Você está na página 1de 139

SUPERSONIC TANDEM WEDGE FLOW IN EXTREME GROUND

EFFECT

_______________

A Thesis

Presented to the

Faculty of

San Diego State University

_______________

In Partial Fulfillment

of the Requirements for the Degree

Master of Science

in

Aerospace Engineering

_______________

by

Alyson George

Spring 2011
iii

Copyright 2011
by
Alyson George
All Rights Reserved
iv

DEDICATION

This thesis is dedicated to my family and Godfather without whom none of this would
have been possible.
v

ABSTRACT OF THE THESIS

Supersonic Tandem Wedge Flow in Extreme Ground Effect


by
Alyson George
Master of Science in Aerospace Engineering
San Diego State University, 2011

A two-dimensional computational analysis was performed on a tandem wingbox that


is the foundation of a supersonic Magnetically Levitated test sled designed and manufactured
at General Atomics and the Air Force at Holloman Air Force Base. The effects of Mach
number, ground clearance and distance between the wingboxes on the supersonic flow
characteristics, and lift and drag loads were investigated. Computational results were
compared against experiments on the tandem wingbox flow performed in the SDSU
supersonic blow down tunnel.
CAD and grid models were generated in Gambit so that grid cells aligned with
shocks. Fluent, a commercially available Computational Fluid Dynamics software, was used
to simulate the flows. The compressible, turbulent flows over the tandem wingbox were
modeled with a Reynolds Averaged Navier-Stokes model closed based on the standard k-
turbulence model. A grid convergence study was conducted with third-order MUSCL
scheme ensuring a minimal effect of numerical errors.
The flow between the wingbox and ground behaves similar to the intake flows of
high-speed propulsions systems. At a lower Mach number of 2, a combination of back
pressure and relatively low frontal clearance area, results in the formation of a detached
normal shock upstream of the tandem wingbox configuration. At a higher Mach number of
4, the detached shock is effectively swallowed by the intake, and an oblique shock system
forms.
The wake behind the front wingbox determines the flow over the rear wingbox. The
wake is deflected away from the ground for all cases considered here, since the pressure
below the wingboxes is higher than the pressure above the wingbox. As a result a shear layer
impinges on the rear wingbox. The pressure force further creates a significant mass flow
from below the wingbox to above the wingbox.
At Mach 2, a decreased ground clearance resulted in an increased pressure along the
bottom of the front wingbox, hence increasing lift and drag. The mass flow is changing due
to the change in ground clearance, therefore decreasing the pressure as ground clearance
increases. At Mach 4 the front wing box load increased with decreasing ground clearance,
whereas the rear wingbox showed a maximum lift and drag at medium range ground
clearances. A moving wall resulted in a more accurate representation of ground effect
because the boundary layers did not develop along the wall as with a stationary wall, which
lead to a separation bubble before the rear wingbox.
The computational results were just in reasonable agreement with experimental
results. The two rods that hold the two wing boxes together were not modeled in the
vi

computation. The interaction of the rods with the wakes and shocks behind the first wingbox
are considered the primary reasons for disagreements between computation and experiment.
vii

TABLE OF CONTENTS

PAGE

ABSTRACT...............................................................................................................................v
LIST OF TABLES ................................................................................................................... ix
LIST OF FIGURES ...................................................................................................................x
LIST OF SYMBOLS ............................................................................................................ xvii
ACKNOWLEDGEMENTS ................................................................................................... xxi
CHAPTER
1 INTRODUCTION .........................................................................................................1
2 COMPUTATIONAL METHODOLOGY ...................................................................10
2.1 Governing Equations .......................................................................................10
2.2 Turbulence Models ..........................................................................................11
2.3 Numerical Methodology ..................................................................................14
3 COMPUTATIONAL MODEL ....................................................................................16
3.1 Experimental Model Configurations................................................................16
3.2 Computational Model ......................................................................................17
3.3 Mesh Configuration .........................................................................................20
4 CONVERGENCE STUDIES ......................................................................................25
5 FLOW CHARACTERISTICS OF TANDEM WINGBOX IN GROUND
EFFECT .......................................................................................................................31
5.1 Tandem Wingbox Configuration at = 2 ....................................................31

5.1.1 Bow Shock and Intake Theory for Area Under the Front
Wingbox ...........................................................................................................31
5.1.2 Flow Over the Top of Front Wingbox ....................................................34
5.1.3 Wake Flow Between the Front and Rear Wingbox ................................35
5.1.4 Flow Over the Top of Rear Wingbox .....................................................36
5.1.5 Flow Under the Bottom of Rear Wingbox ..............................................37
5.2 Tandem Wingbox Configuration at = 4 ....................................................38
viii

5.2.1 Detached Oblique Shock and Flow Over the Top of Front
Wingbox ...........................................................................................................38
5.2.2 Flow Under the Bottom of Front Wingbox.............................................39
5.2.3 Wake Flow Between Front and Rear Wingbox ......................................40
5.2.4 Flow Over the Top of Rear Wingbox .....................................................42
5.2.5 Flow Under the Bottom of Rear Wingbox and the Wake Behind ..........42
5.3 Effect of Mach Number ...................................................................................43
6 EFFECT OF GROUND CLEARANCE AND DISTANCE BETWEEN THE
TANDEM WINGBOXES FOR MACH 2 RUNS .......................................................46
6.1 Effect of Ground Clearance, H .......................................................................46
6.2 Flow Features With the Effect of Distance Between Wingboxes, S ...............55
6.3 Effect of Distance Between Wingboxes, S ......................................................57
7 EFFECT OF GROUND CLEARANCE AND DISTANCE BETWEEN THE
TANDEM WINGBOXES FOR MACH 4 RUNS .......................................................68
7.1 Effect of Ground Clearance, H .......................................................................68
7.2 Flow Features With the Effect of Distance Between Wingboxes, S ...............77
7.3 Effect of Distance Between Wingboxes, S ......................................................78
8 LIFT AND DRAG LOAD ANALYSIS ON TANDEM WINGBOXES ....................88
9 EXPERIMENTAL AND CFD DATA COMPARISON .............................................95
9.1 Pressure Comparison With Experimental Data ...............................................95
9.2 Effect of a Stationary Wall Versus a Moving Wall .......................................100
10 CONCLUSION ..........................................................................................................107
REFERENCES ......................................................................................................................110
APPENDIX
CDF CONTOURS FOR DIFFERENT CONFIGURATIONS ........................................112
ix

LIST OF TABLES

PAGE

Table 2.1. Standard k- vs. SST k- Models: Advantages and Disadvantages .....................13
Table 3.1. List of the Cases Used in Experiment and CFD Analysis ......................................17
x

LIST OF FIGURES

PAGE

Figure 1.1. Single sled configuration for the testing at Holloman Air Force Base....................2
Figure 3.1. Tandem wingbox configuration used for the experiment in the supersonic
wind tunnel...................................................................................................................16
Figure 3.2. Dimensional setup of the Baseline Case, , configuration used in
Fluent for CFD analysis. ..............................................................................................18
Figure 3.3. Baseline Case, , of the tandem wingbox configuration used in
Fluent for the CFD analysis. ........................................................................................18
Figure 3.4. Picture of initial shock theory done on the Baseline Case for mesh
generation references. ..................................................................................................21
Figure 3.5. Breakdown of the near wall regions in terms of the turbulence factor . ...........23
Figure 3.6. Zoomed-in picture of the grid fabrication of the Baseline Case. ..........................24
Figure 4.1. Grid convergence study for the Baseline Case, , of the pressure
coefficient over the front wingbox at = 2 ..............................................................26
Figure 4.2. Grid convergence study for the Baseline Case, , of the pressure
coefficient over the rear wingbox at = 2. ..............................................................26
Figure 4.3. Adaptation of the grid in order to reach a in the desired range of the
log-layer. ......................................................................................................................28
Figure 4.4. Grid convergence study for the Baseline Case, , of the turbulence
factor, , for the front wingbox at = 2. ...............................................................28
Figure 4.5. Example of a Mach contour for the Case when the simulation blows
up at the location where the shock wave impinges on the wall. ..................................29
Figure 4.6. Schlieren image of the Case from the wind tunnel experiment. .................30
Figure 5.1. Mach number contour for the Baseline Case, , with a moving wall at
= 2. ........................................................................................................................31
Figure 5.2. Mach contour with the streamlines plotted of the Baseline Case, ,
with a moving wall at = 2......................................................................................32
Figure 5.3. Mach contour of the wake area of the Baseline Case, , with a
moving wall at = 2.................................................................................................34
Figure 5.4. Pressure contour of the Baseline Case, , with a moving wall at =
2....................................................................................................................................35
xi

Figure 5.5. Pathlines of the wake area for the Baseline Case, , with a moving
wall at = 2..............................................................................................................36
Figure 5.6. Total static pressure in the wake area for the Baseline Case, , with a
moving wall at = 2. ...............................................................................................37
Figure 5.7. Mach contour of the Baseline Case, , with a moving wall at = 4. ........38
Figure 5.8. Mach contour with the streamlines plotted of the Baseline Case, ,
with a moving wall at = 4......................................................................................39
Figure 5.9. Total pressure contour of the Case, , with a moving wall at = 4............40
Figure 5.10. Mach contour of the wake area for the Baseline Case, , with a
moving wall at = 4. ...............................................................................................41
Figure 5.11. Contour of the pathlines in the wake area for the Baseline Case, ,
with a moving wall at = 4......................................................................................41
Figure 5.12. Area ratio reference for analysis of a swallowed shock. .....................................44
Figure 6.1. Pressure coefficient over the front wingbox at various heights above the
moving wall with the wingboxes 1m apart at = 2. ................................................46
Figure 6.2. Pressure coefficient over the rear wingbox at various heights above the
moving wall with the wingboxes 1m apart at = 2. ................................................48
Figure 6.3. Mach contour of the wake area of the Case with a moving wall at
= 2 in reference to the reattachment location of the shear layers on the rear
wingbox........................................................................................................................49
Figure 6.4. Pressure along the moving wall with the wingboxes at various heights
above the wall at 1m apart at = 2. .........................................................................50
Figure 6.5. Mach number along the moving wall with the wingboxes at various
heights above the wall at 1m apart at = 2 ..............................................................51
Figure 6.6. Temperature (K) along the moving wall with the wingboxes at various
heights above the wall at 1m apart at = 2. .............................................................52
Figure 6.7. Turbulence factor, , along the moving wall with the wingboxes at
various heights above the wall at 1m apart at = 2. ................................................53
Figure 6.8. Pressure coefficient over the front wingbox with the wingboxes at various
heights above the moving wall at 1.5m apart at = 2. ............................................54
Figure 6.9. Pressure coefficient over the rear wingbox at various heights above the
moving wall with the wingboxes at 1.5m apart at = 2 ..........................................55
Figure 6.10. Mach contour for the Case with a moving wall at = 2 .......................56
Figure 6.11. Mach contour of the wake area for the Case with a moving wall at
= 2 .........................................................................................................................57
xii

Figure 6.12. Pathline contour of the wake area for the Case with a moving wall
at = 2......................................................................................................................57
Figure 6.13. Pressure contour of the wake area for the Case with a moving wall
at = 2......................................................................................................................58
Figure 6.14. Total pressure contour of the wake area for the Case with a
moving wall at = 2 ................................................................................................58
Figure 6.15. Pressure coefficient over the front wingbox closest to the moving wall
with the wingboxes at different distances apart at =2( and
Cases). ..........................................................................................................................59
Figure 6.16. Pressure coefficient over the rear wingbox closest to the moving wall
with the wingboxes at different distances apart at =2( and
Case). ...........................................................................................................................60
Figure 6.17. Pressure along the moving wall with the wingboxes closest to the wall at
different distances apart at =2( and Case). ........................................61
Figure 6.18. Pressure coefficient over the front wingbox at the middle height from the
wall with the wingboxes at different distances apart at =2( and
Cases). ...............................................................................................................62
Figure 6.19. Pressure coefficient over the rear wingbox at the middle height from the
wall with the wingboxes at different distances apart at =2( and
Cases). ...............................................................................................................63
Figure 6.20. Pressure along the moving wall with the wingboxes at the middle height
from the wall at different distances apart at =2( and Cases). ..........64
Figure 6.21. Pressure coefficient over the front wingbox farthest from the moving
wall with the wingboxes at different distances apart at =2( and
Cases)..................................................................................................................65
Figure 6.22. Pressure coefficient over the rear wingbox farthest from the moving wall
with the wingboxes at different distances apart at =2( and
Cases)..................................................................................................................66
Figure 6.23. Pressure along the moving wall with the wingboxes farthest from the
wall at different distances apart at =2( and Cases). ...........................67
Figure 7.1. Pressure coefficient the front wingbox with the wingboxes at various
heights above the moving wall at 1m apart at = 4. ...............................................69
Figure 7.2. Mach contour of the wake area for the Case with a moving wall at
= 4. ........................................................................................................................69
Figure 7.3. Mach contour of the wake area for the Case with a moving wall at
= 4. ........................................................................................................................70
xiii

Figure 7.4. Pressure coefficient over the rear wingbox with the wingboxes at various
heights above the moving wall at 1m apart at = 4. ...............................................71
Figure 7.5. Pressure along the moving wall with the wingboxes at various heights
above the wall at 1m apart at = 4. .........................................................................73
Figure 7.6. Mach number along the moving wall with the wingboxes at various
heights above the wall at 1m apart at = 4. .............................................................74
Figure 7.7. Pressure coefficient over the front wingbox with the wingboxes at various
heights above the moving wall at 1.5m apart at = 4. ............................................75
Figure 7.8. Mach contour of the wake area for the Case with a moving wall at
= 4..........................................................................................................................75
Figure 7.9. Pressure coefficient over the rear wingbox with the wingboxes at various
heights above the moving wall at 1.5m apart at = 4. ............................................76
Figure 7.10. Mach contour of the wake area for the Case with a moving wall at
= 4. ........................................................................................................................76
Figure 7.11. Mach contour of the wake area for the Case with a moving wall at
= 4. ........................................................................................................................77
Figure 7.12. Pathline contour for the Case with a moving wall at = 4. ..................79
Figure 7.13. Pathline contour for the Case with a moving wall at = 4. ..................79
Figure 7.14. Pathline contour for the Case with a moving wall at = 4. ...................79
Figure 7.15. Pathline contour for the Case with a moving wall at = 4. ...................80
Figure 7.16. Pressure coefficient over the front wingbox closest to the moving wall
with the wingboxes at different distances apart at =4( and
Cases). ..........................................................................................................................80
Figure 7.17. Pressure coefficient over the rear wingbox closest to the moving wall
with the wingboxes at different distances apart at =4( and
Cases). ..........................................................................................................................81
Figure 7.18. Pressure along the moving wall with the wingboxes closest to the wall at
different distances apart at =4( and Cases). ......................................82
Figure 7.19. Pressure coefficient over the rear wingbox at the middle height from the
moving wall with the wingboxes at different distances apart at =4(
and Cases). ........................................................................................................83
Figure 7.20. Pressure along the moving wall with the wingboxes at the middle height
from the wall at different distances apart at =4( and Cases). ..........84
Figure 7.21. Pressure coefficient over the front wingbox farthest from the moving
wall with the wingboxes at different distances apart at =4( and
Cases). ................................................................................................................85
xiv

Figure 7.22. Pressure coefficient over the rear wingbox farthest from the moving wall
with the wingboxes at different distances apart at =4( and
Cases). ..........................................................................................................................86
Figure 7.23. Pressure coefficient along the moving wall with the wingboxes farthest
from the wall at different distances apart at =4( and Cases). ............87
Figure 8.1. Lift coefficient versus the ground clearance for the front, rear, and
combined wingboxes at both = 2 and 4 where the wingboxes are at . ..............89
Figure 8.2. Drag coefficient versus the ground clearance for the front, rear, and
combined wingboxes at both = 2 and 4 where the wingboxes are at . ..............90
Figure 8.3. Lift coefficient versus the ground clearance on the wall for the =2
and 4 flows, where the wingboxes are at . ...............................................................91
Figure 8.4. Drag coefficient versus the ground clearance on the wall for the =2
and 4 flows, where the wingboxes are at . ...............................................................91
Figure 8.5. Lift coefficient versus the ground clearance for the front, rear, and
combined wingboxes at both = 2 and 4 where the wingboxes are at . ..............92
Figure 8.6. Drag coefficient versus the ground clearance for the front, rear, and
combined wingboxes at both = 2 and 4 where the wingboxes are at . ..............93
Figure 8.7. Lift coefficient versus the ground clearance on the wall for the =2
and 4 flows, where the wingboxes are at . ...............................................................93
Figure 8.8. Drag coefficient versus the ground clearance on the wall for the =2
and 4 flows, where the wingboxes are at . ...............................................................94
Figure 9.1. Pressure coefficient over the front wingbox closest to the stationary wall
with the wingboxes at different distances apart at = 2 compared to the
experimental results ( and Cases). .............................................................96
Figure 9.2. Mach contour of the Baseline Case, , with a stationary wall at =
2....................................................................................................................................97
Figure 9.3. Schlieren image of the Case from the wind tunnel experiment.
Source: Ngo, Trieu, Andres Cedeno, and Jeri Perez. Supersonic Ground
Effects. San Diego, CA: San Diego State University, 2009. .......................................97
Figure 9.4. Pressure coefficient over the rear wingbox closest to the stationary wall
with the wingboxes at different distances apart at = 2 compared to the
experimental results ( and Cases). .............................................................99
Figure 9.5. Pressure coefficient over the front wingbox at the middle height above the
stationary wall with the wingboxes at different distances apart at =2
compared to the experimental results ( and Cases). ...............................100
xv

Figure 9.6. Pressure coefficient over the rear wingbox at the middle height above the
stationary wall with the wingboxes at different distances apart at =2
compared to the experimental results ( and Cases). ...............................101
Figure 9.7. Pressure coefficient over the front wingbox farthest from the stationary
wall with the wingboxes at different distances apart at = 2 compared to
the experimental results ( and Cases). .....................................................102
Figure 9.8. Pressure coefficient over the rear wingbox farthest from the stationary
wall with the wingboxes at different distances apart at = 2 compared to
the experimental results ( and Cases). .....................................................103
Figure 9.9. Mach contour of the Case with a moving wall at = 2.........................103
Figure 9.10. Mach contour of the Case with a stationary wall at = 2. ..................104
Figure 9.11. Pathline contour of the Case with a stationary wall at = 2. ..............104
Figure 9.12. Total pressure contour of the Case with a moving wall at = 2. ........104
Figure 9.13. Total pressure contour of the Case with a stationary wall at =
2..................................................................................................................................105
Figure 9.14. Pressure contour of the Case with a moving wall at = 2. .................105
Figure A.1. Mach contour of the Case with a moving wall at = 2. .....................113
Figure A.2. Absolute pressure contour of the Case with a moving wall at =
2..................................................................................................................................113
Figure A.3. Mach contour of the Case with a moving wall at = 4. .....................113
Figure A.4. Absolute pressure contour of the Case with a moving wall at =
4..................................................................................................................................114
Figure A.5. Mach contour of the Case with a moving wall at = 2. .....................114
Figure A.6. Absolute pressure contour of the Case with a moving wall at =
2..................................................................................................................................114
Figure A.7. Mach contour of the Case with a moving wall at = 4. .....................115
Figure A.8. Absolute pressure contour of the Case with a moving wall at =
4..................................................................................................................................115
Figure A.9. Mach contour of the Case with a moving wall at = 2. .....................115
Figure A.10. Absolute pressure contour of the Case with a moving wall at
= 2. .............................................................................................................................116
Figure A.11. Mach contour of the Case with a moving wall at = 4. ...................116
Figure A.12. Absolute pressure contour of the Case with a moving wall at
= 4. .............................................................................................................................116
xvi

Figure A.13. Mach contour of the Case with a moving wall at = 2. ...................117
Figure A.14. Absolute pressure contour of the Case with a moving wall at
= 2. .............................................................................................................................117
Figure A.15. Mach contour of the Case with a moving wall at = 4. ...................117
Figure A.16. Absolute pressure contour of the Case with a moving wall at
= 4 ..............................................................................................................................118
xvii

LIST OF SYMBOLS

Density
Velocity Vector
User Defined Input In Mass Conservation Equation
Stress Tensor

g Gravity
Body Force

Molecular Viscosity

I Unit Tensor
E Energy
P Pressure In Conservation Equations
Heat Flux

Rate of Internal Heat Generation

Viscous Dissipation Function

Operating Pressure

P Local Static Pressure


Molecular Weight

Velocity (x-Direction)

T Temperature
t Time
Average Velocity

Fluctuating Velocity

k Turbulence Kinetic Energy


Specific Dissipation Rate
xviii

G Generation of Turbulence Kinetic Energy


Effective Diffusivity
Y Dissipation Due To Turbulence
User Defined In Transport Equations

User Defined In Transport Equations

S Modulus of The Mean Rate-Of-Strain Tensor


Turbulence Viscosity

Coefficient that Damps Turbulent Viscosity

Dissipation Rate At The Wall

Friction Velocity

Non-Dimensionalized Dissipation Rate For Various Layers of The Velocity Profile

Constant In k- Model

Non-Dimensionalized Velocity

Non-Dimensionalized Turbulence Factor

Center-Cell Value In Grid

Gradient of Upstream Cell In Grid

r Displacement Vector
Closest Ground Clearance To Wall

Middle Ground Clearance To Wall

Farthest Ground Clearance To Wall

Closest Distance Between Wingboxes

Farthest Distance Between Wingboxes

L Length of Wingbox
W Width of Wingbox
xix

Height of Test Domain

Length of Test Domain

Free Stream Temperature

Free Stream Velocity

Free Stream Pressure

Free Stream Pressure In Experiment

Turning Angle
Mach Angle
Free Stream Mach Number

M Mach Number
Mach Reflection Angle
V Velocity
a Speed of Sound
R Ideal Gas Constant
Ratio of Specific Heats

Specific Heat At Constant Volume

Specific Heat At Constant Pressure

Re Reynolds Number
Reference Length

Kinematic Viscosity

Boundary Layer Thickness

y Boundary Layer Thickness ( )

Drag Coefficient

Pressure Coefficient

c Chord of Wingbox
xx

A Area
Absolute Pressure

Total Pressure

Adiabatic Wall Temperature

Temperature of Moving Wall

Velocity of Moving Wall

Pr Prandtl Number
L Length of Domain of Mesh
Lift Coefficient

Inlet Area

Throat Area
xxi

ACKNOWLEDGEMENTS

I would like to thank my committee members, Dr. Jacobs, Dr. Plotkin, and Dr.
Paolini, for their guidance and support throughout this process. Also, I would like to
acknowledge General Atomics and the California Space Grant Consortium for funding this
research. I would like to thank my mom, Cheryl, my dad, David, and my brother, Ryan, for
their unconditional love and support. I would also like to thank my godfather, Allan, for
giving me the opportunity to further my education.
1

CHAPTER 1

INTRODUCTION

General Atomics and the U.S Air Force are developing a High Speed Test Track at
Holloman Air Force Base. The test track magnetically levitates a sled and uses rocket
propulsion to accelerate the sled to supersonic speeds. The objective of the MagLev test rail
is the high-speed testing of a range of systems of relevance to the Air Force that are to be
mounted on the sled [1].
The initial design concept was a single slot tandem wingbox configuration that was
positioned at 1.5 cm of ground clearance (Figure 1.1) [1]. The rockets propel the sled to high
speeds up to 620 ft/s. This high speed results in short test times of up to 5.5 seconds. The
wingboxes are magnetically levitated by a combination of forward motion and the interaction
between the passive copper rails and the magnetic wingboxes. However, the single slot
configuration showed an unacceptable instability due to a rocking effect during levitation.
This prompted General Atomics to modify the initial design and fabricate a dual slot system
with two sets of tandem wingboxes which is more stable.
The operating range of the MagLev test is between Mach .5 and Mach 5. Current
efforts focus on a three stage rocket moving through helium that will allow for testing at even
higher Mach numbers. The speed of sound, which is based on the temperature, ideal gas
constant, R, and the ratio of specific heats at constant pressure and volume, also referred to as
, will be higher in a helium environment than in air because helium is less dense than air.
Therefore, speeds of up to Mach 9 are achievable by covering the test track to make the
environment helium based.
Two of the major issues in the design of the test sled include the weight of the sled
and the magnetic and aerodynamic drag on the sled [1]. At high speeds, the flow around the
tandem wingbox configuration used on the sled is affected by shock waves, shock
reflections, wake effects, and ground clearance that not only lead to high loads, but also
affect the stability of the sled and the heating of the track and the sled. Current
understanding of high-speed, supersonic objects in ground effect is very limited and an
2

Figure 1.1. Single sled configuration for the testing at Holloman


Air Force Base. Source: Hsu, Y., D. Ketchen, L. Holland, A.
Langhorn, D. Minto, and D. Doll. Status of the Magnetic
Levitation Upgrade to Holloman High Speed Test Track. IEEE
Transactions on Applied Superconductivity No. 29 (2008): 2074-
2077. http://maglev.com.www65.yourserver.de/uploads/2008/
MAGLEV_Development_Projects_etc/holland_langhorn_minto_d
oll__status_of_the_magnetiv_levitation_upgrade_to_holloman_hig
h_speed_test_track.pdf.

improved knowledge of the associated aerodynamics is necessary for the structural


engineering of the test sled used on the test track.
The majority of ground effect investigations have focused on subsonic flow in
relation to the take-off and landing of aircraft. A complete review of subsonic ground effect
is well beyond the scope of this thesis. For an excellent review, [2] is referred to and
included in the discussion are some of the more historically important and relevant works.
Tomokita et al. [3] determined in 1933 that when an airfoil is in the presence of the ground,
the lift on the wing increases with small angles of attack. This holds true until the angle of
attack reaches a critical value, at which point there is a decrease in lift [3]. Barber et al. [2]
explained that one of the early challenges was to match experimental results with
computational results. Issues were in the mimicking of the ground in wind tunnel testing or
in CFD as well as the boundary conditions that were appropriate. Typically, wind tunnel
tests used a fixed ground plane that does not represent the true conditions of ground effect.
A moving wall with the same velocity as the free stream air should be used. Raymond [2]
3

found the mirror image technique to increase the lift to drag ratios while conducting ground
effect experiments. The mirror image and a moving ground became important to the
advancement of the research of ground effect. The mirror image is a technique that uses the
mirror image of the airfoil being tested in the wind tunnel to simulate ground effect. The
flow effects on the airfoil being tested are mimicked in the mirror image airfoil, causing the
space between them to act as a ground plane. Another common technique that is used is
potential flow to model ground effect, which results in good test data if the viscous effects
are minimal. However, this model displays discrepancies at small ground clearances due to
boundary layer effects related to a fixed wall and the fact that viscosity is being neglected.
Also referenced was Hirata [2] who used Navier-Stokes solvers on stationary walls in the
wind tunnel and found separation occurs along the ground below the leading edge of the
airfoil. Boundary conditions, user assumptions, and what exactly the term "slip condition"
meant were the reasons for experimental and computational data not matching up in ground
effect research. In CFD, the slip condition refers to zero shear stress at the boundary
resulting in discrepancy with the ground moving at the free stream velocity. Barber et al. [2]
used a Navier-Stokes solver along with a RNG k- turbulence model as well as PIV wind
tunnel experiments to test the boundary condition issues. They tested ground effect with the
slip condition taken as zero shear stress at the boundary, which also allowed for a moving
ground plane depending on the reference to the test object. They found that as the wing gets
closer to the ground, and the area between the body and the ground decreases, the lift and
turbulence kinetic energy increase. As a result of this study, a moving ground was found to
be the most accurate ground boundary condition, and viscous effects are important to the
accuracy of ground effect research [2]. Ahmed et al. [4] tested an airfoil in ground effect
using a moving belt in the wind tunnel. Laminar boundary layers developed on the top
surface of the airfoil near the leading edge, but transitioned to turbulent boundary layers
through a separation bubble resulting from an adverse pressure gradient. While the focus of
the research was on locating the point of transition, it was found that increasing Reynolds
number reduces the separation bubble. It was also found that the area between the underside
of the body and the ground could be equated to a converging-diverging channel. As the
ground clearance decreases, the stagnation point on the airfoil is pushed downstream [4].
While a complete overview of the research is beyond the scope of this paper, the point being
4

made is that subsonic ground effect is now a well established concept in the field of
aerodynamics.
Coulliette and Plotkin [5] investigated ground effect of airfoils with respect to angle
of attack, camber, and thickness ratio using discrete and linear vortex panel methods. It was
determined that smaller angles of attack and camber lead to larger normalized lift, while the
opposite is true for the effect of the thickness ratio. Thickness ratio reduces lift at all values
of ground clearance, and as the thickness ratio increases, the magnitude of lift will increase
[5]. Rozhdestvensky's book [6] discusses extreme ground effect on lifting systems, including
tandem airfoils. He states that for increased stability , the load on the front airfoil should be
greater than on the rear airfoil, and that changing the distance between the airfoils changes
the stability margin. A reduction in the lift coefficient of the tandem airfoil system will lead
to less stability as well [6].
While extreme ground effect for low subsonic flow is well studied, the literature on
supersonic ground effect is much less extensive. Prykhodko et al. [7] performed wind tunnel
and CFD tests on a MagLev vehicle and found that the wind tunnel does not accurately
represent the real conditions no matter how the ground effect is represented in the wind
tunnel, either through a stationary wall, the mirror image, boundary layer control, or a
moveable belt. He found that the nose shape determines the characteristics of the flow. That
data indicated that with a slightly upward nose, the mass flow between the ground and the
body increased which resulted in a decrease the pressure [7]. Supersonic sled tests were
conducted in 1970 on the Blue Flame, a high speed land vehicle, showing how the shock
reflected off the ground from the nose and how this built up the pressure, which increased lift
[8]. Wind tunnel tests were also conducted using a stationary ground plane at a range of
Mach numbers between .5 and 1.5. It was found from these tests that the nose shape and
fineness ratio were key elements in how drag was affected. Drag is affected by ground
clearance because of the pressure build up from the reflected shocks [8].
While not much research has been done on supersonic ground effects, there are
numerical or computational studies that are comparable, in particular, ones dealing with flow
features found in the current research of this paper including supersonic intakes, shock
reflection off of walls, shock interaction with boundary layers, Mach reflections, and shock-
5

wake interaction studies. By breaking down the physical flow field into these individual
parts, one has a reference to draw conclusions about the particular flow being investigated.
Slater [9] performed CFD analysis on supersonic inlets. His study used second order
Reynolds Averaged Navier-Stokes equations on a structured grid for the turbulent,
compressible flow simulation. Inlets decelerate supersonic flow to subsonic flow, and this is
achieved through the shock interaction with turbulent boundary layers. The key with
supersonic inlets is to minimize pressure losses. A concept that is relevant to supersonic
inlets or nozzles is the effect of a choked flow. A normal shock can result from a choked
flow and can move upstream due to the mass flow difference. Through grid convergence
verification, Slater found that the CFD codes that were tested were good representations of
the behavior of the boundary conditions [9]. Ferri and Nucci [10] investigated a new type of
supersonic inlet where the deceleration took place outside of the inlet. They investigated the
pressure recovery with the shock behavior from the supersonic inlet. When a shock occurs at
the inlet, an increase in back pressure will push the shock outside of the inlet and it will stay
there because the size of the throat is now at a size that will not allow for the shock to be
swallowed. An increase in back pressure will push the shock upstream causing the free
stream fluid that enters the inlet to decrease in area [10]. Moase et al. [11] researched choked
nozzle and supersonic diffuser behaviors. They used a fourth order dispersion-relation-
preserving scheme for temporal and spatial discretization for accurate representation of the
exact solutions to test the flow patterns. Their observations indicated that the effect of an un-
start for a ramjet is a normal shock which is propelled out of the diffuser resulting in
unwanted pressure losses. They found that the nonlinear effects of un-choke, un-start, and
over-choke are more significant at "higher Mach numbers upstream and downstream of
choked nozzles" [11].
Shock theory, especially shock reflection off of walls and interaction with boundary
layers, is of particular relevance to supersonic ground effect. Ben-Dor et al. [12] investigated
how pressure downstream will affect the shock reflection in steady flow around a wedge.
One effect is a normal reflection of the oblique shock. Another effect is a Mach reflection
that can be pushed upstream or downstream from the flow before the shock. The resulting
wake pressure will have an effect on the shock reflection. They found that Mach reflections
can interact with the expansion fan that is created at the back edge of the wedge. As a result
6

of this interaction, new expansion fans occur, there is a pressure drop, and the flow's velocity
increases to supersonic conditions. The length and pattern of the Mach wave are going to be
independent of the pressure downstream only if the wake pressure is less than or equal to the
pressure in the expansion fan. When this condition is not met, the Mach reflection is now
dependent on the wake pressure. As a result, the Mach stem is pushed upstream and
increases in length. As the wake pressure continues to increase, the expansion fan will
weaken and eventually become a shock when the wake pressure becomes larger than the
pressure behind the initial oblique shock [12]. Alber [13] researched the flow patterns of
subsonic and supersonic flows over a rearward facing step. The rearward facing step has
been studied for wake and boundary layer research. That research has led to the investigation
of the lip shock. The boundary layer separates due to an adverse pressure gradient, and, as a
result of the compression at the end of the expansion fan, a lip shock forms which separates
the expanding boundary layer and the new viscous sublayer. This becomes a shear layer that
interacts with the recirculation in the wake. The lip shock is a separation shock which results
from disturbances developed in the subsonic portion of the boundary layer. Hama,
referenced in Alber's [13] research, showed that the lip shock can intersect the wake shock
resulting in an expansion fan. Alber also found that turbulent and laminar wakes look the
same. However, the mixing rates are different based on the length scales, therefore any wake
model can be used if the appropriate length scale is applied [13]. Seddon [14] examined the
interaction between turbulent boundary layer and a normal shock using a wind tunnel and no
CFD for his research. Shock/boundary layer interaction is problematic in supersonic intakes
because it can result in an undesirable pressure loss. When the two interact, separation
occurs which leads to an increase in static pressure at the separation point along with a
recirculation bubble which closes at the boundary layer's reattachment point. He noted that
adding constraints to the flow will affect shock interaction, the separation bubble, and
reattachment. Duct area for intakes or spillage requirements would be considered
constraints. Pushing a shock out of the duct will show spillage effects and the separation will
be magnified [14].
Zang et al. [15] looked at numerical solutions to shock wave interactions in turbulent
flows. They used two dimensional Euler equations with a second order, explicit, finite
difference scheme, because at that point, in 1984, two dimensional Navier-Stokes equations
7

were not feasible. At that time, computational methods used a moving shock simulation,
while linear theory used a stationary shock simulation. They found that linear theory
legitimacy is broad for a range of subsonic and supersonic Mach numbers, and there is a
discrepancy between linear and numerical solutions in the area where the "waves with
incidence angles are near or beyond the critical angle" [15].
When using the Reynolds Averaged Navier-Stokes Equations, compressible flows
add to the complexity of the computational modeling. Wilcox [16] discusses turbulence
modeling for compressible flows. He notes that finding a turbulence model is difficult,
because there is a limited selection of models for a wide range of simulations. Such topics as
turbulent heat flux, molecular diffusion, turbulent transport, and dilatation dissipation need to
be considered when modeling turbulence. These topics are interconnected and exhibit a
balance among themselves. For example, addressing dilatation dissipation improves the
mixing layer results as the Mach number increases but adversely effects flows with shock
separation. The reflection of an oblique shock off of a compression corner is one of the most
complex flows to simulate because the Mach number and pressure ratio over the shock have
to be addressed [16]. He also notes that all turbulence models have an issue with heat
transfer at the boundary layer's reattachment point. The surface heat at reattachment is
problematic in that the CFD simulations predict higher values than what is measured.
Wilcox [16] presents the k- model as a good model to use because, unlike the k- model, it
is accurate for the law of the wall and is good at representing shock separated flows for a
range of Mach numbers when a stress-limiter is used.
Research has also been conducted on flow patterns around objects in tandem. A
common theme within this topic is that the rear object affects the forces on the front object.
Yegorov et al. [17] investigated hypersonic viscous flows around cylinders placed in tandem.
They determined that the drag found from the rear cylinder is greatly affected by the distance
from the front cylinder. Reynolds number was also significant for how the parameters
responded. When the Reynolds number increases, the recirculation behind the front cylinder
develops, which affects the rear cylinder's aerodynamics. At small distances, the wake from
the front object dissolves due to the upstream effect of the rear object [17]. Karpov et al. [18]
looked at the separated flow between two bodies. They found that the two bodies will act
independently only if there is a sufficient distance between them, in their case, three to four
8

diameters in length. The opposite was found at smaller distances. The flow separates from
the front body and attaches to the leading edge of the rear body [18]. Hilliard and Springer's
[19] book references tandem bluff body aerodynamics. They say that the flow cannot be
predicted for the tandem configuration based on the individual flow patterns of both bodies
because of the front body's wake interaction with the rear body. The rear body's flow also
affects the upstream flow. Total drag depends on how the flow separates and if it reattaches.
Therefore, distance between tandem bodies plays a key role in drag optimization. Much
research in this area has been done on tractor-trailer systems at low Reynolds numbers [19].
This thesis presents a first examination of aerodynamic ground effect with relevance
to a supersonic magnetically levitated sled configuration used for supersonic testing on the
ground at Holloman Air Force Base. This project has two purposes. The first is an analysis
of supersonic ground effect on a tandem wingbox configuration through the use of
computational fluid dynamics. The second is a comparison of the results to the data received
from the supersonic wind tunnel experiment of the same setup. Six different configurations
were tested. The conditions under study in each configuration were (1) the location from the
ground to test the effect of ground clearance and (2) the distance between the tandem
wingboxes to test wake effects. The tandem wingbox configurations mimic the tests
conducted in the supersonic wind tunnel for comparison [20]. Each configuration was run
with two different free stream Mach numbers to test the effect of overspeeding. All of these
configurations were tested using ANSYS, Inc. Fluent Software for the CFD analysis. The k-
turbulence model was used along with Reynolds Averaged Navier-Stokes Equations for
the CFD analysis.
Chapter Two discusses the computational methodology that was used in Fluent
including the governing equations, turbulence model, and numerical methods. Chapter Three
discusses the computational model and mesh configuration. Chapter Four presents the
convergence studies of the Baseline Case. Chapter Five examines the general flow
characteristics of both the = 2 and = 4 runs, as well as discusses the effect of the Mach

number on the flow. In Chapter Six, the effects of ground clearance and distance between
the tandem wingboxes for the cases with = 2 are analyzed through the use of pressure

comparison plots and pressure and Mach contours. Chapter Seven presents the effect of
9

ground clearance and distance between the tandem wingboxes for the cases with = 4.

The discussion moves to an investigation of the loads experienced by the tandem wingboxes
and the wall at each free stream Mach number in Chapter Eight. This leads, in Chapter Nine,
to the discussion of the comparison of the experimental data to the CFD analysis through the
use of pressure coefficient comparison plots. The difference between a moving wall and a
stationary wall will also be presented in this chapter. The discussion ends with conclusions
on supersonic ground effect in Chapter Ten.
10

CHAPTER 2

COMPUTATIONAL METHODOLOGY

The commercially available CFD software, Fluent, was used to compute the
supersonic flow over the tandem wingbox in ground effect. This chapter begins with a
summary of the governing Reynolds Average Navier-Stokes Equations that Fluent uses as its
model for supersonic flow. Then some of the aspects of the numerical methodology that
Fluent implements to approximate the governing equations are discussed.

2.1 GOVERNING EQUATIONS


The compressible, turbulent flows over the MagLev sled can be determined with the
density-based transient model in Fluent. The model constitutes the flow through the mass,
momentum, and energy conservation equations. The continuity equation is given by
. (2.1)

and the momentum equation as


. (2.2)

where
. (2.3)

is the stress tensor, are body forces, is the molecular viscosity, and I is a unit

tensor. The energy equation is given by


. (2.4)

An equation of state complements the conservation law and is given as


. (2.5)

where is the operating pressure, P is the local static pressure, and is the molecular

weight of the fluid.


11

2.2 TURBULENCE MODELS


The very high Reynolds number flows over the sled are turbulent. Fluctuations in the
velocity of turbulent flow results in fluctuations in the momentum and energy being
transported [21]. To compute the turbulent flows, turbulence models that are based upon the
Reynolds Averaged Navier-Stokes equations are considered. In a Reynolds averaging, the
flow variables are comprised of an average and a fluctuating component:

where the bar indicates the average and the prime indicates the fluctuation. This expression
is substituted into the mass, momentum, and energy equations. For example, the process for
the conservation of mass equation is shown.
. (2.6)

. (2.7)

If the velocity terms are substituted with their new identities, the equation becomes
. (2.8)

. (2.9)

When the fluctuating term is averaged, that term is zero. Therefore these terms will drop
leaving
. (2.10)

This procedure is repeated for the momentum (only the x-direction is shown) and energy
equations and results in the following equations
. (2.11)

. (2.12)

There are now officially four equations for many more unknowns. This is referred to as the
"Closure Problem." A range of turbulence models has been developed to close the Reynolds
stress terms. While no turbulence model has been found to be the best, in the case of the
tandem wingbox flow characteristics, the Standard k- model is used to close the Reynolds
12

stresses encountered in the "Closure Problem." The Standard k- scheme is based on the
Wilcox model and takes into account "low Reynolds number effects, compressibility, and
shear flow spreading" [21]. The Standard k- model was chosen because it is good at
representing the log layer for a wide range of Mach numbers, accurate for wall-bounded and
free shear flows, and good at predicting "shock-separated" flows [16]. This model is based
on the turbulence kinetic energy, k, and the specific dissipation rate, , of the transport
equations presented below:

. (2.13)

. (2.14)

where G is the generation of turbulence kinetic energy and specific dissipation, is the
effective diffusivity, Y is the dissipation due to turbulence, and S is a user defined parameter.
The turbulence generation model used to find G includes in the time averaged terms for the
closure problem discussed earlier.
. (2.15)

where the Boussinesq Hypothesis is used to get


. (2.16)

. (2.16.a)

S in this case is the modulus of the mean rate-of-strain tensor. The turbulence generation and
effective diffusivity are based on the term:
. (2.17)

where will equal 1 for high Reynolds number flows. The k- model is utilized because it

includes the appropriate equations to fix the closure problem and allows the unknowns in the
turbulent flow to be solved for.
A model related to the Standard k- model, the SST k- model, was considered
during the convergence study portion of the research. The Shear-Stress Transport model uses
the accuracy of the near wall modeling in the Standard model along with the accuracy of the
13

far-field modeling in the k- model [21]. This mixture of models makes the SST a good
choice for complex flows. Table 2.1 shows the advantages and disadvantages of these
closely related models [21] [22]. The Standard k- model is ultimately used based on results
obtained in the grid convergence study.

Table 2.1. Standard k- vs. SST k- Models: Advantages and Disadvantages


Advantages Disadvantages
Standard k- accurate near wall modeling lower accuracy when
accurate in predicting free predicting separation
shear flows mesh refinement at walls
good for predicting transition
no damping function for better
numerical stability
SST k- accurate near wall and far-field compressibility option not
modeling given
turbulent shear stress counted mesh refinement at walls
accurate for wide range of dependent on wall
flows distance so not appropriate
better predicts separation for free shear flows

Wall conditions are important factors with this topic of research because ground
effects are being analyzed. At the wall, there are shock and Mach reflections and
shock/boundary layer interactions that are important to the ground effect analysis.
Turbulence greatly affects near-wall representation. Closer to the wall, viscous damping
affects the flow while farther from the wall, turbulence kinetic energy becomes the factor that
influences the flow [21]. The wall conditions are modeled by the following in Fluent:
. (2.18)

where
. (2.19)
14

Because this is a turbulent flow, the logarithmic region (Figure 3.5, p 23) will be the area of
interest and is represented by this equation:

. (2.20)

This turbulence factor, , is an important factor in these calculations. The value of

corresponds to a certain region on the velocity profile. The goal is to have this term in the

logarithmic layer ranging from 30 to around 100. This term is used in the grid convergence
process that is discussed in Chapter 4. As the mesh is refined, the value of decreases.

The value of this term is important because a value that is too large or too small is
undesirable. Convergence and accuracy issues may emerge under two conditions: (1) the
turbulence model not being able to account for low-Reynolds number effects in the sublayer
of the boundary layer if too small, or (2) the wake term becomes larger than the logarithmic
layer if too large [23]. Therefore, the mesh should be coarsened or refined accordingly in
order for the flow characteristics near the ground to be accurately simulated.

2.3 NUMERICAL METHODOLOGY


Fluent uses as a finite volume method to solve the conservation equations. Each grid
cell can be considered a small control volume. Each equation must be discretized and then
linearized in order to get a system of equations with a "sparse coefficient matrix" [21].
Discretization occurs both spatially and temporally. There are many options in terms of
solution methods for spatial discretization. For this research, based on how the different
cases were moving toward convergence, a combination of first-order upwind and third-order
MUSCL was used in the majority of the cases. MUSCL is used because it works well for all
mesh types including hybrid meshes like the ones that were made for this research. The
hybrid meshes that were made contained both triangular and quadrilateral grid cells. This is
generally a good model to use with complex flows. However, shock waves in the flow, or
other discontinuities, can create an issue when using third-order MUSCL due to the fact that
there are no "flux-limiters" [21]. Higher order schemes are more desirable for accuracy
purposes. First-order upwind means that the value at the center of a grid cell is equal to the
edge or face of the grid cell directly upstream. Second-order upwind is more accurate in that
15

it uses the Taylor series expansion at the center of the grid cell for better accuracy at the face
of the grid cell upstream.
. (2.21)

where is the center-cell value, is the gradient of the upstream cell, and r is the
displacement vector. Third-order MUSCL scheme uses a combination of central difference
scheme along with the second-order upwind scheme.
. (2.22)

The Quick scheme was used for the turbulence kinetic energy and dissipation rate, k
and , discretization. This is a higher order scheme for quadrilateral meshes and works best
if the grid cells are in the same direction as the flow. The mesh configuration that was done
for the tandem wingbox system will be explained later in the chapter, but theoretical analysis
was done on the wingbox in order to analyze flow behavior. This initial analysis was done
for this very reason, that Quick works best when the grid cells align with the direction of the
flow. It is also acceptable to use this method on a hybrid mesh such as in this case [21].
Temporal discretization can be done either explicitly or implicitly. In this case, the
first-order implicit scheme was used because it is "unconditionally stable" [21]. This means
that the size of the time step could be any range of magnitude. However, the drawback to
using implicit methods is that now a set of nonlinear algebraic equations has to be solved.

. (2.23)

If the explicit scheme were used, this would not be the case. It would have to meet a stability
requirement and as such, the time step would have to be the same for each step. It can
ultimately end up being a time step of which would make the calculation time

extremely long and expensive.


16

CHAPTER 3

COMPUTATIONAL MODEL

An experimental study of the tandem wingbox performed by students for their senior
design project [20] in the supersonic blow-down wind tunnel at San Diego State University
served as the reference for the computational study. First a brief recap of the experimental
setup is discussed and then the dimensions and mesh of the computational reference case that
is based on the experimental setup is presented.

3.1 EXPERIMENTAL MODEL CONFIGURATIONS


The tandem wingbox model that was used in the CFD analysis stemmed from the
experiment that was done in the supersonic wind tunnel. A three-dimensional CAD model
was generated in Solidworks. The CAD model formed the two-dimensional meshing process
that will be discussed in the next section. Figure 3.1 shows the tandem wingbox
configuration that was mounted in the wind tunnel experiment [20].

Figure 3.1. Tandem wingbox configuration used for the experiment in the supersonic
wind tunnel. Source: Ngo, Trieu, Andres Cedeno, and Jeri Perez. Supersonic Ground
Effects. (San Diego, California: San Diego State University, 2009).
17

The wind tunnel experiment developed six different configurations of the tandem
wingboxes with regards to ground clearance and distance between the wingboxes. The
tandem wingboxes were tested for three different ground clearances at two different distances
apart. Figure 3.1 shows the basic configuration, and Table 3.1 lists the six different
configurations in reference to the figure. H refers to the ground clearance distance and S
refers to the distance between the front and rear wingbox. For future discussion of the
results, each configuration will be stated by its case name. To make the experiment in the
wind tunnel feasible, two rods had to be installed between the tandem wingbox, which was
connected to the top wall by a hinge seen in Figure 3.1.

Table 3.1. List of the Cases Used in Experiment and CFD Analysis
Case Name Height Above Wall Based on Distance Apart Based on
Variable H (m, in) Variable S (m, in)

(Baseline)

3.2 COMPUTATIONAL MODEL


The configuration of the tandem wingboxes at the middle height from the wall and
the smaller distance apart is taken as the Baseline Case. Figure 3.2 shows the Baseline
Case's, , dimensions taken in the CFD analysis, while Figure 3.3 summarizes the

implemented boundary conditions. The Baseline Case is used for initial theoretical testing
including boundary layer and shock reflection analysis, as well as for the grid convergence
process, which is discussed in the next chapter.
18

Figure 3.2. Dimensional setup of the Baseline Case, , configuration used in


Fluent for CFD analysis.

Figure 3.3. Baseline Case, , of the tandem wingbox configuration used in Fluent
for the CFD analysis.

Figure 3.2 shows that the height of the domain around the wingboxes is taken as the
same length of the height of the supersonic wind tunnel test section, as = 6 inches. The

length of the domain was chosen in order for the flow to have sufficient room to develop
downstream. The two rods, as well as the arm seen in Figure 3.1 for the wind tunnel
configuration, are not taken into account in the two-dimensional CFD testing. These are not
taken into account in the computational testing because in the two dimensional modeling, the
19

rods would result in a barrier that separates the flow over and under the wingboxes, whereas
in a three dimensional simulation, the flow would interact in the wake area. Flow
characteristics for the CFD analysis match the experimental results. However, the rods
holding the wingboxes together will later prove to be a significant factor in how the wake
between the front and rear wingbox behaves, causing discrepancy in the experimental data
compared to the CFD results.
The boundary conditions taken in the computational analysis are referenced in Figure
3.3 of the Baseline Case, . The wingboxes are set up with the no-slip condition which

means that directly at the wall of the wingboxes, the velocity is zero. The moving wall,
acting as the ground, is given the same velocity as the free stream. Also, the wall is
configured with zero heat flux through the wall, making the wall adiabatic. The inflow,
outflow, and top boundary edges seen in Figure 3.2 in pink, are set up as pressure far-fields.
The top boundary is not a wall as in the case of the supersonic wind tunnel, therefore no
shocks will reflect off of the top boundary allowing for a more realistic setup.
Similar to the high speed test track setup, the wall that is acting as the ground is
moving, therefore all configurations are tested with a moving bottom wall. The effects of a
stationary wall had to be considered since no attempt to mimic a moving wall during the
wind tunnel experiment was made. Therefore, each case is tested in Fluent with a stationary
wall configuration as well, in order to provide a better comparison with the experimental
data. In Figure 3.3, air flows from left to right, with the top of the wingbox having the sharp
edges, and the bottom of the wingbox having the rounded edges.
Each case is set up in Fluent in the same way, with the same reference parameters.
For the first round of testing, Mach 2 is used as the free steam flow velocity, and with the air
properties taken at a temperature of = 300K, the reference velocity is calculated. The

reference velocity used is = 694.37 m/s. The reference area and length are the same since

this is a two-dimensional model and taken as the chord of the wingbox which is 3 (m, in).
The pressure upstream is = 1 atm or 14.5 psi. For the experiment, the upstream pressure

was = 5.71 psi. The different upstream pressure for the experimental data is used to

non-dimensionalize the pressure, therefore allowing the pressure coefficients to be compared


20

appropriately. Next, Mach 4 is used as the free stream flow velocity at the same pressure and
temperature, leading to a reference velocity of = 1375.05 m/s.

3.3 MESH CONFIGURATION


To determine the mesh topology and refinement, some basic flow features and their
scales are determined using theory. Based on these flow characteristics, one can project how
to define mesh blocks and where to refine them.
Shock-expansion theory enables a preliminary determination of the angles of the
oblique and reflected shocks and the Mach number distribution around the wingbox [24].
With knowledge of the shock angles, an estimate of the location where grid refinement is
required is obtained. Moreover, the mesh can be aligned with the shock to improve shock
capturing and overall accuracy of the simulation. Shock-expansion theory is utilized during
the CFD analysis to verify Fluent's results.
Shock-expansion theory states that a detached shock results when an object with a
blunt nose encounters a supersonic flow. The --Mach diagram [24] is used to determine
Mach angle. In this case, = 2, and , or the turning angle of the wingbox, is around 21.5.

Because the wingbox is not symmetrical (Figure 3.4), the value of is found using known
distances and geometry. Since two parameters are known, the Mach angle, , is retrieved
from the diagram and is 57. This value is important in the meshing process since it is the
angle which the grid cells near the wingboxes will follow, because this is the direction to
which the flow will be turned due to the oblique shock. Shock-expansion theory is applied to
determine the Mach number behind the oblique shock. The Mach number normal to the flow
is calculated from the following equation, and is used in the normal shock tables of Anderson
[24] to get the subsonic Mach number as if this were a normal shock instead of an oblique
shock.
. (3.1)
After is found from the shock tables, the Mach number behind the oblique shock can be
found from:

. (3.2)
21

Figure 3.4. Picture of initial shock theory done on the


Baseline Case for mesh generation references.

Based on the initial calculations, the Mach number behind the oblique is around Mach 1.1.
Oblique shocks impede the flow less than normal shocks. This result is more desirable since
there is less total pressure losses. This process is completed for both the top and bottom of
the wingbox. However, when observing the bottom of the wingbox, the initial detached
oblique shock is reflected off the ground. Reflected shock analysis is conducted to determine
the angle at which it will most likely reflect. The reflection angle, , is calculated to be
around 36.5. The number of reflections depends on the Mach number after each reflection,
and once the flow is no longer supersonic, the shock will cease. Shock reflection only occurs
with the = 4 cases. The pressures that develop in the flow are a factor in the explanation

of why the = 2 cases do not show a shock reflection.

The boundary layer thickness is estimated using the basic parameters of air, the free
stream mach number, parameters that were specified in the experiment, and the Reynolds
number. High Mach number flows, especially in extreme ground effect, are affected by
turbulence. Compressible flow describes a high velocity, high Reynolds number flow where
the density changes throughout the flow based on the pressure variations. The Mach number,
which is based on the free stream velocity and the speed of sound, is used along with the
Reynolds number to determine if the flow is compressible as well as turbulent.
(3.3)

(3.4)
22

In this case the value of is 1.4 for air. The Mach number is known for these cases,

therefore the free stream velocity is found based on reference values. With the velocity, the
Reynolds number is solved for using the following equation:
(3.5)

The cases being tested are clearly compressible and turbulent based on the high of 2 and

4 and the Reynolds number on the order of . Knowing that turbulent boundary layers are

expected, the following equation from Anderson [18] is used to find an estimated value of the
boundary layer around the wingboxes:
. (3.6)

The reference length, , is taken as the length of the top front panel of the wingbox, from the

nose to the sharp edge. This value is around 1 in or .0254 m. The Reynolds number that was
calculated is . The boundary layer value is calculated to be m. This

is used in the meshing process in order for the boundary layer to be observed in the contours
rendered in Fluent.
Initial values of are calculated using the estimated boundary layer value to

determine the range of values in the CFD analysis and what changes will eventually need to
be made in order to get it in the desired range. The following equation from Anderson [24]
was used,
. (3.7)

where y is the boundary layer thickness, is based on the shear stress, and is the viscosity

of air. An important part of this analysis is to see where the turbulence factor range lands

in these initial runs of CFD. By the end of the grid convergence process, the goal is to have
this factor in the range of 30 to 100. This value corresponds to the log layer of a velocity
profile (Figure 3.5) [21]. A finer mesh will result in a smaller value, as noted in the
23

Figure 3.5. Breakdown of the near wall regions in terms of the turbulence factor .
Source: ANSYS Fluent: Theory Guide. Canonsburg, PA: ANSYS, Inc., 2009. CD-
ROM. Version 12.0.

discussion of the grid convergence process in the next chapter. The finest mesh taken results
in a value of in the desired range and averages around 50.
The Baseline Case, , is meshed using Gambit starting with coarser meshes and
finishing with a very fine mesh. Figure 3.6 [21] illustrates how shock-expansion theory is
used for the strategic placement of the grid cells for increased accuracy. The grid cells
follow the same angles of the detached shock at the nose. The grid cells near the wingbox
follow the flow and start out very fine at the wall of the wingbox in order to take into account
the small boundary layer. Moving away from the wall, the grid coarsens. In the flow, a
triangular mesh provides sufficient accuracy.
24

Figure 3.6. Zoomed-in picture of the grid fabrication of the Baseline Case. Source:
ANSYS Fluent: Theory Guide. Canonsburg, PA: ANSYS, Inc., 2009. CD-ROM.
Version 12.0.
25

CHAPTER 4

CONVERGENCE STUDIES

In this chapter, a convergence study of the Baseline Case, , is presented. This

sets the stage for the parametric studies in the remainder of the thesis.
The Baseline Case, , is used for the grid convergence process. The aim of grid

convergence is to ensure that the error associated with the grid-based approximation of the
conservation is reduced to the point where the results are no longer affected by the grid.
Several grids are made for the Baseline Case, from coarse grids to finer grids. The number
of grid cells double as the coarser grids are refined as well as adapted at the near wall region.
Figure 4.1 shows the pressure coefficient results over the front wingbox of the last
three grids generated. Here the grids are labeled fine, finer, and finest with respect to order of
scheme used. The results match up well in general. One area of discrepancy is located at a
point just before the flow expands over the joint causing a decrease in pressure. The other
lies at the end of the front wingbox where an induced oblique shock occurs due to boundary
layer separation. The finest grid is able to fix the resolution problems encountered by the
previous two grids along the top of the front wingbox. The drag coefficient on the front
wingbox for each mesh is measured. The for the fine, finer, and finest meshes are, .214,

.220, and .216 respectively. These values are within 1 to 3 percent of each other. The lift
coefficient for each mesh is within 1 percent of each other. This indicates that the finest
mesh has reached an acceptable level of convergence.
The rear wingbox pressure coefficient data is examined in the same manner as the
front wingbox. Figure 4.2 shows the pressure coefficient data for the same grids used in the
front wingbox evaluation. Here one can see a key area of discrepancy. There is a dramatic
increase in pressure on the lower surface of the wingbox where the wake from the front
wingbox impinges on the rear wingbox just after the normal shock that develops off the
wake. This will be seen later from the pressure and Mach contours, but of greater
significance is the fluctuation in this area. The data in this graph indicates that as the grid
refines, and the order of convergence increases, the steepness of the slope of the peak in
26

Figure 4.1. Grid convergence study for the Baseline Case, , of the
pressure coefficient over the front wingbox at =2

Figure 4.2. Grid convergence study for the Baseline Case, , of the pressure
coefficient over the rear wingbox at = 2.
27

pressure increases. However, the position of the peak of the pressure along the x-direction of
the wingbox is not clearly moving in one direction or the other. The finer mesh moves to the
right of the fine mesh, while the finest mesh moves to the left of the fine mesh. The other
area of slight fluctuation is along the top of the rear wingbox. As with the previous study, the
same type of unclear direction of movement is occurring. The finer mesh is moving down
from the fine mesh, while the finest mesh is moving up from the fine mesh. The rest of the
graph aligns well, and the finest mesh fixes the resolution problems of the previous meshes.
If there were a clear indication of direction, additional refinements could be undertaken.
However, for the extent of this research, it can be assumed that this type of fluctuation will
continue. The lift and drag are calculated for the rear wingbox as well. The for the fine,

finer, and finest mesh are .197, .195, and .202 respectively. These values are within 2.5 and
3.5 percent of each other. The difference in lift coefficient for each mesh is slightly higher,
but within 3 to 6 percent. Therefore, the data from the finest mesh, run with an initialization
of a first-order convergence and finalized with a third-order convergence, is used.
The final parameter to be considered is the turbulence factor, . As previously

stated, a range of 30 to 100 is desired. This means the first grid point from the wall will be in
the log layer of the velocity profile seen in Figure 3.6. Grid adaptation is used near the wall,
which splits up the grid cells making the grid less structured (Figure 4.3). Each mesh
rendered is adapted the same number of times along the near wall of the wingboxes and wall
boundary acting as the ground. Figure 4.4 shows that the fine and finer meshes produce the
same value of with a slight discrepancy along the front of the wingbox. The decrease in

is due to a decrease in the parameter, based on the shear stress. A decrease in shear

stress could indicate a separation point if it is very small or negative. The finest mesh shows
about a 45 percent decrease in value and gives the desired range.
Significantly for some cases, the CFD calculations are blowing up because of the
standing bow shock when it impinges with the wall. The shock has a very high localized
mach number as seen in Figure 4.5. If the wall is stationary, as in the case of the experiment,
28

Figure 4.3. Adaptation of the grid in order to reach a in the desired range of
the log-layer.

Figure 4.4. Grid convergence study for the Baseline Case, , of the turbulence
factor, , for the front wingbox at = 2.
29

Figure 4.5. Example of a Mach contour for the Case when the simulation
blows up at the location where the shock wave impinges on the wall.

a boundary layer develops along the bottom wall. This boundary layer is turbulent, therefore
shock-boundary layer interaction is not be as severe as it would be if this were a laminar
boundary layer. As the boundary layer develops, an adverse pressure gradient causes the
boundary layer to separate. Typically in viscous flow analysis of a boundary layer, as the
boundary layer develops, the velocity profile gets steeper near the wall and the shear stress
becomes very small or even negative. The effect is an adverse pressure gradient and the
boundary layer separates. This separation bubble can also be observed in the Schlieren
images produced from the experiment (Figure 4.6) [20]. It will later be shown in Chapter 9,
when comparing the experimental results to the CFD analysis, that this type of separation
bubble also occurs in the CFD Mach contour downstream in the wake area. The separation
effects the flow by turning it into itself causing another shock to form. This induced shock
causes the flow to turn back towards the wall, reattaching the boundary layer in the process.
This may explain why the Mach number is blowing up at the point where the boundary layer
and the bow shock meet. This might also be a flaw in the software itself and how the
numerical solutions are calculated.
Once the data for several grids is checked, and the turbulence factor, , is in the
correct range, the grid convergence process is complete. Meshes are then fabricated for the
rest of the cases in accordance with the Baseline Case's, , finest mesh, and the Fluent
runs commence.
30

Figure 4.6. Schlieren image of the Case from the wind tunnel
experiment. Source: Ngo, Trieu, Andres Cedeno, and Jeri Perez.
Supersonic Ground Effects. (San Diego, California: San Diego State
University, 2009).
31

CHAPTER 5

FLOW CHARACTERISTICS OF TANDEM


WINGBOX IN GROUND EFFECT

This chapter will present the flow characteristics over the tandem wingbox at =2

and 4. Contours of the cases investigated that are not presented in this chapter are located in
Appendix.

5.1 TANDEM WINGBOX CONFIGURATION AT =2

The discussion of the flow characteristics is broken down into specific areas in
reference to the tandem wingbox configuration at = 2.

5.1.1 Bow Shock and Intake Theory for Area Under


the Front Wingbox
At = 2, a detached oblique bow shock forms off the nose of the front wingbox as

seen in Figure 5.1. Near the wall, this bow shock resembles a normal shock and can be
examined using normal shock theory. Oblique shocks produce fewer losses in total pressure.

Figure 5.1. Mach number contour for the Baseline Case, , with a moving wall
at = 2.

Examining the various heights from the wall at = 2, the detached shock and the

development into a bow shock is a common trend. This bow shock develops as a result of
32

the Kantrowitz condition, a concept in intake theory. Farther distance from the wall is one
option in order to keep the detached shock an oblique shock instead of a standing bow shock.
However, the purpose of this research is to look at the ground effect of a supersonic flow,
therefore, taking the ground far enough away is not an option. Another way to eliminate the
bow shock is to increase the free stream velocity. As the velocity increases, the oblique
shock that forms on the nose of the front wingbox has a smaller shock angle. The --Mach
diagrams show that as the Mach number increases, and knowing that the angle of the
wingbox, , is not going to change, the shock angle will decrease.
For each case, with either a moving or stationary wall, the stagnation point on the
front wingbox is located slightly underneath the nose instead of the tip of the nose. The
stream function contour, Figure 5.2, shows how the streamlines of the flow are interacting
with the wingboxes.

Figure 5.2. Mach contour with the streamlines plotted of the Baseline Case, , with
a moving wall at = 2.

The area near the wall can be compared to an intake. Theory developed for intakes is
applied to understand the flow [25]. The pressure and temperature increase discontinuously
over the shock. The flow is subsonic after the shock. Figure 5.2 shows that spillage occurs.
There is an imbalance between the mass flow that can be accommodated by the intake and
the mass flow that is supplied by the free stream. Some of the streamlines flow around the
intake and accelerate the flow in the process [25]. There will be a buildup of mass and
pressure until it reaches a limit where a shock would have to be formed upstream, such as the
33

bow shock in this flow, which would allow for the spillage. The streamlines come in at the
location they do because of the spillage.
The velocity of the flow from right after the bow shock to the end of the front
wingbox in Figure 5.3, is analyzed using converging-diverging nozzle theory or velocity-area
relations derived below [24].
. (5.1)

. (5.2)

. (5.3)

From the continuity equation, the mass flow-in is equal to the mass flow-out. The mass flow
is based on the density, velocity, and area. This relation is derived assuming adiabatic flow
with no shocks, and that the density changes with pressure isentropically. The density is a
function of the Mach number, therefore the velocity is directly proportional to the area. The
streamlines, positioned like they are due to spillage, set up a diverging nozzle just before the
inlet to the theoretical intake. Because the flow is subsonic after the shock, based on
diverging nozzle theory or the velocity-area relations, the flow continues to decelerate before
it reaches the "intake" because the velocity going in is directly proportional to the area of the
theoretical intake's inlet. This is consistent with the spillage concept in that the flow is
decelerating before it reaches the "intake" in order to reach the appropriate velocity through
the "intake". After the flow decelerates, the flow accelerates due to the converging area of
the theoretical intake. Subsonic flow traveling through a converging nozzle accelerates,
therefore, the flow accelerates and the pressure decreases near the wall. This increase in
velocity will continue until another normal shock forms towards the rear wingbox. Figure
5.2 shows that the streamlines expand in the wake area. Closer examination of the wake
area, as seen in Figure 5.3, reveals that the flow becomes supersonic again at about the
middle of the front wingbox.
When the expansion occurs at the end of the front wingbox, the flow accelerates even
more due to the increase in area (Figure 5.2). Supersonic flow accelerates through a
diverging channel consistent with area-velocity relations. This acceleration leads to the
normal shockwave in front of the rear wingbox.
34

Figure 5.3. Mach contour of the wake area of the Baseline Case, , with a moving
wall at = 2.

5.1.2 Flow Over the Top of Front Wingbox


Along the top surface of the front wingbox, there is an acceleration of the flow over
the nose creating an expansion fan as seen in Figure 5.3. Expansion fans occur when the
supersonic flow is turned away from itself. The expansion wave yields a continuous
decrease in pressure and a continuous increase in Mach number. The expansion is occurring
through the acceleration of the flow across the nose in the midst of the conditions caused by
the detached bow shock. The slight increase in velocity results in a decrease in pressure on
the wingbox which can be seen in the pressure contour of the flow (Figure 5.4). After the
expansion over the nose, a slight increase in pressure is observed because of the oblique
shock after the expansion.This then leads to another expansion at the joint just aft of the nose
of the front wingbox. Because the flow over the expansion is isentropic, the total pressure
remains constant (Figure 5.3). The effect of the detached bow shock is a pressure increase
over the shock and a decrease in Mach number, except for the localized expansion at the nose
and joint.
The flow velocity is increasing after the expansion fan at the joint thus reducing the
pressure and culminating in an oblique shock off the boundary layer. Boundary layer-shock
interaction is a factor here as well. The boundary layer is separating near the end of the
wingbox resulting an induced separation oblique shock off the boundary layer which can be
seen from both the Mach and pressure contours, Figures 5.3 and 5.4.
35

Figure 5.4. Pressure contour of the Baseline Case, , with a moving wall at =
2.

5.1.3 Wake Flow Between the Front and Rear


Wingbox
The wake flow behind the front wingbox interacts with the rear wingbox. On the
bottom-rear corner of the front wingbox, the flow is accelerating due to the expansion of the
flow around that corner. The flow steadily increases through the converging-diverging area,
reaching supersonic speeds. The increase in area at the corner, along with the supersonic
velocity, allows the flow to accelerate further due to velocity-area relations. Off of these
sharp edges, shear layers are forming and interacting with the rear wingbox which can be
seen in Figure 5.3. The shear layers are being pushed upward due to a pressure difference
along the top and bottom of the wingbox from the flow expanding around the bottom-rear
corner.
In the wake, the velocity is relatively low, averaging around Mach .6. Investigating
the vertical flat edge of the front wingbox, separation occurs at the corners of the wingbox
where there are areas of recirculation with low velocity (Figure 5.3). Behind the flat edge of
the front wingbox, there is a turbulent area, where the pressure drag increases due to the
separation. Typically eddies are generated creating a more turbulent flow. The wake area is
producing a large area of circulation (Figure 5.3). The circulation is driven by the flow
outside of the wake. A closer look at the streamlines in the wake area gives a better
understanding of the various circulations.
Figure 5.3 shows that around the nose of the rear wingbox, an increase in velocity
occurs that cuts through this wake recirculation. This acceleration is due to the area in
36

between these areas of recirculation; the large area of recirculation off the back end of the
front wingbox and the smaller separation bubble on the top of the nose of the rear wingbox.
Because they are circulating in the same direction, the flow traveling through this channel
from under the front wingbox near the wall is being accelerated over the top of the rear
wingbox seen in Figure 5.5 of the pathline contour.

Figure 5.5. Pathlines of the wake area for the Baseline Case, , with a moving wall
at = 2.

The shear layer off the bottom corner of the front wingbox interacts with the rear
wingbox, resulting in a large increase in pressure at the bottom wingbox wall. A normal
shockwave forms at the wall, over which there is a small loss in total pressure, seen from the
total pressure contour of Figure 5.6. Limiting the total pressure losses is an ideal case.
Knowing the Mach number before the shock, the normal shock tables, referenced in
Anderson [24], can be used to find this total pressure loss. In this case, there is
approximately a 5% loss. The normal shock does not penetrate the shear layers coming off
of the wake. The shear layers continue the theoretical intake into the rear wingbox, therefore
setting up a normal shock near the inlet of the "intake".

5.1.4 Flow Over the Top of Rear Wingbox


Along the nose of the rear wingbox, the flow accelerates in part due to the channel of
acceleration between the two circulating areas discussed earlier. Also, flows tend to
accelerate over a curved surface, therefore the acceleration may be a result of the flow
expanding over the nose. The streamlines interact with the rear wingbox in the same manner
as the front wingbox, therefore the flow accelerates enough to move around the nose.
Because the flow accelerates around the nose, the flow separates very early, resulting in the
37

Figure 5.6. Total static pressure in the wake area for the Baseline Case, , with a
moving wall at = 2.

separation bubble on the front part of the wingbox, just aft of the nose. Due to an adverse
pressure gradient, coupled with negative shear stress at this location, the flow separates and
moves back upstream. This is similar to the velocity profile for a basic airfoil which shows
the pressure gradient slowly becomes negative and the shear stress decreases over the airfoil
causing the flow to reach a separation point. As the flow continues to the horizontal
edgeafter the joint, the boundary layer reattaches and the pressure stays fairly constant along
the top of the rear wingbox. When the top corner on the back end of the rear wingbox is
reached, the flow expands consistent with expansion fan theory. The flows velocity
increases around that corner and pushes the shear layer down.

5.1.5 Flow Under the Bottom of Rear Wingbox


The bottom portion of the rear wingbox can be treated as an intake similar to the front
wingbox. However, in this case, there is basically a shock located at the inlet. The pressure
of the flow increases over the shock, thereby decreasing the velocity into the theoretical
intake. Since the flow is subsonic, as the area decreases, the flow accelerates due to velocity-
area relations. The flow increases steadily until it reaches the bottom rear corner. At that
point, another expansion occurs and pushes up the shear layer into the wake area. As a result
of this expansion, an oblique shock comes off the wake and reflects off the moving wall,
cutting through the remnants of the wake downstream. These expansions can be seen in
Figure 5.1.
38

5.2 TANDEM WINGBOX CONFIGURATION AT =4

The discussion of the flow characteristics is broken down into specific areas in
reference to the tandem wingbox configuration at = 4.

5.2.1 Detached Oblique Shock and Flow Over the


Top of Front Wingbox
Figure 5.7 shows a Mach contour for the Baseline Case where the free stream Mach
number and the moving wall are at = 4. All the contours used to discuss the flow

characteristics will be of the Baseline Case unless otherwise specified. At this Mach number,
the standing bow shock is swallowed in the intake area under the front wingbox. As the free
stream velocity increases, the oblique shock is closer to the wingbox. When the oblique
shock has a smaller shock angle, , then the flow will not turn as much, and total pressure
loss is reduced. The stagnation point occurs at the tip of the nose. This location is where the
total pressure is highest and the velocity is zero. In the = 2 case, the stagnation point is
just underneath the nose of the front wingbox. Figure 5.8 shows that the streamlines come
straight into the nose due to the fact that the standing bow shock is no longer a factor.

Figure 5.7. Mach contour of the Baseline Case, , with a moving wall at = 4.

After the initial oblique shock, the flow decelerates, and it expands over the joint
along the top of the front wingbox. As a result, the velocity steadily increases through the
expansion until the boundary layer separates and causes an induced separation oblique shock
to come off the boundary layer. This is also seen in the = 2 case contours. The flow
39

Figure 5.8. Mach contour with the streamlines plotted of the Baseline Case, ,
with a moving wall at = 4.

through an expansion wave is isentropic, therefore there will be no further loss in total
pressure until the induced separation oblique shock is reached.

5.2.2 Flow Under the Bottom of Front Wingbox


The swallowed oblique shock reflects off the moving wall under the front wingbox.
This is something that can be determined beforehand analytically, and then compared to what
Fluent shows. From initial calculations, the Mach angle of the flow is 34 with a Mach
number behind the oblique shock to be equal to 2.49. Figure 5.7 shows that the Mach
number behind the oblique shock is close to this calculated value. Using deflection analysis,
the reflection angle, , found is 23.55 with a Mach number behind the reflected shock of
1.56. This calculated value is approximately the value that can be seen from the Mach
contour, right at the reflection site. This Mach number is still above 1, so the shock will
continue to reflect, which can be seen in Figure 5.7 in the wake area behind the front
wingbox. The shock reflects off the shear layers of the wake after it has gained even more
velocity in the area under the wake. The area under the wake is increasing allowing for the
supersonic flow to increase as well. This is consistent with the velocity-area relations
discussed previously. With each reflection there is a loss in the total pressure and entropy is
generated along the bottom of the front wingbox. Figure 5.9 shows the total pressure loss
over the oblique shock. In the = 2 flow cases, the majority of the losses in the flow,
40

Figure 5.9. Total pressure contour of the Case, , with a moving wall at = 4.

around 28%, are due to the standing bow shock. There is a very weak normal shock in the
wake area that results in about a 5% loss of total pressure. Increasing to = 4, the pressure
increases. The result is a bigger pressure loss over the oblique shock, around a 46% loss, and
dropping to a 14% loss after the shock reflection. In the wake area, in response to the
increased Mach number, several shock reflections occur, while in the = 2 runs, only a

very weak normal shock occurs as seen in Figure 5.6. While the oblique shocks result in
more pressure loss, it is operating at twice the free stream velocity. Ideally, a shock that
remains swallowed for the = 2 cases is desired.

5.2.3 Wake Flow Between Front and Rear Wingbox


The wake area behind the front wingbox, when compared to the = 2 cases,

behaves similarly but with a few differences. In the Mach contour in Figure 5.10, the wake
area is shown. The main recirculation is occurring as it did in the = 2 cases, where the

shear layers coming off the rear of the front wingbox behave like moving walls. A large
recirculation in the wake area can be seen in the pathline contour, Figure 5.11.
A smaller recirculation occurs at the top corner, while a bigger recirculation takes up
most of the area. Following the individual pathlines, there is an area where a channel is
produced and the flow from under the front wingbox is pulled up to the area above the rear
wingbox. There is an area of recirculation from a separation bubble on the top of the rear
41

Figure 5.10. Mach contour of the wake area for the Baseline Case, , with a
moving wall at = 4.

Figure 5.11. Contour of the pathlines in the wake area for the Baseline Case, ,
with a moving wall at = 4.

wingbox aft of the nose. This is occurring because there is an adverse pressure gradient
along with the shear stress that is negative. This creates separation due to the expansion of
the flow over the nose of the rear wingbox. The separation bubble is circulating in the same
direction of the main circulation, which accelerates the flow at the nose of the rear wingbox.
As a result, the flow moves from underneath to above the wake area. These are the same
trends that were seen in the = 2 cases as well.

At the bottom corner of the front wingbox, expansion occurs, causing the shear layers

to be pushed up similar to the = 2 cases (Figure 5.10). Here the reflected shock pushes
the shear layer up. There is a shock coming off the bottom corner of the front wingbox
which is cutting through the reflected shock off the moving wall. As shown in Figure 5.10,
42

at the bottom corner of the front wingbox, the shear layer that comes off the rear is moving
downward very briefly before the reflected shock moves the shear layer up. This shear layer
acts like a wall, and in this case, the downward motion makes that small area a concave
corner. In response to this effect, an oblique shock forms off that corner instead of an
expansion wave. The right-running oblique shock runs into the left-running reflected shock,
resulting in an intersection point and a new slip line for the flow. The individual shocks are
refracted and continue. The pressure and the direction of flow are the same over the new slip
line, but the Mach number may not be the same. Entropy over the slip line will be different
as well [24].

5.2.4 Flow Over the Top of Rear Wingbox


Most of the flow changes in the area on top of the rear wingbox occur near the nose.
As discussed earlier, a large separation bubble develops aft of the rear wingbox nose tip
because of the wake area created by the front wingbox. As the boundary layer reattaches,
there is little change in velocity of the flow on the top of the rear wingbox (Figure 5.7).
Farther from the wingbox body, the flow accelerates until it reaches the wake coming off of
the rear wingbox. Here an oblique shock coming off of the wake's shear layers forms similar
to the = 2 cases.

5.2.5 Flow Under the Bottom of Rear Wingbox and


the Wake Behind
As with the front wingbox, a shock reflection occurs in the area under the rear
wingbox. However, the shock reflections create an interesting pattern that is related to the
Mach number behind the reflected shock seen in Figure 5.10. The oblique shock that is
reflected from under the front wingbox is reflecting off the shear layers of the wake. Each
time the shock reflects, the Mach number decreases over the shock. At a certain point, the
Mach number after the shock reaches a limit that does not allow for the proper deflection. A
normal reflection is not possible and the reflected oblique shock curves toward the wall. The
result is that a normal shock occurs at the wall. Due to the normal shock, another reflected
shock forms and continues downstream. This pattern is known as a Mach reflection. The
same pattern is repeated at the bottom of the rear wingbox wall. However, the straight
reflected shock loses velocity and becomes a normal shock at the wall of the wingbox. A
43

reflected shock forms from the normal shock, but there is not enough momentum to continue
this pattern downstream, so this shock pattern ends. What follows is a large amount of
instability in the area toward the end of the rear wingbox. The flow over the normal shock
remains parallel with the moving wall, and continues to lose velocity and total pressure. After
this point, the flow's velocity gradually increases due to the expansion around the bottom
back corner of the rear wingbox [24]
The wake area behind the rear wingbox is shown in Figure 5.7. Following the shock
pattern on the bottom of the rear wingbox, the flow accelerates leading into an expansion.
This accelerated flow pushes up the shear layers coming off the wake. The wake is more
turbulent than the wake off of the front wingbox as a result of the eddy formation. Along the
top of the rear wingbox, the shear layers are being pushed up as well. The flow accelerates
on the top and bottom of the wake area causing oblique shocks to come off of the shear
layers. A strong oblique shock forms off the bottom shear layer of the wake, which reflects
off the moving wall and cuts through the wake further downstream. This strong oblique
shock is common to both = 2 and 4 simulations.

5.3 EFFECT OF MACH NUMBER


The swallowing of the bow shock at = 4 is comparable to the swallowing of a

shock by an intake by overspeeding (Figure 5.7). Initial theoretical calculations are carried
out in order to see exactly when the shock will be swallowed by looking at when a normal
shock appears at the inlet of the "intake." Each case has a different ratio, or inlet area to

throat area ratio. For these ratios, is referenced from the ground to the tip of the nose of

the wingbox, and is the ground clearance for each case (Figure 5.12). This ratio is used to

find the subsonic Mach number behind the normal shock at the inlet. Once the Mach number
behind the shock is found, the standard normal shock properties table [24] is used to calculate
the theoretical . The value of will have to be bigger than the calculated Mach
number in order for the shock to be swallowed.
At = 4, CFD indicates that at each ground clearance, the detached oblique shock
remains swallowed in the theoretical intake. However, based on the initial calculations, only
44

Figure 5.12. Area ratio reference for analysis


of a swallowed shock.

the case where the wingboxes are at should be swallowed because = 4 is greater than
the calculated of 3.02 from an area ratio of 1.393. The case with the wingboxes at
shows that with an area ratio of 1.524, the shock should be swallowed at a greater

than 4.58, which is larger than the actual of 4.

The initial calculations for the case with the wingboxes at are perplexing. The

area ratio, 1.786, is high enough that the corresponding Mach number after the shock does
not register on the normal shock tables, which means that would have to be greater than

50. The point at which sonic conditions are reached at the throat area, , could be farther

upstream, in which case, the area ratio would decrease allowing for a proper to be

calculated. With this in mind, in the case of the smallest ground clearance, if the throat area
was .6 in instead of .5 in, then the calculated would be 3.99, which would then

correspond to the contours generated in Fluent indicating that the shock is swallowed.
To study the conditions under which the detached shock remains swallowed, the
Baseline Case is run at = 2.5. The purpose is to determine if the CFD analysis accurately

predicts the shock being swallowed at a lower Mach number than 4. This will indicate what
the appropriate range of the free stream velocity should be in order to minimize the losses
with this tandem wingbox configuration. With = 2.5, the choked flow still pushes the

detached shock upstream due to the Kantrowtiz violation. This indicates that the a

between 2.5 and 4 is needed to swallow the shock. Finding the ideal would be an

interesting test for future work on this topic. Minimizing the total pressure losses is
45

important for maximizing the performance of the tandem wingbox. However, the purpose of
the MagLev is to evaluate the aerodynamics of the payload being tested on the test rail.
Therefore, while the aerodynamics of the tandem wingbox in ground effect are significant in
how the sled will perform, always testing at maximum performance is unlikely when higher
Mach numbers are desired for payload testing.
46

CHAPTER 6

EFFECT OF GROUND CLEARANCE AND


DISTANCE BETWEEN THE TANDEM
WINGBOXES FOR MACH 2 RUNS

This chapter presents the effect of ground clearance and the distance between the
wingboxes for all of the cases investigated at = 2.

6.1 EFFECT OF GROUND CLEARANCE, H


Figure 6.1 shows the pressure coefficient over the front wingbox for the first three
cases, , , and , where the wingboxes are 1 m apart. When comparing the

pressure coefficients of the flow at the three different heights above the wall, H, the pressure
along the bottom of the front wingbox is higher when the wingbox is closest to the wall. As
the wingbox moves away from the wall, the pressure decreases.

Figure 6.1. Pressure coefficient over the front wingbox at various heights above
the moving wall with the wingboxes 1m apart at = 2.
47

The pressure is higher when the wingboxes are closer to the moving wall because of
the mass flow through the theoretical intake. Due to continuity, after the bow shock the mass
flow coming in will be equal to the mass flow coming out. For each height above the wall,
H, the mass flow is different. For each case there is a detached bow shock, and as the area
increases as the ground clearance increases, the mass flow is going to increase because they
are directly proportional. Pressure is also directly proportional to the mass flow rate when
the ideal gas law is substituted in for the density.
. (6.1)

. (6.2)

Area, however, is the dominant factor in what is changing the mass flow rate for each case.
Therefore, the pressure decreases as the wingbox moves away from the wall. If the pressure
increased along with the area, the mass flow would be even higher. There has to be a balance
between the mass flow rate factors. When the area is dominant, then the density or pressure
and the velocity are going to have to compensate for the difference. Figure 6.1 indicates that
the closer distance to the ground will result in the greatest amount of pressure which means
higher lift and drag loads on the bottom surface of the wingbox. The pressure force along the
bottom of the wingbox is normal to the surface, therefore, the components of the force, the
lift and drag, are larger along the bottom due to ground effect.
Figure 6.2 shows the pressure coefficient over the rear wingbox for the same cases.
The flow along the top of the rear wingbox is minimally affected by the magnitude of the
ground clearance with the exception of the area just aft of the nose. This is the area where
the separation bubble is present and pressure will be at a minimum at this location. Within
the wake, there is an acceleration occurring at the nose of the rear wingbox due to the
recirculation and separation bubble rotating in the same direction. As a result, a portion of
the flow from under the front wingbox is carried to the top of the rear wingbox as discussed
in Chapter 5. Similar to the front wingbox analysis, the ground has more effect along the
bottom surface of the rear wingbox than the top as seen in Figure 6.2. While the distance
48

Figure 6.2. Pressure coefficient over the rear wingbox at various heights above the
moving wall with the wingboxes 1m apart at = 2.

closest to the wall results in the greatest pressure at the front wingbox, the opposite is the
case for the rear wingbox. All ground clearances tested follow a similar trend of a peak in
pressure, but the location and magnitude are different. A major factor in this result is due to
the wake formed from the front wingbox. Recalling literature from chapter one, distance
between tandem objects plays a key role in how the loads will be affected, specifically due to
the way in which the wake reattaches to the rear object, or if it reattaches at all. Another
factor in the difference is the normal shock that develops in the theoretical intake. After the
front wingbox, the flow has been accelerating and a normal shock is formed. The area
between the shear layers and the moving wall have an effect on the location, as well as the
magnitude of the normal shock. As the distance between the moving wall and the wingbox
increases, the normal shock is going to be more developed and therefore stronger. A stronger
shock is going to slow down the flow across the shock more than a weaker shock. However,
49

all shocks that occur are still weak compared to the standing bow shock in front of the front
wingbox, because the total pressure loss is minimal as seen in Figure 5.6.
Applying this analysis to Figure 6.2, it is evident that as the wake's shear layers are
attaching to the rear wingbox, the peak in pressure is reached along with an increase in
pressure after the normal shock. In terms of ground clearance, H, a stronger shock as a result
of a larger H pushes up the shear layers and affects the location of reattachment. They
reattach higher up towards the nose, whereas, the smaller H's normal shock does not affect
the shear layers as dominantly, and therefore they reattach farther down from the nose. The
difference in location of reattachment along the rear wingbox is minimal, around .125 m
further down from the nose. However, it is an effect of ground clearance. In Figure 6.3, the
wake area of the Case indicates the reattachment location of the shear layers to the rear

wingbox.

Figure 6.3. Mach contour of the wake area of the Case with a moving wall at
= 2 in reference to the reattachment location of the shear layers on the rear
wingbox.

Figure 6.4 shows the pressure along the moving wall. This plot shows the flow
patterns through the behavior of the pressure. There is an increase in pressure due to the
standing bow shock in front of the front wingbox. After the bow shock, the pressure will
decrease through the "intake" as the velocity increases, until the normal shock forms just in
front of the rear wingbox. The pressure will increase again, then steadily decrease, until the
50

Figure 6.4. Pressure along the moving wall with the wingboxes at various
heights above the wall at 1m apart at = 2.

oblique shock that is reflected off the ground is reached which can be seen in Figure 5.1.
The pressure plot shows that a smaller ground clearance causes the bow shock to be pushed
farther upstream than the larger ground clearances. The normal shock that develops before
the rear wingbox shows the opposite, similar to the pressure coefficient plots of the rear
wingbox, in that the smaller clearance produces less of a pressure increase than the larger
distances. This is another indication that the distance between the wingboxes influences the
wake effect on the rear wingbox.
The Mach number along the moving wall is shown in Figure 6.5. It is the opposite of
the pressure plot of Figure 6.4 since velocity and pressure are indirectly proportional.
Earlier, it was discussed that in some cases the simulation blew up where the standing bow
shock impinges on the moving wall. This plot of the Mach number along the moving wall
shows this phenomena (Figure 6.5). The Case there is a large jump in Mach number
right before the bow shock at this impingement point. After the bow shock the Mach number
51

Figure 6.5. Mach number along the moving wall with the wingboxes at various
heights above the wall at 1m apart at =2

is going to decrease significantly because of the strength of the bow shock. The Mach
number steadily increases until the normal shock is reached in front of the rear wingbox. The
last build up of velocity comes before the oblique shock off the wake of the rear wingbox
when it reflects off the moving wall before settling back into a Mach number closer to the
free stream velocity.
The velocity at the wall is changing along the moving wall even though the velocity
of the wall was set in Fluent to be equal to the free stream velocity. The moving wall was
also set to be adiabatic, therefore there is no heat flux. Mach number is based on the velocity
divided by the speed of sound, which is dependent on the temperature. Therefore the
temperature along the moving wall must be changing, not from heat flux through the wall,
but from the skin friction along the moving wall. Figure 6.5 shows the Mach number is
changing based on the flow features the wall encounters. Friction generates heat which will
be transferred and affect the flow. Areas where the shock impinges the wall are going to be
52

areas of higher friction, causing a jump in velocity. Figure 6.6 shows the temperature along
the moving wall. There are significant increases in temperature at the particular shock
locations. While the heat flux through the wall is zero, there is still an adiabatic wall
temperature limit that is based on the velocity of the wall [24]. As the velocity increases, this
limit will increase.

. (6.3)

Figure 6.6. Temperature (K) along the moving wall with the wingboxes at
various heights above the wall at 1m apart at = 2.

The factor along the moving wall is shown in Figure 6.7. Ideally, this turbulence

factor should be in the appropriate log layer range discussed earlier. It was found that is

much larger than this range in a few areas. The bow shock is one of these locations. The
extremely large value is indicating that a separation is occurring in this area where the bow
shock impinges the moving wall. The adaptive tool in Fluent was used to better refine the
grid along the moving wall in order to get this number down, which it did over much of the
moving wall. However, the adaptive tool cannot change the shear stress at these particular
points and how it affects the value.
53

Figure 6.7. Turbulence factor, , along the moving wall with the wingboxes
at various heights above the wall at 1m apart at = 2.

The trends of these parameters are similar to the final three cases, , , ,

where the wingboxes are 1.5m apart. The effect of distance apart, S, will be specifically
discussed in the next chapter. Here, the effect of ground clearance is still the focus now that
the wingboxes are slightly farther apart. Figure 6.8 shows the pressure coefficient over the
front wingbox for the various ground clearances above the moving wall. This plot is
identical to the first three cases where the wingboxes are 1m apart. While the front wingbox
would see little or no change when S is only increased by a half of a meter, the slight increase
in S could affect the rear wingbox enough to cause changes to the flow upstream. This plot
reaffirms the earlier trend, that as the wingbox gets closer to the ground, the pressure
increases. The pressure over rear wingbox, however, is slightly different than the previous
cases.
54

Figure 6.8. Pressure coefficient over the front wingbox with the wingboxes at
various heights above the moving wall at 1.5m apart at = 2.

Figure 6.9 presents the data for the pressure coefficient over the rear wingbox with
the wingboxes at the larger distance apart, . Because the wake area is larger, there is a
difference in how the shear layers attach to the rear wingbox. The peak in pressure, where
the wake reattaches, is slightly less than that of the smaller , around a 4.4 percent drop for

and and a 10 percent drop for . The larger ground clearance still creates the greater

build up of pressure due to a stronger shock, as well as the location of reattachment being
closer to the nose. As the ground clearance decreases, the reattachment moves farther from
the nose as with the previous cases. The biggest change in location of reattachment with
respect to S is in the and Cases. There is a 16 percent difference in location with
a smaller S reattaching later than a larger S. After the expansion at the bottom corner of the
55

Figure 6.9. Pressure coefficient over the rear wingbox at various heights above the
moving wall with the wingboxes at 1.5m apart at =2

front wingbox, the wake area is pushed up, and along with the larger wake area, the shear
layers' reattachment to the rear wingbox is being pushed closer towards the nose.

6.2 FLOW FEATURES WITH THE EFFECT OF DISTANCE


BETWEEN WINGBOXES, S
Figure 6.10 shows the Mach contour of the Case. For the = 2 runs, a

standing bow shock is a result of the back pressure pushing the detached shock out of the
theoretical intake. The flow accelerates through the "intake" and expands around the bottom
corner of the front wingbox. This increase in velocity leads, as before, to a normal shock
before the rear wingbox. The difference in distance between the tandem wingboxes, S, is
going to affect how the flow behaves, especially in the wake area, which was a point made in
the literature study of flow around tandem bodies [17,18]. The larger wake area does create a
56

Figure 6.10. Mach contour for the Case with a moving wall at =2

slight difference upstream due to the flow from the rear wingbox. For example, the angle of
the induced oblique shock off the boundary layer separation is larger with a smaller S.
In Figure 6.11, the wake area is inspected in more detail. While the previous contour
had two dominate recirculation areas (Figure 5.3), the larger wake area in the Case
gives the recirculation areas more room to expand. The Mach contour shows three areas of
significant recirculation, which can be seen in even greater detail in the pathline contour of
Figure 6.12. Here, the area of recirculation at the top back corner of the front wingbox is
much larger than the previous cases with a smaller . The larger distance, , allows the
eddies more area to shed and grow. As in previous cases, the separation bubble at the nose of
the rear wingbox is present. This, along with the largest recirculation, cause an acceleration
at the nose, carrying the flow from under the front wingbox to over the top of the rear
wingbox. The wingboxes are still close enough that the shear layers reattach to the rear
wingbox creating that area of high pressure just aft of the nose seen in Figure 6.13 of the
pressure contour.
The normal shock that forms before the rear wingbox is still a weak shock and as the
ground clearance increases, the strength of the shock increases. The shock is weak due to the
small total pressure loss across the shock which can be seen in the total pressure contour of
Figure 6.14. This is the more desirable scenario compared to the standing bow shock before
the front wingbox which creates significant pressure losses. Referring back to Figure 6.10,
after the rear wingbox, the wake follows similar trends from the previous cases with a
smaller . The flow expands over the top and bottom corners on the back of the rear
57

Figure 6.11. Mach contour of the wake area for the Case with a moving wall
at =2

Figure 6.12. Pathline contour of the wake area for the Case with a moving wall
at = 2.

wingbox. As the wake develops, vortex shedding occurs. An oblique shock develops off the
wake and reflects off the moving wall, cutting through the wake father downstream. While
the flow physics have not changed much, the next section will look at the pressure coefficient
over the tandem wingboxes with respect to the distance between them, S, in order to get a
better idea of just how the distance contributes to the flow behavior near the ground.

6.3 EFFECT OF DISTANCE BETWEEN WINGBOXES, S


In this section, the different distances between the wingboxes, S, are compared with
respect to the height above the moving wall, H. Figure 6.15 shows the pressure coefficient
58

Figure 6.13. Pressure contour of the wake area for the Case with a moving wall
at = 2.

Figure 6.14. Total pressure contour of the wake area for the Case with a moving
wall at =2

over the front wingbox for the and Cases. The pressure coefficient over the front

wingbox is identical for both cases. Any differences result from the larger wake area
affecting the flow by the rear wingbox, and the resulting changes to the flow upstream. In
this case, the small change in S has no significant impact on the upstream flow to result in a
difference in the pressure.
59

Figure 6.15. Pressure coefficient over the front wingbox closest to the moving wall
with the wingboxes at different distances apart at =2( and Cases).

Figure 6.16 shows the pressure coefficient over the rear wingbox with respect to
distance between the wingboxes, S, and indicates a difference in the effect of wake area.
Along the top of the rear wingbox, just aft of the nose, there is a slight difference in pressure.
This is where the separation bubble is located. The larger wake area results in a more
adverse pressure gradient, and therefore, a larger separation bubble. The area of significant
difference is where the shear layers reattach to the rear wingbox. The shock just before the
rear wingbox is weakest for the and Case, which results in less of a pressure
increase. Therefore, the magnitudes of the pressure increases are similar, but are the least for
the Case due to the pressure increase after the shock as well as the reattachment of the
shear layers. There is a difference in the location of this reattachment, however. The smaller
wake area reattaches farther from the nose than the larger wake area. After the expansion on
the bottom corner of the front wingbox, the wake is pushed upward. The larger distance, ,
60

Figure 6.16. Pressure coefficient over the rear wingbox closest to the moving wall with
the wingboxes at different distances apart at =2( and Case).

gives the wake area more room to expand, therefore pushing the wake's shear layers up even
more. This allows for reattachment closer to the nose as compared to the case with the
shorter distance between the wingboxes, . With an earlier reattachment, the flow has more
time to accelerate, leading to the lower pressure than the other case, , near the end of the
rear wingbox.
The pressure along the moving wall for these two cases is presented in Figure 6.17.
When the wingboxes are farther apart, the bow shock is pushed farther upstream from the
theoretical intake because the back pressure is larger in the case with the greater distance, .
The Case shows the normal shock just before the rear wingbox to be upstream of the
other case's shock. Therefore, while the magnitude of the shocks along the moving wall, and
61

Figure 6.17. Pressure along the moving wall with the wingboxes closest to the wall
at different distances apart at =2( and Case).

pressure behavior over the wingboxes are on par with one another, the value of S is affecting
the location of the shocks and the reattachment of the shear layers.
The next cases to be compared are the and Cases. Figure 6.18 shows the

pressure coefficient over the front wingbox, while Figure 6.19 shows the pressure coefficient
over the rear wingbox. As in the previous cases, the pressure coefficient over the front
wingbox changes very little in pressure with respect to the value of S. The only area with
some difference is the top back corner of the front wingbox. Just before the end of the top
surface of the front wingbox, the boundary layer separates, creating an induced oblique
shock. As a result, the pressure increases at this point. This is then followed by an area of
recirculation in the wake. The difference lies in the point of separation. The Case
separates earlier than the Case. The data on the pressure coefficient over the rear
wingbox (Figure 6.19) reflects that of the previous cases in the areas of the shear layer
62

Figure 6.18. Pressure coefficient over the front wingbox at the middle height from the
wall with the wingboxes at different distances apart at =2( and
Cases).

reattachment point and the separation bubble aft of the nose. Unlike the previous cases
however, there is not a difference in location of the reattachment, but a difference in the
magnitude of the pressure increase. Figure 6.20 of the pressure along the moving wall shows
that the bow shock's location is not affected by the value of S. However, the normal shock's
location before the rear wingbox is affected. The Case shows a shock upstream of the
Case. This is the opposite effect seen with the and Cases. The has a
higher pressure increase at the reattachment point than the Case. As the ground
clearance increases, the strength of the normal shock increases, therefore the pressure build-
up is going to be larger. The Case, from the previous comparison, showed a higher
pressure increase at the reattachment site as well.
63

Figure 6.19. Pressure coefficient over the rear wingbox at the middle height from
the wall with the wingboxes at different distances apart at =2( and
Cases).

Lastly, the = 2 runs for the and Cases are compared. Figure 6.21
presents the pressure coefficient over the front wingbox, while Figure 6.22 shows the
pressure coefficient over the rear wingbox. The pressure coefficient over the front wingbox,
similar to the previous cases, shows that the increase in distance, S, does not have enough of
an effect on the flow upstream to cause changes to the pressure. At the top corner of the
front wingbox, the boundary layer for the Case is separating from the wingbox earlier
than the Case, which is seen in the previous cases' comparisons as well. The pressure
coefficient over the rear wingbox is where the effect of distance between the wingboxes, S,
occurs at the point of the shear layers' reattachment. The Case has the highest increase
in pressure due to the pressure increase from the shock upstream as well as the reattachment
of the shear layers. However, the difference is less than 5 percent. The Case has a
64

Figure 6.20. Pressure along the moving wall with the wingboxes at the middle
height from the wall at different distances apart at =2( and
Cases).

more sudden increase in pressure at the reattachment point than the Case, which is more
gradual. Figure 6.23 shows the pressure along the moving wall.The standing bow shock is
not affected by the increase in distance between the wingboxes, S, and the normal shock
before the rear wingbox for the Case is upstream from the Case, similar to the
and Cases.
Analysis of the figures featuring the rear wingbox indicates the relationship between
H and S. When the wingboxes start off closest to the wall, , the most significant
difference can be seen with respect to the distance between the wingboxes, S. The
65

Figure 6.21. Pressure coefficient over the front wingbox farthest from the
moving wall with the wingboxes at different distances apart at =2(
and Cases).

reattachment point is affected by the length of the wake. As H increases, the reattachment
point is less affected, as well as the pressure increase due to the reattachment. When the
wingboxes reach a ground clearance of , the wingboxes are reaching a distance where
these changes in H and S present less of an effect.
66

Figure 6.22. Pressure coefficient over the rear wingbox farthest from the moving wall
with the wingboxes at different distances apart at =2( and Cases).
67

Figure 6.23. Pressure along the moving wall with the wingboxes farthest from
the wall at different distances apart at =2( and Cases).
68

CHAPTER 7

EFFECT OF GROUND CLEARANCE AND


DISTANCE BETWEEN THE TANDEM
WINGBOXES FOR MACH 4 RUNS

This chapter discusses the effect of ground clearance and distance between the
wingboxes for all of the cases investigated at = 4.

7.1 EFFECT OF GROUND CLEARANCE, H


This section presents the effect of ground clearance for each Case at = 4. Figure

7.1 shows the pressure coefficient over the front wingbox, and a very different pattern is
observed from the = 2 runs. The various contours for the = 4 runs discussed in

Chapter 5 indicated that the shock is no longer pushed upstream, but is swallowed in the
theoretical intake. The pressure is highest at the nose of the wingbox where the stagnation
point is located. In Figure 6.1, the streamline contour, the streamlines come straight into the
nose, rather than slightly underneath as in the = 2 cases, due to the fact that there is no
longer a standing bow shock in front of the wingbox. This removes the effect of spillage into
the theoretical intake which caused the stagnation point to be just under the nose for the =
2 runs. Along the top of the wingbox, the expansion takes place where the pressure
coefficient drops and then remains unchanged until the end of the front wingbox. The
ground clearance is not affected to the same degree as it was in the = 2 cases until the end
of the front wingbox is reached. At the top of the wingbox, the pressure coefficient at the
end of the front wingbox is increasing for the Case because the boundary layer is
separating from the wingbox surface shown in the Mach contour of Figure 7.2. Along the
bottom surface, shocks are reflecting off the wall and the wingbox itself resulting in the
unusual pressure coefficient trend for the Case. Figure 7.3 of the Mach contour for the
Case shows that the oblique shock reflecting off the wall is interacting with an induced
oblique shock due to separation. This induced shock increases the pressure, and with the
69

Figure 7.1. Pressure coefficient the front wingbox with the wingboxes at
various heights above the moving wall at 1m apart at = 4.

Figure 7.2. Mach contour of the wake area for the Case with a moving wall at
= 4.
70

Figure 7.3. Mach contour of the wake area for the Case with a moving wall at
= 4.

effects of the reflected shock intercepting the induced shock, the pressure will remain higher
until the end of the front wingbox where flow expands into the wake area. For the cases
farther from the moving wall, there is no induced oblique shock due to separation resulting in
a steady increase in velocity along the bottom surface. In the and Cases, the
shock reflection off of the moving wall bypasses the end of the front wingbox and reflects off
the wake instead.
The data in Figure 7.4, pressure coefficient over the rear wingbox, points to the
instabilities that are associated with the increase in . In the = 2 runs, the wake effect

on the rear wingbox plays a key role in the flow behavior around the rear wingbox. The
wake area of the Case (Figure 5.10) shows the same trend as the = 2 runs. There

are two large areas of recirculation causing an acceleration at the nose of the rear wingbox.
However, there is now an oblique shock interfering with the wake instead of a normal shock.
This oblique shock is reflecting off the moving wall, and in the and Cases, is

resulting in Mach reflections (Figures 5.10 and 7.3).The Case (Figure 7.2) shows the
oblique shock reflecting off of the moving wall and bypassing the rear wingbox, leading to a
more stable pressure distribution similar to the = 2 runs. Along the top of the rear
wingbox, the pressure is decreased by the acceleration at the nose of the rear wingbox,
followed by a separation bubble, after which the pressure remains unchanged over the rest of
71

Figure 7.4. Pressure coefficient over the rear wingbox with the wingboxes at
various heights above the moving wall at 1m apart at = 4.

the rear wingbox similar to the = 2 runs. In the Case, along the bottom of the
wingbox, there are two increases in pressure. The first increase in pressure results from the
shear layers of the wake reattaching to the rear wingbox, which can be seen in Figure 7.3 of
the wake area. When the oblique shock reflects off of the moving wall, the resulting
reflection is a Mach reflection at the surface of the wingbox. The stem of the Mach
reflection is a normal shock, therefore a larger pressure increase occurs over the shock. After
this point, the pressure will steadily decrease as the velocity increases and expands over the
back corner of the rear wingbox. The Case follows a similar pattern, with both

reflections of the oblique shocks being Mach reflections as seen in Figure 5.10. The first
pressure increase is also due to the reattachment point, which is farther upstream than the
Case due to the position of the oblique shock reflection off of the wake. The second

increase occurs at the second Mach reflection. The effect of the fluctuations in pressure is
72

high instability until the pressure decreases due to the expansion around the back corner of
the wingbox. There is one increase in pressure in the Case due to the reattachment of

the shear layers of the wake. The pressure decreases steadily through the expansion.
The effect of ground clearance is not as clear with increased . For the front

wingbox, having a swallowed detached shock creates an environment where the slight
change in ground clearance does not affect the pressure until the end of the wingbox is
reached due to the location of the oblique shock reflection. The larger the ground clearance
the further downstream the shock reflection will take place. At a smaller H, the shock
reflection occurs at a point where it will interact with the end of the wingbox, resulting in a
change in flow pattern. The reattachment point on the rear wingbox is not as clearly affected
by ground clearance as in the = 2 cases, and this is because of the shock reflection from

the front wingbox and wake interaction before the rear wingbox. The shock reflection for

is farthest downstream which pushes the reflection off of the wake farther downstream. This
leads to reattachment farther from the nose of the rear wingbox. For , the shock reflection

interacts with the front wingbox before reaching the wake, but this does not cause
reattachment earlier than the other cases. Therefore, shock reflection and wake interaction
are the leading contributors to how and where reattachment happens. As the Mach number
increases, the larger the ground clearance, the more stable the flow will be around the rear
wingbox, causing the loads to be more affected when the wingboxes are closer to the moving
wall.
Figure 7.5 shows the pressure along the moving wall for the , , and

Cases. This Figure shows where the oblique shocks are reflecting and reaffirms the
observation made earlier that as the ground clearance increases, the point of reflection is
located further downstream. The second increase in pressure along the moving wall is
located at the second reflection point. As H decreases, the shock reflections become Mach
reflections which have a normal shock stem. The pressure increase over a normal shock is
more significant than over an oblique shock. This compounds the pressure increase leading
to a much higher pressure increase for smaller ground clearances seen in Figure 7.5.
73

Figure 7.5. Pressure along the moving wall with the wingboxes at various heights
above the wall at 1m apart at = 4.

Figure 7.6 shows the Mach number along the moving wall. Under these conditions,
the velocity decreases over the shock reflections while the pressure increases. At the second
shock reflection location, the Case's Mach reflection is stronger than in the Case.

With the extra ground clearance, the normal shock stem is larger which leads to a stronger
shock, and in turn, increases the loads at this point on the moving wall.
The remaining Cases, , , and , are analyzed with respect to the effect
of ground clearance. Figure 7.7 shows the pressure coefficient over the front wingbox, while
Figure 7.8 shows the pressure coefficient over the rear wingbox. The data in Figure 7.7
reinforces the trend of the previous cases therefore indicating that the increased wake area is
not making significant changes upstream. However, where previously the Case
separated along the top end of the front wingbox, the Case is now the case separating in
this location. Along the bottom surface for the Case, an increase in pressure is
74

Figure 7.6. Mach number along the moving wall with the wingboxes at various
heights above the wall at 1m apart at = 4.

occurring where the reflected shock is interacting with the induced oblique shock off of the
surface of the front wingbox seen in Figure 7.9.
The large ground clearance, , results in one point of reflection (Figure 7.10). This
translates to one increase in pressure where the shear layers are reattaching to the rear
wingbox. In the Case, the reflected shock from the front wingbox hits the wake at the
nose of the rear wingbox seen in Figure 7.11. Therefore, there is an increase in pressure after
the shock at the nose of the rear wingbox. After the oblique shock reflects off of the wake, a
steady increase in velocity will drop the pressure until a Mach reflection forms on the surface
of the wingbox. This causes another peak in pressure which decreases through the expansion
at the end of the rear wingbox. With the Case, the shear layers off of the wake engulf
much of the area under the nose of the rear wingbox keeping the pressure high, and
increasing through the Mach reflection seen in Figure 7.9.
75

Figure 7.7. Pressure coefficient over the front wingbox with the wingboxes at
various heights above the moving wall at 1.5m apart at = 4.

Figure 7.8. Mach contour of the wake area for the Case with a moving wall at
= 4.
76

Figure 7.9. Pressure coefficient over the rear wingbox with the wingboxes at various
heights above the moving wall at 1.5m apart at = 4.

Figure 7.10. Mach contour of the wake area for the Case with a moving wall at
= 4.
77

Figure 7.11. Mach contour of the wake area for the Case with a moving wall at
= 4.

7.2 FLOW FEATURES WITH THE EFFECT OF DISTANCE


BETWEEN WINGBOXES, S
With the increase in , the wake area does encounter changes mainly due to the
reflected oblique shock interacting with the wake. Increased turbulence in the wake area
results from the faster as seen in the pathline contours of the various cases presented. In
the Case, the flow around the bottom corner of the front wingbox expands less than the
Case with a smaller wake area. The effect of the oblique shock off the wake results in a
Mach reflection at the moving wall (Figure 7.9) rather than an oblique shock reflection as
seen in the Case. The larger wake area leads to changes to the flow upstream due to

the flow around the rear wingbox. The flow separates from the front wingbox's top surface
near the back corner, which adds to the shear layers being pushed upward. The Case

shows a similar pattern in that the oblique shock off of the wake forms a Mach reflection at
the moving wall (Figure 7.11). With a larger wake area, the Mach reflection is slightly
upstream of the Case. The shock reflection off of the wall by the front wingbox causes

the wake's shear layers to be pushed up to the nose creating a very different flow
recirculation pattern from the Case. The Case (Figure 7.10) shows an oblique

shock off the moving wall rather than a Mach reflection. The difference in S changes when
the oblique shock reflection from the front wingbox interacts with the wake, which changes
78

the location of the next shock reflection off of the moving wall for the and Cases.

The oblique shocks in these cases result in less total pressure loss than the bow shock for the
= 2 runs.

All cases present a similar flow pattern in terms of shock reflection and interaction.
However, the higher Mach number, which results in increased amounts of turbulence, makes
the behavior in the wake not as predictable as the = 2 runs where the pathline contours
the behavior in the wake not as predictable as the = 2 runs where the pathline contours

showed the same two recirculation areas. Figures 7.12 and 7.13, for the and

Cases, show the recirculation difference and how the flow behaves with increased S. A
larger recirculation at the top corner of the front wingbox results from a larger S, and the
main recirculation area now has two points of recirculation at which the eddies are separating
or shedding. The separation bubble is still present in both cases causing the flow to
accelerate flow particles from underneath the wingbox to the top of the rear wingbox. For
the , , , and Cases, the recirculation does not separate into two. Figures
7.14 and 7.15 show the pathline contours of the and Cases. In the case of a
smaller S, a large separation bubble at the nose of the rear wingbox extends to the back end
of the front wingbox, unseen in any of the other cases. The larger S continues to create two
main areas of recirculation propelling the flow from underneath the wingboxes to above the
wingboxes. While no real trend is seen with increased wake area, with increasing and S,
the flow is more turbulent causing different eddy formations which affect the flow behavior
around the rear wingbox.

7.3 EFFECT OF DISTANCE BETWEEN WINGBOXES, S


The discussion begins with the and Cases. Figure 7.16 shows the pressure
coefficient over the front wingbox. Similar to the = 2 runs, no significant changes are
occurring in the front wingbox due to the effect of distance apart. The Case does show
wingbox is upstream of the smaller wake area.
79

Figure 7.12. Pathline contour for the Case with a moving wall at = 4.

Figure 7.13. Pathline contour for the Case with a moving wall at = 4.

Figure 7.14. Pathline contour for the Case with a moving wall at = 4.
80

Figure 7.15. Pathline contour for the Case with a moving wall at = 4.

Figure 7.16. Pressure coefficient over the front wingbox closest to the moving wall
with the wingboxes at different distances apart at =4( and Cases).

the flow separating near the top back corner of the front wingbox leading to an induced
oblique shock. Due to the larger wake area, separation along the bottom surface of the
wingbox is upstream of the smaller wake area. Figure 7.17 shows the pressure coefficient
over the rear wingbox. As in previous cases, the wake area is going to affect the rear
wingbox more than the flow upstream. There are two significant increases in pressure in the
smaller wake area of the Case, the first due to the reattachment of the shear layers on
81

Figure 7.17. Pressure coefficient over the rear wingbox closest to the moving wall
with the wingboxes at different distances apart at =4( and Cases).

the rear wingbox, and the second due to the Mach reflection at the bottom surface. The
Case presents a similar flow pattern. However, the shear layers engulf the front end of the
rear wingbox, leading to the steady pressure increase resulting from reattachment and the
Mach reflection, which occur in the same location. The reattachment point for the smaller S
is farther upstream than for the larger S. This allows the flow to accelerate before the Mach
reflection, causing a larger jump in pressure. In the Case, the reattachment and Mach
reflection interact which results in less of a pressure increase through the Mach reflection.
Figure 7.18 shows the pressure along the moving wall for the and Cases.

From this comparison, it is evident that the larger wake area causes the shock reflections to
be just upstream of the smaller wake's reflections. There is also a significant difference in
the magnitude of the increased pressure after the second shock reflection, the Mach
reflection.
82

Figure 7.18. Pressure along the moving wall with the wingboxes closest to the
wall at different distances apart at =4( and Cases).

There is no distinguishable difference in pressure over the front wingbox for the
and Cases. Figure 7.19 shows the pressure coefficient over the rear wingbox for
these two cases. The same two pressure increases as in the last two cases occur, one for
reattachment and the other for the Mach reflection. The Mach contour of the Case
(Figure 7.11) shows that the shock reflection from the front wingbox interacts with the larger
wake area in that it pushes the shear layers up toward the nose. The reflected shock off of the
wake's shear layers is essentially at the nose of the rear wingbox. The increase in pressure at
this point resembles a stagnation point similar to that of the front wingbox. Reattachment of
the shear layers occurs farther downstream for the Case, and there is high instability
after the Mach reflection. The location of the Mach reflection for the Case is farther
downstream than the Case because of the shock-wake interaction for the larger wake.
83

Figure 7.19. Pressure coefficient over the rear wingbox at the middle height
from the moving wall with the wingboxes at different distances apart at =4
( and Cases).

Figure 7.20 presents the pressure along the moving wall for the and
Cases. While the first shock reflection off of the moving wall under the front wingbox is
stronger for a smaller S, the Mach reflection for a larger S under the rear wingbox is stronger
than the Mach reflection for a smaller S as indicated by the increase in pressure over the
shock. Due to the shock-wake interaction for the Case, the flow before the Mach
reflection accelerates more than for the Case, which generates the extra increase in
pressure over the Mach reflection.
The last two cases examined are the and Cases. Figure 7.21 shows the
pressure coefficient over the front wingbox, while Figure 7.22 shows the pressure coefficient
over the rear wingbox. The only effect of S seen in the front wingbox is the separation of the
boundary layer for the Case. The pressure coefficient over the rear wingbox shows a
difference with respect to S in the magnitude of the pressure increase at the reattachment
84

Figure 7.20. Pressure along the moving wall with the wingboxes at the middle
height from the wall at different distances apart at =4( and
Cases).

point. A larger increase in pressure occurs in the smaller wake area than the larger wake
area. Trends in the wake area (Figures 7.14 and 7.15) are very different for these two cases.
The larger separation bubble on the top of the nose of the rear wingbox is pushing the shear
layers off the top-rear corner of the front wingbox upstream contributing to the boundary
layer separation. This large separation bubble at the nose is also interfering with the large
recirculation area in the wake, affecting the reattachment to the rear wingbox.
Analysis of the pressure along the moving wall for these two cases (Figure 7.23)
indicates that the shock reflection under the front wingbox is not affected by wake distance.
The location and magnitude are the same for both wake lengths. There is a difference with
the second reflection under the rear wingbox. For the Case, which has a smaller wake
85

Figure 7.21. Pressure coefficient over the front wingbox farthest from the moving
wall with the wingboxes at different distances apart at =4( and
Cases).

area, the second reflection off of the moving wall is upstream from the Case. With
respect to the Case a stronger shock leads to a larger increase in pressure. With the
shock reflection occurring farther downstream, flow acceleration leads to a larger pressure
jump over the shock.
86

Figure 7.22. Pressure coefficient over the rear wingbox farthest from the moving
wall with the wingboxes at different distances apart at =4( and
Cases).
87

Figure 7.23. Pressure coefficient along the moving wall with the wingboxes
farthest from the wall at different distances apart at =4( and
Cases).
88

CHAPTER 8

LIFT AND DRAG LOAD ANALYSIS ON TANDEM


WINGBOXES

Figure 8.1 shows the lift coefficient for the , , and Cases for two free

stream Mach numbers, = 2 and 4 that have been considered so far. For the front

wingbox, in red, as the Mach number decreases the amount of lift increases. As the ground
clearance increases the amount of lift decreases presenting a similar trend that is observed in
subsonic ground effect literature. However, the effect of the wake from the front wingbox
interacts with the rear wingbox. Similar to experiments discussed in Chapter 1, the distance
between the wingboxes does play a significant role on the loading of the rear wingbox. For a
smaller Mach number flow, the lift increases with increasing ground clearance, but for a
larger Mach number, the lift decreases with increasing ground clearance.
The tandem wingbox loads shows that the = 2 flow results in more lift than the
= 4 flow because the change in pressure is lower for the = 4 flow. The change in
pressure is lower because a swallowed shock is the result of the increased Mach number.
Oblique shock reflections form rather than normal shocks thereby limiting the losses in the
flow. For the smaller Mach number flow, a standing bow shock is upstream of the front
wingbox and a normal shock forms in the wake area between the wingboxes. This is
affecting the loads on the tandem wingboxes and results in more lift in the lower Mach
number flow. As increases, the pressure increases as well. However, due to the
difference in pressure loss, , between an oblique shock and a normal shock, the =
2 flow generates more lift and essentially more drag. For the = 2 flow, the maximum lift
occurs at the middle height from the wall due to the increase in lift from the rear wingbox.
Figure 8.2 presents the drag coefficient for the same cases. The smaller Mach number
also results in more drag than the larger Mach number due to pressure losses from the normal
shocks. Data in Figure 8.2 indicates that for the = 2 flow, drag and lift for front and rear
wingboxes are the reverse of one another. For the front wingbox, as the ground clearance
increases, the drag and lift decrease (Figure 8.1). Conversely, for the rear wingbox, as the
89

Figure 8.1. Lift coefficient versus the ground clearance for the front, rear,
and combined wingboxes at both = 2 and 4 where the wingboxes are at .

ground clearance increases, the drag and lift increase. The wake and the pressure increase
from the shear layer reattachment are affecting the loads on the rear wingbox. The drag on
the rear wingbox for the = 4 flow is similar to the lift in that as the ground clearance
increases the drag decreases.
Figure 8.3 presents the lifting loads on the ground with increasing ground clearance.
Both values have similar effects in that the lift on the ground increases as the ground
clearance increases. The increase in lift is more significant with a higher Mach number flow.
Also, negative lift on the ground is a result of a larger Mach number flow. Figure 8.4 shows
the drag coefficient on the wall in terms of ground clearance and free stream Mach number.
This figure shows that the drag is negligible.
The , , and Cases (Table 3.1) are examined in Figure 8.5.
The lift on the front wingbox for both = 2 and 4, is identical to the lift when the
wingboxes are at a closer distance (Figure 8.1). Since the pressure distribution for the front
wingbox are the same for distances and , the loads on the front wingbox are the same.
90

Figure 8.2. Drag coefficient versus the ground clearance for the front, rear, and
combined wingboxes at both = 2 and 4 where the wingboxes are at .

There is a difference in loads with the rear wingbox (Figure 8.5). For a smaller Mach
number, the lift on the rear wingbox increases with an increasing ground clearance. A larger
Mach number results in a maximum lift at and decreases at . The fluctuations in lift
for the individual wingboxes at = 2 cause the combined lift to be steady with increasing
ground clearance. At = 4, however, the lift decreases with increasing ground clearance.
Figure 8.6 shows the drag coefficient for the same cases. The front wingbox data is
the same as the cases in which the wingboxes are closer together. For the rear wingbox at a
smaller Mach number, more drag is the result of the pressure loss incurred from the bow
shock, and it increases with increasing ground clearance similar to the lift. For the larger
Mach number, the drag is at a maximum at and decreases at . The drag is greater on
the rear wingbox than the front wingbox due to the wake effects upon the rear wingbox.
While some cases have clear trends, others do not, indicating that the effect of distance
between the wingbox is an important factor on the loads on the rear wingbox.
91

Figure 8.3. Lift coefficient versus the ground clearance on the wall for
the = 2 and 4 flows, where the wingboxes are at .

Figure 8.4. Drag coefficient versus the ground clearance on the wall
for the = 2 and 4 flows, where the wingboxes are at .
92

Figure 8.5. Lift coefficient versus the ground clearance for the front, rear,
and combined wingboxes at both = 2 and 4 where the wingboxes are at
.

Figure 8.7 shows the lift on the ground with increasing ground clearance. At = 4,
negative lift is increasing as the ground clearance increases, while at = 2, minimum lift
occurs at . Figure 8.8 presents the drag coefficient and similarly indicates that the drag is
negligible. Some cases do follow the trend of decreasing lift with increasing ground
clearance, but it is also apparent that the wake area is affecting the loads on the rear wingbox.
The distance between the wingboxes does factor into the optimization of the performance.
The results also point out that a smaller ground clearance does not always result in the most
lift.
93

Figure 8.6. Drag coefficient versus the ground clearance for the front, rear,
and combined wingboxes at both = 2 and 4 where the wingboxes are at
.

Figure 8.7. Lift coefficient versus the ground clearance on the wall
for the = 2 and 4 flows, where the wingboxes are at .
94

Figure 8.8. Drag coefficient versus the ground clearance on the wall
for the = 2 and 4 flows, where the wingboxes are at .
95

CHAPTER 9

EXPERIMENTAL AND CFD DATA COMPARISON

In this Chapter, the CFD analysis is compared with the experimental results obtained
in the supersonic wind tunnel. The wind tunnel ground is necessarily stationary. Therefore,
a stationary wall is considered in the CFD computations. Also, the effect of a stationary wall
is compared to the effect of a moving wall from the previous chapters. Since the experiments
are performed with = 2, only this free stream Mach number is considered.

9.1 PRESSURE COMPARISON WITH EXPERIMENTAL


DATA
The pressure coefficient measurements between the experiment and computation are
compared. All of the experimental results used the absolute pressure, therefore the following
equation is used for the non-dimensionalization of the pressure taking into account the
different upstream pressure that was mentioned in Chapter 2.
. (9.1)

Figure 9.1 shows the pressure coefficient over the front wingbox along with the
experimental data for the and Cases. The experimental and computational data do

not match up well. Pressure port 1 from the experiment, located at a half an inch back from
the end of the front wingbox, is closer to the numerical results than pressure port 2, which is
located directly in the middle of the back end of the front wingbox. The experimental results
show an increase in pressure from port 1 to 2, while the computational results show a
decrease in pressure. In the computation, the flow expands around the bottom-rear corner of
the front wingbox leading to a decrease in pressure. Boundary layer separation at the top-
rear corner results in a pressure increase. The experimental results also show that the
pressure is lower when the wingboxes are at than when they are at . The computations

show that the pressure for these distances are the same, indicating that the distance between
the wingboxes does not affect the flow upstream for these two cases.
96

Figure 9.1. Pressure coefficient over the front wingbox closest to the
stationary wall with the wingboxes at different distances apart at =2
compared to the experimental results ( and Cases).

The reason for the difference may be found in the comparison of the CFD study's
Mach contour to the Schlieren image of the experiment. Figure 9.2 does show a similar flow
pattern to that of the wind tunnel experiment as seen in the Schlieren picture of the
experiment (Figure 9.3). The Schlieren image shows a detached bow shock upstream of the
front wingbox, which is preceded by an expansion and oblique shock [20]. Along the
stationary wall, a separation bubble is occurring at the shock impingement point on the
stationary wall. An induced oblique shock is a result of the separation bubble that reflects off
of the bottom surface of the front wingbox and continues to reflect off of the stationary wall.
These shock reflections are not observed in the computational result. Figure 9.3 shows that
port 1 is close to the shock reflection off of the front wingbox, which could be a possible
reason for the difference.
97

Figure 9.2. Mach contour of the Baseline Case, , with a stationary wall at =2

Figure 9.3. Schlieren image of the Case from the wind tunnel
experiment. Source: Ngo, Trieu, Andres Cedeno, and Jeri Perez.
Supersonic Ground Effects. San Diego, CA: San Diego State
University, 2009.

Differences are observed in the wake area between the tandem wingboxes. This is
attributed to the effect of the two metal rods that necessarily connect two wingboxes in the
experiment. Pressure port 2 is located in the area between the two metal rods. This
significantly impacts where the flow can travel, and its behavior. This pressure port is
registering a higher pressure than in the CFD results. The velocity of the flow around port 2
in the experiment could be much smaller than that of the computation because of the rods.
Figure 9.2 from CFD shows that the velocity is larger in the middle of the back-end of the
front wingbox where port 2 is located due to the recirculation in the wake. The recirculation
in wake area for the experiment is hindered by the rods, possibly causing the flow to
decelerate much more leading to a higher pressure.
98

Figure 9.4 compares the computational pressure coefficient over the rear wingbox
with the pressure measurements at port 3 and 4 located on the bottom surface of the rear
wingbox in the experiment. The experiment shows a reasonable comparison at port 3 on the
rear wingbox. At the port 4, the difference between the experiment and computation are
larger. The results do show a similar trend in that the pressure decreases between port 3 and
4. The experimental results also show that the larger pressure results when the wingboxes
are at which is opposite of the computational results. The Schlieren images of these cases

in the experiment do not clearly show what is happening in the flow over the rear wingbox.
Port 3 is located in the area where the shear layers reattach, which can be the cause of
significant differences depending on how the rods in the experiment affect reattachment. For
these two cases, there is a comparison within the tolerance at port 3. For port 4, the pressure
is decreasing because the flow is accelerating in the computation. It is unclear from the
experiment as to why the pressure does not decrease as much. If there are shock reflections
under the rear wingbox as well, this could explain why the flow does not accelerate as much
leading to less of a pressure drop.
Figures 9.5 and 9.6 compare the computational pressure to the experimental pressure
on the front and rear wingboxes for the and Cases. Pressure port 1 compares
very well, while pressure port 2 is within reasonable tolerance errors. For the experiment, the
pressure for the larger S increases at port 2, while the pressure for the smaller S decreases,
while in the previous cases, each S case's pressure increased at port 2. The increase in wake
area affects the recirculation which results in a pressure change. However, the pressure
remains higher than the computational analysis indicating that the velocity of the
recirculation is smaller in the experiment. It is likely that the rods are the cause for the
difference in pressure in the wake area. Pressure ports 3 and 4 on the rear wingbox of Figure
9.6 end any sense of a trend developing. In the experiment, the pressure from port 3 to 4 is
decreasing, which is the opposite of the trend of the computational data and the trend of the
previous cases. The smaller S results in a lower pressure than the larger S, while the opposite
occurs in the computation. The rods are interfering with the reattachment of the shear layers
of the wake onto the rear wingbox. Pressure port 3 happens to be located around the
significant area where the shear layer reattach to the rear wingbox that has been discussed
throughout the paper.
99

Figure 9.4. Pressure coefficient over the rear wingbox closest to the
stationary wall with the wingboxes at different distances apart at =2
compared to the experimental results ( and Cases).

Figures 9.7 and 9.8 compares the experimental pressure and the computational
pressure on the front and rear wingboxes for the and Cases. The results show a
similar trend to the and Cases. Pressure port 1 matches well to the numerical
results, while port 2's data is within a reasonable tolerance. In the experiment, the pressure is
increasing from port 1 to 2, with the smaller S resulting in a lower pressure than the larger S.
From to , the pressure at port 2 is registering similar values, .25 and .3 for a smaller S,
and .38 and .375 for a larger S. This indicates that the increase in ground clearance is not
affecting the pressure in the wake as much as the cases. Port 3 in Figure 9.8 shows a
difference from the numerical results, while port 4 is closer in value. The pressure is
decreasing from port 3 to 4, with a smaller S resulting in a lower pressure than the larger S,
similar to the trends of the previous cases. For most cases in the experiment, the pressure is
greater for a larger S over the rear wingbox. However, in the computation, the pressure is
greater for a smaller S over the rear wingbox.
100

Figure 9.5. Pressure coefficient over the front wingbox at the middle height
above the stationary wall with the wingboxes at different distances apart at
= 2 compared to the experimental results ( and Cases).

In general, it is difficult to draw conclusions from the comparison between


computations and experiments. It is difficult to determine from the few cases that are
compared whether the matchup is real or coincidental. The rods that are holding the two
wingboxes together in the experiment are major factors in this discrepancy. During the
experiment, a problem arose when the rods that were used to hold the wingboxes together
were much smaller in diameter. The smaller rods were not stable enough to withstand the
force of the supersonic wind. Because of this, larger rods had to be installed. Any size rod
disrupts the integrity of the flow pattern in the wake area. The rods limits the flow and
change how and where the flow reattaches to the rear wingbox. Pressure port 3 is in this
sensitive location.

9.2 EFFECT OF A STATIONARY WALL VERSUS A MOVING


WALL
A stationary wall is used in the experiment, but a moving ground is required in the
modeling of the MagLev. The team conducting the experiment decided not to use the mirror
image technique in order to simulate a moving ground, therefore a stationary wall was used
101

Figure 9.6. Pressure coefficient over the rear wingbox at the middle height
above the stationary wall with the wingboxes at different distances apart at
= 2 compared to the experimental results ( and Cases).

[20]. A moving wall is going to produce more accurate results, and when testing in a wind
tunnel, the effect of the moving wall is desired as pointed out by the literature in Chapter 1
[2]. To establish the magnitude of the effect of the moving wall, a computational
investigation is conducted in Fluent.
The general flow features of a case with a moving wall and a case with a stationary
wall are very similar (Figure 9.9 and 9.10). There are however, subtle but important
differences. A moving wall combats the growth of a boundary layer along the bottom wall.
If the moving wall has the same velocity as the oncoming flow, then there will be no
boundary layer at all. For a stationary wall, a boundary layer is growing along the bottom
wall that separates from the ground near the aft wingbox. A separation bubble forms at that
location. The adverse pressure gradient along with zero to negative shear stress causes this
separation from the stationary wall. The separation bubble shows a recirculation that is seen
102

Figure 9.7. Pressure coefficient over the front wingbox farthest from the
stationary wall with the wingboxes at different distances apart at =2
compared to the experimental results ( and Cases).

in the pathline contour of Figure 9.11. The negative shear pushes the flow back upstream
creating the recirculation. The flow slows down due to the normal shock just before the
separation bubble and the boundary layer reattaches to the stationary wall.
In both cases, a normal shock forms before the rear wingbox that induces a small total
pressure loss of around 5% (Figure 9.12 and 9.13). Based on the overspeeding discussion, the
shock should be swallowed into the theoretical intake. The way the normal shock is
positioned in the = 2 cases being discussed, the shock is essentially standing at the inlet
of the theoretical intake. According to theory, as the pressure decreases in the intake, the
normal shock, will be swallowed and pushed into the intake. Following the normal shock,
there is an initial increase in the pressure and a decrease in velocity. As the flow continues
through the intake, the flow accelerates, therefore decreasing the pressure continually until
the expansion at the end of the rear wingbox. Figure 9.14 shows the pressure is decreasing.
103

Figure 9.8. Pressure coefficient over the rear wingbox farthest from the stationary
wall with the wingboxes at different distances apart at = 2 compared to the
experimental results ( and Cases).

Figure 9.9. Mach contour of the Case with a moving wall at = 2.


104

Figure 9.10. Mach contour of the Case with a stationary wall at = 2.

Figure 9.11. Pathline contour of the Case with a stationary wall at = 2.

Figure 9.12. Total pressure contour of the Case with a moving wall at = 2.
105

Figure 9.13. Total pressure contour of the Case with a stationary wall at = 2.

Figure 9.14. Pressure contour of the Case with a moving wall at = 2.

This pressure decrease should allow for the shock to be swallowed, but that is not the case.
Hence, there is a balance of pressure that is occurring to keep the shock standing at the
theoretical intake inlet. Referring back to the literature found on oblique and Mach
reflections, the stem of a Mach reflection is a normal shock. It moves up or downstream
based on the pressure. Ben-Dor et al. [12], researched how shock configurations are affected
by the pressure downstream through numerical and analytical investigations. Using a
reflecting wedge in their research, they found that if the downstream pressure, defined as the
wake pressure, is greater than or equal to the pressure of the expansion at the corner at the
back end of the wedge, then the Mach reflection will be independent of the downstream
conditions and will not move up or downstream. This idea is a possible explanation on why
106

the normal shock standing at the inlet of the intake is not swallowed. Depending on where
the downstream pressure is taken in the wake, the pressure after the expansion is close to the
downstream pressure (Figure 9.14).
107

CHAPTER 10

CONCLUSION

CFD analysis was performed on a tandem wingbox configuration in accordance with


the MagLev test rail fabricated and designed by General Atomics and the Air Force.
Computational simulations commenced on several cases with respect to various ground
clearances, H, and distance between the wingboxes, S, at a free stream Mach number of 2
and 4. The CFD analysis was also used to compare results to the experimental wind tunnel
testing on the tandem wingbox configuration. Effect of the Mach number, effect of ground
clearance and distance between the wingboxes, and lift and drag loads were the parameters
used to examine supersonic ground effect.
Grid convergence studies showed that while resolution was reached for the front
wingbox, the rear wingbox remained unresolved in the location of reattachment of the shear
layers. As the grid refined, the fluctuations in this area did not indicate a clear direction for
resolution to warrant further refinement. When the Mach number increased to 4, the
detached shock was swallowed at all values of ground clearance. Increasing the Mach
number to 2.5 still showed the shock being pushed upstream, even at the farthest ground
clearance, indicating that the shock is swallowed between a Mach number of 2.5 and 4. The
ground clearances being tested were not large enough for the lower Mach number flow to
swallow the shock. A swallowed oblique shock is the ideal case in order to limit the losses
over the shocks.
Ground clearance studies for the lower Mach number flow showed that as the ground
clearance decreases the pressure increases along the front wingbox. This translated to an
increase in lift and drag as the ground clearance decreased. The rear wingbox's lift and drag
were affected by the wake effects from the front wingbox. As the ground clearance
increased, the lift and drag on the rear wingbox increased, opposite of the trend of the front
wingbox. The lift on the wall acting as the ground increased with increasing ground
clearance as well. For the higher Mach number flow, the front wingbox showed an increase
in lift with a decrease in ground clearance. However, the drag decreased with a decrease in
108

ground clearance. The rear wingbox showed a decrease in lift and drag as the ground
clearance increased. A significant force was produced along the wall, with the drag being
negligible.
When the wingboxes were farther apart, only slight changes were observed on the
front wingbox such as separation of the boundary layer at the top-end corner. Therefore, the
lift and drag were the same as when the wingboxes were closer together. The distance
between the wingboxes, S, had the most effect on the rear wingbox. The wake from the front
wingbox affected the reattachment location of the shear layers on the rear wingbox, as well
as the magnitude of the pressure increase at that point. As the ground clearance increased the
pressure at this location increased as well. In terms of loads, at = 2, the lift on the rear

wingbox increased with increasing ground clearance. However, at = 4, the lift

maximized at and decreased by 50 percent at . For the drag, the maximum occurred at

as well. The distance between the wingboxes does affect the loads on the rear wingbox

due to the wake effects. This research enhances and extends the evidence in the literature
cited in Chapter 1.
The experimental results did not match well with the CFD analysis. The rods and
hinge were not taken into account in the CFD simulation. As a result, the rods significantly
affected the flow behavior in the wake area. The parameter most impacted was the pressure
difference in the wake area, and the flow coming into the rear wingbox leading to a
difference in pressure at the reattachment location.
Future work on this project should address the effect of the distance between the
wingboxes. The actual Maglev test rail can be referenced with respect to S and used in
numerical simulations for supersonic ground effect analysis. Also, for better experimental
comparison, the wind tunnel experiment setup, including the hinge and rods, can be
incorporated into the CFD analysis in order to determine whether or not the rods are the
reason for the difference in pressure. Also, the concept of an ideal ratio of ground clearance
to distance between the wingboxes that maximizes the performance of the tandem wingbox
configuration can be addressed. A possible configuration, placing the wingboxes far enough
apart so that the flow from the front wingbox does not affect the rear wingbox, would lead to
similar flow characteristics on both wingboxes and perhaps better performance. However,
109

there might be an optimum distance at which the rear wingbox benefits from the flow coming
off of the front wingbox, which in turn optimizes the tandem wingbox as a whole. While
there are some issues regarding certain aspects of this research, a CFD analysis of the flow
field of the tandem wingbox configuration proved to be a useful tool to further an
understanding of the complexities of the developing topic of supersonic ground effect.
110

REFERENCES

[1] Hsu, Y., D. Ketchen, L. Holland, A. Langhorn, D. Minto, and D. Doll. Status of the
Magnetic Levitation Upgrade to Holloman High Speed Test Track. IEEE
Transactions on Applied Superconductivity No. 29 (2008): 2074-2077.
http://maglev.com.www65.yourserver.de/uploads/2008/MAGLEV_Development_Pro
jects_etc/holland_langhorn_minto_doll__status_of_the_magnetiv_levitation_upgrade
_to_holloman_high_speed_test_track.pdf.
[2] Barber, T. J., E. Leonardi, and R. D. Archer. "Causes for Discrepancies in Ground
Effect Analysis." The Aeronautical Journal 1066 (2002): 653-657.
[3] Tomotika, S., T. Nagamiya, and Y. Takenouti. "The Lift on a Flat Plate Placed Near a
Plane Wall, with Special Reference to the Effect of the Ground Upon the Lift of a
Monoplane Aerofoil." Aeronautical Research Institute 97 (1933): 1-60.
http://airex.tksc.jaxa.jp/dr/prc/japan/contents/IS4147494000/IS4147494.pdf.
[4] Ahmed, M. R., T. Takasaki, and Y. Kohama. "Aerodynamics of a NACA 4412
Airfoil in Ground Effect." The American Institute of Aeronautics and Astronautics
45.1 (2007): 37-47.
[5] Coulliette, C., and A. Plotkin. "Aerofoil Ground Effect Revisited." The Aeronautical
Journal 100.992 (1996): 65-74.
[6] Rozhdestvensky, Kirill V. Aerodynamics of a Lifting System in Extreme Ground
Effect. Dusseldorf, Germany: Springer-Verlag, 2000.
[7] Prykhodko, Olexander A., Anatoliy V. Sohatsky, Oleg B. Polevoy, and Andrey V.
Mendriy. "Computational and Wind Tunnel Experiment in High-Speed Ground
Vehicle Aerodynamics." (2006): 1-6. http://www.maglev2006.de/118_Prykhodko/
118_Prykhodko_ok.pdf.
[8] Torda, T. P., and Thomas A. Morel. "Aerodynamic of a Land Speed Record Car."
Journal of Aircraft 8.12 (1971): 1029-1033.
[9] Slater, John W. "Verification Assessment of Flow Boundary Conditions for CFD
Analysis of Supersonic Inlet Flows. Paper presented at the 37th Joint Propulsion
Conference and Exhibit, Salt Lake City, Utah, July 8-11, 2001.
[10] Ferri, Antonio, and Louis M. Nucci. "Preliminary Investigation of a New Type of
Supersonic Inlet." National Advisory Committee for Aeronautics TN 2286 (1951): 1-
41. http://www.dtic.mil/cgibin/GetTRDoc?AD=ADA382078&Location=
U2&doc=GetTRDoc.pdf.
[11] Moase, William H., Michael J. Brear, and Chris Manzie. "The Forced Response of
Choked Nozzles and Supersonic Diffusers." Journal of Fluid Mechanics 585 (2007):
281-304. http://proquest.umi.com.libproxy.sdsu.edu/pqdweb?index=18&did=
1453968581&SrchMode=3&sid=1&Fmt=6&VInst=PROD&VType=PQD&RQT=30
9&VName=PQD&TS=1292181120&clientId=17862&aid=1.
111

[12] Ben-Dor, G., T. Elperin, H. Li, and E. Vasiliev. "The Influence of the Downstream
Pressure on the Shock Wave Reflection Phenomenon in Steady Flows." Journal of
Fluid Mechanics 386 (1999): 213-232. http://www.bgu.ac.il/me/staff/
tov/papers/p213_1.pdf.
[13] Alber, Irwin E. "Integral Theory for Turbulent Base Flows At Subsonic and
Supersonic Speeds." Thesis, California Institute of Technology, 1967.
[14] Seddon, J. "The Flow Produced By Interaction of a Turbulent Boundary Layer with a
Normal Shock Wave of Strength Sufficient to Cause Separation." Ministry of
Technology Aeronautical Research Council R&M No. 3502. London, England: Her
Majestys Stationary Office, 1967. http://naca.central.cranfield.ac.uk/
reports/arc/rm/3502.pdf.
[15] Zang, Thomas A., M. Y. Hussaini, and Dennis M. Bushnell. "Numerical
Computations of Turbulence Amplification in Shock-Wave Interactions." The
American Institute of Aeronautics and Astronautics 22.1 (1984): 13-21.
[16] Wilcox, David C. Turbulence Modeling for CFD. 3rd ed. La Caada, CA: DCW
Industries, 2006.
[17] Yegorov, I. V., M. V. Yegorova, and V. V. Riabov. "Analysis of Hypersonic Viscous
Flow About Bluff Cylinders Placed One After Another." The American Institute of
Aeronuatics and Astronautics (1998): 1-12. http://www.rivier.edu/faculty/vriabov/
Aiaa_98-0171-16395506.pdf.
[18] Karpov, Yu. L., Yu. P. Semenkevich, and A. Ya. Cherkez. "Calculation of Separated
Flow Between Two Bodies." Mekhanika Zhidkosti I Gaza 3.3 (1968): 58-62.
http://www.springerlink.com.libproxy.sdsu.edu/content/l0lk7205q91t8056/fulltext.pd
f.
[19] Hilliard, John C., and George S. Springer. Fuel Economy In Road Vehicles Powered
By Spark Ignition Engines. New York: Plenum Press, 1984.
[20] Ngo, Trieu, Andres Cedeno, and Jeri Perez. Supersonic Ground Effects. San Diego,
CA: San Diego State University, 2009.
[21] ANSYS Fluent:Theory Guide. Canonsburg, PA: ANSYS, Inc., 2009. CD-ROM.
Version 12.0.
[22] M., Jade. "Use of k-epsilon and k-omega models." 28 Apr. 2010. CFD Online. Last
modified January 23, 2011 http://www.cfd-online.com/Forums/main/75554-use-k-
epsilon k-omega-models.html.
[23] ANSYS Fluent: User's Guide. Canonsburg, PA: ANSYS, Inc., 2009. CD-ROM.
Version 12.0.
[24] Anderson, John D. Fundamentals of Aerodynamics. 4th ed. New York, NY: McGraw-
Hill, 2007.
[25] Hill, Philip, and Carl Peterson. Mechanics and Thermodynamics of Propulsion. 2nd
ed. Indianapolis, IN: Addison Wesley Longman, 1992.
112

APPENDIX

CDF CONTOURS FOR DIFFERENT


CONFIGURATIONS
113

Figure A.1. Mach contour of the Case with a moving wall at = 2.

Figure A.2. Absolute pressure contour of the Case with a moving wall at = 2.

Figure A.3. Mach contour of the Case with a moving wall at = 4.


114

Figure A.4. Absolute pressure contour of the Case with a moving wall at = 4.

Figure A.5. Mach contour of the Case with a moving wall at = 2.

Figure A.6. Absolute pressure contour of the Case with a moving wall at = 2.
115

Figure A.7. Mach contour of the Case with a moving wall at = 4.

Figure A.8. Absolute pressure contour of the Case with a moving wall at = 4.

Figure A.9. Mach contour of the Case with a moving wall at = 2.


116

Figure A.10. Absolute pressure contour of the Case with a moving wall at = 2.

Figure A.11. Mach contour of the Case with a moving wall at = 4.

Figure A.12. Absolute pressure contour of the Case with a moving wall at = 4.
117

Figure A.13. Mach contour of the Case with a moving wall at = 2.

Figure A.14. Absolute pressure contour of the Case with a moving wall at = 2.

Figure A.15. Mach contour of the Case with a moving wall at = 4.


118

Figure A.16. Absolute pressure contour of the Case with a moving wall at =4

Você também pode gostar