersonic regions. The viscous boundary layer is
predominantly subsonic, which allows pressure disturbances to be transmitted in both
upstream and downstream directions. The interaction generates large shear gradients
normal to the wall and at the same time the low energy air is dragged downstream.
Within the outer supersonic region the effect of viscosity is relatively small and the flow
can be defined in terms of the shock equations.

The control of the turbulent interaction as applied to a transonic aerofoil will be
addressed in this thesis. However, the work can be applied to the control of the
interaction for numerous other occurrences where a shock meets a turbulent boundary
layer. Chapman et al. (1958) showed that the interaction is relatively independent of
the mode of shock generation. Furthermore, Squire (1996) showed that, for both swept

1-2
normal shock and unswept normal shock interactions, as long as the Mach number
normal to the shock is the same, then the interaction, and therefore its control, should
not differ greatly.

The turbulent interaction was chosen, rather than laminar or transitional, due to the high
Reynolds number realised in many practical applications that could incorporate the
control method presented in this thesis. Its application to a transonic aerofoil was
selected, where the boundary layer over the upper surface interacts with the terminating
shock and the locally supersonic flow returns to subsonic speeds.

Numerous schemes have been suggested to control the interaction. However, they have
generally been marred by the drag reduction obtained being negated by the additional
drag due to the power requirements, for example the pumping power in the case of mass
transfer and the drag of the devices in the case of vortex generators.

Passive control of the interaction, where a porous surface covering a plenum chamber is
placed underneath the shock region, typically produces a significant reduction in drag
above a normal Mach number of 1.25, or for a freestream Mach number over an aerofoil
of 0.81. However, Nagamatsu et al. (1987) state that below a Mach number of about
1.25 the viscous drag rise due to the rough porous surface is more than the wave drag
savings obtained by the passive control. Also, it is not possible to turn passive control
on or off, to control the shock position or the amount of mass transfer and its
distribution during operation. Furthermore, passive control increases the drag
coefficient at off-design conditions, that is either in the absence of a shock or when the
shock location is outside of the control region. Noticeable contributions to the study of
passive control of transonic SBLI have been by Babinsky (1999), Chen (1984), Gibson
(2000), Nagamatsu (1985)(1987) and Raghunathan (1987)(1988).

1.1.1. - Present Approach:
In this thesis a system of piezoelectrically controlled flaps is presented for the control of
the interaction, see fig. 1.1. The flaps would deflect due to the pressure difference
created by the pressure rise across the shock and by piezoelectrically induced strains.
The amount of deflection, and hence mass flow through the plenum chamber, would

1-3
control the interaction. It was originally proposed that the flaps would be capable of
being closed to the flow, providing zero mass transfer/control and a smooth surface
to the flow, unlike in the case of passive control. Additionally, the existence of a matrix
of flaps on a transonic aerofoil, which could open or shut independently, would enable
localised control of the interaction as the swept or unswept shock moves over the
aerofoil at different flight conditions, whilst producing a smooth surface away from
the shock position. It should be pointed out that no feedback system is considered in
this thesis and the term active control refers to controlling the overall flap deflection,
that is the mass transfer and hence degree of separation. It is proposed that the flaps
will delay separation of the boundary layer whilst reducing wave drag and negate the
disadvantages of previous control methods.

using piezoelectric actuators

Piezoelectric material was selected for flap construction due to its bipolar nature, which
is the ability to produce positive or negative strain. If the shock can move over a matrix
of flaps then each flap would need to overcome the aeroelastic deflection, due to the
presence of the shock, regardless of direction. Also, the piezoelectric materials bipolar
characteristic enables it to assist the aeroelastic deflection to promote mass transfer.

The plenum chamber pressure, to a first approximation, would be at a mean pressure of
the upstream and downstream pressure, as indicated by Hafenrichter et al. (2003).
Initially, the piezoelectric flaps are modelled with a constant pressure load to predict tip
deflection. The pressure difference acting across the flap will be in excess of 15kPa, the

1-4
difference between the chamber pressure and the free stream pressure above the flaps.
Improved theoretical studies, to be described in chapter 3, will account for the fact that
there will be a pressure gradient through the plenum chamber and above the flaps.

Experiments initially examined the more complicated swept normal shock (SNS)
interaction. This was primarily due to the fact that provisions to study an unswept
normal shock (UNS) were not available when the project started. The experiments were
designed to study the interaction of a swept shock, generated by an 11 wedge on the
test section floor, with the naturally grown side-wall boundary layer. With a Mach 2
freestream flow a shock was generated with a normal Mach number of 1.3, when the
uncontrolled interaction should be incipiently separated according to existing literature.
It appears that the main requisite for interaction control is to delay or inhibit separation
whilst smearing the shock foot to minimise wave drag. Therefore, the incipiently
separated interaction was initially studied extending to the separated interaction.

The study of the SNS interaction is more complicated than the UNS interaction, as
pointed out by Babinsky (2002). The complications arise from the three-dimensionality
of the SNS interaction compared to the two-dimensional UNS interaction. The
opportunity to examine the UNS interaction at the University of Cambridge arose after
the initial SNS interaction study was completed. The Cambridge facility has a proven
UNS experimental arrangement that allowed improved data capture and interaction
visualisations, with the ability to easily record boundary layer traverses and also to
obtain schlieren photographs. The Australian Research Council (ARC), through a
Linkage grant, and the School of Aerospace, Civil and Mechanical Engineering at
UNSW@ADFA jointly supported this part of the study.

The impetus to control the shock wave boundary layer interaction phenomenon comes
from performance, economical and environmental considerations. Interaction control
on the aerofoil of a typical transonic aircraft would help reduce buffet due to the shock
instability, thus alleviate structural fatigue. Moreover, there would also be an
improvement in the lift to drag ratio. This would in turn help reduce the GTOW,
leading to reduced fuel consumption. Some studies have suggested that a 1% saving in

1-5
overall aircraft drag, with interaction control, could produce as much as a 5% saving in
total fuel consumption on a typical 8hr transcontinental flight, see Bushnell (2004).

1.2. Literature Survey

The Shock Wave/Boundary Layer Interaction (SBLI) occurs on transonic aerofoils at
local Mach numbers, between 1.1 and 1.5. The interaction may have an unswept or a
swept shock, depending on the location of the interaction on the aerofoil, and is usually
with a turbulent boundary layer, due to the high Reynolds numbers of full scale flight.

In this section, the simple UNS interaction is initially described with the type of
interaction, divided into three distinct types depending on the degree of separation; the
unseparated, incipiently separated and the fully separated interaction. However, the
three types of interaction occur at different Mach number ranges. It should be noted
that the progression from unseparated to separated interaction is a continuous process.

The SNS interaction, and its similarities with the UNS interaction, is then discussed.
The SNS interaction, although more complicated, is more widely observed on transonic
transport aircraft, for example on swept wings, at the wing-fuselage junction and the
fin-tail plane interfaces.

Previous methods of active and passive control of the interaction will be discussed,
looking mainly at mass transfer systems. It should be noted that other methods of
control have been identified, for example wall temperature control and surface contour
control, to name a few. The active methods identified include mass injection and
suction. The passive control methods include porous layers over a plenum chamber,
slots, hybrid systems and mesoflaps, which all use a combination of mass
injection/suction. However, the hybrid systems and mesoflaps use a combination of
passive and active control. To reiterate, the interaction is always considered to occur
with turbulent boundary layers.


1-6
1.2.1. The UNS Interaction
For an inviscid flow
*
a shock would meet, be generated or be reflected by a solid
surface generating a discontinuous pressure rise. In practice a boundary layer will
always be present and for a weak shock, say M=1.1, the shock would continue into the
boundary, see fig. 1.2.
Figure 1.2 - Weak UNS Interaction, Green (1969)

Green (1969) noted that for very weak shock waves with an upstream Mach number of
around 1.1 the degree of interaction is relatively weak. The normal shock wave extends
into the turbulent boundary layer virtually as far as the sonic line with little smearing.
The streamlines diverge upstream of the interaction creating compression waves that
coalesce to form the shock, see fig. 1.2. The streamwise extent of the interaction is
typically two or three times the thickness of the boundary layer, see fig 1.3. Provided
that no separation occurs, the interaction does not have an important effect on the global
flow field. However, the occurrence of separation causes divergence and uncertainty in
the expected external flow from what would be predicted with an inviscid flow.

The boundary layer, and its viscous effects, would cause the shock foot to smear due to
the pressure rise across the freestream shock. The high downstream pressure is
transmitted into the boundary layer, where the pressure is continuous everywhere below
the sonic line, and conveyed in both upstream and downstream directions. Of particular
interest is the upstream effect, which causes the boundary layer to thicken. The

*
Shocks are a viscous phenomenon in aerodynamics and the term inviscid refers to a flow without a
boundary layer.

1-7
boundary layer thickening generates a small degree of smearing of the shock foot as the
shock is subjected to reduced Mach numbers as it approaches the surface. In general the
boundary layer, displacement and momentum thicknesses grow throughout the
interaction and no separation is observed. As the shock becomes stronger, the higher
pressure rise leads to greater upstream and downstream influence. This produces
greater boundary layer thickening and results in a greater smearing of the shock foot.
The smearing is the product of compression waves generated by the boundary layer
thickening, creating a localised compression corner.

Figure 1.3 Holographic Interferometry of the uncontrolled
UNS Interaction at M=1.1, Gibson et al. (2000)

A consequence of this viscous interaction phenomenon is a smoothing of the surface
pressure distribution in the shock region where a steady rise replaces the discontinuity,
as can be seen in the smoothing effect of the compression waves in figure 1.3. Green
(1969) simplifies the UNS interaction by only considering the variables of initial
boundary layer conditions approaching the interaction and the pressure rise, which is
directly related to the Mach number. However, he does clarify that the overall pressure
rise is influenced both by boundary conditions in the far field and the behaviour in the
interaction region. For a fully turbulent boundary layer in transonic and supersonic
flows at high Reynolds numbers, the lower the Reynolds number the wider the
spreading of the pressure distribution. Green notes that provided the separated region is
large enough, the flow up to, and possibly downstream of the separation point, depends
solely on the properties of the initial boundary layer. The larger the separated region,
the further downstream of the separation point does the interaction remain independent

1-8
of the downstream conditions. This is a simplification of the free interaction theory by
Chapman et al. (1958) the interactions that are free from direct influences (though
not from indirect influences of downstream geometry) and are free from complicating
influences of the mode of inducing separation, arbitrarily [are] termed free
interactions for brevity.

An increase in Mach number would increase the pressure rise across the shock
promoting separation. It appears that once separation occurs the interaction cannot be
labelled as a free interaction. This is confirmed with the experiments of Chapman et
al. (1958) who studied the Reynolds number effect on the turbulent interaction with
separation, where the model shape and Mach number had to be fixed. It would appear
that if separation can be minimised or removed then the UNS interaction could be
considered quasi-entirely a free interaction, ignoring the downstream subsonic flow
effects. It should be noted that Chapman et al. (1958) did not look at the UNS
interaction with a turbulent boundary layer and that these conclusions are extrapolated
from the study of rear and forward facing steps and compression corners.

1.2.1.1. - INCIPIENT SEPARATION
a) b)
Fig 1.4 a) Wall Shear stress distribution through a SBLI from Green (1969) and
b) The separation bubble growth from Delery (1985)
with increasing Mach number to demonstrate incipient separation

As the shock strengthens, the pressure rise exerted on the flow field increases producing
larger compression waves and greater smearing. Furthermore, the boundary layer is
more susceptible to separation and a limit is reached where incipient separation occurs
at the shock foot. Incipient separation is defined as the threshold when the wall shear

1-9
stress,
w
, at one point equals zero but everywhere else is positive, see fig 1.4a. For a
greater pressure rise
w
will locally go negative, which can produce separation and
reattachment. Due to the complexities of accurately measuring
w
in a transonic
interaction it is easier to calculate when the separation bubble becomes vanishingly
small, see fig 1.4b. That is when the separation (S) and the reattachment point (R) tend
towards one another as indicated by flow visualisation.

According to Pearcey (1955) the growth of the separation bubble from the shock foot
does not depend much on the Reynolds number of the boundary layer provided that it is
fully turbulent. Incipient separation is shown to be weakly dependent on the Reynolds
number with turbulent boundary layers, as shown by Chapman et al. (1958). The
degree of separation and shock smearing would be more dependent on the pressure rise
and shape factor.

Delery and Marvin (1986) note that the influence of Reynolds number is commonly
included in the shape parameter but they are not uniquely linked, for example with a
pressure gradient or transpiration. If the influence of shape factor and Reynolds number
are assumed negligible the criterion at which separation first occurs, for an UNS
interaction, is generally accepted to be when the upstream Mach number is 1.3. This
agrees with Inger (1981), who concluded that unseparated SBLI flows occur up to a
Mach number of approximately 1.3, after which incipient separation takes place. This
value of normal Mach number for the UNS interaction has now been generally accepted
as the limit of incipient separation.

1.2.1.2. - SEPARATION
Incipient separation does not greatly influence the turbulent boundary layer interaction
or the global flow field. However, the rapid onset of separation can have a disastrous
effect. Delery and Marvin (1986) note, such a quasi-explosive increase of the size of
the dissipative region considerably affects the whole flow field and generally leads to a
catastrophic loss in terms of performance. On a transonic aerofoil this would equate to
a loss in lift, rise in drag, different section moment and the on-set of buffet.


1-10
A sizeable separation bubble forms at the shock foot when the upstream Mach number
becomes noticeably greater than 1.3. The bubble is sensitive to external factors and its
streamwise extent can increase dramatically as a consequence of further rises in Mach
number or the action of downstream adverse pressure gradients, such as the one existing
on a highly rear loaded aerofoil. An interaction strong enough to cause separation is
characterised by the existence in the outer flow of a lambda shaped foot of the shock
wave as well as a rapid growth in the separated boundary layer. Referring to figure 1.5,
the separation bubble enhances upstream boundary layer thickening with increased
compression waves, which coalesce to form the oblique leading leg of the lambda
structure. The difference in strength between the oblique shock and the main shock
creates two different flow regions with different pressures and velocities. In order to
make the two adjacent flows compatible the trailing leg of the lambda structure is
generated. The strength of both legs of the lambda structure decrease as they approach
the wall. The weakening of the trailing leg is partly due to the varying upstream
conditions and partly due to the effect of the compression waves generated by the
growth of the displacement boundary layer effects. The formation of the oblique
leading leg is dependent on the boundary layer growth, which is influenced by the
incoming Mach number, the boundary layer thickness and the shape factor.

Figure 1.5 - Lambda structure of the separated UNS interaction, Green (1969)

The virtually normal trailing leg and the oblique leading leg meet the main shock at the
triple point, also known as the bifurcation point. An increased upstream Mach number
leads to greater separation and upstream influence. It is believed that this will cause the

1-11
oblique shock angle to marginally decrease, however, the greater upstream interaction
length leads to an increased triple point height with upstream Mach number.

According to Atkin and Squire (1992), the height of the triple point is approximately 3.5
to 4 boundary layer thicknesses above the surface and the main growth in boundary
layer thickness takes place between the leading and trailing shocks, with the momentum
thickness increasing steadily during the whole interaction process. Their work, with
normal shock waves with turbulent boundary layers, at Mach numbers from 1.3 to 1.55,
showed that, as the upstream Mach number increases the distortion of the shock foot,
the boundary layer thickening becomes increasingly significant. About M = 1.3,
incipient separation takes place, inducing the shock wave foot to bifurcate, with an
oblique leading shock followed by a rear shock, both stemming from the triple point a
few boundary layer thicknesses above the wall. The incoming boundary layer, unable
to overcome the pressure rise imposed by the interaction, separates to form a free shear
layer that subsequently attaches further aft of the rear shock with increased pressure
rise. Atkin and Squire (1992) state that after separation occurs the triple point height is
not greatly altered for increased Mach number without the presence of control. This is
contrary to Gibson et al. (2000), who note that as the Mach number increases after
separation the leading leg becomes more oblique and the triple point lifted away from
the surface. However, Gibson et al. agree with Atkin and Squire that the increase in
length of the separation bubble is also significant.

Referring to figure 1.6, a slip line or vortex sheet is the consequence of the discontinuity
of downstream velocities, greater in section 2 than in section 3. It arises because the
Mach number downstream of the leading and trailing leg of the lambda structure is
greater than the Mach number downstream of the main shock, as the total rise in
entropy is always less through successive shocks than the rise through a single shock for
the same final static pressure.

The difference in shock strength above and below the triple point imparts different
velocities onto the flow. Below a Mach number of 1.4 the flow everywhere is subsonic
downstream of the interaction. However, for Mach numbers greater than 1.4 supersonic
flow is formed downstream of the trailing leg creating a supersonic tongue. The

1-12
supersonic tongue, which grows with increased upstream Mach number, is subsequently
isentropically retarded
due to the downstream subsonic influence, see fig. 1.7. Seddon

(1960) notes in his experiments at a Mach number of 1.47 that the presence of the
supersonic tongue in a predominantly subsonic system, between the vortex sheet and the
separated region, renders the problem very difficult to analyse.

Figure 1.6 - Detailed lambda structure, Atkin and Squire (1992)
Figure 1.7 Supersonic Tongue, Seddon (1960)

without the presence of shocks.


1-13
Seddon goes on to note that the first appearance of separation, a region of reversed flow,
marks an important qualitative change in the UNS interaction. Compared with fully
attached flow, the presence of separation is usually associated with greater losses,
greater uncertainties and, in some cases, a significant difference in the character of the
external flow.

Mateer et al. (1976) worked on UNS interactions at a Mach number of 1.5 and a
Reynolds number range from 10
7
to 8x10
7
. They showed that the length of separation
was proportional to the incoming boundary layer thickness and a supersonic tongue
appears downstream of the shock when the free stream Mach number increases above
1.4, which was also observed by Seddon (1960). The supersonic tongue extends to at
least a height of four upstream boundary layer thicknesses, which is consistent with the
inference of Atkin and Squire (1992). The pressure distribution has an initial steep
pressure rise, followed by an abrupt change in slope and a gradual increase in pressure
to a value that is below the inviscid value. The slope of the initial rise was shown to
increase with Reynolds number, indicating a dependence on Reynolds number above
10
6
. The separation occurs near the onset of the pressure rise which is located
increasingly upstream with increase in Mach number. However, the reattachment
occurred at the same position relative to the shock independent of the Reynolds number.

1.2.2. The SNS Interaction
Figure 1.8 Three dimensional Sharp Fin Induced SBLI
With an unseparated flow structure, Kubota and Stollery (1982)

1-14
Figure 1.8 shows the flow structure proposed by Kubota and Stollery (1982) for an
unseparated SNS Interaction. Analysis by McCabe (1966) shows that incipient
separation occurs at a normal Mach number of 1.2. This supports the conclusions of
Stollery (1989) and Settles and Dolling (1992) that separation occurs more readily in
SNS interactions than in UNS interactions due to the reduction in normal Mach number
with sweep back. As a result most SNS interaction experiments, other than those at low
freestream Mach number (M
) with weak shocks, show evidence of separation.

The pressure downstream of the shock in many UNS interactions quite quickly reaches
the value predicted by inviscid theory. However, in SNS interaction the pressure at the
wall downstream of the shock appears to be strongly influenced by the viscous effects
present in the boundary layer. The total effect of the adverse pressure gradient, the
boundary layer cross flow and the lateral distribution of disturbances produce a large
interaction region downstream of the shock. The size of the interaction region, which
increases with the normal Mach number to the main shock (M
n
), generates an apparent
decreased pressure rise.

West and Korkegi (1972) show, using intersecting wedges, that a strong resemblance
between UNS and SNS interactions occur with high freestream Mach numbers greater
than 3 (M
n
1.36), see figs. 1.3, 1.4 and 1.9. They conclude by saying that a three
dimensional SNS interaction may be viewed locally as [a] two-dimensional [SNS]
interaction with strong cross-flow. Squire (1996) worked on SNS and UNS
interactions and he suggests that correlations between the interactions can be made,
provided this is done in terms of M
n
and pressure rises. This chapter will continue by
presenting interactions and their control in terms of M
n
for comparisons between the
studies.

Settles and Dolling (1992) formulated scaling laws for both SNS and UNS interactions.
The interaction scaling depends on the upstream influence length, local boundary layer
thickness, Reynolds number based on boundary layer thickness and local Mach number
normal to the inviscid shock. This suggests a correlation with the free interaction theory
based on the normal Mach number. Korkegi (1971) indicates that free interaction

1-15
occurs for a turbulent SNS interaction with separation, up to the pressure plateau
(lambda structure or separation point).

Fig 1.9 Surface flow beneath swept normal shocks and the flow viewed parallel to the
shocks for a) unseparated and b) separated flow from Green (1969)

Koide et al. (1996) showed that there is a correlation in SNS interactions produced by
several geometrically dissimilar shock generators, further cementing the concept that
free interactions occur for both SNS and UNS interactions. The primary separation
lines and their angles, which dictate the upstream influence length, were observed to be
proportional to the normal Mach number for the different shock generators.


1-16
One of the first criteria for incipient separation for SNS interactions was proposed by
McCabe (1966) for freestream Mach numbers, M
, above 1.6 with the very simple

approximation

364 . 0
M [Eq. 1.1]

where N is the external flow deflection angle in radians. Korkegi (1973) found that a
better agreement was obtained with 0.3 or P
2
/P
1
=1.5, which equates to M
n
=1.2. This
holds true for high values of Reynolds number and it is assumed that if the Reynolds
number is sufficiently high its influence on incipient separation is negligible.
According to Kubota and Stollery (1982) the incipient separation definitions of McCabe
and Korkegi are conservative since there are experiments which show that the surface
flow can be deflected through angles exceeding the shock wave angle before the
formation of a true separation line.

Figure 1.10 Three dimensional Sharp Fin Induced SBLI with
a separated flow structure, Kubota and Stollery (1982)

The onset of incipient separation can be observed from oil flow visualisation when,
Green (1969) offers, the upstream skin friction lines converge and merge asymptotically
into a separation line (S), or what Kubota and Stollery call a convergence line, see fig.
1.10. Delery and Marvin (1986) state that the separation line runs approximately
parallel to the swept shock. When the downstream streamlines run parallel to the

1-17
separation line
w
normal to the sweep of the shock should equal zero. When the
boundary layer separates the surface flow pattern also reveals the existence of a
reattachment line (R) close to the trace of the shock generator. Kubota and Stollery
(1982) represent the interacting flow with a corner vortex and the inner part of the shock
generator boundary layer is pushed under the sidewall boundary, see fig. 1.10.

Work on SNS interactions by Alvi and Settles (1992) using a sharp fin, with a M
n
range
from 1.4 to 2.45, showed that further increases in the Mach number upstream of the
shock wave continue to affect the details of the shape of the lambda structure. In
particular, there is a tendency for the upstream leg of the lambda structure to become
increasingly oblique with a larger primary separation bubble and an increase in the
triple point height. An increased Mach number corresponds to an increased upstream
influence length as suggested by free interaction theory. Furthermore, above M
n
of 1.64
a secondary separation occurs within the primary separation bubble, seen as a bulge
in the initial reversed flow, as reported by Settles and Dolling (1992), see fig. 1.11. It
appears that the secondary separation is created from the increased primary separation
bubble. However, it does not further complicate the overall interaction, which is more
dependent on the overall size of the separation bubble.

Figure 1.11-SNS Interaction showing secondary separation
from Settles and Dolling (1992)

1-18
1.2.3. - SBLI Control
Korkegi (1976) concluded that the structure of the three dimensional [SNS] interaction
is . primarily dependent on the extent of separation. This holds true for the UNS
interaction as well and it appears that the objective of controlling the interaction is to
extend the limiting boundaries by minimising the amount of separation. This may be
accomplished by: active means, such as tangentially blowing (injection), boundary layer
suction and surface heating or cooling; mechanical means, for example vortex
generators; or by passive means through placing a permeable surface with a cavity
beneath in the shock region. For passive control, the large pressure difference over the
shock will lead to a combined suction and injection effect downstream and upstream of
the shock respectively, see section 1.2.4. Furthermore, hybrid control can be employed
using a combination of the above.

One of the main objectives when designing a wing for transonic speeds is to obtain as
high a drag rise Mach number as possible, subject to certain constraints. In principle,
supercritical aerofoils are shaped to delay the drag rise associated with the energy losses
caused by shock waves and flow separation. At design conditions they have an
extended region of supersonic flow over the upper surface, which is terminated by a
nearly isentropic recompression, thus minimising wave drag. However, they have
limited Mach number range and incidence before off-design conditions are met, for
example a shock is formed on the aerofoil, which drastically alters the aerofoil
characteristics. This can lead to separated flow, increased drag, reduced lift, a radically
different moment and the existence of buffet due to the unsteadiness of the shock.
Therefore, at off-design conditions the control of the SBLI is virtually essential.

1.2.3.1. - MASS TRANSFER (INJECTION OR SUCTION)
Fluid injection in the supersonic region upstream of a shock wave increases the growth
rate of the boundary layer approaching the shock, decreasing skin friction, such that the
effective geometry is changed to produce several weaker shocks, which replace the
original shock. This softens the shock and leads to a reduction in wave drag, which is
the primary desired effect in the case of injection. Pearcey (1961) showed on a
transonic aerofoil with an UNS, at M
=0.9 and a 6Q angle of attack, that slot blowing at


1-19
15% chord minimises separation and greatly improves the pressure distribution,
reducing drag and increasing lift, see fig. 1.12.

Figure 1.12 Shadowgraph and pressure distributions over an aerofoil at 6 with
a) no boundary layer control and b) blowing through an upstream slot, Pearcey (1961)

It should be noted that excessive thickening of the boundary layer approaching the
shock reduces the friction through the interaction and may lead to boundary layer
separation downstream. Delery and Bur (1999) note that low injection velocities result
in a lengthening of the separation bubble. However, a reversal of this effect is observed
after the injected mass flow velocity is increased beyond a certain limit as the
momentum within the boundary layer is increased, resisting separation.

Suction behind the shock considerably reduces the boundary layer and displacement
thicknesses, affecting the whole downstream flow development. Mass suction changes
the shape factor towards unity, which increases the frictional effect and makes the
boundary layer less susceptible to separation. Without suction on an aerofoil, a local
separation bubble, which would expand rapidly, can exist downstream of the shock.
Thiede et al. (1984) showed in their work on aerofoils with double slot suction that
maximum lift, and accordingly the lift coefficient for the onset of buffet, increased
considerably with the application of suction, see fig 1.13. At a pre-shock Mach number

1-20
of 1.44 the flow was still attached and only local separation was occasionally observed.
It was shown that suction delayed shock induced separation and reduced the extent of
the separation bubble. However, it is unsure whether the results are a pure suction
effect or whether there is a passive circulation effect occurring due to the fact that the
two slots are not isolated from one another.
Figure 1.13 Active control on a supercritical transonic aerofoil
by suction through a double slot at a) 4Q and b) 5Q- Thiede et al. (1984)

Krogmann et al. (1985) observed that suction through a perforated strip is less effective,
in the reduction of separation and hence production of lift, at lower Mach numbers as
compared with single and double slot suction
. They also observed that with suction the

shock is located further downstream of the uncontrolled position. Up to the suction
position the flow is further accelerated, increasing the shock strength, which along with
a thinner velocity profile produces additional wave and friction drags. Moreover,
suction reduced pressure drag due to separation and prevented the occurrence of buffet,
even at higher Mach numbers.

with a double slot configuration more effective than a single slot one.

1-21
Green (1969) noted that suction is not an economical way of controlling separation on
transonic aerofoils. However, it is very effective for suppressing separation in
supersonic air intakes, diffusers and is virtually essential above a Mach number of 2 if
high performance is essential.

1.2.4. - Passive Control of Shock Wave/Boundary Layer Interaction (PCSBLI)
Figure 1.14 - Passive Control of Shock/Boundary Layer Interaction

According to Nagamatsu et al. (1985), Mr. Dennis Bushnell and Dr. Richard Whitcomb
of the NASA Langley Research Centre originally suggested PCSBLI in 1979. The
concept consists of a porous surface and a cavity, known as a plenum chamber, located
underneath the shock wave Boundary layer interaction (SBLI), see fig. 1.14.

The high pressure downstream of the shock wave will force some of the boundary layer
fluid into the plenum chamber and out ahead of the shock. This would be equivalent to
a combination of suction downstream and blowing upstream of the shock. The plenum
chamber pressure would, to a first approximation, be an average pressure of the
upstream and downstream pressure.

The plenum chamber would increase the communication upstream of the shock,
which does not happen across an inviscid shock, to create rapid thickening of the
boundary layer approaching the shock. The boundary layer thickening produces an
oblique leading leg shock followed by a nearly normal trailing shock, known as a
lambda structure. The presence of the lambda structure reduces the wave drag, as the
entropy rise is less through successive shocks than through one shock. The combination

1-22
of a weaker trailing leg and suction would reduce the susceptibility of the flow to
separate. Furthermore, the porous surface and the plenum chamber can also dampen
pressure fluctuations and the unsteadiness associated with the SBLI
. It would appear
that passive control produces a lambda structure that is similar to the uncontrolled
interaction with separation, having the advantage of reduced wave drag without its
disadvantages, providing the control is effective in constraining the separation.

1.2.4.1. - THE EFFECT OF PCSBLI ON THE FLOW FIELD
Bahi et al. (1983), Nagamatsu et al. (1985), Nagamatsu et al. (1987), Raghunathan
(1987)(1988), Raghunathan et al. (1987), Raghunathan and Mabey (1987), Savu and
Trifu (1984) showed that the general effect of PCSBLI was to reduce the pressure
gradients and local Mach numbers in the interaction region. The pressure distribution
upstream of the porous region is virtually unaffected. On an aerofoil, the reduced
pressure gradients in the interaction region indicate a reduction in shock strength and,
therefore, lower pressure on the downstream suction surface that will increase lift on the
upper surface. The only sharp changes in pressure gradients are those at the beginning
and end of the porous region, reducing the SBLI effect on the section moment. Savu
and Trifu (1984) suggest that the pressure distribution obtained with PCSBLI can be
used to generate an equivalent solid aerofoil, as PCSBLI on an aerofoil is analogous to
changing the surface geometry. Analysis performed on the equivalent solid aerofoil
showed a similar pressure distribution as the original porous aerofoil. This seems to be
the forerunner to current bump control technology that imitates the separation bubble to
produce the shape of the bump, see Holden (2004).

The application of passive control splits a single shock into a series of weaker shocks
forming a lambda () shock foot. The leading leg of the shock foot is oblique and the
trailing leg of the shock is quasi-normal. Gibson et al. (2000) observed that the
bifurcation point occurs approximately 3-5 boundary thicknesses above the surface.
With increase in freestream Mach number, the leading leg remained anchored at the
start of the porous surface, except at high Mach numbers, and the trailing leg generally
follows the same path as that of the shock on a solid surface. They went on to suggest

The author recognises that the SBLI is steady in the mean but uses the term unsteadiness with
reference to the minor spatial variations of the local SBLI.

1-23
that passive control can be thought to fix the position of the shock in a duct. It appears
that the presence of control increases downstream total pressure, forcing the shock
upstream. The upstream movement would then reduce the effectiveness of the control,
to be discussed in section 1.2.4.2., decreasing downstream total pressure, and therefore
creating a stable shock position. Gibson et al. (2000) observed that PCSBLI has the
effect of stabilising the shock wave, which is of particular relevance in controlling
transonic buffet. A forward movement in shock position led to greater suction, reducing
the effective back pressure and halting the forward movement, while any downstream
movements were halted because the reservoir pressure was sufficient to hold the shock
on the control surface.

The effect of blowing only upstream and suction only downstream of the shock wave on
the pressure distribution, using passive control, shows that blowing reduces the pressure
gradients and skin friction. However, the suction downstream does not increase the
pressure gradients and skin friction as expected; indicating the effects of passive control
are blowing dominated. The dominating influence of blowing in PCSBLI is also
indicated in boundary layer displacement thickness measurements with and without
passive control. Raghunathan (1987) suggests that the reason for this blowing
dominated behaviour is that the development of the boundary layer in the subsonic
region, downstream of the shock, is a function of the upstream influence. With passive
control the boundary layer approaching the shock is thicker and when subjected to a
SBLI, even with a softened shock system, takes a longer distance to rehabilitate itself
and is also less resistant to separation.

One of the major effects of PCSBLI on an aerofoil is on the drag. Delery and Bur
(1999) and Nagamatsu et al. (1985) showed that passive control at low transonic Mach
numbers, M
T
=0.75, increases the overall drag whereas at higher transonic freestream
Mach numbers, above M
T
0.83 (M
n
1.35), it reduces the drag by 33%. Bahi et al.
(1983) stated that significant drag reductions could be achieved at a freestream Mach
number of 0.806 (M
n
1.26), agreeing with Nagamatsu et al. (1987) who suggest that, at
M
greater than 0.8 (M

n
>1.25), the increase in viscous drag is less than the reduction in
wave drag, see fig. 1.15. Raghunathan (1987) showed slight overall drag reduction at
M
= 0.85 (M
n
1.3) depending on the type and amount of porosity. He goes on to show

1-24
that greater levels of overall drag reductions are achieved at M
n
= 1.37. Krogmann et
al. (1985) showed that drag reduction can be obtained over a wide range of Mach
number and incidence, although it is not clear whether viscous drag was included. It
would appear that PCSBLI is beneficial for an UNS interaction above a local Mach
number of 1.3, the point of incipient separation.

Figure 1.15 Drag coefficient variation with freestream Mach number for solid and
porous surface aerofoils from Nagamatsu et al. (1987)

Passive control can delay aerofoil stall and improve the lift to drag ratio. Krogmann et
al. (1985) obtained 20% - 50% increases in the lift to drag ratio and greatly reduced
buffet with passive control, with freestream Mach numbers up to 0.86
**
. They also
showed that PCSBLI can reduce unsteady pressure fluctuation in the interaction region
and substantially raise the buffet boundary for a wing. It is also suggested that the low
velocity plenum chamber, which allows communication across the shock, acts as a
stabiliser for any shock movement.

The drag reduction due to PCSBLI is essentially due to the weakening of the shock,
which leads to a reduction in the entropy rise. However, at off-design conditions
without the presence of a shock to drive the flow there is an increase in drag due to
viscous losses from the unsealed rough holes, see Gibson et al. (2000).

**
The Mach number limit of their study.

1-25
1.2.4.2. - FACTORS AFFECTING PCSBLI
The extent of drag reduction with PCSBLI is dependent upon numerous variables.
Firstly, PCSBLI is more effective at higher Mach numbers (shock strength). Within the
boundary layer the losses are essentially viscous losses, whereas outside the boundary
layer the losses are due to entropy changes across the shock. In order to obtain a net
reduction in drag, the reduction in entropy change has to outweigh the increase in
viscous losses. Comparisons of profiles, both with and without passive control, showed
that the rough surface of passive control increased the viscous losses and the lambda
structure decreases losses due to the increase in entropy across the shock. This appears
to only be possible at sufficiently high Mach numbers, as discussed in section 1.2.4.1.
Raghunathan (1988) shows that appreciable drag reduction is achieved at greater Mach
numbers than those where incipient separation would occur without control.
Furthermore, he appears to suggest that optimum control is achieved when the
separation is just constrained to the control region, agreeing with Nagamatsu et al.
(1985) who suggest that the boundary layer effects are more important than the total
pressure losses. Savu and Trifu (1984) go on to show that for supercritical airfoils at
sub-critical Mach numbers the influence of porosity distribution is negligible. This is
presumably due to the fact that the incipient separation criteria for the uncontrolled
aerofoil have not been met.

Raghunathan (1987) showed that the general effect of increasing porosity is an increase
in the viscous losses near the surface but with a reduction of losses across the shock
system. Whilst an increase in porosity should increase the mass injection upstream of
the shock, it also reduces the pressure gradients in the interaction region, and hence an
increase in skin friction. The optimum porosity for maximum drag reduction, according
to Raghunathan (1988), is in the range: 2 to 3% for normal holes; 1 to 2% for forward
facing holes
; and 1% for slots, where the porosity is situated over a very narrow
region and the passive control mechanism is different. He also showed that the wave
drag generally reduces with an increase in porosity due to a relatively weaker trailing
leg of the lambda shock.

where the holes are inclined 60 forward. It is expected that these promote suction, mass transfer and
upstream boundary layer thickening.

1-26
Gibson et al. (2000) observed that the leading leg of the lambda structure becomes more
oblique with an increase in porosity. This indicates that the strength of the leading leg
increases with porosity due to the amount of mass injection at the start of the control
region, see table 1.1. They went on to say that the main features of the controlled SBLI
vary little with degree of porosity, other than in the greater smearing that occurs for
higher porosities.

Porosity 2% 4% 8%
Oblique shock angle, 52.7 53.5 54.4
Table 1.1 - Varying porosity effect on leading leg angles for a central shock position on
a passive control plate (100mm in length) - Gibson et al. (2000)

Bahi et al. (1983) observed in their work on transonic aerofoils that higher porosity
leads to lower and higher local Mach numbers ahead of and downstream of the
shockwave respectively. This implies weaker leading and terminating shocks to reduce
flow separation, leading to improved total pressure recovery with reduced wave drag.
However, this was only observed after the oblique shock of the lambda structure was
formed. With a porosity of 1.25% the leading leg consisted of compression waves
instead of the oblique shock generated with 2.5% porosity. The existence of
compression waves would imply a weaker shock foot but it is unclear whether there is a
lambda structure.

Shock Position Upstream Central Downstream
Shock location wrt the control plate 33% 50% 66%
Oblique shock angle, 55.3 53.5 50.7
Table 1.2 - Oblique Shock Angles for various shock positions
on a 4%porosity passive control plate (100mm in length) - Gibson et al. (2000)

Gibson et al. (2000) observed the effect of shock location relative to the control region.
A shock location upstream of the central porosity position produces a stronger leading
leg of the lambda structure, see table 1.2. As the passive control action becomes limited
by the small area of the porous plate, through which the higher intensity blowing has to
take place, which in turn increases the strength of the leading shock wave. As the main

1-27
shock approaches the upstream solid surface, the size of the lambda structure decreases
until it adopts the uncontrolled form.

Nagamatsu et al. (1985) indicates that a downstream shock position produces improved
drag production. It is generally accepted that the optimum distribution for uniform
porosity should be 2/3 upstream and 1/3 downstream of the shock position. The
downstream shock position would have a larger lambda structure, as the decreased
leading shock angle, of say 5%, would be overwhelmed by the approximate 33%
increase in upstream interaction length, compared to the mid-position. However, any
further movement downstream would restrict the suction, and therefore total mass
transfer.

Raghunathan and Mabey (1987) observed the controlled interaction at varying shock
positions with respect to the control plate, by increasing the freestream Mach number to
move the shock. However, it is hard to separate the control position effects from the
Mach number effects. Gibson et al. (2000) controlled the downstream total pressure, at
similar Mach numbers, to control the shock position and isolate the effect of shock
position. It would appear that comparison of SBLI control technique efficiencies would
be limited unless similar shock positions and Mach numbers were used.

Chen et al. (1984), Savu and Trifu (1984) and Gibson et al. (2000) have shown that the
type of porosity distribution influences passive control. Chen et al. (1984) conducted
experiments using three types of porosity distribution: uniform distribution, maximum
porosity at the shock position and maximum porosity at the mid section of the porous
region, labelled type A, B and C respectively. Type A and B results indicated that for a
given porosity, drag reduction increases monotonically with Mach number.
Additionally, for a given Mach number, the drag reduction increases with porosity. For
type C distribution, drag reduction increases with Mach number up to a certain Mach
number, above which a smaller reduction in drag was observed for an increase in Mach
number. Presumably, this is due to the shock located downstream of the passive control
region with increased Mach number.


1-28
The type of porous surface is also an influencing factor. The surface can be made with
normal holes, forward facing holes, backward facing holes and slots perpendicular to
the flow, to name a few. Raghunathan and Mabey (1987) showed, with M
n
=1.37 and
1.6% porosity control, that there is a considerable difference in pressure distribution,
shock position, stagnation pressure profiles and drag when the hole inclination is
changed. The forward facing, rearward racing and normal holes produced drag
reductions of approximately 28%, 12% and 2% respectively compared to without
control. It is believed that the forward facing holes have improved mass bleed and
upstream influence. The rear facing holes have less mass injection but the tangential
effect reduces the disruption to the boundary layer compared to the normal holes.
However, the improved wave drag reduction with the inclined holes, rear of forward
facing, comes at the expense of larger viscous losses. The normal holes produced larger
changes in the Mach number with some expansion at the leading edge of the porous
region and improved pressure recovery downstream of the control region, near the
trailing edge.

Bahi et al. (1983) showed that an increase in plenum depth reduced the pressure
gradients in the interaction region, decreased the oblique shock angle (presumably from
lower injection velocity) and increased pressures on the downstream suction surface. A
more oblique leading shock produced by a shallower plenum would reduce wave drag
due to the increased triple point height. It would appear that optimum porosity and drag
reduction is a trade-off between triple point height and oblique shock angle.
Raghunathan (1988), however, suggest that the overall drag reduction is not greatly
affected by the plenum depth with the air velocity in the plenum, being approximately
5% of the free stream air velocity.

1.2.4.3. SLOT AND GROOVE CONTROL
Slot and Groove Control of the SBLI, with a local Mach number of 1.3, was studied by
Smith et al. (2002). The presence of streamwise slots, which are more effective than
grooves, lead to a lambda shock structure that relaxes slowly towards an uncontrolled
interaction with increasing spanwise distance away from a slot, (see fig. 1.16). There is
an increase in boundary-layer thickness that is confined to a very narrow region about
the slot, which results in reduced viscous losses for an array of slots compared to the

1-29
same total control area with passive control. Also, it would appear that slot control does
not constrain the separation to the control region and that it continues downstream of
the interaction. However, it is unclear whether the flow would later reattach or radically
alter the external flow field if applied to a transonic aerofoil.
Figure 1.16 - SBLI Structure with Slot Control, Smith et al. (2002)

1.2.4.4. - HYBRID
Figure 1.17 Slot control positions on the side-wall
to study SNS Interactions from Babinsky et al. (1999)

It has been recognised that the boundary layer usually thickens due to passive control,
leading to an increase in viscous drag. It may even cause flow separation in what would
normally be an unseparated flow field. Babinsky (1999) studied SBLI control
comparing a single spanwise slot to a porous surface with an optional suction

1-30
mechanism on both systems to create active control and reduce thickening of the
boundary layer, see fig. 1.17.

It was observed that applying suction to the control cavity reduces the beneficial effects
of shock smearing while also reducing boundary layer thickness, shape factor and
possibility of secondary separation. Large amounts of mass bleed can remove and even
reverse the shock smearing due to control. It was observed that a slot located upstream
of the inviscid shock position has little influence on the interaction; slightly reducing
upstream static pressures. It should be noted that excessive suction could lead to
thinning of the boundary layer inducing an expansion fan in the upstream interaction
region. The higher local Mach numbers will produce a stronger shock with steeper
pressure gradients, negating the benefits of control. However, a slot at or downstream
of the inviscid shock position has a more pronounced influence due to it being situated
in the subsonic or transonic part of the flow field. It would produce a more smeared
lambda structure with reduced static pressure throughout the interaction.

Figure 1.18 - Hybrid SBLI Control with PCSBLI and
downstream suction, Delery and Bur (1999)

Delery and Bur (1999) suggested a range of hybrid control techniques. Firstly, by
placing a suction slot downstream of the control cavity, combining the advantage of
passive control to decrease the wave drag and the effectiveness of suction to reduce
friction losses, see fig. 1.18. They also considered the possibility of reproducing the
separated flow structure by a local deformation of the surface. A double wedge like

1-31
shape, that imitates the viscous separated fluid, would induce a shock at its origin and a
second shock at the trailing edge. This system only slightly affects the boundary layer,
while substantially reducing the wave drag. A bump with a more progressive contour,
that is a continuous curvature, seems to be very effective at producing a nearly
isentropic compression in the shock region. Bump control of the SBLI is a current area
of interest in the aerodynamics community, see Holden (2004).

1.2.5. - Mesoflaps for Aeroelastic Recirculation Transpiration (MART)
MART technology consists of a matrix of small flaps, rigidly fixed at the upstream end
to form numerous cantilever beams, covering an enclosed plenum chamber, as shown in
fig. 1.19. The presence of a shock creates a pressure difference between the free stream
and plenum chamber. The pressure difference causes the flaps to undergo local
aeroelastic deflection to achieve mass bleed or injection, as shown in fig. 1.19b. The
MART system combines the advantages of rearward and forward facing holes,
producing a flush surface when no shock is present. However, the amount of flap
deflection, and hence mass transfer, is pre-determined by the structural design of the
mesoflaps. A cantilever beam deflects under a pressure load proportional to its length
cubed and inversely to the thickness cubed.

a) b)
Figure 1.19 Mesoflaps for Aeroelastic Transpiration a) Four stream wise flap structure,
Wood et al. (1999) and b) SBLI structure, Jaiman et al. (2003)

Hafenrichter et al. (2003) studied normal SBLI control using mesoflaps at a freestream
Mach number of 1.37, with a unit Reynolds number of 30x10
6
m
-1
, oncoming boundary
layer thickness of 2.6mm and a control region length of 57.15mm above a 19.05mm

1-32
deep plenum. The flaps, manufactured from a nickel-titanium shape memory alloy
termed Nitinol, span 50.8mm and have thicknesses in the range of 63.5-228.6m. Their
work concentrates on four or six streamwise flap matrices with an upstream stagnation
pressure of 210kPa. It was observed that the plenum chamber pressure was
approximately constant, due to the subsonic velocities, at an average of the upstream
and downstream pressures. It should be noted that the extremely thin flaps produced
large deflections, occasionally greater than 2mm, which were often accompanied by
fluttering of the flaps that often lead to mechanical failure of the mesoflaps, namely
cracks.

a)
b)
Figure 1.20 The SBLI at M=1.37 a) uncontrolled and with b) MART Control
of the SBLI with a four flap 191 m array, Lee et al. (2002)

It is observed that the uncontrolled SBLI lambda structure is replaced by a larger
lambda structure consisting of a series of longer leading oblique shocks, see fig. 1.20a
and b. The oblique shocks are generated by both mass injection and compression

1-33
ramps created by upward deflecting mesoflaps. The thicker flaps inject flow more
transversely, compared to the tangential thinner flaps, due to the smaller flap
deflections. Also, the first oblique shock appears to be produced due to the imperfect
transition from the tunnel liner to the control plate and is initiated upstream of the
plenum chamber.

Hafenrichter et al. (2003) state that the 63.54m six-flap array and 127.54m four-flap
array [produced] flap deflections [occasionally] greater than 2mm. This
deflection occurs over MART flap lengths, which are approximately 5.2mm and 7.3mm
for the six-flap and four-flap arrays respectively. Assuming, the flaps produced 2mm
deflection then by simple structural scaling, namely the thickness cubed, the
approximate deflection of the other flaps and their compression ramp angles can be
calculated, see table 1.3. Although the deflections are small relative to the 2.6mm
boundary layer thickness they will still produce oblique shocks, especially when the
deflections are in the order of a displacement thickness (Y*=0.36mm) or greater.

Six-Flap Array (Length = 5.2mm) Four-Flap Array (Length = 7.3mm)
Flap
thickness
(m)
Flap
deflection
(mm)
Flap
deflection
(*)
Ramp
angle
()
Flap
thickness
(mm)
Flap
deflection
(mm)
Flap
deflection
(*)
Ramp
angle
()
63.5 2 5.56 21.04 127.5 2 5.56 15.3
78.2 1.071 2.97 11.6 150.6 1.214 3.37 9.4
101.9 0.484 1.34 5.3 190.5 0.600 1.66 5.0
127.5 0.247 0.69 2.72 228.6 0.347 0.96 2.7
150.6 0.150 0.42 1.65
Table 1.3 MART Flap Deflections for the six-flap and the four-flap arrays

A bow shock is formed by the combination of the upstream flap deflecting into the
flow and transverse mass injection across the sides of the flaps, creating a three-
dimensional flow structure. The slots around the flaps, where the Nitinol is removed to
produce the flap, are 0.4mm wide. These presumably will enhance the three
dimensionality of the flow, due to localised slot control, as they are approximately 15%
of the boundary layer thickness and of similar size to the 0.356mm displacement
thickness. The flow, however, was observed to be quasi two-dimensional over the flaps

Estimated by the Author.


1-34
with separation initiated at the end of the upstream flap. Moreover, reattachment
occurred on the downstream (suction) flaps and from surface flow patterns the
separation does not appear to continue downstream of the control region.

Hafenrichter et al. (2003) recorded surface pressure distributions upstream and
downstream of the control plate. However, the distribution over the control region was
not obtained, which is the region of most interest as this can indicate the strength of the
structures within the lambda foot and the boundary layer characteristics through the
interaction region. Within the test region the pressure downstream of the interaction
tended towards but never reached the inviscid value. It would presumably reach that
value at a position further downstream due to the subsonic flow behind a normal shock.
From boundary layer velocity measurements the optimal flap thickness for the four flap
array was observed to be 190m, with an approximate 0.6mm flap deflection (23% of
the boundary layer thickness). Large deflections cause excessive roughness and
injection/bleed of the boundary layer. Small deflections cause the system to convert to
a conventional slot control device with reduced injection and bleed. Their work
concludes by showing that the MART system provides a lambda shock, and therefore
total pressure recovery benefit, with the above four-flap system. However, the six-flap
array did not demonstrate improved total pressure recovery compared to the solid wall
reference case. This was presumably due to reduced mass transfer and increased
surface roughness compared to the four-flap system.

Lee et al. (2004) measured the skin friction distribution and concluded that the MART
system reduces the centreline skin friction downstream of the SBLI compared to solid
wall values, indicating an increased susceptibility to flow separation. Furthermore, the
optimal mesoflap deflection observed by Hafenrichter et al. (2003) gave a higher skin
friction distribution compared to the other flap arrays, which increases with downstream
distance. This is believed to imply better boundary layer recovery from the shock
induced separation. Lee et al. (2004) showed that minimum skin friction distribution
for the MART system was bounded by the values obtained by a conventional 5%
macroporous passive control system.


1-35
Jaiman et al. (2004) simulated MART control of the UNS interaction at a freestream
Mach number of 1.4 using a CFD programme based on Reynolds averaged Navier-
Stokes equation. To simplify the computational processes the interaction was simulated
at pre-determined amounts of flap deflection, instead of being aeroelastically
determined. The flap deflections were varied from 10% to 50% of the incoming
displacement thickness
, to produce sufficient mass bleed without producing

significant additional separation. It was shown that increased flap deflection; 1)
increases the upstream oblique shock strength, presumably due to increased flap
curvature and mass injection; 2) increases downstream static pressure indicating a
reduced total pressure loss; 3) decreases skin friction downstream of the interaction with
a less full boundary layer, which are trade-offs to the higher total pressure recovery; and
4) increased displacement thickness, indicating that MART control increases the
susceptibility to separation. The optimal upstream and downstream flap deflections
were found to be of the order of half of the incoming displacement thickness. It is
unclear whether this is strictly an optimum value as the study only examined tip
deflections up to this value and not at greater levels. Furthermore it was observed that
to provide increased lambda-foot benefit the upstream flap deflection, 0.5
*
, should
be marginally more than the downstream flap deflection, 0.4
*
. The smaller
downstream flap deflection was attributed to the importance of bleeding only a modest
amount of the boundary layer, whereas the increased upstream deflection provided an
improved oblique shock.

It was found that total pressure recovery rises monotonically with the number of flaps,
but with corresponding deterioration in boundary layer behaviour, by way of increased
displacement thickness. Jaiman et al. (2004) concluded that a two or four flap array
would yield a significant improvement in total pressure recovery with only a minor
increase in displacement thickness. Also, a plenum depth larger than one boundary
thickness did not appreciably affect the total pressure recovery. More importantly, they
did show that without a cavity, that is with no recirculation, there was considerable total
pressure recovery. This indicates that the majority of control benefit comes from the
protrusion of the upstream flaps into the flow. However, recirculation reduced the

It would appear that 10 to 50% of the incoming boundary layer thickness would have been more
appropriate from the 0.6mm deflection produced by the 190Zm thick four flap array.

1-36
displacement thickness and presumably the susceptibility to downstream separation.
My calculations indicate that the upstream and downstream flap deflections for this
study were 50% and 40% of the incoming displacement thickness respectively;
however, this is not stated in the paper. Furthermore, it is unclear how the incoming
displacement thickness, and therefore the flap deflection, was calculated; whether from
experimental values or obtained from CFD. Presumably, if a thinner displacement
thickness eventuated in the CFD than used in the initial calculations then the flap
deflection effects would be greater.

Jaiman et al. (2004) went on to state that the tuning of upstream flap positions [relative
to the inviscid shock position] or length can lead to significant improvement in the total
pressure recovery. Shifting the upstream flap location away from the shock would
generate larger shock smearing, a higher bifurcation point and improved total pressure
recovery. Shifting the downstream flap away from the shock would reduce the total
pressure recovery, which is attributed to a reduced bleed effect after the separation zone.
This appears to be an increased boundary layer thickness effect and not due to the acting
pressure, which should increase with downstream distance to promote mass transfer.

Tharayil and Alleyne (2004) analysed the applicability of the MART system for active
control of an oblique SBLI utilising the Nitinol material, a thermally activated shape
memory alloy. The flaps were annealed flat and in such a way that when heated they
would tend towards the closed, zero deflection, position. However, under a pressure
difference of 4.5kPa the mesoflaps could not return to the closed position, which is
presumably due to their thickness
***
. The author postulates that 4.5kPa is a low acting
pressure but improvements in shape memory alloys will lead to the ability of the system
to overcome greater pressures and allow greater deflection/control. Furthermore, the
MART system cannot be fully utilised for varying shock strength or position. Nitinol
can only produce deflection in one direction. That is it can only assist or inhibit
deflection. If the shock were to move about the mesoflap position then the deflection
would change pole and the flaps, which originally inhibited deflection, would be
passive at best. Also, if the strength of the shock or upstream static pressure, and

***
The thickness of the flaps were stated as 0.152m, however, it appears that the authors meant
0.152mm.

1-37
therefore the amount of mesoflap deflection, changed then the system would be unable
to adapt during flight to the new conditions. It appears that Nitinol can only produce a
predetermined amount of flap deflection, which depends on its structural size and
fabrication. The author speculates that Nitinol cannot alter the degree, or pole, of
overall deflection during flight except for two predetermined positions (aeroelastic
deflection with or without the predetermined Nitinol deflection).

Additionally, there is some question over the capability of the system to overcome the
extreme temperature ranges experienced in practice, from sub-zero on high altitude
transonic aircraft to the hot intake of an engine intake. Nitinol is a thermally activated
material and insulation would presumably be required, which would inhibit the
materials performance.

1.2.6. - Piezoelectric Actuators for Flow Control
The uses of piezoelectric actuators in aerodynamic structures have been widely
considered for dynamic flow control at low subsonic conditions. Jeon and Blackwelder
(2000) used piezoelectric actuators to apply an oscillatory disturbance in the near wall
region of a turbulent boundary layer. Cattafesta et al. (2001) optimised unimorph
vibrations to control the separation over a rearward facing step in low subsonic flow.
Choi et al. (2002) and Seifert et al. (1998) used vibrating actuators on low speed
aerofoils to successfully increase lift by delaying separation. However, there seems to
be minimal literature on the use of piezoelectric material for static control of aerospace
structures, presumably due to the size of material required for appreciable deflections.

Koratkar and Chopra (2000) investigated the use of multilayer piezoelectric actuators
for trailing edge flaps to control the vibration and noise level on helicopters. The rotor
flow field is extremely complex and can include transonic flow on the advancing blade,
dynamic stall on the retreating side of the disc, highly yawed and reversed flow, and
blade-wake interactions with tip vortices from preceding blades. He estimated that the
actuator flap assembly and leading edge weights for mass balancing results in a 17%
increase in rotor-blade mass. This equates to 0.6% increase in gross take-off weight,
compared to the higher 2% weight penalty associated with passive vibration reduction
equipment available.

1-38
1.3. - Present Investigation

1.3.1. - Active Control of Shock/Boundary Layer Interaction (ACSBLI)
Historically, active control has been marred by the drag reduction obtained being
compromised by an additional drag due to the power requirements, for example, the
pumping power in the case of mass transfer and the drag of the devices in the case of
vortex generators. Passive control, generally, produces a reduction in total drag above a
normal shock Mach number of 1.15, or for a freestream Mach number, over a transonic
aerofoil, of 0.80. Below a normal shock Mach number of 1.15 the reduction in wave
drag is less than the increase in friction drag. Furthermore, it is not possible to, turn
passive control off at low Mach numbers, control its position or the amount of mass
transfer and its distribution during operation as discussed in section 1.2.4.
Friction
drag, due to the rough porous surface, also increases at off-design conditions, i.e. in
the absence of a shock or a shock located away from the control region.

A system of piezoelectrically controlled flaps is now presented for the control of the
interaction, see fig. 1.21. The flaps would aeroelastically deflect due to a pressure
difference created by the presence of a shock. The piezoelectric material would then
either assist or inhibit the aeroelastic deflection. The amount of deflection, and hence
mass flow through the plenum chamber, would control the amount of interaction
control. If the piezoelectric material could completely counteract the aeroelastic

To utilise passive control the system (location and size) would have to be predetermined before flight.

1-39
deflection then the flaps could close, providing zero deflection and minimal mass
transfer/control. This would produce a smooth surface to the flow in the absence of a
shock wave and at low local Mach numbers when the presence of passive control
hinders the drag reduction. It should be pointed out, however, that no feedback system
is considered in this thesis and the term active control refers to controlling the overall
flap deflection and hence the bleed/suction rate.

The existence of a matrix of active piezoelectric flaps on a transonic aerofoil, which
could open or shut independently, would enable localised control of the interaction as
the swept or unswept shock moves over the aerofoil at different flight conditions, whilst
producing a smooth surface away from the shock position. For a practical application,
on a transonic transport aircraft aerofoil, the region of control would have to be
approximately 1m long. A porous surface would generate large frictional drag due to
the increased roughness away from the shock compared to a matrix of piezoelectrically
unimorph flaps.

Piezoelectric material was selected for flap construction due to its bipolar nature, which
is the ability to produce positive or negative strain/forces. If the shock can move over a
matrix of flaps then each flap would need to be able to overcome the aeroelastic
deflection due to the presence of the shock regardless of direction. Furthermore, the
bipolar characteristic enables the piezoelectric material to assist the aeroelastic
deflection to promote mass transfer.

The plenum chamber pressure would be constant, at an average of the upstream and
downstream pressure as indicated by Hafenrichter et al. (2003). Initially, the
piezoelectric flaps are modelled with a constant pressure load to predict tip deflection.
The pressure difference acting across the flap will be in excess of 15kPa, the difference
between the chamber pressure and the free stream pressure above the flap. Improved
theoretical studies will account for the fact that there will be a pressure gradient through
the plenum chamber and above the flaps.

A number of design options are considered for the integration of the piezoelectric
ceramic, also known as PZT (Lead Zirconate Titanate), into the flap structure. These

1-40
include the use of unimorphs, bimorphs and polymorphs, with the latter capable of
being directly employed as the flap. The choice of material and structural configuration
for the piezoelectric flaps are discussed in chapters 3 and 4. Active control can be
utilised to optimise the effects of the boundary layer shock wave interaction as it would
allow the ability to control the position of the control region around the original shock
position, mass transfer rate and distribution.

Chapter 2. Experimental Arrangement

2-41
CHAPTER 2 - Experimental Arrangement

2.1. - ADFA Experimental Facility and model

All swept normal SBLI experiments were performed in the supersonic wind tunnel
facility at the School of Aerospace, Civil and Mechanical Engineering, University
College, The University of New South Wales at the Australian Defence Force Academy
(UNSW@ADFA), in Canberra, Australia.

2.1.1. - The Wind Tunnel Facility
The schools supersonic blow-down wind tunnel consists of an air compressor, three
reservoirs, control valve with pneumatic control circuit, settling chamber, first throat
with nozzle liners, second throat, subsonic diffuser and exhaust system with muffler.
The compressor plant includes two oil free compressors, two filters and two silica gel
driers.
Figure 2.1 ADFA Supersonic Blow-down Wind Tunnel Schematic.

The compressor plant fills up the three reservoirs with dried air to 1400kPa. The dried
air is fed to the settling chamber via a control valve system, which maintains a quasi-
steady reservoir pressure throughout the tunnel run. The air from the settling chamber
goes through a series of wire meshes to render the flow uniform and then into the test

2-42
section equipped with a Mach 2 nozzle liner, see fig. 2.1. The flow is steady and
uniform as it accelerates through the test section nozzle to supersonic speeds. In the
present study, the experiments were conducted mostly with a stagnation pressure of
343kPa, which gives a steady running time of approximately 27 seconds. The flow is
then decelerated through a parallel section that acts as a second throat and then onto the
subsonic diffuser that exhausts the flow to atmosphere. Magi (1990) provides more
details about the tunnel, its control system and operation.

The wind tunnel test section has a cross section width and height of 90mm and 155mm
respectively, see figs. 2.2a & b. The upstream Reynolds number, for a stagnation
temperature and pressure of and 294K and 343kPa respectively, is 45.67x10
6
m
-1
.
a)
b)
Figure 2.2 ADFA Supersonic Blow-down Wind Tunnel
a) photo of the Mach 2 liners with wedge on the floor to create a swept shock
and b) Schematic showing dimensions.

2-43
Following Crocco and Lees (1952), with a 1/7
th
power law profile the uncontrolled
turbulent boundary layer initiates at the throat to produce boundary layer and
displacement thicknesses of 7.57mm and 1.79mm respectively at the leading edge of the
wedge, see table 2.1.

Incoming
Mach number
M
Boundary Layer
Thickness
o
o
(mm)
Displacement
thickness
o
o
* (mm)
Momentum
Thickness
u
o
(mm)
Shape Factor
H
Unit Reynolds
number
9 (m
-1
)
2 7.57 1.79 0.59 3.07 45.67x10
6
Table 2.1 UNSW@ADFA Oncoming Boundary Layer Properties.

2.1.2. - Wedge/Shock Generator
Passive control is effective in reducing the total drag after incipient separation occurs.
For a swept normal shock interaction this occurs above a free stream Mach number of
approximately 1.2, as discussed in chapter 1. It was, therefore, decided that the swept
shock should have a normal Mach number of approximately 1.3. The pressure
difference across the shock is proportional to the pressure load on the active control
flaps, which designates the amount of flap deflection available and hence the mass
transfer for control, which is further discussed in Chapter 4.

Figure 2.3 ADFA Test Section with Shock Generator.

To produce a shock wave giving a normal Mach number of 1.3, a wedge was fabricated
to have an angle of 11 to the Mach 2.0 free stream, see fig. 2.3. This would
theoretically produce a shock angle of 40.5 with a static pressure rise ratio of 1.802.

2-44
This implies that, assuming a constant mean pressure in the plenum chamber, each
control flap will be subject to 40.1 % of the upstream static pressure. If the freestream
static pressure in the test section is 44kPa, corresponding to a stagnation pressure of
343kPa, then the control flaps will have an 18 kPa pressure difference acting across
them.

2.1.3. - Model design
a)
b)
c)
Figure 2.4 Control Plate for Swept
Normal SBLI Control showing
a) Pressure Port and Unimorph Layout,
b) Streamlines and c) A Photograph of the
Unimorph Control Plate.

A plenum chamber (C140mm x 20mm) is placed underneath the swept shock in the
sidewall of the test section. The plenum chamber is covered by a plate containing the
control flaps and numerous pressure ports, to record the surface pressures in both
streamwise and transverse directions, see figs. 2.4a & b. The flaps were cut into the
control plate by electro discharge machining (EDM) with a gap of 0.2mm between the

2-45
model skin and the flap. The piezoelectric ceramic is bonded to the underside of the
flap and a voltage is applied to control the actuator deflection and thus mass transfer
rate, see fig. 2.5. The unimorph flap thickness was 0.9mm for the pressure sensitive
paint study and 1.1mm for the discrete pressure measurement survey. A more detailed
sizing of the flaps is discussed in chapter 4.

Figure 2.5 Unimorph Flap Design for Swept Normal SBLI Control.

SBLI control with known amounts of flap deflections was examined to identify the
optimal deflection for control, the flow characteristics and the level of performance
available. Known flap deflections were achieved using a mechanical system to position
the flaps by a pre-determined amount prior to experiments, see section 2.2.3.

2.1.4. - Flow visualisation
Figure 2.6 ADFA Schlieren System.

A schlieren system is an optical, non-intrusive, method based on refraction of light rays
in a varying density flow. It is a common technique for the study of supersonic flow
because it provides good qualitative information about the flow features and is

2-46
relatively simple to set-up. The principle of operation of schlieren is only briefly
mentioned here as it has been widely covered in the literatures, see Liepman and
Roshko (1957) and Kleine (2001).

A light beam is deflected and displaced by different amounts when it travels through an
area of varying density fluid. The angular deflection and the displacement are
proportional to the first and second derivative of the gradient respectively.
Shadowgraph flow visualization utilises the second derivative, the displacement, by
placing a screen behind the test section to show light and dark regions associated with
areas where the light has diverged or converged. When a knife edge is positioned, to
cut-off part of the basic image source, the illumination of deflected light rays at the
viewing screen increases or decreases. The illumination change, due to the light cut-off,
is proportional to the density gradient, the first derivative, and this is called a schlieren
system.

A direction-indicating colour schlieren system is created when the knife edge is
replaced by a circular iris aperture and a colour mask is attached to the light source.
The use of a circular aperture allows density gradients in any direction to be observed.
Whereas, a knife edge will only allow gradients perpendicular to the knife edge to be
observed. Also, the use of a colour mask allows information to be attached to certain
colours, where specific colours indicate the direction of the pressure gradient.
Furthermore, variations in colour hue are easier for the human eye to distinguish rather
than the degree of illumination with a monochrome system. A more detailed discussion
of colour schlieren systems is given by Kleine (2001).

A continuous point light source was used with the exposure time controlled by the
shutter speed of the camera. A short exposure time enabled us to obtain so-called
frozen images, which minimize any movement observed of the shock. The images
were obtained in complete darkness using a standard Canon SLR camera.


2-47
2.1.5. - Oil flow visualization
A mixture of sesame oil and soy sauce was used to obtain initial oil flow visualisation
pictures. It was applied with one brush stroke across the span of the test surface.
Sesame oil by itself is stripped from the surface of the test section during a run and soy
sauce by itself adheres to the test surface but tends to smear. The emulsion of the two
produces the desired trace that stays on the model and does not smear on tunnel start
up/shut down.

Additionally, oil flow visualisation was achieved with an emulsion of paraffin, titanium
dioxide and oleic acid applied to the control plate with the test surface painted matt
black. The flow of the mixture was observed through the runs to ensure that the mix
dried during the experiment and not at tunnel start up or shut down.

2.1.6. - Pressure Sensitive Paints - Couldrick et al. (2004b)
Although pressure sensitive paint (PSP) is a relatively recent flow visualisation
technique, Morris et al. (1993) note that they have been increasingly used in
quantitative aerodynamic applications because of their ability to provide a picture of the
instantaneous pressure field over entire surface without disturbing the flow.

Klein (2000) notes that PSP techniques are based on the deactivation of
photochemically excited organic molecules, the so called luminophores, by oxygen
molecules. This process is called oxygen quenching. A luminophore is shifted to an
excited electronic state (singlet) when it absorbs a photon of appropriate energy. This
excited luminophore can return to the ground electronic state by emitting a photon
(photoluminescence). It can also return to the ground state by a collision with an
oxygen molecule. If this occurs, the oxygen absorbs the excess energy of the
luminophore, and the transition to the ground state is radiationless. If the number of
oxygen molecules increases the chance of excited luminophores undergoing
radiationless transfers increases that is, the photoluminescence decreases.

Radiometric imaging was used to determine the global pressure distribution. It requires
a calibration curve to be established before a pressure distribution can be determined
from the images over the model. The pressure can be calculated using the ratio of the

2-48
ambient pressure intensity and the intensity during a run. The rewritten Stern-Volmer
equation is then given as.

2
3 2 1
) ( ) ( ) (
|
|
.
|
\
|
+
|
|
.
|
\
|
+ =
I
I
T k
I
I
T k T k p
ref ref
[Eq. 2.1]

where I
ref
and I are the intensities at ambient pressure and during the run respectively.
k
1
, k
2
and k
3
are temperature dependant constants.

A Unicoat aerosol PSP Pt(TfPP) with a response time of less than a second was utilised
for ease of application combined with a UV LED 464nm light source, see fig. 2.7. The
camera is perpendicular to the test surface and due to logistics the light source is
positioned 10 lower and downstream of the camera. The angle of the light source
creates a crescent shadow on the lower downstream test surface, which creates pressure
noise in the imaging process in the shadow region.

Figure 2.7 Camera/Test Surface/Light Source Set-up for PSP experiments.


2-49
2.1.7. - Adiabatic Wall Temperature
In order to minimize the effect of heat transfer between the freestream and the PSP/test
surface, the experiments were conducted with flow wall temperatures as close to the
adiabatic wall temperatures as possible. It can be seen from equation 2.1 that the PSP is
temperature dependent and the luminescence is variable with heat transfer. Therefore,
PSP temperature variation across the test surface would create false pressure results.

In order to minimize this effect the upstream and downstream flow adiabatic wall
temperatures, T
aw1
and T
aw2
, were calculated using the incoming Mach 2 flow with a
measured upstream stagnation temperature, T
01
, starting at 287K (K14C) and falling to
285K (K12C) after 20 seconds runtime. The tests were conducted with an initial wall
temperature, T
w
, of 0C 2C. Calculations used a recovery factor, r, and ratio of
specific heats, , of 0.892 and 1.4 respectively for the Mach 2 flow.

8 . 1
2
1
1
2
1
01
=
+ = M
T
T
[Eq. 2.2]
892 . 0
1 01
1 1
=
=
T T
T T
r
aw
[Eq. 2.3]

With a measured stagnation temperature, T
01
, of 287K, an upstream static temperature,
T
1
, of 159K is produced, using eq. 2.2. Using eq. 2.3 this produces an upstream
adiabatic wall temperature, T
aw1
, of 273K. The 11 wedge angle, M, generated a main
shock at an angle, |, of 41 to the flow, with a normal Mach number, M
n1
=M
1
sin|, of
1.312. From normal shock properties:

( ) [ ]
( )
608 . 0
2 / 1
2 / 1 1
2
1
2
1 2
2
=

+
=

n
n
n
M
M
M [Eq. 2.4]
Therefore,
7799 . 0
2
=
n
M [Eq. 2.5]
And
56 . 1
) sin(
2
2
=
=
u |
n
M
M [Eq. 2.6]

2-50
Using the temperature difference equation across a normal shock.

( )
( )
( )
1985 . 1
1
1 2
1
1
2
1
2
1
2
1 2
1
1
2
=
(
+
+
(

+
+ =
n
n
n
M
M
M
T
T

[Eq. 2.7]

This results in a downstream flow temperature, T
2
, of 191K. Using eqs. 2.2 and 2.3
this produces a downstream adiabatic wall temperature, T
aw2
, of 274K. As the
stagnation temperature drops to 285K the upstream and downstream adiabatic wall
temperatures, T
aw1
and T
aw2
, will both by 2K. This would imply minimal heat transfer
to the wall during the run.

The walls of the wind tunnel were rendered cool to 0C 2C by exposing the entire test
facility to the ambient weather in winter overnight. The experiments were conducted at
approximately 5am when the ambient temperature was approximately -6C with fans to
maximized cool air circulation within the facility. The stagnation temperature did not
drop to quite 0C, which is thought to be due to the hot compressors heating reservoir
air.

2.1.8. - Data Acquisition and Controller System
The UNSW@ADFA wind tunnel data acquisition/control system consists of an NEC
(IBM clone) fitted with an Analogue Devices high speed A/D, digital I/O card (815F)
and Metrabyte digital I/O card (PIO-12). The control, acquisition and analysis
programs were written using subroutines from the software package ASYST, which is
written in Forth, see Appendix C2. The system drove the scanivalve and acquired the
stagnation temperature and pressure. The stagnation pressure was measured using a
Druck PCD22 (14bar) pressure transducer feeding a Druck DPI 201 pressure indicator
and the single ended analogue output was fed into an A/D converter. The test-
section/model static pressure was measured using a Druck PCD22 (3.5 bar) pressure
transducer installed into the scanivalve. The transducer was coupled with a Druck DPI
260 pressure indicator and the single ended analogue output was fed into the A/D
converter. Both transducers were calibrated using the gravity assisted weight unit in the

2-51
supersonic wind tunnel. The scanivalve has 48 pressure tappings available and steps,
when commanded, through the 48 tappings.

The stagnation temperature was measured using a J/K thermocouple having a time
constant of 0.1sec. The amplified output, calibrated on a three-point sliding scale, was
fed into the A/D converter.

2.1.9. - Experimental Accuracy
Transducer calibrations were conducted prior to each experimental study. The
barometer used for atmospheric pressure could be read to 0.05%. When these values
are used in the calibration to convert voltage to pressure for the data acquisition system
errors of 0.5% are introduced to the stagnation and static pressure measurements
respectively. These relate to accuracies of 0.003p/P
0
for static pressure measurements
and 3x10
-6
for the calculated drag coefficient C
d
.
The accuracy of the pressure sensitive paint survey is dominated by the data processing
technique used, as discussed in section 5.4.1, which produces a decreased accuracy of
0.02p/P
0
.
2.1.10. - Voltage Generators/Power supplies - Physik-Instrumente (1996)
In order to generate high voltages for the piezoelectric ceramic material (PZT)
excitation, OEM amplifier modules were obtained from Physik Instrumente, see fig. 2.8
and Appendix D1. These amplifiers consist of a Pulse Width Modulator, running at a
nominal frequency of 25 kHz, driving a MOS-FET, which switches low voltage DC into
the primary of a ferrite cored step up transformer. The output of the transformer is
rectified by a string of high peak inverse voltage fast recovery diodes before being
filtered and current limited for operator safety.


2-52
Figure 2.8 High Voltage Amplifier with casing from Physik Instrumente

The output voltage is sampled with a user adjustable resistive divider, which acts as the
output voltage control. This sample was then compared to a reference voltage and the
comparator output was used to drive the pulse width modulator to the desired output
voltage.

The OEM operational amplifiers are low cost amplifiers for high voltage PZTs, with an
output voltage range of 10 V to 1000 V 0.5V. With a 12 V supply, it can output a
peak current of 0.5mA, with an average current of 0.3mA, and can operate with an input
current of only 80mA at 12 V, which means battery operation is possible.

2-53
2.2. - Cambridge experimental facility and model

All Unswept Normal SBLI experiments were conducted in tunnel number 2 of the High
Speed Aerodynamics Laboratory, Department of Engineering, University of Cambridge,
UK, see fig. 2.9.
Figure 2.9 University of Cambridge Blow-down Supersonic Wind Tunnel #2.

2.2.1. - Wind Tunnel arrangement
Tunnel number 2 is a blow-down tunnel that achieves supersonic conditions using air
supplied from compressed air bottles, into a settling chamber, through shaped liners and
then passes to the test section, which has a width of 114mm and a height of 178mm.
The flow passes through a diffuser and is exhausted to atmosphere.
Figure 2.10 Cambridge Supersonic Wind Tunnel Arrangement
with interaction Control Plate.

2-54
The investigation makes use of a normal recovery shock wave, which is generated at
tunnel start up using a second downstream throat, and the naturally grown turbulent
boundary layer on the tunnel floor, see fig. 2.10. By adjustment of the ratio of settling
chamber pressure to back pressure the recovery shock is held at a given position for the
duration of the experiment, allowing the interaction to be studied and the shock wave
position is determined during operation using a schlieren arrangement.

The tunnel was operated at free stream Mach numbers of 1.3 and 1.5. Theoretically,
unimorph control at these Mach numbers should show reduction in overall drag
compared to the uncontrolled case. Holden (2004) has shown that the flow is
incipiently separated at Mach 1.3 and fully separated at Mach 1.5. The tunnel was
operated at a stagnation pressure and temperature of approximately 150kPa and 295K.
The oncoming boundary layers properties at the start of the control plate, calculated
from Crocco and Lees (1952), are shown in table 2.2.

Mach number
M
Boundary Layer
Thickness
o
o
(mm)
Displacement
thickness
o
o
* (mm)
Momentum
Thickness
u
o
(mm)
Shape Factor
H
Unit Reynolds
number
9 (m
-1
)
1.3 7.47 1.34 0.652 2.04 24.38x10
6
1.5 7.54 1.46 0.63 2.30 23.65x10
6
Table 2.2 University of Cambridge Oncoming Boundary Layer Properties.

At these Mach numbers and approximate operation conditions, Holden (2004) has
shown the tunnel to exhibit two dimensional SBLI characteristics up to a spanwise
distance of 60mm across the tunnel centre line. Smith (2002) provides more details
about the tunnel, its control system and operation.

2.2.2. - Co-ordinate system.
Shown in figure 2.11 is the co-ordinate system that will be used throughout this and the
following sections. The streamwise distance (X) is measured from the centre of the
viewing window, which corresponds to the start of the downstream unimorph. The
vertical distance (Y) is measured from the bottom tunnel liner and the spanwise distance
(Z) is measured from the tunnel centerline.


2-55
a)
b)
Figure 2.11 University of Cambridge Control Plate showing pressure port layout for
a) piezoelectric actuation unimorph flap control and
b) mechanically pre-set flap deflection control.


2-56
2.2.3. - Unimorph control device
Figure 2.11a is a plan of the unimorph control plate containing the flaps and numerous
pressure ports. Figure 2.11b shows the increased number of pressure ports that were
available on the flaps for the pre-set flap deflection study. The presence of the
piezoelectric actuators inhibited the number of tappings due to the materials fragility
and brittleness. The individual flaps have a length and width of 52mm and 25mm
respectively and the plenum chamber, 38mm in depth, is placed underneath the control
plate in the floor of the test section. The plenum extends 30mm upstream and
downstream of the flaps positions.

Figure 2.12 Unimorph Flap Design for Unswept Normal SBLI Control.

The flaps were cut into the plate by electro discharge machining (EDM) with a gap of
0.2mm between the model skin and the flap, see figs. 2.11 & 2.12. The gaps so
produced can be considered as providing continual passive slot control of the SBLI.
The slots are longitudinal and lateral in nature for the unimorph edges and ends
respectively. The piezoceramic is bonded to the underside of the flap and a voltage is
applied to control the actuator tip deflection and thus mass transfer rate. The sizing of
the unimorph flaps (flap thickness) is further discussed in chapter 4.

The flaps were also mechanically deflected to certain prescribed levels to analyse the
SBLI control possible at enhanced levels of deflection. Figures 2.13a & b shows the
upper (free stream) and lower (plenum chamber) side of the control plate respectively
with the mechanical devices to control the flap deflections. The mechanical device
allows a continuous amount of flap deflection via a screw and pot system: However,
only SBLI Control with flap deflections 0mm, 1mm, 2mm and 3mm were examined due
to time constraints.

2-57
a)
b)
Figure 2.13 a) Unswept Normal SBLI Control
Plate showing pressure port layout and
b) the mechanical deflection devices

2.2.4. - Measurement locations
a)
b) c)
Figure 2.14 a) Pitot tube set-up with b) a flat head pitot tube
and c) a straight pitot tube - Smith (2002)

2-58
Surface static pressures were measured by tapping the various control plates with holes
of 0.5mm internal diameter and connecting the tappings to pressure transducers. The
locations of surface pressure measurements are shown by the open circle symbols in
figs. 2.11a & b. Pitot pressure traverses were performed at various streamwise and
spanwise positions, with a flat head and straight pitot probe, see figs. 2.14 a c.

2.2.5. - Flow visualization
In order to visualize the flow in the test section a schlieren set-up was utilized with both
horizontal and vertical cut offs, as discussed in section 2.1.4. Surface oil flow
visualization was performed using a mixture of paraffin, titanium dioxide and oleic
acid, which was applied to the tunnel liners and control plate. The flow of the mixture
was observed through the runs to ensure that the mix dried during the experiment and
not at tunnel start up or shut down.

2.2.6. - Calculation of velocity - Smith (2002)
The Cambridge data acquisition program was written by Smith (2002), with the velocity
profiles evaluated from the pitot pressure, P
0
, profiles and the surface static pressure, P
s
,
measurements. The surface static pressure and stagnation temperature are assumed
constant across the boundary layer. Then, using the ratio P
0
/P
s
, the Mach number can
be calculated via the isentropic relationship:

( )
( ) 1
2 02
2
1
1

(

+ =

M
P
P
s
[Eq. 2.8]

for subsonic flows, and

( )
( )
( )
1
2 1
1 2 4
1
2
1
2
2 2
02
+
+
(

+
=

M
M
M
P
P
s
[Eq. 2.9]

for supersonic flows, where a shock will exist upstream of the pitot tube.


2-59
No method of temperature control was available but according to Smith (2002) the wind
tunnel stagnation temperature, T
0
, and tunnel wall temperature, T
w
, are approximately
equal to the ambient laboratory temperature. The stagnation and wall temperatures,
during the runs, remain relatively constant, within 0.8% and 1% respectively. Taking a
recovery factor, r, of 0.89 and using

( )
(

+ =
2
2
1
1
e
e
r
rM
T
T
[Eq. 2.10]

( )
(

+ =
2 0
2
1
1
e
e
M
T
T
[Eq. 2.11]

where the subscripts r, 0, s, w, and e denote the recovery, stagnation, static, wall and the
boundary layer edge values respectively. Then using Crocco velocity-temperature
distribution,

2
1
|
|
.
|
\
|
|
|
.
|
\
|
+
|
|
.
|
\
|
+ =
e e
r
e e
w
e
r
e
w
e
s
u
u
T
T
u
u
T
T
T
T
T
T
T
T
, [Eq. 2.12]

the velocity profiles can be calculated from the Mach number profiles.

2.2.7. - Measurement apparatus calibration and checks
All static and pitot pressures were measured using Druck PDCR200 fast response
differential pressure transducers. The settling chamber stagnation pressure was
measured using a Druck PDCR22 fast response transducer. Atmospheric pressure was
read from a mercury barometer located in the laboratory before every run. In order to
minimize the time lag of the traversing pitot probe the shortest possible length of pipe
was used to connect the probe to the transducer with an adequate response time. More
detailed discussions of the apparatus are given by Holden (2004) and Smith (2002)


2-60
2.2.8. - Experimental Accuracy
The Cambridge data acquisition system was created and analysed by Smith (2002), who
states that errors are introduced into the experiment during the transducer calibration.
The mercury manometer could be read to 0.1% accuracy and the barometer used for
determining atmospheric pressure could be read to 0.005% accuracy. When these
values are used in the calibration to convert voltage to pressure, errors of 0.1% are
introduced into the static and pitot pressure measurements and an error of 0.08% is
introduced into the measure stagnation pressure.

These errors result in uncertainty of the calculated Mach number of the order of 1.6%.
He goes onto state that due to the presence of pressure tappings to measure surface
static pressure will cause the measured values to differ from the true values. However,
the size of the pressure tappings used will result in very small errors that can be
considered negligible. Furthermore, when measuring pitot pressure, probe displacement
effects are significant near the surface. However for the flat head pitot probe d/o~0.02,
where d is the diameter or height of the probe, the errors due to displacement are
thought to be negligible. The straight pitot probe, which has d/o~0.2 but is only used in
the free stream where velocity gradients are minimal, is assumed to produce negligible
additional inaccuracy. The traverse gear was found to be accurate to 0.1mm for a
30mm traverse, which corresponds to a 0.7% error in measured pitot pressure during a
typical traverse. However, this compounds to errors of 8% for calculations of velocity
profiles.

The position of the shock wave is itself subject to a degree of uncertainty due to tunnel
operator control and shock unsteadiness. The shock wave can be positioned to within
4mm of any desired location (3.5% of the control plate length), and any unwanted
movements larger than this can be detected by analyzing surface pressure
measurements.

Chapter 3. Piezoelectric Flap Actuators

3-61
CHAPTER 3 - Piezoelectric Flap Actuators

3.1. Introduction
This chapter considers the use of smart materials for structural control and proceeds to
the use of piezoelectric ceramics for active SBLI flap control. It will examine the use of
piezoelectric ceramics as unimorphs, bimorphs and multimorphs as the flaps and their
ability to produce deflection whilst withstanding the pressure loads expected during
wind tunnel testing. The chapter looks at the piezoelectric ceramic properties and
design configuration options for SBLI control with the sizing of the chosen flaps studied
in chapter 4.

3.2. - Smart Materials
Piezoelectrics are a popular smart material, which undergo deformation when an electric
field is applied across them and, conversely, produce a voltage when strain is applied
and, therefore, can be used as both actuators and sensors. Piezoelectric material
characteristics are relatively linear at low applied fields and bipolar, that is they can
produce positive and negative strains. Also, they exhibit hysteresis, which is important
for dynamic responses. Piezoelectric sheets generate isotropic strains on the surface and
a non-Poissons strain across the thickness. Bimorphs or bending actuators are
available commercially where two or more layers of the piezoelectric ceramic are
stacked with a thin shim between them. If opposite polarity fields are applied to the two
sheets, a bending action is created. Bimorphs produce larger displacement with a
smaller applied voltage compared to single piezoelectric elements bonded to an inert
substrate. Chen et al. (2003) have shown that the dielectric constant of the piezoelectric
material, which is proportional to the induced strain, can be increased by 50% with
additives.

The maximum positive field for piezoelectric actuators is limited by the breakdown of
the dielectric, whereas the maximum negative field is limited by depoling of the
piezoelectric. For PZT-5H, the material selected in this thesis as the actuation material,
the DC depoling field is 5.5kV/cm and if a much higher field is kept for a long time, the

3-62
material gets polarised along the new poling direction. Thus it is possible to repole the
piezoelectric by exciting it with AC or DC fields. Additionally, if a compressive stress
is acting in the plane of the piezoceramic, normal to the polarization direction, it
destroys some of the initial polarisation. That is, it reduces the net polarisation which
can permanently degrade soft ceramics like PZT-5H. It should be remembered that
piezoelectric ceramic material is brittle and high tensile stresses might cause it to crack.
Also, depoling is possible if the operating temperature exceeds the Curie temperature.

Electrorestrictive materials, such as lead magnesium niobate (PMN), are similar to
piezoelectric materials with about twice the strain capability as shown by Ren et al.
(2000). They are very sensitive to temperature, exhibit negligible hysteresis and have a
non-linear relation between the applied field and the induced strain. Chopra (2002),
Mid (2003) and Newport (2003) indicate that much larger strain is capable with
electrorestrictive material, compared to piezoelectric material. However,
electrorestrictive material is monopolar - that is it can only produce positive strain
regardless of the electric field polarity, whereas piezoelectric materials are bipolar; they
can produce both positive and negative strains. As discussed in chapter 4 the flaps must
be capable of inhibiting the aeroelastic unimorph flap tip deflection due to the pressure
difference regardless of its position relative to the shock.

Among other smart materials, shape memory alloys (SMA) appear attractive as
actuators because of the possibility of achieving large excitation forces and
displacements, over an order of magnitude greater than electrorestrictive and
piezoelectric materials. When plastically deformed at a low temperature, these alloys
recover their original undeformed condition if their temperature is raised above the
transformation temperature. The most common SMA material is Nitinol (nickel
titanium alloy) as used in the MART system, see section 1.2.5. Furthermore,
magnetorestrictive materials, such as Terfenol-D, exist which elongate when exposed to
a magnetic field to generate low strains and moderate forces. These materials are
monopolar, non-linear and exhibit hysteresis. Piezoelectrics and electrorestrictors are
available as ceramics, whereas magnetorestictors and shape memory alloys are available
as metal alloys.

3-63
The study of intelligent structures, a subset of active and controlled structures, using
piezoelectric material has been widely researched. In general, the material is bonded to
or embedded into the surface of an inert structure. An electric field is applied inducing
a strain in the piezoelectric, which is constrained by the inert material generating forces
and moments to alter the shape of the structure. A smart piezoelectric beam usually
comprises piezoelectric layer(s) and a host layer(s) on a conventional beam. The
piezoelectric layer is required to perform sensing and actuation tasks, while the host
layer functions as a load bearing element.

A bimorph is a strip made up from two layers of materials with different physical
properties. If the bimorph is subjected to conditions such that the strains in each layer
would be unequal, then it will bend. The best known example is the bimetallic strip.

3.3. - Piezoelectric Ceramic Properties

The following two pages, section 3.3., is a summary of a publication produced by the
manufacturer of the piezoelectric material used, see Morgan Matroc (2000b). Any
comments by the author have been placed as footnotes for clarity.

Piezoelectricity is a material property exhibited by certain classes of crystalline
materials. When mechanical pressure is applied, the crystalline structure produces an
impulse voltage proportional to the pressure. Moreover, when an electric field is
applied to the bipolar material, the crystalline structure produces positive or negative
strain proportional to the applied field. If the piezoelectric ceramic is attached to
another material, then a force and moment will be created changing the composite
materials dimensions.

Piezoelectric, also known as polycrystalline, ceramics are more versatile than natural
piezoelectric crystals. Their physical, chemical and piezoelectric characteristics can be
tailored to specific applications. They are hard, dense ceramics and can be
manufactured in almost any given shape or size. They are chemically inert and immune
to moisture and other atmospheric conditions. In addition, the mechanical and electrical

3-64
axes of these ceramics can be precisely orientated in relation to the shape of the
ceramic. These axes are set during poling, the process that induces piezoelectric
properties in the ceramics. The orientation of the d.c. poling field determines the
orientation of the mechanical, electrical axes and can be applied so the ceramic exhibits
piezoelectric responses in various directions or combinations of directions.

3.3.1. - Aging
Most of the properties of piezoelectric ceramics change gradually with time. These
changes tend to be logarithmic with time after poling. The aging rate of various
properties depends on the ceramic composition and on the way the ceramic is processed
during manufacture. Exact values of various properties, such as dielectric, coupling and
piezoelectric constants may only be specified for a standard time after poling due to
aging. The longer time period after poling, the more stable the material becomes.

3.3.2. - Temperature, Voltage and Stress limitations
Piezoelectric materials have particular operating temperature, voltage and stress limits.
The particular chemical composition of the material determines the limits. Operating
outside these limits may cause partial or total depolarisation of the material, and a
diminishing or loss of piezoelectric properties.

3.3.2.1. - TEMPERATURE LIMITATIONS
As the operating temperature increases, piezoelectric performance of the material
decreases, until complete and permanent depolarisation occurs at the materials Curie
temperature. The Curie temperature, 250C for the PZT-5H material used, is the
limiting operating temperature for a given piezoelectric material. When the ceramic
element is heated above the Curie temperature, all piezoelectric properties are lost. In
practice, the operating temperature must be substantially below the Curie point. The
materials temperature limitation decreases with greater continuous operation or
exposure. At elevated temperature the aging process accelerates, piezoelectric
performance deceases and the maximum safe stress level is reduced.
*
*
It is believed that the applicability of piezoelectric actuators is not degraded for aircraft aerofoil as they
would operate at appreciably lower temperatures. However, it is unclear the suitability to a hot engine
intake.

3-65
3.3.2.2. - VOLTAGE LIMITATIONS
A piezoelectric ceramic can be depolarised by a strong electric field with polarity
opposing the original poling voltage. The limit on the field strength is dependent on the
type of material, the duration of the application and the operating temperature. The
typical operating limit for PZT-5H is between 500Vmm
-1
and 1000Vmm
-1
for
continuous application.
It should be noted that alternating fields can have the same

effect during the half life cycle that is opposite to the poling direction.

3.3.2.3. - MECHANICAL STRESS LIMITATIONS
High mechanical stresses can depolarise a piezoelectric ceramic. The limit of the
applied stress is dependent on the type of ceramic material and duration of the applied
stress. For dynamic stress, impact application, the material behaves non-linearly for
pulse durations of a few milliseconds or more. When the pulse duration approaches a
microsecond the piezoelectric effects becomes linear due to the short application of time
compared to the relaxation time of the domains.

3.4. - Experimentally obtained PZT properties

The piezoelectric ceramic properties had to be acquired for accurate theoretical and
experimental comparison. For initial theoretical analysis, the manufacturer's
piezoelectric property values were used, see Appendix A1. However, for more accurate
theoretical analysis, the piezoelectric ceramic material was tested, due to the
manufacturer claims of a variation in material properties, dependant on age and storage.

3.4.1. - Youngs Modulus of Elasticity & Poissons Ratio
The Youngs Modulus of the piezoelectric ceramic was calculated by attaching strain
gauges to measure strain parallel and normal to an applied stress. The test material was
bonded, using Super Strength Araldite, into a manufactured steel jaw with sufficient
space for the material and Araldite, see fig. 3.1. Strain gauges were attached on both
sides of the test material to detect any offset in applied loads.

A high electrical field can lead to the occurrence of arcing, a localised high energy breakdown. Arcing
can partially depolarise the material and cause possible physical breakdown of the material.

3-66
Figure 3.1 - Piezoelectric ceramic clamping configuration for tensile testing

Figure 3.2 shows the variation of stress versus strain for the piezoelectric ceramic
material. Legend symbols 1 and 2 designate the strain parallel and normal to the load
respectively with F and R designating whether the strain gauge is attached to the front
or the rear of the test piece. The trends shown do not start at the origin due to
instrument limitations. The graph is approximately linear with non linearities due to
instrumental errors. From the slope of the graphs the Youngs Modulus and Poissons
ratio are 64.3 1GPa and 0.43 0.02. Morgan Matroc (2000b) give the Youngs
Modulus and poisons ratio as 64.4GPa and 0.45 respectively, within the values obtained
experimentally.
Figure 3.2 - Stress strain curve for piezoelectric ceramic PZT-5H.

Generally, one side of the piezoelectric ceramic produced less strain than the other,
suggesting that the tensile load was marginally offset. In addition, the side with less
stress also shows a slightly smaller Poissons ratio throughout testing, indicating a small
bending moment. However, the difference is small allowing an average Youngs
Modulus and Poissons Ratio to be used negating the bending motion/offset load.


3-67
3.4.2. - Piezoelectric Material Dielectric Constant
For theoretical deflection modelling the piezoelectric materials dielectric constants
were determined. This was achieved by attaching strain gauges to both sides of the
ceramic and applying an electrical field across the material. The piezoelectric charge
constant, d
31
, which is the strain developed along the piezoelectric ceramic length or
width per volt of applied field, was evaluated. This can vary due to age and storage,
according to the manufactures literature, between -210 to -274 x10
-12
mV
-1
. At the end
of testing the electric field is removed, the strain measured, and the measurement
repeated after one minute to allow for relaxation and dissipation of any remanent
electrical field/force.

V (V)
1F() 2F() 1R() 2R()
0 0 0 0 0
-20 9 10 10 10
-40 18 19 20 20
-60 27 28 30 30
-80 37 38 40 41
-100 47 48 52 52
-120 57 59 63 64
-140 68 70 75 76
-160 80 82 88 89
-180 91 94 102 103
-200 102 106 114 115
0 7 3 9 9
0(1min) 4 1 7 8
Table 3.1 Electrical strain parallel (1) and normal (2) to the applied load
for the front (F) and rear (R) of piezoelectric ceramic.

Referring to table 3.1, it can be seen that the strain induced differs from front to rear due
to the experimental set-up, where the material was simply supported at both ends and
bending was allowed to occur under the materials weight. The d
31
constant is
determined using the average strain from 1F and 1R and is calculated by

( )
V
t
d
R F
* 2
*
1 1
31
+
= [Eq. 3.1]
( ) ( ) { }
V m d / 10 * 250 10 * 5 . 0 *
200 * 2
10 * 7 102 9 114
12 3
6
31

+
=

3-68
Alternatively, the gradient of the graph (P
F
+P
R
)/2 versus the applied voltage can be
utilised to determine the d31 constant, see fig. 3.3.

Figure 3.3 - Stress strain curve for piezoelectric ceramic PZT-5H.

Rearranging equation 3.1, the parallel strain gradient can be used to determine the d
31

constant by

( )
31
1 1
*
2 d
t
V
R F
+
= [Eq. 3.2]

Therefore

( )
V m
V
t
d
R F
/ 10 * 260
10 * 9249 . 1
10 * 5 . 0
2
12
6
3
1 1
31

= =
+
=

A d
31
constant value of -250x10
-12
mV
-1
is used in the theoretical deflection modelling to
produce a conservative tip deflection calculation, which will give a smaller deflection
range for active control.

3-69
3.5. - Bender Actuators

The magnitude of piezoelectric actions is relatively small due to the small amounts of
producible strain. The theoretical maximum extension of a single element is
approximately 12.5m, which is 250 using 500V over a 0.5mm thick and 50mm long
piezoelectric ceramic. Amplification is often required and can be achieved by various
arrangements of the piezoelectric ceramic such as unimorphs, bimorphs and
piezoelectric ceramic stacks (also known as multimorphs or multi layer bender PZT
actuators).

3.5.1. - Unimorphs
In this thesis the term unimorph is used to describe a piezoelectric ceramic bonded to an
inert substrate, see fig. 3.4. Applying an electric field to a free, unbonded, piezoelectric
ceramic will lead to an induced strain, governed by equation 3.3.

pzt
t
V d *
31
= [Eq. 3.3]

Where the electric field is expressed as the applied voltage, V, across the piezoelectric
materials thickness, t
pzt
, and d
31
is the dielectric charge constant. If the piezoelectric
ceramic is bonded to an inert substrate, that is if the ceramic is not free to extend and be
constrained by the substrate, the amount of strain is reduced. The reduced strain will
exhibit a force in the ceramic in the direction of the free piezoelectric strain and
produce an equal and opposite force in the substrate. These two forces are equivalent to
a bending moment about the neutral axis and will deform the unimorph structure.

Figure 3.4 A Basic Unimorph


3-70
3.5.2. - Bimorphs
A bimorph is considered as two pieces of piezoelectric material bonded together, with
opposing poles, see fig 3.5. When subjected to an applied field the two piezoelectric
ceramics will produce equal opposable strains, which induce zero total strain with twice
the bending moment, compared to the unimorph. This can produce relatively large
deflections; typically double that of a unimorph with an inert substrate of similar
thickness and Youngs modulus.

Figure 3.5 A Basic Bimorph

3.5.3. - Multimorphs
For the multi-layer bender PZT actuators selected a tip deflection of up to 1mm was
available with a response time in the millisecond range, according to Morgan Matroc
(2000b), which is equivalent to a 1N force. Referring to figure 3.6, the multi-layer
PZTs are manufactured from 20Sm to 40Sm ceramic layers using a co-firing process.
Multi-layer bender PZTs offer several advantages over the classic bimorph PZTs, which
are manufactured by bonding together two ceramic plates, 0.1 to 1mm thick. Morgan
Matroc (2000b) claims that these advantages include; excellent humidity protection,
wide operating temperature range, vacuum compatibility and high stiffness. The main
advantage, however, is the drastically reduced operating voltage of only 60V
max
for the
1mm deflection range, which is an order of magnitude less than the bimorph &
unimorph voltage required to produce half the multimorph deflection.

Figure 3.6 A Multimorph


3-71
3.6. - Active Control Structural Configurations

3.6.1. - Unimorph configuration
The original unimorph configuration was a piezoelectric ceramic (0.5mm x 25mm x
50mm) simply bonded, using Super Strength Araldyte, to an aluminium plate (1.6mm
x 25mm x 100mm), see fig. 3.7. This would create a flush upper surface to the flow,
with the piezoelectric ceramic producing unimorph capability.

Figure 3.7 Original Unimorph Testing Configuration

However, this design was modified to minimise unimorph tip deflection under a given
pressure load, which is the dominant force in active control, see chapter 4. This was
achieved by embedding the piezoelectric ceramic into the aluminium substrate to
increase the unimorph thickness, hence second moment of area, over which the load is
exerted. Whereas, the original unimorph configuration would only suppress pressure
deflection over the substrate thickness. Also, embedding the piezoelectric ceramic
increases the bonding area available, which to a small extent improves the piezoelectric
actuator deflection. Furthermore, the adapted substrate tip reduces the unimorph
deflection required before mass transfer occurs, see fig. 3.8.

Figure 3.8 The Final Unimorph Configuration

3.6.2. - Bimorph/Multimorph Configuration
The incorporation of bimorphs and multimorphs into the design increases the
complexity of the configuration. As seen in figure 3.9, the basic concept was to bond
the bimorph/multimorph directly to the substrate material.

3-72
Figure 3.9 A Basic Bimorph Flap Configuration

This gives the advantages of increased tip deflection compared to unimorphs, whilst
having a flush surface to the flow providing that an electrical contact could be supplied
to the upper surface. The main disadvantage of this set-up is that the
bimorph/multimorph end would have to be large enough to allow adequate bonding to
the substrate. Epoxy adhesive is the main bonding system used with piezoelectric
ceramics as soldering or welding can easily destroy the piezoelectric properties. The
basic bimorph configuration was disregarded as feasible due to epoxy adhesive working
well in shear but poorly in tension. The bimorph end would have to be excessively
large to provide the required support to overcome the tensile forces that would be
induced by the cantilever structure under a pressure load.

Figure 3.10 Adapted Bimorph Flap Configuration

The adapted bimorph/multimorph configuration clamps the bimorph/multimorph in
the substrate structure, negating the adhesive in tension problem, see fig. 3.10. This
configuration has the disadvantage of a stepped surface to the flow, increasing surface
roughness. The size of the step is a factor of the substrate properties and can be
calculated from the stress produced by a uniform pressure (W) on a cantilever beam:

Moment = Pressure x Area x Length from mid pressure to clamped end

) 5 . 0 (
ef ef
L L bL W M = [Eq. 3.4]


3-73
where b, L and L
ef
are the bimorph width, length and the effective length respectively,
see fig. 3.10. For a pressure, width, total length and effective length of 17.64kPa
(experiments at ADFA), 25mm, 50mm and 45mm respectively, then the induced
moment would be 0.546Nm. The stress, , acting on the substrate step can be
expressed as:

2
6
2
s S
S
t b
M
I
t
M
= = [Eq. 3.5]

where b
s
, t
s
and I are the substrate width (equals actuator width), step height and the
second moment of area respectively. Limiting the maximum stress to half the yield
stress, which for low yield stress steel is 300MPa and rearranging in terms of t
s
gives:

mm
b
M
t
y S
S
935 . 0
10 * 150 * 025 . 0
546 . 0 * 6 6
6
= = =
[Eq. 3.6]

Using a pressure difference of 30kPa for the Cambridge experimental facility and a
safety factor of 1.5, for tunnel start up (45kPa), increases the step height to 2.23mm,
which is approximately a third of the boundary layer. It is believed that the existence of
a rearward facing step of this magnitude, whether upstream of or within the interaction
region will be detrimental. However, its effect could be minimised by filleting, see fig.
3.10. This would inhibit the bimorph deflection properties unless a low Youngs
Modulus material is used.

Another disadvantage of the adapted bimorph design is that the recess jaw in the
substrate would have to be large enough to prevent the piezoelectric ceramic surface
from being scratched during insertion, creating stress/arcing concentrations, and small
enough to provide sufficient support. The concept of having a manoeuvrable lower
surface of the substrate, to clamp the bimorph created excessive design complexities
and was disregarded. Moreover, if the bimorph is clamped too tightly there is a strong
possibility of damaging the fragile piezoelectric ceramic. It should be noted that the
step size required due to the shear across the step height is a magnitude less than the

3-74
step size required due to the stress as previously discussed and, therefore, has not been
considered here.

3.7. Alternate Active Control Configurations

3.7.1. - Alternate Flap Configuration
An alternative active flap configuration was considered using two downward deflecting
flaps, as shown in fig. 3.11. The advantages of this system are that it would minimise
the control region roughness by reducing the flap deflection into the boundary layer.
Also, if the material could be secured in the middle it would allow the use of one piece
of piezoelectric ceramic, reducing power supply requirements. However, the
disadvantages of this system far outweigh the advantages. The upstream flap is not
desirable; if the flap deflected into the freestream due to the aeroelastic behaviour and
was caught in the freestream it could plastically deform under an increased pressure
load, if it was not destroyed. Furthermore, if the upstream flap deflected upwards due
to the pressure difference acting across the flap then the boundary layer would be
scooped into the plenum chamber, cancelling any upstream mass transfer available for
SBLI control. Moreover, as will be discussed, the producible piezoelectric deflection
range is marginally able to overcome the acting pressure deflection across the
piezoelectric flap, let alone produce further downward deflection, see chapter 4.

Figure 3.11 An Alternative Active Flap Control Configuration.

3-75
3.7.2. - Alternate Flapless Configuration
The Flapless configuration, as shown in figure 3.12, consists of a piezoelectric ceramic
to move a grate underneath the surface of the active control section. The grate consists
of multiple slots which when moved line up with slots in the surface material, allowing
mass transfer across the surface of the control region.

This design was not explored as the maximum induced strain, using 1kV, of the
piezoelectric ceramic would be approximately 500, for a 0.5mm thick piezoelectric
ceramic. This would require a piezoelectric ceramic 300mm long to move the grate
0.15mm, the thinnest a slot could be machined. This length of piezoelectric material
could not logistically be incorporated into the wind tunnel facilities. Furthermore, the
presence of the grate on the surface would increase surface roughness similar to passive
control, which one is trying to eliminate.

Figure 3.12 An Alternative Active Control without Flaps.

Other configurations with translating piezoelectric ceramics were considered but all had
the same disadvantage of length of piezoelectric ceramic required to provide sufficient
movement/actuation.

3.8. - Multimorphs/Bimorphs Testing

Initially, multimorphs appeared to be the solution to the magnitude of deflection
required, advertising a 1mm travel (tip deflection) using 0 to 60 Volts. However, they
do not have the mechanical stiffness to withstand the pressure deflection and keep the
1mm travel with 0 to 60V. It was hoped that by increasing the limiting and input
voltages the multimorphs would be able to work under the design pressure.

3-76
3.8.1. - Multimorph Electrical configuration
The multimorph selected from the manufacturer is designed to be driven using a
variable input voltage. The input varies between 30V, the two limiting voltages, as
shown in fig. 3.13a. With a zero input voltage, the multimorph is at zero displacement.
At 30V the multimorph is at maximum negative displacement and conversely for a
30V input. An alternative electrical configuration was utilised, with limiting voltages of
0V and 60V with an input voltage that varied between the two limiting voltages, see fig.
3.13b. It was observed to give the same performance as the designed electrical
configuration with the ability to have a ground.

a)
b)
Figure 3.13 Multimorph Electrical Configurations a) Manufacturer Configuration,
Morgan Matroc) (2000) and b) Alternative #2 Configuration

3.8.2. - Multimorph Electrical Loading No Point Load Applied
Initial experiments were carried out by applying an electrical field across the
multimorph, using electrical configuration #2 (fig. 3.13b), with no load force. For a
given datum voltage (terminal 3 - fixed voltage), the applied input voltage (terminal 2 -
variable voltage) was varied between 0V (terminal 1 ground) and the datum voltage.
The tip deflection was measured using a micro-camera with scale. This electrical
configuration yields zero deflection when the applied voltage is half the datum voltage.

3-77
a) b)
c)
Figure 3.14 - Multimorph Performance
for Datum Voltages of
a) 80V, b) 90V and c) 100V

Figures 3.14 a to c show that the tip deflection, as expected, increases linearly with
applied voltage up to 15V over the fixed voltage datum. At approximately 15V above
the datum voltage the multimorph is observed to breakdown, which is believed to
induce permanent degradation, earlier referred to as aging, in the piezoelectric material.
However, before breakdown occurs a further 40% deflection is approximately obtained
using 15V above the datum value. To reduce the likelihood of multimorph
breakdown/aging, the applied voltage was restricted to equal the datum voltage. Then
an extra 40% theoretical deflection was assumed to be available over the deflection
experimentally obtained see fig. 3.15.

The multimorphs were rated up to a 60V applied voltage, with a travel of 1mm
(15%), which is confirmed in figure 3.15. It was observed that the multimorph
deflection properties are linear up to the rated 60V, and a travel of 1.15mm is obtained,
agreeing with the manufacturers claims. When the voltages are increased over the
manufacturers 60V, the piezoelectric properties are quasi-linear. It is believed that
above 60V the materials physical properties are restricting the piezoelectric deflection
properties. With an applied voltage of 190V the multimorph was observed to fail, in
this case arcing occurred. However, before breakdown occurred an experimental

3-78
deflection of 2.15mm was achieved, corresponding to a theoretical deflection of
3.01mm with the assumed extra 40%.

Figure 3.15 - Multimorph Performance with equal Applied and Datum Voltages.

3.8.3. - Multimorph Electrical Loading Point Load Applied
During bench testing, the multimorphs were tested under load, with a tip point load (P)
applied to produce similar deflection (X) and stresses as an equivalent distributed
pressure load (W). The tip point load was calculated using the tip deflections for a
cantilever beam under a point and pressure loads, see equations 3.7 and 3.8 from Beer
and Johnston (1992). These were equated to produce the equivalent tip point load to a
uniform pressure load, see equation 3.9.

EI
PL
P
3
3
= [Eq. 3.7]
EI
WL
W
8
4
= [Eq. 3.8]

8
3WL
P = [Eq. 3.9]

For a 17.64 kPa pressure load (the ADFA pressure difference) on a 40mm x 11mm
exposed area of a multimorph, an equivalent tip load of 2.91 N (0.3kg) is acquired. For

3-79
active control, the multimorph flap would be required to close, have zero deflection,
against the pressure difference. Therefore, the multimorph was tested to acquire the
applied voltage required to create a zero deflection for a given tip point load.

Referring to figure 3.16, it can be seen that the voltage, required to obtain a zero
deflection, varies quasi-linearly with point load up to 120g, after which breakdown was
observed. The combination of a 150V applied voltage and a 130g point load induced
failure - that is arcing and cracking. On inspection, it was observed that the crack was
initiated at the point of arcing. The crack initiated on the clamp line, the point of
maximum moment/stress, and progressed to a point of maximum shear on the clamping
line, at which point the crack propagated at 45 to the clamp line.

Figure 3.16 Applied Voltage Required
to Obtain Zero Deflection for Variable Point Loads

A 130g point load is equivalent to a uniform pressure load of 7.7 kPa, which is
insufficient to withstand the required steady pressure differential of 17.64kPa, let alone
tunnel start-up, where a higher pressure is experienced before the tunnel settles. The
author suggests that this is further reason why MART technology, as discussed in
section 1.2.5., has only been shown to overcome the aeroelastic deflection produced
from a 4.5kPa pressure load.

If a load bearing, high Youngs Modulus, material is used as the fillet the pressure or
point load could be increased with the increased stress carried by the fillet material.

3-80
However, a high Youngs modulus material would reduce the total amount of deflection
achievable and would create a quasi-unimorph structure.

3.8.4. - Bimorphs
Bimorphs were not experimented on in the same detail as unimorphs and multimorphs.
They have the same theoretical advantages and disadvantages of multimorphs, in that
they allow greater deflection but incur the logistical problem of incorporating them into
the design with a high Youngs Modulus load bearing fillet material required, creating a
quasi-unimorph structure.

3.8.5. - Multimorph/Bimorph Summary
Initially, it was believed that multimorphs would provide the magnitude of deflection
required for active control, boasting a 1mm travel using only 60 Volts. However, it
was shown that they do not have the mechanical stiffness to withstand the pressures
involved whilst retaining the deflection range.

It was observed that multimorphs deflect linearly with applied voltage up to 15V over
the datum voltage. With the tip deflection taken as 100%, when the applied voltage is
equal to the datum voltage, then with a further 15V an approximate 40% increase in tip
deflection was observed. Above 15V over the datum voltage induced breakdown in the
piezoelectric material occurs, which is expected to produce permanent degradation.

No diminished performance was observed with a 150V applied datum voltage.
However, applying a 130g point load under this voltage induced failure of the
multimorph with a crack being initiated by arcing at the point of maximum
moment/stress. It was observed that multimorphs are not capable of the desired
performance under the design pressure and, therefore, were discarded as an option for
active flap control.


3-81
3.9. - Unimorph testing

3.9.1. - Unimorph Substrate
Using tip deflection theory, a suitable substrate for the unimorph can be obtained that
produces the desired deflection characteristics, see chapter 4. The tip deflection varies
with substrate Youngs Modulus and thickness. The selection of substrate is a trade-off
between suppressing the pressure deflection, whilst keeping the unimorph electrically
responsive. This trade-off constricts the substrate to have a Youngs Modulus similar to
that of the piezoelectric ceramic, to enable the unimorph to have maximum range
against a pressure load.

Kaufman (1999) gives the tensile stress, yield stress and Youngs Modulus as 315MPa,
230MPa and 70GPa respectively for Al5083-H321. The Youngs Modulus value was
confirmed, by a three-point test of the aluminium, in order to improve theoretical
unimorph deflection predictions, see Appendix A3.

3.9.2. - Unimorph configuration
The unimorph configuration was examined with an aluminium substrate length of
53mm and a unimorph thickness of 1.92mm, with substrate, bond and piezoelectric
thicknesses of 1.4mm, 0.02mm and 0.5mm respectively. The unimorph was observed
to produce the following tip deflection with varying applied voltage:

-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
-500 -250 0 250 500 750 1000
Applied Voltage (V)
T
i
p
D
e
f
l
e
c
t
i
o
n
(
m
m
)
Figure 3.17. - Unimorph Tip Deflection variation with Applied Voltage.

3-82
From figure 3.17, it can be seen that the unimorph tip deflection varies linearly with
voltage and a 0.33mm deflection results with a 1000V applied voltage. It was observed
that the gradient of the graph (deflection per volt) is non-symmetrical about zero volts
and is greater with a negative voltage, producing negative piezoelectric ceramic strain
.
This would suggest the piezoelectric ceramic has a different Youngs Modulus or
dielectric charge constant, d
31
, in tension and compression. Using a negative applied
voltage to produce a negative strain, the piezoelectric ceramic was observed to
breakdown around -500V, whereas the piezoelectric ceramic withstood +1kV before
arcing/breakdown occurred. The manufacturer claims material voltage limit of
approximately 1000Vmm
-1
, which over the 0.5mm thickness is 500V, agreeing with the
negative strain voltage obtained.

Additionally, the manufacturer claimed different material properties with and without
an applied voltage. Therefore, the unimorph was further tested to determine its non-
symmetrical properties with and without an applied voltage. Point loads up to 875g,
equivalent to a 17.73kPa pressure load on a 25mm x 53mm flap, were applied 1mm
from the tip changing the unimorphs effective length to 52mm. Negative point loads
were applied to an inverted unimorph and axis system, see figs. 3.18 and 3.19.

-0.25
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
0.2
0.25
-900 -600 -300 0 300 600 900
Point Load (g)
T
i
p
D
e
f
l
e
c
t
i
o
n
(
m
m
)
10V
No Voltage
Linear (10V)
Linear (No Voltage)

Figure 3.18 Unimorph Tip Deflection variation with Point Load

A negative applied field produces a positive strain in the piezoelectric material. However, for safety the
substrate is earthed to 0V and a negative applied voltage produces a positive applied field and negative
strain.

3-83
Structurally, it was expected that the unimorph tip would deflect linearly with point
load, which was observed experimentally. It was observed that the deflection produced
with a 10V applied voltage was insufficient to noticeably change the inception point on
the graph. However, creating a closed circuit material changed the gradient of the graph
agreeing with the manufacturers claim of different Youngs Modulus. It was,
therefore, decided that only the closed circuit material properties were to be used for
theoretical analysis as this is the electrical configuration used during wind tunnel tests.

To experimentally validate the unimorph configuration, with a substrate thickness of
1.92mm, the unimorph deflection properties were observed whilst varying applied
voltage under different point loads. The Applied voltage was varied between 500 V to
reduce likelihood of piezoelectric ceramic degradation/breakdown and the point loads
were applied to simulate expected pressure loads.

-0.35
-0.25
-0.15
-0.05
0.05
0.15
0.25
0.35
-500 -400 -300 -200 -100 0 100 200 300 400 500
Applied Voltage (V)
T
i
p
D
e
f
l
e
c
t
i
o
n
(
m
m
)
-575g
-300g
-100g
0g
100g
300g
575g
Figure 3.19 Unimorph tip deflection for given applied voltages
and various point loads

From figure 3.19, it can be seen that tip deflection varies monotonically with applied
voltage. Moreover, for applied voltages above 150V the tip deflection varies linearly
with point loads. The graph is tri-linear as there is a slight decrease of the gradient
between 150V; either side of which the gradient increases. This is attributed to the

3-84
material limits under a combination of voltage and applied load. Furthermore, the
piezoelectric actuation was insufficient to produce zero deflection for unimorph, with
the 1.92mm thick substrate, against a 575g point load, which is equivalent to an 11.7kPa
pressure load.

3.9.3. - Unimorph Summary
Unimorph deflection varies linearly with applied voltage and point load as expected.
Under particular electrical/mechanical load conditions unimorphs are prone to
depolarisation and it is suspected that material degradation occurs with every break
down and depolarisation, referred to earlier as aging.

It was noted during testing that the unimorph properties are non-symmetrical, having
less piezoelectric actuator deflection for a negative applied voltage, compared to
positive applied voltage. However, it appears that the unimorphs have the desired
performance capabilities and were utilised for further wind tunnel testing & finite
element modelling.

3.10. Piezoelectric Actuator Summary

Piezoelectricity is a material property exhibited by certain classes of crystalline
materials. When mechanical pressure is applied, the crystalline structure produces an
impulse voltage proportional to the pressure. Moreover, when an electric field is
applied to the bipolar material, the crystalline structure produces positive or negative
strain proportional to the applied field. If the piezoelectric ceramic is attached to
another material, then a force and/or moment will be created changing the composite
materials shape and dimensions.

The magnitudes of piezoelectric actions are relatively small due to the small amounts of
producible strain. Amplification is often required and can be achieved by various
arrangements of the piezoelectric ceramic such as unimorphs, bimorphs and
piezoelectric ceramic stacks. A unimorph is a piezoelectric ceramic bonded to an
inactive substrate and applying an electric field to the piezoelectric ceramic will lead to

3-85
an induced deflection. The amount of deflection can be doubled using a bimorph,
which is two piezoelectric ceramics bonded together, with opposing poles.

Initially, the author expected that multimorphs would provide the magnitude of
deflection required for active control, boasting a 1mm tip deflection using only 0 to 60
Volts. However, it was shown that they do not have the mechanical stiffness to
withstand the pressures involved whilst retaining the deflection range.

Unimorphs, with an aluminium substrate, produce less deflection than bimorphs and
multimorphs. Moreover, they can withstand the pressure loads for SBLI control and
they have been shown, using an applied voltage of 500V, to overcome a point load
equivalent to an acting pressure of 11.7kPa. It was believed that unimorphs have the
desired performance capabilities and were further examined for active control.
However, a limitation of 500V was applied to minimise the possibility of material
breakdown or depolarisation.

3-86
Chapter 4. Unimorph Actuator Deflection

4-87
CHAPTER 4. - Unimorph Actuator Deflection

4.1. - Introduction

This chapter concentrates on the sizing of the unimorph actuator. Unimorph tip
deflection varies with the substrates Youngs Modulus and thickness. The selection of
substrate is a trade-off between suppressing the pressure deflection, whilst keeping the
unimorph piezoelectrically responsive. Active control of the shock boundary layer
interaction requires that the unimorphs are capable of zero deflection to be regarded as
capable of active control. The aim is to provide maximum tip deflection whilst
retaining zero deflection when the piezoelectric actuation is assisting and inhibiting the
aeroelastic deflection respectively.

Referring to figure 4.1, the final concept for the unimorph configuration was to have the
piezoelectric ceramic embedded within the surface material to allow a greater bonding
area and improved deflection properties, whilst retaining a flush upper surface, see
chapter 3. A lip was then incorporated into the design to minimise the deflection
required before mass transfer occurred.

Figure 4.1 - Modified Unimorph Flap configuration.

In this chapter two distinct pressure loads of 17.64kPa and 30kPa will be examined,
which represent the pressure difference across a unimorph for the swept normal SBLI
control (M
n
=1.3) at UNSW@ADFA and the unswept normal SBLI control (M=1.5) at
the University of Cambridge experimental arrangements respectively. A pressure
difference of 21.79kPa is also examined for the unswept normal SBLI control (M=1.3)
at the University of Cambridge. However, the unimorph is sized for the M=1.5 (30kPa)
study as the pressure deflection dominates the piezoelectric actuation. Section 4.5. has
been published in Structural Engineering & Mechanics: an international journal and is
an abridged version of the paper, see Appendix F1.

4-88
4.2. - Literature Survey

The basic assumptions for modelling a piezoelectric beam structure with classic
laminate theory were noted by Aldraihem and Khdeir (2000) as: the substrate layers
are made of isotropic or specially orthotropic materials; the bonding layers between two
adjacent layers is perfect; both faces of a piezoelectric layer are completely covered
with electrodes; and the actuation is performed only by the piezoelectric layer.

A review of smart structures by Chopra (2002) indicated that numerous theories for
prediction of flexural response of laminated plates exist that include in progression of
accuracy and computational cost; classical laminated plate theory (CLPT), first order
shear distribution theory (FSDT), higher order shear distribution theory (HSDT) and
layer wise shear distribution theory (LWSDT). CLPT is the most widely used which
assumes a linear variation of bending strain across thickness and the effects of
transverse shear are neglected. This is quite similar to the Bernoulli-Euler beam theory
and implies a perfect bonded condition between the actuators and substrates. This
theory is strictly applicable for thin plates (length/thickness > 30), where transverse
shear effects are negligible.

Gehring et al. (2000) used the classic laminate theory to determine piezoelectric
bimorph deflections and emphasised the importance of considering the effect of
clamping a cantilever across one end due to piezoelectric and Poisson ratio effects
altering the shape of the cantilever at the clamped end. That is clamping will inhibit the
natural curvature across the bimorph width resulting in a more flexible structure.

Soares et al. (1999) worked on finite element modelling of laminate plates, based on
higher order and first order displacement fields, showed significant errors with FSDT
models for highly anisotropic and or moderately thick piezoelectric laminated plates.
Chattopadhyay and Seeley (1997) agree with Soares et al. showing significant non
linearity for very thick laminates where laminate length to thickness ratio was four.
However, for moderately thick laminates, when the ratio is ten, the laminate is only
slightly non-linear with good agreement between CLPT and the HSDT.

4-89
Crawley and Luis (1987) use a linear strain distribution for the piezoelectric to predict
the effectiveness of the piezoelectric material in transmitting strain to the substructure, a
cantilever beam. They showed that both experimentally and analytically the
piezoelectric actuators are a viable concept for dynamic vibration and static shape
control of structures. They note that the stress-strain relationship for the piezoelectric is
similar to that of a thermoelastic material.

Wang and Noda (2001) studied single piezoelectric ceramic actuators covering the
length of the support structure, a cantilever beam (15mm thickness x 12mm width x
100mm length and Length to thickness ratio (L/t)=6.67). A thermally induced 0.48mm
deflection was generated and gradually eliminated with the application of increasing
voltages in the PZT layer. The stresses in the piezoelectric ceramic increased as higher
voltages were applied, reflecting the increase in piezoelectric ceramic strains. The
potential trade-off for minimising thermal deformation is increased stresses. The stress
distribution is very important to ensure that the structure does not crack and no plastic
deformation appears within the material. It was observed that due to the discontinuity
of the two materials, large stress discontinuities can induce delamination on the
interface of the two layers. It appears that minimising the stress distribution within the
piezoelectric ceramic is important to reduce material failure.

Eisenberger and Abramovich (1997) showed using CLPT that it was feasible for a 0.5m
long cantilever beam with piezoelectric ceramic layers on either side (L/t=125), under a
distributed lateral load, to have zero tip deflection, the unloaded case, with the
application of voltage fields to the piezoelectric ceramics. Also, Donthireddy and
Chandrashekhara (1996) use embedded piezoelectric layers to produce very similar
results (L/t=173). However, they did not achieve zero deflection due to the applied
voltage limitation of 250V, compared to Eisenberger and Abramovichs 500V.

Bruch Jr. et al. (2000) give the maximum allowable applied field for PZT-5H
piezoelectric ceramic material as approximately 500 - 1000Vmm
-1
. Their work
indicated that a single actuator (1mm x 10mm x 1000mm) was sufficient, on an inert

4-90
beam (5mm x 10mm x 1000mm) simply supported (L/t=167) with a 50N point load in
the middle, to return the beam to its unloaded condition with a 500V applied voltage.

Finite element modelling by Matthew et al. (2001) of piezoelectric actuators for active
flow control applications shows that a shear lag model gives better dynamic response
predictions than the perfect bond model. The deflection amplitudes were predicted
reasonably well for both FEM models but the benefits of the shear lag model may be
realised in situations with large excitation voltages, with larger deformations in the bond
layer and possible non-linearities in the shear-stress profile. The perfect bond model
assumes that an infinitely rigid bond exists, between the piezoelectric material and the
substrate, with all loads from the piezoelectric material fully transmitted to the
substrate. This creates an artificially stiff model and a shear lag model was utilised to
improve the model with a shear stress acting at the interface between the piezoelectric
material/bond and the piezoelectric material/substrate. However, for static control the
perfect bond model is a valid assumption.

The use of piezoelectric material to control structural deflections has been extensively
studied with numerous studies concentrating on the dynamic behaviour of piezoelectric
actuators in the design of aerospace structures. However, Donthireddy and
Chandrashekhara (1996) state that the use of piezoelectric material to statically control
the shape of structures has received less attention. Their work investigated the shape
control of laminated beams with surface bonded or embedded piezoelectric actuators
using a layer-wise theory. This allows a continuous displacement field through the
thickness and provides an adequate framework for the analysis of both thin and thick
laminated beams. The lateral strains, which are often neglected in one dimensional
beam analysis, are also incorporated. Donthireddy and Chandrashekhara note that
neglecting lateral strains in conventional beam models results in a significant error in
the analysis, concurred by Gehring et al. (2000). Lateral curvature will stiffen a
structure to resist longitudinal curvature.


4-91
The use of piezoelectric material has been incorporated in structures for sensors and
actuators for dynamic and static control. Hwang and Park (1993) studied the dynamics
of piezoelectric actuator stimulation on a cantilever beam. A piezoelectric film sensor
was incorporated into the design to sense the beam dynamics and then another
piezoelectric actuator was utilised to control and dampen large amplitude vibration.
Whereas, Vipperman and Clark (1996) used the piezoelectric component
simultaneously as both sensor and actuator, a sensoriactuator, for dynamic control. This
creates a complex system due to the sensor output is approximately two orders of
magnitude less than the input for similar actuation. Moreover, for static structure
control the sensor output is dynamic, tending to zero with time, and its accuracy is
sensitive to system noise.

The work of Lin and Hsu (1999) studied the optimal voltage distribution for static shape
control of smart beam plates laminated with sine sensors and actuators. It was shown
that a laminate beam of sensors and actuators can be constructed as a feasible way of
shape feedback control. It should be stressed that the sensors and actuators were
separate entities although they were connected in a closed loop. The ability to use the
same piezoelectric material for simultaneously sensing and actuating of static structures
has not been effectively achieved to date. Furthermore, Lin and Hsu showed that one or
two pairs of sensors and actuators were adequate to bring the structural deflection back
to the original form under various load conditions. However, more complex loads
required more sensors and reactors.

Mukherjee and Joshi (2002) examined the optimal piezoelectric distribution for shape
control of a long aluminium cantilever beam (3.5mm x 20mm x 100mm) subject to a
uniform pressure load. The objective was to find the actuator layout that gives the
original undeformed, zero load, shape under the uniform pressure load. It was shown
that it is difficult to control coupled bending-twisting deformation using isotropic
actuation. To control the deformation, anisotropic actuation is necessary and
unidirectional piezoelectric layers with cross-ply orientation were used. These are
termed piezo-fiber composites, which can be tailored to achieve desired stiffness and
actuation.

4-92
Yam and Yan (2002) investigated the optimal thickness and embedded depth of
piezoelectric actuators for multi piezoelectric-laminated structures. Their work
optimised the structure to create the largest possible actuating force for active vibration
control. This work highlighted that little work has been achieved on optimising piezo-
laminated structures by independently varying numerous material parameters.

Finally, Crawley (1994) reviewed piezoelectric material for the use in aerospace
structures and emphasised that actuation materials with three to ten times larger strain,
than commercially available piezoelectrics, must be developed to truly achieve the
desirable level of control for many structural applications. It is believed that this may
still be the case, as shown in this thesis.

4.3. - Classic Theory

Classic theory uses a combination of composite laminate plate theory (CLPT) and
simple beam bending theory. This theory was chosen due to its relative simplicity and
has been shown in previous works to be adequate for the unimorphs with a length to
thickness ratio greater than 10 - Chattopadhyay and Seeley (1997). Furthermore, it has
been shown to be truly applicable for thin plates with length to thickness ratios greater
than 30 when transverse shear is negligible - Chopra (2002). The final unimorphs have
L/t ratio of 20 and 31 for the swept and unswept normal SBLI control designs
respectively.

Figure 4.2 CLPT Model showing no deflection
or slope at the start of the unimorph


4-93
Classic theory models the unimorph as a beam with a constant cross section and no
substrate tip or support structure modelling, see fig. 4.2. However, it does allow lateral
curvature to freely occur over the entire unimorph length. The effective bending
stiffness of the unimorphs is calculated assuming an infinitely strong bond between the
piezoelectric ceramic and substrate. The effective stiffness is then used in classic beam
theory to determine the tip deflection of the unimorph, considered as a cantilever beam
under a uniform pressure load and an applied voltage to the piezoelectric ceramic layer.
The classic theory, for predicting deflection, was implemented using a spreadsheet to
give instantaneous results. It assumes that the total uniform deflection is a summation
of the deflections due to the pressure difference and the piezoelectric action.

The piezoelectric actuation is comparable to thermal actuation and a direct analogy is
derived in Appendix B1. The unimorph deflection obtained when the piezoelectric
ceramic has an applied voltage across it,
V
, is given by

2
*
2
L
y
V

= [Eq. 4.1]

where
y
and L are the induced curvature along the unimorph length due to the
piezoelectric actuation and the unimorph length respectively. Also, the unimorph
deflection under a given pressure load,
W
, is given by

( )
eff
W
EI
WbL
* 8
4
= [Eq. 4.2]

where W, b and (EI)
eff
are the pressure load, unimorph width and effective bending
stiffness respectively. The effective bending stiffness of the composite beam along its
length is given as
( )
11
d
b
EI
eff
= [Eq. 4.3]
where d
11
is the first element of the bending element of the compliance matrix that is the
inverse of the laminate stiffness matrix (see eq. B1.7 in Appendix B1).

4-94
4.4. - Finite Element Modelling (FEM)

ANSYS v6.0
*
was utilised as the FEM program for the design optimisation. A
parametric design language program was written to enable the analysis to be repeated
simply, quickly and to minimise manual work time, see Appendix C1. It allows the
parameters to be changed easily and the results recorded as table arrays to minimise
memory usage.

The program models the entire unimorph as well as the supporting structure. This could
not be performed with the classical theoretical analysis due to the non-uniform cross
section. Furthermore, FEM allows modelling of the solid substrate tip and analysis of
complex pressure loadings, which satisfy aerodynamic boundary conditions, whereas
classic theory can only apply simple approximate loads.

Figure 4.3 Unimorph Finite Element Model in ANSYS 6.0.

The final FEM model consists of 3144 Solid95 Brick elements with 1344 elements to
model the unimorph structure, see fig. 4.3. The grid resolution was chosen as a balance
between computational cost and sufficient accuracy, see appendix B2. Zero deflection
boundary conditions were placed on all edges of the supporting structure allowing the
unimorph to have non-zero slope and deflection at the start of the actuator. As a guide,
each surface optimisation figure in section 4.6. calculated with ANSYS took
approximately 110 hours of CPU time on an Enterprise 4500 with an Ultraspark 2
processor.

*
ANSYS is a proprietary name for a FEM programme by ANSYS, inc.

4-95
The piezoelectric ceramic dimensions are 0.5mmx25mmx50mm with Youngs Modulus
(E
P
), d
31
, d
33
and Poissons Ratio (
P
) of 64GPa, -250x10
-12
mV
-1
, 450x10
-12
m V
-1
and
0.45 respectively. The substrate support structure dimensions are 4mmx104mmx52mm
with the unimorph tip extending 2mm beyond the uniform cross section. The substrate
(support structure, unimorph and tip) is modelled as one structure with a Poissons
Ratio (
S
) of 0.3.
4.5. - 17.64kPa Results Line Optimisation (Couldrick et al. (2003))

4.5.1. - Theoretical deflection variation with substrate thickness for no load
The substrate thickness was altered to observe how the tip deflection would vary with a
500V applied voltage, 1000Vmm
-1
piezoelectric field, see fig. 4.4. This analysis used a
substrate Youngs Modulus of 70GPa (aluminium) and the substrate thickness is
normalised to the piezoelectric ceramic thickness (x=t
s
/t
pzt
). The results are compared to
experimentally obtained data for validation. It is seen that as the substrate thickness
increases, the tip deflection increases initially, reaches a peak and then reduces.
Initially, as the substrate constrains the piezoelectric strain the bending moment
increases. However, as the thickness is further increased the unimorphs second
moment of area is increased reducing the overall unimorph deflection. The theoretical
results show good agreement with experimental data.

0.00
0.10
0.20
0.30
0.40
0.50
0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 5.00
Normalised Substrate Thickness (x)
D
e
f
l
e
c
t
i
o
n
(
m
m
)
Classical Theory
FEM
Experimental data

Figure 4.4 Unimorph deflection vs. Normalised substrate thickness
due to 500V applied voltage (1000Vmm
-1
).


4-96
The two prediction theories show good correlation with each other whilst varying
substrate thickness with FEM predicting slightly higher tip deflections compared to the
classic theory. This is attributed to the support structure in FEM restricting the lateral
curvature at the start of the unimorph, making it less resistant to the piezoelectric
actuation.

4.5.2. - Theoretical deflection variation with substrate Youngs Modulus for no load
The substrates Youngs Modulus was varied to observe the tip deflection
characteristics with a 500V applied voltage, see fig. 4.5. This analysis used a substrate
thickness of 1.1mm (x=2.2) and the substrates Youngs Modulus is normalised to the
piezoelectric ceramics Youngs Modulus (y=E
s
/E
pzt
). A legend of suggested substrate
material has been included for the range of Youngs Modulus considered.

0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.0 1.0 2.0 3.0
Normalised Substrate Young's Modulus (y)
D
e
f
l
e
c
t
i
o
n
(
m
m
)
Classical Theory
FEM
Aluminium
and alloys
|----|
Steels
|---------|
Tantalum
and alloys
|--------------|
Copper alloys
|----------------|
Titanium .
Alloys .
|----------------|
Tin
alloys
|----|
GFRP
|----------------|

Figure 4.5 Deflection vs. Normalised Substrate Youngs Modulus
for a 500V applied voltage (1000Vmm
-1
).

It is seen that as the substrate thickness/stiffness increases, the tip deflection increases
initially, reaches a peak and then reduces. Note that at zero substrate stiffness there will
be no bending deflection as the piezoelectric ceramic is free to expand under the
induced strain. As the substrate thickness/stiffness increases from zero, its increasing
restraint on the in-plane expansion of the PZT produces an increasing out-of-plane
moment causing the composite beam to bend more and more. However, as the substrate
becomes stiffer due to the increasing unimorphs effective Youngs Modulus, it
becomes more effective in restraining the in-plane expansion of the beam with less
deflection, since the applied voltage is unaltered.

4-97
The two prediction theories show good correlation with each other whilst varying the
Youngs Modulus with FEM predicting slightly higher tip deflections compared to the
classic theory. This is attributed to the support structure in FEM reducing the lateral
curvature across the unimorph.

4.5.3. - 17.64kPa Constant Pressure Load Predictions
The two modelling systems were tested to predict the deflection under a uniform
pressure load of 17.64kPa, a value halfway between the theoretical upstream and
downstream pressures for the swept normal SBLI study at UNSW@ADFA, see figs.
4.6a & b. No experimental data could be obtained as this is a first order theoretical
prediction of the pressure gradient and is hard to physically simulate.

a)
-3.5
-3.0
-2.5
-2.0
-1.5
-1.0
-0.5
0.0
1 2 3 4 5
D
e
f
l
e
c
t
i
o
n
(
m
m
)
Classical Theory (-500V)
FEM (-500V)
Classical Theory (+500V)
FEM (+500V)

b)
-2.0
-1.8
-1.6
-1.4
-1.2
-1.0
-0.8
-0.6
-0.4
-0.2
0.0
0.3 0.8 1.3 1.8 2.3 2.8 3.3
D
e
f
l
e
c
t
i
o
n
(
m
m
)
Classical Theory (-500V)
FEM (-500V)
Classical Theory (+500V)
FEM(+500V)

Figure 4.6 - Tip deflection for a 17.64kPa uniform pressure load for varying
substrate: a) thickness (E
s
= 70GPa); and b) Youngs Modulus (t
s
= 1.1mm)


4-98
The trend between the classic theory and the FEM results are good for varying both
substrate thickness and Youngs Modulus. Classic theory predictions, for varying
substrate thickness, are within 0.1mm of the FEM predictions. Furthermore, the classic
theory predictions for varying substrate Youngs Modulus are within 0.17mm of the
FEM predictions. However, after the substrate normalised Youngs Modulus is greater
than 0.4 the error is reduced to within 0.1mm. When the normalised substrate Youngs
Modulus and thickness are 1.1 and 2.2 respectively, deflections ranges of -0.93mm to
-0.41mm and -0.99mm to -0.43mm are predicted for FEM and classic theory
respectively.

The differences between the two unimorph deflection prediction theories are suspected
to be due to the deviation from perfect built-in condition at the cantilevered end in the
FEM model, which realistically simulates the support structure around the unimorph
allowing non-zero slope and reducing lateral curvature at the start of the flap.

It is also seen that a negative applied voltage field assists deflection whilst a positive
applied voltage field inhibits deflection. An aluminium flap, with a Youngs modulus
of 70GPa (y=1.1) cannot be assumed to have zero deflection, i.e. minimum mass
transfer, within the substrate thickness and Youngs Modulus range considered.

4.5.4. - Quadratic Pressure load predictions
Figure 4.7 Quadratic pressure loading.


4-99
The main advantage of FEM is that complex pressure loadings can be applied to satisfy
the boundary conditions. The basic boundary condition is that there is no pressure load
at the flap edges. A parabolic load distribution was applied to the FEM model with the
numbering system used corresponding to the elements on the FEM model, see fig. 4.7.

A maximum pressure of 17.64kPa is used, which is the theoretical pressure difference
across the unimorph for the swept normal SBLI control experiments conducted at
UNSW@ADFA. The pressure drops off to zero at the flap edges producing a more
realistic loading and, hence, better deflection predictions. This loading cannot be
implemented using the classic beam theory in which the pressure can be varied only
along the length of the beam. No experimental data points could be obtained, during
wind tunnel runs, for unimorph deflection under this pressure loading as all measuring
techniques available are invasive and would alter the flow conditions being observed.
a)
-1.4
-1.2
-1
-0.8
-0.6
-0.4
-0.2
0
0.2
1 2 3 4 5
D
e
f
l
e
c
t
i
o
n
(
m
m
)
-500V
0V
500V

b)
-0.8
-0.7
-0.6
-0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3 0.8 1.3 1.8 2.3 2.8 3.3
D
e
f
l
e
c
t
i
o
n
(
m
m
)
-500V
0V
500V

Figure 4.8 FEM tip Deflection predictions for a quadratic load for
varying substrate: a) thickness; and b) Youngs Modulus.

4-100
The tip deflections of the flap under the quadratic pressure distribution are plotted
against the normalised substrate thickness (x) and the normalised substrate Youngs
Modulus (y), see figs. 4.8a & b. It is seen that as the substrate thickness and Youngs
Modulus are increased the tip deflection decreases. This is to be expected as the thicker
and stiffer the substrate the more constrained the unimorph will be.

If the unimorph has to be capable of zero deflection to provide active control then, for
with 500V, the substrate has to be at least 0.9mm thick (x=1.8) for a 70GPa Youngs
Modulus. Alternatively, a flap with a thickness of 1.1mm requires a minimum Youngs
Modulus of 50GPa (y=0.78) for complete theoretical closure. When the normalised
substrate Youngs Modulus and thickness are 1.1 and 2.2 respectively, a deflection
range of -0.49mm to -0.51mm is predicted.

4.5.5. - 17.64kPa Pressure Deflection Summary
The classic theory gives good correlation and accuracy with the FEM for the uniform
17.64kPa pressure load. FEM predicts greater piezoelectric actuation due to the support
structure is able to deflect. This creates a non-zero slope that inhibits lateral curvature at
the start of the unimorph. The main disadvantage of the FEM over classic theory is its
higher computational cost in terms of time and effort compared to the instant
spreadsheet ability of the classic theory.

The unimorph design can be easily optimised to provide active control, zero deflection
with 500V. The thinner or more compliant the substrate the more mass transfer or
deflection can occur. The zero deflection criterion for active control limits the substrate
to minimum thickness and stiffness. FEM gives the ability to analyse more complex
pressure loadings that reflect the real boundary conditions, which enables improved
predictions of minimum thickness and stiffness over the classic theory.

From the above study it was determined that the unimorph substrate should be
aluminium (y=1.1) with a thickness of 1.1mm (x=2.2). These values, using FEM, give
0.50mm for magnitude of maximum deflection whilst theoretically ensuring zero
deflection for a quadratic 17.64kPa pressure load. However, under a 17.64kPa uniform

4-101
pressure load the predicted deflection ranges, using 500V, are -0.99mm to -0.43mm
and -0.93mm to -0.33mm for CLPT and FEM respectively.

4.6. - 30kPa Results Surface Optimisation

It was shown in section 4.5. that the unimorph deflection produced by the 17.64kPa
pressure load is larger than the deflection induced via the piezoelectric actuation. It is,
therefore, desirable to have improved unimorph deflection predictions for the increased
30KPa pressure load that simulates the pressure load acting on the unimorph for the
normal SBLI control experiments conducted at Cambridge University.

The Line optimisation, performed in section 4.5., varied the substrate Youngs Modulus
and thickness independently. The following surface optimisation varies them
simultaneously, creating improved optimisation. The surface optimisation can be
considered as a 40 x 40 line optimisation.

The substrate thickness and Youngs Modulus were varied simultaneously to observe
the magnitude of tip deflection using a 500V applied piezoelectric field. This analysis
used substrate thickness normalised by the piezoelectric ceramic thickness (x=t
s
/t
pzt
) and
substrate Youngs Modulus normalised by the piezoelectric ceramics Youngs
Modulus (y=E
s
/E
pzt
) as variable parameters.

4.6.1. - Deflection for no load
Figures 4.9a & b show that maximum deflection, due to the piezoelectric actuation,
occurs for a normalised substrate thickness, x, and normalised Youngs Modulus, y, of
approximately 0.25 and 7 respectively. A thinner substrate does not constrain the
piezoelectric as much, producing less induced strain, a smaller moment arm and hence
generating less deflection. As the substrate thickens the amount of deflection is reduced
due to the substrates second moment of area is increasing.

The effect of substrate Youngs Modulus on unimorph deflection is more pronounced
for thinner substrates. Tip deflection increases with Youngs Modulus to a maximum as
the substrate increasingly constrains the piezoelectric ceramic. Initially, the effect of

4-102
substrate thickness on unimorph deflection is similar to the Youngs Modulus, that is
deflection increases with thickness to a maximum as the substrate constrains the
piezoelectric ceramic. After the peak deflection, an increase in thickness, and therefore
stiffness, reduces unimorph deflection. That is as the induced strain (substrate thickness
and Youngs Modulus) initially increases from zero the restraint on the in-plane
expansion of the PZT increases. This increases the out-of-plane moment, creating more
deflection of the composite beam. However, as the stiffness further increases, the beam
becomes more effective in restraining the in-plane expansion of the beam, causing less
deflection. Note that at zero substrate stiffness and Youngs Modulus there will be no
bending deflection as the piezoelectric ceramic is free to expand.
a)
b)
Figure 4.9 Contours of unimorph deflection (mm), with a 500V applied field,
for a) Classic Theory and b) FEM.

4-103
The FEM program does not allow the substrates thickness or Youngs Modulus to go
to zero. However, the two prediction methods show good correlation with each other
with FEM predicting larger maximum tip deflections than the classic theory, 0.85mm
and 0.78mm respectively. The differences are attributed to the support structure
inhibiting lateral curvature in the FEM model, producing a more flexible structure with
increased deflection.

4.6.2. - Deflection for 30kPa Constant Pressure Load
a)
b)
Figure 4.10 Contours of unimorph deflection (mm) using Classic Theory,
with a uniform pressure load, for applied voltages of
a) -500V open flaps and b) 500V closed flaps

4-104
The two prediction theories were applied to predict the deflection under a uniform
30kPa pressure load, a value halfway between the theoretical upstream and downstream
pressures for the unswept normal SBLI (M=1.5) study conducted at the University of
Cambridge, see figs. 4.10 & 4.11. A negative applied voltage assists flap deflection
with the 500V applied voltage labelled flaps open. Conversely, a positive voltage
inhibits tip deflection with the +500V applied voltage labelled flaps closed.

For both theories it is seen that maximum deflection is obtained with minimum
substrate thickness and Youngs Modulus. This would create a lower second moment
of area and stiffness to resist a given pressure deflection. Minimum deflection was
observed to occur with substrate Youngs Modulus and thickness ratios of 0.05 and
0.125 respectively. Whereas, maximum deflection was observed to occur at substrate
Youngs Modulus and thickness ratios of 7.5 and 7.25 respectively. The term
negative has been included in brackets next to the magnitude of minimum deflection
to indicate that the flaps cannot produce absolute zero deflection within the 500V
applied voltage limit.

The pressure deflection is at least an order of magnitude greater than the piezoelectric
deflection, see figs. 4.9, 4.10 & 4.11. The applied pressure dominates the unimorph
deflection, creating similar trends and results between the applied voltages. The two
theories are in good agreement with both theories predicting a quasi-closed unimorph
with a 1Um minimum deflection. The more critical area of interest is the 500V graph,
which will inhibit unimorph deflection, with the ability to induce zero deflection, that is
to close the unimorph and produce active control. The FEM predicts a slightly larger
region, combinations of substrate Youngs Modulus and thickness, where total
deflection is less than -0.05mm. The differences are small but are suspected to be due
to the deviation from perfect built-in condition at the cantilevered end in the FEM
model, which realistically simulates the support structure around the unimorph allowing
a non-zero slope at the beginning of the unimorph and inhibiting lateral curvature.


4-105
a)
b)
Figure 4.11 Contours of unimorph deflection (mm) using FEM,
with a uniform pressure load, for applied voltages of
a) -500V open flaps and b) 500V closed flaps

Referring to figures 4.10b and 4.11b, neither theory indicates that absolute zero
deflection is possible, suggesting that active control is not possible within these
substrate range considered with an applied voltage range of 500V for a 30kPa uniform
pressure load. When the normalised substrate thickness and Youngs Modulus are 4
(2mm) and 1.1 (70GPa) respectively, the predicted deflection ranges for CLPT and
FEM using 500V are -0.45mm to -0.18mm and -0.45mm to -0.16mm respectively.

4-106
4.6.3. - Quadratic Pressure load predictions
Figure 4.12 Quadratic pressure loading showing isobars

Again, the main advantage of FEM is that complex pressure loadings can be applied
that more accurately reflect the real conditions, with the basic boundary condition of no
pressure load at the unimorph edges. A quadratic load distribution was applied to the
FEM model, see fig. 4.12. A maximum pressure of 30kPa is used with the pressure
decreasing to zero at the unimorph edges. A quadratic (1-x
2
) is used to model the
pressure variation from the unimorph centre line and support structure/unimorph
boundary.

Figures 4.13a & b show the unimorph deflection under the quadratic pressure
distribution whilst simultaneously varying substrate thickness and Youngs Modulus. It
is seen that maximum deflection is obtained with minimum substrate thickness and
Youngs Modulus. As the unimorph becomes thicker and stiffer, the substrate
increasingly constrains the piezoelectric ceramic against pressure and piezoelectric
deflection, reducing total unimorph tip deflection. That is, as the normalised substrate
thickness and Youngs Modulus ratio are increased, the tip deflection decreases,
approximately inversely to the ratios. This is to be expected as the thicker and stiffer
the substrate, the more constrained the unimorph will be.


4-107
a)
b)
Figure 4.13 Contours of unimorph deflection (mm) using FEM, with a quadratic
pressure load, for applied voltages of a) -500V flaps open and b) 500V closed flaps

The unimorph deflection, for a quadratic load is approximately a third of the unimorph
deflection with a uniform pressure load of 30kPa, see figs. 4.11a & b. This is expected,
as analytically for a cantilever beam, the moment generated from the uniform pressure
load is three times greater than the moment from the quadratic pressure load. The
unimorph deflection is not dominated as much by the pressure deflection, resulting in a
noticeable difference between the trends and results of the two applied voltages.
Moreover, the reduced pressure effect has produced a region, range of substrate

4-108
Youngs Modulus and thickness, where zero deflection with 500V is theoretically
possible.

From figures 4.13a & b, it is seen that when the normalised substrate thickness and
Youngs Modulus are 4 and 1.1 respectively the total unimorph tip deflection is
0.24mm and +0.04mm for applied voltages of -500V and +500V respectively.
Moreover, at these substrate conditions active control, where zero deflection is required,
is feasible within the 500V applied voltage range.

It is predicted that if the unimorph has to be capable of zero deflection to provide active
control then for a 500V limit an aluminium substrate thickness has to be at least
1.25mm (x=2.5). The final unimorph design used a substrate thickness and Youngs
Modulus of 2mm (x=4) and 70Gpa (y=1.1, aluminium) respectively. These values give
a 0.24mm magnitude of maximum tip deflection, whilst theoretically ensuring closure,
and can be easily manufactured in aluminium without defects.

4.6.4. - Classic Theory for Quadratic pressure loads
Complex loads cannot be easily applied with the classic theory, which only allows a
linear variation of pressure along the unimorph length. However, classic theory can be
used to predict the deflection to a first approximation of a quadratic load. A uniform
pressure load is applied which generates the same pressure moment on the unimorph as
the quadratic load. This is calculated by integrating the pressure, multiplied by the
distance from the unimorph end, over the entire surface. Equating this to the moment
caused by a uniform pressure load gives a uniform pressure load that is a third of the
maximum given quadratic pressure load. This is re-enforced by the quadratic pressure
deflections being approximately a third of the deflection with a uniform pressure load.

The result, see figs. 4.14a & b, is a prediction to the same order of magnitude and trends
as the FEM, see figs. 4.13a & b. With a normalised substrate thickness and Youngs
Modulus of 4 and 1.1 respectively the deflection ranges from 0.24mm to +0.03mm,
with an applied voltage range of 500V, compared to the FEM prediction of -0.24mm
to +0.04mm. The main difference is that FEM predicts a larger region where zero

4-109
deflection is viable, suggesting a larger combination of feasible Youngs Modulus and
thickness.

a)
b)
Figure 4.14 - Unimorph deflection (mm), using Classic Theory for a quadratic load,
whilst varying normalised substrate thickness and Youngs Modulus for
applied fields of a) -500V open flaps and b) 500V closed flaps

Classic theory predicts that if the unimorph has to be capable of zero deflection to
provide active control then for a 500V limit an aluminium substrate thickness has to be
at least 1.5mm (x=3), compared to the FEM prediction of 1.25mm. The final unimorph

4-110
design used a substrate thickness and Youngs Modulus of 2mm (x=4) and 70Gpa
(y=1.1, aluminium) respectively. These values give a 0.24mm magnitude of maximum
tip deflection, whilst theoretically ensuring closure, and can be easily manufactured in
aluminium without defects. Whilst FEM is presumed to more accurately predict a
0.24mm magnitude of maximum deflection at these substrate values, it is believed that
the CLPT gives a very good prediction at considerably less computational cost. For this
reason it is believed that CLPT can be utilised as a very good first order deflection
prediction method for assisting (-500V) and passive (0V) pressure deflections. A
slightly degraded prediction was produced, using CLPT, whilst inhibiting pressure
deflection (+500V), however, it was shown in these few examples to be adequate as a
first approximation.

4.6.5. - 30kPa Pressure Load Summary
The classic theory gives good correlation and accuracy with the FEM for the 30kPa
pressure load. The main difference between the two theories is that the FEM models
the support structure, which allows a non-zero slope and also restricts lateral curvature.
The main disadvantage of the FEM over classic theory is its higher computational cost
in terms of time and effort compared to the instant spreadsheet of the classic theory.

It is believed that the classic theory can be utilised as a very good first order deflection
prediction method for assisting (-500V) and passive (0V) pressure deflections with a
slightly degraded prediction whilst inhibiting pressure deflection (+500V). However,
classic theory was shown in these few examples to be adequate when the predicted
deflection produced positive tip deflection. That is when it completely negated the
unimorph deflection due to the pressure load.

From the above study it was determined that the unimorph substrate should, again, be
aluminium (y=1.1) with an increased thickness of 2mm (x=4) to minimise the increased
dominant pressure load. These values give 0.24mm with both theories for magnitude of
maximum deflection whilst theoretically ensuring zero deflection for a quadratic 30kPa
pressure load. However, under a 30kPa uniform pressure load the predicted deflection
ranges, using 500V, are -0.45mm to -0.18mm and -0.45mm to -0.16mm for CLPT and
FEM respectively.

4-111
4.7. Improved FEM Prediction

The final FEM grid resolution, used in section 4.4., 4.5. and 4.6., was chosen on a trade
off of tip deflection accuracy against computational cost. Each surface optimisation
required 1600 individual solutions to be performed. It was seen in the grid resolution
study that improved accuracy could be obtained with a 20x20x5 grid, however, it took
400 times longer to solve, see appendix B2. This grid resolution was therefore only
utilised for the final unimorph structure to provide a more accurate prediction and
produced the following results, see table 4.1. It should be noted that the deflection for
the normal shock experiments at M=1.3 with an acting pressure of 21.79kPa was not
performed with the final grid resolution. This was because the unswept normal SBLI
arrangement was sized for the M=1.5 experiments.

Voltage CLPT
(mm)
Final Grid
Resolution (mm)
20x20x5 Grid
Resolution (mm)
Swept Normal Shock Control M
n
=1.3 (17.64kPa)
500V (closed flaps) -0.427/0.045 -0.411/0.051 -0.394/0.092
-500V (open flaps) -0.989/-0.517 -0.929/-0.489 -0.992/-0.505
Unswept Normal Shock Control M=1.3 (21.79kPa)
500V (closed flaps) -0.088/0.062 N/A -0.097/0.069
-500V (open flaps) -0.363/-0.213 N/A -0.382/-0.216
Unswept Normal Shock Control M=1.5 (30KPa)
500V (closed flaps) -0.173/0.034 -0.160/0.044 -0.188/0.042
-500V (open flaps) -0.448/-0.241 -0.445/-0.241 -0.473/-0.242
Table 4.1 Tip deflection predictions for CLPT, Final FEM Grid Resolution and
FEM with Increased 20x20x5 Grid Resolution for Uniform/Quadratic Pressure Load

From table 4.1 the predicted unimorph tip deflection ranges, for the swept normal SBLI
study at UNWS@ADFA (M
n
=1.3) with a 1.1mm thick aluminium substrate, with a
uniform and quadratic pressure load are -0.99mm to -0.39mm and -0.51mm to
+0.09mm respectively.


4-112
From table 4.1 the predicted unimorph tip deflection ranges, for the unswept normal
SBLI study at the University of Cambridge (M=1.3) with a 2mm thick aluminium
substrate, with a uniform and quadratic pressure load are -0.38mm to -0.10mm and
-0.22m to +0.07mm respectively.

From table 4.1 the predicted unimorph tip deflection ranges, for the unswept normal
SBLI study at the University of Cambridge (M=1.5) with a 2mm thick aluminium
substrate, with a uniform and quadratic pressure load are -0.47mm to -0.19mm and
-0.24m to +0.04mm respectively.

4.8. - Unimorph Tip Deflection Contour

It is assumed in the above analysis that the tip deflection, in the Z direction, is constant
across the unimorph span and, therefore, only the unimorph centreline deflection is
considered. FEM, using ANSYS, has the ability to predict the total unimorph and
support structure deflection to be observed in single or combined modes.

It is observed, from figure 4.15a that the assumption of constant deflection is valid
when the unimorph is subject to a -500V to piezoelectrically assist the deflection under
a 17.64kPa pressure load. However, when a +500V is applied to inhibit the pressure
deflection, see fig. 4.15b, a degree of contouring can be observed.


4-113
a)
b)
Figure 4.15 Top Profile of the unimorph deflection, Z axis,
with a uniform 17.64kPa pressure for applied fields of a) -500V and b) +500V

It is observed from the top profile, fig. 4.15b, and the side profile, fig. 4.16b, that the tip
deflection at the unimorph sides varies by than less than 5% of the centreline deflection
with an applied 500V. This equates to 0.05mm or 1% of the oncoming boundary
thickness suggesting it will have minimal effect on the SBLI control due to surface

4-114
contouring along the span. However, it is believed that under certain conditions the
unimorph could shut (have zero deflection) at the sides whilst retaining an amount of
deflection along the centreline enabling mass transfer to occur, producing a degree of
SBLI control.

a)
b)
Figure 4.16 Exaggerated Side profile of the unimorph deflection, Z axis,
with a uniform 17.64kPa pressure for applied fields of a) -500V and b) +500V

4-115
4.9. - Conclusion

This chapter examines two distinct pressure loads of 17.64kPa and 30kPa, which
represent the pressure difference across a unimorph for the swept normal shock SBLI
with M
n
=1.3 (UNSW@ADFA) and the unswept normal shock SBLI M=1.5 (University
of Cambridge) experimental arrangements respectively.

The classic theory, using CLPT, gives good correlation and accuracy with the FEM.
The main difference between the two theories is that the FEM models the support
structure, which allows a non-zero slope and also restricts lateral curvature. The main
disadvantage of the FEM over classic theory is its significantly higher computational
cost in terms of time and effort compared to the instant spreadsheet ability of the
classic theory. However, the present investigation shows that CLPT can be utilised as a
very good first order deflection prediction method.

Unimorph design, substrate Youngs Modulus and thickness, can be easily optimised to
provide active control. A thinner or more compliant substrate will allow more mass
transfer or deflection to occur. Zero deflection for active control limits the substrate to
minimum thickness and stiffness. FEM gives the ability to analyse more complex
pressure loadings that reflect the real boundary conditions. This enables improved FEM
predictions of minimum thickness and stiffness over the classic theory.

It was determined that the unimorph substrate should be aluminium (y=1.1) and have a
thickness of 1.1mm (x=2.2) for the swept normal SBLI study at UNWS@ADFA
(M
n
=1.3) with a quadratic 17.64kPa pressure load. These values, using CLPT and
FEM, give 0.52mm and 0.51mm respectively for magnitude of maximum deflection
whilst theoretically ensuring zero deflection. However, under a 17.64kPa uniform
pressure load the predicted deflection ranges, using 500V, are -0.99mm to -0.43mm
and -0.99mm to -0.39mm for CLPT and FEM respectively.

For the unswept normal SBLI study at the University of Cambridge (M=1.5), it was
determined that the unimorph substrate should, again, be aluminium (y=1.1) and have
an increased thickness of 2mm (x=4) to minimise the increased dominant pressure load.

4-116
These values give 0.24mm with both theories for magnitude of maximum deflection
whilst theoretically ensuring zero deflection against a quadratic 30kPa pressure load.
However, under a 30kPa uniform pressure load the predicted deflection ranges, using
500V, are -0.45mm to -0.18mm and -0.47mm to -0.19mm for CLPT and FEM
respectively.

The unswept normal SBLI Control study at the university of Cambridge with an
upstream Mach number of 1.3 used the same control plate as the M=1.5 study. CLPT
and FEM predicts the 2mm thick aluminium substrate give 0.21mm and 0.22mm
respectively for magnitude of maximum deflection whilst theoretically ensuring zero
deflection against a quadratic 21.79kPa pressure load. However, under a 21.79kPa
uniform pressure load the predicted deflection ranges, using 500V, are -0.36mm to
-0.09mm and -0.38mm to -0.10mm for CLPT and FEM respectively.

Chapter 5. Swept Normal SBLI Control

5-117
CHAPTER 5. Swept Normal SBLI Control

5.1. - Introduction
This chapter examines the swept normal shock wave/turbulent boundary layer
interaction using two similar methods of control. Firstly, unimorph control is examined
where the degree of flap deflection is controlled using piezoelectric actuators. This
comprises a study of discrete pressure data and also the use of pressure sensitive paints
and is discussed in section 5.2 to 5.4. Appendices F2 and F3 are the published versions
of these studies.

The second part of this chapter, section 5.5., examines the use of dimensionally
identical flaps that mechanically set the level of deflection to control the swept normal
SBLI. It was proposed that optimal control could be provided with a deflection between
1mm and 3mm. However, these levels of deflection cannot at present be achieved using
piezoelectric actuation, as elucidated in chapter 4.

5.2. - Uncontrolled Swept SBLI

5.2.1. Schlieren
Referring to figure 5.1, the flow is from left to right, with the main shock at
approximately 41 to the flow direction. With an 11 wedge and a shock normal Mach
number of 1.3, the flow would be incipiently separated according to Korkegi (1973).
Also, according to Kubota and Stollery (1982), for a freestream Mach number of 2,
incipient separation occurs for a wedge angle between 10 and 12 so that the present
flow is on the verge of separation. It is observed that this is, indeed, the case as the
main shock is smeared at the tunnel wall as indicated by the darker region upstream of
the main shock. The dashed white line marks the approximate start of this darker region
which indicates the foot of the leading leg of the uncontrolled lambda structure (ULS),
in agreement with the works of Alvi and Settles (1992), Squire (1996). The foot of the
lambda structure is seen to initially diverge from the swept shock angle by
approximately 4.5, which also agrees with the trend of Alvi & Settles work on the
swept SBLI without control. This divergence angle is observed to increase at the top of

5-118
the figure as the swept shock interacts with both wall and ceiling boundary layers, as
pointed out by Green (1969).

Figure 5.1 Schlieren photograph of the uncontrolled swept SBLI.

A low pressure region, seen as a dark purple region, exists above the wedge as the flow
is expanded over the top of the wedge. Furthermore, two weak shocks (shown) can be
observed that originate far upstream of the test section. They are believed to be
relatively weak as there is no apparent hue change either side of them. It is assumed
that the expansion fan and the weak shocks will not significantly influence the discrete
pressure data or the PSP results.

5.2.2. - Oil Flow visualisation
Tests were conducted on the interaction region using oil flow visualisations, see figures
5.2a to c. A water-soluble pen was used to trace and enhance the surface streamlines,
allowing the main surface flow features to be easily observed, see fig. 5.2c. The flow is
from left to right, with the wedge shock parallel to the convergence line. It can be seen
that the shock is at approximately 41, with an inviscid flow deflection angle of 11.

5-119
Figures 5.2a to c indicate that the flow is on the verge of separation, consistent with the
schlieren data. This is verified by the fact that the surface streamlines upstream of the
shock converge to become parallel to the wedge shock, creating the convergence line.
A region of separation is observed as the oblique shock, generated by the wedge,
interacts with both the sidewall and ceiling boundary layers, as seen in figure 5.2a. It is
understood that this region is sufficiently removed from the unimorph flaps and
pressure ports to be significant in having any influence on the data obtained.

a)
b)

5-120
c)
Figure 5.2 Oil Flow visualisation of the uncontrolled swept SBLI
a) tunnel sidewall, b) control surface and c) pen enhanced oil flow.

5.3. - Piezoelectric Flap Actuator Controlled SBLI

Figure 5.3 shows the flow features of swept normal SBLI control using unimorphs. Oil
Flow visualisation was only achieved with 0V for safety reasons. A 500V applied
field could not be applied as it is possible that the oil would act as a conductor and earth
the power supply through the tunnel and instrumentation. It can be concluded, from the
parallel streamlines down stream of the convergence line, that the flow is separated and
a divergence line can also be observed, which is a characteristic of re-attaching flow.
This divergence line is influenced by the corner vortex, resulting from the flow
separating from the wedge and re-attaching to the sidewall. S1, S2, and S3 indicate
separation lines - without control, with flap control and with flap and slot control
respectively. S2 and S3 are confirmed from pressure distribution readings, as will be
discussed in section 5.3.2. The wedge generated shock, as shown in fig. 5.1, is found to
occur at approximately the same angle and position as the convergence line and has
been omitted from the above figure to prevent over crowding.


5-121
Figure 5.3 - Pen enhanced oil flow visualisation of the swept SBLI
with a 0V unimorph control.

The main shock generated by the wedge (see fig. 5.4a) will occur parallel to the
convergence line and has an angle, |
s
, of 41 to the flow direction. The S1 shock is the
leading leg of the lambda foot without control and indicates the initiation of separation.
The S1 angle, |
fs
, is 45.5, which gives a A
1
[= (|
fs
- |
s
)] of 4.5. This agrees with the
trend of Alvi and Settles (1992) work of swept interaction without control, as shown in
figure 5.4b.

The S2 and S3 shock foot angles indicate the leading leg shock at the start of the
interaction control with unimorphs and combined unimorph slots/flaps respectively. S2
and S3 have shock angles, |
fs
, of 57.5 and 71 respectively, which correspond to A
1
of
16.5 and 30 respectively
*
, see fig. 5.4b. These signify leading leg shocks that are
anchored further away from the main shock, indicating a weaker front shock and
consequent reduction in wave drag.

*
|
fs
and |
s
are measured to an accuracy of 1 and 0.5 respectively. This gives an accuracy of 1.5
for A
1
.

5-122
a)
b)
Figure 5.4 a) Co-ordinate system and b) Variation of A
1
[= (|
fs
- |
s
)]
for front shock with normal Mach number (Accuracy 1.5), after Squire (1996).

5.3.2. - Surface Pressures
The surface pressure ratios were recorded for actuator control at various actuator
deflections (applied voltages), normal to the original shock position and along the end
of the upstream unimorph, (figs. 5.5a). Figures 5.5b and 5.6 show the pressure
distribution for swept normal SBLI control with - flap control using smart flaps with
the trend shown for; increasing voltage, the theoretical inviscid pressure rise and slot
0
5
10
15
20
25
30
35
40
1.3 1.5 1.7 1.9 2.1 2.3 2.5
M
n
A
1
(
o
)
Without Control
Unimorph control
Unimorph control with slots
Alvi & Settles

5-123
control (to indicate the passive control with zero flap deflection). The discrete pressure
data has an accuracy of 0.003 in p/P
0
.
Figure 5.5b shows the pressure rise normal to the original shock, from pressure port 8
through to 26, up to the plateau as observed by Babinsky (1999). The rise is gradual
compared to the pressure rise across an inviscid shock. It is believed that increasing the
unimorph deflection increases the upstream pressure by allowing more mass injection
from the plenum. It also has the effect of decreasing the plateau pressure, which
suggests a weaker shock. That is, increased deflection gives more control and implies
lower drag.

Figure 5.5 a) Control Plate showing Pressure Port and Unimorph Layout
and b) the Surface Pressure distribution along the normal to the wedge shock.


5-124
Figure 5.6 shows the pressure rise along the end of the upstream unimorph; from
pressure port 8 through to 16 (see fig. 5.5a). A maximum and minimum can be
observed, due to the rise in pressure across the leading leg of the lambda shock and the
thickening of the boundary layer due to the mass injection respectively. The pressure
rises at the end as the overall pressure has to rise across the shock.

Figure 5.6 - Surface Pressure distribution
along the back edge of the upstream unimorph.

5.3.2.1. BOUNDING STREAMLINE L1
The surface pressure ratios were recorded for actuator control at various unimorph
deflections along the two bounding streamlines L1 and L3 as shown in fig. 5.7a. Figure
5.7b shows the corresponding pressure distribution for various conditions of flap
control. In the figure slot control refers to passive control with zero flap deflection. On
the abscissa, streamwise distances are measured from the wedge shock position with
negative distances implying upstream. The discrete pressure data, as shown in figures
5.7b & 5.8, has an accuracy of 0.003 in p/P
0
.
Referring to figure 5.7b, the initial pressure rise without control from -4<X/o
0
<-1.5 is
due to the front leg of the bifurcated lambda-shock structure. This is followed by a
plateau region up to X/o
0
=2, which rises downstream towards the inviscid value.


5-125
a)
b)
Figure 5.7 - a) Bounding Streamlines L1 and L3 and
b) the surface pressures along L1 for swept SBLI

The pressure rise with full flap control shows the following: a local pressure maximum
(A) which indicates the leading leg of the lambda shock; a local pressure minimum (B)
which is a result of expansion waves and boundary layer thickening at the end of the
upstream unimorph; and a small pressure plateau (C) (-1.5< X/o
0
<1.5). This is very
similar to the results of Babinskys slot control with suction. These attributes indicate
that the shock wave not only has a lambda-like structure but also has spread extending
the shock foot.

The slot control pressure rise is similar to the no control pressure rise, indicating that
minimal interaction control has been produced. However, there are subtle differences
with a more gradual pressure rise from the lower minimum (denoted as B at X/o
0
=-4) to

5-126
the less prominent plateau (0< X/o
0
<1.5). This is attributed to the passive slots control
that initiates an interaction at a distance of seven oncoming boundary layer thicknesses
upstream of the inviscid shock position. However, the level of control is small due to
the relative size of the 0.2mm slots compared to the 7.57mm oncoming boundary
layer.

The upstream pressure ratio with slot control is greater (-10< X/o
0
<-7.5) than with full
flap control. The author speculates that full flap control has reduced mass injection
across the sides of the unimorph as the deflection allows less resistant injection across
the end. Furthermore, slot control does not produce sufficient mass injection to
bifurcate the foot of the shock and therefore no localised maximum is observed.
However, the mass injection does thicken the boundary layer and creates a localised
pressure minimum (X/o
0
=-4).

The increased mass injection across the end of the unimorph with full flap control
produces a more gradual pressure rise (-4< X/o
0
<-1.5) and a lower plateau pressure
(-1.5<X<1.5). It is thought that the increased mass transfer bleeds off more of the
higher pressure fluid downstream of the shock into the plenum chamber. The pressure
plateau without control is at approximately 70% of the total pressure rise. Whereas,
with flap control the plateau is at approximately 50% of the total pressure rise,
suggesting an extended region of interaction.

5.3.2.2. - BOUNDING STREAMLINE L3
Figure 5.8, showing the pressure rise along L3, allows the interaction to be observed for
an increased downstream control region and has similar characteristics to figure 5.7b.
The local maximum (A) is greater because the leading leg of the lambda shock is closer
to the main shock position. This represents a more nearly normal shock to the flow and
therefore an increased pressure rise. The local minimum (B) is again greater than for
L1. However, it is uncertain whether the magnitude of the difference between
maximum and minimum is less along L3 due to its closer proximity to the maximum
(A) and the plateau (C) or if the boundary layer has not thickened to the same degree.
For similar reasons it is uncertain why the plateau is at a lower pressure than the local
maximum at A. The pressure fluctuations downstream are thought to be due to the local

5-127
interaction between the trailing leg of the lambda shock and the sidewall boundary
layer.
Figure 5.8 - Surface Pressures for swept SBLI control along L3.

Figures 5.7b and 5.8 show that the final pressure ratios are less than the inviscid value.
The final downstream pressure ports are just outside of the interaction region and it is
expected than they are still being influenced by the interaction control.

The results indicate that a 500V applied electric voltage to the piezoelectric ceramic
produces zero deflection of the unimorph. However, interaction characteristics are still
observed due to the passive slot control. Furthermore, it is not certain that with a
500V applied voltage the deflection is exactly zero as precise deflection measurement
devices are unavailable.

5.3.2.3. LEADING LEG OF THE LAMBDA SHOCK
From figure 5.7b and 5.8 the pressure rise across the leading leg of the lambda shock
can be calculated as the pressure ratio of the local maximum divided by the pressure
ratio upstream of the local maximum (pressure port 5 and 11). In both cases the
pressure ratio across the leading leg is approximately 1.22, which equates to a leading
leg shock angle, |
ll
, of 33. This is an improvement over the Alvi and Settles (1992)
prediction of 45 without control, see fig. 5.9. The more acute the angle the weaker the
leading leg shock will be, leading to a smaller increase in entropy and hence smaller
wave drag.

5-128
Figure 5.9 Variation of |
ll
with Normal Mach number, Squire (1996).

5.3.3. - Drag Coefficient
Losses in total pressure, P
0
, correspond to increase in entropy and hence drag. It may be
seen that at each pressure tap the scanivalve measures a point drag (c
d
). By integrating
over the test length, the total drag coefficient can be determined. Nagamatsu et al.
(1985) present the following formula for determining the point drag coefficient, c
d
:

( )
( ) ( )
(
(
(
(
(
(
=

|
|
.
|
\
|
2
1
1
01
1
1
0
1
2
1
1
01
1
1
0
1
1
1
01
0
1
1
1
1
1
2 '
P
P
P
P
P
P
P
P
P
P
P
P
c
t t
t
t t
d
[Eq. 5.1]

From this relation, it can be calculated that the point drag is zero for all the pressure
ports before the shock, as P
0t
is equal to P
01
, which makes the last two terms in equation
5.1 have zero value. The point drag coefficients are integrated across the test section
and the following integral for the total drag is obtained:

}
=
l
d d
dx c
l
C
0
'
1
[Eq. 5.2]

where l is the interaction length. Along L1 l corresponds to the length along the
pressure ports from 1 to 34. Drag Coefficients were calculated for various unimorph
20
25
30
35
40
45
50
1.3 1.5 1.7 1.9 2.1 2.3 2.5
M
n
|
l
l
(
o
)
Unimorph control with slots -500V
'Without control' prediction
Alvi & Settles

5-129
deflections, as shown in fig. 5.10. A positive voltage implies reduced deflection,
whereas a negative voltage implies increased deflection. It was not possible to reach the
piezoelectric ceramic 1000V applied voltage limit due to arcing limitations observed
above 500V (see chapter 3).

A trend can be clearly observed from figure 5.10 and as expected, a positive applied
voltage, closing the flap, increases the drag coefficient. Conversely, a negative voltage,
assisting tip deflection, reduces the drag coefficient. A greater tip deflection will
produce more mass transfer, improving the control effectiveness to some degree.
Figure 5.10 - Drag Coefficient vs. Applied Voltage
for unimorph controlled swept SBLI.

The results show that a 1% reduction in drag coefficient is achieved for the control area
for the full extent of control (+500V to -500V). The author recognises that this is small
when applied to a wing, which is attributed to the small deflection range available.

At this stage in the research the author postulated that increased deflection would
produce improved drag reduction. This test section had a slot to control area of 1 %.
Chen et al. (1984) suggests that passive control using slots has an optimum porosity of
1 %, which should give a visible drag reduction. However, flap control does not exhibit
the exact characteristic of slot control as a compression corner is produced and the mass
transfer is angled with the flow compared to the normal inject/suction of slot control.

5-130
5.3.4. - Swept Normal SBLI Unimorph Control Summary
The surface pressure distributions show a local pressure maximum that indicates an
extended foot of the leading leg shock followed by a local pressure minimum that is a
result of boundary layer thickening. In addition to a pressure plateau, these features
indicate that the shock wave bifurcates to form a lambda-like structure and, therefore, a
reduced entropy rise.

The pressure ratios before the plateau are reduced for increased voltage, reduced
unimorph deflection, giving active control. Increased unimorph deflection leads to
greater mass transfer from the higher pressure plenum chamber. This is reversed for
downstream pressure ratios as increased mass transfer bleeds off the higher pressures to
the lower pressures in the plenum chamber. It is assumed that with an applied 500V
electric voltage to the piezoelectric ceramic zero deflection of the unimorph results.
Interaction characteristics are still observed for 500V control due mainly to mass
transfer that occurs through the surface slots, which exists at nominal zero deflection.

As expected, a positive applied voltage, closing the actuator flaps, increases the drag
coefficient, and a negative voltage, assisting tip deflection, reduces the drag coefficient.
A greater tip deflection will produce more mass transfer, improving the control
effectiveness to a certain degree. The fluctuations observed were the consequences of
the inherent noise in the instrumentation.

This discrete pressure data shows that unimorph flap actuators provide a degree of
beneficial control. The full flap control produces an extended lambda structure as
indicated by the local pressure maximum and the lower pressure plateau as compared to
the no control and slot control, zero unimorph deflection, cases. However, the results
would also suggest that unimorph actuators could not provide real active control as
the pressure distribution, using 500V, with zero unimorph deflection is different from
the no control distribution. This indicates that a form of control is still present when
voltage is applied and the control is not turned off. This is presumably due to some
leakage across the 0.2mm gap between the model skin and the flap, creating passive
slot control.


5-131
5.4. Pressure Sensitive Paint (PSP) Investigation of Swept SBLI Control

5.4.1. - PSP Technique
Intensity values, I, were for the images of I
ref
/I were obtained by taking a reference
image of the model at atmospheric pressure (wind off) and an image when the tunnel
was running (wind on). The pressure distribution was then determined using equation
2.1. The coefficients, k
1
, k
2
and k
3
, were determined using measured pressure data and
intensity ratios at numerous locations, see fig. 5.5a. The images were processed using
MATLAB by breaking the images into 576 x 720 matrices of intensities and then
averaging over 4x4 element (1mm x 1mm) tiles to attempt to reduce the noise in the
data. The reduction of noise is a trade off with high-resolution information, providing
an accuracy of 0.02 in p/P
0
. The positions of the unimorph flaps, main shock (MS)
and the front foot of the uncontrolled lambda structure (ULS), as calculated from figure
5.1, have been overlaid on the processed figures for ease of comparison, see figs. 5.11,
5.12, 5.14 & 5.15.

PSP was applied over the circular test surface (fig. 2.7) and the experiments were
conducted with the test surface at approximately 0C 2C. This is the flow adiabatic
wall temperature and is used to minimise any heat transfer between the flow and the
control plate, see section 2.1.7. As indicated by equation 2.1, the luminescence of the
paint is temperature dependent so any heat transfer will produce spurious results. The
luminescence is then recorded by a digital camera and post-processed.

Furthermore, it is believed that a crescent of approximately 10% of the test section area
in the bottom right hand section of the picture is indicating higher pressure than actual
due to the shadowing effects of the light incident angle to the test section, see fig. 2.7.
It is not precisely known to what extent this affects the results but it does not extend as
far as the downstream unimorph.

5.4.2. - Uncontrolled SBLI
Figure 5.11 is observed to closely agree with the schlieren for the location of the leading
leg of the uncontrolled lambda structure (ULS), see fig. 5.1. Furthermore, it seems to
be in close agreement with the discrete pressure data obtained, see figs. 5.7b & 5.8. It is

5-132
observed that the pressure upstream of the ULS is uniform, at 0.14 0.02, and increases
at 0.14 0.02 as it travels under the lambda structure. Downstream of the main shock
(MS) the pressure further rises and achieves approximate inviscid levels as it reaches
the end of the control plate. The smaller region of lower pressure (green region) that is
observed at the bottom right of the picture is generated by the flow expanding over the
wedge.

Figure 5.11 - Pressure distribution map of the uncontrolled swept SBLI using PSP.

5.4.3. - Closed Unimorph (500V)
It was theoretically calculated that zero deflection should be achievable with a 500V
applied voltage across the piezoelectric ceramic (see section 4.7). Figure 5.12 shows
the surface pressure distribution map using a 500V applied voltage. A distinct pressure
rise across the position of the ULS can be seen with minimal extra smearing compared
to the uncontrolled pressure distribution map, see fig. 5.11. This indicates that the
global flow field is approximately uncontrolled and the extra smearing observed,
especially over the downstream unimorph, is expected to be due to the limited localised
passive slot control.


5-133
A slight pressure drop towards the sides of the upstream unimorph (yellow tinge along
the sides) is believed to be due to localised expansion as a result of boundary layer
thickening from the passive control provided by the streamwise slots. Similar flow
features are seen with the slot control of Smith et al. (2003) and Smith et al. (2002).
This slot control effect can be considered to be the result of a 3D bubble forming just
upstream of the slot position and expanding with downstream distance (see fig. 5.13a).

Figures 5.12 - PSP Pressure distribution map of unimorph swept SBLI control
with 500V applied voltage (Closed).

The bubble of influence dissipates into the global flow field away from the slot
position. It is thought that the size of the bubble, located on top of the separation line, is
related to the size and strength of the control region, see fig. 5.13b. However, the
observed slot effect, using PSP, is minimal due to the 0.2mm width of the slots
compared to the 7.57mm thick oncoming boundary layer. Whereas, Smith et al. had
slot widths of the same order as the oncoming boundary layer thickness.


5-134
a)
b)
Figure 5.13 - a) 3D Bubbles of Influence of slot control of the SBLI
and b) the effect of slot strength, Smith et al. (2002).

5.4.4. - Uncontrolled Unimorph (0V)
Figure 5.14 shows the interaction with a 0V applied voltage across the piezoelectric
ceramic producing SBLI control with an uncontrolled inert unimorph. The aeroelastic
tip deflection achieved is a product of the pressure difference, acting on a cantilever
beam. It should be noted that there is increased noise due to the trade off with
resolution, as discussed in section 5.4.1.

The leading leg of the lambda structure is not seen to have significantly moved but a
higher degree of smearing can be observed, compared to the closed flaps (500V) case
which, therefore, would suggest a smaller increase in entropy. The smearing has spread
below the unimorph region as the larger lambda structure has an increased influence on
the neighbouring flow field, as can be identified by the spread of the green region.
However, in the top right hand section of the test surface the pressure rise is sharper as

5-135
the leading foot of the lambda structure is seen to diverge more, becoming increasingly
normal to the flow, possibly due to a higher downstream total pressure.
with a 0V applied voltage. (Open actuator inactive)

Furthermore, the low-pressure region on the upstream unimorph flap has increased in
area, with a decreased pressure, as the slot control bubbles of influence have grown
due to the increase in mass injection (slot control) as a result of increased unimorph
deflection.

5.4.5. - Fully Deflected Unimorph (-500V)
An increase in unimorph deflection should be theoretically achieved with a negative
applied voltage. A 500V voltage was applied to the piezoelectric ceramics to observe
the interaction, as shown in figure 5.15.

The leading leg of the lambda structure is observed not to have appreciably moved with
similar smearing to that produced with a 0V applied voltage. There is a similar
termination of the lambda structure, in the top right hand section of figure 5.15.
However, there is reduced shock smearing in the neighbouring lower flow field, as seen
at the bottom of the figure. It is not known whether this is a reduced noise effect or the

5-136
increased unimorph tip deflection is producing larger/stronger slot control, which
creates a stronger terminating leg of the lambda structure in the neighbouring flow field.

with a -500V applied voltage. (Open)

The low-pressure region on the upstream unimorph has increased uniformity, with a
reduced pressure compared to the 0V and 500V cases. This suggests the stronger slot
control is producing larger bubbles of influence above the upstream unimorph.

The oil flow visualisation and pressure distribution experiments reported in sections 5.1
to 5.3 were conducted approximately one year prior to this series of PSP experiments.
The manufacturer states, as indicated in the literature survey, that the piezoelectric
material has an operating life during which it permanently degrades. This would
produce less deflection for the open flaps case of the current experiment than was
previously achieved. This is explains why a significant local pressure maximum region
ahead of the shock is not observed in figure 5.15, unlike those in figs. 5.7 and 5.8
(labelled A). To a lesser extent the free stream conditions (temperature and humidity)
would also have changed to produce a slightly different SBLI behaviour.


5-137
5.4.6. - Flap deflection effects
In the above discussion, the effects of unimorph deflection, per se, have not been
mentioned. Indeed, there will be effects due to an upstream compression ramp and
downstream expansion corner. However, with the amount of deflection being less than
1mm along the 52mm long unimorph, it is expected that the deflection effects will be at
least an order of magnitude less than the mass transfer effects. Furthermore, this
deflection will create non-linearity in the PSP intensity recorded by the camera but it is
expected that such discrepancies would be far outweighed by the noise in the pressure
sensitive paint imaging.

5.4.7. - PSP Summary
Experiments have shown that SBLI control using piezoelectric actuators is feasible.
From the discrete pressure data obtained, the concept of piezoelectric actuator flap
control is shown to be an effective method of controlling the shock wave/boundary
layer interaction. However, the unimorph actuators were unable to completely
eliminate the control effects.

Using the discrete pressure data, in conjunction with the PSP, it is seen that the pressure
distribution is more complicated and the pressure maps obtained show that, globally,
active control is achieved with unimorph flaps. However, the presence of slot control
gives a locally continuous passive control of the interaction.

It was assumed that, with a 500V applied voltage, the unimorph flaps are closed and
have zero deflection. The pressure distribution map indicates that the global flow field
is approximately uncontrolled with a small amount of extra smearing, resulting in a
limited localised passive slot control.

When the unimorphs are free to deflect, with 0V applied voltage, the leading leg of the
lambda structure is not seen to change significantly. However, a higher degree of
smearing can be observed, compared to the flaps closed case, indicating thereby a
broader lambda shock structure and hence suggesting reduced wave drag.


5-138
With a 500V applied voltage, to assist unimorph deflection, the leading leg of the
lambda structure is observed not to move appreciably and the shock smearing is similar
to that observed with a 0V applied voltage. However, it is not known whether the
increased unimorph tip deflection, from 0V to 500V, produces improved slot control
or if the increased deflection has exceeded the optimum amount for control.

5.5. Mechanically deflected flap SBLI Control
Flap deflection of approximately 0.5mm is possible with the piezoelectric actuators,
which is clearly not adequate for optimal control and greater deflection is needed.
Therefore, the flaps were mechanically deflected up to 3mm in order to evaluate the
swept normal SBLI control performance and the characteristics of optimal control.
Furthermore, a set deflection of 0mm experiment was conducted as the piezoelectric
actuators were only theoretically shown to produce zero deflection.

5.5.1. - Oil Flow Visualisation
5.5.1.1. - 0MM FLAP DEFLECTION
Figure 5.16 - Oil flow visualisation of swept SBLI control
with 0mm deflected flaps test section.


5-139
Figure 5.16 shows the flow features of the swept normal SBLI control with 0mm flap
deflection. Compared to figures 5.2a to c, it can be seen that 0mm flap deflection has
produced a quasi-uncontrolled surface flow footprint with the parallel streamlines
downstream of the convergence line indicating that the flow is incipiently separated.
The convergence line is observed to deflect from 40 to 41 as it progresses from the
lower region of the control plate, at the bottom of figure 5.16, through the unimorph
region that indicates that minimal control is being produced from the passive control
provided by the slots. The divergence (shock generator reattachment) line can be
observed, which is a characteristic of re-attaching flow; a region of separation is
observed as the oblique shock generated by the wedge interacts with both the sidewall
and ceiling boundary layers, at the top right of figure 5.16.

Figure 5.17 - Oil flow visualisation of Swept SBLI Control at M=1.3
with a 1mm flap deflection

It can be seen, from figure 5.17, that with a 1mm flap deflection there are three main
differences of the oil flow visualisation characteristic compared to the 0mm deflection
case. In the lower region of the test surface, the convergence line is unchanged at 40

5-140
with similar downstream streamlines deflected to 41 and a divergence line. However,
as the convergence line experiences flap control, it is observed to deflect to 63
indicating an increased level of control. The streamlines on the downstream flap are
also seen to further deflect to 47.
The second main difference is the streamlines emerging from the upstream flap, with a
1mm flap deflection creating transverse mass injection across the sides of the flap. This
is indicated by the estimated mass transfer separation line, see fig. 5.17. However, it is
only observed on the upper side of the upstream flap as it is thought that the incoming
flow on the lower side deflecting upwards onto the convergence line negates the
transverse mass injection. The convergence line above the flaps is being strongly
influenced by the transverse mass injection and is more a separation line than the
convergence line of the wedge generated shock.

Lastly, a small region of separation is observed at the end of the upstream flap due to
the flap creating a rearward facing step combined with the presence of mass injection
(labelled as separation region on fig. 5.17). This region of separation is seen to be
biased to the upper side due to the lower incoming fluid deflecting onto the
convergence line inhibiting the separation on the lower end of the upstream unimorph.

5.5.1.3. - 2MM AND 3MM FLAP DEFLECTION
Referring to figure 5.18a, a 2mm flap deflection accentuates these characteristic
changes with the flap convergence line increasing to 66. The increased deflection
promotes the mass injection across the sides and spillage off of the upstream flap,
creating a larger separation region on the upper side of the flap control region. This
separation region is believed to completely dominate the surface flow, over the
convergence line. The estimated separation region at the end of the upstream flap is
observed to extend to a small degree downstream and lower as the rate of mass injection
increases. However, it is still contained by the flap convergence line and the end of the
upstream flap.


5-141
a)
b)
Figure 5.18 - Oil flow visualisation of Swept SBLI Control at M=1.3
with flap deflections of a) 2mm and b) 3mm.


5-142
The effects of the mass suction from the downstream flap are now observed to affect the
neighbouring flow. This is only observed on the upper side of the downstream flap as
the deflected streamlines below the flap control region produce similar effects.

With a 3mm flap deflection the convergence line angle below the flaps reduces to 34
and increases to 76 as it experiences flap control, see fig. 5.18b. The increased
deflection produces greater mass transfer and hence enhanced control of the swept
SBLI. The combination of greater mass injection across the sides and the spillage off of
the upstream flap dominate the lower neighbouring flow, producing a small separation
line. Above the upstream flap this spillage effect is greater, producing a separation line
that extends further into the upper section of the control surface. The increased
downstream suction also has a greater influence on the neighbouring flow above the
flap. Also, the separation region at the end of the upstream flap, still bound by the flap
convergence line, is seen to extend lower and link up with the separation line created by
the mass injection across the lower side of the upstream flap.

5.5.1.4. - LEADING LEG SHOCK ANGLE
The main shock generated by the wedge occurs parallel to the uncontrolled convergence
line and has an angle, |
s
, of 41 to the horizontal. From the oil flow visualisation
experiments, measuring the angles of the leading legs of the lambda feet for different
levels of control has great inaccuracy due to the complex flow patterns. Therefore, it is
assumed that the angle of the leading leg of the lambda shock due to the flap control is
the angle of the flap convergence line minus the 41 shock angle plus the 4.5 angle for
the uncontrolled SBLI, Squire (1996). For the quasi-uncontrolled 0mm deflection the
flap convergence line angle, |
cl
, is 41, which gives an equivalent A
2
[= (|
cl
- |
s
+ 4.5)]
of 4.5.
For 1mm, 2mm, 3mm flap deflection, the convergence line angles are, |
cl
, are of 63.5
and 66 and 76 respectively, which relate to A
2
of 27.0, 29.5 and 39.5 respectively,
see fig. 5.19. These signify leading leg shocks that are anchored further away from the
main shock, indicating a weaker front shock and consequent reduction in wave drag.

5-143
An increased error of 2 has been generated due to the ambiguity of the control
provided with 0mm and the complex oil flows.

Figure 5.19 Variation of A
2
[= (|
cl
- |
s
+ 4.5)]
for front shock with normal Mach number, after Squire (1996).

From figure 5.19, it can be seen that increased flap deflection leads to improved SBLI
control due to greater mass transfer. It would appear that unimorph control with open
flaps produces a deflection somewhere between a half to two-thirds of a millimetre and
that the addition of the longitudinal slots provides control approximately equal to a
flap deflection of 2mm. The level of control is significantly improved with 1mm flap
deflection compared to 0mm flap deflection with a marginal increase with 2mm
deflection. It appears from the trend of 0mm to 2mm deflection that further deflection,
and hence mass transfer, would asymptote the A
2
to approximately 30. However, the
observed 39.5 with 3mm deflection is believed to be a result of deleterious separation
effects.

It is, therefore, suggested that a 1mm flap deflection provides optimal SBLI control as
the separation produced is minimal whilst inducing a significant amount of control.
Also, at 1mm flap deflection the separation effects around the slots are minimal.

5.5.2. PSP Results
The instantaneous pressure fields measured with pressure sensitive paints were obtained
for the mechanically set flap deflection for control of the swept normal SBLI. The

5-144
uncontrolled swept SBLI experiment was repeated to ensure repeatability of the PSP
and experimental set-up
.
5.5.2.1. - THE UNCONTROLLED CASE
Figure 5.20 is observed to be in close agreement with the schlieren picture taken for the
location of the leading leg of the uncontrolled lambda structure (ULS), see fig. 5.1. The
pressure upstream of the ULS is generally uniform and increases in pressure as it travels
through the lambda structure. The pressure further rises downstream and achieves
approximate inviscid levels as it reaches the end of the control plate. The smaller region
of lower pressure that is observed at the bottom right of the picture is generated by the
flow expanding over the wedge.

Compared to the previous PSP uncontrolled SBLI study (Fig. 5.11), the pressure map
obtained shows good repeatability with two subtle differences. The pressures are
greater by 0.01 p/P
0
in figure 5.11 but they lie within the accuracy of 0.02 in p/P
0
.
Secondly, the shadowing effects previously observed in the bottom right hand section of
the figures have been minimised, suggesting an improved light/camera set-up.
Figure 5.20 - Pressure distribution map of the uncontrolled swept SBLI using PSP
The two experiments were conducted approximately a year apart.


5-145
Figure 5.21 - Pressure distribution map of the Swept Normal SBLI
using PSP with flaps deflected to 0mm
Referring to figure 5.21, the pressure distribution map for 0mm flap deflection control
indicates that a quasi-uncontrolled SBLI has been created. A uniform lower upstream
pressure of 0.130.02 can be seen, which gradually changes to a higher downstream
pressure after the ULS of 0.200.02. A reduced pressure on the downstream flap of
0.160.02 is observed, indicating a larger lambda structure that is produced from the
passive control provided by the longitudinal slots. Evidence of slot control is further
observed with a higher pressure indicated at the start of upstream flap.

As seen from figure 5.22 with 1mm flap deflection, the degree of smearing of the
lambda structure is greater with a larger region of lower pressure on the downstream
flap. The high pressure region observed at the start of the upstream flaps with 0mm flap
seems to have been eliminated. It is thought that the mass injection across the end of
the flap exerts less resistance to the flow than injection through the 0.2mm slots.
Therefore, the amount of mass injection at the start of the upstream slots will be
minimal. Furthermore, flap deflection creates a rearward facing step to the flow and,
combined with the mass injection, this induces boundary layer thickening, which then

5-146
reduces the pressure at the end of the upstream flap. No effects can be seen in the
neighbouring flow fields above and below the control flaps.

Figure 5.22 - Pressure distribution map of the Swept Normal SBLI
using PSP with 1mm deflected flaps.

5.5.2.4. - 2MM AND 3MM FLAP DEFLECTION
A 2mm flap deflection induces further smearing of the lambda structure across the flaps
due to increased mass transfer. The smearing below the flaps, however, is reduced
producing a sharper rise in pressure due to its proximity to the main shock. It is
believed that the increased transverse mass injection across the sides of flap creates
regions of higher pressure adjacent to the end of the upstream flap. Also, the higher
pressure on top of the upstream flap is attributed to the deflection creating a
compression corner that produces localised compression waves. This then expands
over the rear facing step to a lower pressure.

The combination of transverse mass injection and the deflection of the upstream
streamlines on the convergence line inhibit the lower pressure region at the end of the
upstream unimorph on its lower side. Conversely, it promotes its influence on the upper
side, as seen with the separation region indicated with oil flow in section 5.5.1.

5-147
a)
b)
Figure 5.23 - Pressure distribution map of the Swept SBLI using PSP with flaps
deflected to a) 2mm and b) 3mm.

The 3mm flap deflection generally accentuates the 2mm deflection characteristics with
an increased lambda smear on the downstream flap, a higher pressure on the upstream
flap due to the larger compression ramp and the larger mass injection. This has the

5-148
effect of creating a larger region of higher and lower pressure in the neighbouring flow
fields at the end of the upstream flap on its lower and upper side respectively.

5.5.2.5. COMPARISON WITH CONTROLLED PSP DATA
The PSP results obtained for the piezoelectric actuated unimorphs (figs. 5.12, 5.14 and
5.15) are observed to produce results equivalent to the 0mm and 1mm set flap deflection
PSP (see figs. 5.21 & 5.22). Secondly, it is seen that unimorph control with a 500V
applied voltage does not reproduce the 0mm flap deflection PSP map, suggesting that
the unimorphs are not closed, that is they do not achieve true zero deflection.
Furthermore, the open flaps pressure map, produced with a -500V to assist flap
deflection, does not quite reproduce the 1mm deflection pressure map, which indicates
considerably enhanced shock smearing.

The PSP data indicate that the piezoelectric actuation provides flap deflections of less
than 1mm with open flaps and approximately 0.2mm
with closed flaps. This would

suggest that the pressure distribution acting on the unimorph is somewhere between the
uniform and quadratic distributions modelled (see chapter 4). From section 4.7. the
predicted deflections, using CLPT and FEM, are 0.52mm and 0.51mm respectively for
magnitude of maximum deflection whilst theoretically ensuring zero deflection for a
17.64kPa quadratic pressure load. However, under a 17.64kPa uniform pressure load
the predicted deflection ranges, using 500V, are
-0.99mm to -0.43mm and -0.99mm to -0.39mm for CLPT and FEM respectively.

It is believed that under increased flap deflections, greater than 1mm, the acting
pressure on the unimorph may tend toward the quadratic pressure distribution,
especially when spillage and compression corner effects are introduced. Conversely,
the acting pressure will tend to a uniform load as the deflection approaches 0mm and
the mass transfer across the sides of the flaps is minimised.

to allow mass transfer across the unimorph lip.


5-149
5.5.3. Discrete Pressure data
5.5.3.1. - THE L1 STREAMLINE
It can be seen from figure 5.24 that the surface pressure along L1 with 0mm flap
deflection provides a similar pressure trace to the previous study with slot control, see
fig. 5.7b.

The 0mm flap deflection pressure trace is similar to the no control pressure rise
indicating a similar amount of SBLI control. However, the main pressure rise with
0mm flap deflection (-4< X/o
0
<-1.5) is more gradual due to the passive slot control
provided, indicating a weaker/larger lambda structure. Furthermore, the higher pressure
observed upstream of the main pressure rise (-10< X/o
0
<-7.5), compared to no control,
is believed to be due to the mass injection through the upstream longitudinal slot.

With 1mm flap deflection, the recorded pressure trace is similar to the piezoelectric
actuator unimorph flap pressure trace with a -500V applied voltage to assist deflection,
see section 5.3.2. A pressure maximum (labelled as A at X/o
0
=-5.5) is observed due to
the leading leg of the lambda shock structure with a localised minimum (labelled as B at
X/o
0
=-4) due to boundary layer thickening before the main pressure rise and an
observed pressure plateau (labelled as C at -1.5< X/o
0
<1.5) that indicates separation.

Figure 5.24 - Surface Pressures along L1 for swept SBLI control

5-150
With a 2mm flap deflection, a lower pressure maximum (A) is observed to spread
upstream compared to 1mm flap deflection, which is indicative of a larger series of
weaker shocks that are induced further upstream. It is believed that this is produced by
the combination of increased transverse mass injection and upstream surface curvature
(compression ramp) at the greater flap deflection. The main pressure rise is observed to
initially reproduce the 0mm pressure rise but then plateaus at a higher pressure
compared to the 1mm case due to the increased mass injection. It is believed that the
localised maximum and minimum in the pressure plateau region with 2mm deflection
indicates a less steady region of separation. The final pressure ratio tends towards the
inviscid value and is expected to reach the uncontrolled value at greater downstream
distance.

A 3mm flap deflection enhances the pressure trace changes observed with 2mm flap
deflection with: a lower pressure maximum that is spread further upstream due to the
increased mass transfer and surface curvature; a similar pressure minimum due to
boundary layer thickening and a slightly higher plateau pressure, with increasing
fluctuations downstream. The increased mass injection and surface curvature with 3mm
flap deflection does produce a weaker series of compression shocks that are initiated
further upstream. However, separation effects around the slots are expected to outweigh
the weaker lambda structure.

5.5.3.2. THE L3 STREAMLINE
Figure 5.25, showing the corresponding pressure rises along L3, allows the interaction
to be observed for an increased downstream control region and has similar
characteristics to figure 5.24. The local maximum (A) is greater because the leading leg
is closer to the main shock position. This presents a more normal shock to the flow and
therefore an increased pressure rise. The local maximum is seen to be at a lower
pressure and spread further upstream with increased flap deflection due to the increased
mass transfer and surface curvature.


5-151
Figure 5.25 - Surface Pressures along L3 for swept SBLI.

The local minimum (B) is not as obvious along L3 and it is uncertain whether the
magnitude of difference between maximum and minimum is less, compared to that
along L1, due to its proximity to the maximum (A) and the plateau (C) or if the
boundary layer has not thickened to the same degree. For similar reasons it is unclear
why the plateau is at a lower pressure than the local maximum. The plateau pressure is
observed to reduce with 1mm flap deflection compared to 0mm due to mass transfer
and a larger lambda structure.

With increased deflection the average pressure plateau pressure, between X/o
0
=-V and
X/o
0
=V, is marginally affected. However, greater fluctuations in the plateau pressures
are produced indicating a less steady lambda structure
. The pressure fluctuations

downstream are assumed to be due to the local interaction with the trailing leg of the
lambda structure and the thinning of the boundary layer. It can be seen that the
fluctuations increase with flap deflection. The lower plateau pressure and reduced
downstream fluctuations would again suggest that 1mm flap deflection is optimal.


5-152
5.6. Swept Normal SBLI Summary
The results indicate that optimal control of the swept normal SBLI is attained with a
flap deflection between 1mm and 3mm. However, to obtain the deflection required for
optimal performance a more powerful piezoelectric actuator material is required than is
currently available.

It was seen that 0mm flap deflection produced a quasi-uncontrolled SBLI with slight
deviations due to the continual passive control provided by the slots. The main
pressure rise across the SBLI, with 0mm flap deflection, is more gradual due to the
mass transfer provided, indicating a weaker/larger lambda structure. Also, a higher
pressure is observed upstream of the main pressure rise due to upstream mass injection
via the longitudinal slots.

A 1mm flap deflection produces a significant level of beneficial SBLI control with the
convergence line deflecting to 63, compared to the 41 for uncontrolled convergence
line. A region of separation is also produced at the end of the upstream flap due to a
combination of the flap creating a rearward facing step and the presence of mass
injection.

Increasing the flap deflection provides greater levels of tangential and transverse mass
transfer, which has greater influences on the global flow field. However, increased flap
deflection also promotes spillage of the fluid above the upstream flap over the sides.
Greater spillage and transverse mass injection creates a weaker leading leg that is
initiated further upstream causing a separation region that extends further into the
neighbouring flow fields. In addition, the increased 2mm and 3mm deflections are
observed to produce an unsteady influence, which the author speculates will progress
downstream of the interaction region.

It is suggested that a 1mm flap deflection provides near optimal SBLI control as the
separation and unsteadiness produced are minimal whilst a significant amount of control
is produced. Also, with 1mm flap deflection, the separation effects are minimised in the
neighbouring flow fields above and below the flap region.


5-153
The PSP results obtained for the piezoelectric actuated unimorphs suggest flap
deflections between 0mm and 1mm. It is believed that the piezoelectric actuation
provides a maximum flap deflection between half and two thirds of a millimetre with
open flaps and approximately 0.2mm with closed flaps. This would suggest a
pressure force distribution is acting on the unimorph somewhere between the uniform
and quadratic pressure distributions modelled in chapter 4. CLPT and FEM show
unimorph deflection ranges of -0.99mm to -0.43mm and -0.99mm to -0.39mm under a
17.64kPa uniform pressure load. With the quadratic pressure load the corresponding
values are -0.52mm to +0.05mm and -0.51mm to +0.09mm.

Under increased flap deflections, greater than 1mm, the pressure force acting on the
unimorph may tend toward the quadratic pressure distribution, especially when spillage
and compression corner effects are introduced with surface curvature. However, the
pressure force tends to act as a uniform load as the deflection approaches 0mm.

5-154
Chapter 6. Unswept Normal SBLI Control At M=1.3

6-155
CHAPTER 6. Unswept Normal SBLI Control At M=1.3

The experiments with an unswept normal shock were primarily concentrated with a
Mach number of 1.5, as any beneficial SBLI characteristics should be more apparent at
M=1.5 than at M=1.3. Therefore, the unimorph structure for the unswept normal SBLI
studies was designed for the 30kPa pressure difference acting on the unimorphs with the
M=1.5 flow. However, experiments were conducted also with M=1.3 with a 21.8kPa
acting pressure difference. It was estimated that the unimorph deflection ranges
available are -0.38mm to -0.10mm and -0.22mm to 0.07mm for the uniform and
quadratic pressure distribution models respectively, as discussed in chapter 4.

6.1. - Uncontrolled SBLI

The uncontrolled SBLI at M=1.3 has been previously studied by numerous researchers
using the Cambridge facility. It was therefore decided to utilise the work of Holden
(2004) for data comparison for our Mach 1.3 unimorph controlled actuator experiments.
However, it should be noted that a different second throat had to be incorporated for the
unimorph control experiments to enable steady shock positioning in the test section.
Therefore, any comparison between the uncontrolled Mach 1.3 experiments by Holden
(2004) and the present experiments can only be qualitative.

Referring to figure 6.1a, the flow is from left to right, with the normal shock,
M
= M
n
= 1.3, interacting with the naturally grown boundary layer on the tunnel floor.
Imperfections in the tunnel have caused spurious Mach waves upstream of the SBLI
that have negligible effect on the flow before the interaction. From oil flow pictures it
is observed that the flow is incipiently separated, as shown by the slight oil flow build
up at the main shock location as shown in figure 6.2. It is also seen that the flow is
affected by the sidewall boundary layers.


6-156
a)
b)
Figure 6.1 a) Schlieren pictures and b) total pressure traverses
of the uncontrolled SBLI at M=1.3, Holden (2004).

The incipiently separated flow structure is in agreement with the literature, where
separation first occurs at M=1.3 with compression waves upstream of the shock
location. The leading leg of the lambda structure, compression wave, has a 52.0 1
shock angle and this region has a reduced total pressure loss. The schlieren picture
gives a triple point height of approximately 31mm 1mm, as shown by the coalescence
of the leading compression wave with the main shock, see fig. 6.1a. The total pressure
traverse, measured 90mm downstream of the shock position, gives a triple point height
of approximately 32mm 1mm, below which the total pressure ratio rises above the

6-157
theoretical value behind a normal inviscid shock of the same strength, see fig. 6.1b. The
triple point height difference is expected to be created from the shear layer lifting away
from the surface with downstream distance. From the total pressure traverse, the
boundary layer thickness is shown to be approximately 10mm 1mm. The error bars, of
the measured shock angles and structure heights, are due to the low Mach number. The
low Mach number creates: subtle Mach waves that appear light in the schlieren;
increased smearing; and gradual triple points on the total pressure traverses.

Figure 6.2 Oil Flow visualisation of the uncontrolled SBLI at M=1.3, Holden (2004).

6.2. - Unimorph Controlled Flap Deflection

Oil Flow visualisation was only conducted with a 0V applied voltage for safety reasons.
A 500V applied voltage could not be utilised as it is likely that the oil will act as a
conductor and earth the power supply through the tunnel and instrumentation.

Referring to figure 6.3, the flow is from left to right with the main shock located above
the start of the downstream unimorph. The oil flow is slightly smeared, especially at the
end of the downstream flap, due to tunnel shut down removing the aeroelastic
deflection
1
. The figure has been divided into three distinct regions: a unimorph flap

1
During experiments the author ensured the oil flow on the exposed control surface dried. However,
during runs a small amount of oil flow mixture would enter the plenum via the downstream flap. This oil
flow mixture would not be exposed to the main flow and subsequently not dry. On tunnel shut down the
small amount of mixture would be forced back on to the control surface via the downstream flap.

6-158
control region bounded by the sides of the unimorphs; two longitudinal slot control
regions; and two uncontrolled region outside of the longitudinal slots.

Figure 6.3 - Oil Flow visualisation of SBLI
with Unimorph Control at 0V at M=1.3 and Main Shock at X/
o
=0.

Referring to figure 6.3, a separation bubble is initiated, for the localised flap control at
the end of the upstream unimorph as a result of the aeroelastic effect. This results in a
small amount of mass transfer and it is thought that this will lock the unsteady shock
into position. This fixing of the shock position would minimise the unsteady
characteristics of the shock, for example the deleterious effects of noise, buffet, etc.
The separation bubble is approximately the same size as the length of separation in the
uncontrolled region. The reattachment point is at a distance of approximately 3
0
, where
0
is the thickness of the undisturbed upstream boundary layer (see Chapter 2). Also, a
region of reversed flow, across the main shock position, is observed that fans out as it
approaches the end of the upstream unimorph.

The flap deflection, although small, creates an upstream compression ramp to the
incoming boundary layer. The higher pressure experienced on top of the flap will
promote fluid spilling off the sides, as seen at the end of the upstream unimorph in fig.
6.3. Furthermore, the flap deflection will increase the orifice area at the sides of the
unimorph, for transverse mass injection to take place. Injection will occur
approximately normal to the side orifice opening and, therefore, will be angled away

6-159
from the flap. The presence of the separation bubble and spillage off the top of the
unimorph enhances this transverse mass injection. The transverse mass injection
separation is shown to be initiated towards the end of the upstream unimorph and is
terminated by mass suction via the downstream unimorph.

Horseshoe vortices can also be seen at the leading edge of the unimorph longitudinal
slots, due to the mass injection across the sides of the unimorph. As discussed in
section 2.2.3., the flaps were machined into the aluminium substrate with 0.2mm cuts,
which creates slots. The horseshoe vortices indicate the presence of passive slot
control, as shown by Smith (2002). However, the amount of control produced by the
slots is minimal due to their small size compared to the 7.47mm thickness of
0
.
6.2.2. - Schlieren and Total Pressure traverses
Figures 6.4a to c have similar first oblique shocks upstream of the main shock that
originate from the start of the test section control plate. The shock angle is 53.0, which
is an increase from the 52, of the spurious wave, for the uncontrolled case, see fig.
6.1a. It is uncertain if the observed angle change is created by the difference between
the unimorph control plates (fabricated here in Australia) and the uncontrolled test plate
(fabricated in the workshops at Cambridge). Secondly, the stagnation pressure, which is
manually controlled, would have an effect on the oncoming boundary layer and local
Mach number. Thirdly, the presence of the unimorph control theoretically should
reduce the total pressure loss and the resulting higher downstream total pressure could
affect the position and nature of the SBLI itself.

The results indicate that the second oblique shock is formed by localised boundary layer
thickening due to fluid injection across the upstream unimorph sides, the so called
longitudinal slots. The shock angle is seen to increase from 53.5, closed flaps,
which are at approximately 0mm deflection, to the aeroelastic deflected 55.5 of the
inert flaps. As the unimorphs start to deflect there will be an increase in mass
injection across the end of the unimorph creating a larger compression corner effect
from the thicker boundary layer. However, with open flaps the second oblique shock
angle reduces to 54. It is believed that this is due to an increase in mass transfer across
the end of the unimorph. Streamwise mass transfer across the unimorph end, with a

6-160
larger orifice area, will exhibit less resistance to the fluid compared to across the thin
slots at the unimorph sides. This will decrease the amount of upstream mass
injection/boundary layer thickening via the longitudinal slots and induce a smaller
second oblique shock angle.

a)
b)
c)
Figure 6.4 Schlieren pictures of unimorph controlled SBLI at M=1.3 with
a) closed flaps (500V), b) inert flaps 0V & c) open flaps (-500V).


6-161
Shock angle data has been tabulated for ease of comparison of the trends created with
piezoelectric actuation (table 6.1). It should be noted that the uncontrolled case does not
have a second oblique shock angle as this is a feature of unimorph control.

Flap Deflection First Oblique
Shock Angle ()
Second Oblique
Shock Angle ()
Leading Leg
Shock Angle ()
Uncontrolled 52.0 (Spurious wave) - 52.0
+500V closed flaps 53.0 53.5 58.5
0V inert flaps 53.0 55.5 58.0
-500V open flaps 53.0 54.0 58.0
Table 6.1 Summary of shock Angles for the uncontrolled
and unimorph controlled SBLI at a nominal Mach number of 1.3, (accuracy of 1).

The bottom of the normal recovery shock is observed to bifurcate but a vortex sheet is
not seen. This is possibly due to the fact that the strength of the normal shock is weak.
The angle of the leading leg of the lambda structure is 58.0 for the open flaps.
However, it does not change perceptibly for the closed flaps configuration, which
seems to be within the experimental accuracy of the measurements. It is possible that
the effects of mass injection with open flaps produce similar boundary layer
thickening to the closed flaps due to an increased mass injection at a more oblique
angle.

No boundary layer traverses were performed for unimorph SBLI control at M=1.3 due
to time and experimental constraints. Therefore, the triple point height and boundary
layer thickness have been estimated from the schlieren photographs. These values have
been tabulated in table 6.2 for ease of comparison of the trends created with
piezoelectric actuation. The triple point height and boundary layer thickness as a result
of the unimorph control along the centreline (Flap Control) and along the longitudinal
slots (Slot Control) (see fig. 6.3a) have been labelled as FC and SC respectively. It
should be noted that schlieren is an integration of the flow structures through all
spanwise positions and, therefore, is a product of the uncontrolled, the slot controlled
and the unimorph flap controlled SBLI structures. The heights specified in table 6.2 for
the various SBLI control structures were confirmed via total pressure traverses. These,
however, have been omitted as they offer no extra information.

6-162
Flap Deflection FC Triple
Point Height
(mm)
SC Triple
Point Height
(mm)
FC Boundary
Layer Thickness
(mm)
SC Boundary
Layer Thickness
(mm)
Uncontrolled
31.0 10.0 -
500V closed flaps
28.5 89.0 10.5 17.0
0V inert flaps
26.5 86.0 8.5 17.0
-500V open flaps
30.0 100.0 9.5 16.0
Table 6.2 Summary of triple point height and boundary layer thickness
for the uncontrolled and unimorph controlled SBLI at M=1.3, (accuracy of 1mm).

The closed flaps configuration is expected to produce approximately zero unimorph
deflection. This would generate passive control effects from a combination of
longitudinal and lateral slot control. It is observed that with closed flaps the FC
triple point height is reduced and the boundary layer is marginally thicker, suggesting a
smaller region of reduced total pressure loss and, therefore, a slightly degraded
performance in respect of the wave drag, compared to the uncontrolled case. Also, it is
expected that the FC boundary layer is thicker due to the presence of the flaps.

Slot control is seen to give a three fold larger localised lambda shock foot due to the
increased upstream distance of the start of the longitudinal slot, at approximately 7
0
compared to 1.5
0
with flap control along the centreline. The SC boundary layer is also
twice as thick and it is believed that the combination of the smaller FC and larger SC
lambda structures produces a beneficial overall reduction in total pressure loss in spite
of the localised boundary layer thickening.

With a 0V applied voltage the unimorphs will deflect due to the aeroelastic effect of a
cantilever beam under a pressure load. These inert flaps create a small level of
deflection, of the order of 0.25mm, that reduces the size of both FC and SC lambda
structures, in height and reduced upstream influence as inferred from the increased
second oblique shock angle coupled with its reduced triple point height, see tables 6.1 &
6.2. It is expected that the deflection will promote a more oblique mass injection across
the end of the upstream unimorph. Furthermore, it is believed that flap deflection
reduces the amount of mass injection at the start of the longitudinal slot and promotes
it towards the end of the slot where the gap is larger due to tip deflection. This would

6-163
reduce the slot control upstream influence but make it stronger further downstream to
create a stronger/smaller second oblique shock. The reduced FC triple point height of
the unimorph lambda structure is believed to result from the mass injection, angled
more rearward due to the flap deflection, thus resulting in less upstream influence.
Also, the boundary layer has thinned with a 0V applied voltage as a result of the
increased mass suction across the downstream unimorph. It is possible that the reduced
boundary layer thickness and triple point height provides a similar reduction in total
pressure loss in the closed flaps case.

A -500 V applied voltage will assist the unimorph deflection under a pressure load. The
increased deflection configuration, labelled open flaps, allows greater mass transfer to
occur, which enlarges the unimorph lambda structure compared to the inert and
closed unimorphs. The FC lambda structure is comparable to the uncontrolled lambda
structure, which is believed to be due to the decreased upstream influence,
approximately 1.5
0
and increased leading leg shock angle due to the mass injection.
However, the presence of the larger SC lambda structure will produce an overall
reduction in total pressure loss. The increased unimorph deflection has thickened the
boundary layer over the unimorph region, which influences the slot control region by
reducing the boundary layer thickness, toward the FC boundary layer thickness value.

No measurements were taken of the actual tip deflection achieved during wind tunnel
runs due to the confined area within the sealed plenum chamber. Furthermore, no
intrusive measuring system could be used externally to the plenum as this would
interfere with, and change, the flow structure over the test section.

6.2.3. Surface Pressure Measurements
Surface pressures were measured at various locations, (fig. 3.11a), normalised to the
upstream stagnation pressure and compared to the uncontrolled case, as shown in
figures 6.5a to 6.5c. Non-dimensional distances are shown relative to the main shock
location (X/
o
=0) with a negative distance implying upstream.


6-164
a)
b)
c)
Figure 6.5 Discrete pressure data of unimorph control at M=1.3 with
a) closed flaps (+500V), b) inert flaps 0V & c) open flaps (-500V).

6-165
There is no distinct pressure plateau for the uncontrolled case due to the flow being
incipiently separated. The start of the pressure rise is approximately 2
0
upstream of the
main shock location and indicates the foot of the leading shock of the lambda structure.
Only a small number of pressure taps were available within the unimorph control
region.
The pressure distribution with unimorph control is very similar for all three flap
positions, with the pressure ratio upstream of the control region being the same as the
uncontrolled case, indicating similar incoming flows. A pressure rise is observed at the
start of the control region due to the higher pressure mass injection through the
longitudinal slots. The similar pressure rise, for all three unimorph deflections,
indicates a similar amount of control. However, the overall pressure rise is less
suggesting smaller entropy rise through the shock structure with unimorph control.

The differences between the three pressure distributions are subtle but it can be seen that
the pressure rise adjacent to the shock seems to occur at a greater downstream position
with increased unimorph deflection. These streamwise variations are within
experimental errors and at increased deflection, and hence mass transfer, the trailing leg
of the lambda structure is pushed further downstream.

6.2.4. - Unimorph Control Limitation
As seen from figures 6.5a to 6.5c, the differences between the open and closed flaps
case seem to be quite subtle. This is due to the small amounts of unimorph deflection
achievable, as discussed in chapter 4. It is obvious that greater deflection is required for
optimal control. To achieve this, piezoelectric material with greater strain production,
than current commercially available material, is called for (see Crawley (1994)).

6.3. Mechanically Fixed Flap Deflections 0mm

In order to determine the deflection required for optimal unimorph control a series of
tests were conducted at mechanically fixed flap deflections to evaluate the
characteristics and performance possible. A set deflection of 0mm was examined to
determine the unimorph control performance available when zero deflection is produced
by the piezoelectric actuators. It was intended that 0mm deflection would provide a

6-166
flow similar in characteristics to the uncontrolled and the closed flaps cases, to show
that the unimorphs could be shut off to provide active control.

6.3.1. Schlieren and Total Pressure Traverses
The presence of spurious waves can be observed from the schlieren picture shown in
figure 6.6a, with the first oblique shock originating at the leading edge of the control
plate, which joins the test section nozzle. The second oblique shock is formed by
boundary layer thickening due to mass injection through the longitudinal slots and is a
localised spanwise effect.

a)
b)
Figure 6.6 a) Schlieren pictures and b) Total pressure traverses of flap control of the
SBLI with 0mm deflection at M=1.3.


6-167
The bottom of the normal recovery shock is seen to bifurcate, with a leading leg angle
of 52 1, which indicates that a quasi-uncontrolled SBLI system has been produced.
The 0mm flap SBLI structure does deviate from the uncontrolled structure with a
thicker boundary layer due to mass transfer, a smaller lambda structure along the
centreline. This is attributed to the mass injection at a distance of 1.5I
0
upstream or the
inviscid shock position, compared to the leading foot of the uncontrolled lambda
structure at a distance of 2I
0
upstream. In addition a larger localised spanwise lambda
structure due to mass injection via the longitudinal slots is created with unimorphs.

The schlieren picture indicates that 0mm deflection was not the same as that of the
closed flaps configuration where the leading leg of the lambda shock was at 58.5
compared to the 52 observed here. The reason for this is not clear but it may be due to
the fact that true 0mm deflection was not achieved.

It was determined from the total pressure traverses, see fig. 6.6b, that the height of the
triple point and boundary layer thickness on the centre line, Z=0, is 29.0mm and
10.5mm respectively, compared to 32.0mm and 10.0mm for the uncontrolled case
respectively. The reduced triple point height is believed to be the result of the mass
injection. The combination of a similar shock angle with a smaller streamwise length
will produce a lower triple point height. The slight increase in boundary layer thickness
may also be due to the presence of mass transfer across the lateral slot at the end of the
upstream flap.

The triple point height and boundary layer thickness at the spanwise station Z=13, on
the side of the flaps (longitudinal slots), are 35mm and 16.5mm respectively. At this
position the amount of mass transfer would be greater than at Z=0 due to the continual
mass transfer across the longitudinal slots. Compared to the centre line, the mass
transfer into the boundary layer commences considerably upstream of the interaction,
which results in increased downstream boundary layer thickness and also triple point
height.

The triple point height and boundary layer thickness at Z=9, which is on the flaps and
4mm from the side, is 28.5mm and 15.5mm respectively. At this spanwise position the

6-168
SBLI characteristics are influenced by both the longitudinal and lateral slots at the side
and end of the flaps respectively. It appears that the triple point is more influenced by
the lateral slot with a similar height to the Z=0 and the boundary layer is more
influenced by the longitudinal slot with the thickness tending to the Z=13 value.

The triple point height and boundary layer thickness at Z=22mm, which is 9mm outside
of the flap control region, are 32.0mm and 11.0mm respectively. At this position,
approximately one oncoming boundary layer thickness outside of the flap control
region, the results indicate that this SBLI is comparable to an uncontrolled SBLI as
there is minimal flap/longitudinal slot control effects experienced at this spanwise
position. The increased boundary layer thickness is expected to be due to the
dissipation of the longitudinal slot control into the global flow field.

It is thought at Z=30 that the SBLI is not being controlled by the flap but that the
increased triple point height is produced from the influence of the sidewall SBLI, as
seen from the oil flow visualisation fig. 6.7.

Figure 6.7 - Oil Flow visualisation of SBLI
with 0mm Flap Deflection at M=1.3 and the Main Shock at X/
o
=0.


6-169
The flow is from left to right with the main shock located above the start of the
downstream unimorph, see fig. 6.7. The figure has been divided into three distinct
regions: a flap control region bounded by the sides of the flaps; a longitudinal slot
control region above the flap sides and an uncontrolled region outside of the
longitudinal slot region where the SBLI is comparable to the uncontrolled case.

In the uncontrolled region the flow is expected to be incipiently separated, as indicated
by the small extent of oil flow accumulation downstream of the shock position.
Additionally, it is believed that separation has occurred when the shock interacts with
the tunnel side wall boundary layer and is tripped by the presence of fixture screws on
the control plate, see fig. 6.7.

It is thought that the flap control region has separated, as indicated by the thinning and
the build up (reattachment) of oil flow upstream and downstream of the main shock
position respectively. The similar oil flow in the flap control region and the
uncontrolled region suggests a quasi-uncontrolled SBLI is created with 0mm flap
deflection.

The control effect of the longitudinal slots is minimal due to their size compared to the
boundary layer thickness, which are an order of magnitude smaller. However, small
horseshoe vortices are observed at the leading edge of the longitudinal slots, due to the
minimal mass transfer, which indicates the presence of passive slot control, (see Smith
et al. (2002))

Surface pressures were measured at various locations, (fig. 3.11b), normalised to the
upstream stagnation pressure and compared to the uncontrolled SBLI. Non-dimensional
distances are shown relative to the main shock position with a negative distance
implying upstream. It can be seen in figure 6.8, with a 0mm deflected flap, the pressure
rise is similar to the no control case suggesting a quasi-uncontrolled interaction is
created, however, significant differences are observed. The slight pressure rise in the
upstream interaction region (X/
o
-1.5) is thought to be the slot effect at the leading

6-170
edge of the flap. The pressure rise downstream is generally less than in the uncontrolled
case, indicating a smeared shock structure.

Figure 6.8 Discrete pressure data of flap control with 0mm deflection at M=1.3.

6.4. - Mechanically Set Flap Deflections 1mm to 3mm

6.4.1. Schlieren and Total Pressure Plots
It is believed that at flap deflections of 1mm, and greater, the concept of longitudinal
slot control is invalid as the mass transfer, across the sides of the flaps, is not
approximately perpendicular to the surface and that the deflected flap produces a
compression ramp, which results in a pair of counter rotating vortices at the sides. The
SBLI should now be considered a three dimensional control effect and not a
combination of isolated regions of slot and flap control.

The shock angle data has been tabulated for ease of comparison of the trends created
with flap control of the SBLI, see table 6.3. It should be noted that the uncontrolled
case does not have a second oblique shock angle as this is a feature of the unimorph
control. The second oblique shock is formed by the presence of localized boundary
layer thickening, created by fluid injection into the boundary layer across the sides of
the flap. This is a localised spanwise effect and, therefore, the secondary shock appears

6-171
weaker in the schlieren than the first oblique shock, as seen by the lighter lines, see figs.
6.9a, 6.10a and 6.11a.

Shock Angle ()
Second Oblique
Shock Angle ()
Leading Leg
Shock Angle ()
Uncontrolled 52.0 - 52.0
0mm 53.0 50.5 52.0
1mm 53.0 51.0 57.0
2mm 53.0 52.5 64.0
3mm 53.0 53.0 54.0/58.0
Table 6.3 Shock Angles for the uncontrolled
and flap controlled SBLI at M=1.3, (accuracy of angles 1).

a)
b)
Figure 6.9 a) Schlieren pictures and b) Total pressure traverses of the M=1.3
SBLI control with 1mm flap deflection


6-172
Figures 6.9a, 6.10a and 6.11a have similar upstream oblique shocks that originate from
the start of the test section control plate and are created at the joint between the nozzle
and the control plate. The leading oblique shock is observed to remain at 53.0
regardless of flap deflection. This is the same angle measured for the 0mm deflected
flap control and more than the uncontrolled case.

a)
b)

The bottom of the normal recovery shock is seen to bifurcate forming larger lambda
structures at increased flap deflection, see figs. 6.10a and 6.11a. The angle of the
leading shock of the lambda structure for 1mm and 2mm flap deflections is seen to

6-173
increase to 57 and 64 respectively, compared to the 0mm and uncontrolled value of
52. Flap deflection allows increased mass transfer to occur, producing greater
boundary layer thickening and along with the surface curvature, induces a more normal
shock.
a)
b)

Referring to figure 6.11a, at 3mm flap deflection a much larger lambda structure is
produced with two observable leading legs. It is believed that the first leading leg is
primarily caused by surface curvature of the upstream unimorph and the second leading
leg produced from the mass injection across the end of the flap. The series of leading
legs/shocks of the lambda structure produce a more contoured trailing shock. Such a

6-174
contoured trailing shock is also observed with mesoflap technology, which generates a
series of leading shocks from a matrix of deflected flaps and mass injections, as seen in
figure 6.12 (after Lee et al. (2002)).
Figure 6.12 Schlieren picture of Mesoflaps with Aeroelastic Transpiration
Control of the SBLI at M= 1.4 , Lee et al. (2002).

The triple point height and boundary layer thickness have been tabulated for the
different spanwise position for ease of comparison of the trends created with flap
control of the SBLI, see table 6.4. It should be noted that the uncontrolled values at all
spanwise positions have been assumed to be equal to the Z=0 uncontrolled value and
that the flow is two dimensional up to and including Z=30. Furthermore, the 3mm flap
deflection triple point heights have been omitted due to the complexity of the SBLI
structure and its size relative to the boundary layer traverse instrumentation limits.

Uncontrolled 0mm
Deflection
1mm
Deflection
2mm
Deflection
3mm
Deflection
Z
(mm)
TP
(mm)
(mm)
TP
(mm)
(mm)
TP
(mm)
(mm)
TP
(mm)
(mm)
TP
(mm)
(mm)
0 32.0 10.0 29.0 10.5 30.0 14.0 31.5 17.0 - 18.5
9 32.0 10.0 28.5 15.5 29.0 14.0 31.5 15.0 - 17.5
13 32.0 10.0 35.0 16.5 35.0 16.0 31.5 15.5 - 16.5
22 32.0 10.0 32.0 11.0 32.5 14.0 31.5 14.5 - 15.0
30 32.0 10.0 35.0 10.0 35.0 10.0 35.0 10.0 - 10.5
Table 6.4 Summary of triple point heights and boundary layer thicknesses for
flap controlled SBLI at M=1.3 for various Z positions and flap deflections
as determined from figs. 6.6 & 6.9, (accuracy of 1mm).

6-175
In general the triple point height and boundary layer thickness are both observed to
increase with flap deflection up to 3mm. The increased deflection produces a larger
lambda structure due to the increased mass transfer and surface curvature of the
upstream flap. However, there are two exceptions to this rule with 1mm flap deflection.

Firstly, the boundary layer initially thins with 1 mm deflection at the spanwise station of
Z=9, which is attributed to the the flow over the upstream flap spilling off the sides,
increased downstream suction and the mass injection across the end of the flap becomes
oblique with initial deflection. However, at greater flap deflection the increased surface
roughness of the flaps into the flow, the presence of rearward/forward facing steps and
increased mass transfer induces the boundary layer to further thicken.

Secondly, along Z=13 the triple point height and boundary layer thickness are not
greatly affected with 1mm flap deflection. It is believed that the increased mass
injection occurs at an oblique angle compared to the normal injection with 0mm
deflection resulting in a similar effect. However, above 1mm deflection both the triple
point height and boundary layer thickness reduce due to the combination of increased
surface roughness, mass injection and spillage off the upstream flap.

Along Z=22, the triple point height and boundary layer thickness are intermediate
values of the Z=13 and uncontrolled values, indicating the influence of the transverse
mass injection and spillage off the upstream flap, on the global flow field. Greater
deflection increases the flaps influence over the uncontrolled SBLI, with the Z=22
values tending away from the uncontrolled values, towards the Z=13 values with
increased deflection.

The triple point height and boundary layer thickness values, at Z=30, seem to be little
affected with flap deflection up to 2mm. It is believed at this spanwise distance the
effects of flap control have dissipated into the global flow field. At 3mm deflection, the
boundary layer marginally thickens suggesting a small degree of influence being
exerted at this spanwise position by the flap control.


6-176
6.4.2. - Total Pressure Maps
The assumption of symmetry of the total pressure traverses about the centreline (Z=0)
allowed the data to be processed, smoothed and interpolated using MATLAB, to
produce total pressure maps, see figs. 6.13 to 6.15. It should be noted that they are
interpolations of the boundary layer traverses and a more accurate picture could be
created with more pressure traverses, which could not be achieved due to time
constraints. The boundary layer is shown by the red to light blue transition, 40 to 0%
total pressure loss. The triple point height
2
is shown by the more subtle dark blue to
intermediate blue transition, 0 to 2% total pressure loss, where 2% is the total pressure
loss behind a normal M= 1.3 shock.

Figure 6.13 Total Pressure Maps with 0mm flap deflection at M=1.3.

The total pressure maps clearly show the influences on the SBLI structure, especially on
the boundary layer thickness, with increased flap deflection. Referring to figure 6.13, at
0mm flap deflection the triple point height is relatively uniform over the control region
at approximately Y=32, with a slight reduction above the flaps to Y=29. The boundary

2
Depending on the colour print-out the triple point is not clear due to the 2% change compared to the
obvious boundary layer with a 40% change. The author recommends the electronic version for
observations.

6-177
layer, however, is observed to noticeably grow from Y=10 to Y=15 with the dominating
localised SBLI control provided by the longitudinal slots.

a)
b)
Figure 6.14 Total Pressure Maps of the M=1.3 SBLI control
flaps deflected to a) 1mm and b) 2mm

Referring to figure 6.14a, with 1mm flap deflection the triple point height has lifted
over the flap to be uniform at a level of Y=30, due to increased mass injection and
surface deflection. Moreover, the boundary layer thickness over the entire flap

6-178
(-12.5<Z<12.5) is observed to be comparable to that produced by the slots at 0mm
deflection. The influences on the boundary layer of the two forms of control have
started to merge to generate localised control over the flaps with minimal influence on
the neighbouring flow field.

At 2mm flap deflection the triple point height is uniform across the entire flow field and
the boundary layer has uniformly thickened over the flaps to Y=15, see fig. 6.14b. The
influences on the boundary layer, of the two forms of control have, in effect, merged to
generate localised SBLI control over the flaps with minimal influence on the
neighbouring flow field.

For 3mm flap deflection the two control effects have combined and the boundary layer
has considerably thickened over the flap. It is expected that the increased flap
deflection is dominating the SBLI and influencing the neighbouring flow field, see fig.
6.15.
Figure 6.15 Total Pressure Maps of the M=1.3 SBLI control
with 3mm flap deflection.

The overall effect of greater fixed deflection is to increase the triple point height and
boundary layer thickness uniformly over the control area thereby achieving a shock
smearing effect. The total pressure mapping shows this quite effectively.

6-179
a)
b)
c)
Figure 6.16 - Oil Flow visualisation of
flap control of the SBLI deflected to a) 1mm, b) 2mm and c) 3mm.


6-180
Increased flap deflection increases the amount of mass injection across the sides of the
flap, which produces transverse mass injection and a separation line in the oil flow.
Initially, with 1mm flap deflection, the transverse mass injection is small and is sucked
back into the plenum chamber via the downstream flap, see fig. 6.16a. At 2mm flap
deflection, the upstream transverse mass injection starts to dominate the surface flow
and the downstream mass suction is insufficient. The separation line is then observed to
continue downstream of the control region, see fig. 6.16b. A 3mm deflection increases
the spanwise mass transfer, which is initiated further upstream creating a wider
separation line that again continues downstream of the control region, see fig. 6.16c.

The upstream flap deflection would create a finite compression that generates a pair of
counter rotating vortices from either side, which are shed downstream. Perpendicular
and transverse mass injection pushes the vortices up into the flow and away from the
flaps. The vortices are then strengthened by mass suction on the downstream flap
sucking the flow in from the lower part of the boundary layer. These vortices increase
in strength with greater deflection as seen from the wider separation lines.

A reversed flow region is produced on the downstream flap due to the presence of a
localised separation region initiated at the downstream end of the upstream flap. It is
believed that the reversed flow interaction with the mass injection across the end of the
upstream flap further promotes the transverse mass injection. This localised separation
region appears to be situated above the downstream flap.

Small horseshoe vortices can be observed, for all flap deflection configurations, at the
leading edge of the upstream flap, indicating the passive SBLI control produced via the
flap sides.

Non-dimensional surface pressures were measured at various locations, (fig. 3.11b), and
compared to the uncontrolled case. Non-dimensional distances are shown relative to the
main shock position (X/
o
=0) with a negative distance implying upstream, see figs.
6.17a to c.


6-181
a)
b)
Figure 6.17 Discrete pressure data for SBLI control at M=1.3
with flaps deflected to a) 1mm and b) 2mm.

Referring to 1mm flap deflection (fig. 6.17a), the initial pressure ratios upstream of the
shock are somewhat higher than with the uncontrolled case for all flap deflections. This
indicates the boundary layer is thickened upstream, which is due to mass injection
across the sides of the flap as seen in the total pressure maps of figure 6.14a. There will
be a 1.1 compression corner effect from the flap deflection against the main flow
3
. It

3
A 1mm deflection creates a 1.1 compression ramp along the 52mm flap, which would produce a QP/P
0
of approximately 0.02. Similarly for a 2mm and 3mm deflection, QP/P
0
values of 0.04 and 0.07 are
produced respectively.

6-182
is clear that the effect of shock smearing is occurring due to flap deflection as seen by
the gradient for the rise across the interaction region. This is consistent with the total
pressure data of figure 6.14a.

With 2mm flap deflection (fig. 6.17b), the above effects are stronger due to higher mass
injection. The shock smearing is seen to be considerable compared to the no control
case. These results are once again consistent with the total pressure map data of figure
6.17b.

Figure 6.18 shows results with 3mm flap deflection. We note that a larger initial
pressure rise is observed due to the increased surface curvature and mass transfer.
There are considerable deviations in spanwise distributions upstream and a pressure
plateau (-2 X/
o
0), indicating a large separation region, is seen to exist upstream of
the main shock. Comparing with the total pressure map, it would appear that a
separation bubble would be under the lambda structure. The lambda structure itself still
appears to exist; although from downstream pressures, it would be very weak and
distorted.

Figure 6.18 Discrete pressure data for SBLI control at M=1.3
with 3mm flaps deflection.


6-183
6.4.5. - Velocity Traverses
a)
b)
Figure 6.19 Velocity traverses for M=1.3 SBLI control
with flaps deflected to a) 0mm and b) 1mm

The velocity traverses have been non-dimensionalised to the uncontrolled flow velocity
at the edge of the boundary layer, U
e
= 283ms
-1
. A 0mm flap deflection produces a
similar velocity traverse over the flap control region at Z=0, Z=22 and Z=30 suggesting
minimal effect of the flap on the SBLI compared to the uncontrolled regions, see fig.
6.19a. However, the longitudinal slot at Z=13 produces a thicker boundary layer that

6-184
is expected to influence the Z=9 traverse, which is less than a boundary layer thickness
away.
a)
b)
Figure 6.20 Velocity traverses for M=1.3 SBLI control
with flaps deflected to a) 2mm and b) 3mm.
4
Referring to figure 6.19b, with 1mm flap deflection the centreline (Z=0) boundary layer
thickens due to increased mass transfer. However, away from the centreline the

4
The smoothing method used for graphical visualisation of the data has created a false apparent anomaly
of two data value at one height at approximately Y=2mm on the Z=0 traverse.

6-185
boundary layer thins due to spillage off the upstream flap and transverse mass injection
across the flap sides. The increased transverse mass injection combined with spillage
off the top thickens the boundary layer away from the flap, which attenuates with
distance away from the flap with the Z=30 traverse less affected than at Z=22. These
effects are accentuated with further deflection, (see figs. 6.20a & b).

Referring to figure 6.20a, the flow on the flap (Z=9) is accelerated over the freestream
velocity, which is attributed to be due to weaker lambda structure. This may indicate
the presence of a supersonic tongue and also optimal deflection. From previous
sections it is implied that optimal control is produced with 2mm deflection providing
SBLI control that extends into the neighbouring flow field. However, a degree of
boundary layer unsteadiness, on the centreline, is generated with a 1mm flap deflection,
which is promoted with increased flap deflection. Although, it does not appear to be
influencing the neighbouring streamlines and, therefore, is only a localised effect until a
deflection of 3mm is produced.

6.5 - Summary

It is seen that the uncontrolled SBLI with a normal Mach number of 1.3 is incipiently
separated, which is in agreement with past research. The uncontrolled SBLI is seen to
be relatively two dimensional with slight three-dimensional effects where the floor and
sidewall boundary layers interact.

The unimorph installation process produces longitudinal slots which result in
continuous passive control. The effect, however, is small due to the relative size of the
slots compared to the oncoming boundary layer thickness. Unimorph deflection
generates a compression wave that is formed by localised boundary layer thickening due
to fluid injection. The bottom of the normal recovery shock is seen to bifurcate leading
to a lambda structure. The vortex sheet emanating from the triple point, however, was
not seen although it must exist. In the closed flaps unimorph configuration, the orifice
area for mass transfer is small, producing minimal mass transfer across the edges of the
unimorphs. The increased mass transfer, with unimorph deflection, will promote

6-186
boundary layer thickening further upstream compared to the closed flaps
configuration.

With 0V applied, the unimorphs are free to deflect as cantilever beams due to the
aeroelastic effects of a pressure load. This reduces the size of the lambda structure in
height and upstream influence. It is expected that the deflection has promoted a more
tangential mass injection across the end of the upstream flap and reduced the mass
injection across the sides at a more transverse angle.

A -500V applied voltage, labelled open flaps, will assist the aeroelastic deflection
allowing greater mass transfer to occur that enlarges the lambda structure. It is believed
that unimorph control will lock the unsteady shock into position and minimise the
unsteady characteristics of the shock.

The pressure distribution with unimorph control is very similar for all three flap
positions, indicating a similar amount of mass injection that influences the global flow
field. The main pressure rise with unimorph controlled SBLI is approximately the same
shape as the uncontrolled SBLI. However, the overall pressure rise is less suggesting
greater smearing of the shock structure. The differences between the three pressure
distributions are not large, mainly due to the small amounts of unimorph deflection
available, but it can be seen that the main pressure rise occurs at a greater downstream
position with increased unimorph deflection.

In order to determine the deflection for optimal unimorph control a series of runs was
conducted at fixed flap deflections to evaluate the characteristics and performance. A
set deflection of 0mm produces a quasi-uncontrolled SBLI, with a thicker boundary
layer, a smaller flap lambda structure and a localised larger slot induced lambda
structure. The schlieren picture indicates that true 0mm deflection was not achieved
using the piezoelectric actuators with closed flaps at 500V.

It was found that at flap deflections of 1mm the SBLI should be considered as three
dimensional with greater flap deflection allowing increased mass transfer to occur. The
combination of greater boundary layer thickness and increased flap deflection induces a

6-187
stronger leading leg of the lambda shock. This results in a larger lambda structure,
implying greater smearing and improved pressure recovery.

Increased flap deflection produces higher pressures compared to the no control case
due to the increased surface curvature and mass transfer, which increasingly influences
the neighbouring flow field. Increasing flap deflection induces a growth of a separation
bubble under the lambda structure and a reduced final pressure ratio suggesting a
distorted lambda structure. The data from pressure distributions, total pressure maps
and velocity profiles gave a consistent picture, supporting the above conclusions.

It is not obvious from the above flap experiments that optimal control has been
produced within the deflection range tested. However, the results indicate that a
deflection between one and two millimetres may in fact provide near optimal control.


6-188

7-189
CHAPTER 7. Unswept Normal SBLI Control at M=1.5

The unimorph flaps for the M=1.5 studies were calculated to have a 30kPa acting
pressure difference. The unimorph deflection ranges were calculated to be -0.47mm to
-0.19mm and -0.242mm to 0.042mm for the uniform pressure and quadratic pressure
respectively, as discussed in chapter 4.

7.1. - Uncontrolled SBLI

a)
b)
Figure 7.1 a) Schlieren pictures and b) total pressure traverses
of the M=1.5 SBLI uncontrolled, after Holden (2004).

7-190
Referring to figure 7.1a), the flow is from left to right, with the normal shock interacting
with the naturally grown boundary layer on the tunnel floor. From oil flow pictures it
was observed that the flow has separated, as shown by S & R in figure 7.2.
Imperfections in the tunnel have caused spurious Mach waves upstream of the SBLI,
which have negligible effect on the flow before the interaction.

The flow structure is in agreement with the separated SBLI, as defined by Seddon
(1960), with a slip line occurring from the triple point separating the higher velocity
flow, which has passed through the lambda structure. The leading leg of the lambda
structure has a 450.5 shock angle and this region is seen to have reduced total
pressure losses. The schlieren picture gives a triple point height of approximately
26mm, as shown by the abrupt change in shock angle, see fig. 7.1a. Referring to figure
7.1b, the total pressure plots measured 90mm downstream of the shock position
corresponds to a triple point height between 27 and 28mm, where the total pressure ratio
rises considerably above the theoretical value behind a normal shock. The triple point
height difference is expected to be created from the shear layer lifting away from the
surface with downstream distance. The boundary layer thickness is seen to be
approximately 13mm.

Figure 7.2 Oil Flow visualisation of the uncontrolled M=1.5 SBLI, Holden (2004).


7-191
Referring to figure 7.2, the flow is from left to right showing the separation line (S) and
the position of reattachment (R) with oil flow visualisation. The separation line is
located at the start of the lambda structure, as seen in figure 7.1, with the reattachment
positioned after the rear of the lambda structure. The separation line is seen to be
relatively normal to the free stream flow with slight three-dimensional effects as the
floor and side wall boundary layers interact.

It should be noted that Holdens work was conducted at a different stagnation pressure
and with a different second throat. Therefore, any comparison with the unimorph
control can only be qualitative. It was not possible to repeat the uncontrolled
experiments due to time constraints at the Cambridge facility.

7.2. - Unimorph Controlled SBLI

7.2.1. - Schlieren and Total Pressure traverses
Figures 7.3a & 7.4a have similar upstream oblique shocks that originate from the start
of the test section control plate. The first oblique shock is observed to remain at 42,
which suggest that it is a Mach wave, referred to as a spurious wave in section 7.1a.

The second oblique shock is formed by the presence of localized boundary layer
thickening created by fluid injection into the boundary layer across the unimorph sides.
It is weaker in the schlieren than the first oblique shock, as seen by the lighter line. The
second oblique shock angle is seen to increase with an applied 500V voltage, closed
flaps, as the reduced unimorph deflection has decreased fluid injection across the end of
the unimorph. This is expected to result in an increased mass transfer across the slots,
leading to increased boundary layer thickening and a stronger shock. However, this
change is of the order of 1 due to the small amount of slot effect provided. The
slots are 0.2mm wide compared to the 7.54mm oncoming boundary layer thickness.

The bottom of the normal recovery shock is seen to bifurcate with a shear layer forming
and continuing downstream of the triple point. The angle of the leading leg of the
lambda structure changes from 54 (open flaps) to 56.5 (closed flaps), see figs. 7.3a
& 7.4a and table 7.1. With closed flaps the smaller orifice area reduces mass transfer.

7-192
This creates a more normal leading leg due to increased boundary layer thickening.
Also, the leading shock of the lambda structure is more smeared with unimorph control,
compared to the uncontrolled case, indicating three dimensionality of the flow. The
increased smeared structures, recorded with schlieren, reduce the accuracy of the angle
measurements.

a)
b)
Figure 7.3 a) Schlieren pictures and b) Total pressure traverses of
unimorph control of the M=1.5 SBLI with closed flaps (+500V).


7-193
a)
b)
unimorph control of the M=1.5 SBLI with open flaps (-500V).

Shock Angle ()
Second Oblique
Shock Angle ()
Leading Leg
Shock Angle ()
Uncontrolled 43.0 0.5 - 45.0 0.5
+500V Closed flaps 42.0 0.5 44.0 0.5 56.5 1
-500V Open flaps 42.0 0.5 43.0 0.5 54.0 1
Table 7.1 Summary of shock Angles for the uncontrolled
and unimorph controlled SBLI at M=1.5
1
.
1
The accuracy of these values has been calculated by taking 4 separate samples including the maximum
and minimum. Then a 2-tale test is applied with 90% confidence.

7-194
The triple point height and boundary layer thickness, as calculated from the total
pressure traverses, have been tabulated for the different spanwise position for ease of
comparison of the trends created with piezoelectric actuation/unimorph deflection, see
table 7.2. It should be noted that the uncontrolled values have been equated with the
nearest traverse data available. It is observed that the height of the triple point on the
centre line, Z=0, is approximately 27.5mm and 30.0mm for closed flaps and open
flaps respectively, compared to the 26mm uncontrolled triple point. This trend is
expected because the open flaps configuration produces a greater level of control,
mass transfer and smearing of the shock. The increased triple point height, and
therefore reduced total pressure loss, indicates the efficacy of using smart unimorph
flaps. However, the level of control is small due to the small amounts of strain, and
hence deflection, achievable by the present piezoelectric material, whose limitations are
discussed in chapter 4. With closed flaps the increased triple point height, compared
to the uncontrolled case, is believed to be created from mass injection across the end of
the flap. This at best will provide passive lateral slot control if zero deflection is
achieved. Unfortunately no measurements could be taken of the actual tip deflection
achieved during a wind tunnel run due to the confined area within the sealed plenum
chamber. Furthermore, no intrusive measuring system could be used externally to the
plenum as this would interfere with and change the flow structure over the test section.

Uncontrolled +500V
Closed flaps
-500V
Open flaps
Z
(mm)
TP
(mm)
(mm)
TP
(mm)
(mm)
TP
(mm)
(mm)
0 26.0 13.0 27.5 16.5 30.0 16.5
9 26.0 13.0 28.0 18.0 28.0 18.0
13 26.0 12.5 28.0 20.0 28.0 20.0
22 26.0 12.0 24.0 18.0 24.0 18.0
Table 7.2 Summary of triple point heights and boundary layer thicknesses
for flap controlled M=1.5 SBLI at various spanwise positions and flap deflections
as determined from figs. 7.3a & 7.4b.


7-195
The boundary layer thickness on the centreline, as determined from the total pressure
traverses for both flap positions, is approximately 16.5mm. This is 3.5mm larger than
the 13mm uncontrolled SBLI boundary layer due to the presence of the flaps.
The lambda structures at Z=9 and Z=13 are similar for both flap positions suggesting
that the longitudinal slot passive control is the driving influence over the small amount
of unimorph control available. The triple point height and boundary layer thickness at
Z=9, which is on the unimorph and 4mm from the side, is 28mm and 18mm
respectively for both flap positions. The triple point and boundary layer thickness at
Z=13, on the side of the unimorph (longitudinal slots), are 28mm and 20mm
respectively for both flap positions. It is expected that the thicker boundary layer, at
Z=13 compared to Z=9, is created due to its proximity to the longitudinal slot, which
provides increased mass transfer.

The triple point and boundary layer thickness at Z=22mm, which is 9mm outside of the
unimorph control region, is 24mm and 18mm respectively for both flap positions. It is
believed that this SBLI is comparable to an uncontrolled region as there is minimal
unimorph control effects experienced at this spanwise position. However, the
differences between the uncontrolled region on the unimorph control plate and the
uncontrolled case, from Holden (2004), are expected because of he different tunnel
settings for the two cases.

Oil Flow experiments were conducted with a 0V applied voltage for safety reasons. A
500V applied field could not be applied as it was thought that the oil will act as a
conductor and earth the power supply through the tunnel and instrumentation.

Referring to figure 7.5, the oil flow is smeared, especially at the end of each flap, due to
the tunnel shut down removing the aeroelastic deflection
2
. The flow is from left to right
with the main shock located above the start of the downstream unimorph. The figure
has been divided into three distinct regions: a unimorph control region bounded by the

2
During experiments the author ensured the oil flow on the exposed control surface dried. However,
during runs a small amount of oil flow mixture would enter the plenum via the downstream flap. This oil
flow mixture would not be exposed to the main flow and subsequently not dry. On tunnel shut down the
small amount of mixture would be forced back on to the control surface via the downstream flap.

7-196
sides of the unimorphs; two longitudinal slot control regions; and two uncontrolled
region outside of the longitudinal slots.

Figure 7.5 - Oil Flow visualisation of unimorph controlled SBLI at M=1.5with 0V.

The separation point for the localised unimorph control is located at the end of the
upstream unimorph flap, the start of mass injection into the boundary layer. This will
lock the unsteady shock into position, with the foot of the leading leg anchored at
this position. This fixing of the shock position would minimise the unsteady nature of
the shock. The separation bubble, compared to the uncontrolled SBLI, has
approximately doubled in length as the reattachment point is at the trailing edge of the
downstream unimorph - the point of mass suction. This produces a larger lambda
structure and hence improves the reduction in total pressure losses.

Horseshoe vortices can be seen at the start of the upstream unimorph longitudinal slots.
This is due to the mass injection across the sides of the unimorph. As discussed in
chapter 2, the flaps were machined into the aluminium substrate with 0.2mm cuts,
which create these slots. The horseshoe vortices indicate the presence of slot control, as
described in Smith et al. (2002). However, the amount of control produced by the slots
is minimal due to their relative size compared to the oncoming boundary layer.


7-197
7.2.3 - Discrete Pressure Measurements
a)
b)
c)
Figure 7.6 Discrete pressure data of unimorph control for the M=1.5 SBLI
a) closed flaps (+500V), b) inert flaps 0V & c) open flaps (-500V)

7-198
Non-dimensional surface pressures were measured at various locations (see fig. 3.11a)
and compared to the uncontrolled case. Non-dimensional distances are shown relative
to the main shock position with a negative distance implying upstream. Only a small
number of pressure taps were available within the unimorph control region.

The pressure trace for unimorph control is very similar for all three flap positions, see
figs. 7.6a to c. The initial pressure ratio upstream of the control region is the same as
that for the uncontrolled case, followed by a pressure rise at the start of the control
region due to mass injection through the longitudinal slots, whose influence extends
further into the uncontrolled region than oil flow visualisation suggests. The pressure
dip just before the main pressure rise is more smeared than in the uncontrolled case as
verified from the oil flow visualization. This is believed to be due to the boundary layer
thickening occurring over a longer streamwise distance. The main pressure rise with
unimorph controlled SBLI is approximately the same shape as that for the uncontrolled
SBLI tending to the same downstream values. However, the main pressure rise is
located increasingly upstream with greater unimorph deflection.

It is expected that an increase in the level of control will be achieved with -500V, open
flaps, due to higher mass transfer through the plenum chamber. It may also be noted
that the very slight dip observed at X/
o
= -K, the knee of the curve, could be the effect
of the rearward facing step generated by the deflected unimorph. The rearward facing
step generates a localised region of expansion and possibly separation, which grows
with increased deflection.

7.2.4. - Unimorph Control Limitation
The pressure data show that the differences between the open and closed flaps cases
are minimal, see figs. 7.6a & c. This is due to the small magnitudes of unimorph
deflection achievable. Clearly, greater deflection is required to achieve optimal
unimorph control. This requires a piezoelectric material with greater strain production,
which is currently unavailable (see Crawley (1994)).


7-199
7.3. Mechanically Set Flap Deflections 0mm
A series of runs were conducted at mechanically set flap deflections to evaluate the
performance achievable and to predict the amount of deflection required for optimal
control. A set deflection of 0mm was examined first and it was intended that it would
provide similar characteristics to the uncontrolled and closed flaps cases and also to
verify if the unimorphs could be actively shut off.

7.3.1. - Schlieren and Total Pressure Traverses
Referring to figure 7.7a, the presence of spurious waves can again be observed, which
as previously indicated, are characteristics of the experimental facility. The second
oblique shock is formed by boundary layer thickening above the longitudinal slots and
is a localised effect.

The bottom of the normal recovery shock is seen to bifurcate with a leading leg angle of
46 1 and a vortex sheet emanates from the triple point. The increase in leading leg
shock angle, from the uncontrolled 45 1, is believed to be the result of the mass
injection across the end of the flap structure, producing passive control. The schlieren
picture indicates that 0mm deflection was not achieved using the piezoelectric actuators.

Three dimensionality can still be observed with 0mm deflection with a more smeared
leading shock of the lambda structure, compared to the uncontrolled case, see fig. 7.1.
The two observed trailing legs are attributed to the flap and slot control regions, as
discussed in section 7.2.2.

Referring to figure 7.7b, it is observed that the triple point height and boundary layer
thickness on the centre line, Z=0, is 25.5mm and 14.5mm respectively, compared to the
approximate 27mm and 13mm for the uncontrolled case. The slightly reduced triple
point height is the result of a reduced upstream influence with the mass injection at
approximately 1.5L
o
compared to the uncontrolled case where the leading foot is
positioned at approximately 2L
o
. The increased boundary layer thickness is attributed to
the presence of the lateral slot creating a roughened surface. The combination of
reduced triple point height and increased boundary layer thickness would produce an

7-200
increase in total pressure loss and is, therefore, less efficient than the uncontrolled case
as regards wave drag.

a)
b)
Figure 7.7 a) Schlieren pictures and b) Total pressure traverses
of the M=1.5 SBLI with flap control at 0mm deflection.

From the traverses, the triple point and boundary layer thickness at Z=13, on the side of
the flaps (longitudinal slots), are 25mm and 18.5mm respectively. At this position the
amount of mass transfer is greater than at Z=0 due to the larger streamwise porosity of
the longitudinal slots. Compared to the centre line, the mass transfer into the
boundary layer commences further upstream of the interaction, approximately 8
o
.

7-201
The triple point height and boundary layer thickness at Z=9, which is on the flaps and
4mm from the side, are 25mm and 17mm respectively. At this spanwise position SBLI
characteristics are influenced by both the longitudinal and transverse slots at the sides
and end of the flaps respectively.

The 23.5mm triple point height and 16mm boundary layer thickness at Z=22, 9mm
(approximately one oncoming boundary layer thickness) outside of the flap region,
show the influence of the passive control on the global uncontrolled flow field. This
influence is reduced, as expected, with greater distance from the flap. The author
suggests that this SBLI is quasi-uncontrolled as there is minimal flap/longitudinal slot
control effects experienced at this spanwise position.

The triple point height and boundary layer thickness at Z=30mm, which is 13mm
outside of the flap control region, are 18.5mm and 11.5mm respectively. At this
position, approximately two oncoming boundary layer thicknesses outside of the flap
control region, the flaps have minimal effect on the SBLI. The observed differences,
compared to the Z=22 SBLI, are attributed to the floor and sidewall SBLI.

Figure 7.8 - Oil Flow visualisation of the M=1.5 SBLI
with 0mm flap deflection control.

The flow is from left to right with the main shock located above the start of the
downstream unimorph, see fig. 7.8. The figure has been divided into three distinct

7-202
regions: a flap control region bounded by the sides of the unimorphs; a longitudinal
slot control region above the flap sides; and an uncontrolled region outside of the
longitudinal slot region.

In the uncontrolled region, juxtaposed to the flap, a separation and reattachment line is
produced from the shock interaction with the boundary layer. Furthermore, a separation
structure can be observed when the shock interacts with the tunnel side wall, as
indicated at the bottom of figure 7.8.

The separation point for the flap control region is located at the end of the upstream
flap, which is the start of mass injection into the boundary layer. The mass injection
across the end of the flap will lock the front shock into position and minimise the
unsteady SBLI characteristics. The separation bubble, compared to the uncontrolled
SBLI separation, has approximately doubled in length as the reattachment point is at the
end of the downstream unimorph, where suction takes place. The larger separation
region would produce a more smeared lambda structure leading to reduced total
pressure loss.

The control effects of the longitudinal slots is minimal due to their size compared to
the oncoming boundary layer thickness. Also, no horseshoe vortices can be observed at
the upstream end of the slots, suggesting reduced passive SBLI slot control.

7.3.3. - Discrete Pressure Measurements
Non-dimensional surface pressures were measured at various locations, (fig. 3.11b), and
compared to the uncontrolled pressure distribution. Non-dimensional distances are
shown relative to the main shock position with a negative distance implying upstream.
It can be seen in figure 7.9, with a 0mm deflected flap the pressure rise is similar to the
no control case. However, some significant differences are observed: a slight pressure
rise can be seen at the start of the control region indicating the presence of some mass
injection; a pressure dip can be observed between X/
o
=-4.7 and X/
o
=-3.3, indicating
localised thickening of the boundary layer; a further dip can be observed at X/
o
=-K
across the end of the upstream flap, which is likely due to boundary layer thickening
created by the mass injection across the lateral slot, located at X/
o
=-1.3.

7-203
Figure 7.9 Discrete pressure data of flap control of the M=1.5 SBLI
with 0mm flap deflection control.

7.4. - Mechanically Set Flap Deflections 1mm to 3mm

It is clear that when flap deflections are large (>1mm) the concept of longitudinal slot
control becomes invalid and the flow structure becomes complex and three dimensional.

7.4.1 Schlieren and Total Pressure Traverses
Figures 7.10 to 7.12 have similar upstream oblique shocks that originate from the start
of the test section control plate, as discussed earlier. The leading oblique shock is
observed to remain at 42.5 regardless of flap deflection, which is slightly larger than
the theoretical 41.8 Mach angle at M = 1.5.

The second oblique shock is formed by the presence of localized boundary layer
thickening, created by fluid injection into the boundary layer across the flap sides. This
is a localised spanwise effect and, therefore, the second shock appears weaker in the
schlieren than the first oblique shock, as seen by the lighter line. Referring to figure
7.10a, the second oblique shock angle for flap deflections of 1mm and greater is
constant at 42, which is significantly greater than the 39 with 0mm flap deflection. It
is expected that flap deflection increases the flap side orifice area allowing greater
transverse fluid injection, leading to increased boundary layer thickening and a stronger

7-204
shock. This is a reversal from the unimorph controlled case, where we note an increase
in the second oblique shock angle with flap closure.

a)
b)
the M=1.5 SBLI control with 1mm flap deflection.

The bottom of the normal recovery shock is seen to bifurcate with a vortex sheet
emanating from the triple point. Referring to figure 7.10a and 7.11a, the angle of the
leading leg of the lambda structure for 1mm and 2mm flap deflections is 52, which is
greater than for 0mm flap deflection. Flap deflection allows mass transfer to occur,
producing greater boundary layer thickening. This is different from the unimorph
control case where the leading leg has an angle of 56.5. It is possible that with closed
flaps configuration and nominal zero deflection certain mass transfer (leakage) occurs

7-205
normal to the surface. This would then increase boundary layer thickening to create the
more oblique shock angle of 56.5. As the flap deflects more, the mass injection angle
is tilted towards the rear producing less boundary layer thickening resulting in a lower
leading leg shock angle.

a)
b)

With higher deflections, the flow becomes increasingly three dimensional. This leads to
smearing of the lambda structure as seen by the schlieren pictures (fig. 7.12a). There is
also an indication that the leading leg has split into two; one due to the compression
ramp effect and the other due to the mass injection.


7-206
The pressure traces for 2mm and 3mm (figs. 7.11b and 7.12b) indicate that the SBLI
becomes increasingly unsteady with flap deflection. The author is unsure of the reason
for the observed variations. The unsteady nature is exacerbated by the movement of the
pitot traversing through the boundary layer, producing unrepeatable results below
Y=12mm for Z=0. However, they have been included as they highlight the sensitive
nature of the interaction whilst maintaining a steadiness to the larger structure.

a)
b)

The shock angle data has been tabulated for ease of comparison of the trends created
with flap control of the SBLI, see table 7.3. It should be noted that the uncontrolled

7-207
case does not have a second oblique shock angle as this is a feature of the unimorph
control.

Shock Angle ()
Second Oblique
Shock Angle ()
Leading Leg
Shock Angle ()
Uncontrolled 43.0 - 45.0
0mm 42.0 39.0 46.0
1mm 42.5 42.0 52.0
2mm 42.5 42.0 52.0
3mm 42.5 42.0 -
Table 7.3 Shock Angles for the uncontrolled and
flap controlled SBLI at M=1.5, (accuracy of 1).

The triple point height and boundary layer thicknesses have been tabulated for the
different spanwise position for ease of comparison of the trends created with flap
control of the SBLI, see table 7.4. It should be noted that the uncontrolled values have
been equated with the nearest traverse data available, for example Z=8 values have been
applied to Z=9. The triple point height and boundary layer thicknesses, as a general
rule, are seen to increase monotonically with increased flap deflection up to 2mm.

The triple point height is observed to be greatest along Z=13 where the maximum mass
transfer occurs along the flap sides. The boundary layer thickness is observed to be
greatest along the centreline, Z=0, suggesting that the surface curvature, with flap
deflections greater than 1mm, is dominant. The triple point height and boundary layer
thickness are then smaller with greater spanwise distance from the flap, as the flap
influence is absorbed into the neighbouring flow field.

Uncontrolled 0mm Deflection 1mm Deflection 2mm Deflection 3mm Deflection
Z
(mm)
TP
(mm)
(mm)
TP
(mm)
(mm)
TP
(mm)
(mm)
TP
(mm)
(mm)
TP
(mm)
(mm)
0 26.0 13.0 25.5 14.5 28.5 21.5 28.5 20.0 - 24.5
9 26.0 13.0 25.0 17.0 29.5 19.5 30.0 20.0 - 24.5
13 26.0 12.5 25.0 18.5 31.0 17.5 30.0 17.5 - 21.0
22 26.0 12.0 23.5 16.0 24.5 16.5 31.0 17.0 - 20.0
30 26.0 12.0 18.5 11.5 21.0 12.5 21.5 13.0 35.0 14.0
Table 7.4 Summary of triple point heights and boundary layer thicknesses for
flap controlled SBLI at M=1.5 for various Z positions and flap deflections
as determined from figs. 7.6 & 7.9a to c, (accuracy of 1mm).


7-208
It should, however, be cautioned that, as seen from table 7.4, the measurements of the
triple point height and boundary layer thickness are not exact because of the
complicated nature of the flow around the flaps and also the fact that the flow
boundaries are not easily defined.
3
The numerical values should, therefore, be
considered indicative.

The lambda structure, with a 3mm deflected flap, could not be fully mapped with a total
pressure traverse due to the unsteady nature of the large separation bubble that is
produced and propagates downstream of the control region (fig. 7.12b). The flow
unsteadiness could also suggest that optimal control is not achieved with a 3mm flap
deflection.

7.4.2. - Total Pressure Maps
The total pressure maps clearly show the influences on the SBLI structure with
increased flap deflection. At 0mm flap deflection the triple point is slightly raised to a
relatively uniform level over the flap compared to the uncontrolled region at Z=22. The
dominating SBLI control experienced from the longitudinal slots has a localised
influence due to their relative size with respect to the oncoming boundary layer, see fig.
7.13a. The slots create a thicker boundary layer, as shown by the dark blue to red
transition, due to the mass injection.

With 1mm flap deflection (fig. 7.13b) the boundary layer thickness has increased across
the span due to mass transfer and flap deflection. The overall effect is a small reduction
in total pressure loss as the triple point height has lifted by a similar amount to the
boundary layer. Although, the higher triple point indicates that the shock is more
smeared.

With 2mm flap deflection (fig. 7.14a) the control effects of the slots and flaps have
combined to increase the boundary layer thickness and triple point height. The results
indicate that the total pressure loss has increased (consequently increasing the drag
coefficient) as compared to the 1mm deflection case. This is attributed to the benefit of

3
The accuracy of these values has been calculated by taking 4 separate samples including the maximum
and minimum. Then a 2-tale test is applied with 90% confidence. A 95% confidence would have
produced an accuracy of 1.6mm. The author did not measure the temporal variations at a single location.

7-209
the higher triple point being negated by a reduction in the total pressure recovery behind
the SBLI. The author speculates that a small amount of separation has occurred, which
is suggested by the respective pressure traverse, see fig. 7.11b. This indicates that a flap
deflection between 1mm and 2mm may be optimal before incipient separation.

a)
b)
Figure 7.13 Total Pressure Maps for flap control of the SBLI


7-210
A 3mm deflection produces a thicker boundary layer with a reduced integrated area of
total pressure loss and increased unsteadiness. This suggests that although a larger
smearing of the shock takes place the SBLI control performance has degraded from that
produced with 2mm flap deflection.

a)
b)
Figure 7.14 Total Pressure Maps for flap control of the SBLI


7-211
a)
b)
c)
Figure 7.15 - Oil Flow visualisation of the M=1.5 SBLI control with
flaps deflected to a) 1mm, b) 2mm and c) 3mm.

Increased flap deflection increases the amount of mass injection across the sides of the
upstream flap, which produces transverse mass injection and separation line around the
slots, see fig. 7.15a. Initially, with 1mm flap deflection, the transverse flow is
constrained by the downstream flap and sucked into the plenum chamber. However,

7-212
with 2mm flap deflection the transverse mass injection dominates the surface flow and
the mass suction is insufficient to contain it, so that the separation line continues
downstream of the control region, see fig. 7.15b. At 3mm deflection, this effect is
increased with a wider separation line continuing downstream, see fig. 7.15c.

Reversed flow is produced on the downstream flap due to the presence of a localised
separation region initiated at the trailing edge of the upstream flap and is believed to
terminate on the downstream flap. This separated region produces a smaller separation
line, within the larger separation line, as the reversed flow interacts with the streamwise
mass injection across the end of the upstream flap.

A deflected upstream flap produces a finite ramp/compression corner, which generates
stronger counter rotating vortices from either side with increased deflection. Mass
injection across the sides of the upstream flap pushes & strengthens the co-rotating
vortices, produced from either side of the flaps, up into the flow and away from the
flaps. These vortices would then be strengthened by mass suction of the downstream
flap.

7.4.4. - Discrete Pressure Measurements
Surface pressures were measured at various locations, (fig. 3.11b), normalised to the
upstream stagnation pressure and compared to the uncontrolled case. Non-dimensional
distances are shown relative to the main shock position with a negative distance
indicating an upstream position of the shock.

The initial pressure ratio upstream of the control region, for all flap deflections, is the
same as that for the uncontrolled case, indicating similar oncoming flow conditions and
minimal upstream influence. A pressure rise is then observed at the start of the control
region due to the higher pressure mass transfer from the plenum across the sides of the
upstream flap, see fig. 7.16a. There is a slight dip in pressure ratio for a 1mm flap
deflection just before the main pressure rise, which is due to boundary layer thickening.
At increased flap deflection the compression ramp effect negates the pressure dip and
increases the pressure ratio upstream of the main pressure rise, especially above the
flaps (Z=0 to Z=12), see figs. 7.16a to c.

7-213
a)
b)
c)
Figure 7.16 Discrete pressure data for the M=1.5 SBLI control
with flaps deflected to a) 1mm, b) 2mm and c) 3mm.


7-214
Increased flap deflections move the leading leg of the lambda structure upstream, as
seen by the main pressure rise across the shock occurring further upstream, see figs.
7.16a to c. The pressure ratio is seen to reach the uncontrolled value further
downstream with increased deflection as the lambda structure and separation bubble
grow.

A pressure plateau is observed to grow in length with increased flap deflection
suggesting a growth in separation region. This is also seen from the oil flow
visualisation in figures 7.8 and 7.15a to c. Furthermore, at 1mm flap deflection the
pressure plateau (separated region) is seen to be localized on the downstream flap, with
the Z=17.5 trend not having a significant pressure plateau, see fig. 7.16a. However, the
separated region widens with 2mm flap deflection generating a pressure plateau on the
Z=17.5 trend, see fig. 7.16b, and further spreads to Z=21.5 with 3mm flap deflection,
see fig. 7.16c.

7.4.5. - Velocity Traverses
The velocity traverses have been non-dimensionalised to the uncontrolled flow velocity
at the edge of the boundary layer, U
e
= 295ms
-1
. A 0mm flap deflection produces
similar velocity traverses over the flap at Z=0 and Z=9, suggesting similar amounts of
control and little influence from the longitudinal slots at these spanwise positions, see
fig. 7.17a. The Z=13 traverse, on the longitudinal slot produces a thicker boundary
layer, due to increased upstream mass injection. A velocity difference is observed at
about the triple point, above which the flow seems to be retarded. The velocity
difference creates a shear layer between the two regions. The Z=22 traverse, 9mm from
the flap , shows a fuller boundary layer with a lower shear layer than between Z=0 and
Z=13. The author suggests that the slot influence is being dissipated into the global
flow field and at Z=30 the boundary layer is seen to be thinner as the flow tends further
to the uncontrolled SBLI. The absence of a higher retarded flow region at Z=30, does
not necessarily signify the absence of lambda structure. However, it is possible that the
flow in this region is being influenced by the sidewall boundary layer.


7-215
a)
b)
Figure 7.17 Velocity traverses for flap control of the SBLI

Referring to figure 7.17b, from Z=0 to Z=13, the profile seems to be relatively uniform
for the 1mm flap deflection with an increased boundary layer thickness and triple point
height compared with 0mm flap deflection, suggesting a uniform degree of control. At
Z=22 the triple point has risen compared to the 0mm flap deflection. This is attributed

7-216
to the growing influence of flap control on the neighbouring flow field with flap
deflection. The velocity distribution at Z=30 has not significantly changed from 0mm
flap deflection, suggesting that it is not being influenced by the flap control. Moreover,
a degree of unsteadiness is observed along the flap centreline.

a)
b)
Figure 7.18 Velocity traverses for flap control of the SBLI

7-217
Deflecting the flap to 2mm raises the triple point at Z=22 to the quasi-uniform level of
Z=0 to Z=13, which is of a similar height to the 1mm flap deflection values. The
boundary layer is observed to become thinner with distance from the centreline, as the
control influence is initially merged with that of the flap side and then with the
uncontrolled structure. Again, the velocity distribution at Z=30 has not significantly
changed from the 0mm flap deflection, suggesting that it is not affected by the flap
deflection. The unsteadiness of the local boundary layer is seen to increase in
amplitude.
4
Increasing the flap deflection to 3mm causes the boundary layer to have a uniform
thickness from Z=0 to Z=13, which becomes thinner with increasing spanwise distance.
At Z=30 the boundary layer has thickened and a gradual retardation can be observed
form Y=10 to Y=35. The outer flow appears to be uniform at all spanwise positions
with no apparent triple point beneath Y=37. The unsteadiness of the flow within the
boundary layer is more apparent with 3mm flap deflection, compared to 2mm, as the
separation region is propagated downstream of the control region. It is possible the
SBLI instabilities observed with 2mm and 3mm flap deflections might indicate that
1mm flap deflection would be nearer the optimum for a normal shock at M=1.5.

7.5. - Normal SBLI Control at M=1.5 Summary

The uncontrolled M=1.5 SBLI, with an unswept normal shock interacting with the
naturally grown boundary layer on the tunnel floor, indicates separated flow. A vortex
sheet is formed from the triple point separating the higher velocity flow, which has
passed through the lambda structure, from the flow that went across the normal shock
resulting in smaller total pressure losses. The uncontrolled SBLI is largely two
dimensional with small three-dimensional effects near the side wall.

The presence of unimorph control creates an oblique shock that is formed by the
presence of localized boundary layer thickening created by fluid injection. This is due
to passive control via longitudinal slots produced as a result of fabrication. Greater

4

7-218
SBLI control is produced with increased unimorph deflection. However, the level of
control is small due to the small amounts of strain, and hence deflection, achievable by
the piezoelectric material.

In order to determine the deflection required for optimal unimorph control a series of
runs was conducted at mechanically set flap deflections to evaluate the performance
achievable. The 0mm flap deflection was shown to produce a quasi- uncontrolled SBLI
with deviations created from mass transfer across the sides and end of the flap structure.
However, the control provided by the longitudinal slots is minimal due to their size
compared to the boundary layer thickness. Furthermore, it was inferred that 0mm
deflection was not achieved using the piezoelectric actuators in the closed flaps
configuration.

At flap deflections of 1mm, and greater, the concept of longitudinal slot control
becomes invalid as the mass transfer is such that the flow becomes three dimensional.
Also with increased flap deflection levels, at and above 1mm, the angle of the leading
leg and the size of the lambda structure increases due to the increased mass transfer.
Flap deflection induces transverse mass injection across the flap sides, which is initially
constrained by the downstream flap and sucked into the plenum chamber. However, at
flap deflections greater than 1mm, the transverse mass injection dominates the surface
flow and the downstream mass suction is insufficient to contain it. The separation
produced is observed to continue downstream of the control region and influence the
global flow field.

With 1mm flap deflection the increased boundary layer thickness is comparable to that
produced by the slots. However, the two forms of control can still be identified and
their combination produces a larger region of reduced total pressure loss. At 2mm flap
deflection the control effects of the slots and flaps have combined and cannot be
isolated from one another, producing an increased region of reduced total pressure loss
with slight unsteadiness within the lower part of the boundary layer. This would
suggest flap deflection between 1mm and 2mm may be optimal.


7-219
A 3mm deflection produces a considerably larger region of reduced total pressure loss
but with increased unsteadiness in the boundary layer. However, the overall pressure
loss appears to be greater than with 2mm flap deflection, suggesting a higher drag.

7.6. Normal SBLI Control Summary

It is established that the uncontrolled SBLI with a normal Mach number of 1.3 and 1.5
are incipiently separated and separated respectively. Unimorph control generates an
oblique shock, which is formed by localised boundary layer thickening due to fluid
injection across the flap sides, the longitudinal slots, and the bottom of the normal
recovery shock is seen to bifurcate. In the closed flaps unimorph configuration, the
orifice area is small, producing minimal mass transfer across the end of the unimorphs.
The increased mass transfer, with unimorph deflection, will promote boundary layer
thickening further upstream compared to the closed flaps case. It is shown that
deflection promotes the mass injection across the end of the upstream flap and reduces
the amount across the sides. It is suggested that the presence of unimorph control will
lock the unsteady shock into position and minimise the unsteady characteristics of the
shock, such as noise, buffet, etc.

A set flap deflection of 0mm, at both Mach numbers, produces a quasi-uncontrolled
SBLI, with a thicker boundary layer, a smaller lambda structure along the flap
centreline and a localised larger slot induced lambda structure. Schlieren pictures
indicate that 0mm deflection was not achieved using the piezoelectric actuators with
closed flaps (500V).

With flap deflections of 1mm, and greater, the concept of longitudinal slot control is
invalid because the mass transfer, across the sides of the flaps becomes three
dimensional. Increased flap deflections produce a higher pressure rise upstream of the
main pressure rise due to the increased surface curvature and larger mass transfer, which
increasingly influences the global flow field. This results in increased upstream
influence, a larger separation bubble under a wider lambda structure and a reduced final
total pressure ratio.


7-220
It is not clear from the flap experiments at set deflections that optimal control for the
unswept normal SBLI at M=1.3 has been produced within the 0mm to 3mm deflection
range. The author suggests that a deflection between one and two millimetres does, in
fact, provide optimal control.

For an unswept normal SBLI at M=1.5, the SBLI instabilities with flap deflection above
2mm suggest a flap deflection between one and two millimetres is optimal for reducing
total pressure loss without any SBLI unsteadiness produced due to separation. A 3mm
deflection produces a considerably larger region of reduced total pressure loss with
increased unsteadiness in the thicker boundary layer.

Chapter 8. Theoretical Optimisation

8-221
CHAPTER 8. Theoretical Optimisation

This chapter provides a theoretical model of shock wave/boundary layer interaction
(SBLI) control using an upstream and downstream flap. This has been performed to
assist in the understanding of the control influences produced in the SBLI. It is applied
to the MART system of Hafenrichter et al. (2003) as well as to validate and improve the
understanding of unimorph flap control. From the optimal flap deflections obtained
from the theory and experiments in chapters 5 to 7, the strength of piezoelectric material
required for optimal control are then predicted using the structural theory of chapter 4.

8.1. - Theoretical 2-D Aerodynamic Model

Figure 8.1 Theoretical SBLI Flap Control Model

Theoretical modelling of interaction control using flaps first examines the flow
downstream of the interaction. Referring to figure 8.1, it is assumed that the
downstream flap does not deflect, which in practice will deflect into the plenum
chamber, see fig 8.2a. However, the deflection is ignored as the boundary layer is close
to separation and, therefore, will have minimal streamwise velocity at the reattachment
point on the downstream unimorph. This allows the mass suction to be modelled as a
result of the pressure difference acting across a gap, which increases proportionally with
streamwise length, L
G
, see fig 8.2b.


8-222
a) b)
Figure 8.2 a) Actual Downstream Flap Tip Deflection
and b) Modelled Downstream Flap Tip Deflection

Using conservation of mass, the mass injection upstream is assumed equal to the mass
suction downstream. The angle at which the mass flow is injected is calculated from
structural analysis and using trigonomic functions, the displacement thickness growth
angle and a leading shock angle for the lambda structure is predicted. Using this
leading leg shock angle and the distance from the point of mass injection to the inviscid
shock position a triple height can then be estimated.

8.1.1. - Downstream Mass Flow (suction) through a single hole/slot
The mass suction across the downstream unimorph is modelled using the perforated
plate model of Doerffer and Bohning (2000), see fig. 8.3. It is assumed that the mass
transpiration across the downstream flap is a product of the pressure drop across a slot
whose span is unity. The freestream conditions pertain to those existing downstream
of the shock and the driving pressure difference across the flap, dP, is half the pressure
rise through the main shock, the other half being across the upstream flap.

Figure 8.3 Flow configuration and definitions (after Doerffer and Bohning (2000))

The subscripts s, 0, e, w and h denote the conditions of the static, stagnation, freestream
(edge of the boundary layer), wall and the hole (slot) respectively. The pressure drop
across a single hole, irrespective of size, is given by Doerffer and Bohning (2000) as


8-223
( )
(
(
+
|
.
|
\
|
=
1
1
1 1
55 . 0
1
55 . 0
1
0
2 . 1
1
c
h h
h
M B d M
P
dP
[Eq. 8.1]
Doerffer and Bohning (2000)
where
2
2 1
h h
w
u
B
t
= [Eq. 8.2]

c
1
=1.52, d
1
=25, P
0h
=P
s
=P
2
and dP=0.5(P
2
-P
1
)
where the subscripts 1 and 2 designate the freestream conditions upstream and
downstream of the shock respectively. In equation 8.1., it is assumed that the stagnation
pressure on the high pressure side of the hole, P
0h
, is the reservoir pressure and equal to
the freestream static pressure above the hole, P
s
. This means that the slot is underneath
a large cavity so that stagnation conditions are present and any tangential dynamic
pressure has no effect. Such an assumption is made and justified by Doerffer and
Bohning. In the present circumstances this can only be justified for a slot of unit width
and for theoretical calculations this is not unreasonable.

If the wall shear stress, t
w
, goes to zero equation 8.1 can be simplified to

(
(
|
.
|
\
|
=
55 . 0
1
55 . 0
1
2
2 . 1
1
h
M
P
dP
, [Eq. 8.3]

or rearranging gives

( )
1
55 . 0
2
1
55 . 0
2
1 7 . 0 2 . 1 M f
P
P
P
dP
M
h
=
(
|
|
.
|
\
|
~
|
|
.
|
\
|
= , [Eq. 8.4]

where M
h
is the effective Mach number at which the fluid flows through the slot across
the entire dimension. Doerffer and Bohning place a limit that M
h
should be less than or

8-224
equal to 0.57 for their study of sub-millimetric diameter holes, after which choking
occurred. This limit has been removed as the slot area is considerably larger than that
of a sub-millimetric diameter holes.
*
The mass flow through the slot is then given as

( )
2
2
1 2
1
2
2
1
1
w
h
h
h h
RT
P
M
M
A u A m

+
|
.
|
\
|
+
= = & , [Eq. 8.5]

where A is the cross sectional area.

The mass flow per unit width is then given by

( )
2
2
1 2
1
2
2
1
1
w
h
h
G
RT
P
M
M
L
b
m
+
|
.
|
\
|
+
=
&
, [Eq. 8.6]

where b is the slot width and L
G
is the length of the slot gap, see fig. 8.1a.

8.1.2. - Total Slot Length and Mass Flow Injection Angle
a) b)
Figure 8.4 The Slot Created by Upstream Flap Deflection
a) dimensions and b) mass flow injection angle

Referring to figures 8.2 and 8.4, the total slot length, L
G
, is equal to the total gap due to
flap deflection, G
T
. This is calculated from the gap height, G
Y
, which equals the
unimorph flap deflection, A
U
, and the streamwise length, G
X
. G
X
is the initial slot width
at 0mm deflection due to the manufacturing process, G
I
= 0.2mm, and the decrease in

*
This only effects the unswept normal SBLI study at M
1
=1.5, whose M
h
is 0.615, see section 7.3.1

8-225
unimorph flap length along the X axis due to the flap deflection, AL
U
.
L
G
is calculated
using

( )
2 2 2 2
U I u X Y T G
L G G G G L A + + A = + = = [Eq. 8.7]

The mass flow injected upstream of the shock is equal to the mass flow removed from
downstream of the shock due to conservation of mass. The flow injection angle,
inj
m&
u , is
assumed to be normal to the deflected flap opening, see figs. 8.4b and 8.5a & b.
a) b)
Figure 8.5 - Mass injection a) without flap deflection and b) with flap deflection.

8.1.3. - The displacement thickness growth angle
Fugure8.6 Calculated Angles (Not to scale)

Referring to figure 8.6, the total displacement thickness growth angle, u
T
, can be
considered as the sum of the mass flow angle,
m&
u , the unimorph deflection angle, u
U
,
and the uncontrolled displacement thickness growth angle, u
o*
, using equation 8.8.

Ignoring the AL
U
term produces errors of less than 0.5% for the 3mm unimorph flap deflection
considered. The streamwise opening of the gap is made obvious in figure 7.9 for 1-3mm deflections

8-226
* o
u u u u + + =
U m T &
[Eq. 8.8]

where
U
U
U
L
A
= u [Eq. 8.9]

where L
U
is the unimorph length, see fig. 8.2b.

Figure 8.7 Model of Mass Flow Angles

Referring to figure 8.7 the mass flow angle, the angle at which the displacement
thickness grows due to the mass injection, can be calculated using the sine rule as

( )
inj
m
Total
injection
m
m
m
& &
&
&
u u = 180 sin sin [Eq. 8.10]

and

( )
inj
m injection injection Total
m m m m m
&
& & & & & u
o o
cos * * 2
*
2
*
2
+ + = [Eq. 8.11]

where the incoming mass flow within the displacement thickness is given by

|
|
.
|
\
|

(
(
|
|
.
|
\
|
=
5
7
2
5
7
2
2
5
7
4
5
7
2
3
5 4 *
*
1 ln 12
* 6 * 4 *
12
1
2
7
C C C C
C C m
o o o o
o
& [Eq. 8.12]

and

8-227
1
7
1
1 1
4
2
rU
C T
C
P
o
= [Eq. 8.13a]
2
1
7
2
1
5
2
rU
C T
C
P W
o
= [Eq. 8.13b]

where , T, C
P
, r, U, J and J* are the density, temperature, specific heat at constant
pressure, recovery factor, velocity, boundary layer thickness and displacement thickness
respectively. The subscripts 1 and W denote the upstream static and wall conditions
respectively. Equations 8.12 and 8.13a and b are derived in Appendix E1.

8.1.4. - Leading Leg Shock Angle (|)
Figure 8.8 Model for calculating the shock angle from
the total deflection angle for the upstream leg of the lambda shock

The leading leg shock angles is iteratively calculated using the attached oblique shock
equation, due to the total deflection angle, u
T
, given by Anderson (1990).

( )
(
+ +
=
2 2 cos 2
1 sin
cot 2 tan
2
1
2 2
1
11
|
|
| u
M
M
T
[Eq. 8.14]
(Anderson (1990))


8-228
where u
T
, and M
1
are the total displacement thickness growth angle, ratio of specific
heats and the upstream freestream Mach number respectively, see fig. 8.8.

8.1.5. - Triple Point Height
The triple point height, h
tp
, was assumed to be the combination of the unimorph
deflection, L
U
, the total boundary layer thickness at the end of the unimorph, o
T
, and the
height due to the leading leg angle to the main shock position, o
|
, see equation 8.15 &
fig. 8.9.

|
o o + + A =
T U tp
h [Eq. 8.15]

Figure 8.9 The Modelled Lambda Structure including heights

The total boundary layer thickness at the end of the unimorph, o
T
, is assumed to be the
uncontrolled oncoming boundary layer thickness, o
o
, increased by the uncontrolled
growth angle, u
o
, along the unimorph as calculated from Crocco and Lees (1952).

) tan (
o
u o o
U o T
L + = [Eq. 8.16]

The height due to the leading shock interacting with the main shock position, o
|
, is
calculated by the tangent of the shock angle, |, multiplied by L
|
, the distance upstream
of the main shock from where the leg originates, using

8-229

| o
| |
tan L = [Eq. 8.17]

where L
|
is the addition of the distance from the end of the unimorph flap to the
inviscid shock position, L
inj
, plus the upstream influence length due to the existence of
the shock, L
inf
, and of the mass injection,
m
L
&
, see fig. 8.10.

m inj
L L L L
&
+ + =
inf |
[Eq. 8.18]

Figure 8.10 The Modelled Lambda Structure including Lengths

The upstream influence due to the pressure rise across the shock on the boundary layer
is given by

( )
o
M L o 1
1 inf
= [Eq. 8.19]

Also, there will be an upstream influence of the injection and it is assumed that this will
vary linearly with unimorph flap deflection with the maximum upstream influence
length of 0.3o
o
at zero unimorph deflection and is assumed to be

( )
o u m
L o 1 . 0 3 A =
&
[Eq. 8.20]

This is because there will be no shock effect for a Mach wave.


8-230
An upstream influence limit of one boundary layer thickness is applied, however, this is
not required for the current analysis as the maximum possible upstream influence is
0.8J
o
with the unswept normal SBLI control at M
1
=1.5.

i m
L L o s +
& inf
[Eq. 8.21]

8.2. - Analysis of the MART System for an UNS Interaction at M=1.4

Analysis of the MART system assumes an upstream flow with a Mach number of 1.38,
a reservoir pressure of 210kPa, a boundary layer thickness of 2.6mm and an oncoming
displacement thickness of 0.36mm. This would create a pressure difference across the
downstream flap of 35.8kPa and with a skin friction coefficient (C
f
) of 0.0005 induces
suction across the end of the downstream flap at a Mach number of 0.51.

8.2.1. - The MART Six-Flap System
Applying the theoretical model to the MART system of Hafenrichter et al. (2003) and
using published data, the displacement thickness growth angles have been calculated for
given mesoflap deflections, see fig. 8.11.
Figure 8.11 Displacement Thickness growth angle variation for the MART
six-flap deflection


8-231
It can be seen from figure 8.11, that the mass injection effect initially reduces for
mesoflap deflections up to 1mm as the injection angle,
m&
u , becomes more oblique.
After 1mm deflection the mass injection influence increases at a constant rate. It
appears that the minimum total displacement thickness growth angle, u
Tmin
, for the six-
flap MART system is 23.4, which corresponds to a mesoflap deflection of 0.50mm or
1.39o
o
*. It should be noted that u
Tmin
is considerably larger than the maximum 9.3
deflection angle, for a Mach 1.4 flow with an attached oblique shock, as given by
Liepman and Roshko (1957). This would suggest that the leading leg of the lambda
structure would detach from the end of the flap and initiate further upstream.

a) b)
c)
Figure 8.12 - Schlieren pictures of
the MART Six-Flap System
with mesoflap thicknesses of
a)63.5m, b)101.9m and c)150.6m,
- Hafenrichter et al. (2003)
In the experiments of Hafenrichter et al. (2003) the 101.9m thick MART systems
produces optimal control with a deflection of approximately 0.484mm, as calculated by
the author in section 1.2.5. This would suggest that the minimum total displacement

Angles estimated by Author.


8-232
thickness growth angle provides optimal control. However, Hafenrichter et al. go on to
show that no six flap system produces overall beneficial effects compared to the solid
wall (uncontrolled) case.

From the published schlieren photographs it is observed that the six-flap MART system
generates numerous oblique shocks, resulting in a complex shock structure, see figures
8.12 a to c. It should be noted that the shock angles have been estimated for this
analysis, as they were not published.

The maximum deflection angle for an attached oblique shock at M=1.5 is approximately
12.1, which is considerably less than u
Tmin
. It is, therefore, assumed that the leading leg
is not attached to the end of the upstream unimorph, the point of mass injection. It
would appear that the leading leg of the lambda structure is generated from the
compression ramp effect of the deflected mesoflaps and the naturally growing boundary
layer.

From structural analysis it is calculated that mesoflap deflections of 2mm, 0.48mm and
0.15mm are produced with flap thicknesses of 63.5m, 101.9m and 150.6m
respectively, see section 1.2.5. Also, according to Hafenrichter et al. (2003), the second
flap is believed to deflect half the magnitude of the first flap, giving a deflection of
0.075mm for the second 150.6m thick mesoflap.

The oblique shock angle, or the leading leg angle as it would be, is calculated using a
deflection angle consisting of the mesoflap deflection combined with the displacement
thickness growth.
**
Referring to figure 8.13, the leading leg shock angles are observed
to closely agree with the schlieren when the mass injection is ignored. With this in
mind, it is not surprising that Jaiman et al. (2004) observed a beneficial effect of
mesoflap deflection without the presence of mass transfer, as the compression ramp
effect is present, see section 1.2.5. The fact that mass transfer improved the total
pressure recovery is believed to be due to boundary layer suction inhibiting separation,
which would otherwise continue downstream. The increase in measure experimental
leading leg shock angle with reduced mesoflap deflection is thought to be a combination

**
The effect of mass injection has been ignored.

8-233
of the presence of mass injection and an increased uncontrolled displacement
thickness growth angle. Furthermore, the local Mach number for the second 150.6m
thick mesoflap will be less than the modelled 1.38 due to the interaction of the first flap.
This will increase the leading leg shock angle of the second flap.

Figure 8.13 The leading leg shock angle variation for mesoflap deflection
of the MART six-flap deflection assuming no mass injection effect.

8.2.2. - The MART Four-Flap System
Performing a similar analyses on the four flap system, it can be seen that the minimum
total displacement thickness angle, u
Tmin
, for the four-flap MART system is 20.8,
which corresponds to a mesoflap deflection of 0.67mm or 1.86o
o
*, see fig. 8.14. This
amount of deflection is approximately achieved with the 190.5m thick MART system,
which is predicted to produce 0.6mm deflection, see section 1.2.5. Again, Hafenrichter
et al. (2003) showed that this was the optimal four flap system of those considered.
Moreover, it was also seen to produce an overall beneficial effect on the total drag
compared to the uncontrolled SBLI. It is believed that the reduced number of flaps,
compared to the six-flap MART system, has created a smoother surface to the flow,
reducing the viscous losses within the boundary layer.


8-234
Figure 8.14 Displacement Thickness growth angle variation for deflection of the
MART Four-flap system

Referring to figure 8.15, it is observed that four-flap system generates numerous oblique
shocks. These are believed to be generated primarily from the compression ramp effect
of the deflected mesoflaps as u
Tmin
is considerably larger than the maximum deflection
for an attached oblique shock. The leading leg angles have been estimated by the author
and structural analysis predicts the first and second flap on the 190.5m four flap
system will deflect 0.6mm and 0.3mm respectively into the flow, see section 1.2.5.
Figure 8.15 - Schlieren pictures of the MART Four-Flap System
with a mesoflap thickness of 190.5m Hafenrichter et al. (2003)
Angles estimated by Author.


8-235
Again ignoring the mass injection effect, the deflection angle is assumed to be the
combination of the mesoflap deflection and the uncontrolled displacement thickness
growth angles. The generated oblique shock (leading leg) angle is then observed to
agree closely with the schlieren photography, see fig. 8.16. The increased leading leg
shock angle is due to a combination of the mass injection effect and an increased
displacement thickness growth due to the boundary layer interacting with the upstream
oblique shocks. Furthermore, the local Mach number will be less for the second flap,
compared to the modelled constant local Mach number of 1.38 due to the interaction
produced by the first flap. This will lower the local Mach number and increase the
shock angle produced by the second flap with 0.3mm deflection, see figure 8.16.

Figure 8.16 The leading leg shock angle variation for
mesoflap deflection of the MART four-flap deflection

8-236
8.3. - Analysis of the Unimorph System

Analysis of the unimorph system uses the oncoming boundary layer and displacement
thicknesses calculated in chapter 2. These values along with the pressures, Mach
numbers, coefficients of friction and velocities involved are given in table 8.1.

UNS Interaction SNS Interaction
Upstream Normal Mach number, M
n1
1.3 1.5 1.3
Reservoir Pressure, P
01
(kPa) 150 150 343
Boundary Layer Thickness, o
0
(mm) 7.5 7.5 7.6
Displacement Thickness, o
0
* (mm) 1.4 1.5 1.8
Downstream slot pressure difference, dP (kPa) 21.79 29.8 17.6
Skin Friction Coefficient, C
f 0.001 0 0.0002
Suction Mach number, M
h 0.42 0.615 0.45
Table 8.1. Theoretical Model Inputs

8.3.1 - The Unimorph System for an UNS Interaction at M=1.5
Figure 8.17 Displacement Thickness growth angle variation
for Unimorph flap deflection at M
1
=1.5

The theoretical model described in section 8.1 is applied to the unimorph system for the
control of the unswept normal shock (UNS) interaction at M
1
=1.5, see fig. 8.17. The
flap manufacturing process created slots, with an initial width (G
I
) of 0.2mm, which
provides continual passive control at 0mm flap deflection. However, the amount of
control is minimal at true zero deflection due to relative size of the oncoming boundary

8-237
layer. Using an initial slot width of 0.2mm for the theoretical model was observed to
produce excessive mass transfer at zero deflection. Therefore, an initial slot width of
0.1mm was used for the theory. This provided minimal mass transfer at zero deflection
and did not significantly change the amount of mass transfer with flap deflections
greater than 0.05mm.

Referring to figure 8.17, the mass injection effect reduces for increased unimorph flap
deflection as the injection angle,
m&
u , becomes more oblique. However, it appears to
asymptote with a balance of the reducing injection angle and the increasing amount of
mass injection. The theoretical model estimates a minimum total displacement
thickness growth angle of 7.6 with a unimorph flap deflection of 1.5mm, or
approximately 1J
o
*. This suggests that optimal performance, regarding the boundary
layer disturbance and the frictional losses, will be achieved. From structural modelling
and deflection calculations this amount of deflection cannot be produced using a
piezoelectrically controlled unimorph for active control, see chapter 4. For this reason
flap SBLI control was studied using a mechanical device to study the flow
characteristics at flap deflections greater than those available using piezoelectric
actuators.
Figure 8.18 The leading leg shock angle variation for unimorph deflection.


8-238
It can be seen from figure 8.18 that that the leading leg shock angles, measured from
schlieren photography in chapter 7, agree closely with the theory. The piezoelectric
actuator controlled unimorph cases of flaps closed and flaps open are estimated to
produce deflections of 0.09mm and 0.35mm respectively. However, these are only
approximate deflections as they could not be measured during the experiments and are
assumed to be the mean deflection produced by the uniform and quadratic pressure
distributions, see section 4.7. It is predicted that for the optimal unimorph deflection of
1.5mm, inferred from figure 8.17, a leading leg shock angle of 52 will be generated.

For the uncontrolled case a 45 leading leg angle is produced when there are no
deflecting flaps and no mass transfer present, (fig. 8.18). If the displacement thickness
growth is solely considered as the deflection angle then the oblique shock equation
gives a leading leg angle of 43. The difference is presumably from the presence of the
SBLI, which will have an upstream influence and increase the displacement thickness
growth to approximately 2.8 (the equivalent deflection angle for a 45 oblique shock).


Referring to figure 8.19, there is good correlation between the theoretically estimated
triple point height and the experimental values, which are within the experimental error.
It is unclear whether 1.5mm is the optimal unimorph deflection as it appears that a flap

8-239
deflection between 1mm to 2mm produces similar effects. There does appear to be a
trade-off between the total pressure and friction losses for optimal deflection angle,
within the limit of inhibiting boundary layer separation. When considered with the
experimental data, this would suggest that the 1mm flap deflection is optimal as it is
less likely to induce flow unsteadiness and separation effects.

8.3.2. - The Unimorph System for a SNS Interaction at M
n
=1.3
n1
=1.3.

The theoretical model was applied to the unimorph system for controlling the swept
normal shock (SNS) Interaction with M
n
=1.3, see fig. 8.20. It can be seen that the mass
injection effect is relatively constant regardless of the amount of deflection above
0.01mm, suggesting that the main difference in SBLI control will come from the
compression ramp effects produced by the deflected unimorphs. The theoretical model
calculates a minimum total displacement thickness growth angle of 3.6 with 0mm
unimorph flap deflection. The total displacement thickness growth angle increases with
unimorph deflection suggesting increased friction losses within the boundary layer.

Referring to figure 8.21, the leading leg shock angle increases, as expected, with
unimorph deflection due to the increasing total displacement thickness growth angle.
The leading leg shock angle, with flaps open, is estimated from experiment data to be

8-240
33, see section 5.3.2., and is calculated to have a 0.75mm deflection, a mean of the
uniform and quadratic pressure deflections, see section 4.7, agreeing with the theory.



The increasing leading leg shock angle of the lambda structure results in a higher triple
point, producing a larger lambda structure with an implied improvement in total
pressure recovery, see fig. 8.22. However, the increased total pressure recovery comes

8-241
at the cost of increased frictional losses within the boundary layer, see fig. 8.20, but it is
unclear what deflection is optimal for the SBLI control. Unfortunately no experimental
data could be achieved for the triple point height of the swept normal SBLI control
experiments due to the control plate being on the tunnel side-wall inhibiting schlieren
photography.

8.3.3. - The Unimorph System for an UNS Interaction at M=1.3
1
=1.3

The theoretical model was used to analyse the unimorph system to control the UNS
interaction at M
1
=1.3, see fig. 8.23. Again, the mass injection effect decreases for
increased unimorph flap deflection as the injection angle,
m&
u , becomes more oblique. It
then asymptotes as the reducing injection angle is balanced by an increasing amount of
mass injection. The theoretical model calculates a minimum total displacement
thickness growth angle of 6.1 with a unimorph flap deflection of 0.75mm or 0.56J
o
*.
However, a unimorph deflection in the range 0mm to 1.5mm induces similar amounts of
interaction control. This appears a reasonable result as the interaction should be
incipiently separated and, therefore, may not require control.

It should be noted that this is approximately the maximum 6.7 deflection angle for an
attached oblique shock. The author suggests that the leading leg, for the M=1.3 flow,

8-242
will detach regardless; due to the proximity of the total displacement thickness growth
angle to the maximum deflection angle for attached flow. Furthermore, it is believed
that the upstream influence will increase, beyond the theoretical, regardless of
attachment due to the leading leg being close to detaching. Theory predicts that the
leading leg will not be attached to the end of unimorph above the maximum 6.7
deflection angle, which corresponds to a unimorph deflection of 2mm. For these
reasons the theory has been modified to account for a detached leading leg.

8.3.3.1. - MODIFIED THEORETICAL MODEL
The theoretical model was modified for the M=1.3 unswept normal SBLI, to take into
account the displacement thickness growth angle being approximately the maximum
flow deflection angle for an attached oblique shock. This is achieved by calculating the
triple point height assuming no upstream influence. Then the leading leg angle is
predicted using the fixed triple point height and the upstream influence length of mass
injection and the deflected unimorph. Equation 8.18 was modified to

inj
L L =
11
|
[Eq. 8.22]

That is the leading leg reaches upstream of the inviscid shock position only to the end of
the unimorph flap and that there is no further upstream influence due to the pressure rise
across the SBLI or the mass injection.

The attached oblique shock angle is initially calculated using the theoretical deflection
angle (eq. 8.14), to calculate a triple point height (eq. 8.16), see fig. 8.24. It is assumed
that if u
T
is larger than the maximum deflection angle for an attached oblique shock
then the leading leg angle will be the maximum (Q
ll
=69).

From figure 8.24, it can be seen that the theoretical triple point height agrees with those
calculated from the experiments. The triple point height rises gradually after 2mm
unimorph deflection due to the combination of the constant lambda structure size after
detaching (h
tp
=L
inj
tan69) and the increasing unimorph deflection. It is unclear if the
theory over predicts for the 2mm unimorph deflection, compared to the experimental

8-243
estimate, or whether the model has in fact broken down after the maximum attached
deflection angle is reached. It is seen in experiments that the boundary layer becomes
unsteady at 2mm and 3mm deflections, which is believed to be due to a detached
unsteady leading leg of the lambda structure, see section 6.4.5.


With the triple point height determined, a new leading leg shock angle is calculated to
account for the upstream influence of the mass injection and the pressure rise across the
main shock using

|
|
.
|
\
|
+ +
=

m inj
ll inj
ll
L L L
L
new
& inf
1
tan
tan
|
| [Eq. 8.23]

where the upstream influence length of mass injection,
m
L
&
, is the product of the amount
of mass injection and its angle. Therefore, equation 8.20 has been revised to

( )
o m Total m
m L o u
& &
& sin = [Eq. 8.24]


8-244
The upstream influence length of the pressure rise across the main shock, L
inf
, is the
same for an attached leading leg, see eq. 8.19. However, after detachment occurs the
upstream influence is assumed to be an entire oncoming boundary layer thickness, using

( )
o
M L o 1
inf
= (Attached leading leg)
or
o
L o =
inf
(Detached leading leg) [Eq. 8.25]

with the total upstream influence length limit of equation 8.21 that

i m
L L o s +
& inf

It can be seen from figure 8.25 that the initial leading leg angle, Q, is over calculated,
which is attributed to the leading legs proximity to detaching and that there is no
upstream influence. However, when the leading leg angle is revised by holding the
triple point constant and considering the upstream influence of the interaction the theory
is in good agreement with the values estimated from the schlieren. After 2mm flap
deflection the leading leg detaches and is initiated 1o
o
upstream of the point of mass
injection (the end of the upstream unimorph flap).



8-245
Two lambda structure leading legs, at 54 and 58, were observed to exist with a 3mm
unimorph deflection, see section 6.4.1. These are attributed to the compression ramp
and mass injection effects respectively. Referring to figure 8.23, the predicted 7.8 total
displacement thickness growth angle would cause the leading leg to detach pushing its
foot upstream onto the unimorph. A 3mm unimorph deflection would create a
compression ramp angle of 3.3 with a corresponding attached oblique shock (leading
leg) angle of approximately 55. Downstream, if the flow on top of the unimorph has
already been turned by 3.3 then the defection angle due to mass injection would be
approximately 4.5, with a corresponding 58 oblique shock (leading leg) angle. The
second oblique shock angle due to the mass injection may in fact be greater than 58, as
the first oblique shock would slightly retard the flow, increasing the second shock angle
due to a reduced local Mach number. However, they are within the experimental
accuracy.

Referring to figure 8.23 and 8.24, it would appear that the optimal unimorph deflection
for the UNS interaction at M
1
=1.3 is approximately 0.75mm for minimal total
displacement thickness growth. At this deflection the disturbances to the boundary
layer, and therefore the frictional losses, due to the total displacement thickness growth
angle will be minimal (fig. 8.23). However, a slightly greater deflection, say 1mm,
would produce a larger lambda structure with a taller region of increased total pressure
recovery with similar frictional losses (fig. 8.24).

For the uncontrolled case a 52 leading leg angle is produced when there are no
deflecting flaps and no mass transfer present, (fig. 8.25). If the displacement thickness
growth is solely considered as the deflection angle then the oblique shock equation
gives a leading leg angle of 51. These are within experimental error but it is believed
that presence of the SBLI will increase the displacement thickness growth upstream of
the interaction. This effect is only minor as the shock is relatively weak.


8-246
8.4. - Piezoelectric Actuated Optimised SBLI Control

8.4.1. - Optimal Deflection for SBLI Control
Using experimental data, unimorph deflections of 1mm and between 1mm to 2mm are
believed to produce optimal control of the swept normal shock (SNS) and the unswept
normal shock (UNS) interactions respectively, see chapters 5 to 7. The theoretical
model predicts a slightly reduced optimal deflection range of 0mm to 1mm for the UNS
interaction at M
1
=1.3 but it does not identify an optimal deflection for the SNS
interaction, see table 8.2.

SNS Interaction
Control (M
n
=1.3)
UNS Interaction
Control (M=1.3)
UNS Interaction
Control (M=1.5)
Experiments 1mm 1mm to 2mm 1mm to 2mm
Theoretical Model - 1mm 1mm
Table 8.2 Optimal deflections for unimorph controlled SBLI

8.4.2. Acting Pressure Distribution
At deflections of 1mm and greater, the results indicate that the acting pressure
distribution on the unimorph will be approximately quadratic. Above 1mm deflection
there will be increased spillage off the high pressure side and mass transfer across the
sides of the unimorph. However, these effects will be minimal at below 1mm deflection
and the pressure distribution will be more uniform, see chapters 5 to 7.

8.4.3. Piezoelectric Requirements
The amount of unimorph deflection predicted by theory and inferred from experiments
can then be used to calculate the strength of the piezoelectric actuation required for
optimal control. The classic structural modelling, from chapter 4, is utilised to
determine the d
31
charge constant required for the piezoelectric material. Quadratic and
uniform pressure distributions are used to assist (-500V) and inhibit (500V) the
aeroelastic deflection respectively.

8.4.3.1. SNS INTERACTION CONTROL (M
N
=1.3)
Referring to figure 8.26, contours of constant deflection for a quadratic 17.64kPa
pressure load can be observed with an applied -500V to assist the aeroelastic deflection.

8-247
The piezoelectric charge constant is -625x10
-12
mV
-1
, which is two and a half times
greater than currently available material. The grey area shows the range of substrate
thickness and Youngs Modulus where zero deflection is achievable under a uniform
17.64kPa pressure load with a maximum applied voltage of 500V to inhibit the
aeroelastic deflection.

It was determined from experiments that optimal control is approximately provided with
1mm unimorph deflection. From figure 8.26, it can be seen that there is a range of
thickness and Youngs Modulus where the 1mm deflection contour and the 0mm (grey)
region overlap. This signifies that optimal active control using piezoelectric actuation
would be feasible with a strain production two and a half times greater than currently
available. The substrate material would need a minimum Youngs Modulus of 128GPa
(y=2) and an approximate substrate thickness of 0.85mm (x=1.7). It is expected that
steel would have to be utilised for practical use, because it has a Youngs Modulus of
approximately 210GPa (y=3.28). This would imply that a 0.83mm (x=1.66) steel
substrate thickness would be optimal for active control.

Figure 8.26 Deflection Contours for a Quadratic 17.64kPa pressure load
with a -500V and a piezoelectric charge constant d
31
of -625x10
-12
mV
-1
.
The use of steel in aerodynamic structures is not common practice, especially for
transonic aerofoils, due to its weight to strength ratio. However, the use of carbon fibre

8-248
composites is becoming more commonly used in the production of wings. Carbon Fibre
material can have various Youngs Modulus from 200GPa to 600GPa. However, a
standard isotropic material will have a Youngs Modulus comparable to steel and will
be taken as 230GPa (y=3.6) from SOFiCAR (2005). Therefore, a 0.8mm (x=1.6)
carbon fibre substrate thickness would be optimal for active control.

Historically, aluminium has been more readily utilised and therefore a greater d
31
charge
constant would be required. Referring to figure 8.27, contours of constant deflection for
a quadratic 17.64kPa pressure load can be observed with an applied -500V to assist the
aeroelastic deflection. The piezoelectric charge constant is -750x10
-12
mV
-1
, which is
three times stronger than currently available material. Again, the grey area shows the
range of substrate thickness and Youngs Modulus where zero deflection is achievable
under a uniform 17.64kPa pressure load with a 500V, or less, applied voltage to inhibit
the aeroelastic deflection.

31
of -750x10
-12
mV
-1
.
It can be seen that using a piezoelectric material with triple the d
31
charge constant, than
currently available, allows the use of aluminium (y=1.1) to provide optimal control.
This allows a 0.94mm (x=1.88) to 1.17mm (x=2.34) range of substrate thicknesses were
0mm and 1mm deflection is achievable within the 500V applied voltage range.

8-249
Moreover, using a mean thickness value of 1.05mm (x=2.1) would allow a lower
applied voltage range of approximately 450V. This would reduce the likelihood or
piezoelectric breakdown or arcing to occur.

8.4.3.2. UNS INTERACTION CONTROL (M=1.3)
a)
b)
31
of
a) -750x10
-12
mV
-1
and b) -1000x10
-12
mV
-1
.
Experimental data and theoretical modelling predict that optimal control of the UNS
interaction at M1=1.3 is approximately achieved with deflections of 2mm and 1mm
respectively. Referring to figure 8.28a and b, contours of constant deflection for a

8-250
quadratic 21.79kPa pressure load can be observed with an applied -500V to assist the
aeroelastic deflection. The piezoelectric charge constants are -750x10
-12
mV
-1
and
-1000x10
-12
mV
-1
, which are three and four times respectively more powerful than
currently available material. The grey area shows the range of substrate thickness and
Youngs Modulus where zero deflection is achievable under a uniform 21.79kPa
pressure load with a maximum applied voltage of 500V to inhibit aeroelastic deflection.

It can be seen that there is a range of substrate thickness and Youngs Modulus where
the 1mm deflection contour and the 0mm, grey, region overlap, see fig 8.28a. This
signifies that active control with an optimal 1mm deflection is feasible using a
piezoelectric material with a d
31
charge constant three times greater than currently
available. For an aluminium substrate a thickness range of 1.14mm (x=2.28) to 1.21mm
(x=2.42) can be utilised. A thinner substrate will allow greater maximum deflection but
would increase the applied voltage required to achieve zero deflection. The use of
carbon fibre would require a substrate thickness of 0.86mm (x=1.72).

Referring to figure 8.28b, to obtain a unimorph deflection range of 0mm to 2mm a
piezoelectric charge constant of -1000mV
-1
is required, which is four times greater than
currently available. If it were achievable then an aluminium substrate thickness range
of 0.82mm (x=1.64) to 0.87mm (x=1.74) would provide the 2mm optimal deflection
required for active control. Also, a carbon fibre substrate would require a thickness of
0.65mm (x=1.3).

8.4.3.3. UNS INTERACTION CONTROL (M=1.5)
The experimental data and the theoretical model both predict that optimal interaction
control at M1=1.5 is approximately produced with a 2mm unimorph deflection. Figures
8.29a and b, contours of constant deflection for a quadratic 30kPa pressure load can be
observed with an applied -500V to assist the aeroelastic deflection. The piezoelectric
charge constants are -875x10
-12
mV
-1
and -1250x10
-12
mV
-1
respectively, which are three
and a half and five times respectively greater than currently available material. The
grey area shows the range of substrate thickness and Youngs Modulus where zero
deflection is achievable under a uniform 30kPa pressure load with a 500V, or less,
applied voltage to inhibit the aeroelastic deflection.

8-251
It can be seen that there is a range of thickness and Youngs Modulus where the 1mm
deflection contour and the 0mm (grey) region overlap, see fig 8.29a. This signifies that
active control with an optimal 1mm deflection is feasible using a piezoelectric material
with a d
31
charge constant three and a half times larger than currently available. For an
aluminium substrate a thickness range of 1.33mm (x=2.66) to 1.39mm (x=2.78) can be
utilised. The use of carbon fibre would require a substrate thickness of 1.02mm
(x=2.04).

Figure 8.29 Deflection Contours for a Quadratic 30kPa pressure load
31
of
a) -875x10
-12
mV
-1
and b) -1250x10
-12
mV
-1
.

8-252
Referring to figure 8.29b, to obtain a unimorph deflection range 0mm to 2mm a
piezoelectric charge constant of -1250mV
-1
is required, which is five time greater than
currently available. If it were achievable then an aluminium substrate thickness range
of 0.95mm (x=1.9) to 1.01mm (x=2.02) would provide the 2mm optimal deflection
required for active control of the UNS interaction at M
1
=1.5. A carbon fibre substrate
would require a thickness of 0.74mm (x=1.48).

8.5. Optimal Shock Location
a)
b)
c)
d)
Figure 8.30 Unswept Normal SBLI Control with a rear shock position at M=1.5
using flaps deflected to a) 0mm, b) 1mm, c) 2mm and d) 3mm.

8-253

The primary goal of SBLI control is to reduce the overall drag, which is achieved by
reducing the wave drag and minimising the increase in frictional drag. One way to
reduce the wave drag is to induce a larger (more smeared) lambda structure by having a
shock position further downstream. It is believed that this should produce a
significantly higher triple point and lightly thicker boundary layer. Schlieren
photography was achieved for a shock positioned to the rear of the interaction region
with 66% of the control plate situated upstream, see figs. 8.30a to d.

Referring to figure 8.30 a & d, it can be seen that the triple point has approximately
doubled in height with 3mm deflection compared to the 0mm flap deflection control
with the rear shock position. Moreover, this is approximately an extra 50% over the
triple point height produced with a 3mm flap deflection and a middle shock position,
see fig. 7.9. However, the 0mm unimorph deflection triple point for a rear shock
position appears approximately 10% higher than with a middle shock position, see
figures 7.6 and 8.30a. This is due to the passive control provided by the longitudinal
and lateral slots produced during the manufacture process.

8.6. Further work

Using an alternate shock position it is unclear whether the unsteadiness observed with
the middle shock position will still be present. Also, it is unknown to what extent the
viscous losses and total pressure recovery will be affected. This is clearly an area for
further work as this could not be fully examined due to time constraints at the
Cambridge facility.

Furthermore, a more powerful piezoelectric or other smart material actuator is
required than is currently available, as elucidated at the conclusion of chapter 5. This
would enable increased flap deflections to levels indicated in this work for optimal
SBLI control.

In addition, it should be remembered that for practical application to a transonic wing
the height of triple point cannot extend indefinitely to produce increased total pressure

8-254
recovery. The normal shock on a transonic wing will have a finite height and,
moreover, the triple point will only be able to extend to a fraction of this height before
the supersonic region above the wing is adversely affected. Moreover, in view of the
degradation of the material over time one should be mindful of the operating limits of
piezoelectric material.

8.7. - Conclusion

A theoretical model is derived to predict the effect of unimorph flap deflection on the
displacement thickness growth angles, the leading shock angle and the triple point
height.

The results indicate that minimal frictional effects are produced with the minimum total
displacement growth angle. However, optimal deflection for SBLI control is a trade-off
between reducing the total pressure losses, which is implied with increasing the triple
point height, and minimising the frictional losses. This suggests that optimal flap
deflection for SBLI control occurs at slightly differing amounts to those dictated by the
total displacement thickness growth angle.

From theoretical analysis of the UNS interaction at M
1
=1.5, it is predicted that the
minimal displacement thickness growth angle is produced with a unimorph flap
deflection of 1.5mm or 1J
o
*. However, it is suspected that optimal SBLI control will be
produced at a reduced flap deflection with a higher triple point and its implied
improvement in total pressure recovery.

The theoretical analysis of the UNS interaction at M
1
=1.3 predicts the minimum
displacement thickness growth angle is produced with a unimorph flap deflection of
0.75mm or 0.56J
o
*. However, it is suspected that optimal SBLI control will be
produced at an increased flap deflection with a higher triple point and its implied
improvement in total pressure recovery.

It is unclear from the theoretical analysis of the SNS interaction what unimorph flap
deflection is optimal for control. It predicts that increasing flap deflection increases

8-255
both frictional and total pressure losses. Optimal control will be produced when the
combination of the two losses are minimal, however, this cannot be determined from the
theoretical model.

For total displacement thickness growth angles of approximately, or greater than, the
maximum deflection angle for an attached oblique shocks, there are separate interaction
control effects. These effects are produced from the compression ramps, due to the
deflected unimorph flaps, and the mass injection from the plenum. Before the
maximum deflection angle is achieved the results indicate that the interaction control is
a combination of the two effects with one leading leg formed at the end of the upstream
flap. However, when the maximum deflection angle is reached, or exceeded, the
leading leg appears to detach from the end of the unimorph and is replaced by two
oblique shocks. The first oblique shock originates at the start of the deflected flap and
is produced from the compression ramp. The second oblique shock originates at the end
of the flap due to the mass injection.

Using the experimental and theoretical deflections for optimal control, the power of
piezoelectric material required can be determined. For the three test cases examined in
this thesis a d
31
charge constant between -750x10
-12
to -1250x10
-12
mV
-1
is required.
However, this is three to five times more powerful than current commercially available
material.

8-256
Chapter 9. Conclusion

9-257
CHAPTER 9. Conclusion

The control of the turbulent shock wave/boundary layer interaction as applied to a
transonic aerofoil is addressed in this thesis. However, the work can be applied to the
control of the interaction for numerous other situations where a shock meets a turbulent
boundary layer. It is shown that, for both swept normal shock and unswept normal
shock interactions, as long as the Mach number normal to the shock is the same, then
the interaction, and therefore its control, is the same.

A system of piezoelectrically controlled flaps is used to control the interaction. The
flaps aeroelastically deflect due to the pressure difference created by the pressure rise
across the shock and by piezoelectrically induced strains. The amount of deflection and
hence mass flow through the plenum chamber then controls the interaction. It should be
pointed out that no feedback system is considered in this thesis and the term active
control refers to controlling the overall flap deflection, that is, the mass transfer and
hence degree of flow separation. It is shown that the flaps could delay separation of the
boundary layer and decrease the wave drag thereby showing some advantage over
previous control methods.

A number of design options were considered for the integration of the piezoelectric
ceramic, also known as PZT (Lead Zirconate Titanate), into the flap structure. These
included the use of unimorphs, bimorphs and polymorphs, with the latter capable of
being directly employed as the flap. It was found that multimorphs and bimorphs do
not have the mechanical stiffness to withstand the pressures involved. The unimorphs
can withstand and overcome the pressure loads associated with SBLI control and hence
they were used. Their main disadvantage was that they produced smaller deflections.

It is shown that classic laminate plate theory (CLPT) gives good correlation and
accuracy with the finite element modelling (FEM) for predicting flap deflection. FEM
gives the ability to analyse more complex pressure loadings that reflect the real
boundary conditions. This enables improved FEM predictions of minimum thickness
and stiffness over the classic theory. The main disadvantage of FEM over CLPT is its
significantly higher computational cost in terms of time and effort. However, the results

9-258
indicate that CLPT can be utilised as a very good first order deflection prediction
method. Both systems were utilised to optimise the unimorph design for substrate
Youngs Modulus and thickness to provide active control. The thinner or more
compliant the substrate the more mass transfer, deflection, can occur. Zero deflection
for active control limits the substrate to minimum thickness and stiffness.

It was found that near optimal control for interactions at M
n
=1.3 and 1.5 is attained with
flap deflections between 1mm and 3mm. However, to obtain the deflection required for
truly optimal performance a more powerful piezoelectric actuator material is required
than is currently available.

For all three cases studied, it was observed that a 0mm flap deflection produced a quasi-
uncontrolled interaction with slight deviations due to the continuous passive control
provided by the slots. The main pressure rise across the SBLI, with 0mm flap
deflection, is more gradual due to the mass transfer provided, indicating a weaker/larger
lambda structure. Also, a higher pressure was observed upstream of the main pressure
rise due to upstream mass injection via the longitudinal slots. A 1mm flap deflection
produces a significant level of beneficial SBLI control with a region of separation
produced at the end of the upstream flap due to a combination of the flap creating a
rearward facing step and the presence of mass injection.

For the swept normal shock (SNS) interaction, the experimental data indicates that a
1mm flap deflection provides near optimal control, as the separation and unsteadiness
produced are minimal whilst a significant amount of control is produced. Also, with
1mm flap deflection, the separation effects are minimised in the neighbouring flow
fields above and below the flap region. The PSP results indicated the piezoelectric
actuated unimorphs provide a maximum flap deflection is between half and two thirds
of a millimetre with open flaps and approximately 0.2mm with closed flaps. This
suggests that a pressure force distribution acting on the unimorph was somewhere
between the uniform and quadratic pressure distribution. The results indicate that with
flap deflections greater than 1mm the pressure force acting on the unimorph may tend
toward the quadratic pressure distribution, especially when spillage and compression

9-259
corner effects are introduced with surface curvature. However, the pressure force tends
to act as a uniform load as the deflection approaches 0mm.

It was established that the uncontrolled unswept normal shock (UNS) interaction with a
normal Mach number of 1.3 and 1.5 indicate incipient separation and full separation
respectively. Experimental data indicate that a deflection of between one and two
millimetres does, in fact, provide optimal control for the M=1.3 interaction. At M=1.5,
the SBLI instabilities with flap deflection above 2mm suggest that a flap deflection of
between one and two millimetres is optimal. A 3mm deflection produces a
considerably larger region of reduced total pressure loss with increased unsteadiness in
the thicker boundary layer.

A theoretical model has been developed to predict the effect of unimorph flap deflection
on the displacement thickness growth angles, the leading shock angle and the triple
point height. It predicts optimal control is produced with a unimorph flap deflection of
0.75mm (0.56>
o
*) and 1.5mm (1>
o
*) for M
n
=1.3 and 1.5 respectively. The theoretical
model agrees fairly well with the UNS experimental data of both the present unimorph
control study and the MART control study performed by the research group at the
University of Illinois Urbana Champaign. However, it does not predict a clear optimal
deflection for the SNS interaction. It shows that increasing flap deflection increases
both frictional and total pressure losses. Optimal control will be produced when the
combination of the two losses is minimal; however, this cannot be determined from the
present theoretical model.

Using the experimental and theoretical deflections for optimal control, the power of
piezoelectric material required can be determined. For the three test cases examined in
this thesis a d
31
charge constant between -750x10
-12
to -1250x10
-12
mV
-1
is required.
However, this is three to five times more powerful than the current commercially
available material.

9-260
References 261
REFERENCES
Aldraihem, J. and Khdeir, A. A. (2000) "Smart beams with extension and thickness-
shear piezoelectric actuators", Smart Materials and Structures 9: pp. 1-9.

Alvi, A. and Settles, G. S. (1992) "Physical Model of the Swept Shock Wave Boundary-
Layer Interaction Flowfield", AIAA Journal 30(9): pp. 2252-2258.

Anderson, J. D. (1990) "Modern Compressible Flow with Historical Perspective". New
York, McGraw-Hill.

Atkin, C. J. and Squire, L. C. (1992) "A Study of the Interaction of a Normal Shock
Wave with a Turbulent Boundary Layer at Mach Numbers Between 1.30 and
1.55", Eur. J. Mech., B/Fluids 11(1): pp. 93-118.

Babinsky, H. (1999) "Control of Swept Shock Wave/Turbulent Boundary-Layer
Interactions". ISSW22, July 18-23, Imperial College, London, UK. Paper 0050

Babinsky, H. (2002) Personal Communication. AIAA Reno Conference.

Babinsky, H., Inger, G. R. and McConnell, A. D. (1999) "A Basic
Experimental/Theoretical Study of Rough Wall Turbulent Shock/Boundary Layer
Interaction". ISSW22, July 18-23, Imperial College, London, UK.

Bahi, L., Ross, J. M. and Nagamatsu, H. T. (1983) "Passive Shock Wave/Boundary
Layer Control for Transonic Airfoil Drag Reduction". AIAA 21st Aerospace
Sciences Meeting, January 10 - 13, Reno, NV, USA., AIAA-1983-0137.

Beer, F. P. and Johnston, E. R. (1992) "Mechanics of Materials, 2nd edition in SI
Units". Berkshire, England, McGraw Hill.

Bruch Jr., J. C., Sloss, J. M., Adali, S. and Sadek, I. S. (2000) "Optimal piezo-actuator
locations/lengths and applied voltage for shape control of beams", Smart Materials
and Structures 9: pp. 205-211.

Bushnell, D. M. (2004) "Shock Wave Drag Reduction", Annual Review of Fluid
Mechanics(36): pp. 81-96.

Cattafesta, L., Garg, S. and Shulka, D. (2001) "Development of Piezoelectric Actuators
for Active Flow Control", AIAA Journal 39(8): pp. 1562-1568.

Chapman, D. R., Kuehn, D. M. and Larson, H. K. (1958) "Investigation of Separated
Flows in Supersonic and Subsonic Streams with Emphasis on the Effect of
Transition," NACA Report 1356.

Chattopadhyay, A. and Seeley, C. E. (1997) "A Higher Order Theory for Modelling
Composite Laminates with Induced Strain Actuators", Composites Part B 28B: pp.
243 - 252.
References 262
Chen, C., Chow, C., Holst, T. and Dalsem, W. V. (1984) "Numerical Study of Porous
Airfoils in Transonic Flow", AIAA Journal 22: pp. 989-991.

Chen, T.-Y., Chu, S.-Y., Wu, S.-J. and Juang, Y.-D. (2003) "Effects of Strontium on the
Dielectric and Piezoelectric Properties of Sm-Modified PbTiO
3
Ceramics",
Ferroelectrics 282: pp. 37-47.

Choi, J., Jeon, W.-P. and Choi, H. (2002) "Control of Flow Around an Airfoil Using
Piezoceramic Actuators", AIAA Journal 40(5 - Technical Notes): pp. 1008 - 1010.

Chopra, I. (2002) "Review of State of Art of Smart Structures and Integrated Systems",
AIAA Journal 40(11): pp. 2145 - 2186.

Couldrick, J. S., Shankar, K., Gai, S. and Milthorpe, J. (2003) "Design of "Smart" Flap
Actuators for Swept Shock Wave/Turbulent Boundary Layer Interaction Control",
Structural Engineering and Mechanics: An International Journal 16(5): pp. 519-
532.

Couldrick, J. S., Gai, S., Milthorpe, J. and Shankar, K. (2004a) "Active Control of
Swept Shock Wave/Turbulent Boundary Layer Interactions", The Aeronautical
Journal 108(2): pp. 93-102.

Couldrick, J. S., Gai, S., Milthorpe, J. and Shankar, K. (2004b) "Investigation of Active
Control of Swept Shock Wave/Turbulent Boundary Layer Interactions using
pressure sensitive paints", The Aeronautical Journal 108(9): pp. 483-490.

Couldrick, J. S., Gai, S., Milthorpe, J. and Shankar, K. (2005) "Normal Shock
Wave/Turbulent Boundary Layer Interaction Control using "Smart" Piezoelectric
Flap Actuators", The Aeronautical Journal 109(11): pp. 577-583.

Crawley, E. F. (1994) "Intelligent Structures for Aerospace: A Technology Overview
and Assessment", AIAA Journal 32(8): pp. 1689 - 1699.

Crawley, E. F. and Luis, J. (1987) "Use of Piezoelectric Actuators as Elements of
Intelligent Structures", AIAA Journal 25(10): pp. 1373 - 1385.

Crocco, L. and Lees, L. (1952) "A Mixing Theory for the Interaction Between
Dissipative Flows and Nearly Isentropic Streams", Journal of the Aeronautical
Sciences 19(10): pp. 649 - 676.

Delery, J. (1985) "Shock Wave/Turbulent Boundary Layer Interaction and it's Control",
Progress in Aerospace Science 22: pp. 209-228.

Delery, J. and Bur, R. (1999) "Shock Wave/Boundary Layer Interaction and control
techniques: a physical description". ISSW22, July 18-23, Imperial College,
London, UK, Paper 6010.

References 263
Delery, J. and Marvin, J. G. (1986) "Shock-Wave Boundary Layer Interactions". E.
Reshotko, North Atlantic Treaty Organization: Advisory Group for Aerospace
Research and Development: pp. AGARDograph No. 280.

Doerffer, P. P. and Bohning, R. (2000) "Modelling of perforated plate aerodynamics
performance", Aerosp. Sci. Technol. 4: pp. 525-534.

Donthireddy, P. and Chandrashekhara, K. (1996) "Modeling and Shape Control of
Composite Beams with Embedded Piezoelectric Actuators", Composite Structures
35: pp. 237 - 244.
Eisenberger, M. and Abramovich, H. (1997) "Shape control of non-symmetric
piezolaminated composite beams", Composite Structures 38(1-4): pp. 565-571.

Gehring, G. A., Cooke, M. D., Gregory, I. S., Karl, W. J. and R., W. (2000) "Cantilever
unified theory and optimisation for sensors and actuators", Smart Material and
Structures 9: pp. 918-931.

Gibson, T. M., Babinsky, H. and Squire, L. C. (2000) "Passive Control of Shock
Wave/Boundary Layer interactions", Aeronautical Journal: pp. 124-140.

Green, J. E. (1969) "Interactions Between Shock Waves and Turbulent Boundary
Layers", Progress in Aerospace Science 11: pp. 253-340.

Hafenrichter, E. S., Lee, Y. L., Dutton, J. C. and Loth, E. (2003) "Normal
Shock/Boundary-Layer Interaction Control Using Aeroelastic Mesoflaps", Journal
of Propulsion and Power 19(3): pp. 464-472.

Holden, H. A. (2004) "Transonic Shock/Boundary Layer Interaction Control using
Three-Dimensional Devices." - PhD Thesis. The Engineering Department, The
University of Cambridge, Cambridge, UK.: pp. 154.

Hwang, W.-S. and Park, H.-C. (1993) "Finite Element Modeling of Piezoelectric
Sensors and Actuators", AIAA Journal 31(5): pp. 930 - 937.

Inger, G. R. (1981) "Application of a Shock-Turbulent Boundary Layer Interaction
Theory in Transonic Flowfield Analysis". AIAA Transonic Perspective
Symposium, 18 - 20 February, NASA/Ames Research Center, Moffett Field,
California, USA, AD-A103244. pp. 621-636

ISSI (2005) "Unicoat PSP (Aerosol Can)", Innovative Science Solutions Inc. Accessed -
1st August 2005: http://www.innssi.com/UnicoatPSPAerosolCan.htm
Jaiman, R. K., Loth, E. and Dutton, J. C. (2003) "Simulations of Normal Shock-
Wave/Boundary Layer Interaction Control Using Mesoflaps". 41st AIAA
Aerospace Sciences Meeting & Exhibit, 6-9 January 2003, Reno, Nevada, USA,
AIAA-2003-445.

References 264
Jaiman, R. K., Loth, E. and Dutton, J. C. (2004) "Simulations of Shock/Boundary-Layer
Interaction Control Using Mesoflaps", Journal of Propulsion and Power 20(2): pp.
344-352.

Jeon, W. P. and Blackwelder, R. F. (2000) "Perturbations in the wall region using flush
mounted Piezoceramic actuators", 28: pp. 485-496.

Kaufman, J. G. (1999) "Properties of Aluminium Alloys - Tensile, Creep, and Fatigue
Data at High and Low Temperatures". Materials Park, Ohio, The Aluminium
Association.

Klein, C. (2000) "Application of Pressure Sensitive Paint (PSP) for the determination of
the instantaneous pressure field of models in a wind tunnel", Aerospace Science
Technology 4: pp. 103 - 109.
Kleine, H. (2001) "Measurement Techniques and Diagnostics". (in 'Handbook of Shock
Waves, Volume 1: Theoretical, Experimental, and Numerical Techniques by D. A.
Russell'). Edited by G. Ben-Dor, O. Igra and T. Elperin. New York, Academic
Press: pp. 683-740.

Koide, S., Saida, N. and Ogata, R. (1996) "Correlation of Separation Angles Induced by
Glancing Interactions", AIAA Journal 34(10 - Technical Notes): pp. 2198 - 2200.

Koratkar, N. A. and Chopra, A. (2000) "Analysis and Testing of Mach-Scaled Rotor
with Trailing-Edge Flaps", AIAA Journal 38(7): pp. 1113-1124.

Korkegi, R. H. (1971) "Survey of Viscous Interactions Associated with High Mach
Number Flight", AIAA Journal 9(5): pp. 771-784.

Korkegi, R. H. (1973) "A Simple Correlation for Incipient Turbulent Boundary-Layer
Separation due to a Skewed Shock Wave", AIAA Journal 11(11): pp. 1578-1579.

Korkegi, R. H. (1976) "On the Structure of Three-Dimensional Shock-Induced
Separated Flow Regions", AIAA Journal 14(5): pp. 597-600.

Krogmann, P., Stanewsky, E. and Thiede, P. (1985) "Effects of Suction on
Shock/Boundary-Layer Interaction and Shock-Induced Separation", Journal of
Aircraft 22(1): pp. 37-42.

Kubota, H. and Stollery, J. L. (1982) "An Experimental Study of the Interaction
Between a Glancing Shock Wave and a Turbulent Boundary Layer", Journal of
Fluid Mechanics 116: pp. 431 - 458.

Lee, Y., Hafenrichter, E. S., Dutton, C. and Loth, E. (2004) "Skin Friction
Measurements for Recirculating Normal-Shock/Boundary-Layer Interaction
Control", AIAA Journal 42(4): pp. 806-814.

Lee, Y., Hafenrichter, E. S., Jaiman, R. K., Orphanides, M. J., Dutton, J. C., Grafton,
W., Wilkins, L. J. and Loth, E. (2002) "Skin friction measurements in normal
References 265
shock wave/turbulent boundary-layer interaction control with aeroelastic
mesoflaps". 40th AIAA Aerospace Sciences Meeting and Exhibit, 14-17 January,
Reno, Nevada, USA, AIAA-2002-979.

Liepman, H. W. and Roshko, A. (1957) "Elements of Gasdynamics". New York, John
Wiley & Sons Inc.

Lin, C.-C. and Hsu, C.-Y. (1999) "Static shape Control of Smart Beam Plates with Sine
Sensors and actuators", Smart Materials and Structures 8(5): pp. 519 - 530.

Magi, E. C. (1990) "Investigations into the flow behind castellated blunt trailing edge
Aerofoils at Supersonic Speed" - Doctor of Philosophy. Department of
Mechanical Engineering,, University College, The University of New South
Wales, Canberra: pp. 243.

Mateer, G. G., Brosh, A. and Viegas, J. R. (1976) "A Normal Shock-Wave Turbulent
Boundary-Layer Interaction at Transonic Speeds." 14th AIAA Aerospace Sciences
Meeting, Washington, USA, AIAA-1976-161.

Matthew, J., Sankar, B. and Cattafesta, L. (2001) "Finite Element Modelling of
Piezoelectric Actuators for Active Flow Control Applications". 39th AIAA
Aerospace Sciences Meeting and Exhibit, 8-11 January, Reno, Nevada, USA.

McCabe, A. (1966) "The Three-Dimensional Interaction of a Shock Wave with a
Turbulent Boundary Layer", The Aeronautical Quarterly: pp. 231-252.

Mid (2003) "Active Materials". Accessed - 15/12/03:
http://www.mide.com/active_materials
Morgan Matroc, I. (2000a) "Piezoelectric Ceramics for Designers",

Morgan Matroc, I. (2000b) "Piezoelectric Technology data for Designers",

Morris, M. J., Donovan, J. F., Kegelman, J. T., Schwab, S. D. and Levy, R. L. (1993)
"Aerodynamic Applications of Pressure Sensitive Paint", AIAA Journal 31(3): pp.
419 - 425.

Mukherjee, A. and Joshi, S. (2002) "Piezoelectric Sensor and Actuator Spatial Design
for Shape Control of Piezolaminated Plates", AIAA Journal 40(6): pp. 1204 -
1210.

Nagamatsu, H. T., Ficarra, R. V. and Dyer, R. (1985) "Supercritical Airofoil Drag
Reduction by Passive Shock Wave/Boundary Layer Control in the Mach Number
Range .75 to .90". AIAA 23rd Aerospace Sciences Meeting, January, Reno,
Nevada, USA, AIAA-1985-0207.

Nagamatsu, H. T., Mitty, T. J. and Nyberg, G. A. (1987) "Passive Shock
Wave/Boundary Layer Control of a Helicopter Rotor Airfoil in a Contoured
References 266
Transonic Wind Tunnel". AIAA 25th Aerospace Sciences Meeting, January,
Reno, Nevada, USA, AIAA-1987-0438.

Newport (2003) "Electrostrictive vs. Piezoactuators". Accessed - 15/12/03:
http://www.newport.com/Support/Tutorials/OptoMech/electropiezo.asp
Pearcey, H. H. (1955) "Some effects of shock induced separation of turbulent boundary
layer in transonic flow past aerofoils", ARC R& M No. 3108,

Pearcey, H. H. (1961) "Shock-Induced Separation and it's Prevention by Design and
Boundary Layer Control". (in 'Boundary Layer and Flow Control - It's Principles
and Application -Volume 2'). Edited by G. V. Lachmann. London, Pergamon
Press Ltd. 2: pp. 1170-1361.

Physik-Instrumente (1996) "E-461 HVPZT Amplifiers", Physik Instrumente. Accessed -
3/5/05: http://www.physikinstrumente.de/products/prdetail.php?secid=6-30
Physik-Instrumente (1996b) "PL-122 PL-140 Piezoelectric Multilayer Bender
Actuator". Accessed - 25/11/02:
http://www.physikinstrumente.com/pztactuators/1_30a.html
Raghunathan, S. (1987) "Effects of Porosity Strength on Passive Shock-Wave/Boundary
Layer Control", AIAA Journal 25(5): pp. 757-758.

Raghunathan, S. (1988) "Passive Control of Shock-Boundary Layer Interaction",
Progress in Aerospace Science 25: pp. 271-296.

Raghunathan, S., Gray, J. L. and Cooper, R. K. (1987) "Effects of Inclination of Holes
on Passive Shock Wave Boundary Layer Control". AIAA 25th Aerospace
Sciences Meeting, January, Reno, Nevada, USA, AIAA-1987-0437.

Raghunathan, S. and Mabey, D. G. (1987) "Passive Shock-Wave/Boundary Layer
Control on a Wall-Mounted Model", AIAA Journal 25(2): pp. 275-278.

Ren, W., Masys, A. J., Yang, G. and Mukherjee, B. K. (2000) "The Field and Frequency
Dependence of the Strain and Polarisation in Piezoelectric and Electrorestrictive
Ceramics". 3rd Asian Meeting on Ferroelectrics (AMF3), 12-15 December, Hong
Kong.

Savu, G. and Trifu, O. (1984) "Porous Airfoils in Transonic Flow", AIAA Journal 22(7
- Technical Notes): pp. 989 - 991.

Seddon, J. (1960) "The Flow Produced by Interaction of a Turbulent Boundary Layer
with a Normal Shock Wave of Strength Sufficient to Cause Separation" -, HMSO
ARC, R&M, London: pp. No. 3502,.

Seifert, A., Eliahu, S. and Greenblattm, D. (1998) "Use of Piezoelectric Actuators for
Airfoil Separation Control", AIAA Journal 36(8 - Technical Notes): pp. 1535 -
1537.
References 267
Settles, G. S. and Dolling, D. S. (1992) "Swept Shock-Wave/Boundary Layer
Interactions", Progress in Astronautics and Aeronautics 141(Ed - A.R. Seebass):
pp. 505 - 574.

Smith, A. N. (2002) "The Control of Transonic Shock Wave/Turbulent Boundary Layer
Interactions using Streamwise Slots" - PhD Thesis. Department of Engineering,
University of Cambridge, Cambridge: pp. 169.

Smith, A. N., Babinsky, H., Dhanaesekaran, P. C., Savill, A. M. and Dawes, W. N.
(2003) "Computational Investigation of Slot and Groove Controlled Shock Wave/
Boundary Layer Interactions(AIAA Paper 03-0446)". 41st AIAA Aerospace
Sciences Meeting and Exhibit, 6-9 January, Reno, NV, USA. AIAA 2003-0446

Smith, A. N., Babinsky, H., Fulker, J. L. and Ashill, P. R. (2002) "Normal Shock
Wave/Turbulent Boundary Layer Interactions in the presence of Streamwise Slots
and Grooves", The Aeronautical Journal 106(1063): pp. 493 - 500.

Soares, C. M. M., Soares, C. A. M. and Correia, V. M. F. (1999) "Optimal Design of
Piezolaminated Structures", Composite Structures 47: pp. 625 - 634.
SOFiCAR (2005) "Societe des Fibres de Carbone", SOFiCAR. Accessed - 12th October
2005: http://www.soficar-carbon.com/uk/index.html
Squire, L. C. (1996) "Interaction of Swept and Unswept Normal Shock Waves with
Boundary Layers", AIAA Journal 34(10): pp. 2099-2101.

Stollery, J. L. (1989) "Glancing Shock-Boundary Layer Interactions". Special Course on
Three-Dimensional Supersonic and Hypersonic Flows Including Separation, 8-12
May.

Tharayil, M. and Alleyne, A. G. (2004) "Modeling and Control for Smart Mesoflap
Aeroelastic Control", IEEE/ASME Transactions on Mechanics 9(1): pp. 30-39.

Thiede, P., Krogmann, P. and Stanewsky, E. (1984) "Active and Passive
Shock/Boundary Layer Interaction Control on Supercritical airfoils." AGARD CP-
365,

Vipperman, J. S. and Clark, R. L. (1996) "Implementation of an Adaptive Piezoelectric
Sensoriactuator", AIAA Journal 34(10): pp.

Wang, B. L. and Noda, N. (2001) "Design of a smart functionally graded
thermopiezoelectric composite structure", Smart Material And Structures 10: pp.
189-193.

West, J. E. and Korkegi, R. H. (1972) "Supersonic Interaction in the Corner of
Intersecting Wedges at High Reynolds Numbers", AIAA Journal 10(5): pp. 652-
656.

References 268
Wood, B., Loth, E. and Geubelle, P. (1999) "Mesoflaps for Aeroelastic Transpiration
for SBLI Control". AIAA 37th Aerospace Sciences Meeting, January, Reno,
Nevada, USA, AIAA-1999-0614.

Yam, L. H. and Yan, Y. J. (2002) "Optimal Design of Thickness and Embedded Depth
of Piezoelectric Actuator in Piezo-laminated Structures". 2nd Advances in
Structural Engineering and Mechanics, 21-23 August, Pusan, Korea.

Appendix A1. Piezoelectric Properties

A1-1
A1. Piezoelectric Material (Morgan Matroc, I. (2000a))
Appendix A2. Aluminium 5083-H321

A2-2
A2. Aluminium 5083-4321 Properties - ASM (2005)

Appendix A2. Aluminium 5083-H321

A2-3
Thermal Properties
References

ASM, A. S. M. I. (2005) "Aluminum 5083-H116; 5083-H321", ASM Aerospace
Specification Metals Inc. Accessed - 8th November 2005:
http://asm.matweb.com/search/ SpecificMaterial.asp?bassnum=MA5083H116
Appendix A3. Aluminium Substrate Three Point Test A3-4
A3. Aluminium Substrate
Three Point Test
Kaufman (1999) gives the tensile stress, yield stress and Youngs Modulus as 315 MPA,
230 MPA and 70 GPa respectively for Al5083-H321. In order to evaluate the Youngs
Modulus a three-point test was undertaken, see fig A3.1. The yield & tensile stresses
were not calculated as they are not required for the calculation of tip deflection and the
designs would work considerably away from these limits.
Figure A3.1. - Three-point test dimensions

The deflection for a three-point test is governed by the beam theory in Appendix D of
Beer and Johnston (1992) as:
( )
EIL
d L Pd
MAX
3 9
2
3
2
2
2
2

= [Eq. A3.1]
12
3
wt
I =
w=25mm
t=4mm

For L and d
2
of 8cm and 3.6cm respectively
MAX
occurs at d
m
where:

cm
d L
d
m
125 . 4
3
2
2
2
=
= [Eq. A3.2]

Testing yielded the following results:
P(N)
test
(mm)
theory
(mm) Error (%)
0 0 0 0
20 -0.023 -0.023 0
30 -0.034 -0.034 0
35 -0.037 -0.039 -5.13
40 -0.042 -0.045 -6.67
45 -0.048 -0.051 -5.88
50 -0.055 -0.056 -1.79
55 -0.061 -0.062 -1.61
60 -0.066 -0.068 -2.94
65 -0.073 -0.073 0
70 -0.08 -0.079 1.28
Average Error -2.07
Figure A3.1. Aluminium deflection in a three-point test

Plotting the test data and applying a trendline gives
Equation of Trendline
y = -887.28x + 0.8638
0
10
20
30
40
50
60
70
-0.08 -0.06 -0.04 -0.02 0
Maximum Deflection (mm)
P
(
N
)
Figure A3.2. Maximum Deflection variation for a given point load

Rearranging equation A3.1; the gradient of figure A3.2. can be given as
( )
E
d L d
IL P
MAX
2
3
2
2
2
2
3 9
[Eq. A3.3]

The gradient of the figure A3.2. is -887.28 Nm
-1
and applying this to equation A3.3
gives a Youngs Modulus of 70.05GPa.

There is an instrumentation error of 1N for the force, however, the Youngs Modulus
can be assumed to be 70 GPa, agreeing with the data books.
References

Beer, F. P. and Johnston, E. R. (1992) "Mechanics of Materials, 2nd edition in SI
Units". Berkshire, England, McGraw Hill.

Kaufman, J. G. (1999) "Properties of Aluminium Alloys - Tensile, Creep, and Fatigue
Data at High and Low Temperatures". Materials Park, Ohio, The Aluminium
Association.

Appendix B1. Classic Laminate Plate Theory B1-1
Appendix B1 - Classic Laminate Plate Theory

B1.1. Hygrothermal Stresses in Laminates

Figure B1.1 Classical theory unimorph configuration

It has been shown that the piezoelectric actuation is comparable to thermal actuation,
Chen et al. (2003). A direct analogy is now derived with the forces and moments on a
laminates being directly related to the strains, shears and curvatures written as

(
(
(
(
(
(
(
(
xy
y
x
xy
y
x
xy
y
x
xy
y
x
D D D B B B
D D D B B B
D D D B B B
B B B A A A
B B B A A A
B B B A A A
M
M
M
N
N
N
k
k
k
c
c
0
0
0
66 26 16 66 26 16
26 22 12 26 22 12
16 12 11 16 12 11
66 26 16 66 26 16
26 22 12 26 22 12
16 12 11 16 12 11
[Eq. B1.1]

where N, M, c, , k are the forces, moments, strains, poisons ratio and curvature
respectively along the x and y axis. The laminate extensional stiffnesses, A
ij
, are given
by

( ) ( ) ( )
1
1
2
2

=
= =

}
k k
N
k
k
ij
t
t
k
ij ij
z z Q dz Q A [Eq. B1.2]

The laminate coupling stiffnesses, B
ij
, are given by

( ) ( ) ( )
2
1
2
1
2
2
2
1
= =

}
k k
N
k
k
ij
t
t
k
ij ij
z z Q zdz Q B [Eq. B1.3]

The laminate bending stiffnesses, D
ij
, are given by

( ) ( ) ( )
3
1
3
1
2
2
2
3
1
= =

}
k k
N
k
k
ij
t
t
k
ij ij
z z Q dz z Q D [Eq. B1.4]

where Q
ij
are components of the stiffness matrix. In partitioned form equation B1.1 can
be written as
)
`
=
)
`
k
c
0
D B
B A
M
N
[Eq. B1.5]

where
(
D B
B A
is the stiffness matrix

In equation B1.5, the extensional stiffness matrix, A
i
, relates the induced mid-plane
strains, c
o
i
, to the induced in-plane forces, N
i
, and the bending stiffness matrix, D
i
,
relates the curvatures,k
i
, to the induced moments, M
i
. The coupling stiffness matrix, B
i
,
couples the induced mid-plane strains, c
o
i
, to the induced moments, M
i
, and the
curvatures,k
i
, to the induced in-plane forces, N
i
.
The inverted version of equation B1.5 is

)
`
=
)
`

M
N
D B
B A
1
0
k
c
[Eq. B1.6]

where
1
(
D B
B A
is the compliance matrix

or alternatively

)
`
=
)
`
M
N
D B
B A
' '
' '
0
k
c
[Eq. B1.7]

Induced Strain along the x axis due to a change in temperature, c
ATx
, is written as

[ ] T
x T
x
A =
A
o c [Eq. B1.8]

where AT and o
x
are the change in temperature and material thermal coefficient
respectively. Using B1.8, the thermal forces and moments due to temperatures change
are

[ ] ( )
1
1

=
A
A =
k k k
N
k
k T
z z Q T N o [Eq. B1.9]
[ ] ( )
2
1
2
1
2

=
A

A
=
k k k
N
k
k T
z z Q
T
M o [Eq. B1.10]

where [ ] Q is the stiffness matrix, equation B1.5. Using the piezoelectric/thermal action
analogy the induced strain from the piezoelectric ceramic, c
pzt
, is given by

[ ] T
t
V d
x
p
y
pzt
x
pzt pzt
A ~ = = = o c c c
31
[Eq. B1.11]

where d
31
, t
p
and V are the piezoelectric charge constant [250 *10
-12
mV
-1
in both x and
y direction], thickness and the applied voltage respectively. The in-plane forces, N, and
moments, M, are induced by the voltage applied to the piezoelectric material and can be
calculated using

[ ] ( )
1
1
31
=
=
k k
N
k
k
P
pzt
z z Q
t
V d
N [Eq. B1.12]
[ ] ( )
2
1
2
1
31
2

=
=
k k
N
k
k
P
pzt
z z Q
t
V d
M [Eq. B1.13]

The strain is induced within the piezoelectric layer, see fig. B1.2., and therefore the z
component of equations B1.12 and B1.13 can be written as

( )
p p k k
t t z z = =

0
1
[Eq. B1.14]
( )
s p
s p s p
k k
t t
t t t t
z z =
(
+
=

2 2
2
1
2
2 2
[Eq. B1.15]

This simplifies B1.12 and B1.13 to

[ ]
pzt pzt
Q V d N
31
= [Eq. B1.16]
[ ]
s pzt pzt
t Q
V d
M
2
31
= [Eq. B1.17]
and Q can be reduced to

[ ]
(
(
(
=
(
(
(
=
66
22 21
12 11
66 62 61
26 22 21
16 12 11
0 0
0
0
Q
Q Q
Q Q
Q Q Q
Q Q Q
Q Q Q
Q [Eq. B1.18]

where

2
21 12
1
11
1 1
p
p
E
E
Q
v v v
=
= [Eq. B1.19]

2
21 12
2 12
21 12
1 1
p
p p
E
E
Q Q
v
v
v v
v
= = [Eq. B1.20]

11 2
21 12
2
22
1 1
Q
E
E
Q
p
p
=
=
v v v
[Eq. B1.21]

P
G G Q = =
12 66
[Eq. B1.22]

[ ] [ ]
y x
Q Q Q Q Q Q = + = + = ) ( ) (
22 12 12 11
[Eq. B1.23]
substituting equations B1.19 to B1.21 into equations B1.23, B1.16 and B1.17 gives

[ ]
( ) ( )
p
p
p
p
p
p
p
y
pzt
x
pzt
VE d
t
E
t
V d
N N
v v
v
+ = =
1 1
1
31
2
31
[Eq. B1. 24]

[ ]
( )
s p
p
p
p
p
y
pzt
x
pzt
t t
E
t
V d
M M
2
31
1
1
2 v
v
+ = =
( )
x
pzt
s
s
p
p
N
t
t
VE d
2 1 2
31
=
=
v
[Eq. B1.25]

where E
P
, d
31
, V, v
P
and t
s
are the piezoceramic Youngs Modulus, piezoelectric charge
constant, applied voltage, piezoceramic Poissons ratio and substrate thickness
respectively.

B1.2. Unimorph deflection for a pressure load [o
W
]
The unimorph tip deflection, using CPLT under a given uniform pressure load, W, for a
cantilever beam, is given by

( )
eff
W
EI
WbL
* 8
4
= A [Eq. B1.26]

where L, M, and N are the total unimorph length, external applied moments and loads
respectively. Then the effective bending stiffness of the composite beam along its
length can be obtained as

( )
11
d
b
EI
eff
= [Eq. B1.27]
where b is the beam width and d
11
is the first element of the bending element of the
compliance matrix, that is the inverse of the laminate stiffness matrix, see eq. B1.7.

B1.3. Unimorph deflection for an applied voltage [o
V
]
Using piezoelectric and laminate theory the deflection obtained when the piezoceramic
has an applied voltage across it, o
V
, is given by

2
*
2
L
iy
V
k
= A [Eq. B1.28]

where k
iy
is the induced curvature along the y azis and is calculated by substituting
equations B1.24 and B1.25 into equation B1.7.

Chen, T.-Y., et al. (2003). "Effects of Strontium on the Dielectric and Piezoelectric
Properties of Sm-Modified PbTiO
3
Ceramics." Ferroelectrics 282: 37-47.

Appendix B2. FEM Grid Resolution B2-7
Appendix B2 FEM Grid Resolution

The final FEM grid resolution was selected on a trade-off between accuracy and
computational cost, see fig. B2.1. The computational process time required to generate
a solution is a function of the number of elements squared. In reality there is also a
limitation of the memory used by ANSYS to produce a solution, which increases with
the number of elements.

Figure B2.1 - ANSYS model showing final grid resolution

The model was divided into; the unimorph structure and the larger support structure
along the X-axis. The support structure is subdivided into three sections along the Y-
axis to create 4 sections (3 support structure and 1 unimorph structure). Each section is
further divided into 2 layers on the Z-axis, a lower 0.5mm layer and a higher 2mm
layer, creating 8 subsections. Only the 0.5mm layer subsection on the unimorph
structure has piezoelectric material properties, with the rest have aluminium material
properties. The initial grid resolution studies, in the subsequent sections, vary the
number of element along all three axes for each of the eight subsections.

B2.1. - Unimorph deflection -0kPa /-500V

A grid sensitivity study was conducted with the number of elements in all three planes
allowed to change. Initially the zero pressure condition was examined with only the
piezoelectric deflection considered, (figs. B2.2a to e). The contours are of constant
deflection as a percentage of the deflection prediction with an X, Y and Z resolution of

a)
b)
c)
d)
e)
Figure B2.2 Contours of percentage
accuracy of the FEM grid result whilst
varying the number of cells along the X
and Y axis under a 0 kPa pressure load
and a -500V applied voltage for
a) z=1, b) z=2, c) z=3, d) z=4 and e) z=5

20, 20 and 5 respectively. The value at this grid resolution was chosen under the
assumption that, due to it being the highest grid resolution, it would give the most
accurate prediction. It is observed that convergence on the exact solution is achieved
with minimal number of cells on the Y axis, whereas convergence is improved with an
increase in number of cells in the X axis. Furthermore, increasing the number of cells
in the Z axis improves the results. However, the skewed results imply that a minimal
number of cells on the Y axis would produce accurate results. This is misleading and
unintuitive as the reverse effect of the X and Y axis cell numbers negate each other
suggesting that the maximum number of cells does not produce the most accurate result.
It is believed that this is due to the poisons ratio, creating a curvature along the
unimorph X and Y axes increasing the unimorphs effective strength along the X-axis
which is more accurately represented with increased number of cells along the Y-axis.

A simplified representation of the contour graphs are shown in figure B2.3a & b, whilst
only varying one axis grid cell number at a time. It is observed, in figure B2.3a, the
effect of increasing the number of cells in the X axis with the number of cells in the Y
axis constant at 20. The initial effect is dramatic in improving the accuracy of the
solution, up to approximately the fifth element, when the results are within 1% of the
solution, and with more than eight grid cells the accuracy is within 0.5%.

a) b)
Figure B2.3 Percentage accuracy of FEM grid results under a 0 kPa pressure load and
a -500V applied voltage whilst varying number of cells
on the a) X-axis (Y=20) and b) Y-axis (X=20).

Again, changing the number of elements in Y plane has an initial dramatic effect on the
accuracy of the solution up to approximately the fifth element, which is within 0.2% of
the exact solution, see fig. B2.3b. The effect of further increasing the number of cells
along the Y axis improves the accuracy of the result. However, with a suspected
marked increase in computational cost.

The effect of changing the number of cells on the Z axis was observed to have a
minimal effect after the second element, with only the single Z element producing
markedly differing results. It is believed that using a Z resolution of 3 or more cells is
less efficient due to the increased computational time compared to the accuracy of the
solution
From the above results it was inferred that a minimal grid resolution of 8, 5 and 2
elements is required in the X, Y and Z axis to produce valid results. However,
increased element numbers would improve the validity at increased computation cost.

B2.2. - Unimorph deflection -17.64kPa /-500V

a)
b)
c) d)
e)
Figure B2.4 Contours of percentage
and Y axis under a -17.64 kPa pressure
load and a -500V applied voltage for
a) z=1, b) z=2, c) z=3, d) z=4 and e) z=5
A second grid sensitivity study was conducted, again, with the number of elements
along all three axes allowed to vary. The unimorph was subject to a uniform -17.64kPa
pressure load with a -500V applied voltage in order that the piezoelectric actuation
assisted the pressure deflection, see figs. B2.4a to e. The contours are of constant
deflection as a percentage of the deflection prediction with an X, Y and Z resolution of
20, 20 and 5 respectively. The value at this grid resolution was chosen under the
assumption that, due to it being the highest grid resolution, it would give the most
accurate value. The results are more intuitive than the previous grid sensitivity study as
convergence is observed with an increase in the number of cells along any of the three
axes. It is believed that the removal of the unintuitive poisons ratio effects (fig. B2.2a
e) are due to the pressure load dominates the unimorph deflection, see chapter 5.

Again, a simplified representation of the contour graphs are shown in figure B2.5a & b,
whilst only varying one axis grid cell number at a time. It is observed, in figure B2.5a,
the effect of increasing the number of cells in the X axis with the number of cells in the
Y axis constant at 20. The initial effect is dramatic in improving the accuracy of the
solution, up to approximately the fifth element, when the results are within 3.5% of the
solution. The accuracy converges to within 1% with more than eleven cells along the
X-axis.
a) b)
Figure B2.5 Percentage accuracy of FEM grid results under a -17.64 kPa pressure
load and a -500V applied voltage whilst varying number of cells

Increasing the number of elements in Y axis has an initial dramatic effect on the
accuracy of the solution up to approximately the fourth element, which is within 1% and
0.8% of the exact solution for Z=1 and Z=2 respectively, see fig. B2.5 b. The effect of
further increasing the number of cells along the Y axis improves the accuracy of the
result. However, it is believed that the efficiency of the accuracy is doubtful above the
eighth element.

The effect of changing the number of cells on the Z axis was observed to have a
minimal effect after the third element, with the second Z element producing slightly
under predictions and the single Z element producing markedly differing results. It is
believed that using a Z resolution of more than 3 cells is less efficient due to the
increased computational time compared to the accuracy of the solution

From the above results it was inferred that a minimal grid resolution of 11, 5 and 3
elements is required in the X, Y and Z axis to produce valid results. However,
increased element numbers would improve the validity at increased computation cost.

B2.3. - Unimorph deflection -17.64kPa /+500V

A third grid sensitivity study was conducted, again, with the number of elements along
all three axes allowed to vary. The unimorph was subject to a uniform -17.64kPa
pressure load with a -500V applied voltage in order that the piezoelectric actuation
inhibit the pressure deflection, see figs. B2.6a to e. This is the more critical of the three
grid sensitivity studies as zero deflection, when piezoelectric actuation negates the
pressure deflection, will enable active control.

The contours are of constant deflection as a percentage of the deflection prediction with
an X, Y and Z resolution of 20, 20 and 5 respectively. Again, this grid resolution was
chosen under the assumption that, due to it having the highest grid resolution, it would
give the most accurate solution. The results are similar to section B2.2. in that they are
more intuitive than the section B2.1. grid sensitivity study as convergence is observed
with an increase in the number of cells along any of the three axes. As stated earlier this
is the most critical of the three case studies as the total unimorph deflection is an order
of magnitude of less than the previous studies and this is used to determine whether zero
deflection is feasible for active control. Therefore, percentage errors can be
deceivingly higher with a comparison value closer to absolute zero. This is shown with
a contour of 0% percentage accuracy, see figs B2.6 a-e.

a)
b)
c) d)
e)
Figure B2.6. Contours of percentage
and Y axis under a -17.64 kPa pressure
load and a +500V applied voltage for
a) z=1, b) z=2, c) z=3, d) z=4 and e) z=5

A simplified representation of the contour graphs are shown in figure B2.7a & b, whilst
only varying one axis grid cell number at a time. It is observed, in figure B2.7a, the
effect of increasing the number of cells in the X axis with the number of cells in the Y
axis constant at 20. The initial effect is dramatic in improving the accuracy of the
solution from approximately 0% with only one element to 85%, within 15% of the exact
solution, with five elements on the X axis. Further convergence, to within 5%, is
observed with more than ten cells along the X-axis.
a) b)
Figure B2.7 Percentage accuracy of FEM grid results under a -17.64 kPa pressure
load and a +500V applied voltage whilst varying number of cells

Increasing the number of elements on the Y axis has an initial smaller effect on the
accuracy of the solution than on the X-axis. The accuracy of the result with five
elements on the Y-axis within 6%, 4% and 3% of the exact solution for Z=1, Z=2 and
Z=3 respectively, see fig. B2.7b. Increasing the number of cells along the Y axis to
eleven improves the accuracy of the result to with 1 % for Z=3.

Again, the effect of changing the number of cells on the Z axis was observed to only
have an effect on the Y Grid resolution. The single Z element producing markedly
differing results which converges to the solution with a second element. It is believed
that using a third cell is more than sufficient and any further increase in Z resolution is
less efficient due to the increased computational time compared to the minimal accuracy
gained.

From the above results it is inferred that a minimal grid resolution of 10, 11 and 3
elements is required in the X, Y and Z axis to produce valid results. Increasing the
number of elements further improves the results. However, it is doubtful of the worth
as the computational cost would be significantly higher. Furthermore, percentage
accuracy of results may be less appropriate than the previous grid sensitivity studies
because the comparison (100%) value is close to zero and, therefore, difference in
predicted deflection may be a more appropriate comparison.
B2.4. - Final Grid Sensitivity

The above grid resolution studies varied the number of elements on the X-, Y- and Z-
axis by the same amounts for all eight subsections. An improved grid resolution study
would vary the number of elements along the three axes independently in each
subsection. However, the FEM 20, 20, 5 solution was utilised to identify the regions of
comparatively large stress and deflection. At these points the grid resolution was
increased and in regions of comparatively low stress and deflection the grid resolution
was decreased.

The final grid resolution to minimise error and computational cost was selected as 10-20
elements on the X-Axis, 10-16-10 elements on the Y axis and 1-3 elements on the Z
axis, as shown in fig. B2.1. The time for a single solution reduced from 150 minutes for
the 20,20,5 grid resolution to three and a half minutes for the final grid resolution, see
table B2.1.

Applied Loads Predicted Tip Deflection (mm)
Pressure
(kPa)
Applied
Voltage (V)
20,20,5 Grid
(9000 secs)
Final Grid
(210 secs)
CLPT
(<1 sec)
Difference
(m)
Error (%)
0 -500 -0.1473 -0.1473 -0.1190 0 / +28.2 0 / +19.16
-17.64 -500 -0.3563 -0.3533 -0.3121 +3 / +44.2 +0.84/ +12.41
-17.64 500 -0.0618 -0.0596 -0.0656 +2.2 / -3.8 +3.56 / -6.15
Table B2.1 - Comparisons of CLPT, final grid resolution to the exact solution.

It is stated, in chapter 5, that a unimorph deflection graph took 110hours to obtain the
400 unimorph deflection predictions, which is four fold the time to produce 400
individual solutions. The parametric design language program which ran all 400
solutions as one job was considerably retarded by the memory space required and
available to compute such a job and, hence, took longer than first considered. However,
the autonomous nature of the program freed up operator time and it is believed that to
run 400 jobs individually would have taken considerably more than 110 hours of
operator time compared to the single submission of the larger program.

Appendix C1. ANSYS Program C1-1
Appendix C1 ANSYS Program

C1.1. Parametric Deign Language
/PREP7

*SET,TREPEAT,20 ! Number of Substrate Thickness
*SET,YOREPEAT,40 ! Number of Substrate Youngs Modulus
*SET,VOLT,0 ! Applied Voltage
*SET,JOBS

JOBS=TREPEAT*YOREPEAT

*SET,YO1
*SET,YO11
*SET,YO12
*SET,YO2,6.4E10
*SET,K

*DIM,YO,,YOREPEAT
YO1=0.05 ! Starting Youngs Modulus Ratio
YO12=0.05 ! Starting Youngs Modulus Ratio Increment

*DO,I,1,YOREPEAT !Number of Youngs Modulus Increment

YO(I)=YO1
YO11=YO1+YO12
YO1=YO11
*IF,I,EQ,9,THEN
YO12=0.1
*ENDIF
*IF,I,EQ,14,THEN
YO12=0.25
*ENDIF
*ENDDO

*DIM,NAME,CHAR,1
NAME(1)='NULDC3p2' ! Output Filename
*DIM,LOAD,CHAR,1
LOAD(1)='loadc30' ! Input Load Filename

*SET,Z1,0
*SET,Z2 ! Thicknesses
*SET,Z3
*SET,Z4,-0.005
*SET,Z
*SET,Z11
*SET,Z12
Z12=2.5 ! Starting Thickness Ratio T
S
/T
PZT

Z11=0.25 ! Starting Thickness Ratio Increment
*SET,E1,10 ! X Substrate Plate Divisions
*SET,E2,20 ! X Unimorph Divisions (Maybe Changeable)
*SET,E3,10 ! Y Substrate Plate Divisions
*SET,E4,16 ! Y Unimorph Divisions
*SET,E5,3 ! Z Substrate Plate Divisions
*SET,E6,1 ! Z Unimorph Divisions
*SET,E7,1 ! Z Support Plate Divisions

*SET,E8 ! Input Load Divisions
*SET,E9 ! And Variables
*SET,E10
*SET,E11

E8=E2+2
E9=E4+1
E10=(E2*E4)+E4
E11=E2+1

*SET,ECOUNT
*SET,ECOUNT1
*SET,ECOUNT2
*SET,ECOUNT3

*SET,X1,0 ! Lengths
*SET,X2,0.052
*SET,X3,0.102
*SET,X4,0.1021
*SET,X5,0.1027
*SET,X6,0.104

*SET,Y1,0 ! Widths
*SET,Y2,0.0395
*SET,Y3,0.0645
*SET,Y4,0.104
Y6=(Y2+Y3)/2 ! Unimorph Tip Mid Position

*SET,P1,-17640
*SET,P2
*SET,P3
*SET,P4
*SET,P5

*SET,NX1 ! Input Load Variable
*SET,NY1
*SET,NX2
*SET,NY2

*DIM,FLAPLOAD,TABLE,E8,E9,,X-COORD,Y-COORD

*TREAD,FLAPLOAD,%LOAD(1)%,txt,,1

*DIM,RESULTS,TABLE,TREPEAT,YOREPEAT,,SUBTHICK,YOUNGMOD

*DO,I,1,YOREPEAT
RESULTS(0,I)=YO(I) ! Labelling Results Matrix
*ENDDO
K=1
N=1
!* Geometry
FINISH
*DO,J,1,TREPEAT
! Load Structure
Z2=Z12*0.1*Z4 ! Set Substrate Thickness
Z3=Z2-0.0005

M=1
YO1=YO2*YO(K) ! Set Substrate Youngs Modulus

ET,1,SOLID95 ! Substrate Material Thickness
R,1,.01
MP,EX,1,YO1
MP,NUXY,1,0.3
MP,DENS,1,2700
MP,ALPX,1,0

ET,2,SOLID95 ! Piezoelectric Material Thickness
R,2,.01
MP,EX,2,YO2
MP,NUXY,2,0.45
MP,DENS,2,7700
MP,ALPX,2,5E-7
MP,ALPY,2,5E-7
MP,ALPZ,2,-9E-7

FINISH
/SOLU

FLST,2,21,5,ORDE,15
FITEM,2,4
FITEM,2,-6 ! Set Boundary Conditions
FITEM,2,8
FITEM,2,11
FITEM,2,-13
FITEM,2,17
FITEM,2,-19
FITEM,2,21
FITEM,2,24
FITEM,2,-26
FITEM,2,30
FITEM,2,-32
FITEM,2,34
FITEM,2,37
FITEM,2,-39
/GO
DA,P51X,ALL,0

FLST,2,3,5,ORDE,2
FITEM,2,1
FITEM,2,-3
/GO
SFA,P51X,,PRES,P1 ! Apply Load to Support Plate

ECOUNT=1801
ECOUNT2=1
ECOUNT3=1

*DO,I,1,E10

NX1=ECOUNT3
NY1=ECOUNT2 ! Apply Input Load to Substrate Plate
NX2=ECOUNT3+1
NY2=ECOUNT2+1

P2=FLAPLOAD(NX2,NY1)


FLST,2,1,2,ORDE,1
FITEM,2,ECOUNT
/GO
SFE,P51X,5,PRES,,P2,P3,P4,P5

ECOUNT1=ECOUNT+3
ECOUNT=ECOUNT1

ECOUNT1=ECOUNT2+1
ECOUNT2=ECOUNT1

*IF,ECOUNT2,EQ,E9,THEN
ECOUNT2=1
ECOUNT1=ECOUNT3+1
ECOUNT3=ECOUNT1
*ENDIF

*IF,ECOUNT3,EQ,E11,THEN
ECOUNT=3081
E11=E8
*ENDIF

*ENDDO
FINISH

Z=-1000*Z2

RESULTS(J,0)=Z

*DO,K,1,YOREPEAT
/SOLU
FLST,2,13,6,ORDE,2
FITEM,2,1
FITEM,2,-13
/GO
BFV,P51X,TEMP,VOLT ! Input Applied Voltage to Results Matrix

YO1=YO2*YO(K)

MP,EX,1,YO1 ! Input Youngs Modulus to Results Matrix

NUM=NODE(X6,Y6,Z1) ! Select Nearest Mode to the unimorph tip centre

SOLVE
FINISH

DEFL=UZ(NUM)

/PREP7
RESULTS(J,K)=DEFL ! Input Deflection Solution to Results Matrix
SAVE,'%NAME(1)%','db','',ALL
FINISH

ECOUNT=N+1
N=ECOUNT

ECOUNT=M+1
M=ECOUNT
*ENDDO

/PREP7
FINISH

Z=Z12+Z11
Z12=Z

*ENDDO

*CREATE,ansuitmp ! Write Results Matrix to File
*CFOPEN,'%NAME(1)%','TXT',''
*VWRITE,RESULTS(0,0),RESULTS(0,1),RESULTS(0,2),RESULTS(0,3),RESULTS(0,4),RE
SULTS(0,5),RESULTS(0,6),RESULTS(0,7),RESULTS(0,8),
(F16.7,F16.7,F16.7,F16.7,F16.7,F16.7,F16.7,F16.7)
*CFCLOS
*END
/INPUT,ansuitmp

Appendix C2. ASYST Program C2-6
Appendix C2 ASYST Program

C2.1.

\ Scanivalve controller
\
\ This program is designed to control the Scanivalve and to take
\ Stagnation Pressures/Static Pressure/ Temperature measurements in order

\ Original program written by Eric Magi, modified by Andrew Roberts
\ and Robert Green for subsonic wind tunnel.
\
\ Modified by Jon Couldrick July 2002

FORGET.ALL

RTI-800/815

INTEGER DIM[ 2400 ] ARRAY DATA.BUFFER1
REAL DIM[ 1 ] ARRAY DATA.MEAN
12 STRING FILENAME
INTEGER SCALAR COUNT

1 1 A/D.TEMPLATE CHNL01
DATA.BUFFER1 TEMPLATE.BUFFER
50 TEMPLATE.REPEAT
1.5 CONVERSION.DELAY

50 TEMPLATE.REPEAT

20 TEMPLATE.REPEAT

0 DIGITAL.TEMPLATE SCAN.STEP
BINARY
01000000, DIGITAL.MASK

0 DIGITAL.TEMPLATE SCAN.HOME
10000000, DIGITAL.MASK
DECIMAL

: HOME.SCAN
SCAN.HOME
1 , DIGITAL.OUT
0 , DIGITAL.OUT
5 MSEC.DELAY
1 , DIGITAL.OUT
BEGIN
STACK.CLEAR
CR ." Waiting for scanivalve to go home"
DIGITAL.IN DUP .
1. = IF TRUE
ELSE FALSE
THEN
UNTIL
CR ." scanivalve homed"
;
: STEP.SCAN
SCAN.STEP
1 , DIGITAL.OUT
0 , WRITE.BITS
5 MSEC.DELAY
1 , DIGITAL.OUT
CR ." Scanivalve stepped 1 position"
;
: CLEAR.DATA
0 DATA.BUFFER1 :=
0 DATA.BUFFER2 :=
0 DATA.BUFFER4 :=

: COLLECT.DATA
CLEAR.DATA

CHNL02
A/D.IN>ARRAY
CR
?BUFFER.INDEX .

48 0 DO

CHNL01
A/D.IN>ARRAY
CR
?BUFFER.INDEX .
STEP.SCAN

CHNL02
A/D.IN>ARRAY
CR
?BUFFER.INDEX .

CHNL04
A/D.IN>ARRAY
CR
?BUFFER.INDEX .

LOOP

: INITIALISE
DAS.INIT
SCAN.STEP
1, WRITE.BITS
HOME.SCAN

: STORE.RAW
CR ." NAME OF FILE TO BE CREATED ? "
CR
"INPUT FILENAME ":=

FILENAME DEFER> OUT>FILE
CONSOLE.OFF
." BASE PRESSURE " CR
1 COUNT :=

2400 0 DO
DATA.BUFFER1 [ COUNT ] . CR
1 COUNT +
COUNT :=
LOOP

CR ." STAGNATION PRESSURE " CR
1 COUNT :=
2450 0 DO
1 COUNT +
COUNT :=
LOOP

CR ." STAGNATION TEMPERATURE " CR
1 COUNT :=
960 0 DO
1 COUNT +
COUNT :=
LOOP
CONSOLE
OUT>FILE.CLOSE

: COLL.DATA1
INITIALISE
50 MSEC.DELAY
COLLECT.DATA
STORE.RAW

Appendix C3. PSP Program

C3-9
%PSP.m
%Program to compare the intensity ratio of wind off (uniform pressure) and
%wind on (variable pressure) images using pressure sensitive paints

clear
close all

readwoffs %A Program to read all the Wind Off image
readwons %A Program to read all the Wind On image
calccentres %A Program to calculate the centre of the test surface
inputPmin1 %A Program to input all the minimum pressures
inputPmax1 %A Program to input all the maximum pressures
inputPandI %A Program to input the pressures and intensity ratios
footprint %A Program to read the coordinates of the
%uncontrolled structure lines

for runnum=1:dateruns %For every run
for picno1:daterunpics(runnum) %For every image of a run
divno=4; %The number of linear pixels to be smoothed
divarea=divno*divno; %The number of area pixels to be smoothed
imredon1=imredon{runnum,picno}; %Select Wind on image
imgreenon1=imgreenon{runnum,picno};
imblueon1=imblueon{runnum,picno};

x_position = 0; %Start at the beginning of the image
y_position = 0;

y_no=240/divno; %Reduce image size to number of
x_no=320/divno; %smoothed tiles
y_nomin1=y_no-1;
x_nomin1=x_no-1;

datenum1=2;
runno1=woff1(runnum);

imredoff1=imredoff{datenum1,runno1}; %select Wind off image
imgreenoff1=imgreenoff{datenum1,runno1};
imblueoff1=imblueoff{datenum1,runno1};

intensity = zeros(y_no,x_no); %define intensity matrix

Pmin=Pmint{runnum}(picno); %Read minimum pressure
Pmax=Pmaxt{runnum}(picno); %Read maximum pressure
Pdif=dP{runnum}(picno); %Read pressure scale

C3-10
for k=1:y_no
for l=1:x_no
x1=(x_position+(divno*(l-1)));
y1=(y_position+(divno*(k-1)));
WONintensity = zeros(divno,divno);
for j=1:divno
for i=1:divno
x2=x1+j;
y2=y1+i;
%Calculate Widn On intensity
WONintensity(i,j)=imredon1(y2,x2)+imgreenon1(y2,x2)+imblueon1(y2,x2);
end
end

for j=1:divno
for i=1:divno
x2=x1+j;
y2=y1+i;
%Calculate Widn Off intensity
WOFFintensity(i,j)=imredoff1(y2,x2)+imgreenoff1(y2,x2)+imblueoff1(y2,x2);
end
end

WONintensity_sum=0;
WOFFintensity_sum=0;
for j=1:divno
for i=1:divno
WONintensity_sum=WONintensity_sum+WONintensity(i,j);
WOFFintensity_sum=WOFFintensity_sum+WOFFintensity(i,j);
end
end

%Average pixel values to get one value for divno*divno pixel patch
av_WONintensity=WONintensity_sum/divarea;
av_WOFFintensity=WOFFintensity_sum/divarea;

if av_WOFFintensity == 0
Pressure = 0;
else
Iratio = av_WONintensity/av_WOFFintensity;
%Change intensity ratio into a pressure
Pressure = (Patm-(I1*Iratio)+(I2*Iratio^2))/PStag;
end

if Pressure <= Pmin
Pressure = Pmin;
end

C3-11
if Pressure >= Pmax
Pressure = Pmax;
end

if p>=r1 %Only calculate for the test surface
Pressure = Pmin;
end
end
end

run=int2str(runno1); %Compute Figure Title and
run1=strcat(m1,run); %Select Directory to save figure
date1=date(datenum1);
date2=datea;
pic=(((picno-1)*3)+1);
pz=int2str(daterunspics{runnum}(pic));

rab=int2str(runnum);
cd (rab)

pab=int2str(picno);
cd (pab)

end

figtitle=strcat(w1,run1,'0',date1,w2,run2,'0',date2,'-',pz,'Pmin-',Pminz,'Pmax-',Pmaxz);
mfig=strcat('Pmin-',Pminz,'Pmax-',Pmaxz,'.jpg');
Range=[Pminj Pmaxj];
figure;
surf(x,y,intensity); %Print Figure
shading interp
caxis(Range);
colormap(hsv) %Select Colour Range
colorbar %Legend
title(figtitle)
view(0,-90);
hold on

plot(fpx1,fpy1,'Color','k','LineWidth',0.5) %Plot Uncontrolled Footprint
hold on
plot(fpx2,fpy2,'Color','k','LineWidth',0.5)
hold on
hold on
hold on
end

C3-12
Appendix D1. OEM Amplifiers

D1-1
D1. OEM Amplifiers
Appendix D1. OEM Amplifiers

D1-2
Appendix E1. Mass Flow within the Boundary layer

E1-1
Mass Flow within the Boundary Layer between any distances
Assuming a perfect gas,

RT P = . [Eq. E1.1]

This make C
P
constant (C
P
=1.005 kJkg
-1
K
-1
) and

) (
) (
y T
T y
e
e
=
[Eq. E1.2]

Also, assuming a one seventh power law boundary layer profile gives

7
1
) (
|
.
|
\
|
=
o
y
U
y U
e
[Eq. E1.3]
and
} }
= =
2
1
2
1
) ( ) (
) ( ) (
y
y
e
e
e
e
y
y
dy U
U
y U y
dy y U y m
& . [Eq. E1.4]

Substituting equation E1.2 & E1.3 into equation E1.4 gives

} }
= =
2
1
2
1
) ( ) (
7
1
7
1
7
1
7
1
y
y
e e e
y
y
e
e e
dy
y T
y T U
dy
y
y T
T
U m
o
o
& . [Eq. E1.5]

The temperature profile in the boundary layer can be written as

( )
P e
W AW W
C
U r
U
U
T T T y T
2
) (
2
+ = [Eq. E1.6]

Assuming adiabatic wall conditions, that is

1 =
W
AW
T
T
, [Eq. E1.7]

equation E1.6 reduces to

( )
2
2
2
2
1
2
) ( U r T C
C C
U r
T y T
W P
P P
W
= = . [Eq. E1.8]

Where r is the recovery factor and is taken as 0.892 at M=2.

E1-2
Then combining equations E1.5 and E1.8 gives

} }

=
=
2
1
2
1
2
7
1
7
1 2
7
1
7
1
2
2
2
y
y
W P
P e e e
y
y
P
W
e e e
dy
U r T C
y C T U
dy
C
U r
T
y T U
m
o
& . [Eq. E1.9]

Further simplification using

2
2
2 2
2 ) (
) (
e
e
U
y U
rU y rU U r = = , [Eq. E1.10a]

combining equations E1.3 with E1.10a gives

7
2
7
2
2
2
o
y
rU U r
e
= [Eq. E1.10b]
and

|
.
|
\
|
=
7
2
2
7
2
7
2
2
2
1
2 y rU T C U r T C
e W P W P
o
o
[Eq. E1.11]

Then combining equation E1.11 and E1.9 gives

} }

=
=
2
1
2
1
7
2
3 2
7
1
1
7
2
2
7
2
7
1
7
1
2
2
y
y
y
y
e W P
P e e e
dy
y C C
y
C dy
y rU T C
y
C T U m
o
o & [Eq. E1.12a]

where
7
1
1
2 o
P e e e
C T U C = [Eq. E1.12b]
W P
T C C
7
2
2
2o = [Eq. E1.12c]
2
3 e
rU C = [Eq. E1.12d]

Equation E1.12a can be reduced to

} }

=
=
2
1
2
1
7
2
5
7
1
4
7
2
3
2
7
1
3
1
y
y
y
y
dy
y C
y
C dy
y
C
C
y
C
C
m& [Eq. E1.13a]
where
e
P e e
rU
C T
C
C
C
7
1
3
1
4
2 o
= = [Eq. E1.13b]
2
7
2
3
2
5
2
e
P W
rU
C T
C
C
C
o
= = [Eq. E1.13c]

E1-3
Integration of equation E1.13 gives

|
|
.
|
\
|
+
(
(
|
|
.
|
\
|

|
|
.
|
\
|

(
(
|
|
.
|
\
|

=
5
7
2
1
5
7
2
1
2
5
7
4
1
5
7
2
1
5
7
2
2
5
7
2
2
2
5
7
4
2
5
7
2
2
3
5 4
1 ln 12
6 4
12
1
1 ln 12
6 4
12
1
2
7
1 2
C
y
C
y
C
y
C
y
C
y
C
y
C
y
C
y
C C m
y y
& [Eq. E1.14]

When y
2
is the displacement thickness (o*) and y1 is a flat plate (0mm) this reduces to

|
|
.
|
\
|

(
(
|
|
.
|
\
|
=
5
7
2
5
7
2
2
5
7
4
5
7
2
3
5 4 *
*
1 ln 12
* 6 * 4 *
12
1
2
7
C C C C
C C m
o o o o
o
& [Eq. E1.15a]

e
P e e
rU
C T
C
7
1
4
2 o
= [Eq. E1.15b]
and
2
7
2
5
2
e
P W
rU
C T
C
o
= [Eq. E1.15c]

Proof of integration (Equation E1.13 to equation E1.14)
Equation E1.14 is applied from y
2
=y and y
1
=0 giving

|
|
.
|
\
|

(
(
|
|
.
|
\
|
=
5
7
2
5
7
2
2
5
7
4
5
7
2
3
5 4
1 ln 12
6 4
12
1
2
7
C
y
C
y
C
y
C
y
C C m
y
& . [Eq. E1.16]

Using the substitution of

5
7
2
C
y
X = [Eq. E1.17]

equation E1.16 becomes

( ) ( )
)
`
= X X X X C C m
y
1 ln 12 6 4
12
1
2
7
2 3
5 4
& [Eq. E1.18]
or
( )
)
`

|
|
.
|
\
|
= X X
X X
C C m
y
1 ln
2 3 2
7
2 3
3
5 4
& [Eq. E1.19]

E1-4
Then differentiating equation E1.19 gives

( )
)
`

|
|
.
|
\
|
= = X
dX
d
X
X X
dX
d
dy
dX
C C
dy
dX
dX
m d
dy
m d
y y
1 ln
2 3 2
7
2 3
3
5 4
& &
. [Eq. E1.20]

Also, differentiating equation E1.17 yields

5
7
5
7
2
C
y
dy
dX

= [Eq. E1.21]

Combining equation E1.20 and E1.21 produces

( )
)
`
+ =

X
X X y C C
dy
m d
y
1
1
1
2
7
5
2
5 4
&
[Eq. E1.22]

Finally, substituting equation E1.17into equation E1.21 results in

|
|
.
|
\
|
+ =

2
5
7
2 2
5
7
2
2
5
7
4
7
5
2
5 4
1
1
1
C
y
C
y
C
y
y C C
dy
m d
y
&
[Eq. E1.23]
or
7
2
5
7
1
4
7
5
3
5
7
5
2
5
7
3
5
7
1
4
1
y C
y C
y y
C
y
C
y
C
y
C
dy
m d
y
=
&
. [Eq. E1.24]

The integral of equation E1.13a and therefore proving that equation E1.13a integrates to
equation E1.14.

Whole

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Whole

Uploaded by

Copyright:

Available Formats

School of Aerospace, Civil and Mechanical Engineering

due to the downstream subsonic influence, see fig. 1.7. Seddon

without the presence of shocks.

) with weak shocks, show evidence of separation.

, above 1.6 with the very simple

=0.9 and a 6Q angle of attack, that slot blowing at

. They also observed that with suction the

greater than 0.8 (M

Estimated by the Author.

, to produce sufficient mass bleed without producing

It should be noted that alternating fields can have the same

The two experiments were conducted approximately a year apart.

with closed flaps. This would

to allow mass transfer across the unimorph lip.

. The pressure fluctuations

This is because there will be no shock effect for a Mach wave.

Angles estimated by Author.

Angles estimated by Author.

& . [Eq. E1.4]

& . [Eq. E1.9]

You might also like