Vous êtes sur la page 1sur 82

DESIGN, CALIBRATION AND TESTING OF A FORCE BALANCE FOR

A HYPERSONIC SHOCK TUNNEL

by

PRAVIN VADASSERY

Presented to the Faculty of the Graduate School of


The University of Texas at Arlington in Partial Fulfillment
of the Requirements
for the Degree of

MASTER OF SCIENCE IN AEROSPACE ENGINEERING

THE UNIVERSITY OF TEXAS AT ARLINGTON


MAY 2012

Copyright by Pravin Vadassery 2012


All Rights Reserved

ACKNOWLEDGEMENTS
Foremost, I am thankful to God for having blessed me throughout my life, without whom
nothing is possible. Next thanks go to Dr Frank Lu and Dr Don Wilson for their constant support
and for giving me the opportunity to work at the ARC (Aerodynamics Research Center). Again, I
am thankful to Dr Lu for his determination, enthusiasm and vast knowledge. His words of
encouragement, Making mistakes is all part of the learning process, helped me to overcome
the hardships during my research. A special thanks to Eric M Braun for his help, quick
suggestions and for always being around.
I acknowledge my fellow team mates in doing an excellent job of reconstructing the
UTA Hypersonic Shock Tunnel and getting it back on running condition. Special thanks go to
Tiago Rolim for his endless support and always assisting me in the times of repair, machining
and discussions. Thanks also to Derek Leamon, Nitesh K Manjunatha, Raheem Bello and
Dibesh Joshi.
I appreciate the work of all the technical staff involved in the Mechanical and Aerospace
Department. Special credit to Kermit Beird, Sam Williams and Rod Duke for fabrication of all
necessary parts and for sharing their practical knowledge. I sincerely thank everyone in the
ARC, also for making this place lively and loud.
Finally, I would like to thank my parents, family and friends for their patience and for
supporting me.
April 17, 2012

iii

ABSTRACT
DESIGN, CALIBRATION AND TESTING OF A FORCE BALANCE FOR
A HYPERSONIC SHOCK TUNNEL

Pravin Vadassery, M.S


The University of Texas at Arlington, 2012
Supervising Professor: Frank K. Lu
The forces acting on a flight vehicle are critical for determining its performance. Of
particular interest is the hypersonic regime. Force measurements are much more complex in
hypersonic flows, where those speeds are simulated in shock tunnels. A force balance for such
facilities contains sensitive gages that measure stress waves and ultimately determine the
different components of force acting on the model. An external force balance was designed and
fabricated for the UTA Hypersonic shock tunnel to measure drag at Mach 10. Static and
dynamic calibrations were performed to find the transfer function of the system. Forces were
recovered using a deconvolution procedure. To validate the force balance, experiments were
conducted on a blunt cone. The measured forces were compared to Newtonian theory.

iv

TABLE OF CONTENTS
ACKNOWLEDGEMENTS ................................................................................................................iii
ABSTRACT ..................................................................................................................................... iv
LIST OF ILLUSTRATIONS..............................................................................................................vii
LIST OF TABLES ............................................................................................................................. x
Chapter

Page
1. INTRODUCTION ........................................... 1
1.1 Literature Survey .............................................................................................. 1
1.2 Force Measurement Techniques ..................................................................... 2
1.2.1 Internal Force Balance ..................................................................... 3
1.2.2 External Force Balance .................................................................... 3
1.2.3 Strain Gages .................................................................................... 4
1.2.4 Piezoelectric Film ............................................................................. 4
1.2.5 Accelerometer .................................................................................. 5
1.3 Convolution ...................................................................................................... 5
1.4 Objective of Research ...................................................................................... 7
2. FACILITY........................................................................................................................ 8
2.1 UTA Hypersonic Shock Tunnel at the Aerodynamics
Research Center .......................................................................................... 8
2.2 Reconstruction of the UTA Hypersonic Shock
Tunnel ........................................................................................................ 13
2.3 Diaphragm Test .............................................................................................. 13
3. DESIGN AND EXPERIMENTAL SETUP .................................................................... 15
3.1 Force Balance Design .................................................................................... 15
v

3.1.1 Finite Element Analysis ................................................................. 18


3.1.2 Force Balance Construction .......................................................... 24
3.2 Calibration Technique .................................................................................... 29
3.2.1 Static Calibration ............................................................................ 29
3.2.2 Dynamic Calibration ...................................................................... 32
3.3 Shock Tunnel Testing ................................................................................... 42
4. RESULTS AND DISCUSSION .................................................................................... 43
4.1 Force Measurement Prediction ..................................................................... 43
4.1.1 Modified Newtonian Theory .......................................................... 43
4.1.2 Coefficient of Drag Calculation using
Pitot Pressure .......................................................................... 46
4.2 Experimental Results ..................................................................................... 48
5. CONCLUSION AND FUTURE WORK ........................................................................ 53
5.1 Force Balance in the UTA Hypersonic Shock Tunnel ................................... 53
5.2 Future Work and Recommendations ............................................................ 55

APPENDIX
A. LIST OF DESIGN DRAWINGS..................................................................................... 56
B. MATLAB PROGRAM FOR FORCE ESTIMATION ..................................................... 63
C. INSTRUMENTATION DETAILS .................................................................................. 67
REFERENCES ............................................................................................................................... 70
BIOGRAPHICAL INFORMATION .................................................................................................. 72

vi

LIST OF ILLUSTRATIONS
Figure

Page

1.1 Linear input-output system (a) continuous (b) discrete .............................................................. 6


1.2 Convolution in time and frequency domain ................................................................................ 6
2.1 Schematic of the UTA Hypersonic Shock Tunnel ..................................................................... 9
2.2 Panorama view of the UTA Hypersonic Shock Tunnel .............................................................. 9
2.3 Schematic of the double diaphragm section ........................................................................... 10
2.4 Photograph of double diaphragm section ............................................................................... 10
2.5 Steel diaphragms
(a) scored diaphragm (b) ruptured diaphragm after test ........................................................... 14
3.1 Different preliminary designs .................................................................................................... 17
3.2 Fabricated force balance .......................................................................................................... 18
3.3 Generated mesh of the force balance ...................................................................................... 19
3.4 FEA analysis settings ............................................................................................................... 20
3.5 Strain concentration in stress bars .......................................................................................... 20
3.6 Simulated input load of 350 N .................................................................................................. 21
3.7 Response to simulated impulse at
(a) location1 (b) location2 ......................................................................................................... 22
3.8 (a) Simulated step load of 222.4 N (b) step response of location 2 ........................................ 22
3.9 Animated result of stress wave propagation ............................................................................ 23
3.10 Blunt cone model (a) side view (b) front view ........................................................................ 24
3.11 Hardened steel bolt hinge ..................................................................................................... 25
3.12 Installed model and balance in the test section .................................................................... 26
3.13 Attached strain gages ........................................................................................................... 27
3.14 Installed model and balance in the test section, front view .................................................... 28
vii

3.15 Schematic of static calibration procedure ............................................................................. 29


3.16 Static loading and unloading of force balance ...................................................................... 30
3.17 Average film output versus hammer force ............................................................................ 32
3.18 Schematic of a cut weight test .............................................................................................. 33
3.19 Vertical cut weight test .......................................................................................................... 33
3.20 Schematic of impulse hammer calibration ............................................................................ 34
3.21 Raw data of hammer impulse test.......................................................................................... 35
3.22 Sample hammer impulse ...................................................................................................... 35
3.23 Check signal for both raw and modified hammer pulse ........................................................ 36
3.24 Detail view of check signal with error bar ............................................................................... 37
3.25 Simulated unit step input ....................................................................................................... 37
3.26 Modified hammer signal ......................................................................................................... 38
3.27 Enlarged view of the modified hammer pulse ....................................................................... 38
3.28 Impulse response obtained from FFT and JMECG .............................................................. 39
3.29 Power spectral density plot of FRF ....................................................................................... 40
3.30 Enlarged power spectral density plot of FRF for first 12 kHz ................................................ 40
3.31 Enlarged phase spectrum of FRF ......................................................................................... 41
3.32 Spectrogram of the FRF ........................................................................................................ 41
4.1 Plot of (a) coefficient of drag (b) coefficient of lift .................................................................... 45
4.2 Coefficient of drag from recovered force (condition 1) ............................................................ 48
4.3 Recovered drag force and predicted force .............................................................................. 49
4.4 Raw pitot pressure signal ....................................................................................................... 49
4.5 Detailed view of the pitot pressure signal and drag ................................................................. 50
4.6 Coefficient of drag from recovered force (condition 2) ............................................................. 50
4.7 Recovered drag force .............................................................................................................. 51
4.8 Raw pitot pressure signal ........................................................................................................ 51
viii

4.9 Detailed view of the pitot pressure signal ................................................................................ 52


A.1 Force balance drawing ............................................................................................................ 57
A.2 Blunt cone model drawing ...................................................................................................... 58
A.3 PCB pressure transducer holder drawing ............................................................................... 59
A.4 Hinge joint part 1 drawing ....................................................................................................... 60
A.5 Hinge joint part 2 drawing ....................................................................................................... 61
A.6 Scoring pattern on steel diaphragm drawing .......................................................................... 62
C.1 Amplifier circuit diagram for piezoelectric film ......................................................................... 69

ix

LIST OF TABLES
Table

Page

2.1 Rupture properties of diaphragm tests .................................................................................... 14


3.1 Properties of some metals/alloys ............................................................................................ 16
3.2 Static calibration results .......................................................................................................... 31
3.3 Test condition .......................................................................................................................... 42
4.1 Force prediction using modified Newtonian theory condition 1 .............................................. 46
4.2 Force prediction using modified Newtonian theory condition 2 .............................................. 46
4.3 Comparison of experimental to theoretical drag ..................................................................... 52

CHAPTER 1
INTRODUCTION
The forces acting on a flight vehicle are critical for determining its performance. Of
particular interest is the hypersonic regime. Research in hypersonics has led to successful tests
of scramjet (supersonic combustion ramjet) based vehicles, such as, NASA X-43A. In
hypersonic vehicle design, significance is laid on propulsion system integration, engine
performance, aerodynamics and thrust measurements. Specifically ground-based test facilities
have limited steady test time thus making force measurement complex in this very short
duration of time.
1.1 Literature Survey
Hypersonic wind tunnels have been in use since the 1950s and have
developed into different types, namely, continuous and impulse types. Impulse facilities include
shock tubes, reflected shock tunnels and expansion tunnels. The basic principle of these
impulse facilities is to suddenly release a highly compressed gas in the so-called driver tube
through rupturing a diaphragm. The sudden release of the compressed gas propagates a shock
wave into a so-called driven tube filled with the test gas at low pressure.

The shock

compresses and heats the test gas to the desired conditions, after which, it is expelled by a
nozzle to hypersonic conditions. For example, the T4 free piston shock tunnel at the University
of Queensland is an impulse-type facility that simulates hypersonic flows [1].

1.2 Force Measurement Technique


Throughout this thesis, force measurements refer to techniques for impulse
facilities unless noted otherwise. Force measurement is complicated in impulse facilities due to
the short test duration that will likely prevent the force balance from attaining a steady state.
This limitation of short test times in such facilities was overcome by applying the stress wave
force measurement technique (SWFM), proposed by Sanderson and Simmons [2]. Due to
impulsive aerodynamic loading, stress waves that are created, propagate and reflect through
the model and support structure, which are measured and analyzed by this method. Extension
of this work by Daniel and Mee [3] using finite element modeling led to the design of a threecomponent force balance. The SWFM technique is based on the principle that, when stress
waves travels, no force equilibrium is reached in such a short duration of time so that the strain
histories are the crucial feature for developing force measurement techniques.
An investigation into internal and external force balances was undertaken by Robinson
et al. [4] which showed that a higher accuracy of the recovered force and moment loads was
attained using an external force balance. Also, for a blunt body, these authors found that the
interaction of external balance on the model forces was negligible when compared to that of an
internal balance. Some of the recent developments include comparing the experimental
measurements with CFD calculations by Boyce and Stumvoll [5], which showed good
agreement for a range of Mach numbers and test gases.
On the other hand, accelerometer-based force balance were used by Kulkarni and
Reddy [6] and Sahoo et al [7], which was a single-component accelerometer force balance. The
data were in accordance with modified Newtonian theory. Sahoo et al. [8], found that the drag
measured on a 30 degree semi-apex angle blunt cone model at Mach 5.75 with an acceleratorbased balance agreed closely to the SWFM technique.

1.2.1 Internal Force Balance


Internal force balances are defined as those that have the measuring instruments like
strain gages, accelerometers placed inside the model. The mounting system (sting) has to
adapt depending on the location of the force balance. Models are generally attached to a long
sting and placed in the test section of the tunnel. The geometry of the model has to
accommodate the sting. A common way to measure forces is by strain gages. Strain gages
work on the principle that when a load is applied, the stretching or deformation of the gage
causes a change in electrical resistance, details can be found in Section 1.2.3.
1.2.2 External Force Balance
External balances are those where the measuring instruments are located outside the
model but may be within the test section. The definition of external balances used to be
restricted those that are mounted outside the test section, which has been updated to balances
that are specifically external to the model, but which can be within the test section. The principle
of external balances is similar to that of internal balances but the difference is that the
measuring devices are placed on a supporting structure, such as a sting. The forces on the
model are transmitted as stress waves to the sting, which on deformation or bending creates
strain that is measured by the attached strain gages. Specifically for hypersonic shock tunnels,
external force balances that use stress wave propagation are named as Stress Wave Force
Balances (SWFB) [2].
During a run in an impulse facility, the sudden aerodynamic load initiates stress waves
in the model. The stress waves propagate and reflect between the model and the support
structure. A steady state of force equilibrium cannot be achieved between the model and the
support structure, since the duration of steady flow time is very minute. The SWFB concept
operates on the principle that no steady-state force equilibrium is achieved [9]. The force
balance forms a linear system, where the forces can be obtained by a deconvolution technique,
which is discussed further in a later chapter. A SWFB is suspended from the test section by
3

means of thin wires, with the strain gages mounted on the supporting sting. Different kinds of
materials have been used for SWFB construction such as brass, aluminum or steel.
1.2.3 Strain Gages
Strain gages are sensors that are used to measure strain or deformation. Strain gages
work by the principle that a strain in a metal or semi-conductor causes a change in resistance,
which when measured can be related to the strain. There are different types of strain gages,
namely, metallic foil gages and semiconductor strain gages, which can be either piezo-resistive
or piezoelectric. All resistance-based strain gages require an excitation voltage. A Wheatstone
bridge arrangement increases the sensitivity of the strain gage, thus allowing small changes in
strain to be measured.
1.2.4 Piezoelectric Film
Piezoelectric film gages are a type of transducer for measuring dynamic strain and are
used in high-frequency applications .Some of the features of piezo film are its flexibility, varying
thickness, lightweight and easy application. These gages have a large frequency range of up to
the order of 1 GHz. Some other properties include its dynamic range, high mechanical strength,
and temperature and humidity stability. Piezoelectric film can be adapted to various shapes and
can be bonded with commercial adhesives. Another feature is high voltage output, that can be
as high as 10 times higher than normal strain gages. Some disadvantages of piezoelectric films
are that they are sensitive to electromagnetic radiation, in such cases shielding becomes
important to avoid any kind of interference and ensuring a good signal-to-noise ratio.
Piezoelectric film gages do not require an external power source or excitation voltage.
Marineau [11] showed that a piezoelectric force balance has a higher frequency response than
a strain gage force balance. Both balances showed comparable levels of accuracy. The
piezoelectric balance shows a 350% increase in frequency response and 400% increase in
sensitivity.

1.2.5 Accelerometer
An accelerometer as the name suggests is a device that measures acceleration of an
object. It measures the rate of change of the velocity of the object relative to an inertial frame of
reference. The most common measuring unit is g. An accelerometer can also measure a
quantity of weight per unit mass (test mass), which has the dimensions of acceleration and is
also known as the g-force. Accelerometers are used in force measurements due to their high
sensitivity to vibrations and their high-frequency range.
1.3 Convolution
Convolution is mathematically an operation that involves multiplication, shifting and
addition. The reverse operation called deconvolution is used to calculate the input signal, when
the system's impulse response and its output signal are known. It can be difficult to understand
the convolution and deconvolution concepts in the time domain. More often, deconvolution is
carried out in the frequency domain. Multiplication in the frequency domain is equivalent to the
convolution operation in the time domain and likewise division in frequency domain acts like the
deconvolution operation in the time domain.
The expression for convolution is given by the formula

(1.1)
(1.2)

where, y(t) is the output of the system , x(t) input to the system and h(t) is the transfer function
of the system. Convolution expressed in both continuous and discrete form is represented in
figure 1.1.

x(t)

Linearsystem
h(t)

y(t)

(a)

x(n)

Linearsystem
h(n)

y(n)

(b)

Figure 1.1 A linear input-output system a) continuous, b) discrete.


Due to the large number of multiplications and additions that must be performed in the
convolution algorithm, it can be inefficient when a large amount of data needs to be processed.
As stated above, the convolution could be made easy by multiplication in the frequency domain
via Fourier and inverse Fourier transforms, represented in equation (1.3).

(1.3)

A block diagram showing the input-output relationship in the time and frequency
domains is depicted in Fig. 1.2. Fourier transform is used to change a signal from time domain
to frequency domain. The reverse is done by inverse Fourier transform. The frequency
response is a complex function of frequency that can be expressed by a magnitude and a
phase spectrum.

x(t)

y(t)

h(t)

x(f)

h(f)

FT

FT

IFT

Frequency
Domain

IFT

Time
Domain

y(f)

Figure 1.2 The relationship of convolution in time domain and in the frequency domain [12].
6

1.4 Objective of Research


The goal of this research is to design, calibrate and test a simple force balance system
that is capable of measuring drag on various models. As the test time is of very short duration,
force measurement becomes a challenge. To design this force balance system, an approach is
utilized to model the response of the balance using FEA. ANSYS Explicit Dynamics solver is
used for the dynamic analysis. Piezoelectric films were used to measure stress waves due to
aerodynamic loading. Deconvolution was used to determine the system transfer function and to
recover the force. The drag on a spherically blunted cone was measured and the drag
coefficient was compared with that obtained from modified Newtonian theory. Future efforts
would consist of extension of the force balance to measure other components of force, such as
lift and pitching moment. Force measurement on other models, such as conical model, inlets
and scramjet vehicles would be included.

CHAPTER 2
FACILITY
2.1 UTA Hypersonic Shock Tunnel at the Aerodynamics Research Center
The Hypersonic Shock Tunnel at the UTA Aerodynamics Research Center is a reflected
type. It was designed and built in the late 1980s [13]. The main components of the hypersonic
shock tunnel include the driver section, driven tubes, nozzle, test section, diffuser and dump
tank, which are shown in figure 2.1.
This facility is able to simulate high Mach numbers and high enthalpy flows. The main
parts of this shock tunnel are:

Driver tube

Diaphragm section

Driven tubes

Nozzle

Test section

The shock tube is fabricated in four sections for ease of transportation, installation and
maintenance [14]. The driver tube is a single section which is designed for a maximum
operating driver pressure of 41.4 MPa (6000 psi) and hydrostatically tested to 62.1 MPa (9000
psi). The driver tube is 3 m (10 ft.) long with an internal diameter of 15.24 cm (6 in.) and a wall
thickness of 2.54 cm (1.0 in.). One end is closed off with a hemispherical end cap.

Figure 2.1 Schem


matic of the UTA Hypersonic
H
Shock Tunnel and its dimensions
d

9
Figure 2.2 Pan
norama photogra
aph of the UTA Hy
ypersonic Shock Tunnel

The other
o
end has
s a 48.26 cm (19 in.) diam
meter 11.43 cm
m (4.5 in.) thicck flange, which allows the
e
drive
er tube to be bolted
b
to the diaphragm
d
se
ection and the
e driven sectio
on [15]. The fllange has two
o
O-rin
ng grooves machined in them to accom
mmodate two O-rings
O
to en
nsure a tight, high-pressure
e
seal [15].
ble diaphragm
m section se
eparates the driver from the driven tu
ube, so-called
d
The doub
d to hold two diaphragms. This section has the sam
me dimension as that of the
e
because it is used
ges. Steel boltts of 2 in. diam
meter are use
ed to bolt the flanges togetther.
flang

Figurre 2.3 Schem


matic of the do
ouble diaphrag
gm section be
etween the drriver and drive
en tube [15].

ograph of the
e double diap
phragm section (middle), driver sectio
on (right) and
d
Figurre 2.4 Photo
drive
en section (lefft). The steel pipe
p
is used to
t pressurize the double diiaphragm secction.
10

The driven tube is constructed in three segments of 2.74 m (9 ft) length each. The three
segments are connected to each other with a flange at each end identical to the one on the
driver section. The internal diameter is the same as the driver tube. Two O-rings are located
between each connection to provide a good high-pressure seal.
The expansion nozzle was developed by LTV Aerospace and Defense Company,
presently a part of Lockheed Martin Missiles and Fire Control. It was part of an arc-driven
hypervelocity wind tunnel facility and was subsequently donated to UTA. The end of the driven
tube has a special coupling for the nozzle insert and secondary diaphragm. The coupling
consists of several parts that form a locking system for the throat insert. The nozzle has
interchangeable throat inserts to provide a discrete test section Mach numbers of 5 to 16. The
nozzle has a length of 2.57 m (101 in), with an exit diameter of 33.6 cm (13.25 in) at the test
section [15].
The test section has a dimension of 53.6 cm (21.1 in.) in length and 44 cm (17.5 in.) in
diameter. It has circular access windows of 23 cm (9 in.) diameter facing each other on either
side. These two ports can be used as mounting ports or for optical windows [10]. The rear of
the test section has a conical converging section which leads into the diffuser. The dimensions
of the converging section are 38.1 cm (15 in) in diameter within the test section and it contracts
to a diameter of 31 cm (12.2 in) to the entrance of the diffuser. The flow is captured by the
converging section and generates the first shock wave necessary to slow the flow down in the
diffuser [15].
The dump tank is located outside the building and has a volume of 4.25 m3 (150 ft3).
The vacuum system vacuums the shock tunnel from the tank to the secondary diaphragm. A
35.6 cm (14 in) vacuum pipe is connected directly from the dump tank by a flange joint with a
double O-ring seal. A smaller 7.62 cm (3 in) diameter piping is used for connecting the vacuum
pump to the vacuum tank.

11

A high-pressure system is used to pressurize the driver tube. Another lower pressure
system is used to operate the remote control valves and the booster pump (Haskel model55696) on the high-pressure system. The high-pressure system consists of a 5-stage
compressor and a booster pump. The 5-stage compressor (Clark Model CMB-6) which is
located in the adjacent compressor building can provide dry air at up to 14.5 MPa (2100 psi).
The booster pump (Haskel Model 55696) is a two-stage booster pump which is used to attain
pressures of up to 41.4 MPa (6000 psi) in the driver tube. The Haskel pump uses dried, filtered
compressed air from the main compressor or helium supplied from 2200 psi bottles. The
pressurized gas is stored in a one meter diameter spherical storage tank which can hold
pressures up to 41.4 MPa (6000 psi).
The low pressure is generated by another compressor (Kellogg American inc. modelDB462-C) which supplies dry air at 1.2 MPa (175 psi). Regulators are used to reduce the
pressure to 689.5 kPa (100 psi), which is needed for the booster pump operation. The lowpressure compressed air (175 psi) is used by both the vacuum pump isolation valves and the
booster pump in the high-pressure system.
A secondary diaphragm separates the driven section from the test section. Both the
driven section and the test section including the nozzle have their own vacuum pumps. The
driven tube is vacuumed by a vacuum pump (Sargent-Welch Model 1376). This pump has a
free-air displacement of 300 liters per minute and is able of pumping down to 0.001 mmHg. The
test section is vacuumed by another vacuum pump (Sargent-Welch Model 1396) which is
connected to the dump tank. This pump is capable of a free-air displacement of 2800 liters per
minute and is able of achieving low pressures of up to 0.0001 mmHg. Vacuum is measured in
both the driven tube and the dump tank by a pressure gauge (MKS Baratron Type 127A). The
gauge has a full-scale range of 1000 mmHg and an accuracy of 0.1 mmHg.

12

2.2 Reconstruction of the UTA Hypersonic Shock Tunnel


The hypersonic shock tunnel had not been in use for many years and had been
disassembled for a long time due to other research activities. Reconstruction was needed and
began in 2010, where some of the parts had to be repaired or replaced with redesigned parts.
The hypersonic shock tunnel began full operation by mid 2011.
The first steps involved were setting up the driven tube sections which included removal
of corrosion and cleaning the inner tube. The diaphragm section was attached back to the driver
segment. For obtaining a good vacuum, the system had to be rechecked to ensure good seal. A
schedule 40-steel pipe of 76 mm (3 in.) internal diameter and 2.13 m (7 ft.) length had to be
replaced and customized for convenient attachment to the external dump tank. Safety valves
from the dump tank had to be replaced. Due to corrosion of the inner surface of the tank, it was
cleaned and treated with Enrust to prevent further occurrence. Some components had to be
refabricated or redesigned. The throat locking mechanism for the nozzle inserts had to be
fabricated in 4340-stainless steel. A diffuser section was designed for convenient sting
installation. This section has five ports used for model mounting and instrumentation purposes.
2.3 Diaphragm Test
As mentioned before, the driver and driven tube are separated by double diaphragms
made of 1008 steel (10/12gage, 0.03 in. thickness). New thickness tests had to be conducted
for higher pressure in the range of 20~30MPa (3000~4500 psi). Thickness and scoring play
important roles for achieving proper rupture. In some tests, the petals were torn off, which are
undesirable. These steel diaphragms must be scored with a cross pattern on each run, for
perfect rupture. Several tests were conducted on the scoring depth and thickness of the plate,
to improve the quality of the rupture and to contain the needed pressure. The tests ensured a
clean rupture and minimal petal fragmentation. Detailed drawing is given in appendix A. The
special cross pattern, known as a cross potent in heraldry was made [15] with a CNC machine
for quick manufacturing and reduced cost. Moreover, CNC machining helps in maintaining
13

conssistent scoring
g depths. The
e figure 2.5 shows
s
the ste
eel diaphragm
m before and after the testt.
Table
e 1.1 specifie
es the differen
nt rupture presssures for diffferent scoring
g depths.

Table 2.1 Ruptu


ure propertiess of diaphragm
m tests.

High-pre
essure Test

Diaphrag
gm scoring de
epth
(inches)

Rupture Presssure
R
(Psia)

1
2
3

0.040
0.035
0.030

2200
2548
3050

Figurre 2.5 Steel diaphragms,


d
(a) Scored Dia
aphragm, (b) Ruptured dia
aphragm afterr a successfu
ul
test.

14

CHAPTER 3
DESIGN AND EXPERIMENTAL SETUP
3.1 Force Balance Design
An investigation was conducted by Robinson et al. [4] on both internal and external
balances to measure forces and moments using FEA (finite element analysis). Their analysis
showed that greater accuracy of the recovered forces and moments could be obtained with the
external force balance design. For the present work, the design of an external force balance
(stress wave force balance) was investigated. This balance has the ability of mounting a variety
of models. Some of the conceptual design requirements included:
1. Size constraint.
2. Strength of Balance and other components.
3. Model and support attachment
4. Strain gage and transducer placement.
6. Calibration ease
7. Machining simplicity

The force balance can only be accommodated in the limited room given by the
dimension of the test section, including the model. The design should be able to adapt to the
test section of the UTA Hypersonic Shock Tunnel, which has a dimension of 53.6 cm (21.1 in.)
in length and 44 cm (17.5 in.) in diameter. The strength of the balance is important in deciding
on the type of material. Different type of metals including steel, stainless steel, brass and
aluminum were investigated.

15

Table 3.1 shows some properties of specific metals/ alloys.


Material

Young's Modulus
(GPa)

Maximum tensile
strength
(MPa)

Density
(kg/m)

Aluminum-6061-O

69

310

2700

Brass

97-105

550

8000-8730

Stainless steel, AISI 302 cold-rolled

207

860

8190

Steel, API 5L X65

200

531

7600

For ease of manufacture, reduced weight and high strength, aluminum (Al-6061) was
chosen as a suitable material. Other materials used for the model and support bolts include
hardened steel and stainless steel. An FEA was used to assist in selecting a suitable force
balance design amongst a number of candidates. The first step was modeling simple stress
bars to understand the propagation of stress waves in solids. These stress bars were analyzed
in ANSYS using the Explicit Dynamics solver. All preliminary conceptual designs were modeled
using Catia V5 and analyzed using the FEA solver. Many other types of designs were analyzed,
as shown in figure 3.1.

16

Fig
gure 3.1 Different prelimina
ary designs.

n gage placem
ment the slan
nt faces are the
t preferred stress bars. The differen
nt
For strain
prelim
minary designs were ana
alyzed using FEA by appllying a simullated impulse
e force to the
e
front, at the mode
el mount locattion, details limited to (f) iss explained in
n section 3.1.1
1. Designs (a)
and (b) showed a large stresss concentratio
on where the
e bolts are loccated. Design
ns (c) and (d)
show
wed higher in
nternal reflecttions of stresss waves tha
an (f). Design
n (e) was co
omplex, which
h
would not be prac
ctical. FEA de
emonstrated that
t
most stra
ain is seen on
n the stress bars
b
of design
n
(f). Design
D
(f) was
s chosen con
nsidering all th
he factors tha
at are mentioned above. Itt was decided
d
to mo
odel the force
e balance as a single solid piece for fabrication ease.
The force
e balance wass fabricated in
n 6061 alumin
num alloy from
m a single so
olid block. The
e
single block desig
gn helped in reducing the
e stress concentration, which
w
tends to
o accumulate
e
more
e at joints wh
here memberrs are fastened together. Figure 3.2 shows
s
the fab
bricated force
e
balan
nce. The force balance has a dimension
n of 20.9 cm (8.25 in.) in le
ength, 10.8 cm (4.29 in.) in
n
heigh
ht and 2.5 cm
m (1 in.) in wid
dth. The stresss bars have a 0.88 cm (0.35 in.) thickn
ness. Detailed
d
draw
wing of the forc
ce balance iss given in appendix A.

17

Figure 3.2 Fabricated forcce balance (6


6061- aluminu
um alloy)

3.1.1
1 Finite Eleme
ent Analysis
Finite elem
ment analysiss which helpss to determine
e stress distriibution, displa
acements and
d
deforrmations, was
s used to verify the stresss concentration and respo
onse of the force
f
balance
e.
The ANSYS
A
Explicit Dynamics solver is use
ed to understa
and the dynam
mic response of a structure
e
unde
er a time-vary
ying load typiccally with dura
ations of less than 1 secon
nd. This solve
er can also be
e
used
d for impact analysis, sho
ock propagattion and stre
ess wave pro
opagation. Th
he differentia
al
equa
ation of motion
n in structural dynamic ana
alysis is given
n by,

(3.1)
wherre
m iss the mass off the system,
c iss the damping
g coefficient ,
k iss the stiffness
s constant ,
u iss the displace
ements vectorr
p(t) iss the vector of
o the time-varying load.

18

At each point in time, the vectors of displacement , velocity

and acceleration

are

of particular interest to determine stress concentration of the balance and understand its
dynamic response.
A mesh was generated using the explicit meshing feature. Tetrahedral elements were
used to mesh the force balance. A total number of 34088 elements was used in the simulation.
A uniform mesh was generated with default size elements to respond to high frequencies of the
stress wave. Mesh refinement was used for computational efficiency, by maintaining larger
elements to insignificant areas and increasing relevance to areas of higher stress concentration.
Care was taken in this process, as a coarse mesh was not able to transmit high-frequency
information to the finer mesh [1]. Damping was not used in the simulation, so as to acquire high
frequency stress waves. For computational ease, the balance was analyzed without the other
small components. Figure 3.3 shows the mesh of the force balance.

Figure 3.3 Generated mesh of the force balance using explicit dynamics.

The force balance was modeled as a rigid body, the top surface, a fixed support and
the simulated force was applied from the front face, which is shown in figure 3.4. The time step
19

is co
ontrolled by th
he smallest element
e
size, which is use
ed to progresss the solution
n in time. The
e
soluttion time step was 0.085 ss.

ure 3.4. Analysis settings, force is applied from the front


f
surface.
Figu

Fig
gure 3.5 Strain concentratio
on in stress bars
b
of the forrce balance.
20

From the solution of the stress analysis, it can be seen that stresses are prominent on
the two stress bars, figure 3.5. Animation results show that stress waves move from the front of
the axial bar towards the stress bar and then to the rear of the axial bar. Reflections of stress
waves occur in the stress bars. These results are shown in figure 3.9.
First analysis was to determine the response of the force balance to a simulated
impulse force, figure 3.6. The pulse of 220 s width, was applied to the front surface, with
maximum amplitude of 350 N (78.6 lbf). Strain was monitored on two locations, on the top
surface of the axial bar (location 1) and on the stress bar (location 2), as shown in figure 3.4.
Figure 3.7 (a) shows the strain output on location 1, which resembles the input impulse. The
response of location 2 is shown in figure 3.7 (b). Many reflections are seen in the response of
this location due to various wave reflections in the stress bar. From the simulation, a transfer
function was obtained from the strain-history of locations 1 and 2. The next analysis was to find
the response of the force balance to a simulated step load. The step load (100 s rise time) of
222.4 N (50 lbf) was applied to the model for a period of 4 ms. The step load and step
response on location 2 can be seen in figure 3.8.

Figure 3.6 Simulated input load of 350 N. The pulse starts at 90 s.


21

The input step lo


oad was reco
overed by de
econvolution of
o the step response
r
with
h the transfe
er
functtion. FEA sho
owed the dyn
namic behavio
our of the forcce balance to
o impulsive fo
orces and also
o
demo
onstrated pos
sitions to placce the strain gages.
g
It helpe
ed in showing
g that input fo
orces could be
e
successfully recov
vered.

(b)

(a)

onse to the simulated


s
imp
pulse at (a) lo
ocation 1 and
d (b) location
n 2.The pulse
e
Figurre 3.7 Respo
startss at 90 s.
.

(a)

(b)

8 (a)Simulated
d step load off 222.4 N and
d (b) step resp
ponse of locattion 2.
Figure 3.8
22

Figure 3.9 Animated result shows the


e stress wave propagation in the force balance.

23

3.1.2
2 Force Balan
nce Constructtion
A 12.7 mm
m (1/2-in.) threaded hole in the fro
ont of the ba
alance is use
ed for mode
el
attacchment. Two 12.7 mm 1/2--in. threaded holes were available at the
e top to attacch the balance
e
convveniently into the test secction. Two 6.3
3 mm (1/4-in
n.) threaded holes were lo
ocated at the
e
botto
om for attachm
ment as need
ded. Figure 3..2 shows the fabricated forrce balance with
w a detailed
d
draw
wing given in appendix
a
A.
A blunt co
one model wa
as chosen to
o validate the force balancce by measuring the forcess
and comparing th
hem to the prrediction, which will be disscussed in se
ection 5.2. Th
he blunt cone
e
mode
el was made
e of steel with
h a base radius of 40 mm
m (1.575 in.) and a semi--angle of 18.5
5
degre
ees. The mod
del was 88.9 mm (3.5 in.) long and it we
eighs about 907.1
9
g (2 lbss). Figure 3.10
0
show
ws a photogrraph of the blunt
b
cone. For
F simultane
eous pitot pre
essure measurements the
e
mode
el was design
ned to hold a PCB 113A21
1 pressure tra
ansducer. A pressure
p
transducer holde
er
was designed to firmly embracce the transd
ducer. This ho
older was tigh
htened from the centerline
e
throu
ugh the base of the mode
el to the nose
e of the cone
e. The pitot pressure
p
transducer had a
recesss of 4.5 mm
m (0.18 in.). The
T
design specification
s
of both the blunt
b
cone model
m
and the
e
presssure transduc
cer holder are
e given in the Appendix A.

Figure 3.10
0 Blunt cone model
m
(a) side
e view (b) fron
nt view.
24

Another component that was used was a two-sided bolt, with a hole drilled through it.
This bolt was used to connect the model to the force balance and take out the coax cable from
the pressure transducer. A 12.7 mm (1/2 in.) thread on one side was used for balance
attachment and a 17.4 mm (11/16 in.) thread was used for model attachment. The overall length
of the bolt was 96.5 mm (3.8 in.). Another feature of this bolt is its hinge design, so that the
angle of attack of the model can be changed, from 5 to +5 deg. Due to this feature, the bolt
made of 4140 steel was additionally hardened and drawn, making it strong enough to withstand
high impact loading. Detailed drawings of parts of the bolt are given in appendix A.

Figure 3.11 Hardened steel bolt, the larger pin is used to change the angle of attack and the
other pin is used to lock the position.

The hardened steel bolt was screwed into the base of the model with required wiring
lead taken out. This assembly consisting of the model and the hardened steel bolt was attached
25

to the front of the force balance, thus in the centerline. Piezoelectric film gages (Measurement
Specialties Model DT1-052k) were used to measure the stress waves. Two gages were used,
one on the stress bar and the other behind the model on the balance aligned with the axis. The
gages were shielded with copper foil to prevent EMI (electromagnetic interference). Care was
also taken to protect the gages from direct pressure exerted during flow by wrapping PVC/
rubber around them, which was then sealed with electrical tape.

Figure 3.12 Installed model and balance in the test section. The support structure can be seen
on the top.
Additional strain gages (Omega Model SGD-3/120-LY13) were used for static
calibration. These gages were installed on the centerline of the first stress bar in a Wheatstone
bridge arrangement. A full-bridge mode was chosen since it gives maximum sensitivity to strain
and also provides temperature compensation. M-bond 200 adhesive and conditioner were used
as adhesive for attaching the strain gages. The finished gages were then given a protective
coating of M-coat c. Figure 3.13 shows the installed gages on the front bar. The strain gage

26

signal was amplified with a strain gage amplifier (Paine Model strain gage amplifier) with a gain
of 100 and an excitation of 10 V.

Figure 3.13 Two strain gages are seen on the stress bar, the other two gages of the full bridge
are attached to the lower surface of the stress bar.

Two holes of 12.7 mm (1/2-in.) diameter, a distance of 11.6 mm (4.6-in.) apart, were
drilled from the top of test section. The force balance assembly was attached to the ceiling of
the test section by two hardened steel bolts. A 5.5 mm (7/32-in.) diameter hole was drilled into
each bolt for channeling wiring from the test model and force balance to outside the tunnel.
These holes were later sealed from both inside and outside. For adjusting the alignment and
27

height of the force balance an aluminum block of dimensions 2.5 cm 2.5 cm 15.2 cm (1 in.
1 in. 6 in.) was installed between the balance and the ceiling. The model-balance assembly
was attached to the aluminum block by two 12.7 mm (1/2-in.) steel bolts. Steel washers and
rubber washers/bushings where placed in all bolt connections for damping. Figure 3.14 shows
the force balance-model assembly attached in the test section.

Figure 3.14 The force balance with the model, view from front of the nozzle.

28

3.2 Calibration Techniques


3.2.1 Static Calibration
Static calibration is done by loading the force balance system with known weights and
measuring the output for each increasing load. After loading the balance unloading is performed
in like manner. This procedure helps to characterize the linearity and the possibility of hysteresis
in the system. Figure 3.15 shows a sketch of the static calibration procedure.

Figure 3.15 Schematic of the static calibration procedure.

Static calibration was performed on a thrust stand by holding the force balance rigidly.
Steel wire rope was attached to the blunt cone using the pressure transducer holder by
tightening the holder inside the cone. The steel wire was tied to a digital scale (AWS model-TL440), which had a maximum load range of 1957 N (440 lbf). A turnbuckle was used to connect
the weighing scale rigidly to an anchor bolt in the thrust stand. Tension was applied to the
model-balance assembly by tightening the turnbuckle. The force was progressively applied, up
to a maximum of 1423 N (320 lb). The strain was noted down for each load. Table 3.2 shows
29

the applied loads. For this calibration the foil strain gages were used as they can measure the
applied static loads. The same procedure was done for unloading the force. Figure 3.16 shows
the strain gage output voltage versus applied force on the balance system, for both the loading
and unloading case. No significant hysteresis was seen in the static calibration. The trendline
equation V=0.003F+0.003, R2=0.999, for loading and the trendline equation is V=0.0003F
+0.009, R2=0.995 clearly show a linear relationship between the applied load and strain.

0.45
0.40

Voltage,(V)

0.35
0.30
0.25
0.20
0.15
0.10

Loading

0.05

Unloading

0.00
0

200

400

600

800
1000
Force(N)

1200

1400

1600

Figure 3.16 Static loading and unloading of force balance.

30

Table 3.2 Static calibration results.


Mass(lbs)

Force(N)

0
20
40
60
80
100
120
140
160
180
200
220
240
260
280
300
320

0.0
88.9
177.9
266.8
355.8
444.8
533.7
622.7
711.7
800.6
889.6
978.6
1067.5
1156.5
1245.5
1334.4
1423.4

Straingageoutputvoltage(V)
Loadingofweight
Unloadingofweight

0.000
0.025
0.047
0.071
0.097
0.119
0.141
0.167
0.188
0.212
0.232
0.254
0.276
0.298
0.320
0.349
0.369

0.000
0.030
0.054
0.077
0.101
0.132
0.156
0.178
0.209
0.232
0.257
0.280
0.304
0.321
0.339
0.355
0.373

Another approach that was performed for static calibration was using an impulse
hammer (PCB Model 086C01) and averaging the output for different input hammer forces. For
this method, piezoelectric films were used to measure the dynamic strain. Since these
piezoelectric films measure dynamic forces, the output forces were averaged. Given that the
output of the piezoelectric film gages oscillates around zero, the average would give values
close to zero, therefore the rms value of the output voltage was calculated. Thus the sensitivity
constant can be determined. The trendline equation of V=7E-05F +0.003, R2=0.953 of figure
3.17, shows the linear relation formed by averaging output of different hammer hits, with
different force values.

31

0.050
0.045
AveragedVoltage(V)

0.040
0.035
0.030
0.025
0.020
0.015
0.010
0.005
0.000
0

100

200

300

400

500

Force(N)
Figure 3.17 Plot of average film output versus applied hammer force.
3.2.2 Dynamic Calibration
Dynamic calibration is performed to characterize the balance behavior when a sudden
load (impulse force) acts on a system. Dynamic calibration is performed by different methods
including hammer impulse excitation. A perfect impulse will excite all the frequencies of a
system, which consist of the model and the force balance. Other ways of calibrating is creating
a step load by hanging weights from a wire and cutting the wire. This type of calibration can be
done both horizontally and vertically. Drop test calibration is a procedure of suspending the
balance and cutting the wire above the balance, thus creating a step response. For the drop test
calibration, care must be taken to prevent damage from impact. Figure 3.18 is a sketch of a cut
weight test. The balance and model system are attached from the top and a known mass is
suspended from the end of the model through a pulley. The wire is then cut near to the model
creating a step response, measured with the gages.

32

Figure 3.18 Sketch of a cut weight test.


Cut weight test was also performed vertically, since vibrations can occur in the step response
when using a pulley as support. This arrangement is seen in figure 3.19.

Figure 3.19 Vertical cut weight test


33

Dynamic calibration is performed to


o find the forrce balance characteristics
c
s, also known
n
as th
he transfer fun
nction. Calibra
ation is done by striking th
he model with
h an instrumen
nted hammerr.
A scchematic of th
his calibration
n is shown in
n figure 3.20. A PCB Model 086C01 impulse force
e
hamm
mer was used
d with the mo
odel installed in the test se
ection. A signa
al conditionerr (PCB Model483A
A) was utilize
ed for the ham
mmer signal. The piezoelectric films were
w
used to measure the
e
outpu
ut strain. Da
ata were reco
orded using an oscillosco
ope (Tektronix Model DP
PO 4054, see
e
Appe
endix C for details).
d
The sampling
s
rate
e for the calib
bration was 25MS/s.
2
Diffe
erent hamme
er
tips, consisting off metal, plastic and rubberr were tested to determine
e one that is most
m
suitable
e.
Testss showed tha
at the metal tip excited the system with
w
higher fre
equencies. The
T
rubber tip
p
damp
ped higher fre
equency and created impu
ulses of largerr pulse width. Both the mettal and plasticc
tips were
w
used to create the im
mpulse. Figurre 3.21 illustra
ates the raw data of a ham
mmer impulse
e
and the
t strain output. A detail of
o the pulse with
w the metal tip is shown in figure 3.22
2, the hamme
er
signa
al takes more
e than 1 ms to
o reach back to its steady state, right after
a
the pulse
e. When using
g
the im
mpact hammer care was taken
t
to obta
ain a good pu
ulse. A poor strike
s
will crea
ate a string of
o
small pulses due to
t the bounce
e of the hamm
mer.

Figure 3.2
20 Schematicc of impulse hammer
h
calibrration

34

Figurre 3.21 Raw data of a ha


ammer impulsse (above) an
nd the respon
nse of the ha
ammer impacct
taken
n over duratio
on of 150 ms (below).

F
Figure
3.22 Sa
ample hamme
er impulse cre
eated by strikking the metall tip on the mo
odel cone.
35

To form the transfer function (impulse response), the obtained strain output is
deconvolved with the hammer impulse. As mentioned before, a poor hammer strike can result in
obtaining an inaccurate transfer function. The hammer strike can be verified, since theoretically
convolution of an ideal impulse with a unit step results in a perfect step response. Matlab was
used to create a unit step (start at t=0) as shown in figure 3.25. The hammer impulse was
convolved with this unit step. The resultant convolved signal is illustrated in figure 3.23. This
signal was compared with a simulated step response, which is formed by convolving a modified
impulse with the unit step [9]. This modified impulse was created by padding zeroes right after
the pulse of the hammer strike, as shown in figure 3.26. The pulse was identified to have a
width of approximately 487s, as illustrated in figure 3.27.

Figure 3.23 Check signal of both raw hammer signal and modified hammer signal.

A detailed view of the check signal is shown in figure 3.24. The error bar shows the
deviation of the signal from the modified pulse signal. The hammer check signal agrees with the
perfect step response. An error estimate on the signal shows variation of 0.6 %. It may be
36

conccluded that the


ese variationss in the hamm
mer check sig
gnal is becausse the mean right after the
e
pulse
e of the hamm
mer strike, (fig
gure 3.22) is not
n zero.

Figure
e 3.24 Detaile
ed view of che
eck signal witth error bar

Fiigure 3.25 Siimulated unit step input.


37

The pulse
e width chang
ges for each hammer impu
ulse, the calib
bration pulse width ranged
d
from 200 to 500 s.

Fig
gure 3.26 Mo
odified hamm
mer impulse.

Figurre 3.27 Enla


arged view of
o the pulse of the modiified hammerr impulse. Pulse width iss
appro
oximately 487
7s.
38

Througho
out this work, Matlab wa
as used for data
d
processing. The impu
ulse response
e
was found by dec
convolution of
o the pulse output
o
with th
he hammer im
mpulse. The deconvolution
d
n
was tested in two
o ways, one using
u
FFT alg
gorithm and th
he other itera
ative deconvolution method
d
using
g functional minimization
m
w extended
with
d conjugate gradient
g
algorithm (JMECG
G) [16]. From
m
figure
e 3.28 it can be seen tha
at both the sig
gnals, obtaine
ed by FFT an
nd JMECG ag
gree well with
h
each
h other. Both methods ca
an be used in
n determining
g the impulse
e response. The obtained
d
deco
onvoluted sign
nals had to be
b filtered aga
ain, since the
e deconvolution method amplifies
a
highfrequ
uency compon
nents. A ten-p
point moving--average filterr was used fo
or this samplin
ng rate.

Figurre 3.28 Com


mparison of the impulse response ob
btained using
g two metho
ods (FFT and
d
JMEC
CG)
The impu
ulse response
e in the frequ
uency domain
n is known as
a the frequency response
e
functtion (FRF). From
F
the FRF
F shown in fiigure 3.29, th
he higher frequencies mig
ght be due to
o
39

intern
nal reflection of stress waves. The re
esulting frequ
uency respon
nse describess the balance
e
syste
em characteristics. A deta
ailed view of the
t first 12 kHz
k
is shown in figure 3.30. The phase
e
specctrum of the siignal is illustra
ated figure 3.31.

Figurre 3.29 Powe


er spectral density plot off frequency response
r
function showing the variouss
frequ
uencies.

3
Enlarged
d power specctral density plot of FRF forr the first 12 kHz.
k
Figure 3.30
40

Figure
e 3.31 Enlarge
ed view of phase spectrum
m of FRF for the first 12 kH
Hz.

Matlab was used to


o create a spe
ectrogram of the
t frequencyy response fu
unction (FRF)),
which
h is shown in
n Figure 3.32
2. The spectrrogram was calculated
c
ussing the shorrt-time Fourie
er
transsform. The spectrogram shows
s
the fre
equency variiation with tim
me. At any point
p
of time
e,
frequ
uencies are excited, this iss due to the hammer impullse. The modes of vibration are seen ass
red. The
T first mode shows the maximum am
mplitude.

F
Figure
3.32 Sp
pectrogram of the FRF.
41

3.3 Shock Tunnel Testing


The experiments were conducted with the UTA Hypersonic Shock Tunnel, using air as
the driver gas. Steel diaphragms were used in the double diaphragm section as mentioned in
chapter 1. The secondary diaphragm was made of mylar (0.010 in. thickness). For the tests, a
Mach 10 nozzle insert was used.
CEA (Chemical Equilibrium with Applications) was used in calculating reflected shock
conditions. The flow was considered a frozen composition. Shock velocity was calculated using
two pressure transducers (PCB Model-111A23), which were located 82.5 cm (32.5 in.) and
219.7 mm (86.5 in.) from the end of the driven section. The free stream flow conditions were
found using the perfect gas relations, which are summarized in table 3.3. For the experiments
all the signals including the pitot pressure, force data and pressure transducers in the driven
tube (CH1 and CH2) were recorded simultaneously using an oscilloscope (Tektronix Model
DPO 4054, see Appendix C for details). A rising edge trigger was used for the first pressure
transducer (CH1) for a level of 300 mV. It was set to ensure no loss of data and capture all
signals. The data were sampled at 25 MS/s and for a duration of 40 ms.

Table 3.3 Test conditions.


Condition
No:

P0
(MPa)

o
(kg/m3)

T0
(K)

p
(Pa)

(kg/m3)

T
(K)

V
(m/s)

H0
(MJ/kg)

9.441

2.68

10.22

914.6

72.89

4.95E-03

51.3

1344.9

0.65

9.427

2.705

10.21

923.7

73.76

4.93E-03

52.0

1352.3

0.66

Condition
1
Condition
2

42

CHAPTER 4
RESULTS AND DISCUSSIONS
4.1 Force Measurement Prediction
4.1.1 Modified Newtonian Theory
Newtonian theory assumes that the oncoming flow can be considered of as continuous
stream of particles.

When the particles hit a surface at high speeds, they lose all their

momentum perpendicular to the surface. The pressure coefficient predicted by Newtonian


theory is given by

2sin

(4.1)

This equation shows that the pressure distribution is related to the square of the inclination
angle. The modified Newtonian theory was proposed by Lees in 1955 so that the pressure is a
function of M:

(4.2)

where, Cp,max is the maximum pressure coefficient behind a normal shock wave, at the
stagnation point.

is calculated using the Rayleigh pitot formula [18] :

(4.3)

43

The axial force coefficient is calculated by the following relation [17],

2C

0.25 cos

0.125 sin

cos

cos

sin

cos

(4.4)

0.50 sin

cos

tan

sin

cos
cos

2tan

where,
RN is the nose radius of the blunt cone
RB is the base radius of the blunt cone
c is the half cone angle
is the angle of attack
The normal force coefficient is calculated by the following relation [17],

2C

0.25


cos

cos

(4.5)

The following relation is used to calculate Lift-to-Drag ratio:

44

(4.6)

Drag
g is calculated
d by using the
e relation

v SC (4.7)

(4.8
8)

(4.9
9)

4.9) into (4.7) yields


Eqnss. (4.8) and (4

(4.10)

The axial and norrmal force coefficients, coe


efficients of liift and drag were
w
estimate
ed for a range
e
of an
ngles of attack
k from 0 to 15
5 deg which are
a summarize
ed in Tables 4.1
4 and 4.2.

Figurre 4.1 Plot of (a) coefficien


nt of drag vs. angle of atta
ack and (b) co
oefficient of lift vs. angle of
o
attacck.
45

Table 4.1 Force prediction using modified Newtonian theory using condition 1.
Angle of Attack
()
0
5
10
15

Ca

Cn

Cl

Cd

L/D

0.2697
0.2736
0.2852
0.3041

0
0.1427
0.2780
0.3991

0
0.1183
0.2243
0.3067

0.2697
0.2850
0.3291
0.3971

0
0.4152
0.6814
0.7725

Table 4.2 Force prediction using modified Newtonian theory using condition 2.
Angle of Attack
()
0
5
10
15

Ca

Cn

Cl

Cd

L/D

0.2692
0.2732
0.2848
0.3038

0
0.1427
0.2781
0.3991

0
0.1184
0.2244
0.3069

0.2692
0.2846
0.3287
0.3967

0
0.4160
0.6826
0.7736

4.1.2 Coefficient of Drag Calculation using Pitot Pressure


The following equation shows how force coefficients are found by normalizing the force
history by the pitot pressure using a suitable scaling factor [9],

From equation 4.7

(4.11)

The Rayleigh-pitot formula given in equation 4.5 can be also written as [9],

(4.12)
46

For high values of M, when 2

1 , equation 4.12 can be approximated as,

(4.13)

(4.14)

(4.15)
Substituting these into equation 4.13 we get

Substituting

(4.16)

in equation 4.16

(4.17)

For a given flow condition

and S remain constant, therefore drag coefficient is expressed as

(4.18)

Measuring the pitot pressure along with the force during an experiment can be used to estimate
the force coefficient, which also varies with time.
47

4.2 Expe
erimental Ressults
The predicted values using the mo
odified Newto
onian theory and the reco
overed forcess
were
e compared. Figure 4.2 is a plot of the coefficie
ent of drag history
h
obtain
ned from the
e
recovvered force of
o condition 1. This data was
w normalized with the pitot
p
pressure
e as shown in
n
section 4.1.2. An average
a
taken from 175 ss to 275 s iss also shown in the figure.

Figu
ure 4.2 Coefficient of drag from the reco
overed force (condition
(
1).

It can be seen that ap


pproximately the
t first 120 s is the flow
w commencin
ng stage, after
the fllow arrival the
e drag coefficcient remains steady for a period of 100
0 s and is th
hen noticed to
o
decre
ease. This is due to increa
ase in pressure at the base
e of the cone, which would
d describe the
e
decre
ease in the coefficient
c
of drag
d
[9]. A lo
ow frequency oscillation was
w found in the
t test signa
al
which
h would also account for the unsteadyy drag force. Further invesstigations nee
ed to be done
e
on th
his low freque
ency vibration. Figure 4.5 shows
s
that the
e recovered fo
orce data follo
ows the same
e
trend
d as the pitot signal.
s
48

Figure 4.3
3 Recovered drag
d
force an
nd theoretical force.
The raw
r
pitot pres
ssure signal iss shown in fig
gure 4.4, a de
etail of the pito
ot pressure fo
or the first 400
0
s is shown in figu
ure 4.5. The data
d
was filterred using a 20 kHz low-pass Butterwortth filter.

Figure 4.4
4 Raw pitot signal.

49

Figurre 4.5 First 40


00 s of the pitot
p
pressure and drag. A 20
2 kHz low pa
ass filter was used.
The results for condittion 2 are sh
hown in figurres below. Figure 4.6 is a plot of the
e
Coeffficient of drag history obta
ained from th
he recovered force of condition 2. Figu
ure 4.7 showss
the re
ecovered forc
ce.

Figu
ure 4.6 Coefficient of drag from the reco
overed force (condition
(
2).
50

Figure 4.7 Recovered


R
dra
ag force.
The raw
r
pitot pres
ssure signal iss shown in fig
gure 4.8, a de
etail for the firsst 400s is sh
hown in figure
e
4.9. The
T data was
s filtered using
g a 20 kHz low
w-pass Butterworth filter.

Figure 4.8
8 Raw pitot signal.
51

Figure 4.9
9 First 400 ss of the pitot pressure.
p
A 20 kHz low-pa
ass filter was used.
u
From
m the experim
mental resultss it is seen that the reco
overed drag is in accorda
ance with the
e
meassured pitot pressure. The
e experimenta
al to theoreticcal drag com
mparison is su
ummarized in
n
table
e 4.3. % is
i the differeence betweeen the theorretical and experimental
e
values. The
e
varia
ations seen in
n the recovere
ed drag, migh
ht be due to several facto
ors, which are
e discussed in
n
detaiil in chapter 5.
5

Table 4.3 Comparison


C
o experimenta
of
al to theoreticcal drag

C
Conditionno
:

Dragexp
e
(N)

Draagtheory
(N)

Coefficienttofdrag
coefficient

16.98
8.9%

1
17.9

5.9

0.1997.8

13.619.6%

1
18.0

32.3

0.12827.9

52

CHAPTER 5
CONCLUSION AND FUTURE WORK
Due to the short duration in impulse hypersonic shock tunnels, it is difficult to measure
accurately the aerodynamic forces. Since the test time is very small, force equilibrium may not
be reached between the model and the balance structure. The stress wave force balance
(SWFB) is a method used to analyze the stress waves formed during aerodynamic loading.
5.1 Force Balance in the UTA Hypersonic Shock Tunnel
A force balance was designed for measuring forces on aerodynamic models. In the
present work, drag was measured for a blunt cone using the force balance. The force balance
system included the model, which was fabricated in steel and the force balance, made of 6061
aluminum alloy. Several designs were investigated before the actual fabrication. Some of the
limitations in the design included the size of the balance, amount of load the balance must
withstand, model and support attachment and machining simplicity. The force balance was
designed to fit in the hypersonic shock tunnel test section, allowing room for support
attachments, model attachment and wiring. FEA (Finite Element Analysis) was used to
determine the dynamic characteristics of the force balance under high impact loading. It was
also performed to be certain of the maximum loads the force balance could resist.
Piezoelectric film gages and strain gages were used to measure forces. Both static and
dynamic calibrations were performed. The static calibration involved loading the force balance
with increments of known weights and measuring the strain for each weight. The same
procedure was also done for unloading the weights. These results were plotted to obtain a
linear relationship. Dynamic calibration was done by pulse excitation using an impulse force
hammer. The input by a hammer hit and the output was obtained from the piezoelectric film
53

gages. From the impulse hammer tests an impulse response/ transfer function was determined.
Another dynamic calibration carried out was vertical and horizontal cut weight tests. These tests
involved hanging known weights to the model by steel wires and cutting the wire, thereby
creating a step load. These step loads were deconvoluted with the obtained transfer function to
recover the step input.
The model used in experiments was a blunt cone made of steel. It was designed such
that a pitot pressure measurement was taken simultaneously. The model and balance assembly
was then installed from the top in the test section. The required wiring was taken out of the test
section through drilled bolts, which were sealed. The instrumented gages were shielded to
prevent electromagnetic interference and sealed in rubber and electrical tape. Tests were
conducted with conditions mentioned in chapter 4. All the data signals including the pitot
pressure, force data and pressure transducers in the driven tube (CH1 and CH2) were recorded
simultaneously using an oscilloscope ( see Appendix C).
The strain data was processed and deconvolved with the transfer function, to recover
the drag. Deconvolution was performed using an iterative algorithm [16]. A low frequency
vibration was noticed in the signal of the measured force, before the incident shock reached the
pressure transducers in the driven tubes. From each of the primary tests, this mentioned
vibration occurred, which led to the conclusion that a vibration is due to the test section
movement. An approach was made to tighten down the test section using steel wire rope and
turnbuckles. Tests showed that these oscillations persisted. Therefore, it may be concluded that
stress waves formed by the sudden rupture of the diaphragms have affected the drag
measurements.

54

5.2 Future Work and Recommendations


The following points may be included for future work for force measurements:

Develop an isolation system to reduce the initial vibrations in the force signal.

Force measurement with other models

Analyzing the flow with CFD simulations.

Tests at different enthalpy levels.

Calculate three components of force, such as, lift, drag and pitching moment.

Compensation of inertial forces due to tunnel movement using accelerometers.

Analyze signals in time-frequency representation using wavelet transform.

From the recovered drag, it can be seen that the signal follows the pitot pressure but
that there is also a fluctuation in drag, which is accounted by the occurrence of these low
frequency vibration. Further investigations need to be done on developing methods to decrease
these low frequency oscillations, by using springs, rubber dampers etc. Some of the
recommendation include, but are not limited to tests with simple shield design and compare the
results. Also, test the model at different angle of attacks and compare the drag. Care must also
be taken with the piezoelectric films, since they are very sensitive to EMI (electromagnetic
interference). Any power source close to the tunnel must be avoided, which can result in
erroneous values.

55

APPENDIX A

LIST OF DESIGN DRAWINGS

56

57
Figure A.1 Force balance drawing.

58
Figure A.2 Blunt cone model drawing.

59
Figure A.3 PCB pressure transducer holder drawing.

60
Figure A.4 Hinge joint part 1 drawing.

61
Figure A.5
5 Hinge part 2 draw
wing.

62
Figure A.6 Scoring pattern on steel diaphragm drawing (courtesy Tiago Rolim).

APPENDIX B
MATLAB PROGRAM FOR FORCE ESTIMATION

63

%xxx Program to calculate Axial, Normal force coefficient and L/D xxx%
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
clc
clear all
for j=1:4
in=input('*********************\nTo find Axial,Normal force
coefficient and L/D press 1: \nTo find Cpt press 2:\nTo find Pressure
ratio press 3:\n********************* ');
Rb=1.65;
Rn=0.355;
theta=9;
phi=asin(cos(theta));
r=0;
L=0;
d=Rb/Rn;
gamma=1.4;
M=4.3;
if in==3
%%%xxxxxxxxxxxxxxxxxxx pressure ratio calculation xxxxxxxxxxxxx
Cpt=input('Enter Cpt , if known :
') ;
gamma=input('Enter specific heat ratio :
');
M=input('Enter Mach no: ');
pressure_ratio =(((gamma+1)*M^2)/2)^(gamma/(gamma1))*((gamma+1)/((2*gamma*M^2)-(gamma-1)))^(1/(gamma-1))
x=pressure_ratio;
plot(M,x ,'*b');
xlabel('Mach no: ,M')
ylabel('Pressure ratio , Pt2/P1')
%xxxxxxxxxxxxxxxxxxxxxxxxxxxxxx Cpt calculation xxxxxxxx
elseif in~=(1:3)
return
elseif in==2

x=input('Enter pressure ratio : ');


gamma=input('Enter specific heat ratio : ');
M=input('Enter Mach no:
');
Cpt=(((x)-1)*(2/(gamma*M^2)))

%%%xxxxxxxxxxxxxxxxxxxxxxx L/D
elseif in==1

xxxxxxxxxxxxxxxxxxx

64

for i=1:1
Cpt= input('Enter Cpt: ');
Rb=input('Enter Base Radius: ');
Rn=input('Enter Nose Radius: ');
theta=input('Cone half Angle : ');
phi=asin(cos(theta));
r=0;
x=0;
L=0;
d=Rb/Rn;
gamma=input('Enter specific heat ratio : ');
M=input('Enter Mach no:
');
alpha=-15:0.5:15;
Cp=Cpt*(sind(theta))^2
plot(alpha,Cp,'*r')
hold on
xlabel('Angle of Attack,Alpha')
ylabel('Axial Force Coefficient, Ca')
a=2*Cpt*((Rn^2)/Rb^2);
b=0.25.*(cosd(alpha).^2).*(1(sind(theta)^4))+(0.125.*(sind(alpha).^2)*(cosd(theta).^4));
c=((tand(theta).*((cosd(alpha).^2)*(sind(theta).^2)+0.5.*(sind(alpha).
^2).*(cosd(theta).^2))).*((((dcosd(theta)).*cosd(theta))/tand(theta))+(((dcosd(theta)).^2)/(2*tand(theta)))));
Ca=a*(b+c);
plot(alpha,Ca,'*-r')
hold on
title('Axial Force Coefficient vs Angle of Attack (Alpha)')
xlabel('Angle of Attack, Alpha')
ylabel('Axial Force Coefficient, Ca')
p=2*Cpt*((Rn^2)/Rb^2);
q=0.25.*sind(alpha).*cosd(alpha).*(cosd(theta)^4);
r=(sind(alpha).*cosd(alpha).*sind(theta).*cosd(alpha).*((((dcosd(theta)).*cosd(theta))/tand(theta))+(((dcosd(theta)).^2)/(2*tand(theta)))));
Cn=p*(q+r);
figure
plot(alpha,Cn,'*-black')
title('Normal Force Coefficient vs Angle of Attack (Alpha)')
xlabel('Angle of Attack,Alpha')
ylabel('Normal Force Coefficient, Cn')
CL=((Cn.* cosd(alpha))-(Ca.*sind(alpha)));
CD=((Cn.* sind(alpha))+(Ca.*cosd(alpha)));
L_D= CL./CD
figure
65

plot(alpha,L_D,'*-g')
drawnow
title('L/D ratio vs Angle of Attack (Alpha)')
xlabel('Angle of Attack,Alpha')
ylabel('L/D ')
hold on
end
end
end

66

APPENDIX C

INSTRUMENTATION DETAILS

67

C.1

Data acquisition
Manufacturer: Tektronix Digital Phosphor Oscilloscope.
Model: DPO 4054
Features: Analog bandwidth 500 Mhz
Sample rate 2.5 GS/s
Record length 20 M points
Analog channels 4

C. 2

Strain measurements
Manufacturer: Measurement specialties
Model: DT1-052k
Features: Min. impedance 1M
Output voltage mV to 100s of volt
Operating temp -40 to 60C

Manufacturer: Omega engineering, Inc.


Model: SGD-3/120-LY13
Features: Max Vrms 4.5
Nom. Resistance 120
Gage Factor 2.0 5%
Operating temp -75 to 200C

C.4

Pressure Transducers
Manufacturer: PCB Piezotronics, Inc.
Model: 111A23
Features: Measurement range 10kpsi
Sensitivity 0.5mV/psi
Maximum pressure 15kpsi
Operating temp -73 to 135C

68

Manufacturer: PCB Piezotronics, Inc.


Model: 113A21
Features: Measurement range 200psi
Sensitivity 25mV/psi
Maximum pressure 1000psi
Operating temp -73 to 135C

C.7

Strain gage amplifier


Manufacturer: Paine instruments, Inc.
Model: Strain gage amplifier
Features: Excitation voltage range 0-10V
Gain 100

Amplifier circuit diagram used for piezoelectric films

Figure C.1 Circuit diagram of piezoelectric film amplifier. Circuit uses a LM-386 IC.

69

REFERENCES
[1] Robinson, M., Simultaneous Lift, Moment and Thrust Measurements on a Scramjet in
Hypervelocity Flow, Ph.D. dissertation, University of Queensland, 2003.
[2] Sanderson, S.R. and Simmons, J.M., Drag Balance for Hypervelocity Impulse Facilities,
AIAA Journal, Vol. 29, No. 12, pp. 21852191, 1991.
[3] Daniel, W.J.T. & Mee, D.J., Finite Element Modelling of a Three-Component Force Balance
for Hypersonic Flows, Computers and Structures 54 (1), 35{48}, 1995.
[4] Robinson, M., Schramm, J.M. and Hannemann, K., An Investigation into Internal and
External Force Balance Configurations for Short Duration Wind Tunnels, Notes on
Numerical Fluid Mechanics and Multidisciplinary Design, Volume 96/2008,129-136, 2008.
[5] Boyce, R. R. and Stumvoll, A., Re-entry Body Drag: Shock Tunnel Experiments and
Computational Fluid Dynamics Calculations Compared, Shock Waves, 16 6: 431-443,
2007.
[6] Kulkarni, V. and Reddy, K.P.J., Accelerometer-Based Force Balance for High Enthalpy
Facilities, J. Aerosp. Engrg. 23, 276 doi:10.1061/(ASCE), 2010.
[7] Sahoo,N, Mahapatra, D.R., Jagadeesh, G., Gopalakrishnan, S. and Reddy, K.P.J., Design
and Analysis of a Flat Accelerometer-based Force Balance System for Shock Tunnel
Testing, Measurement, 40 (1).pp.93-106, 2007.
[8] Sahoo, N., Suryavamshi, K., Reddy, K.P.J. and Mee, D.J., Dynamic Force Balances for
Short-Duration Hypersonic Testing Facilities, Experiments in Fluids, 38 (5). pp. 606-614,
2005.
[9] Mee, D.J., Dynamic Calibration of Force Balances, Centre for Hypersonics, The University
of Queensland, Australia. Tech. Rep. 2002/6, Jan 2003.
70

[10] Smith, A. L.; Mee, D.J., Drag Measurements in a Hypervelocity Expansion Tube, Shock
Waves, Volume 6, Issue 3, pp. 161-166,1996.
[11] Marineau, E., Force Measurements in Hypervelocity Flows with an Acceleration
Compensated Piezoelectric Balance, Journal of Spacecraft and Rockets, 0022-4650
vol.48, no.4 (697-700), 2011.
[12] Smith, S.W., The Scientist and Engineer's Guide to Digital Signal Processing.
[Online],http://www.dspguide.com/, 2012.
[13] Murtugudde, R.G., "Hypersonic Shock Tunnel," Master's Thesis, Department of
Aerospace Engineering, The University of Texas at Arlington, Arlington, TX, 1986.
[14] Stuessy, W.S, "Hypersonic Shock Tunnel Development and Calibration," Master's
Thesis, Department of Aerospace Engineering, The University of Texas at Arlington,
Arlington, TX, 1989.
[15] Stuessy, W.S., Murtugudde, R.G., Lu, F.K. and Wilson, D.R., "Development of the UTA
Hypersonic Shock Tunnel," Paper 90-0080, AIAA 28th Aerospace Sciences Meeting,
January 8-11, Reno, Nevada, 1990.
[16] Prost, R., Goutte, R., Discrete Constrained Iterative Deconvolution Algorithms with
Optimized Rate of Convergence, Signal Process.7(3), 209230,1984.
[17] Bertin, J.J. Hypersonic Aerothermodynamics. American Institute of Aeronautics and
Astronautics, Inc., Washington, DC, 1994.
[18] Anderson, J.D. Fundamentals of Aerodynamics. New York, NY: McGraw-Hill, 2001.

71

BIOGRAPHICAL INFORMATION

Pravin Vadassery graduated with a Bachelors degree in Aeronautical engineering, his


endeavor to learn new things, lead him to the Masters degree in Aerospace engineering. His
passion for experiments and hands-on jobs helped him during his research at the Aerodynamic
Research Center. He has worked on many projects during his undergraduate and graduate
years, which included areas of design, analysis, and comparative studies. He plans to start his
career with all experience he gained and eventually establish his own company.

72