 

: ..

.1: Ethyl,
.. .

3
1.. . 5.
3‘: i: .2!

( fut:

V I... .. .4 s
:9...‘
‘11::

4. A
.9 3.4.1...

5.. .

3|:
:3 ‘ .. .
I 3:12.

 

 

 

 

 

 

 

 

 

 

 

. :35.
.:..3:-.
X. 3.2.1.

in...

735...:
;. ~
3 .

:2... 5‘. u :
xx. .1.

a.

n. ,

a:

 

 

 

THtSIs

(Raul

M
l —

' llBRARY
Mu Michigan State
‘ University

 

 

 

This is to certify that the

dissertation entitled

EXPERIMENTAL STUDY OF HEAT TRANSFER AND PHASE CHANGE CONDENSATION [N A
DEVELOPING, TWO-DIMENSIONAL WALL JET FLOW FIELD wrrH AN ISOTHERMAL
BOUNDARY CONDITION

presented by

Paul Bryan Hoke
has been accepted towards fulﬁllment

of the requirements for

PhD. degree in Mechanical Engineering

knit/buzz

ﬂ \ Major professor

MS U i: an Afﬁrmative Action/Equal Opportunity Institution 0-12771

Date J/2’§/0 /

 

 

 

PLACE IN REI'URN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

MAY 0 03

new“? mm
W

M4

 

 

 

 

 

 

 

 

 

 

 

 

 

 

BIOI cJCIFICJDatoDinpes-pts

 

_—-——~

EXPERI

EXPERIMENTAL STUDY OF HEAT TRANSFER AND PHASE CHANGE CONDENSATION
IN A DEVELOPING, Two-DIMENSIONAL WALL JET F Low FIELD
WITH AN ISOTHERMAL BOUNDARY CONDmON

By

Paul Bryan Hoke

A DISSERTATION

Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Mechanical Engineering

2001

Dave;

nears
deﬁes
was U‘

0? the

ABSTRACT

Experimental Study of Heat Transfer and Phase Change Condensation in a
Developing, Two-dimensional Wall Jet Flow Field with an Isothermal Boundary
Condition

By

Paul Bryan Hoke

Benchmark data were obtained in a developing wall jet ﬂow ﬁeld as a
means to validate computational models utilized for automotive
defroster/demister flow design. A non-impinging, two-dimensional wall jet ﬂow
was utilized due to the relatively simplicity of the ﬂow geometry and the similarity
of the boundary conditions to the actual defroster ﬂow ﬁeld.

Isothermal, non-isothermal, non-condensing and non-isothermal,
condensing ﬂow regimes were measured and compared to results obtained from
a two-dimensional computational model. The experimental, mean velocity
proﬁles at streamvvise locations greater than 7.62 slot widths downstream agreed
with results in the literature to within 1% [Launder and Rodi, 1981]. The
computational models tended to over-predict velocity and temperature gradients
at the wall surface and over-predicted growth of the inner shear region by as
much as 60% while under-predicting the jet-spreading rate. The model
predictions of the mean velocity and the streamwise-normal stress agree within

10% and 20% respectively throughout the developing region investigated here.

I
uﬁizng
neaSur
accura:
tectniq
tathl
pmhde

otcrn

conder

A new technique was developed to measure small aliquots of water
utilizing headspace gas chromatography with a flame ionization detector. This
measurement technique provided measurements of water from 0 to 3 mg with an
accuracy of $0.038 mg at a 95% conﬁdence level for the regression model. This
technique is more robust and exceeds the capability of gravimetric equipment
that is readily available. The development of this technique was essential to
provide an accurate measure of condensate in the required range and to meet
other requirements, such as specimen isolation, of the experimental program.

A novel optical technique was developed to measure transient, local
condensate thickness on reflective surfaces. Surface reﬂectance measurements
to 8-12 pm wavelengths (infrared) were correlated with H20 condensate
thickness present on the surface. This technique proved to be highly sensitive
with a dynamic range from 0 to 5 pm. Measurement accuracy depends on the
target utilized. Accuracy of :t: 0.17 pm at a 95% conﬁdence level was attained for
ensemble surface calibrations (multiple target average). It appears that a single
calibrated target would provide greater measurement accuracy. This technique
was utilized to provide measurements of transient condensation development in
the wall jet flow and estimates of the local, instantaneous mass transfer
coefﬁcient were calculated. Results indicated that the technique is an extremely
sensitive condensate thickness indicator but that uncertainty in the mass transfer
coefﬁcient can be of the same order of magnitude or larger than the
measurement due to the potential error in determining the concentration

difference in the ﬂow ﬁeld.

Copyright by
Paul Bryan Hoke
2001

To Carol, Madeleine, and Isabelle

and Si
900d ‘
helpln
has hi

have I

his tlm
often r

DGFSpe

LOCont.
p°$sibl
18::th

and ma

Sappm

ACKNOWLEDGMENTS

I could never have accomplished this body of work without the assistance
and support of my family, mentors, friends and colleagues. I promised my very
good friend Alptekin Aksan that he would be ﬁrst in this list once when he was
helping me out of a jam and indeed his assistance has been beyond measure, as
has his friendship. Scott Morris was also invaluable as a resource and a friend. I
have learned more in discussions with Scott than lever did in class.

Dr. John McGrath, my advisor and mentor, gave me support, advice and
his time. I thank him for that; he was always available. Dr. Craig Somerton was
often my moral support and my listening ear. He often helped me keep
perspective during this process. I'd also like to thank the other members of my
committee, Dr. Ahmed Naguib and Dr. Dennis Gilliland. They always took the
time to answer my questions and share their experience and knowledge. Dr.
Bashar AbdulNour had the conﬁdence in me to supply us with the funding to
make this projectpossible. Dr. AbdulNour is a wonderful advocate and his input
has been invaluable.

Others here at MSU offered assistance, input and advice. Dr. Paul
Loconto and Yan Pan made the analytical chemistry portions of this thesis
possible. Arno Kinnen was a joy to work with, I will always remember our
technical discussions with a smile. Doug Neal, Damien Fron, Molly Desjardins
and many, many others have made this a wonderful place to work and learn.

Finally, I'd like to thank my family, immediate and extended. Their undying

support kept me going and gave me conﬁdence in myself.

vl

TABLE OF CONTENTS

List of Tables ....................................................................................................... ix
List of Figures ....................................................................................................... x
Nomenclature .................................................................................................... xvii
Introduction ........................................................................................................... 1
Chapter 1. Statement of problem and review of literature ..................................... 4
1.1 STATEMENT OF PROBLEM .................................................................................. 4
1.2 REVIEW OF LITERATURE ..................................................................................... 9
Chapter 2. Facility development and characterization ........................................ 17
2.1 FACILITY DESCRIPTION ..................................................................................... 19
2.2 EXPERIMENTAL CAPABIerIES ............................................................................ 23
2.3 DATA ACQUISITION AND MOTION CONTROL ......................................................... 25
2.4 MEASUREMENT PARAMETERS AND CONTROL ..................................................... 26
2.4.1 PRESSURE MEASUREMENT ............................................................................ 26
2.4.2 VELOCITY MEASUREMENTS ............................................................................ 27
2. 4. 3 TEMPERATURE MEASUREMENT ...................................................................... 30
2.4.4 HUMIDITY MEASUREMENT AND CONTROL ........................................................ 31
2. 4. 5 CONDENSA TION VISUAUZA TION ...................................................................... 32
2.5 THERMALLY ACTIVE TEST SECTION .................................................................... 33
2.6 CALIBRATION AND CHARACTERIZATION EXPERIMENTS ......................................... 36
2.6.1 TEST CONTRACTION MEASUREMENTS ............................................................. 36
2. 6.2 SPATIAL FLOW UNIFORMITY MEASUREMENTS ................................................... 38
2. 6. 3 TEMPORAL STABILITY OF THE FMTF TEMPERATURE AND HUMIDITY ................... 41
Chapter 3. lsothennal flow .................................................................................. 44
3.1 EXPERIMENTAL RESULTS .................................................................................. 45
3.2 COMPUTED RESULTS AND COMPARISON To EXPERIMENTAL DATA ......................... 52
3.3 DISCUSSION OF RESULTS ................................................................................. 56
Chapter 4. Steady state, non-isothermal, non-condensing ﬁow .......................... 60
4.1 EXPERIMENTAL METHODS ................................................................................ 60
4.2 EXPERIMENTAL UNCERTAINTY ........................................................................... 63
4.3 EXPERIMENTAL RESULTS OF THE NON-ISOTHERMAL, NON-CONDENSING

EXPERIMENTAL REGIME .......................................................................................... 65
4.3.1 NON-ISOTHERMAL VELOCITY PROFILES ........................................................... 68
4.3.2 NON-ISOTHERMAL TEMPERATURE PROFILES ................................................... 72
4.4 SUMMARY OF EXPERIMENTAL RESULTS .............................................................. 77
4.5 COMPARISON OF EXPERIMENTAL DATA To CF D RESULTS ................................... 79

vii

Chapter 5. Steady state, non-isothermal, condensing flow ................................. 83
5.1 TEMPERATURE AND VELOCITY MEASUREMENTS .................................................. 83
5.2 STEADY STATE MASS TRANSFER MEASUREMENTS ............................................... 88

Chapter 6. Optical, transient mass transfer measurement utilizing infrared

thermography .................................................................................................... 100
6.1 EXPERIMENTAL METHODS .............................................................................. 106
6. 1. 1 MEASUREMENT OF TRACE CONDENSATE ...................................................... 107
6. 1. 2 CAUBRA TION OF THE SURFACE REFLECTIVITY AS A FUNCTION OF CONDENSATE
THICKNESS ......................................................................................................... 116
6.1.3 EFFECT OF DISTANCE AND ANGLE ON THE IR-HZO OPTICAL CAUBRA TION ....... 123
6.2. VALIDATION OF THE OPTICAL MEASUREMENT CALIBRATION WITH A
NON-HOMOGENEOUS CONDENSATE FIELD ..................................................... 124
6. 2. 1. NON-HOMOGENEOUS BtPERIMENTAL PROCEDURE AND RESULTS .................. 130
Chapter 7. Optical, transient mass transfer measurements
in the developing wall jet ................................................................................... 135
7.1 INTRODUCTION .............................................................................................. 135
7.2 EXPERIMENTAL CONFIGURATION ..................................................................... 135
7.3 TRANSIENT OPTICAL DATA .............................................................................. 138
7.4 RESULTS OF THE TRANSIENT CONDENSATION EXPERIMENTS IN THE F MTF ......... 142
7.5 SHERWOOD NUMBER DEPENDENCE ON JET REYNOLDS NUMBER ........................ 152
7.6 UNCERTAINTY ESTIMATES FOR THE OPTICAL INFRARED
REFLECTANCE TECHNIQUE .................................................................................... 154
Chapter 8. Summary ......................................................................................... 159
8.1 FUTURE WORK .............................................................................................. 163
Appendix A. Cold wire calibration device and methodology .............................. 168
Appendix 8. Equipment and Settings Utilized
for Reflectance Calibration Experiments ........................................................... 175
Appendix C. Control settings utilized for GC-F ID with Automated Sampler ...... 177
Appendix D. Comparison of the heat transfer associated
with the phase change event compared to the sensible heat transfer .............. 179
Appendix E. Catalog of condensation and mass transfer measurements
obtained with the optical infrared reflectance technique ................................... 183
Appendix F. Error analysis in the Sherwood number determined
with the optical infrared reﬂectance technique .................................................. 195
Appendix G. Publications related to this work ................................................... 205

viii

LE?

Cha
Tabl
FAT?

Cha;
Tabk
intte
Tabk

ADDEI
Table

docun

LIST OF TABLES

Chapter 2

Table 2.1. Experimental parameter control and measurement capabilities of the
FMTF .................................................................................................................. 24
Chapter 5

Table 5.1. Comparison of experimental and computed velocity gradients

at the thermally active surface ............................................................................ 98
Table 5.2. Comparison of experimental and computed temperature gradients

at the thermally active surface ............................................................................ 98
Chapter 7

Table 7.1. Variables and uncertainties utilized to calculate the uncertainty

in the Sherwood number ................................................................................... 155

Table 7.2. Contribution of the experimental variables to the total uncertainty .. 157

Appendix F
Table F.1. Mathematica” variable names and corresponding variables in

document .......................................................................................................... 204

Craz
Figu

in ail

LIST OF FIGURES

Chapter 1

Figure 1.1. Ratio of latent heat transfer to sensible heat transfer for water vapor
in air ...................................................................................................................... 8
Chapter 2

Figure 2.1. Schematic of the wall jet ﬂow ﬁeld in the FMTF. ............................. 18
Figure 2.2. The Ford MSU Test Facility (FMTF) ................................................ 19
Figure 2.3. Photograph of the Ford MSU Test Facility. ..................................... 20
Figure 2.4. Thermally active test section - aluminum heat exchanger. ............. 22

Figure 2.5. Schematic of the data acquisition and motion control systems. ...... 25
Figure 2.6. Hot-wire conﬁguration for two wire probe technique; top view;

(Flow direction in x-direction, into the page). ...................................................... 28
Figure 2.7. Temperature reference and thermocouple calibration equipment. ..31
Figure 2.8. Interior ﬂow pattern of the thermally active test section. ................. 34
Figure 2.9. Bypass loop schematic for transient thermal tests .......................... 35

Figure 2.10. Transient temperature history of the thermally active
test section .......................................................................................................... 35
Figure 2.11. Comparison of test contraction velocity (from pressure tap

measurements) and Pitot tube measured velocity .............................................. 38
Figure 2.12. Z-direction (transverse) velocity proﬁle; y/w = 0.5 ......................... 40
Figure 2.13. Jet temperature and humidity uniformity study (w = 2 cm). ........... 41

Figure 2.14. Typical transient response data for the wall jet ﬂow in the FMTF..43

Chapter 3

Figure 3.1. Deﬁnitions of features and zones in the two-dimensional

wall jet ﬂow .......................................................................................................... 45
Figure 3.2. Non-dimensional velocity proﬁle at xlw = 0.0. ................................. 47
Figure 3.3. Non-dimensional velocity proﬁle at xlw = 1.59. ............................... 48
Figure 3.4. Non—dimensional velocity proﬁle at xlw = 3.18. ............................... 48

'ﬂ'ﬂ'ﬂm

RI
Fi
at
Fi
at
Fig
Fit
to 1

Chi
Fig
Fig
COD
Fig
CDC
Fig
Fig
Fig
Fig
Fig
Fig
Fig
Fig.
Fig1
Fig;
at X:

Figure 3.5. Non-dimensional velocity proﬁle at xlw = 4.45. ............................... 49
Figure 3.6. Non-dimensional velocity proﬁle at xlw = 7.62 ................................ 49
Figure 3.7. Non-dimensional velocity proﬁle at xlw = 10.8 ................................ 50
Figure 3.8. Comparison of experimental data to results from the literature. ...... 51
Figure 3.9. Computed velocity results for the wall jet ﬂow ﬁeld.

Reynolds stress model utilized ........................................................................... 53
Figure 3.10. Comparison of experimental and computed velocity results

at xlw = 0.0 ......................................................................................................... 54
Figure 3.11. Comparison of experimental and computed velocity results

at xlw = 10.8. ...................................................................................................... 54
Figure 3.12. Comparison of jet spreading rates (location of 52) .......................... 55
Figure 3.13. Boundary layer proﬁles at xlw = 0.0 compared

to the Biasius laminar proﬁle ............................................................................... 56
Chapter 4

Figure 4.1. Schematic of the hot-wirelcold-wire dual probe sensing head. ........ 63
Figure 4.2. Comparison of velocity proﬁles in the non-isothermal, non-

condensing and isothermal experimental regimes; X/w = 10.8 ............................ 66
Figure 4.3. Comparison of the self preserving momentum proﬁle in inner

coordinates to data from Wygnanski et al. [1992]; xlw = 10.8 ............................. 67
Figure 4.4. Velocity proﬁle at xlw = 0.0; non-isothermal; T. = 4 °C; n = 1 .......... 69

Figure 4.5. Velocity proﬁle at xlw = 1.59; non-isothermal; T. = 4 °C; n = 2 ....... 70
Figure 4.6. Velocity proﬁle at xlw=4.45; non-isothermal; T. = 4 °C; n = 1 ......... 70
Figure 4.7. Velocity proﬁle at xlw = 7.62; non-isothermal; T. = 4 °C; n = 4 ........ 71
Figure 4.8. Velocity proﬁle at xlw = 10.8; non-isothermal; T, = 4 °C; n = 3. ....... 71
Figure 4.9. Non-dimensional temperature proﬁle at xlw = 1.59; T. = 4 °C ......... 75
Figure 4.10. Non-dimensional temperature profile at xlw = 4.45; T8 = 4 °C ....... 75

Figure 4.11. Temperature proﬁle at xlw = 7.62; T. = 4 °C .................................. 76
Figure 4.12. Temperature proﬁle at xlw = 10.8; T. = 4 °C. ................................. 76
Figure 4.13. Comparison of the experimental and computed temperature proﬁles
atx/w=1.59;T,=4°C ....................................................................................... 80

Figul
at XIII
Figur

at AM

Figure 4.14. Comparison of the experimental and computed temperature proﬁles

at xlw = 4.45; T. = 4 °C. ...................................................................................... 80
Figure 4.15. Comparison of the experimental and computed temperature proﬁles
at xlw= 10.8; T.=4°C ....................................................................................... 81
Chapter 5

Figure 5.1. Comparison of velocity proﬁles in the non-isothermal, condensing
and isothermal experimental regimes. ................................................................ 85
Figure 5.2. Comparison of temperature proﬁles in the condensing and non-
condensing, non-isothermal experimental regimes; T, = 4 °C ............................ 86
Figure 5.3. Steady state condensation - collection and measurement system... 89
Figure 5.4. Comparison of the experimental and computed steady-state
condensation results ........................................................................................... 91
Figure 5.5. Temperature variation on the thermally active test section for a

constant temperature circulating bath temperature of 34 °C ............................... 94
Figure 5.6. Average, steady-state ShenIvood number comparison between the
experimental, computational, and published results ........................................... 96
Chapter 6

Figure 6.1. Thin ﬁlm absorption model. ............................................................ 101
Figure 6.2. Thin ﬁlm model results of the Inframetrics 600L sensitivity ............ 104

Figure 6.3. Structure of the condensate formation on the thermally active test
section .............................................................................................................. 105
Figure 6.4. Typical chromatogram obtained during HS-C-GC-F ID

calibraﬁon ......................................................................................................... 111
Figure 6.5. Typical calibration result for GC-F ID system response to the

volume of water added. (n = 11 calibration samples) ........................................ 113
Figure 6.6. Schematic of the lR-H20 calibration experiment ............................ 117
Figure 6.7. Photograph of the lR-Hzo calibration experiment .......................... 118

Figure 6.8. Ensemble correlation of reflectance ratio to estimated thickness
of condensate present ...................................................................................... 120

xii

Fig
P'9
Fig
one
Fig
ﬁon
ka
RESL
Figl
beﬁ:
FigL
non-
Figu
Duet

each

Cha;
Figul
Figui
dunn
Figu,
tans
Figu.
i852 s
FiQUI
test s
Fi9m
SURE:
FiQUr
and ti
“PEI

Figure 6.9. Correlation of reﬂectance ratio to estimated thickness of condensate
present for each aluminum target utilized ......................................................... 122
Figure 6.11. Temperature distribution in aluminum target derived from a
one—dimensional ﬁn analysis ............................................................................. 127
Figure 6.12. Temperature distribution in 0.8 mm thick, aluminum target derived
from a one-dimensional ﬁn analysis ....... - ........................................................... 128
Figure 6.13. Concentration gradient distribution along the aluminum target
resulting from the ﬁn temperature distribution ................................................... 129
Figure 6.14. Non-homogeneous concentration ﬁeld on an aluminum target
before and after a portion was dried with nitrogen gas ..................................... 131
Figure 6.15. Measured thickness results on aluminum target #4 during a
non-homogeneous condensation ﬁeld test ....................................................... 132
Figure 6.16. Results of the non-homogeneous concentration ﬁeld tests utilizing
the HS-C-GC-FID technique as the reference standard. Coefﬁcient of variation in

each non-uniform image presented .................................................................. 133
Chapter 7

Figure 7.1. Experimental setup in the Ford-MSU test facility. ......................... 136
Figure 7.2. Surface temperature history of the thermally active test section

during a typical transient condensation experiment ......................................... 137
Figure 7.3. Typical infrared images of the thermally active test surface during
transient condensation experiment and the resulting surface reﬂectance ......... 139
Figure 7.4. Reﬂectance measurements at t = Os and t = 69s of thermally active
test surface ....................................................................................................... 140
Figure 7.5. Typical condensate thickness measurements on the thermally active
test surface during transient condensation (T j = 24.1 °C. %th = 51.1) ........... 141
Figure 7.6. Typical temperature proﬁle at t = 59s on the thermally active test
surface during transient condensation (T j = 24.1 °C. %th = 51.1). ................. 144

Figure 7.7. Typical concentration difference at t = 593 between the wall jet ﬂow
and the thennally active test surface during transient condensation
experiment ........................................................................................................ 145

xiii

Figure 7.8. Typical ShenNood number results on the thermally active test surface

during transient condensation experiment (T, = 24.1 °C, %Rh,- = 51.1) ............ 147
Figure 7.9. Compiled ShenNood number results on the thermally active

test surface (Rew=13300) .................................................................................. 148
Figure 7.10. Comparison of experimental and published [Mabuchi and Kumada,
1972] Sherwood number results ....................................................................... 149
Figure 7.11. Experimental Sherwood number results ﬁtted

with a ﬁfth order curve ....................................................................................... 151
Figure 7.12. Experimental Sherwood number results as a function of the

jet Reynolds number ......................................................................................... 153

Figure 7.13. Uncertainty of the experimental Sherwood number results as a
function of the temperature of the thermally active wall surface (T.).

Tm... ~ 283 K ................................................................................................ 156
Appendix A

Figure A.1. Schematic of a cold-wire probe ..................................................... 168
Figure A.2. Schematic of the cold-wire calibration device ................................ 170
Figure A.3. Temperature conditioning section of the cold-wire calibrator ......... 171
Figure A.4. Cold-Wire calibration device ........................................................... 171

Figure A.5. Temperature proﬁles at the exit plane of the four
calibration nozzles ............................................................................................ 172
Figure A.6 Time-averaged mean temperature and standard deviation of the cold

wire measurement as a function of time ........................................................... 173
Appendix C

Figure C.1. Analytical syringe and 22 mL headspace vial and crimp cap utilized
for chromatography experiments ...................................................................... 178
Appendix E .

Figure E.1. Condensate thickness results for Rew = 6560 test

(11 June 2001. Experiment #2) ......................................................................... 184

I...
r

Figure E.2. Temperature log for ReW = 6560 test

(11 June 2001. Experiment #2) ......................................................................... 184
Figure E.3. Sherwood number results for ReW = 6560 test ( 11 June 2001.
Experiment #2) .................................................................................................. 185
Figure E.4. Condensate thickness results for Rew = 8400 test

(15 June 2001. Experiment #4) ......................................................................... 185
Figure E.5. Temperature log for R9,, = 8400 test

(15 June 2001. Experiment #4) ......................................................................... 186
Figure E.6. Sherwood number results for ReW = 8400 test (15 June 2001.
Experiment #4) .................................................................................................. 186
Figure 5.7. Condensate thickness results for Rew = 10200 test

(15 June 2001. Experiment #3) ......................................................................... 187
Figure E.8. Temperature log for ReW = 10200 test (15 June 2001. Experiment
#3) ..................................................................................................................... 187
Figure E.9. Sherwood number results for Re... = 10200 test (15 June 2001.
Experiment #3) .................................................................................................. 188
Figure E.10. Condensate thickness results for Rew = 13300 test

(15 June 2001. Experiment #2) ......................................................................... 188
Figure E.11. Temperature log for ReW = 13000 test

(15 June 2001. Experiment #2) ......................................................................... 189
Figure E.12. ShenIvood number results for Rew = 13300 test ( 15 June 2001.
Experiment #2) .................................................................................................. 189
Figure E.13. Condensate thickness results for Rew = 13300 test

(5 June 2001. Experiment #1) ........................................................................... 190
Figure 5.14. Temperature log for Rew = 13000 test

(5 June 2001. Experiment #1) ........................................................................... 190
Figure E.15. Sherwood number results for ReW = 13300 test (5 June 2001.
Experiment #1) .................................................................................................. 191
Figure E.16. Condensate thickness results for ReW = 13300 test

(20 June 2001. Experiment #1) ......................................................................... 191

Figure 5.17. Temperature log for Re... = 13300 test (20 June 2001.

Experiment #1) .................................................................................................. 192
Figure E.18. Sherwood number results for Re... = 13300 test (20 June 2001.
Experiment #1) .................................................................................................. 192
Figure E.19. Condensate thickness results for Re... = 13300 test

(20 June 2001. Experiment #2) ......................................................................... 193
Figure 6.20. Temperature log for Re... = 13300 test (20 June 2001.

Experiment #2) .................................................................................................. 193
Figure E.21. Sherwood number results for Re... = 13300 test (20 June 2001.
Experiment #2) .................................................................................................. 194

NOMENCLATURE

A .............................................................................................................. Area (m2)
AID .............................................................................. Analog to digital conversion
D .................................................................................... Diffusion coefﬁcient (m zls)
DIA .............................................................................. Digital to analog conversion
C .......................................................................................... Concentration (kg/m3)
CFD ......................................................................... Computational ﬂuid dynamics
cp ...................................................... Speciﬁc heat at constant pressure (J kg'1K' )
E ........................................................................................................... Voltage (V)
f ...................................................................................................... Frequency (Hz)
F D .................................................................................. Flame ionization detector
FMTF .................................................................................... Ford—MSU test facility
HS ......................................................................................................... Heads ace
h ...................................................................... Heat transfer coefﬁcient (W m’ K")
h... ............................................................................ Mass transfer coefﬁcient (mls)
IR ................................................................................................................ infrared
k ........................................................................... Thermal conductivity (W rn'1 K‘ 1)
k-e .......................................................................................... k-e turbulence model
9 ........................................................................... Gravitational acceleration (In/$2)
Go ..................................................................................... incident radiation (Wlm2 )
GC ........................................................................................ Gas chromatography
J ................................................................................................... Radiosity (Wlm )
j" ............................................................................................. Mass ﬂux (kg m‘2 s' 1)
L ................................................................................................ Latent heat (KJlkg)
L .................................................................................................. Jet slot length (m)
L ................................................................ Thermally active test section length (m)
Le ............................................................................................ Lewis Number (ct/D)
m ............................................................................................................. mass (kg)
MTBE ................................................................................... Methyl tert-butyl ether
n .............................................................................................. Number of data sets
P ..................................................................................................... Pressure (kPa)
ppm ............................................................................................... Parts per million
q" .................................................................................................. Heat ﬂux (Wlmz)
q..." ................................................. Heat ﬂux associated with mass transfer (Wlm: )
R .................................................................................... Gas Constant (kJ kg'1 K )
Rex .................................................................................. Reynolds Number (U xlv)
RSM .................................................................. Reynolds stress turbulence model
Shx ............................................................................... Sherwood number (h... x/D)
SNS ..................................................... Streamwise normal stress (100 (u' liJ.......2 ))
St ....................................................................................... Strouhal number (f BIU)
T ........................................................................................... Temperature (°C or K)
T ........................................................................................................ Thickness(m)
t .................................................................................................................. Time (s)
U .................................................................................... Streamwise velocity (mls)

xvii

 

U+ ............................................................................................ Shear velocity (mls)

u' ...................................................... Root mean square streamwise velocity (mls)
V .................................................................... Velocity normal to test surface (mls)
V ....................................................................................................... Velocity (mls)
W ...................................................................................... Spanwise velocity (mls)
w ........................................................................................... Wall jet slot width (m)
X (or x) ............................................................... Streamwise direction/location (m)
A ................................................................................................... Mean value of A
Y (or y) ............................................... Normal to test surface direction/location (m)
y ............................................................................................ Wall coordinates (m)
Z (or z) .................................................................. Spanwise direction/location (m)
%Rh ............................................................................................. Relative humidity
Greek

(1 .......................................................................................................... Absorptivity (m")
e ...................................................................................................................... Emissivity
A ............................................................................................... Delta (e.g. X... - X.)
82 .................................................... Distance from test surface where U = 0.5 Um
p ................................................................................................... True mean value
p ...................................................................................................... Density (kg/m3)
p ........................................................................................................... Reﬂectance
v ..................................................................................... Kinematic viscosity (mzls)
c .............................................. Stephan Boltzmann constant (5.67X10‘8 W rn'2 K")
o ................................................................................................ Standard deviation
9 ............................................................................... Non-dimensional temperature
9 ...................................................................................... Momentum thickness (m)
9' ................................................. root mean square Non-dimensional temperature
to .......................................................................... Speciﬁc humidity (kg Hzolkg air)
SUBSCRIPTS

cal .......................................................................................................... Calibration
cw ............................................................................................................ Cold-wire
f ....................................................................................................................... Final
hw .............................................................................................................. Hot-wire
j ......................................................................................................... Jet conditions
jet ...................................................................................................... Jet condih'ons
m .................................................................................................................... Mirror
max .......................................................................................................... Maximum
0 ............................................................................................................. Initial (t=0)
ref .......................................................................................... Reference conditions
3 .................................................................................................................. Surface
sat ........................................................................................................... Saturation

xvili

t ..................................................................................................................... Target

tot .................................................................................................................... Total
w ................................................................................................................... Water
w ..................................................................................................................... Wire
co ............................................................................................... Far ﬁeld conditions

datats
oi defr
Deﬁes

miﬁnL

 

 

INTRODUCTION

The primary objective of this investigation is to establish an experimental
database to be used to enhance computational ﬂuid dynamics (CFD) simulations
of defrosting/demisting on the interior surfaces of automotive Windshields.
Defroster/demisting performance is a key factor in customer satisfaction and
minimum performance standards have been mandated by the federal
government to ensure passenger safety. To develop accurate computational
capabilities, comprehensive ﬂow and condensation investigations are necessary.
CFD simulation models require experimental data for calibration and veriﬁcation
before they can be declared valid for the engineer's use in design and
development. One of the primary goals of this effort was to contribute to the
improvements of these analytical tools at Ford Motor Company.

A ﬂow facility has been constructed as a joint venture between Ford Motor
Company and Michigan State University to investigate the momentum, heat, and
mass transfer in a developing region of a two-dimensional wall jet ﬂow. The
facility generates a wall jet ﬂow on a thermally active test section inside a sub-
atmospheric chamber. The humidity and velocity of the wall jet as well as the
temperature of the thermally active test section can be controlled. Phase change
occurs as the humid jet ﬂows across the cooled thermally active test section.
Three boundary layers (momentum, temperature, and water concentration)
develop along the length of the test surface. These are interrogated

experimentally.

proﬁie
non45
regnné
comp.
locatio
jet res.
tended
over-p'

model 1

next. T
and we
Spaﬁah.
compute
transfer

this Dre:

tempera

 

"licknes.
Capable

measure

P

—_-_————

The ﬁrst objective of this research is to obtain velocity and temperature
proﬁle measurements in three different experimental regimes. Isothermal ﬂow,
non-isothermal, non-condensing ﬂow, and non-isothermal, condensing ﬂow
regimes were measured and compared to results from a two—dimensional
computational model. The experimental, mean velocity proﬁles at streamwise
locations greater than 7.62 slot widths downstream agreed to within 1% with wall
jet results in the literature [Launder and Rodi, 1981]. The computational models
tended to over-predict velocity and temperature gradients at the wall surface and
over-predicted growth of the inner shear region by up to 60%. The computational
model tended to under-predict the jet spreading rate by approximately 30%.

The mass transfer to the thermally active test section was investigated
next. The steady state mass transfer to the surface was measured by collecting
and weighing the condensate run off. These results provide a measure of the
spatially averaged, steady state mass transfer coefﬁcient of 0.0407 mls. The
computational model over-predicted the spatially averaged, steady state mass
transfer coefﬁcient by approximately 50%. A zonal turbulence model corrected
this prediction but did not address the over estimation of the velocity and
temperature gradients at the wall surface.

A novel optical technique was developed to measure condensate
thickness on reﬂecﬁve surfaces in the ﬁnal phase of the project. The technique
capable of obtaining quantitative measurements of the condensation onset was
desired to study the windshield fogging phenomenon. Surface reﬂectance

measurements in the 8-12 pm wavelength range (infrared) were correlated to

prc
Ilii

0.1

at:

tra'

an
unc
ma;
den
the

stat
con
Iher

test

bee.-

H20 condensate thickness present on the surface of test targets. The technique
proved to be highly sensitive with a dynamic range of 0 to 5 pm of condensate
thickness. Measurement accuracy depends the target utilized. Accuracy of i
0.17 pm at a 95% conﬁdence level was attained for ensemble surface
calibrations (multiple target average). it appears that greater accuracy would be
attainable with a single calibrated target. This technique was utilized to measure
transient condensation development in the wall jet ﬂow and estimates of the
mass transfer coefﬁcient were calculated. Results indicate that the technique is
an extremely sensitive method to measure condensate thickness but that the
uncertainty in the mass transfer coefﬁcient can be of the same order of
magnitude or larger than the measurement. This is due to the potential error in
determining the concentration gradient in the ﬂow ﬁeld. The results obtained with
the optical infrared reﬂectance technique agree with the results of the steady
state mass transfer experiments. The order of magnitude of both results are
consistent with the literature but the mass transfer coefﬁcient distribution on the
thermally active test surface appears to be strongly inﬂuenced by the Ford-MSU
test facility.

Note that this thesis contains color images. Symbols or shading have

been used to differentiate between results where possible.

Chapter 1

STATEMENT OF PROBLEM AND REVIEW OF LITERATURE

1.1 STATEMENT OF PROBLEM

The motivation for this project developed from a model study of an
automotive defroster/demister system. Experimental benchmark data in a model
ﬂow with geometrical similarity to an automotive defroster was desired. These
data provide the means to calibrate and validate the computational ﬂuid
dynamics (CFD) software models utilized for defroster design. Turbulence
closure models used in the CF D software, such as the k-e model, have
parameters that are typically calibrated to agree with ﬂat plate boundary layer
data and/or other canonical ﬂows. The defroster ﬂow ﬁeld differs signiﬁcantly
from the calibration ﬂow ﬁelds. The model results are therefore suspect in the
application flow. The intent of the research was to provide a benchmark data set
in a model ﬂow ﬁeld that more closely resembles an automotive
defroster/demister system. These data could then be utilized to calibrate the
turbulence model or, at the very least, to offer insights on how the CF D model
might differ from the actual ﬂow ﬁeld.

Examples of important technological processes combining heat, mass and
momentum transfer include vehicle windscreen defrosting/defogging, heating and
air conditioning, accumulation on control or ﬂight surfaces on aircraft,
refrigeration processes, boilers, and heat exchangers. Mass transfer can cause

other secondary effects beyond the enhancement of the heat transfer. For

example, visibility can become a critical safety factor if phase change occurs on a
windshield during vehicle Operation. Similarly, control surface icing in aircraft can
create a signiﬁcant hazard. Defogger and defroster design is also of critical
importance to customer satisfaction in automotive applications [AbdulNour,
1999]. Investigation of automotive defroster/demister performance motivated
this project.

The approach adopted here was to start with the simplest case to be
followed by later development of the experimental investigation into more
complicated ﬂow regimes. Initially, the isothermal ﬂow ﬁeld was studied. Thermal
effects were then introduced into in the ﬂow ﬁeld by lowering the temperature of
the thermally active test section by circulating chilled water through its interior.
Finally, mass transfer and phase change (condensation) were then incorporated
into the experiment by suppressing the temperature of the thermally active test
section below the dew-point temperature of the air in the ﬂow.

The isothermal ﬂow ﬁeld data obtained provides validation for the
computational ﬂuid dynamics code and the turbulence model(s) utilized. The
conjugate momentum and heat transfer problem was then investigated by
creating a thermal boundary layer in the ﬂow. The validity of using a decoupled
(velocity solution independent of the temperature solution) solution scheme for
the momentum and energy equations was evaluated experimentally using these
data. Finally, the ﬂow with momentum, heat and mass transfer was interrogated
experimentally and the data obtained utilized for comparison and validaﬁon of the
CFD models.

Heat transfer can be enhanced by an order of magnitude by the
introduction of a phase change at a surface [Legay-Desesquelles and Prunet-
Foch, 1986]. Thus thermal system performance and energy exchanges can be
improved by orders of magnitude. It is important to note that the energy or heat
transfer associated with the mass transfer, even if the total amount of mass is
small, is typically greater than the sensible heat transfer. This is due to the large
latent energy release or absorption required to effect the change of phase.

The following derivation demonstrates the relative importance of the
energy transfer associated with mass transfer as part of the overall heat transfer
in a convective condensation phenomenon. Mass transfer and heat transfer can
be modeled in an analogous fashion [Bejan, 1995]. Speciﬁcally, the heat transfer
and mass transfer coefﬁcients can be related as shown in Equation 1.1

[lncorpera and DeWitt, 1990]

k 1
_ = Lei-n, ~_
h DLe“ pc" n 3 (1'1)

 

For a condensation processes consisting of water vapor in air, the ratio of
hlh... is equal to approximately 1000. If the heat ﬂuxes associated with the
sensible heat transfer and the heat transfer associated with the phase change
are considered however, the contribution from the latent heat (L) must also be
accounted for. The latent heat for water vapor at 283 K is 2477 lekg [Cengel
and Boles, 1989]. The sensible heat ﬂux (q") and the heat ﬂux associated with

the latent heat (q...") are presented in Equations (1.2) and (1.3) respectively.

q": hA’I‘ (1 .2)

qm"= LhmAC (1.3)
The latent heat transfer is a function of the concentration gradient in the
ﬂow ﬁeld. The concentration gradient in the ﬂow is a function of the temperature
gradient. Thermodynamic equilibrium is assumed to exist everywhere in the ﬂow
and speciﬁcally at the chilled surface. The concentration gradient can then be
written in terms of the temperature gradient utilizing the Clausius-Clapeyron
equation [Cengel and Boles, 1989] if the temperature and humidity of the free
stream (ambient) air and temperature of the chilled surface are known. The ratio
of the latent heat transfer (Equation (1.3)) to the sensible heat transfer (Equation

(1.2)) can be expressed as shown in Equation (1.4).

 

 

 

(I... =
q"
‘l'-'I I-i ‘
%Rhiekw TM Tj PTot-Pref Rw TM Tj L[ I l]
Lhum. _e"" Tet Ts (1.4)
h(T,- -T.)Rw L_[-L__'_] T.
Rw Tref Tj
T]. PTm_me%the

 

 

Equation (1.4) incorporates three assumptions: that the density of the
liquid water is much greater than water vapor, the relative humidity (%Rh.) at the
chilled surface is 100%; and that the water vapor/air mixture can be treated as an
ideal gas. The complete derivation of Equation (1.4) is presented in Appendix D.
Figure 1.1 displays the ratio of the latent heat transfer to the sensible heat
transfer as a function of free stream temperature (T.) with a surface temperature
of 277 K and a jet humidity (%Rh.) of 70%. These conditions are typical of the
tests that were conducted and of appropriate scale for the windshield
condensation problem. Note that the plot starts at the dew-point temperature for
the conditions speciﬁed (283.6 K). Clearly, the heat transfer associated with the
latent heat release is at least as important as the sensible heat transfer in this

regime.

 

2.5 , /

2.0 h /

 

 

qmﬂlq"
3

1.0 I /

/
0.5: /

 

 

 

 

 

 

 

 

 

 

 

ALA A;-- AAA-

285 290 295 300 305 310 315 320
TJ (T,=277 K)

 

Figure 1.1. Ratio of latent heat transfer to sensible heat transfer

for condensing water vapor in air.

Slow fog build up and ﬂash fogging on the windscreen can occur if
conditions are correct [Peters, 1972]. Both conditions can severely effect
passenger safety and vehicle operation. Speciﬁcally, the initiation of condensate
formation is critical to vehicular applications since small quantities of liquid can
cause drastic changes in windshield clarity. If the conditions necessary for
condensation can be avoided or altered in such a way as to make it more difﬁcult
for condensation to develop, the possibility of a safety hazard can be delayed,
minimized or avoided altogether. Therefore understanding the onset of
condensate development is critical and tools appropriate for experimental

analysis are necessary.

1.2 REVIEW OF LITERATURE

The two-dimensional wall jet has been studied rigorously. The ﬁrst
experimental publication in the literature dates back to 1934 [F6rthmann, 1934].
The ﬂuid mechanics of this ﬂow conﬁguration were thoroughly investigated in the
1960's and 1970's by researchers such as Bradshaw and Gee [1960], Patel
[1962], Tailland[1970], and many others. The developed velocity proﬁle, the
development of the turbulence signature in the ﬂow, and the jet growth in the wall
jet were well categorized in this body of literature. Recent studies of the wall jet
have been conducted with new types of equipment such as laser Doppler
velocimetry (LDV) and particle image velocimetry (PIV), because this ﬂow has
been so well documented previously [Eriksson et al., 1997]. The two-

dimensional wall jet has been studied over ﬂat surfaces and surfaces with a

variety of proﬁles, with pressure gradients (favorable and unfavorable), and with
various free stream velocities and initial jet turbulence intensities. The results of
these works are well summarized in The Turbulent Wall Jet [Launder and Rodi,
1981]. The bulk of the work reviewed by Launder and Rodi [1981] focuses on
the fully developed (turbulent) wall jet with the comparison of studies performed
at streamwise positions of 20-100 slot widths from the jet oriﬁce (this eliminates
effects of the jet oriﬁce on the ﬂow). This literature serves as a reference base
and allows for comparison of the computational model at distances further
downstream than were studied in the present work. Experimental evaluation of
the developing region is essential to evaluating the computational model since
the developing region of the ﬂow is strongly inﬂuenced by the conﬁguration of the
facility [Mabuchi and Kumada, 1972]. The defroster/demister ﬂow on the
windshield is also a developing ﬂow ﬁeld. Finally, this ﬂow ﬁeld will then
establish the environment for a study of the heat and mass transfer phenomena
relevant to the motivation for this project, the automotive defrosterldemister. The
experimental tools currently available to measure mass transfer will be reviewed
next.

Mass transfer to surfaces has been well studied as a steady state
phenomenon as a means to enhance the heat transfer [lncorpera and DeWitt,
1990, Bejan, 1995]. This enhanced energy transfer associamd with the phase
change event is the basis of utilizing a ﬂuid undergoing phase change in a vapor
compression refrigeration cycle. However, in many vehicular applications, the

process is inherently transient with respect to the boundary conditions and ﬂow

10

properties, especially the ﬂow humidity. The desire to study this phenomenon in
these applications led to developing new technology to capture transient
measurements of condensate thickness on the order of micrometers.

Several techniques exist for the measurement of mass transfer
coefﬁcients. In particular, sublimating surfaces such as naphthalene or benzoic
acid have been used to measure the mass transfer rate from surfaces subjected
to various ﬂow ﬁelds [Goldstein, 1995, Mabuchi and Kumada, 1972, Alenezi and
Abdo, 1991]. Optical measurement and detection of water and ice on surfaces
have been documented by Anderson et al. [1996], Bearman et al. [1998], Utton
and Sheppard [1985], and Pekdemir [1994]. Electroplating and electrode
reduction measurements have also been used to obtain local measurements of
mass transfer rates [Noui-Mehidi et al. 1999, DeBecker et al. 1995, Zhae and
Trass, 1997, and others]. These techniques will be discussed next.

The naphthalene sublimation method is capable of obtaining accurate
measurements of time averaged local mass transfer coefﬁcients or time and
spatially averaged mass transfer coefﬁcients [Goldstein and Cho, 1995]. This
technique has been widely used since the 1950's. It cannot be utilized in certain
ﬂows such as high velocity ﬂows due to aerodynamic healing. Natural
convective (or very low velocity) ﬂows are also problematic since temperature
ﬂuctuations over long experimental runs create substantial error in the
measurements [Goldstein and Cho, 1995]. It also cannot measure mass transfer
coefﬁcient changes that occur temporally such as the mass transfer coefﬁcient

evolving during a transient ﬂow event. The technique requires a measurable

11

amount of naphthalene to be removed over a period of time, thus providing a
temporal integration of the mass transfer coefﬁcient at a speciﬁc location.
Goldstein and Cho [1995] estimate that the methodology produces
measurements of the Sherwood number with a 7% uncertainty at a 95%
conﬁdence level (1 20). The properties of the naphthalene, most notably the
vapor pressure, are the largest contributors to the experimental uncertainty.

Electrode reduction methods utilized liquid chemicals as the ﬂow medium.
These techniques allow for time series information to be obtained experimentally,
but do not provide local information as a function of time. This is because the
electrode must have a reasonable surface area exposed to the ﬂow to function
properly. Typically, a wire is utilized as the electrode so that data are obtained
along a line on the surface of interest [Debecker et al., 1995]. Thus this method is
capable of providing area averaged mass transfer coefﬁcients as a function of
time or time-averaged, local mass transfer coefﬁcient measurements. The local
mass transfer as a function of time can be deduced from an iso-ﬂux line in a two-
dimensional ﬂow ﬁeld with this technique.

The method developed by Bearman et al. [1998] is capable of optical
measurements of ice and liquid water thickness over a range of 500 pm to 2000
pm. The method utilizes images of the surface and a reference standard
acquired over narrow (10 nm) wavelength bands. A ratiometric reﬂectance
calculation in each wavelength band is used to measure water thickness on the
surface. The reﬂectance proﬁle in the wavelengths tested is then compared to a

calibration database to obtain a thickness measurement. The authors did not

12

present measurements above 2000 pm or below 500 pm. The absorptivity of
water over the wavelengths utilized (0.850 pm to 1.050 pm) only varies from
0.0433 cm'1 to 0.363 cm“1 which would result in less than 2% absorption of the
incident light at a ﬁlm thickness of 500 pm. This relatively small change in the
signal appears to limit the minimum measurement range. There is an absorption
peak at the 980 pm wavelength due to the resonance of the hydrogen-oxygen
bond, so it may be possible to utilize this speciﬁc wavelength to expand the
capabilities of the method to a lower limit by increasing the absorbed light. The
authors specify long exposure times (up to 40 seconds) and 21 images of the
same condition to obtain a thickness measurement. Therefore, this technique is
not capable of rapid thickness eg. it would not be suitable for automotive fogging
where 1: ~ 102 seconds. A similar technique is described by Mao et. al. [1992] for
the measurement of frost on a surface, the measurement range presented is
from 1 to 4 mm with an accuracy of 13%, the authors do not specify the
conﬁdence interval of the uncertainty estimate.

An optically based, photo-evaporative technique was developed by Utton
and Sheppard [1985], Davies et al. [1990, 1991], and summarized by Pekdemir
and Davies [1999]. The technique utilizes a calibrated paper soaked in solvent
attached to the test surface. This technique is based on measuring the
reﬂectivity and transmissivity of a calibrated ﬁlter paper. Pekdemir [1994]
calibrated the target ﬁlter paper in the same ﬂow ﬁeld that experimental
measurements were obtained. Applicability of the calibration to other ﬂows of

interest was not discussed explicitly. Furthermore, the calibration values of the

13

target paper vary by over 100% in the same calibration experiments and can
change drastically with the adhesive utilized to attach the paper to the test
surface. The surface roughness of the target paper can also effect the
measurement results, thus potentially adding artifacts to the data obtained. The
need to attach the calibrated paper limits the size and geometry of the surface of
interest that was studied [Pekdemlr, 1994].

The photo-evaporative technique can only be effectively utilized in quasi-
steady state ﬂows as the technique can not measure quickly changing mass
transfer rates. Solvent migration in the ﬁlter paper also limits the maximum mass
transfer rates that can be studied [Pekdemir and Davies, 1999]. The calibration
and experimental data show initial transient and potentially non-reproducible
measurements. This is further complicated by drift in the paper calibration with
time and the fact that the paper requires several experimental runs prior to
calibrating to insure calibration stability. The evaporating solvent utilized for this
technique is also toxic. Temperature variations between the calibration and the
experimental conditions contribute to the measurement error. Pekdemir and
Davies [1999] estimate the accuracy of this technique to be 112%, but do not
state the conﬁdence level of this estimate. The variation in the calibration results
is overcome in the Pekdemir [1994] publication by calibrating against known
results (from other experimental methods) in the actual test ﬂow, which severely
limits the potential applications of this technique to ﬂows that have already been
studied by other techniques. Utton and Sheppard [1994] also indicates that the

technique would have only limited applications in natural convective ﬂows

14

because the density of the solvent vapor would strongly inﬂuence the natural
convective ﬂow ﬁeld.

The current state of the art indicates that an opportunity exists to develop
and implement a technique capable of transient mass transfer measurements of
the water condensation phenomena. The initial onset of the condensate is of
critical importance to automotive defrosterldemisters. Currently,
defroster/demister performance is measured in an arbitrary fashion that is highly
dependent on human input and error. Defroster/demister testing is performed to
meet Federal Motor Vehicle Safety Standard 103 (FMVSS 103). The FMVSS 103
requires defroster testing as outlined in SAE Recommended Practice J902 [SAE,
2000]. This document deﬁnes the deﬂected area as the region of the windshield
that is ”dry, covered with partially melted ice or wet" [SAE, 2000, p.34.19]. It
further elaborates than an observer inside the vehicle should mark an outline of
the defrosted region at 5 minute intervals for engineering purposes [SAE, 2000].
Clearly, this testing procedure and the data acquired are of limited use and prone
to human inﬂuence. An experimental technique with PC based data acquisition
would obviously provide a more objective measure of defroster/demister
performance. The technique, which will be discussed next, developed as part of
this project presents a potential solution to obtain quantitative data to address
these issues.

The ﬁnal phase of this project investigated transient mass transfer. A new
optically based technique based on the scatter and absorption of infrared

illumination on a reﬂective surface was developed. Although this technique was

15

not extended to automotive glass in more than a preliminary fashion, the
technique was successfully applied to an aluminum test surface in the model wall
jet flow. The transient measurements obtained with the method developed in this
document are capable of condensate thickness measurements between 0 and 5
micrometers at a measurement rate of 0.2 Hz with the equipment utilized. The
accuracy of the technique depends on whether calibrated targets are utilized for
the measurements or if a similar surface technique is employed (eg. that is,
calibration is conducted on surfaces similar to the test surface). The most
accurate measurement results are obtained with calibrated targets in the ﬂow of
interest; it is estimated that condensation thickness measurements can be

obtained with an uncertainty of 1 0.10 pm under these conditions.

16

Chapter 2

FACILITY DEVELOPMENT AND CHARACTERIZATION

A facility was designed and constructed to provide a model ﬂow ﬁeld to
obtain the benchmark data required for the project. A two-dimensional wall jet
ﬂow ﬁeld with controlled velocity, wall surface temperature and inlet humidity was
selected as the model ﬂow. A schematic of a two-dimensional wall jet ﬂow ﬁeld is
presented in Figure 2.1. The wall jet ﬂow ﬁeld was selected as the model ﬂow
because it has some geometric similarity to a defrosterldemister system but is
simple enough to avoid complications such as fan turbulence, windshield
curvature or jet impingement. The design required temperature control of the
wall surface to add thermal and mass transfer effects to the ﬂow. Mass transfer is
induced when the wall temperature is below the dew-point of the ﬂow and
condensation develops. The use of a wall jet ﬂow ﬁeld also allows the

experimental data to be compared with other studies in the literature.

17

 

 

 

 

w I “-9-
r4
. L- 82 _.. 7
Velocrty Proﬁle ‘65
* ~11 o
_\ " [_.
\\\\ I. Umax H... g
Temperature and UI/2 V\\\\ ,/ 6,: ‘3
Concentration Proﬁles <3
7 .4- 2 £‘
C en TI}? C3 8
T°° I/ g C3
/ o ‘t':
_-- U)

 

+
X

Figure 2.1. Schematic of the wall jet ﬂow ﬁeld in the FMTF.

The following sections outline the design of the Ford-MSU test facility
(FMT F) and ﬁre functionality of the complete system. The data acquisition,
motion control, and the capabilities of speciﬁc measurement transducers and
devices utilized are also included. The measurement accuracy and resolution of
each component are deﬁned as well as how each measurement device
contributes to the experimental project objectives. Finally, the results of the

characterization experiments are presented.

18

2.1 FACILITY DESCRIPTION

A ﬂow facility was designed and constructed to meet the goals and

experimental goals of the project. The ﬂow facility is presented in Figures 2.2

and 2.3. The system was built with a modular design so that components or

conﬁgurations could be modiﬁed with minimal down time. The major system

components will be described in the order that the airﬂow proceeds through the

 

 

 

 

 

 

 

 

 

 

 

system.
Flow 1. Inlet
I 2 2. Steam Supply
3,} 3. Humidiﬁer
4. Condensate Return
\~ 4 5. Duct
4/5 6. Passive Mixing Box
6 7. Flow Conditioner
ff 8. Two Dimensional Contraction
9. Splitter Plate
10. Auxiliary Nozzle
I 11. Thermally Active Test Section
I 12. 3 Axis Positioning System
__7.~;7___j 13. Optical Access
=l/ 8 14. Base
10 9 E; g 15. Condensate Collector
“ 16. Prime Mover
8 12 ‘3 17. Exit w/ Throttle Valve
Flow
_,_ t/ .7
l4 - .—.—. - ‘ H: T“ —~
”:3 /-- 16 T
15 x \ If Ti 0.76m
x f‘—’1 ---- /I I

 

 

 

 

 

Figure 2.2. The Ford MSU Test Facility (FMTF).

19

 

Figure 2.3. Photograph of the Ford MSU Test Facility.

The steam humidiﬁcation system, the passive mixing box and the
associated ductwork (items 2-6 in Figure 2.2) extend above the ﬂow conditioner
and the two-dimensional contraction. This conﬁguration allows the steam
humidiﬁcation to completely mix and thus insures a homogenous ﬂow. Water
droplet evaporation or removal is essential for the hot-wire/coid-wire probes to
function properly as droplet impact would destroy the sensors immediately. The

system is capable of adding 0.0018 lgls of water vapor to the airﬂow. This is

20

enough to raise the humidity of the wall jet from 20% Rh to 75% Rh at a 300
CF M ﬂow rate.

The ﬂow exits the passive mixing box and enters the ﬂow conditioner (item
7 in Figure 2.2). The ﬂow conditioner consists of a one-inch-thick honeycomb
structure with 0.20 inch diameter cells. The honeycomb structure removes large
scale eddies from the air stream. The ﬂow then proceeds through 3 wire screens
placed 3 ‘A inches apart. These screens consist of a mesh of 30 by 30 wires per
square inch. This system helps break down any further turbulent motions in the
ﬂow and provides an airflow with minimal disturbances at the entrance of the two-
dimensional contraction.

The two-dimensional contraction is shown as item 8 in Figure 2.2. The
two-dimensional contraction accelerates the flow and ejects it through two 2 cm
wide X 35.6 cm long slots (1 on each side of the thermally active section) into the
sub atmospheric test chamber. The contraction was designed to provide a jet at
the exit plane with a uniform velocity proﬁle [Morel, 1979]. The splitter plate
creates a thin momentum boundary layer at the jet exit, but it also allows the
thermally active test surface to be mounted in the center of the jet exit.

The thermally active test section (item 11 in Figure 2.2) is shown in Figure
2.4. The test section is essentially an aluminum heat exchanger. The surface is
maintained at a constant temperature by pumping a water-glycol solution from a
constant temperature circulator bath through the test section. The internal details

of the test section will be discussed in section 2.5.

21

Two-dimensional
contraction exit

Thermally active test
section

    
 

Positioning
system
Splitter plate

Figure 2.4. Thermally active test section — aluminum heat exchanger.

The positioning system and ﬂow measurement sensors (item 12 in Figure
2.2) are also contained in the sub-atmospheric test chamber. The details of the
sensors and the motion control system are presented in sections 2.3 and 2.4.

The ﬂow exits the sub-atmospheric test chamber into the base (item 14 in
Figure 2.2) of the FMTF. The base houses the prime mover for the ﬂow system,
which operates in suction mode to prevent fan turbulence from entering the sub
atmospheric test chamber. The airﬂow rate through the FMTF system is throttled

at the exit of the prime mover by a sluice valve (item 17 in Figure 2.2).

2.2 EXPERIMENTAL CAPABILITIES

Table 2.1 presents a summary of the FMTF experimental parameters
along with the associated control and measurement capabilities. In general, the
facility can measure the velocity, temperature, and humidity of the wall jet ﬂow
and the ambient conditions in the sub atmospheric test chamber. The surface
temperature of the thermally active test section can also be measured and
controlled. The humidity and the velocity of the jet ﬂow can be controlled as
described in Table 2.1. The speciﬁc controls and instrumentation used for

measurements will be discussed in detail in the following sections.

23

Table 2.1. Experimental parameter control and measurement capabilities of the

 

 

 

 

 

 

 

 

F MTF.

FMT F Variable Range of Control Measurement Resolution
Humidity Ambient to ambient + 11.25%

55%Rh at 300 CFM
Velocity (Bulk Flow) 0 tol3.5 m/s 12% reading
Pressure 0 to 0.5 in-H20 0.5% full scale

(0 to 124 Pa)
Local Velocity Measurements l to 13.5 m/s :2%
Sensor Positioning 120 mm of motion with <200 pm for full range of
(y-axis Traverse) 2.5 pm steps motion
Local Temperature T/C: -20 °C to 100 0C 0.2 °C (T/C)
Measurements cold-wire: -20 °C to 100 0.2 °C (Cold-wire)
Thermocouple (T/C) and Cold - °C

Wire Probe Measurements

 

 

 

 

 

Temperature Reference -l76 °C to 180 °C 0.01 °C (NTST traceable)
Thermally Active Test Surface -20 °C to 100 °C with 0.4 °C (stability limited)
(Neslab Bath Control Limits) current working ﬂuid
Condensation Visualization Pixel size of 391 pm x Visualization teclmique
(CCD Camera - 640 x 480 391 pm (if complete test only
detector ) surface is viewed)
Condensation Visualization Pixel size of 1.91 mm x O to 5 pm condensate
(Infrared Camera) 1.91 m (if complete test thickness -
surface is viewed) $0.17 um @
95% conﬁdence level.

 

 

 

24

 

2.3 DATA ACQUISITION AND MOTION CONTROL

Data acquisition and stepper motor motion are controlled and recorded
electronically. The data acquisition and motion control systems are shown

schematically in Figure 2.5.

 

Pressure Transducer
I __!____

- Sub-Atmospheric

PC With 2

Th [Pm ﬁ if: 300 _—_I_ Test Chamber

software “met“ Hot-wire} Cold-wire
I 3

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Humidity and
Temperature
Probe
pC with DIAS-1800 Thamocouplﬁ
QuickBasic ___+ 4 D/A, 2 Am (T-Type)
program Channels 6
H Stepper Motor F
Controller
10
PC with
DAS-TC
3:333" 16 AID
Channels

 

 

 

 

 

 

 

 

 

Figure 2.5. Schematic of the data acquisition and motion control systems.

The stepper motor positioning system is automated and controlled with
software written in Microsoft QuickBasic° and implemented on a PC. The
software accesses a DAS-18008T board (Keithley Metrabyte) which provides

four, 12 bit D/A output channels and 2 high speed (300 kHz) input channels to

25

control the stepper motor. The stepper motor (Vexta, Model Ph268M—E1.5B-C7)
drives a precision screw (Specialty Motions Inc., Model KBO801-150) in order to
move the sensor-positioning traverse. The stepper motor moves the traverse in
the y-direction (refer to Figure 2.1) in increments of 2.5 um. This measurement
step is smaller than the hot-wire or cold wire probe diameter (5 pm) and thus
does not limit the spatial measurement resolution obtained.

The DAS-TC board is capable of sampling at 1 Hz per channel when
multiple channels are utilized. This sampling rate limits the data inputs to
relatively slow frequency sensors and thus the DAS-TC system is utilized

exclusively for thermocouple and humidity sensor input.

2.4 MEASUREMENT PARAMETERS AND EQUIPMENT

2.4.1 PRESSURE MEASUREMENT

A MKS Baratron Model 225A pressure transducer is utilized to monitor a
Pitot tube mounted at the exit of the two-dimensional contraction. The pressure

transducer has a measurement range of 0 in-HzO and 0.5 in-HzO, and an
uncertainty of 0.0025 in-HzO (0.5% of full scale).

Pressure measurements are used to calculate the inviscid core velocity of

the wall jet using the Bernoulli equation (without body forces),

Bill: 31.122.

p 2 p 2. (2.2)

4‘

V1 is the inlet velocity, P1 is the inlet pressure, V2 is the velocity in the Pitot
tube (which is equal to zero), P2 is the stagnation pressure, and p is the density
of air. The velocity measured with the Pitot tube is utilized to calibrate the hot-
wire sensors and to determine the wall jet velocity. A typical test is performed
with a maximum inviscid core velocity of 10 m/s. The pressure transducer is

capable of resolving this value to within 10.025 m/s.

2.4.2 VELOCITY MEASUREMENTS

Velocity measurements are performed with an lFA300 hot-wire
anemometer (TSI) using the constant temperature anemometer (CTA) module.
The hot-wires are constructed with 5 pm diameter tungsten wire. These wires
are heated electrically to maintain a ﬁxed, user-deﬁned resistance value and thus
a constant operating temperature. The heat transfer from the wire is proportional
to the voltage supplied and to the ﬂow velocity. It is this phenomenon that allows
the hot-wires to be calibrated and utilized as a ﬂow velocity measurement tool.

The governing relationship between the hot-wire voltage (E) and the
velocity (V) [Collis and Williams, 1959] is given as

E2 = (A + BV")(Tw — To, ), (2.4)
where A, B, and n are the calibration coefﬁcients obtained from a 3
parameter ordinary least squares curve-ﬁt of the calibration data; T... is the

average wire temperature and T... is the air temperature local to the hot wire.

27

However, the lFA300 system does not use a physically based equation to ﬁt the
calibration data, but rather a fourth order curve ﬁt. Similar accuracy is attainable
with either method and limited by the quality of the calibration procedure [TSI
Incorporated, 1997]. A hot-wire and a cold-wire probe can be run simultaneously
and the experimental conﬁguration utilized is presented in Figure 2.6. This
provides local temperature compensation for the hot wire velocity measurements
in the thermal boundary layer. Details of this technique are presented in Chapter

4.

SUpport bracket Dial indicator —
near wall locator

z "\" Thermally active
test section

 

Hot-wire / cold-wire

probe body Broaches

and 5 micron sensing i
wire

 

 

Figure 2.6. Hot-wire conﬁguration for two wire probe technique; top View;

(Flow direction in x-direction, into the page).

A well-calibrated single hot-wire is capable of achieving a velocity
measurement accuracy of i 1% at the turbulence intensity level present in the

calibration ﬂow in the Ford-MSU ﬂow facility [Bruun, 1995]. This degree of

28

accuracy relies on calibrating the hot-wire before and after each experiment and
using an average of the two calibrations to process the data. Post-calibration
improves measurement accuracy by accounting for hot-wire calibration drift due
to wire oxidation, slight ambient temperature or humidity changes, and/or impacts
from dust particulate. The measurements presented here were made in a
boundary layer near a solid surface. The hot-wire was often destroyed or
damaged by contact with the surface, thus eliminating the possibility of post-
calibration. Despite this, the experimental standard deviation in the isothermal
velocity data was always less than 3% of the peak velocity value (or in absolute
units, always less than 0.30 m/s). A single hot-wire is incapable of providing
velocity ﬁeld information. Hot-wires can provide high frequency velocity data, up
to 300,000 Hz depending on sensor geometry, [TSI incorporated, 1997]. Peak
frequency of the velocity ﬂuctuations in the wall jet flow were determined from the
power spectrum density obtained experimentally. The power spectrum density
diminished by three orders of magnitude at approximately 1500 Hz. All hot-wire
data were sampled at 8 kHz or above based on these ﬁndings to insure the
Nyquist criteria was met. The integral length scale of the ﬂow was determined to
be approximately 8 mm using an autocorrelation in the inner boundary layer.
This clearly scales with the ﬂuctuations in the outer shear layer and not with the
inner boundary layer since the boundary layer thickness is less than 2 mm at any
point on the thermally active test section. The integral length scale provides

insight into the sensor spacing required for hot-wire temperature compensation.

29

2.4.3 TEMPERATURE MEASUREMENT

Three different methods are currently utilized for temperature
measurement in the FMTF. These consist of thermocouples, a thermistor, and
the cold-wire probe. The thermocouples are type-T (copper-constantan) and are
manufactured on site. These thermocouples have a measurement bead diameter
of approximately 800 pm and a time constant of 1 s (for typical experimental ﬂow
velocities). Special type-T thermocouples [Omega Inc.] have also been acquired
with a measurement bead diameter of 60 pm and an approximate time constant
of 0.02 seconds in the ﬂow ﬁeld. The thermistor is an integral part of the General
Eastern M2-Plus dew-point monitoring system utilized to monitor the wall jet
temperature and humidity level.

An Omega DP-95 RTD NIST (National Institute of Standards and
Technology) calibrated temperature measurement system is utilized as the
temperature standard for the laboratory. The thermocouples are calibrated
against this standard by placing them in an aluminum cylinder in the Neslab
RTE-140 bath with the RTD probe. The calibration equipment is presented in
Figure 2.7.

m H
“Witt
NW Eas'
techno/0W
“W“ at.

“Sim C

Model 90

 

Figure 2.7. Temperature reference and thermocouple calibration equipment.

2.4.4 HUMIDITY MEASUREMENTAND CONTROL

Humidity is measured at the exit of the two-dimensional contraction with a
General Eastern M2 Plus dew-point monitor. The system utilizes chilled mirror
technology to provide a humidity measurement accurate to :l:1.25%Rh. The
system also provides a temperature output accurate to within 10.2 °C. The
system can adjust to changes in the dew-point temperature (and thus the relative
humidity) at a rate of 1.5 °C/s [General Eastern, 1999].

The humidity of the wall jet ﬂow can be increased with the Armstrong
Model 90 steam humidiﬁer incorporated into the facility. The system is capable of

adding 0.0018 kg/s of water vapor to the airﬂow. This is enough to raise the

31

humidity from 20% Rh to 75% Rh in a 300 CFM ﬂow. Experiments have
produced 95% humidity levels and above at the wall jet exit plane at less than
three-quarters of the Armstrong unit capacity with ambient conditions at 50%Rh.
The slow time response of the humidity probes in the control circuit caused
instability and oscillation in the control circuit provided with the unit. Therefore,
the system is controlled in an open loop fashion. The system has a measured
variability of less than 12% Rh when controlled in the open loop mode during the

duration of a typical experiment.

2. 4. 5 CONDENSATTON VISUALIZA TION

Both a CCD camera and an Inframetrics 600L infrared camera are
available for the purpose of condensation visualization and quantiﬁcation. The
image information can be stored on video cassette recording equipment and time
marked with a FOR-A video timer (Model VTG-33) to capture transient
phenomena. The image information is processed with image processing
software (lmagePro Plus v.4.1, Media Cybernetics). This software is capable of
multi-functional image analysis.

Measurement resolution is determined in part by the ﬁeld of View required.
A CCD camera with 640 x 480 elements will have pixels that cover a 391 pm x
391 pm square area if the entire thermally active test surface is viewed (A. =
0.0445 m2). The pixel area scales linearly with the area of the surface viewed,

i.e. if half the surface is viewed, each pixel will be half as large. Thus the ﬁeld of

32

view and the spatial resolution are intimately related and can be adjusted by
lenses or physical arrangement of the optical system.

A technique to measure transient condensate development was also
desired. At the time of construction of the facility, no such method was decided
upon or implemented. Details of the development and implementation of a

suitable technique are included in Chapter 6.

2.5 THERMALLY ACTIVE TEST SECTION

Circulation of refrigerated ﬂuid in the interior of the thennally active test
section provides temperature control of the experimental surface (Figures 2.4
and 2.8). The interior conﬁguration of the ﬂow channels can be seen in Figure

2.8. Previous tests using the test section in a similar ﬂow ﬁeld showed that 80%

of the test surface temperature varied by less than 0.3 °C at wall jet velocities as
large as 10 rnls [AbdulNour et al., 1997]. The thermally active test section can
be heated or cooled depending on the setting of the temperature bath. The
Neslab NTE-140 temperature bath utilized is capable of supplying 500 watts of
cooling at 0 °C or 800 watts of heating and has a range of - 40 °C to 100 °C.

l<— 7”—>l

 

 

 

 

 

 

 

 

 

 

n Flow Channels
10.4” //
v 1
JL n n
K /“

Inlet and Exit

Figure 2.8. interior ﬂow pattern of the thermally active test section.

Transient surface temperature changes are obtained by using a bypass
loop in the coolant circuit (Figure 2.9). The bypass loop can be closed and the
coolant pumped into the therrnaliy active test section. This results in a transient
surface temperature as shown in Figure 2.10. These results show that a rapid
change in the surface temperature of the thermally active test section can be
obtained in 45 seconds and that the temperature can be monitored and recorded
for modeling purposes. This change could be utilized to model transient changes
in the automotive environment such as a sudden rain or snow fall (thus changing
windshield temperature). While not directly analogous to a speciﬁc automotive
event, the dynamic modeling capabilities of the software can be veriﬁed with the

data obtained.

Bypass loop and valve

Neslab ‘ ‘ 7 Thermally Active
Constant ' , Test Section

Temperature
Bath/Circulator ‘—

 

I

 

 

 

 

 

 

 

Figure 2.9. Bypass loop schematic for transient thermal tests.

 

   
   

 

 

 

 

 

25
20 —- Bath temperature
A at to 1 °C at i=0
0
0
E 15 _-
3 __
——-Waii Temperature
g — inlet Temperature
5 10 T — Outlet Temperature
P _ _ __._-,W
5 -_
o l . .
0 100 200 300 400 500 600

Time (s)

Figure 2.10. Transient temperature history of the thermally active test section.

2.6 CALIBRATION AND CHARACTERIZATION EXPERIMENTS

Several veriﬁcation experiments were conducted to insure that certain
assumptions regarding the FMTF system were correct. This section will detail
the characterization experiments, which include a Pitot tube check of the velocity
at the exit plane of the contraction. A Spanwise (z - direction) velocity proﬁle was
also obtained to verify the two-dimensionality of the jet proﬁle issuing from the
two-dimensional contraction. The contraction (nozzle) was designed speciﬁcally
for this facility. The thermal and humidity stability of the FMTF ﬂow facility was
also investigated to deﬁne the experimental steady-state and quantity transient

response times of the FMTF system.

2. 6. 1 TEST CONTRACTION MEASUREMENTS

The ﬂuid velocity above the contraction is not equal to zero because of
ﬂow conﬁnement in the duct. The geometry of the ﬂow is known and the
continuity equation is utilized to substitute for V1 in the Bemouili equation (2.2) for
ﬂow through the two dimensional contraction. The area ratio of A1 to A2 for the
two-dimensional contraction is 5:1, thus V. is equal to 20% of V2. Thus the

velocity terms in Equation (2.2) can be simpliﬁed as

V
V22 - V,2 = V22 — (ff = (l- 0.04)sz = 0.96sz. (2.3)

Substituting this result into Equation (2.2) and solving for the exit velocity

from the two-dimensional contraction (V2) yields

2(1)] - P2)
V, = I_o.96p . (2.4)

The comparison between the Pitot tube measurements and the pressure
tap measurements are presented in Figure 2.11 for the two-dimensional
contraction. The Pitot tube method and the static pressure tap method yield
results that differ by less than 0.5% at the test ﬂow velocity of 10 m/s, thereby
verifying that the pressure tap method or the Pitot tube allows for accurate
measurement of the inviscid core jet velocity. The velocity measurement of the
inviscid core of the jet (Uj) will be utilized primarily to present the velocity

measurement results in a standard non-dimensionalized format.

37

 

 

 

 

 

 

 

 

14
12 -- x3
10 -- /AX/
—; y = 1.0015x A?(
E 8 g R2 = 0.9993 /’/
5‘ .A/
/
I; 6 g / xPitoiathw=o5
§ / A Pitot at xlw = 1.5
a /
4 >(
2 ‘7 : l ; +
2 4 6 8 1o 12 14

Pressure tap velocity (mls)

Figure 2.11. Comparison of test contraction velocity (from pressure tap

measurements) and Pitot tube measured velocity.

2. 6.2 SPATIAL FLOW UNIFORMITY MEASUREMENTS

The ﬂow characteristics of the wall jet were measured. The goal of the
ﬂow conditioning was to provide a wall jet that was spatially uniform with regard
to velocity, temperature, and humidity. The characteristics of interest were the
free stream turbulence intensity as well as the uniformity and stability of
temperature and humidity. These variables are indications of the quality of the

ﬂow conditioning and the two-dimensional contraction design.

The free stream turbulence intensity was measured with a single hot-wire
probe at the exit plane (xlw = 0) of the test contraction. The RMS (root mean
square, u’) of the velocity measurement is a quantitative measurement of the
ﬂuctuation level present in the flow. The measured value of u’lU was less then
1.5% in the inviscid core of the jet at the xlw = 0.0 position.

A Spanwise (z-direction) traverse of the wall jet was performed with a
single hot-wire to verify the two-dimensionality of the jet ﬂow. The z—direction
mean velocity (W2) is assumed to be zero and the root mean square (RMS) is
assumed to be very small compared to the X and y-dlrecﬁons. if this assumption
is valid, the wall jet proﬁle would be variable only in the x and y-directions and
therefore the z-direction proﬁle should be ﬂat and independent of 2 location near
the center of the wall jet. Figure 2.13 presents U/Um as a function of z/L. The
variable z/L is the location in the z—direction non-dimensionalized by the slot
length (L = 35.6 cm). Figure 2.12 shows the sharp increase of the velocity at the
outer edge of the slot (zlL = -0.5) and the ﬂat proﬁle from the outer edge of the
slot past the center of the slot (zIL = 0). Note that only one edge of the ﬂow was

interrogated due to the symmetric design of the facility.

 

 

UIUmu
c
or

 

 

 

 

 

WJ 0.14
a - e

-0.7 -0.5 -0.3 01 0.1 0.3
2 location (zIL)

 

Figure 2.12. Z-direction (transverse) velocity proﬁle; ylw = 0.5.

The temperature ﬁeld within the jet ﬂow was measured by positioning
type-T thermocouples at multiple locations across the exit plane of the jet. These
measurements are presented along with the humidity data in Figure 2.13. The
temperature and humidity of the ﬂow are spatially uniform within the
measurement resolution capabilities. The data presented in Figure 2.13 show a
homogeneous temperature and humidity ﬁeld along the centeriine of the
thermally active test surface (z/L = 0). The gradients near the edges of the
thermally active test surface (near zlL = -0.5 and z/L = 0.5) are created by the jet

mixing with the air in the sub-atmospheric test chamber. All velocity and

temperature measurements presented are taken along the centeriine of the
thermally active test surface unless speciﬁed otherwise. The spatial uniformity

along the jet exit is essential to insure the two dimensionality of the ﬂow.

 

 

 

 

 

 

 

so 30
7 ~ .-
8 6 Q I” 4’ {It 6 29
76 - T 28
74 - ~— 27 8
0v
2, 72 - § ~~ 26 g
z ‘5 s
m 701 ‘4” 25 g
at 68 cu <9 6 m c P a, .- 24 a
‘P 5
66 ~ A 23 1-
34 0Humidity l T 22
52 OTemperature -- 21
60 T 7 . r I ‘l’ 20
-0.8 -0.6 -o.4 -o.2 o 0.2 0.4 0.6 0.8
zIL location

Figure 2.13. Jet temperature and humidity uniformity study (w = 2 cm).

2. 6.3 TEMPORAL STABILITY OF THE FMTF TEMPERATURE AND HUMIDITY

Experiments were conducted to determine the temporal stability of
temperature and humidity within the FMTF sub-atmospheric test chamber and
Within the wall jet ﬂow. These data help deﬁne the amount of time the system

requires to reach steady-state as well as the stability of the steady-state

41

condition. These experiments were executed with the Armstrong Model 90
steam humidiﬁer providing ﬂow humidiﬁcation and with the therrnaliy active test
section chilled to the point where condensation was actively occurring
(approximately %Rh,- = 70, T. = 22 °C, and T. = 12 °C).

The wall jet reacted very quickly ( < 120 s) to steam humidiﬁcation from
the Armstrong Model 90 humidiﬁer. The environment in the sub-atmospheric test
chamber was slower to equilibrate (~ 30 min.) due to the low entrainment
velocities present and the volume (1.62 m3) of the test chamber. The volumetric
ﬂow rate through the system for a typical test is 0.14 m3/s, but little of the wall jet
is dispersed throughout the chamber.

The wall jet temperature is typically several degrees warmer (2 - 3 °C)
than the sub-atmospheric test chamber upon system start up due to the location
of the intake near the top of the laboratory. The addition of steam humidiﬁcation
also increases the wall jet temperature. Figure 2.14 shows the thermal response
of the wall jet. The humidity measured in the wall jet equilibrates much quicker
than the temperature and is also included in Figure 2.14. The tests were initiated
from the “cold start up” condition where the F MTF had equilibrated with the
ambient laboratory conditions. Two different velocities were included to
demonstrate the limitation of the humidiﬁcation system, which can be seen when
comparing the two tests. Note that ambient humidity was approximately 11%Rh
for both tests. The overshoot and settling time of the General Eastern M2-Plus

sensor is also clearly evident in Figure 2.14.

42

100 . 30

 

 

 

 

 

 

 

 

 

 

 

90 a ‘— %th ~- 29

A 80 ~ -- 28
g 70 - -~ 27 E
g 60 .. ~- 26 g
E 50 . — __ __ _ ‘i 25 g
g 40 4 w T. ~ 4. 24 g.
3 so 4 ’ "T' -- 23 .2

& 2° ‘Ej -——U, = 13 m/s 1' 22

10 . ——U,= 10 m/s ..21

0 . . . 20

0 500 1000 1500 2000
Time (a)

Figure 2.14. Typical transient response data for the wall jet ﬂow in the F MTF.

Figure 2.14 indicates that the FMTF response to humidity changes is
faster than to temperature changes and that steady-state conditions were
reached within 8.3 minutes (500 s). The steady state jet conditions were stable
within 2% Rh and 1 °C. These results deﬁned the necessary settling time for the
facility to obtain steady state operation. They also began to establish the

limitations of the transient changes that may be implemented experimentally.

Chapter 3

ISOTHERMAL FLOW MEASURMENTS

The wall jet ﬂow ﬁeld was ﬁrst investigated as an isothermal ﬂow. The
isothennai, two—dimensional wall jet has been rigorously studied as detailed in
Chapter 1. The purpose of this study was to investigate the developing region of
the wall jet for three speciﬁc reasons:

1. To establish conﬁdence in the experimental techniques through direct

comparison with previous studies in the literature.

2. To detail the developing region of the wall jet which is subject to unique
features of the ﬂow facility. These effects damp out as the ﬂow
develops down stream.

3. To establish baseline data for comparison to future non-isothermal
tests and to the CFD simulation results.

The isothermal experimental technique will ﬁrst be discussed, followed by
presentation of the measurement results. The experimental data will then be
compared to the results from the literature and from the CF D model. Finally, the
experirnentai procedure and the results will be evaluated.

The several features of the wall jet will be discussed. The deﬁnitions of
various features and zones in the wall jet ﬂow are deﬁned in Figure 3.1. The
inner and outer shear layers are deﬁned based on the derivative of the time

averaged velocity (U) gradient with respect to the wall normal direction (y).

 

lnviscid core

I
—.___ ‘g‘ k .2 _
-~. , _-
- ' 4
NW»; ‘— .__E
\ 1

ll] l7

qﬁﬁ

*7?—

I Inner shear layer

l

T‘-~~\ . I. 7 dU/dy > 0

,‘/

a,

7"“
.

. ., :31
i \.\

Outer shear layer
dU/dy < 0

/ /’ vx

(U/Umax =05) Umax

i
..i
l

'3(1 ‘4: ]“T “(gm

 

Figure 3.1. Deﬁnitions of features and zones in the

two-dimensional wall jet ﬂow.

3.1 MEASUREMENT RESULTS

lsotherrnal test data were obtained with a single hot-wire probe (velocity)
and a calibrated thermocouple (temperature). The electrical resistance of the hot-
wire is a function of temperature. Therefore, an average temperature is
equivalent electronically to an average resistance. The hot-wire was maintained
at a constant average temperature (~230 °C) by the anemometer control circuit
The control circuit utilizes a balanced Wheatstone bridge and Joule heating to
maintain the sensing wire at a ﬁxed average resistance. The hot-wire sensor is
only responsive to changes in the convective cooling velocity in an isothermal

ﬂow and can therefore be calibrated easily to measure velocity. Although these

45

tests were performed for isothermal conditions (the jet and the thermally active
test surface were not thermally active), temperature compensation is necessary
to insure accurate hot-wire measurements if the ambient temperature during
testing is different than the ambient temperature during hot-wire calibration. The
hot-wire and thermocouple were traversed across the wall jet width with the
stepper motor positioning system. The stepper motor and the output data were
controlled/recorded with the systems described in Chapter 2. These
measurements were conducted at ambient laboratory conditions. These were
typically T.o ~ 22 °C and %Rh ~ 60%. The velocity of the inviscid core of the jet
was set at 10 m/s for each test to provide a flow velocity comparable to the
automotive flow ﬁeld.

The isothermal velocity proﬁles are presented in their non-dimensional
form (U/Um as a function of y/62) in Figures 3.2 - 3.7. The peak velocity in each
proﬁle is deﬁned as Um; Um differs at each xlw location due to slight variations
in the experimental set up and decay of the velocity as the jet progresses down
stream. The ordinate is deﬁned as y/52, where y is the distance from the test
surface and 82 is the distance from the test surface in the y-direction to the
location where the velocity (U) has decreased to 50% of its maximum value
(Um) in the outer shear layer. This format collapses the data to a single proﬁle
when the jet has reached the self-preserving, fully-developed flow conﬁguration
[Potter and Foss, 1982]. The streamwise normal stress (SNS) is also included in

each ﬁgure. The SNS is deﬁned as

02

u
SNS =100'I'J—2 (3_1)

The SNS is an alternate method of presenting the root mean square

(RMS) of the velocity ﬂuctuations or the turbulence intensity.

 

 

 

 

 

 

1 WWﬁ‘ 7
o 9 « " "
. >1! X xU/Umax
‘ x 4r 6
0.8 'lx * . +SNS
0.7 “,1: § 4‘ 5 of;
0.6 4 § 5
2 ’3 . “ ‘ ~§
J X
g 0.5 x x 8
x 1’ 3 3
0.4 x U)
3
0.3 it .. 2
i
0.2 x
« 1
0.1 x
o t . s . o
0 0.5 1 1.5 2

Y’52

Flgure 3.2. Non-dimensional velocity proﬁle at xlw = 0.0.

47

 

 

 

 

1
(W
0.9.3 g
i ,5 .- 6
(x
0.8 g X
0.7 . x le/Umax . 5 "A
3 ; +§ES a
0.6 g x 2
2 * "-
05 a
g .5 3 a
0.4 i 5
x in
03 + 5 z
. + «2 (o
0.2 + ’1
+ ’1; ..1
0.1 T 15‘
+ + * " It at >lk at x“ x
o LALL+++ ﬁ¢++ + .L 1+ .9. o
o 0.5 1 1.5 2

via:

Figure 3.3. Non-dimensional velocity proﬁle at xlw = 1.59.

 

 

 

 

 

 

 

 

1 W 7
l y: xix
0.9 . x “Sr
)8 5‘ " 6
0.8 < x x
X x x UlUmax
X & -.. 5 A
0.7 l x + SNS N2
- X
0.6 3. n 0 4 "a
2 " '5
a 0.5 x 8
D +t+xx -» 3 F
0.4 4, + ..
+ X (I)
+ +)& z
0.3 -.
41 ﬁx 2 m
0.2 f“
I + 93‘ xx ‘* 1
0.1 t + + X x
f X XXX X x
o J+++++** . + +1344?" 4):. ﬁx it 0
O 0.5 1 1.5 2
WE:
Figure 3.4. Non—dimensional velocity proﬁle at xlw = 3.18.

 

 

 

 

 

1 {M 7
0.9 r
i a“
0.8 >
s k x U/Umax A
07 g ‘& +SNS 0 5 NE
D
N 06 I ‘x L 4 NT
5 0.5 up 3
\ 4» +x o
D + $4.: + _ 3 v-
0.4 1 , ++l§ ¥* V
1* “t to
0.3 If ‘15,; 2 g
.
0.2 ‘ +1“ ‘54
+ + t + :1 Q’; -- 1
0.1 1#*"‘¢_‘jbf+m§:+ +.+ W
i "wiry 03* ' *1” '
0 . . . ' 0
o 0 5 1 1 5 2
Y’s:

 

 

 

 

 

7
«6
xLl/Unax “’5":
+SNS g
N "4N?
g 3
~ 0
3 t l 3 2
* ‘E i in
+ 2 z
l ‘ m
i‘ 1
ﬁxik
“f3
. . 0
1 1.5 2
WM

Figure 3.6. Non-dimensional velocity proﬁle at xlw = 7.62.

49

 

 

 

 

 

 

 

 

5‘
0 9 g m 6
°8 3 xxx“
5 .1"
0'7 ’5‘ x U/Umax g
. SNS
g 0 6 x + “F 4 a
E 05 '3
D + + ++++ + J 3 8
0.4 ++++ + +3 +++ 3
+1“ + 1* g
0.3 .1 +1 '5!“ -. 2 CD
fa + “21‘
0.2 1 {*pr ++I++ ”it
I- +¢t$+ &‘* 1
0.1 4 x
0 . . . 0
O 0.5 1 1.5 2

Figure 3.7. Non-dimensional velocity proﬁle at xlw = 10.8.

The widening of the jet proﬁle as the ﬂow progresses in the streamwise
direction is evident in Figures 3.2 - 3.7 as the ﬂow develops from the "top hat"
proﬁle at the jet exit to the smooth proﬁle at the downstream positions. The
inviscid core region of the flow is evident in the proﬁles for xlw < 5. This region
appears as a ﬂat section in the proﬁle and is very evident in the xlw = 0.0 and xlw
= 1.59 proﬁles (Figures 3.2 and 3.3). The spreading of the jet into the stagnant air
is also evident as the jet advances downstream. The inviscid region becomes
narrower and disappears as the flow advances and the outer sheared region of
the jet and the inner boundary layer advance toward the center of the jet and

eventually meet (at approximately xlw ~ 6).

It is expected that once the jet becomes fully—developed that the data will
collapse to a single curve. Data from x/w = 7.62 and xlw = 10.8 are presented in
Figure 3.8 along with data from Tailland [1967] and Eriksson et al. [1997] at xlw =
600 and x/w = 20 respectively. Note that the decay rate of the maximum velocity
can also be compared between other fully developed wall jets as a means to
verify the two dimensionality of the jet and to insure the self-preserving nature.
This study focused on the developing region however and only two
measurements of the maximum velocity did not include the inviscid core. At xlw
= 7.62 and 10.8, the maximum velocity had decreased to approximately 98% and

94% of the jet velocity respectively.

 

 
  
  
 
 
  
  
  

 

 

 

 

 

 

7
UlUmaxxM=7.82
UlUmaxxM=10.8 + 6
Tailland.)dw=100,1987
mm.m=20.1996
A
SNS-mates ‘r 5 of‘
Taland-SNS-xlwstoo a
Edhem-SNS-Wazo 2
3 ~- 4 r:
3
v
a a
D 7” 3 v-
v
fl)
2
-_ 2 a)
1 1
0
2

via:

Figure 3.8. Comparison of experimental data to results from the literature.

51

It is concluded from the results presented in Figure 3.8 that the self-
preserving, mean velocity proﬁle develops between the measurement locations
of xlw = 7.62 and xlw = 10.8 in the FMTF facility. The experimental data at xlw =
10.8 agrees to within 1% of Um with the fully developed wall jet proﬁles in the
literature. Note that the streamwise normal stress does not agree as well as the
mean velocity proﬁle. The SNS proﬁle measured at xlw = 10.8 is 20% to 30%
lower than Tailland’s data across the jet. This indicates that the mean ﬂow has
just assumed the fully developed form and the higher moments are still
developing [Tennekes and Lumley, 1994] and that the initial jet conditions are still

effecting the ﬂow.

3.2 COMPUTED RESULTS AND COMPARISON TO EXPERIMENTAL DATA

A computational model of the FMTF facility was developed at Ford Motor
Company by Dr. B. AbdulNour. A 2-dimensional, structured, quadrilateral mesh is
generated from the geometry of the flow facility. The mesh contains 188,353
cells and is heavily clustered near the thermally active test surface to improve
resolution and accuracy in this region. The general-purpose CFD package
FLUENT© 5 is used for the numerical simulation. The grid was adapted iteratively
per the solution results in terms of y+ and the velocity gradient. The y+ dimension
of the ﬁrst computational node along the surface ranges from 0.975 to 1.23. Both

a modiﬁed k-e and a Reynolds stress model (RSM) were utilized to model ﬂow

52

turbulence. Mean velocity and SNS results were nearly identical for the two
turbulence models.

The results of the computational model are presented in Figures 3.9, 3.10
and 3.11. Figure 3.9 presents an overview of the CF D results. The development
of the mean velocity and streamwise normal stress proﬁles are evident in this
ﬁgure. Figures 3.10 and 3.11 show the data for xlw = 0.0 and xlw = 10.8.
Experimental velocity data for xlw = 0.0 and 10.8 are included in Figures 3.10

and 3.11 for comparison purposes.

 

 

 

 

 

 

 

 

 

 

_ .- 7
1 )7,
0.9 '
08 + w=159 1“ 6
' —xlw=4.45 ﬂ 5 N...
0-7 —)dw=10.8 g
0.6 I o mass-sue _4 2
=2 05 TL _W=4.45'Sm .:
5 ' —-)dw=10.8—SNS _,3 8
0.4 . t:
ll. (0
0.31 - '_ ~2 5
02 \
+ 1
O1 \
0 L‘ , . ....... 0
0.0 0.5 1.0 1.5 2.0 2.5

ylﬁz

Figure 3.9. Computed velocity results for the wall jet ﬂow ﬁeld. Reynolds stress

model utilized.

   
         
 

 

 

ﬁt: 7
x UlUmax l
0.9 x l
l—WUnax-CFDRSMmOdel‘ 1» 6
0.8 + SNS l
07 —SNS-CFD RSMmodel " 5 "g
0.6 .1 4 or
g 0.5 g
\
D .. 3 2
0.4 v
‘é’
0.3 .1 2 (n
0.2
» 1
0.1
0 . m j o
o 0.5 1 1.5 2
yl82

Figure 3.10. Comparison of experimental and computed velocity results

 

 

 

 

 

 

at x/w = 0.0.
1 1 7
0'9 . x U/Umax
——UIUmax - CFD RSM model ~ 6
‘ + SNS
—SNS -CFD RSM model . 5 N:
a
5
.1. 4 N-\
a
3 8
5
3
... 2 m
, <— 1
\.
0 . . ' 0
o 0.5 1 1.5 2
ylbz

Figure 3.11. Comparison of experimental and computed velocity results

at x/w = 10.8.

The computed jet spreading rate was also compared to the experimental
results. The rate of the spread of the jet can be monitored using the location of
82 as the flow develops. Figure 3.12 contains the jet spreading data. It can be
seen that initially, the computational jet is wider but that the experimental ﬂow
growth rate exceeds the CFD result. It is of interest to note that this is the
approximate location where the experimental ﬂow has lost the inviscid core and
is expected to begin transition to turbulent ﬂow. The experimentally measured jet
growth rate is 32% greater than the computation result, although the jet width at

xlw = 10.8 is only different by approximately 5%.

 

   

 

 
 

+ CFD
+ Experimental 1'

 

 

 

 

 

0 2 4 6 8 1O 12

Figure 3.12. Comparison of jet spreading rates (location of 62).

3.3 DISCUSSION OF RESULTS

The measurements in the developing region of the wall jet illustrate the
transition of the jet from the slug like proﬁle at the contraction exit (xlw = 0.0) to
the onset of the fully developed mean velocity proﬁle. The inner boundary layer
of the experimental flow is laminar like at the exit of the two dimensional
contraction. Figure 3.13 shows a comparison of the inner and outer boundary
layer proﬁles at the contraction exit to the Biasius similarity solution proﬁle

[Bejan, 1995].

 

 

 

 

 

 

 

 

 

1.2
1 _ LY W13 G
A
A ‘6 °
0 8 7 A0
a 40
ﬂ A
2 0-6 130° —- Biasius solution
3 AG
o_4 - 0 wall edge, y =0 mm
9 A outer edge, y =20 mm
0.2 1;
0 I I
0 2 4 6 8

Figure 3.13. Boundary layer proﬁles at xlw = 0.0 compared

to the Biasius laminar proﬁle.

This laminar like proﬁle indicates that the ﬂow must begin to undergo a
transition to turbulence prior to xlw = 10.8. The transition phenomenon can be
inferred since the mean velocity proﬁle at xlw = 10.8 conforms to the fully
developed, turbulent wall jet proﬁle. The experimental results agree
exceptionally well (typically < 1% of Um) with the fully developed mean proﬁle
results extracted from the literature as seen in Figure 3.8.. This level of
agreement is utilized to validate the measurement techniques and establishes
conﬁdence in the methods and equipment utilized for this study.

Comparing the measured velocity results with the CFD model illustrates
two major discrepancies. The boundary layers at the contraction exit (xlw = 0.0)
are thicker in the computed result than those measured. The inner boundary
layer also thickens more rapidly as the ﬂow proceeds downstream in the
computational model than in the experiment even though the computed wall jet
proﬁle growth rate is less than the experimental result. The experimental 62
grows from 18.2 to 27.7 mm from the slot exit to xlw =10.8 while the
computational result grows from 19.3 to 25.7mm in the same region.

The most signiﬁcant difference between the experiment and the
computation is in the inner boundary layer region near the thermally active test
surface. The computational boundary layer grows thickens at a greater rate than
the experimental measurements indicate. This is clearly evident if the peak
location of U/Um is considered in Figures 3.10 and 3.11. The CF D code over-
predicts the growth of the inner boundary layer. The code is utilizing a

turbulence model that is active throughout the ﬂow ﬁeld even though the

57

experiment indicates that there is a laminar like boundary layer at the jet exit (xlw
= 0). Thus the computational ﬁow does not develop a laminar boundary layer
that undergoes a transition to turbulence. This would also impact the velocity
gradient (shear stress) at the wall, which would be expected to strongly inﬂuence
the heat and mass transfer. In light of these considerations, a computational
model with a divided mesh in which non-turbulent regions of the ﬂow could be
inserted and modeled is recommended to attain more accurate simulation
results.

The computed result does approximate the experiment reasonably well.
The largest variations in the mean velocity proﬁles between the experiment and
the CFD occur in the inner boundary layer region. At xlw = 4.45, y/82 = 0.1, there
is a 13% difference in the mean velocities. This is the largest variation between
the results and it decreases to an 8% variation at x/w =10.8. The decrease in the
maximum mean velocity at each streamwise location agrees to within 1.5%
between the experimental measurements and the CFD results.

The Stream-wise Normal Stress (SNS) proﬁles agree within the
experimental variation at every outer peak (in the outer shear layer) location. The
SNS peak in the inner boundary layer differs by as much as 28% between the
experimental and the computed results. This difference is probably also due to
the use of a turbulence model throughout the ﬂow. The inner and outer peaks of
the computed SNS proﬁles at xlw = 10.8 agree with the experiment within the
experimental variation. Results from a CFD model with non-turbulent grid zones

incorporated near the jet oriﬁce have not been provided for comparison at this

time, but they are expected to improve the comparison between the experiment

and the computation, especially in the inner boundary layer.

59

Chapter 4

STEADY STATE, NON-ISOTHERMAL, NON-CONDENSING MEASUREMENTS

The next phase of the project was to measure thermal effects in the ﬂow
ﬁeld. The thermally active test section was chilled to add a thermal boundary
layer to the ﬂow. The temperature of the test surface was carefully monitored to
prevent condensation from forming. This ﬂow ﬁeld added signiﬁcant
complications to the experimental technique as it required local temperature
compensation for the hot-wire probe to be utilized. The hot-wire probe operates
based on convective heat transfer and is therefore sensitive to temperature
changes in the ﬂuid ﬂowing over the sensing wire. The data are utilized to verify
the results of the CFD model with energy (heat) transfer occurring in the ﬂow
ﬁeld. The following sections detail the experimental methods, uncertainty
analysis, and experimental results. Comparisons to the isothermal work are
made and the experimental data are compared to results from the computational

model.

4.1 EXPERIMENTAL METHODS

A single 5 pm diameter tungsten hot-wire was utilized for non-isothermal,
non-condensing ﬂOw measurements to measure the streamwise velocity. Hot-
wire sensors can be utilized in non-isothermal ﬂows if the local ﬂuid temperature
can be measured and implemented to compensate the velocity measurement

[Goldstein, 1996]. A cold-wire resistance thermometer constructed from 5 pm

diameter tungsten wire was utilized to measure the local ﬂuid temperature. A
cold-wire uses the temperature—electrical resistance material property to measure
temperature electronically. The cold-wire probe was calibrated to determine the
temperature-resistance relationship empirically. A very small sensing current (~1
mA) was used to measure the resistance (and therefore, the temperature) in the
experimental wall jet ﬂow ﬁeld.

The velocity and temperature measurements in the non-isothermal ﬂow
were calculated from the hot-wire and the cold-wire signal using the following

procedure. The temperature was calculated ﬁrst from the cold-wire voltage

 

signal with
E—C
T o ~wire = 1 4.1
c ld C2 ( )

The quantities C1 and C2 were obtained from an ordinary least squares
curve ﬁt of the cold-wire calibration data. The measured voltage, Emma, from
the hot-wire was compensated utilizing the temperature output from the cold-wire

with

T —T
E 22E m 2[ hw cal] 42
Compensated M ured -——Thw—Tcw (.)

61

Equation (4.2) compensated the hot-wire voltage for changes in the heat
transfer relationship due strictly to variation of the local temperature from the
temperature present when the hot-wire was calibrated. The velocity was then
calculated with the compensated voltage value, Ecompensated, utilizing Equation

(4.3) presented by Collis and Williams [1959]:

E2 = A + BVn (4.3)

The quantities A, B, and n were obtained from an ordinary least squares
curve ﬁt of the hot-wire calibration data. The cold-wire and the hot-wire probes
must be calibrated prior to every experiment. Post calibration was also conducted
if the probes survived the experiment intact. Note that Equation (4.3) was utilized
until the TSI 1050 anemometer was replaced with the TSI lFA300 anemometer
system. The lFA300 software replaced Equation (4.3) with a fourth order
polynomial ﬁt to the calibration data.

The sensors were aligned parallel to the wall along the same Spanwise
axis. The active sections were separated by 2.2 mm. Both sensors were
constructed on site from 5 pm diameter tungsten wire. A TSI anemometer was
utilized to control the sensors and a 12 bit Keithley Metrabyte AID card recorded

their output at 8000 Hz for later data processing.

Wire sensors

/
I3mm

  
 

<:"

1 Probe body

  

Figure 4.1. Schematic of the hot-wire/cold-wire dual probe sensing head.

The sensor positioning and data acquisition systems are conﬁgured as

described in Chapters 2 and 3.

4.2 EXPERIMENTAL UNCERTAINTY

A hot-wire sensor is calibrated by positioning the probe in a ﬂow with a
known velocity as described in Chapter 3. The uncertainty in the thermally active
experiments was slightly larger than the isothermal uncertainty due to the
additional experimental variance added with the addition of the second sensor
(the cold-wire). The experimental uncertainties in the non-isothermal ﬂow of the
velocity and temperature were calculated to be 3.5% of Um and :l: 0.5 °C
respectively. The uncertainty in the velocity measurements was determined from
the experimental variation from test to test. The temperature uncertainty was
determined from the variance of the data utilized for the calibration data.

A cross correlation study was conducted with two hot-wires placed 4 mm
apart in the ﬂow. A correlation coefﬁcient of 0.63 was measured indicating that
the hot-wires were not necessarily measuring the same ﬂow structure. The

temperature and velocity are not necessarily correlated in this ﬂow, but it is

63

essential that the hot-wire and cold-wire probes are measuring the same ﬂuid
mechanical event in order for the hot-wire temperature compensation to function
properly. The worst scenario resulting from uncorrelated sensors is examined
next.

The hot-wire measurement is compensated as indicated in Equation (4.2).
The worst case scenario can be realized if the temperature and the velocity
measurements are completely uncorrelated and out of phase. For example, the
maximum temperature would occur locally at the hot-wire probe while the
minimum (and incorrect) temperature ﬂuctuation occurs at the cold wire probe.
The maximum effect of this test case can be analyzed using the temperature
data obtained in the ﬂow. These data indicate that the maximum temperature
ﬂuctuation can be estimated as approximately 0.15 (T, - T.) °C. The value of 0.15
is twice the RMS (standard deviation) measured at xlw = 10.8. This equates to 3
°C in a typical experiment. The error (AE) created in the hot-wire voltage can be

calculated utilizing Equation (4.4).

AB:

1 l

E (T... —T...) 5 _ (T... —T...> 5 (4.4)
“W“ T,w 4“,, —0.15(T,. —Tp) Thw 4,, +0.15(T,. —Tp)

 

 

Equation (4.4) results in a 1.26% error deviation from the correct
compensated voltage. This will cause a maximum error of approximately 6.7% in

the velocity at U= Um (67cm/s at 10 mls) utilizing a typical hot-wire calibration

result. However, the peak temperature ﬂuctuation does not coincide with the
maximum velocity. Thus the error can be more accurately estimated at 1.9% of
Um (19 cm/s) in the inner boundary layer of the ﬂow. Also note that this is the
two standard deviation ﬂuctuation level and thus 95% of the actual error is likely
to be signiﬁcantly less. Recall that the integral length scale of me ﬂow was
determined to be approximately 8 mm. This level of error was deemed
acceptable since the actual wire sensing sections are closer together than 4 mm

and the correlation coefﬁcient is not actually zero.

4.3 MEASUREMENT RESULTS OF THE NON-ISOTHERMAL, NON-CONDENSING
EXPERIMENTAL REGIME

The wall jet temperature was typically 24 °C (but varied with ambient
conditions slightly from day to day (:t 1 °C)) and the temperature of the thermally
active test section was set at 4 °C. The ambient humidity varied from 15% to 25%
Rh from day to day, but was constant for the duration of each non-isothermal
experiment to within 2% Rh. Note that for these conditions the dew-point never
exceeds 3.6 °C.

Within the measurement uncertainty, the velocity proﬁles measured in the
non-isothermal, non-condensing ﬂow showed no deviation from the isothermal
velocity measurements. This was expected due to me small changes in the ﬂuid
properties over the temperature range present in the thermal boundary layer.

This result was further supported by a comparison of the Grashoff number and

the Reynolds number. This comparison indicated that the momentum effects
were several orders of magnitude larger than the buoyancy effects. Based on
these assumptions, it was expected that the thermal boundary layer would have
a negligible effect on the velocity proﬁle in the wall jet ﬂow. The data displayed in
Figure 4.2 show the comparison between the measured velocities at xlw = 10.8
in the isothermal and non-isothermal, non-condensing experimental regimes.
Figure 4.2 contains 3 non-isothermal, mean velocity data sets and the ensemble
average of 5 isothermal experimental data sets. These non-isothermal data sets
were statistically indistinguishable from the isothermal data with a null hypothesis

test at a 95% conﬁdence level.

 

 

x MMsMal UlUmax ’
olsotherrnal U/Umax " 6
o Non-isothermal SNS
nisothermal SNS

 

UIUHIIX
SNS

 

 

 

0 0.5 1 1:5 2
W82
Figure 4.2. Comparison of velocity proﬁles in the non-isothermal, non-

condensing and isothermal experimental regimes; xlw = 10.8.

The isothermal estimation of the wall shear stress was utilized to non-
dimensionalize the velocity data into wall coordinates due to isothermal velocity
measurements obtained closer to the wall than the non-isothermal, non-
condensing measurements. The single hot-wire isothermal measurements were
simpler to perform which made obtaining near wall velocity data easier. The near
wall, isothermal velocity measurements were compared to other experimental
measurements of the wall jet proﬁle such as the data from Wygnanski et al.

[1992]. This comparison is presented in Figure 4.3.

24
20 -

 

 

 

 

 

1 10 100 1000

1|-

Y

x U+ (10.8 - Present results)
0 Wygnanski etal., Re=19000

_ Wygnanski et al. ﬁt, Re=13000

 

 

 

Figure 4.3. Comparison of the self preserving momentum proﬁle in inner

coordinates to data from Wygnanski et al. [1992]; xlw = 10.8.

67

Figure 4.3 indicates the agreement between the 5 individual, isothermal
experimental data sets and that of Wygnanski et al. This comparison indicates
that the near wall velocity measurements are consistent with data in the
literature. The shear velocity was estimated from the linear velocity proﬁle

measured near the test surface. Wygnanski also utilized near wall (y < 400 pm)

velocity measurements to calculate the shear velocity, U,. This method tends to
be less ambiguous than the Clauser method of estimating shear velocities. Note
that the larger difference between the results beyond y“ = 100 is due to the larger

velocities in the outer shear layer associated with a higher slot Reynolds number.

4.3.1 NON-ISOTHERMAL VELOCITY PROFILES

Detailed velocity proﬁles of the non-isothermal wall jet ﬂow ﬁeld were
taken, but were indistinguishable from the isothermal proﬁles measured
previously. This result was critical to the experimental program because it
demonstrated that the hot-wire temperature compensation with the cold-wire
measurement was functioning properly. Error in the compensation technique or
in the cold-wire measurement would cause measurement errors in the velocity
proﬁle. Figure 4.4 through Figure 4.8 present the measured time-averaged
velocity data and the streamwise normal stress (SNS) in the non-isothermal wall
jet ﬂow for the experimental locations xlw = 0.0, 1.59, 4.45, 7.62, and 10.8,
respectively.

The number of data sets, 11, is indicated in each ﬁgure title. The reduction

of the inviscid core due to growth of the inner boundary layer and outer shear

layer is evident in these ﬁgures. The ﬁgures also show the increasing magnitude
of velocity ﬂuctuations near the wall and the inward propagation of the shear

layer turbulence as the ﬂow advances in the streamwise direction.

 

 

 

 

 

 

1 wx%x if ‘f RX 7
XX)“ X X xx
x x _
x x xU/Umax +6
0.8a x A
x x +SNS b5 01‘
x X i
0.61)< X *4 3
2 x x a
a 4
5 x -3 8
0.44 X S
(I)
X #2 a
0.2— x
x ~1
X x
Oft‘“++++ +T+ + ++++f3+ + o
.5 1 1.5
° 0 WM

Figure 4.4. Velocity proﬁle at xlw = 0.0; non-isothermal; T. = 4 °C; 11 = 1.

69

 

08|

06

'IE:\D

04

02

Figure

08
0.6
04

x IED\D

02
o
1 F 19071

 

 

 

 

 

 

 

1‘WWW§X
§ x
x xU/Umax 1* 6
_l
08 x x +SNS A
x: x *5 NA
0.6 1“ g
2 i! x "*4 “I
x 3
a v
3 E .. 3 s
0.4 J 4'“ S
U)
+ I __ 2 z
+ x (D
0.2 J ++ +X
++ +1
++ 4» xx
1&2, + x"xXxxxxxxxxxx
0 -I_m+m4+_ttd______r_£_u+¢++++++++++— 0
0 0.5 1 1.5 2
via:

Figure 4.5. Velocity proﬁle at xlw = 1.59; non-isothermal; T. = 4 °C; 11 = 2.

 

 

 

 

 

 

 

 

1 _, ir‘v‘w‘xXXXXx 7
x
x ~~ 6
0.3 _ g: x xUIUmax
g x +SNS 4, 5 N:
a: g
0.8 ~ x __4 a?
g :1
x V
5 «- 3 3
0.4 d + + X F
+ V
m
+ + X «2 z
+ x m
0.2 J + x
<—1
1' X
s + + " x x
o +'H'+++ 1' + + Y 1 1 + + .1. i, 0
0 0.5 1 1.5 2
WM

Figure 4.8. Velocity proﬁle at xlw=4.45; non-isothermal; T. = 4 °C; n = 1.

7O

 

 

 

 

 

1 -—W 7
.1“ ‘1‘“
we r—‘Aﬂ
1‘ gt,“ x U/Umaxl ~- 6
0.8 g N. + SNS
E .. "5 E
aux
x 0.6 x ”Xx 4 "T
5 s a
\ ’5 o
D ~L 3 c
0.4 ~ + 1: $15!; 4+ + r,
I *4' )‘x1r : ++ w
+ + z
+ 411 ”(111): I ++ 2 I")
02 , + ++ ++ I:
air, at? u“ t #1 1
‘ "Aw: t
0 r r I I o
0 0.5 1 1.5 2

Y’5 2

Figure 4.7. Velocity proﬁle at xlw = 7.62; non-isothermal; T. = 4 °C; n = 4.

 

 

 

 

 

 

 

 

X
—~6
Xx
0.8 g Xx xU/Umax
X
a xxx +SNS 55“.
x X X
20.6; xx». «4 :1
3 XXX "1:
5 + X x 3 v
+ -1—
0.4 J t+t+tl~+ﬁ+++ ++++++++ X §
+ V
++11+ + +¥ ¥ * }_ 2 m
+ 2
16+ it
02 3+“ * + w
' Etc 1 * I
+*" « 1
0 l w I 0
0 0.5 1 1.5 2
WE:

Figure 4.8. Velocity proﬁle at x/w = 10.8; non-isothermal; T; = 4 °C; n = 3.

71

 

expect
momer
experlr
numbe
implerr
lndicat.

solve tl

menu
mamet
The re-
exPeril
thus m
The Ch
each 6

0.003 1

separa

Can be

The conclusion drawn from the velocity measurements Was that, as
expected , the presence of the thermal boundary layer did not affect the
momentum characteristics of the ﬂow to an extent that was detectable with the
experimental methods applied. This veriﬁed the results of the Grashoff-Reynolds
number comparison and provided evidence that the cold-wire measurement was
implemented properly to compensate the hot-wire output. This result also
indicates that the CF D simulation could utilize an uncoupled methodology to

solve the momentum and energy equations without introducing signiﬁcant error.

4.3.2 NON-ISOTHERMAL TEMPERATURE PROFILES

The temperature data were taken directly from the cold-wire
measurements utilizing Equation (4.1). The cold-wire was constructed from 5 pm
diameter tungsten wire, 1 mm in length with a typical resistance of 3-5 ohms.
The resistance of the wire was measured with a 1 mA sensing current during the
experiments. This low current resulted in negligible heating of the sensor and
thus minimized the cold-wire sensitivity to changes in the convective velocity.
The change in the resistance as a function of temperature was calibrated prior to
each experiment. The coefﬁcient of resistivity for tungsten is approximately
0.003 mm °C). A very low noise ampliﬁer is required to process the signal.

The frequency response of the 5 um cold wire is approximately 400 Hz
[T hole, 1992]. The ﬂow ﬁeld frequency based on the momentum thickness of the
separating laminar boundary layer and the free stream (inviscid core) velocity

can be calculated from the Strouhal number [Ho and Huerre, 1984].

72

 

 

Fhf
ﬁec
cnu
ﬂuc
the

ten1
Sen:
0f1:
Dan
obna

Were

dlan-
aver

_ f 0
" TIT (4.5)

..J
2

St

The Strouhal number for laminar boundary layer separation is 0.032,
which results in a frequency of approximately 100 Hz for the conditions in the
FMTF. Thus the cold-wire temperature sensor is capable of measuring at a
frequency of more than twice the ﬂuctuation frequency, meeting the Nyquist
criterion, and capturing temperature ﬂuctuations associated with velocity
ﬂuctuations (transport). A faster temperature sensor could be utilized to verify
the dynamics measured with the 5 pm sensor. A greater sensitivity to
temperature and a faster time constant can be obtained with a smaller diameter
sensor. A 2.5 pm diameter tungsten wire has an estimated frequency response
of 1200 Hz (Thole, 1992] and was incorporated to realize these improvements.
Data obtained with the 2.5 pm wire probes were indistinguishable from the data
obtained with the 5 pm wire probes. The 5 pm probes were utilized since they
were easier to manufacture and generally more robust.

The temperature proﬁles at xlw = 10.8 were veriﬁed with a 7.6 pm
diameter micro-thennocouple. The thermocouple measured the same time-
averaged temperature proﬁle as the cold-wire sensor, but it measured a slightly
lower RMS proﬁle. The lower ﬂuctuation measurement is most likely the result of

the slower time response of the thermocouple damping out ﬂuctuations.

73

 

 

 

nond
respel
the let

showr

SEt of
layer.
to the
ﬂat pl;
from U
differe

the inr

Figure 4.9 through Figure 4.12 present the cold-wire temperature data at
non-dimensional streamwise locations of xlw = 1.59, 4.45, 7.62 and 10.8,
respectively. The temperature proﬁles have been non-dimensionalized based on
the temperature difference between the test surface (T3) and the jet exit (T,) as

shown in Equations (4.6) and (4.7).

e—T_n
_ T} -T; (4.6)
._ T'

6 — Tj —Ts (4.7)

All the ﬁgures contain 3 data sets except Figure 4.10, which contains one
set of data. The ﬁgures reveal the expected thickening of the thermal boundary
layer. Other inferences based on ﬂat plate boundary layer theory are limited due
to the signiﬁcant differences in the general ﬂow features of a wall jet ﬂow and a
ﬂat plate boundary layer ﬂow. Speciﬁcally, the propagation of the ﬂuctuations
from the outer shear layer into the inner boundary layer represents a signiﬁcant
difference and results in quicker initiation of the transition to a turbulent ﬂow in

the inner boundary layer.

74

 

 

 

 

 

1JAoﬁoﬁAoooﬁo°o°oQ>oo°°°A°<°°°ao<b<bQ
0
s
0.8 ~ 0°
<3 ° 9 l
3 no"
0..., __l
ED 8
co”
0.4 l
0.2 4
o- . l
0 0,2 0.4 0.6 0.8 1
via:

Figure 4.9. Non-dimensional temperature proﬁle at xlw = 1.59; T. = 4 °C

 

 

 

 

 

 

 

 

1 o°°0v 6%7° 6 ° 0 ° ° ° 0 o o o
o
0.8 4 5’
o
o 06
o
0.6 - o A 9'
ED
co“
0.4 ~
0.2 —
oﬁdAAAAIAAAAA|AA ATA A A A
0 0.2 0.4 0.6 0.8 1
Y152

Figure 4.10. Non-dimensional temperature proﬁle at xlw = 4.45; T. = 4 °C.

75

 

 

 

 

 

 

 

1 ofw‘o 08 O: 08 v80 09 0° ‘09 00° 0% oo 0
03° 0 0 ° 0
$
0.8 o
J ° 9
0
o 6 w; A 9
. I %
co 2
co“ 0
0.4 -
0.2 -
0 W “A“
0 0.2 0.4 0.6 0.8 1
Y’az

Figure 4.11. Temperature proﬁle at x/w = 7.62; T. = 4 °C.

<>
<9
<>

 

 

 

 

 

1 oo<9°o®<vg 0 Voeooovvv
W o o
0.8- [09
de'
o.6«°
ED 0
d O
0.4‘0
0
0.2‘
o MMAMMAMMA AA AA A‘A AA AA AA AA AA
0.6 0.8

 

0 0.2 0.4

Y’s:
Figure 4.12. Temperature proﬁle at xlw = 10.8; T8 = 4 °C.

76

4.4 SUMMARY OF EXPERIMENTAL RESULTS

The measured velocity proﬁles in the non-isothermal jet are unaffected by
the thermal gradients near the wall. This information was inferred from a
Grashoff-Reynolds number comparison that indicated that the inertial effects
were several orders of magnitude larger than the buoyancy effects. The small
changes in the ﬂuid properties associated with the temperature gradient within
the thermal boundary layer also support this assumption. The comparison
between the measured velocity proﬁles in the isothermal and the non-isothermal,
non-condensing experimental ﬂows conﬁrm that no measurable differences exist.

The comparison of the isothermal and non-isothermal velocity data at xlw
= 10.8 to published data in the fully-developed, self-preserving region indicate the
quality of the measurements taken. The agreement with the published, fully
developed proﬁles provides a high level of conﬁdence in the measurements
taken in the developing region of the wall jet.

The developing region of the wall jet is of particular interest because of the
similarities to the defroster/defogger ﬂow ﬁeld. Both ﬂow ﬁelds exit from high
aspect ratio openings and issue across thermally active surfaces. Actual
defroster ﬂows impinge on the surface of interest and generally have higher
turbulence intensities than the ﬂow studied here. The developing region of the
wall jet is sensitive to the particular geometry of the test facility and must
therefore be quantiﬁed experimentally for accurate comparisons to computational
results to be performed. The velocity and temperature measurements from the

FMTF provide a powerful opportunity for direct comparison of the experimental

results and the CFD predictions of the ﬂow ﬁeld. This comparison can be
implemented to validate or improve the computational models utilized for
practical design applications.

The excellent match of the isothermal and non-isothermal experimental
velocity proﬁles at each xlw location indicate that the hot-wire temperature
compensation in the thermal boundary layer is functioning properly. The velocity
proﬁles from the two different experimental regimes are statistically
indistinguishable. The hot-wire velocity measurement is corrected for local
temperature variation with the cold-wire temperature measurement. The
combination of these two probes in the boundary layer allows simultaneous
measurement of the streamwise velocity, temperature and their ﬂuctuating
components. This arrangement allowed local temperature compensation for the
hot-wire to improve velocity measurement accuracy. These measurements
provide a useful set of benchmark data for comparison to the temperature ﬁeld
calculations from the CFD simulations.

The experimental data also indicate that the CF D simulations can be
performed with an uncoupled methodology. This methodology refers speciﬁcally
to whether the momentum and energy equations need to be solved
simultaneously. The fact that the experimental velocity proﬁle is independent of
the temperature proﬁle indicates that an uncoupled solution can be implemented.

The uncoupled model is much more efﬁcient to implement computationally.

78

4.5 COMPARISON OF EXPERIMENTAL DATA To CFD RESULTS

The computational model utilized an uncoupled solution scheme. The
velocity proﬁle was solved independent of the thermal effects. This is the
equivalent of solving the isothermal ﬂow ﬁeld. The isothermal velocity solution
then provides the necessary information to solve the energy equation and arrive
at a solution for the thermal boundary layer. This solution technique was
assumed adequate since the experimental measurements indicated no
measurable change in the velocity proﬁles (see Figure 4.2). This technique
allows for the simulation to be conducted in less time. The comparisons of the
velocity results are identical to the results presented in Section 3.2 and will not be
repeated here.

Figures 4.13 through 4.15 present the comparison of the thermal

boundary layer in the experimental facility and the computational results.

79

 

 

 

 

 

 

 

 

103AA‘o‘ooovo°o‘ero°vo OOOQQQ$
0.8-
9 °9
9 A 9'
. 0-6‘ —CFD-RSM model
CD
a3"
0.4"
02.
OMAAAAAAAA:AMMMAAMA A? AAM¢AMMM
0 02 0.4 06 08 1

y/82

Figure 4.13. Comparison of the experimental and computed temperature proﬁles

at xlw =1.59;T8 = 4 °C.

 

    
    
  

 

 

 

o o o o
0 6
ll 9'
——CFD-RSMmodel
0.2«
OMAAAAAAILAAA'AA A‘A A‘A A
0 0.2 0.4 0.6 0.8 1

WE 2

Figure 4.14. Comparison of the experimental and computed temperature proﬁles

at xlw = 4.45; T; = 4 °C.

80

A AA
v vovovouovVVﬁ

 

 

 

 

 

 

 

 

CD
a)" —CFD - RSM model
0.4'
0.2~
DWAymﬁMMM'AAAAAﬁAAAAAgAAAAA
0 0.2 0.4 0.6 0.8 1

W52

Figure 4.15. Comparison of the experimental and computed temperature proﬁles

at xlw =10.8;Ts = 4 °C.

The mean (time - averaged) temperature proﬁles indicate that the thermal
boundary layer (as determined by 87(x/w)) in the CF D simulation thickens at a
greater rate than the experimental measurements indicate. This is probably the
result of the turbulence model once again and is directly coupled to the fact that
the momentum boundary layer (8(x/w)) thickened at a greater rate in the
simulation than in the experiment. This information supports the use of a
transition model to activate the turbulence model or a zonal grid structure to allow
the turbulence model to be implemented at a predetermined ﬂow location.

The mean temperature proﬁles at xlw = 1.59 agree within the experimental

uncertainty. The comparison between the experiment and the computation

81

continues to degrade as the boundary layers thicken at different rates as the ﬂow
progresses downstream. At xlw = 10.8, the proﬁles differ by approximately 8% of
the maximum temperature difference (or about 1.6 °C). The temperature
gradient at the test surface (dT/dy I y=0) is up to 61% larger in the computed result
than in the experimental data. This indicates that the computational model over-
predicts the heat transfer at the thermally active surface. The greater estimate of
the heat transfer would lead to under-prediction of the time necessary to defrost
or defog a windscreen. Federal law speciﬁes defrosting times for passenger
vehicles (FMVSS103). Successful modeling would reduce or prevent
modiﬁcations to existing designs once prototype testing is initiated. The data
obtained provides the opportunity to tune the computational model to agree with

this model ﬂow ﬁeld and potentially improve the computational capabilities.

Chapter 5

STEADY STATE, NON-ISOTHERMAL, CONDENSING MEASUREMENTS

The steady state, non-isothermal, condensing regime was studied with
two separate methodologies. Velocity and temperature proﬁles in the ﬂow ﬁeld
were measured with the same techniques detailed in Chapters 3 and 4. The
overall mass transfer rate to the thermally active test section was also studied.
The steady state mass transfer was studied as a limiting case to determine
overall system performance in the Ford-MSU test facility and to measure the time
averaged and spatially averaged mass transfer coefﬁcient on the thermally active
test section. These data provide insight into the global performance of the
system and established another validation data set for direct comparison to the
computational work being performed at Ford Motor Company. The average mass
transfer coefﬁcient (hm) on the thermally active test section in the FMTF wind
tunnel was measured at 0.0407 m/s for a jet Reynolds number of 13,300. This
result was within 10% of the computed result obtained with non-turbulent grid

zones incorporated near the jet oriﬁce.

5.1 TEMPERATURE AND VELOCITY MEASUREMENTS

Hot-wire and cold-wire measurements in the non-isothermal, condensing
ﬂow regime were conducted. The steady state, non-isothermal, condensing
experiments were conducted with the thermally active wall temperature set at

approximately 4 °C and the relative humidity of the wall jet set at 50%. The

ambient temperature was approximately 24 °C (:l: 1 °C) for all of the experiments.
The time-averaged velocity proﬁle is expected to agree with the isothermal
velocity proﬁle due to the small changes in the ﬂuid properties over the
temperature range utilized and the low rate of mass transfer associated with the
condensation. The three experimental velocity measurements shown in Figure
5.1 bear out these expectations when compared to the ensemble average of the
isothermal velocity proﬁles. These measurements were complicated by the
presence of a thin layer of liquid H20 (the condensate) on the test surface. It is
believed that this ﬂuid destroyed the hot-wire when contact occurred. This made
near wall velocity measurements difﬁcult to obtain. However, the cold-wire was
more robust than the hot-wire due to the very low overheat utilized. The cold-
wire was infrequently destroyed during the experiments and was capable of

obtaining temperature data closer to the test surface than the hot-wire probe.

 

 

 

 

 

 

 

 

 

xU/Umax
xx ObotliermalWU'nax «l 6
0:55 OSNS
”'5 UlsothermalSNS .L 5 A
2‘ of‘
’6“ i
"‘ i
o ’13. ﬂ 4 '1
o 0 ° x753 3
° ° ° cm 8
0 c0 ..
aaﬂbpﬂgawﬁq’geﬁtx" 3 a
o 0153:”)
0 Do 30 5
0° )Q‘g ‘* 2
0" M
00
~ 1
Y r I f ' O
0.6 0.8 1 1.2 1.4
Y’s:

Figure 5.1. Comparison of velocity proﬁles in the non-isothermal, condensing

and isothermal experimental regimes.

The temperature proﬁles in the ﬂow were measured next. Results from
the literature indicated that a measurable shift in the temperature proﬁle might be
present change [Legay-Desesquelles and Prunet-Foch, 1986]. The time-
averaged temperature proﬁles presented in Figure 5.2 show a statistically
signiﬁcant difference in the near wall region (ylw<0.03) between the non-
isothermal, non-condensing experiments and the non-isothermal, condensing

experiments. A 95% conﬁdence level was utilized for the null hypothesis test.

 

 

 

  

 

 

 

1 I xoﬁo—xqu‘mf ’99 “YT—ax ,.
l x wg‘g‘ x x
I
a»)
q
0'8 ,. x6 Condensing
:2.
43* 06 Non-Condensing
« e
0.6 ﬂ .“Q 00 Condenslng
_ x0 (80' Non-Condensing
CD. 0
CD x
.4 ‘ x
0 O
O
0.2 J
MMﬁmAﬁwAgwm °R° °R° W0 299 0 0 A90 0

 

 

 

O r
0 0.25 0.5

yl82

Figure 5.2. Comparison of temperature proﬁles in the condensing and non-

condensing, non-isothermal experimental regimes; T; = 4 °C.

The effect of the latent heat release created by the condensation could be
evident as an increase in the slope of the temperature proﬁle at the wall. The
temperature gradient would increase if the latent energy is released
volumetrically near the wall, if mixing is improved due to the additional surface
roughness created by the ﬂuid structure, or if a portion of the latent heat is
convected from instead of conducted into the test surface. The saturation
temperature for the experimental conditions was computed to occur (for constant

total pressure) at y/62 of 0.1 (y z 450 pm). This could result in volumetric heat

88

generation within a condensing vapor adjacent to the wall as the latent heat is
released. The difference in the temperature data in Figure 5.2 is consistent with
this assumption. HoWever, no vapor layer was apparent during the experiments,
which is consistent with Matuszkiewicz and Vernier [1991]. Matuszkiewicz and
Vernier [1991] state that droplet formation in ﬂows with Le<1 is impossible in the
condensation boundary layer if the free stream droplet mass fraction is zero and
the temperature gradient is small. The conditions in the FMTF ﬂow are
consistent with this speciﬁcation and therefore no vapor is anticipated.

The regression curves ﬁt to the non-isothermal, non condensing and the
non-isothermal, condensing temperature measurements are statistically different
for values of ylw less than 0.03 at a 95% conﬁdence level, assuming that all the
errors present are purely random. The difference was determined by comparing
the conﬁdence intervals for each regression result. The data were taken with a
hot-wire/cold-wire dual probe and the experiment was terminated when the hot-
wire signal indicated failure. This limited the nearest temperature measurement
to the wall to be at a y" location of approximately 10, using the isothermal shear
velocity (U,). The isothermal shear velocity was utilized because it was better
resolved than the non-isothermal, condensing velocity measurements and no
signiﬁcant difference in the velocity proﬁles were anticipated or detected (see
Figure 5.1).

The increase in the non-isothermal, condensing temperature proﬁle is
consistent with predicted and observed trends in the literature for convective

ﬂows with phase change [Legay-Desesquelles and Prunet-Foch, 1986]. The

87

addition of mass transfer to the ﬂow increases the effective heat transfer and
thus raises the effective Nusselt number. This alteration of the heat transfer
directly impacts the temperature proﬁle in the flow. The surface temperature was
maintained at a constant value. However, the heat ﬂux into the surface was not
measured during these experiments. Veriﬁcation of the offset could be
investigated by installing a heat ﬂux meter in the test surface and comparing the
change in the heat ﬂux conducted into the surface to the total quantity of mass
condensing (and thus the total latent heat released) on the surface. The
difference in the heat ﬂux into the thermally active test surface and the total latent
heat released would represent the total additional heat ﬂux convected away by

the wall jet ﬂow ﬁeld.

5.2 STEADY STATE MASS TRANSFER MEASUREMENTS

Steady-state condensation measurements were conducted by collecting
the ﬂuid draining from the bottom edge of the thermally active test section. The
thermally active test section is aligned vertically to allow the ﬂuid to ﬂow off
during experiments (see Figure 2.2, item 9 and Figure 5.3). The collection and
measurement system utilized to measure the steady state condensation rate is
presented in Figure 5.3 [Kinnen, 2000]. A funnel with a narrow inlet was
attached to the lower edge of the thermally active test section. A beaker was
constructed with a narrow inlet to collect and contain the condensate. The
beaker provided 3 narrow slits for the airﬂow to escape. This anangement was

chosen to minimize the possible evaporation of the condensate collected. Tests

were also conducted with no active condensation occurring on the test section to
measure the quantity of ﬂuid evaporating from the beaker. Flow humidity levels
were varied throughout the experimental range during the evaporation tests. The
steady state condensation measurements were corrected for evaporation based
on these results. The correction for the evaporative losses was of the order of 5
to 10% of the total condensate volumes and was a linear function of the

concentration difference as expected.

Thermally active test
section

Funnel

Partitioned and
enclosed beaker

Digital Scale

 

Figure 5.3. Steady state condensation - collection and measurement system.

The mass of the condensate in the beaker was recorded with an electronic
scale with an RS-232 output linked to a PC. The experiment was allowed to
equilibrate for at least 30 minutes before data acquisition was initiated. Data were
acquired for 30 minutes and then the total mass transfer rate was calculated. The
mass accumulation data as a function of time were ﬁt with a linear regression
model. The results were highly linear with r2 = 0.98 or larger.

The experimental results and numerical steady-state condensation
predictions are presented in Figure 5.4. The data are plotted as the total
condensation rate to the thermally active test section (grams/min) as a function of
the concentration difference in the ﬂow ﬁeld (kg/m3). The concentration
difference is based on the humidity and temperature at the exit of the two
dimensional contraction and the average surface temperature at the centerline of
the thermally active test section. Note that the presence of the condensate layer
was calculated to have a negligible effect (< 0.01 °C) on the accuracy of the
thermocouple measurements. This is consistent with the literature, which
indicates that the interface and surface temperature may be expected to be the
same for ﬂows with speciﬁc humidities (co) less than 0.1 kg H20! kg air
[Matuszkiewicz and Vernier, 1991]. The speciﬁc humidity in the experimental ﬂow
was always less than 0.02 kg H20/ kg air. The computational results were

provided by Dr. B. AbdulNour (Ford Motor Company) and were computed with

Fluent 5° with a user deﬁned function (UDF) incorporated into the software. The

UDF utilized the calculated concentration gradient at the wall and Fick's law to

calculate the mass transfer to the surface.

 

Exp = 84.772x + 0.06
1-75 . R2 = 0.9519

1.5 < NH: = 140.71x + 0.0202
1.25 . rk-e> xlw=7.5=101.18x
gm > xlw=12.5 = 71.571):

   
 
  
 

Condensation Rate (glmln)

 

 

 

 

0.000 0.002 0.004 0.006 0.008 0.010 0.012 0.014 0.016
AConcantratlon (kg/m“) = (m~p)o.., - (m‘p)(mm

 

0 Experimental Data — 95% Cl on SLR Model of Data — - - -SLR model of exp. data

 

rk-e>x/w=7.5 ——rk-c>x/w-12.5

 

rk—e

 

Flgure. 5.4. Comparison of the experimental and computed steady-state

condensation results.

These results were published in Hoke et al. [2000]. The slope of the
predicted condensation rate, with the turbulence model active in the entire ﬂow
ﬁeld, is 55% larger than the measurement results. The numerical simulation
assumes that the ﬂow is turbulent throughout the ﬂow ﬁeld. The experimental
measurements show a laminar-like, Blasius velocity proﬁle at the jet exit.

Mabuchi and Kumada [1972] show a 45% increase in the average mass transfer

91

M--...-_L_ _ .._..

coefﬁcient between a wall jet with an laminar to turbulent transition and a wall jet
with an fully turbulent initial boundary layer. This is indicative that the difference
between the experiment and the CF D results may be caused by the turbulence
model utilized in the entire ﬂow ﬁeld simulation.

The numerical model was modiﬁed as recommended in Chapter 2 to
incorporate zones where the turbulence model could be turned on and off to
simulate the effects of the laminar to turbulent transition. The experimental
velocity measurements of the inner boundary layer indicate that the transition
region begins before xlw = 5. This estimate is based on the sharp rise in the
streamwise normal stress in the inner boundary layer that begins to become
pronounced at xlw = 4.45. Two simulations were conducted where the transition
location was set at xlw = 7.5 and 12.5, respectively [Chui, 2000]. The results from
these simulations bracket the experimental results and are shown in Figure 5.4.

The slope of the condensation rate measured experimentally is 84.7
glmin/(kg/m") or 0.0014 mals. Note that dividing this parameter by the thermally
active test surface area (0.045 m2) yields results for the mass transfer coefﬁcient
in more usual dimensions of mls, producing II... = 0.031 m/s. Measurements of
the test surface temperature by AbdulNour et al. [1997] have shown that the
edges of the thermal test surface are signiﬁcantly affected by conduction to the
splitter plate and that only 80% of the thermally active test surface could be
considered isothermal. The center of the thermally active test surface was
reported to be isothermal to within 1 °C [AbdulNour, et al., 1997]. Temperature

measurements of the thermally active test section were conducted with the

infrared camera and are presented in Figure 5.5. The measurements were
conducted with the thermally active test surface heated to avoid complications
with condensation on the surface. The condensation on the surface could create
larger temperature variations on the surface; however, the heat ﬂux associated
with the mass transfer in a typical experiment is less than 20 W. This ﬂux was
estimated to effect the surface temperature by less than 1 °C at any location.

Note that the circulating bath is capable of 500 W of cooling.

93

24°C

 

Figure 5.5. Temperature variation on the thermally active test section for a

constant temperature circulating bath temperature of 34 °C.

Based on the results presented in Figure 5.5, it is expected that the edges
of the thermally active test section are 5 to 6 °C warmer than the center during
the condensation experiments. This temperature variation would reduce the
concentration difference by approximately 60 to 80% compared to the center of

the test section in a typical experiment. The thermal ﬂux from the higher

condensation rates is expected to create larger temperature variations on the
surface. The surface temperature variation results in an effective isothermal area
of approximately 76% of the total test section area. If the experimental
measurements are adjusted for this loss of effective area, the mass transfer
coefﬁcient would be equal to approximately 0.00181 m3/s (0.0407 mls), which is
within 10% of the CF D results with the transition location at xlw = 7.5.

The realizable k-e models utilizing the transition locations of 7.5 and 12.5
produced slopes of 0.00168 m3/s and 0.00119 m3/s respectively, under-
estimating the corrected experimental result of 0.00181 male. This indicates the
importance of properly modeling the transition location to obtain valid results in
the developing ﬂow ﬁeld. A recommendation to incorporate a linear stability
model or other technique to predict the onset of the transition region to improve
the model was made to Ford Motor Company based on these results. The
transition model is especially critical if the computation is to be utilized as a
design tool in similar ﬂow ﬁelds since an experimental prototype may not be
feasible/available. The xlw locations of 7.5 and 12.5 for the transition zones
were chosen based on the experimental velocity measurements and the test
surface geometry. Also note that the transition region in the experiment is
expected to begin before xlw = 5. Recall that transition does not occur at sharply
demarcated boundary

The spatial average of the Sherwood number (hm w/ D) over the thermally
active test surface can be determined from the steady-state condensation

measurements by calculating an average mass transfer coefﬁcient from the slope

95

of the results in Figure 5.4. Local information cannot be obtained from the
steady-state experiments since the data measured are of the mass transfer to
the entire surface. These experimental and computational average results are
compared in Figure 5.6. to averages taken from data published by Mabuchi and
Kumada [1972]. The results from Mabuchi and Kumada are referenced as M&K
in the ﬁgure along with a descriptor indicating the jet Reynolds number (U, w / v)
in the experiment. The average Sherwood number results from the experimental
data were corrected for the effect of the temperature variation on the test surface
as previously explained. Note that the data from Mabuchi and Kumada [1972] is

representative of much of the data in the literature such as Akﬁrat [1966].

 

Average 8h.

 

(Re=13300)
Exp (Corrected
for Area)
CFD
(Re=13300
all turbulent)
CFD
(Re=13300,
xlw =7.5
transition)
M&K(Re=9140),
Untripped)
M&K(R¢.=17800,
Untripped)

i
t

Figure 5.6. Average, steady-state Sherwood number comparison for the

experimental, computational, and published results.

Utilizing linear interpolation to obtain an estimate for ReW = 13,300 based
on the untripped results published by Mabuchi and Kumada [1972] provides an
estimate of Shw= 41. The area-corrected steady state condensation results (ShW
= 38.2 t 3.4) presented in Figure 5.6 agree with the expected results
extrapolated from the data published by Mabuchi and Kumada [1972] to within
the 95% conﬁdence region. This indicates that the ﬂow may be undergoing
transition similar to that indicated by the published results.

The variation in the simulation results indicates the sensitivity of the CF D
mass transfer calculations to the location of the inner boundary layer transition to
turbulent ﬂow. The "error bars" on the CFD result reﬂect the difference between
the predications obtained with the RSM and the k-a turbulence models. The
uncertainty indicated by the "error bars" on the other results are the
experimentally determined variation of the speciﬁed uncertainty of the
experimental technique utilized. Direct comparisons of the data show that the
CFD model with a zonal transition at xlw = 7.5 agrees to within 10% with the
experimental results.

Comparison of the velocity and temperature gradients at the thermally
active test surface as discussed in Chapters 3, 4, and 5 could also explain the
discrepancy between the CFD model and the experimental results. Tables 5.1
and 5.2 contain the comparison between the velocity and temperature gradients

measured and calculated at the thermally active wall surface.

97

at the thermally active surface.

Table 5.1. Comparison of experimental and computed velocity gradients

 

 

 

 

 

 

 

Experimental 2:3! :ZZUSS k- % Variation

Ensemble results 8 CFD from Exp.
model

xlw 31%?” r2 xlw 31%?” 211%?“ xlw % Variation

0 11132 0.84 0 22193 22832 0 105.1

1.59 10595 0.90 1.59 22270 21315 1.59 101.2

3.18 12268 0.94 3.18 21827 20213 3.18 64.8

4.45 9390 0.86 4.45 21420 19576 4.45 108.5

7.68 12179 0.87 7.68 20490 18472 7.68 51.7

10.8 14534 0.81 10.8 20134 18208 10.8 25.3

 

 

 

 

 

 

 

 

 

 

 

 

 

gradients at the thermally active surface.

Table 5.2. Comparison of experimental and computed temperature

 

 

 

 

 

 

 

 

 

 

Experimental 3;; again: k-a % Variation
Ensemble results model CF D from Exp.

”W 303%; r2 xlw ‘5”? Egg; xlw % Variation
1.59 16123 0.96 1.59 42671 41303 1.59 156.2

4.45 22450 0.98 4.45 36226 30150 4.45 34.3

7.68 20000 0.92 7.68 33067 27739 7.68 38.7

10.8 16147 0.98 10.8 31224 26451 10.8 63.8

 

 

 

 

 

 

 

 

gradients at the wall surface. The % variation in Tables 5.1 and 5.2 is deﬁned as

The CFD model consistently results in larger velocity and temperature

the 100 times the difference between the CF D and the experimental results

divided by experimental result.

 

 

The over-prediction in the wall gradients by the CF D model is probably
also responsible for the over-estimation of the steady state mass transfer rate.
The transition region should begin before xlw = 7.62 based on the experimentally
measured increase in the streamwise normal stress proﬁle in the inner boundary
layer. Correcting the wall gradient prediction should lower the overall mass
transfer calculation result and allow the zonal turbulence model transition location

to better reﬂect the experimentally determined position.

Chapter 6

OPTICAL MEASUREMENT OF TRANSIENT MASS TRANSFER UTILIZING
INFRARED THERMOGRAPHY

A non-contacting, optical measurement technique was sought that could
be utilized to obtain transient mass transfer measurements on a surface in
convective ﬂow applications. A non-contacting, optical technique is highly
desirable since no probes or sensors are added to the ﬂow or surface under
interrogation. Presence of a probe or sensor can create experimental artifacts
effecting the results. Additionally, infrared wavelengths were utilized to obtain a
high degree of sensitivity due to the spectral absorptivity characteristics of liquid
water. There is evidence in the literature that the presence of liquid water and
other liquids are detectable with infrared sensors [Bearman et al. 1998, Pekemir
and Davies, 1999].

Utilizing therrnographic reﬂectance measurements to measure local
condensate thickness is based on the absorptivity of the condensate. Figure 6.1
presents a model schematic of a uniform thin ﬁlm of water on a reﬂective surface.
This ﬁgure greatly simpliﬁes the actual experimental conditions but it does
demonstrate a theoretical basis for the technique. The condensate surface
Observed in the experiment is more complex than the thin ﬁlm assumed in the
model and appeared to consist of a ﬁlm mixed with randomly distributed droplet-
like structures. For long enough times, these droplets eventually coalesced into

larger droplets in the tests in the FMTF and ﬂowed off the thermally active test

section.

100

Condensate ﬁlm
of thickness T \

   
   

ptGoe-GT
Gears! =ptGoe-2aT
Reﬂected radiatlon Reﬂective target
Incident irradiation / T
G0 Goo“

 

 

Figure 6.1. Thin ﬁlm absorption model

In this model, the incident irradiation (6,) would be absorbed by the
condensate ﬁlm over two thicknesses as it was reﬂected from the target surface.
The quantity of energy absorbed is an exponential function and is dependent on
the absorptivity (a) and the thickness (T) of the condensate. Note that the angles
of incidence in Figure 6.1 are exaggerated for illustrative purposes. The actual
experiment utilizes illumination and viewing angles nearly perpendicular to the
target surface. A simple analysis of the physical situation presented in Figure
6.1 would result in the reﬂectance of the sample with a thin condensate ﬁlm being

calculated as

 

(6.1)

101

In principle, Equation (6.1) could be solved for the condensate thickness
(T). A complication arises from the fact that the intensity of the reﬂected energy
cannot be measured directly. The infrared camera is only capable of measuring
the radiosity (J) from the target surface, which includes emitted and reﬂected
energy (the targets utilized are opaque). However, the reﬂectance can be
obtained by measuring the total radiation from the surface when the target is
independently illuminated by two sources, A and B, with different intensities
[lnframetrics, 1988]. The emitted energy from sources A and 8 must also be
measured directly. Assuming that the target surface temperature is the same in
both measurements, then (e.oT4 )A = (e.eT‘)B Subtracting the two measurements
eliminates the emitted radiation from the image data. The target reﬂectance can

then be calculated by

JM T JsB _ PsGA + (55074» T psGB + (33074»
J - J

MA 018 pmGA + (£m0TM4 )A - pmGB + (EmO'Tm4 )B _
= psGA -psGB = psGA = psGB .
pmGA T Pm GB F». GA Pm GB

S

 

(6.2)

Note that the reﬂectivity of the mirror (pm) was determined to be ~0.97. If
the reﬂectance of the aluminum target and the intensity of the irradiation (G) from
the two illumination sources are assumed to be constant, the thickness can be

calculated from the following ratio

102

 

 

 

 

( pste—ZaT \
£1 _ GI _ _
ln[pso J — ln Egg - ZaT. (6.3)
G
K 0 J

The sensitivity of the optical reﬂectance method was explored based on
this thin ﬁlm model. The dynamic range of the technique is expected to be
dependent on the wavelength of infrared radiation utilized and the sensitivity and
resolution of the optical system used. The IR camera utilized, an Inframetrics
Inc. model 600L, has an 8 bit detector sensitive to wavelengths between 8 to 12
pm. The average absorptivity of liquid water in this spectrum range is
approximately 1300 cm'1 [Modest 1993]. The 8 bit range limits the useful
sensitivity in the low reﬂectance range and deﬁnes the measurement resolution
obtained with the current equipment. It was expected that the reﬂectivity could
be accurately determined to approximately 0.1 before the signal to noise ratio
became untenable. Equation (6.1) was utilized to determine the response of the
IR system to condensate accumulation on aluminum targets to estimate the
maximum thickness measurable with the experimental equipment available. The
results of this analysis are presented in Figure 6.2. The typical aluminum sample

reﬂectance of 0.7 was utilized in the calculation.

103

 

 

 

 

 

 

 

 

 

 

0.9 .

'8 0.8 - —— % of Illumination reﬂected
g 0.7 - — - --- Minimum reﬂectance

l:

2

C

0

3

(I

.E

E

2

.\°

0 . r
0 0.005 0.01 0.015

Thickness (mm)

Flgure 6.2. Thin ﬁlm model results of the Inframetrics 600L sensitivity.

The results presented in Figure 6.2 indicate that the useful dynamic range
of the optical condensate measurement technique developed with the equipment
available would be limited to a maximum thickness measurement of
approximately 7 to 8 pm. Thus the technique is highly sensitive to initial
condensate development but has a limited dynamic range.

The relation determined experimentally is not as simple as the thin ﬁlm
model indicates. However, this log relation provides a useful linearization of the
data [Hoke at al., 2000]. The experimental results differ signiﬁcantly from the thin

ﬁlm model predictions because the condensate does not develop as a uniform

104

thin ﬁlm but is observed to be a mixture of ﬁlm and droplets as shown in Figure
6.3.

 

Figure 6.3. Structure of the condensate formation on the thermally active test

section.

The droplets seen in Figure 6.3 scatter and absorb light differently than a
thin ﬁlm and the optical path length through the condensate would not be uniform
across a droplet on the target surface. Due to these effects, an empirical relation
was sought between the reﬂectance measurement and the thickness of the
condensate present.

Individually characterizing the droplets is not necessary due to the optical
properties of the infrared system utilized. The infrared camera obtains a digital

image of the surface and each pixel of the image is an integrated average of an

105

area of thesurface observed. The Inframetrics 600L long wave camera utilized
for these experiments integrates over a length scale of 0.67 mm to 1.19 mm for
viewing distances of 336.5 mm and 616 mm respectively [lnframetrics, 1988].
This is approximately one to two orders of magnitude larger than the largest
droplets observed on the test surface. The reﬂectance calibration presented
here takes advantage of this area integration and assumes a stationary droplet

distribution in each optically integrated measurement area (optical pixel).

6.1 EXPERIMENTAL METHODS

The development of the experimental technique required four phases. The
ﬁrst phase developed an independent methodology to measure the trace
amounts of water on the calibration target surfaces. The second phase
consisted of calibrating the IR camera measurement of the surface reﬂectance as
a function of the condensate present on the test samples. The effects of viewing
angle and distance were also tested during the second phase. The third step
involved validation of the technique under conditions where non-uniform
condensate thicknesses were expected on the aluminum calibration targets. The
fourth and ﬁnal step involved quantifying the error present in the method,

examining the limitations and establishing the accuracy of the technique.

106

6.1.1 MEASUREMENT OF TRACE CONDENSATE UTILZING HS-C-GC-FID

The spectral sensitivity of the infrared camera utilized limits the estimated
maximum dynamic range of the optical thickness measurement to approximately
8 pm. This estimate is based on the thin ﬁlm model. This is the equivalent of
having 0.8 pg of condensate per square centimeter of target surface area.
Conventional techniques for water measurement such as gravimetric or titration
techniques were not suitable for a quantity this small. Furthermore, since the
target is cooled below the ambient dew-point, the calibration target must also be
isolated quickly to prevent additional condensation or evaporation. Simply
moving the target to a secondary measurement location without isolating it from
the environment could cause signiﬁcant measurement error. An error estimate in
the condensation measurement based on the target temperature and an ambient
humidity of 50% Rh was calculated from a simple energy balance between the
internal energy of the target and the latent energy of condensation or

evaporation. This ﬁrst order estimate was calculated with

(meAT) = (mL)condensate (6-4)

aluminum

This would result in a 1% error in the mass measurement from additional
condensation. This result indicates that further condensation is not of critical
importance. However, a more signiﬁcant problem is evaporation of the

condensate from the target. Visual observation of the wetted aluminum targets

107

indicates that exposure to dry ambient air can remove a signiﬁcant amount of the
condensate from the target.

An indirect measurement technique for the measurement of trace
quantities of water using head-space, capillary, gas chromatography (HS-C-GC)
was developed by Loconto at al. [1996] based on the work of Loeper and Grob
[1988]. In this method, a sample containing an unknown quantity of water (H20)
is added to a head-space vial. A stoichiometric excess of calcium carbide (C8C2)
is added and the vial is quickly sealed. The calcium carbide reacts with the water
in the vial and produces acetylene (Csz) and calcium hydroxide (Ca(OH)2)
according to Equation (6.5) [Noller, C., 1966].

(:ch2 m+ 2H20 01-) €sz (g,+ Ca(0H)2 m (6.5)

The acetylene gas can be detected with the headspace, capillary, gas
chromatography system utilizing a ﬂame-ionization detector (HS-C-GC—FID) and
measured with a high degree of accuracy. The HS-C-GC—FID system must be
calibrated against samples with known quantities of water in order to be utilized
as an effective measurement method.

All the materials and hardware utilized for the HS-C-GC-F ID calibration
were prepared as if they were to be utilized in the condensation experiment. The
headspace vials were cleaned with de-ionized water and hydrochloric acid
solution and then rinsed with de-ionized water and acetone. The headspace

vials were dried in an oven at 110 °C for 24 hours and then stored in a desiccator

108

for later use. This treatment provided clean vials with a low measured blank level
that was repeatable.

Aluminum test targets (10 mm x 32 mm x 0.8 mm) were polished with a
ﬁber-free pad, cleaned with acetone and allowed to soak in the ambient
atmosphere for at least one hour. To prevent deviation from the actual testing
conditions and methods, a dry aluminum test target was added to each vial to
account for any effect the aluminum may have on the chemical reaction or the
acetylene concentration levels. Each sample (including the controls) included an
aluminum target.

Two different methods for adding trace water were utilized. Initially, a
solution of water dissolved in a polar organic compound was utilized. Solutions of
10,000 parts per million (ppm) of water in methyl tert-butyl ether (MTBE) were
prepared. These solutions were delivered to headspace vials with a glass liquid
handling syringe. Calcium carbide was added and the vials were sealed and
shaken to insure that physical contact occurred between the water and the
calcium carbide. The specimens were then stored for 24 hours at ~10 °C. The
samples were analyzed by sampling the headspace of each 22 mL HS vial using
a 100 pL Pressure-Lok" gas-tight syringe (Precision Sampling, Inc.). The
headspace sample was then injected in to a Autosystem‘ID GC (Perkin-Elmer).
The analog output from the C-GC-FlD was connected to a PC via a 600°
interface (PE-Nelson). The PC used Turbochromo (PE Nelson) software for

chromatographic data acquisition and data reduction. The control settings for the

109

HS-C-GC-FID are presented in Appendix C. The output was utilized to calibrate
the GC—FID system.

The calibration was veriﬁed by adding water directly to the 22 mL
headspace vials with 20 pL Pipetman [Gilson]. It was found that the two
calibrations did not agree within the margin of error expected. Initially, the
accuracy of the direct H20 addition was suspect due the difﬁculty of precisely
delivering aliquots of less than 5 pl of H20 into the vials.

The density of H20 at ambient temperature is ~1mg/pL. The use of
conventional 22 mL HS vials for calibration was abandoned since it is
experimentally difﬁcult to transfer from 0.5 to 3 UL aliquots of H20 precisely using
liquid handling glass syringes. Use of 125 mL HS vials with crimp tops enabled
the volume of H20 to be increased by a factor of 125/22 or 5.68 while
maintaining the concentration of H20 at the level that would be found if the more
conventional 22 mL HS vials were utilized.

A new calibration was then conducted by scaling up the entire process to
reduce the experimental difﬁculties with delivering small volumes of water. The
following procedure was utilized to calibrate the HS-C-GC-FID: 5.68, 11.36 and
17.04 pL of H20 were added to each 125 mL vial with a 20 pl PipetmanQ
(Gilson). Note that these volumes produce C2H2 concentrations equivalent to 1,2
and 3 pL of H20 in a 22 mL HS vial. A stoichiometric excess of calcium carbide
was then added and each vial was immediately sealed and shaken vigorously on
a VWR vortex mixer [model K-550-G]. The specimens were then stored for 24

hours at ~10 °C. Three controls were prepared for each calibration run to

110

establish a baseline. The controls consisted of an aluminum target and calcium

carbide, but no H20.

The samples were analyzed in an identical manner as before. First, the

headspace of each 125 mL HS vial was sampled using a 100 pL Pressure-L0H”
gas-tight syringe and the sample injected into an Autosystem® GC. The analog
output from the GC-FlD was recorded with a PC using the 600® interface. The

PC used Turbochme for chromatographic data acquisition and data reduction.

The peak area measurement was obtained and the control average subtracted
from each run. The control subtraction was performed to eliminate the effect of
variable atmospheric humidity levels between test dates. A typical

chromatographic result is presented in Figure 6.4

 

 

 

 

 

 

 

5.0E+05
A 4.0E+05 —
>
3
‘5 3.0E+05 «
a
a
3
O
9 2.0E+05 —«
“.-
o
0 1.0E+05 .

0054-00 '——'I f: . . . . . .

o 50 100 150 200 250 300 350 400

Tlme (s)

Flgure 6.4. Typical chromatogram obtained during HS-C-GC-FID calibration.

111

The results of the scaled up calibration agreed with the direct H20 method
in the 22 mL vials within the margin of error but the results did not agree with
those obtained from the technique utilizing the H20 / MTBE solution. The highest
conﬁdence is placed in the scaled up calibration method since it introduces the
fewest unknowns. The potential causes for error in the H20 I MTBE solution
calibration method include the solubility of the acetylene in the MTBE and
incomplete reaction of the CaC2 with the H20 when in the MTBE solution.
Research into the exact nature of the difference between the calibration
techniques was not conducted. The actual IR- H20 calibration experiments
contain only H20 and since the effect of the MTBE in calibration samples could
not be easily quantiﬁed, the H20 / MTBE methodology was discarded.

The success of the scaled up calibration technique is based on the
measurement characteristics of the ﬂame-ionization detector. The F ID is
sensitive to the total number of carbon atoms injected into the detector, thus it is
sensitive to mass not concentration [Skoog and Leary, 1992]. Since a volume of
headspace gas is injected, the total mass injected is directly dependent on the
concentration of carbon (C2H2 in this procedure) in the measurement volume.
The concentration of the C2H2 depends on the initial volume of air and the total
mass of water present prior to the conversion reaction. If the headspace vial
volume and the quantity of water are increased by the same ratio, identical
concentrations of C2H2 per unit volume are produced.

The HS-C-GC-FID system utilized in these tests was being used for other

experiments while these tests were conducted. The instrument is sensitive to

112

numerous settings including the hydrogen and air gas pressure settings. Due to
the imprecise control of several of the system inputs, new calibrations were
conducted with every experiment to compensate for drift of the HS-C-GC-FID
system. A typical calibration of the HS-C-GC-FID system is presented in Figure

 

 

 

 

 

 

 

 

6.5.
3.5
y = 2.639E-06x + 2.916E-02
3 ‘ R2 = 999215-01 ,,.»-:.:;;:f:?1*
2.5 if,
9 2 a"? I;
:I:
E’ 1.5 -
1 y,
x Data (n=11)
0.5 ~ —— 95% Conﬁdence interval
¢//’ - > ._ - -- Linear regression model
0 it . , . . .
0.0E+00 2.0E+05 4.0E+05 6.0E+05 8.0E+05 1.0E+06 1.2E+06
Peak Area (In V-e)

Flgure 6.5. Typical calibration result for GC-F ID system response to the volume

of water added. (n = 11 calibration samples).

113

The result presented in Figure 6.5 can be utilized to measure the volume
of water present in an unknown sample. This measurement has a 95%
conﬁdence region indicated by the solid red lines in Figure 6.5. The 95%
conﬁdence intervals range from :I: 46 pg at 0 mg, :l: 21 [19 at 2 mg to the
maximum deviation of 1 38 pg at 3 mg. Typically, the range from 0 to 2 mg was
sufﬁcient to encompass the range of condensate mass deposited on the
aluminum targets utilized in the empirical calibration presented here. Note that
the calibration relation is often formatted with the axes reversed, this causes a
negligibly small change in the conﬁdence region and is essentially equivalent to
the presentation here.

This result provides an independent means to determine the mass (and
therefore, volume) of water present on an optical calibration target. An
experiment to calibrate the infrared camera response to local condensate
thickness was designed such that the optical targets could be easily captured in
the 22 mL headspace vials necessary for the gas chromatography technique to
be utilized.

An investigation of other methods to measure the condensate on the
aluminum target surface was conducted. These included Nuclear Magnetic
Resonance (NMR), Magnetic Resonance Imaging (MRI) technology, laser
speckle methods, and shadow moire techniques. The NMR and MRI
technologies present the opportunity for very ﬁne measurements of the water

present, but cannot be utilized on metal surfaces [Gregory, 1997]. It also doesn't

114

allow sample isolation and there is limited access to this type of equipment due
to its cost and operating expense.

Laser measurement methods were also investigated. Laser speckle
techniques can provide very accurate in-plane displacement measurements but
are not very sensitive to out of plane motion [Cloud, 1998]. Independent
validation (calibration) would also be necessary if this method were utilized to
calculate condensate volumes on the surface. Local measurements of ﬂuid
thickness during condensation are also possible with laser interferometry.
However, the limitations of this measurement are that It only provides information
along the fringe lines and has a minimum thickness capability of approximately
100 pm [Schlesinger et al., 1986]. Thus neither of these techniques are
applicable for the range of the experiments conducted here.

The shadow moire technique was also investigated. This technique
projects a moire or shadow grid pattern on the surface of interest. Changes in
the geometry of the surface can then be measured by monitoring changes in the
pattern [Cloud 1998]. The resolution necessary to obtain reasonable
measurements of the condensate pattern is not feasible with visible light thus
requiring expensive optics and detectors. Data reduction with this technique
would also be very complicated and would require external validation/calibration.
Based on these ﬁndings, the HS—C—GC—F ID technique was adopted for trace

water measurement to calibrate the infrared reﬂectance response.

115

6.1.2 CALIBRATION OF THE SURFACE REFLECTIVITYAS A FUNCTION OF
CONDENSATE THICKNESS

An experiment was developed to quantify the change in the reﬂectivity of
the aluminum test surface with the volume of condensate present. The intent
was to develop an empirical relationship based on the local change in the
reﬂectance that could be utilized to provide an optically based measurement of
the condensate thickness as a function of position and time. The result of this
measurement could then be utilized to calculate the local mass transfer rate or, if
the free stream concentration could be measured or was known and
thermodynamic equilibrium assumed at the surface, a local mass transfer
coefﬁcient. Since this technique would not rely on mechanical or chemical
alterations of the surface and could be applied in any situation where optical
access is available, artifact-free measurements of transient mass transfer
coefﬁcients would be possible.

Aluminum was chosen as the test surface due to its relatively high
reﬂectance (typically 0.60 to 0.70 as prepared), high thermal conductivity, low
thermal capacitance, and wide applicability in heat transfer devices. The high
reﬂectivity simpliﬁed the measurements since more of the dynamic range of the
Inframetrics 600L infrared camera could be utilized. The high thermal
conductivity and low thermal capacitance simpliﬁed the experimental design by
reducing the cooling power necessary and helped to insure an acceptably
isothermal test surface. Small samples were manufactured from aluminum sheet

(10 mm x 31 mm x 0.8 mm). This size was selected so that it would easily ﬁt in

116

the headspace vials utilized with the HS-C-GC-FID technique to determine the
total amount of water present.

The experimental conﬁguration utilized in the calibration experiments is
shown schematically in Figure 6.6 and photographically in Figure 6.7. A detailed

list of the equipment and settings utilized is presented in Appendix 8.

Mechanical Clamp
Peltier Device

Infrared Camera Chilled N2 Flow

Heat Exchanger

Infrared Source ”XX'ﬁlJ;
(Heater) 7

Shutter

 

Surface details shown
in Figure 2.1

Figure 6.6. Schematic of the lR-H20 calibration experiment.

117

 

Figure 6.7. Photograph of the lR-H20 calibration experiment.

The initial calibration experiment was performed with the IR camera at a
distance from the calibration samples that could be replicated in the FMTF
facility. The test was designed in this fashion so that if further testing indicated
distance was an important factor, extensive recalibration would not be necessary
prior to use in the FMTF. The Inframetrics camera was positioned with the
bayonet mount 616 mm (24.25") from the calibration target and viewed the target

at an angle of 8° from perpendicular. The viewing angle was necessary to

118

illuminate the sample with the infrared source (heater). The aluminum calibration
target subtended an angle of 102° of the total 20° ﬁeld of view of the IR camera.

The aluminum sample was cleaned with acetone and allowed to dry in the
ambient environment. The heater providing the infrared illumination was allowed
to stabilize so that the reﬂection from the reference mirror was approximately 49
°C and the shutter was activated. The sample holder was heated and dried with a
hot air gun. The aluminum target was then placed in the mechanical clamp and
pressed down so that the back of the target was in ﬁrm contact with the sample
holder. The Inframetrics camera signal was then recorded and the chilled
nitrogen ﬂow was initiated.

Current (0.6 amps) was supplied to the Peltier device once the nitrogen
ﬂow had chilled the target holder to approximately 15 °C (monitored with a
surface mounted thermocouple). This reduced the target temperature below the
ambient dew-point within 180 seconds. The current to the Peltier device was
adjusted if necessary to maintain the target above the freezing point. The
temperature of the target was inferred from a thermocouple mounted next to the
target on the aluminum clamp holding the target. The clamp and target were of
identical thickness and were in good thermal contact with one another.

The target was then allowed to soak in the ambient environment for
various lengths of time to allow condensate to build up on the front surface of the
target. The target was then dropped into a 22 mL headspace vial, calcium
carbide was added and the vial was sealed for later analysis with the HS-C-GC-

F ID technique outlined in section 6.1.1. The total mass of water present was

119

divided by the surface area of the sample exposed to the atmosphere to obtain
an estimate of the average condensate thickness. The condensate was
assumed to be uniformly dispersed on the exposed aluminum surface based on
the uniformity of the changes in the surface reﬂectance. The coefﬁcient of
variation of the surface reﬂectance was typically 3.8% and always less than 5%.
The average thickness was utilized since the IR camera measurement integrates
over each measurement pixel. The initial and ﬁnal reﬂectance measurements of
the targets were obtained from the recorded video data. The results of this

process are presented in Figure 6.8.

 

 

 

 

 

 

 

 

5
4 5 J .
A y = -2.9316x
4 ' xxx.
3 “ 8 o R2 = 0.9399
3 5 T Q \¥‘\.
g \7*\\ ‘
.1: 3 ~ 90%;?
.2 ."“;~Ll;f:\ .0
.c 25 r 0‘ 5:73.-
.— \\’\
2 . ° *4 Sex
15 - . Data .
—~—»-—~ 95% conﬁdence Interval gig—xx .
—i e . . \T: \‘x
.3 1 —— —— Linear regressron model :\\
o 5 ~ “““‘§;?ii':‘\-
\‘* 9
0 T T l r T T \
-1.6 -1.4 -1.2 -1.0 -0.8 —O.6 -0.4 -0.2 0.0
Ln(pr'pr)

Figure 6.8. Ensemble correlation of reﬂectance ratio to estimated thickness

of condensate present.

120

The data are represented by the solid diamonds in Figure 6.8. A linear
regression model was ﬁtted to the data. The model was forced to have a zero
intercept since the physics of the problem indicates that a dry sample should not
change reﬂectance without modiﬁcation of the surface. This causes a very slight
change in the model. If an intercept is incorporated into the model, the ﬁtted line
is y = -2.893x + 0.0329. However, the null hypotheses of zero intercept is
retained (t-test p-value = 0.81) so the ﬁtted line given in Figure 6.8 was used for
the calibration. The 95% conﬁdence interval of the mean regression result
presented in Figure 6.8 is less than 1 0.17pm. The scatter in the data about the
regression model is larger and could be interpreted as a truer estimate of the
variability of the technique since the calibration targets only approximate the test
surface. The standard deviation of the data about the log-linear regression line is
o = 0.30 pm. The usefulness of this result is dependant on the calibration
surfaces being representative of the test surface. The type of surface material
and ﬁnish are critical to maintaining this assumption.

The variability of the data in Figure 6.8 was investigated. It was
discovered that the residuals were dependent on the target utilized. Plotting the
data for each target shows a higher degree of linearity in the results for each
target. Figure 6.9 presents the data from Figure 6.8 for each target and

demonstrates the effect of the target on the result obtained.

121

 

 

 

, T(5,=-3.1509Ln(pm) T3 =_3.1546Ln(pm)
4-5 x R2 = 0 9951 ( ’
- R2 = 0.9832
4 n A. o
X‘ ~ \
i 3.5 2 .‘x.\ Tm = -2.9464Ln(pr/n)
g 3 . 9. :25..- R2 = 0.8914
.e\ . o x...\\
g 2.5 ~ T(1)=-2.535Ln(p,/pi)s .
g 2 - R2 = 0.9928 tj‘iisa
1-5 ‘ Tami“: \
1 2 oSampleZ ° “\
us ple3 _ “\
0.5 _ ”:20.“ TM) - -3.0888Ln(n,/p.)
IxSampleSI R2 = 0.9926 \' .
o . . * T ‘

 

 

-1.6 -1.4 -1.2 -1.0 08 -0.6 -o.4 -o.2 0.0
Lnlprlpr)
Figure 6.9. Correlation of reflectance ratio to estimated thickness of condensate

present for each aluminum target utilized.

The aluminum targets were handled in an identical fashion for each
calibration experiments. Targets #1, #4, and #5 were polished with an abrasive
pad to create a homogeneously roughened surface. The surfaces of targets #2
and #3 were left with the original cold rolled ﬁnish. The differences between
samples could be the result of minute surface variations. The high degree of
linearity for each test result (with the possible exception of target #2) indicates
that the simple reﬂectance model does provide an appropriate non-dimensional

form for the data. This result indicates that the best condensate thickness

122

measurement can be obtained by utilizing a calibrated target in the ﬂow ﬁeld of
interest. A composite calibration of representative surfaces can be utilized but
the uncertainty in the measurements would increase. This can be observed
directly by comparing the squared correlation coefﬁcients (R2) in Figures 6.8 and
6.9 as well as the obviously improved linearity of the results when separated by

individual target specimens.

6.1.3 EFFECT OF DISTANCE AND ANGLE ON THE IR-H20 OPTICAL CALIBRATION.

The distance between the IR camera and the surface of interest is not
expected to effect the correlation results. The slit response function (SRF) of the
Inframetrics 600L infrared camera utilized in these experiments was empirically
determined previously [Hoke, 1998]. At the distance utilized for all the
experiments presented here, the pixel size is 6.1 mm and the 95% SRF is 7.0
mm. The camera is incapable of resolving objects of the scale of the individual
droplets on the aluminum targets. The infrared detector effectively integrates a
local reﬂectance in each pixel, thus averaging the effect of multiple droplets.

The angle of incidence between the infrared camera, the surface and the
infrared source is not anticipated to signiﬁcantly affect the results over angles
less than the Brewster angle of the condensate. The optical path length through
the droplets/ﬁlm is a weak function of angle at small angles and thus small
variation in the viewing and illumination angle should not effect the calibration
signiﬁcantly. However, empirical calibration of angles much greater than were

utilized here (8°) is recommended prior to application.

123

6.2. VALIDA‘HON OF THE OPTICAL MEASUREMENT CALIBRATION WITH A NON-
HOMOGENEOUS CONDENSATE FIELD

A method to validate the optical measurement was sought as a means to
conﬁrm the accuracy of the experimental correlation developed. Two concepts
were explored before a methodology was decided upon. The ﬁrst method was to
create a "sandwich" of liquid water between two materials with a gap width of
known thickness. The second method explored was to create a non-
homogeneous condensate ﬁeld on the aluminum targets. The volume
measurement obtained from the HS-C-GC-FID method could then be compared
with a pixel by pixel, surface inteng of the condensate volume obtained from the
infrared, surface reﬂectance measurements.

The sandwich method required surfaces with an acceptable ﬂatness and
smoothness tolerance. The surfaces would have to be smooth to within 0.25 pm
and would have to maintain a ﬂatness speciﬁcation across the target surface of
similar dimensions in order to provide a useful measurement. Variation in the
surface larger than 0.25 pm would create larger uncertainty than that obtained
with the HS-C-GC-F ID calibration and would limit the usefulness of the validation
technique. 0ne material would also have to be transparent to infrared radiation
and the other material would have to be highly reﬂective to the same
wavelengths. Aluminum and germanium were selected as possible materials
based on these criteria. Four signiﬁcant problems arose as this method was

further investigated. These complications were:

124

1. Aluminum target surface would not match current target surface
characteristics. The current calibrated aluminum targets would not be
useable for this technique.

2. A germanium lens with appropriate speciﬁcations would cost approximately
$2000 for a 25mm diameter specimen.

3. Condensate In the gap would not be consistent with the structured
condensate observed in the calibration or F MTF experiments and could
therefore create different results. This could limit the usefulness of any
comparisons made.

4. Selecting an appropriate gap material to space the aluminum and germanium
materials.

Due to these difﬁculties, the non-homogeneous method was then
explored. This method requires that a non-uniform condensate ﬁlm develop on
the aluminum targets. Altering the aluminum target surface temperature, altering
the concentration gradient over the target surface, or altering the mass transfer
coefﬁcient could create the desired non-uniform condensate ﬁeld. The total
condensate volume on the target could then be measured with the HS-C-GC-FID
technique. The change in the surface reﬂectance of the aluminum target would
then be evaluated pixel by pixel and a thickness calculated for each pixel with the
correlations presented in Figures 6.8 and 6.9. Since the pixel size is known, a
total volume of condensate on the surface can be calculated. The resulting
volume from the infrared technique can then be compared directly to the HS-C-
GC-F ID result.

125

The non-uniform method provided a much more direct and simple means
of validation and allowed direct comparison to the existing data set with the
fewest number of possible complications and was therefore selected. The
weakness of the non-uniform method was that it also utilizes the gas
chromatographic technique so an error hidden in this technique would also affect
the non-uniform tests. However, since the condensate ﬁlm would not be uniform
but would qualitatively have the same structure, it was decided that this method
was superior to creating an artiﬁcially uniform liquid ﬁlm with the known gap
"sandwich" technique.

Initially, a non-uniform, ﬁn like temperature distribution was chosen for
creating the non-homogeneous condensate ﬁeld. A model was created to
estimate the temperature distribution that might be obtained with varying sample
thicknesses. An estimate of 5 W m‘2 K’1 for the convective heat transfer
coefﬁcient was made assuming a zero velocity far ﬁeld with a temperature of 20
°C. The model was based on a one-dimensional ﬁn analysis since only the lower
portion of the target would be cooled and the back of the aluminum target would
be placed against an insulating Teﬂon surface to insure that condensate only
developed on the side of the target facing the infrared camera. The model
predications for the temperature distribution are presented in Figure 6.11. The
axes in Figure 6.11 represent the location on the target surface (x (m)), the target

material thickness (m), and the vertical axis is the target temperature (K).

126

     

,, ’9 thickness (m)
/, 0.0006

 

0 0.0008

Figure 6.11. Temperature distribution in aluminum target derived

from a one-dimensional ﬁn analysis.

Since the inﬂuence of the speciﬁc targets was previously identiﬁed as
discussed in section 6.1.2, the target thickness was ﬁxed at 0.8 mm so that the
calibrated targets could continue to be utilized. This resulted in a temperature

proﬁle on the aluminum target as shown in Figure 6.12.

127

280 I

Q I
7: I
279 I

I

278I

I

_,a k "A _ A i ._. _.4 A ,L_ _ —A.—___4* v a. ___—__J.

0.005 ‘— 001 0.018 7 0.02

 

X (m)
Figure 6.12. Temperature distribution in 0.8 mm thick, aluminum target derived

from a one-dimensional ﬁn analysis.

The temperature distribution presented in Figure 6.12 results in a spatial
variation in the concentration gradient driving the mass transfer to the target
surface. Equation (6.6) was utilized to covert the temperature gradient to a

concentration gradient along the target [Cengel and Boles, 1989].

 

 

I.( 1 _ 1 )
R T T(x)
O.622¢Prefe w ref
AC=CJ°- (6.6)
Rairnx)

128

Utilizing (6.6) results in a spatial distribution of the concentration difference

along the length of the 0.8 mm thick, aluminum target as shown in Figure 6.13.

 

 

 

Concentrationdifference(kglm’)
0.004 I.
0.0035I
I
_..L 7 ..L_ i _._ A._ x m
I 0005 x 0.01 0.015 002 ( )
I
0.0025 I

Flgure 6.13. Concentration gradient distribution along the aluminum target

resulting from the ﬁn temperature distribution.

This temperature gradient results in an approximate 40% variation of the
concentration gradient along the surface. Initial experimental data did not agree
with the model results. Only a 5-15% variation in the target surface temperature
was observed. This was attributed to the fact that a quiescent far ﬁeld did not
exist and that the Teﬂon insulation cooled much quicker than the model
indicated. A more pronounced difference in the condensate pattern was desired

so a means was sought to amplify the variation.

129

Dry nitrogen gas is currently used to cool the heat exchanger attached to
the Peltier device (see Figure 6.6). The nitrogen ﬂow exits the heat exchanger
system with an approximate temperature of 10 °C. Experiments revealed that
diverting this ﬂow to the surface for an instant could dry 8 portion of the aluminum
target and since the temperature was above the freezing point, there was no risk
of causing icing of the surface. This method was adopted to enhance the

variation in the condensation ﬁeld.

6. 2. 1. NON-HOMOGENEOUS EXPERIMENTAL PROCEDURE AND RESULTS.

The calibration procedure outline in section 6.1.2 was utilized with one
additional step, dry nitrogen gas was utilized to evaporate a region of
condensation on the target prior to capture in the 22 mL headspace vials. This
produced a non-homogeneous condensate ﬁeld. Examples of two typical
therrnograms before and after a region was exposed to the nitrogen ﬂow are
presented in Figure 6.14.

130

. luminum
arget

- lum inu m
arget

   

Before application of N2 jet After application of N2 jet
Figure 6.14. Non-homogeneous concentration ﬁeld on an aluminum target

before and after a portion was dried with nitrogen gas.

The infrared image was then evaluated pixel by pixel with the correlation
previously obtained for the uniform condensation, lR-H20 calibration result
(Figure 6.9). The calculated thickness was multiplied by the area per pixel to
obtain a volume of condensate present. The thickness variation on a target in a
typical experiment is presented in Figure 6.15. The coefﬁcient of variation (COV)
of the image data in Figure 6.15 is 27.8%. The average COV observed during the

non-uniform tests was 27.1% and ranged from 3.7% to 47.7%.

3.0

Y (cm)

3
8
8
7‘3
2:
l-

A

   

0 1.1
x (cm)

Flgure 6.15. Measured thickness results on aluminum target #4 during a non-

homogeneous condensation ﬁeld test.

The volume obtained with the pixel by pixel analysis was compared to the
result obtained from the HS-C-GC-FID technique. A perfect result would present
a 1:1 slope when the IR and the HS-C-GC-FID results were plotted against one

another. The results of the comparison are presented in Figure 6.16.

 

 

1.2
I x Volume (IR) 50

 

   

 

1 1:1 slope - 2 _. 45
1 -1TTTT95% Conﬁdence interval x
i - COV ,. -2 40
I' , ~Regression model ,
-- 35
0.8 . , «
“X

" 30

Area corrected volume (pl)
0
O)
M
(II
Coefﬁcient of variation of IR image (%)

 

 

 

-- 20
0.4 -
“ 15
0.2 « 7 1°
_ T 5
0 . , r . - o
0 0.2 0.4 0.6 0.8 1 1.2

60 measured volume (pl)

Figure 6.16. Results of the non-homogeneous concentration ﬁeld tests utilizing
the HS-C-GC-F ID technique as the reference standard.

Coefﬁcient of variation in each non-uniform image presented.

Data from 18 tests are presented in Figure 6.16. The result in Figure 6.16
agrees with the expected slope of 1.0 within the statistical uncertainty of the data.
This indicates that the method is self-consistent and capable of resolving the total

mass transfer to a surface with a non-homogeneous condensate ﬁeld.

The correlations derived for each aluminum target were utilized to covert
the change in surface reﬂectance to a local thickness (Figure 6.9). The surface
area contained in each pixel was then determined from the optical arrangement.
Multiplying the pixel area by the calculated condensate thickness and summing
across the surface results in a total volume estimate derived from the infrared
reﬂectance measurement. The entire aluminum target was not analyzed
because the edge effects where the target is held create non-physical artifacts in
the infrared image. Approximately 88% of the surface was interrogated optically
and the assumption was made that the remainder of the surface contained an
identical condensation pattern. Thus each optically integrated IR result was
multiplied by a factor of 1.136 to correct for the area. This assumption could

account for some of the scatter observed in Figure 6.16.

134

Chapter 7

OPTICAL, TRANSIENT MASS TRANSFER MEASUREMENTS
IN THE DEVELOPING WALL JET

7.1 INTRODUCTION

The ﬁnal phase of the project was to measure the mass transfer to the
thermally active test surface in the Ford-MSU test facility with the new optical
infrared reﬂectance technique. This represents the culmination of the project to
collect a benchmark data set for the purpose of calibrating the CF D model.
These measurements provide condensate thickness data as a function of time
and location. Surface temperature measurements combined with temperature
and humidity measurements in the wall jet provide the necessary data to
determine the concentration difference in the ﬂow ﬁeld. This combination of data
allows the mass transfer coefﬁcient and the Sherwood number to be calculated
as a function of time and location. The results of the transient data measured
with the optical technique will be compared to the steady state results presented
in Chapter 5 and with steady state results in the published literature. These
results indicate that the optical infrared reﬂectance technique is functioning as

expected and provide insight into application of the technique.

7.2 EXPERIMENTAL CONFIGURATION
The infrared camera was positioned to view the thermally active test
section from a distance of 616 mm at an angle of 8°. This distance and angle are

identical to those utilized in the calibration experiments presented in Chapter 6.

135

A hole was cut in the Lexan° window in the FMTF in order to provide optical
access for the infrared camera. A 3" x 12" heater thermally bonded to a
blackened aluminum plate was utilized as a source of infrared illumination.
Electrical current sufﬁcient to heat the aluminum plate to approximately 49 °C
was supplied. Figure 7.1 presents a photograph of the experimental
conﬁguration in the FMTF. Note that the infrared source (heater) and shutter are

hidden on the far side of the aamera.

Inframetrics
6001. IR camera

Thermally
active test
surface

 

Figure 7.1. Experimental setup in the Ford-MSU test facility.

The constant temperature bath was set to the desired temperature
(typically, 4 °C) and allowed to stabilize with the ﬂow to the thermally active test

section turned off. The thermocouple and video data acquisition systems were

activated and the initial reﬂectance of the surface was measured. The ﬂow to the
thermally active test section was initiated and the mechanical shutter was
activated to capture reﬂectance measurements of the thermally active test
section as the surface temperature decreased below the wall jet dew-point
temperature. Figure 7.2 contains the thermocouple temperature record of the

thermally active test section.

 

 

 
 
 
 
 
 

 

 

 

 

 

 

 

30 o xlw=0.5
n xlw=1.5
25 ﬂ x’w=2.5
r X X/W=3.8
x xlw=5
8 20 1 o xlw=9
°~ + xIw=12
a
g Tjet
l- 15 4 ..... — - Tdewpoint
10 . """""""""
5 . , . .
0 20 40 60 80 100

Tlme (s)

Figure 7.2. Surface temperature history of the thermally active test section

during a typical transient condensation experiment.

137

The thermocouples are mounted on the back of the thermally active test
section at the streamwise locations indicated in Figure 7.2. The thermocouples
are mounted off the centerline of the test section at zlw = -1.0. This location
corresponds to the measurement location interrogated by the optical infrared

reﬂectance technique on the opposite surface.

7.3 TRANSIENT OPTICAL DATA

Infrared images of the surface were recorded with a video cassette
recorder for the duration of the experiment for post processing. The images were
time marked during recording with a video timer. The raw video images were
captured as digital images with Image Pro Plus 4.1 (Media Cybernetics) and
intensity data were extracted from the images for further analysis. Two images
are utilized to calculate the apparent surface reﬂectance utilizing the same
techniques detailed in Chapter 6. An example of the raw 8 bit images and the
resulting surface reﬂectance obtained from the image data is presented in Figure

7.3.

138

Pseudo- colored gray scale data Reflectance data

 

 

 

 

 

 

 

OE w_—:o F ‘ rem—w] . I y“ ~ 04
xlw ~0.3 xlw~0.3 ‘m xlw~0.3 ‘_"
”w ”7'4 . xIw~7 .4 ' xIw~7 .4 ‘
zlw~-1.2to-0.2 z/w~-.12to-0.2 zlw~-1.2t_O_-0.2
Image A Image B Reﬂectance Result

Figure 7.3. Typical infrared images of the thermally active test surface during

transient condensation experiment and the resulting surface reﬂectance.

The two-dimensionality of the results presented in Figure 7.3 is expected
to be due to the temperature variations on the thermally active test surface. This
pattern is not evident in the measurement of the initial surface reﬂectance and
changes throughout the experiment. The standard deviation of the condensate
thickness measurement across the image never exceeds 0.25 pm at any instant
and the maximum difference between the maximum and the minimum is typically
less than 0.6 pm. The initial surface reﬂectance and the surface reﬂectance att

= 693 are presented in Figure 7.4 for comparison.

139

Rmnm data Reﬂectance data

0
.0
cn
o
O
b

xlw ~0.3 xlw ~0.3

 

x/w ~7.4 xlw ~7-4
z/w -1.2 to -0.2 z/w ~-1.2 to -0.2

 

Inltlal reﬂectance (t=0) Reﬂectance at t=69s

Figure 7.4. Reﬂectance measurements at t = 0s and t = 69s of thermally active

test surface.

The surface reﬂectance measurements of the thermally active test section
were converted to condensate thickness measurements with the result presented
in Figure 6.8. The conversion to thickness of the image data was conducted pixel
by pixel. The mapping from optical surface reﬂectance to condensate thickness
provided spatial and temporal data of the mass transfer to the thermally active

test section. Sample measurements of the condensate thickness development

on the test surface are presented in Figure 7.5. Further experimental

measurement results of the transient condensate development are included in

 

 

 
     

 

 

 

 

 

 

Appendix E.
6
, . t=+31
5 - _ _ . t=+42
5 - . M A t=+46.5
A 4 4 35' - Ix t=+52
E ..,. ..
=4 .. 2: t=+59
g 3 _ . t=+63
.5 $4: + t=+69
E - t=+73
'— 2 d J - t=+75
1 - Ave—A , a- a A _ .4-
ﬁl‘“ _ h L: * J
o T :w—v rt;- Aw W‘
0 2 3 4 5 6 7 8
xlw

Figure 7.5. Typical condensate thickness measurements on the thermally active

test surface during transient condensation (T j = 24.1 °C, %Rh] = 51.1).

The data collected can be utilized to calculate the concentration difference

in the ﬂow ﬁeld and mass ﬂux to the surface. thus providing a measure of the

mass transfer coefﬁcient as a function of location and time. The transformation of

141

the image data to the transient condensate thickness measurements and
discussion thereof will be presented in the next section. Temporal variation of the
mass transfer coefﬁcient is not anticipated in the results since the velocity ﬁeld is
steady with only the surface temperature changing. The rate of change in the
surface temperature is less than 0.5 °C/s while the ﬂow is resident on the
thermally active test surface for less than 0.05 seconds. The concentration

difference does develop transiently with the surface temperature.

7.4 RESULTS OF THE TRANSIENT CONDENSATION EXPERIMENTS IN THE FMTF

The condensate thickness and surface temperature measurements were
converted to local mass transfer coefﬁcient data. Initially the condensate
thickness as a ﬁrnction of time was converted to a mass ﬂux utilizing Equation

(7.1):

 

j,.=(T(t2)TT(ti)]pH20 (7.1)

tZ-tl

The concentration difference between the thermally active test surface
and the wall jet ﬂow was calculated using the Clausius-Clapeyron equation
[Cengel and Boles, 1989]. The temperature and humidity of the wall jet and
temperature of the chilled surface were used for the calculation. The assumption
of thermodynamic equilibrium is employed at the thermally active test surface.

The result is:

142

 

 

 

 

ac = (mp).- —(wp). =
I [Eff '_--'.]] [2544232)] I
R T Tj R T Ts
_ Pi%Rhi° Ref _ Pse Ref
—O°622PRef [Efg[ 1 ,_.1_ ]] [Efg[ l __L]] (7.2)
R T T' R T Ts
\ PTot _ PRefe Ref J PTot _ Ref 6 Ref )

 

 

The complete derivation of Equation (7.2) is presented in Appendix D.
Note that the temperature of the thermally active test surface (T .) is a function of
time and location. The temperature proﬁle on the test surface was interpolated
from the thermocouple results presented in Figure 7.2. An example of the
resulting temperature proﬁle on the test surface is presented in Figure 7.6. Note
that the discrete data points in the ﬁgure indicate the surface temperature

measurements and thus the actual thermocouple locations.

143

 

 

 

 

 

 

 

 

14 . T,(x/w) = -0.0568(x/w)3 + 0.7183(xlw)2 - 2.8164(x/w) + 14.373
R2 = 0.9058
A l
cg 13 “ e Thermocomle data
O.
E — Temperatue proﬁle estimate
0 12 4 _
j—
11 -
1o . i T
0 1 2 3 5 6 7

Flgure 7.6. Typical temperature proﬁle at t = 593 on the thermally active test

surface during transient condensation experiment (T, = 24.1 °C, %Rh, = 51 .1).

The temperature proﬁle in Figure 7.6 was then used with Equation (7.2) to
estimate the concentration difference along the thermally active test surface.
The estimated concentration difference at t = 598 for experiment #2 on 15 June,

2001 is presented in Figure 7.7.

144

 

3.00E-03

2.50E-03 -

2.00E-03 ~

1.50E-03 -

AC (kg/m“)

1.00E-03 ~

5.00E-04 .

 

 

0.00E+00 f . i T . . . .
0 1 2 3 4 5 6 7 8

 

Flgure 7.7. Typical concentration difference between the wall jet ﬂow and the

thermally active test surface at t = 59s during transient condensation experiment.

The concentration difference and the mass ﬂux can be used with Fick’s
law, Equation (7.3), to calculate the mass transfer coefﬁcient. Note that the
average concentration difference between the measurement times t2 and t1 was

utilized in the mass transfer coefﬁcient and the Sherwood number calculations.

j"
h =—
m A0 (7.3)

145

The ShenNood number can be applied to non-dimensionalize the data
obtained with the optical infrared reﬂectance technique. The Sherwood number
based on the jet oriﬁce width, with Equations (7.1), (7.2), and (7.3), is equal to

 

 

 

 

Sh, = hrw =
D
T t -T t
[ (2) (.)]prW
_ tZ—tl
I [Es[_1__i]] [Pg[_l__;]] I (7.4)
R T T] R T Ts
pJ%RhJe 1‘" pse R‘f
l-“iI—L—ii H—t—J—I D
R T T. R T Ts
\ PTot T Refe Ref J PTot PRet‘e Ref

 

 

The results of a typical experiment transformed into non-dimensional (Sherwood
number) form are presented in Figure 7.8. The legend indicates the approximate
center time of the data utilized to calculate the Sherwood number and indicates
the total time separation (dt = t2 - t1) in seconds between the condensate
thickness measurements used to calculate the mass ﬂux and the concentration

difference.

146

 

 

  
  
  
  
   

 

 

 

 

 

 

120 e t=51s. dt=125
. e t~553, dt=65

100 ‘ t~555. dt=17s

e t~588. dt=11s

8° 3 xt~84s.d1=10s

=3 +t~68$, dt=1OS

m 60
40 -
20 7
0 r T i
0 1 2 3 4 5 6 7 8

xlw

Figure 7.8. Typical Sherwood number results on the thermally active test surface

during transient condensation experiment (T, = 24.1 °C, %Rh] = 51 .1).

Note that the scatter near the onset of condensation (Wnt‘ 40s) is the
result of the increased uncertainty in the concentration difference. The
uncertainty will be discussed in detail in section 7.6.

The entire length of the thermally active test section was surveyed In two
separate experiments since the ﬁeld of view of the infrared camera was not
sufﬁcient to visualize the entire test surface at the working distance utilized. The
ﬁrst section of the thermally active test surveyed extended from xlw ~ 0.3 to 7.4.
The second ﬁeld of view produced data from xlw ~ 5.1 to 12.1. The compiled

results are presented in Figure 7.9.

147

 

   
  
  
  
   

 

 

 

 

 

a Raw-1m.w20I01TestZ

100 4 x Rew=13300, 08/20/01 Testi

e Raw-13300, 08/15/01 Test2

80. o Revw13300. 08/15/01 Teetz

: a Raw-13300. 08/05/01 TeatZ

f, °° -Rew-13300, 08/05/01 Testl
4o
20

0 .
0 2 4 s s 10 12

Flgure 7.9. Compiled Sherwood number results on the thermally active test

surface (Rew=13300).

The Sherwood number results presented in Figure 7.9 are internally
consistent and of the correct order of magnitude to be consistent with
expectations. Figure 7.10 presents a comparison of the data in Figure 7.10
(collapsed to a single curve ﬁt) to the steady state results published by Mabuchi
and Kumada [1972]. A ﬁfth order curve was ﬁt to the experimental data. The
experimental ShenNood number is not be expected to be a polynomial but a
function of R8,,“5 and/or a Rex”. These relationships are the classical results for
laminar or turbulent ﬂow ﬁelds respectively. However, since the exact form of the

data is unknown, a ﬁfth order ﬁt was deemed acceptable. Note that in Figure

7.10, the tripped results from Mabuchi and Kumada, which conform to a Rex”,

show the same proﬁle as the present results.

 

140

 

e Max (Re(w )=17800, untripped)
— — mK (Reiw)=17800. tr'pped)

A Ma-K (Re(w)=9140, trbped)

0 Max (Re(w)=s14o, mtripped)
-——Fittoexperinemine(w) =13300)

 

 

 

Shw

 

 

 

 

 

 

Flgure 7.10. Comparison of experimental and published
[Mabuchi and Kumada, 1972] Sherwood number results.

The experimental result in F igure 7.10 does not indicate the laminar to
turbulent transition that is evident in the Re] = 17,800 result from Mabuchi and
Kumada [1972]. The transition is region is visible in the Sherwood number
increasing after reaching a local minimum (at approximately xlw = 5). This is not
necessarily unexpected since the developing region of the jet is inﬂuenced
strongly by the particular jet oriﬁce and upstream conditions (xlw < 25) [Mabuchi

and Kumanda, 1972]. The turbulence intensity at the jet exit can also strongly

149

inﬂuence the location and nature of the transition region. The results published
by Federov at. al. [1997] indicate that the average Sherwood number increases
by 19% as the turbulence intensity increases from 0% to 1%. The study by
Federov et. al. was conducted in a developing channel ﬂow but the results
indicate that the transition location is responsible for the dramatic change in the
Nusselt and Shemood numbers. Small changes in the turbulence intensity were
observed to create large changes in the heat transfer in the laminar range by
Kestin at al. [1961] in a ﬂat plate boundary layer. Kestin at. al. attribute this
effect solely to the change in the transition location in the ﬂow ﬁeld. Analogous
effects could be expected in the developing wall jet ﬂow. The turbulence
intensity upstream of the jet oriﬁce in the Ford-MSU test facility is 1.7%. The
proﬁle of the measured ﬂuctuations at the wall jet oriﬁce is detailed in Figure 3.2.
Mabuchi and Kumada [1972] do not specify the turbulence intensity at any
location in their test ﬂow.

The data obtained with the optical infrared reﬂectance technique were
then integrated spatially for direct comparison to the steady state results
presenbd in Chapter 5. The ﬁfth order curve ﬁt of the experimental Sherwood
number results was integrated to estimate the spatially averaged result on the
entire thermally active surface. The ﬁt to the experimental data is presented for

clarity in Figure 7.11.

150

100

 

y = 737115032 + 2.7269E«01x‘ - 3.7143E+00:r3 + 2.2808901)? -
6.2910E+01x + 1.0444E+02

80 ~ ‘3 R2 =8.7875E-01

Shw

 

 

 

 

Figure 7.11. Experimental Sherwood number results ﬁtted with a ﬁfth order

curve.

The ﬁt to the data was integrated with Equation (7.5) to determine the

spatially averaged Sherwood number.

1 L
Sh, =1- (I f (x)dx (7.5)

This results in a spatially averaged Sherwood number of 42.1. This result
and the experimental steady state result (Shw = 38.2 1 3.4) agree to within the

151

overlapping 95% uncertainty regions. This result is signiﬁcant because it
indicates internal consistency for the experimental results obtained in the FMTF
ﬂow ﬁeld. Since the transient results and the steady state results are expected to

agree since the velocity ﬁelds in the two experiments are identical.

7.5 SHERWOOD NUMBER DEPENDENCE ON JET REYNOLDS NUMBER

Experiments were conducted by varying the wall jet velocity to further
investigate the optical infrared reﬂectance technique. The Sherwood number
results were expected to increase with increasing Reynolds number, similar to
the trends apparent in Mabuchi and Kumada [1972]. Figure 7.12 presents the

experimental results as a function of the jet Reynolds number.

152

120

 

     
   
     

  
     
 

 

 

 

A . Rew=13300. 08/11/01 Test T
3 o Rew=13300, 08/15/01 Test2
f - Rew=10200. 08/15/01 Tests
100 - .~ aRew=84<X).06/15l01Test4
’2'. - Rew=6560.06l11101Teet2
I ..-
.¢-~ -
so _ .11 f lncreasrng
1,. '3‘};- Rew
1' w.
I: 60 «I .7." I . ' 0 . o e “an. exhixm.
. - - z§:~.&§g‘% .: ". '
a‘ . v . W
40 J ‘ \s
".‘o d;
20 -
0 1
0 1 2 3 4 5 0 7 8

Flgure 7.12. Experimental Shem/cod number results as a function of the jet

Reynolds number.

The trends measured with the optical infrared reﬂectance technique show
the expected increase in the ShenNood number with increasing Reynolds
number. None of the Reynolds number results show clear evidence of a laminar
to turbulent transition region. The proﬁle measured in the FMTF is consistent
between experiments however, and may be the distinctive signature of the wall

jet geometry, turbulence intensity, and the upstream ﬂow conditioning in the

Ford-MSU test facility.

7.6 UNCERTAINTY ESTIMATES FOR THE OPTICAL INFRARED REFLECTANCE TECHNIQUE
An uncertainty estimate was calculated based on the observed
uncertainties in the optical infrared reﬂectance technique. The technique of
partial derivatives as outlined by Moffat [1988] was utilized. This technique
results in the uncertainty in a ﬁinction being estimated by the sum of squares of
the partial derivatives multiplied by the measured variability (or estimated
variability) in each variable. Equation (7.6) is utilized to estimate the uncertainty

in R which is a function of the variables Xi :

2
5R
5R = 21676570] (7.6)

Applying Equation (7.6) to the Sherwood number result presented in
Equation (7.4) provides an estimate of the uncertainty in the Sherwood number.
The result of this calculation is lengthy and is included in Appendix F. The

variables and the uncertainty associated with each is presented in Table 7.1.

154

Table 7.1. Variables and uncertainties utilized to calculate the uncertainty

 

 

 

 

 

 

 

in the Sherwood number.
Uncertainty (95% confidence level
Varlable 0") typically) (6X1)
Tlt) (thickness) 608)” pm (from regressron model in ﬁgure
t (time) 10.5 s
w (jet width) 10.0005 m
D (diffusivity) 12x10"7 mzls
T‘ (wall 10 25 °C
temperature) '
T] (jet temperature) 10.25 °C

 

 

 

 

Applying Equation (7.6) to the Sherwood number( Equation (7.4)), with
the values speciﬁed in Table 7.1, results in the uncertainty estimate in the
Sherwood number presented in Figure 7.13. This result is not exact for each
experiment conducted since the speciﬁc uncertainty estimates will depend on the
experimental conditions realized. Note that the uncertainty is the greatest when

the test surface temperature is near the calculated dew-point temperature.

155

800
43? r
.E I ~
to I. 5
E 600] I
g I I
3 I
g 400:
0° I
38 l

200 I

_ . _._. ___. ~..—.—

276 278 280 282
1}”

Flgure 7.13. Percent uncertainty in the experimental Sherwood number results
as a function of the temperature of the thermally active wall surface (T.).

Tam-point ~ 283 K.

The uncertainty in the Sherwood number is great enough to reduce the
usefulness of the comparison, especially when the surface temperature is near

the dew-point of the wall jet ﬂow. Table 7.2 presents the contributions to the total

error for typical experimental conditions.

156

Table 7.2. Contribution of the experimental variables to the total

 

 

 

 

 

 

 

 

 

 

 

 

 

uncertainty.

Parameter Value (Sndxf‘ (ﬂ... I 10.0 °C (S,,dx)z @ T... = 8.0 °C
Tw 10.0, 8.0 °C 6023 132

T] 24 °C 5517 152
Rh] 46% 13860 381

w .02 m 25.2 4.2

D 2.48 X10"5 mzls 2.6 0.45
‘P1 2.0 pm 332 53.9
‘1’2 3.9 pm 332 53.9
At 10 s 100.7 16.8
% w/AC 96.9% 83%
% w/A‘i’ 2.5% 13.6%
Shw 142, 82 200.6 25

 

 

 

 

The experimental parameters involved with the determination of the

concentration difference in the ﬂow constitute the majority of the uncertainty in

the Sherwood number calculation. The relative humidity measurement is the

single largest contributor to the uncertainty. The current system utilized in the

facility is a chilled mirror system that cost $2500 with an accuracy of 11.25%.

Clearly, more accurate determination of the relative humidity in the jet and the jet

and test surface temperatures would greatly enhance the Shem/cod number

determination.

The lowest attainable uncertainty with typical experimental parameters

decreases to approximately 1 22% when the test surface temperature decreases

to 2 °C. However, the optical infrared reﬂectance technique sensitivity to

condensate is great enough that the measurement is saturated well before the

surface temperature is reduced that far. The strength of the optical infrared

157

 

reﬂectance technique is detecting and measuring transient, local condensate
thickness on the surface of interest. Improved estimates of the Sherwood
number would depend on developing an independent and more accurate method
to determine the concentration difference in the ﬂow ﬁeld. Such a method would

replace monitoring the thermally active test surface temperature.

158

Chapter 8

SUMMARY

The original motivation for this project was to develop benchmark data for
use as a calibration and validation tool for a computational ﬂuid dynamics model
utilized in automotive defrosterldemister design. The developing region of a two-
dimensional wall jet was selected as a model ﬂow ﬁeld and a facility was
constructed.

The isothermal, wall jet ﬂow ﬁeld was initially studied in the experimental
facility. Hot-wire anemometry was utilized to survey the ﬂow at the jet oriﬁce and
ﬁve downstream locations. The measurement results were compared to data
from the literature to validate the facility and the measurement techniques
utilized. The mean velocity proﬁle in the Ford-MSU test facility conformed to the
canonical self-preserving, two-dimensional wall jet proﬁle at streamwise locations
greater than xlw > 7.62.

Comparison of the experimental results with the computed results
indicates that the CFD model, with both the Reynolds stress and the k-e
turbulence models, over-predicted the velocity gradient at the wall and the inner
layer boundary growth rate. The jet spreading rate, as measured by 62, was
under predicted by the CF D model by approximately 30%. The discrepancy in
the inner boundary layer and speciﬁcally in the velocity gradient at the wall

surface is critical to successfully model the heat and mass transfer. This

159

discrepancy indicates that the model will over-predict the heat and mass transfer
at the surface.

The non-isothermal, non-condensing (T . > Twat) ﬂow was invesﬁgated
next. Hot and cold-wire anemometry were utilized during this experimental
phase. The cold-wire probe provided temperature measurements in the ﬂow.
The cold-\M're data were also utilized to temperature compensate the hot-wire
velocity measurements. Hot-wire and cold-wire measurements were conducted
at the same streamwise locations as the isothermal experiments. The velocity
data were found to agree with the isothermal measurements to within the
experimental uncertainty. This result was expected from Reynolds number -
Grashoff number comparisons and the small property variations associated with
the small temperature changes in the experimental ﬂow ﬁeld. This result also
indicated that a decoupled momentum and energy solution technique could be
utilized in the computational model with the introduction of minimal error.

Only the experimental temperature data were compared to the CF D result
since the velocity data from the experiment and the CFD were unchanged. The
computational model over-predicted the temperature gradient at the wall surface
and the growth rate of the thennal boundary layer (as measured by 87) by
approximately 30% and 10% respectively. This level of agreement is
approximately equal to the results from the velocity comparison. The thermal
penetration from the surface into the ﬂow is controlled by the transport properties

in the inner boundary layer and therefore, this level of agreement is expected.

160

The non-isothermal, condensing (T . < Tm) ﬂow regime was
investigated next. Velocity and temperature proﬁles were obtained at the oriﬁce
and at the streamwise location xlw = 10.8. It was found that the velocity proﬁle
again agreed with the isothermal measurements to within the experimental
uncertainty. This result was again expected due to the Reynolds - Grashoff
number comparison, the small property changes in the ﬂow, and the low mass
transfer rate. The mass transfer velocity to the surface is 0.3% of the jet velocity
and was not anticipated to effect the velocity proﬁle.

A measurable offset was detected in the near wall temperature proﬁle at
xlw = 10.8 between the non-isothermal, non-condensing tests and the non-
isothermal, condensing experiments. This result is consistent with results in the
literature for ﬂows with condensation phase change [Legay-Desesquelles and
Prunet-Foch, 1986]. This increase is the result of the enhanced heat transfer
resulting from the latent heat released at the wall surface.

Steady state condensation to the thermally active test section was
investigated next. The average mass transfer coefﬁcient was determined to be
h... = 0.031 m/s. Later studies of the temperature of the test surface indicate that
the mass transfer in the steady state case is under-predicted due to temperature
variations of the thermally active surface. A corrected estimate of the mass
transfer coefﬁcient of hm: 0.0407 m/s ls deduced based on the observed
temperature variations of the thermally active test surface.

A novel optical mass transfer tool was developed in the next phase of the

experimental work to study the early transient stages of the condensation

161

phenomena. Infrared measurements of the surface reﬂectance were found to be
highly sensitive to the presence of condensation on the reﬂective surface. The
apparent surface reﬂectance was calibrated as a function of the measured
thickness of water on the aluminum targets. A new technique to measure small
quantities (<5 mg) of H20 was developed to facilitate this calibration.

The lR-H20 reﬂectance technique was determined to be very sensitive to
condensate thickness and was capable of obtaining transient condensate
thickness measurements. The technique has a measurement range of 0-5 pm
and is accurate (on calibrated surfaces) to 1 0.17 pm at a 95% conﬁdence
interval. The technique has limited usefulness in determining the mass transfer
coefﬁcient due to the uncertainty associated with the concentration difference in
the ﬂow ﬁeld. Small errors in the temperature (~0.2 °C) can result in orders of
magnitude shifts in the uncertainty in the mass transfer coefﬁcient, especially
when the temperature of the surface is near the dew-point temperature of the
ﬂow.

The lR-H2O reﬂectance technique was applied to the developing wall jet
ﬂow ﬁeld in the Ford-MSU test facility. It was found that the mass transfer
coefﬁcient measurements were consistent with the expected Reynolds number
dependence. The results did not show the expected laminar transition to a
turbulent proﬁle but did conform to a turbulent Rex°'8 type proﬁle. The difference
between the published results and the data obtained in the FMTF could be the
effect of the facility design on the developing region. The particular facility nozzle

geometry and ﬂow conditioning have been noted to inﬂuence the jet up to 25 jet

162

widths downstream of the jet exit [Mabuchi and Kumada, 1972]. This is
consistent with the results published by Hsiao and Sheu [1996] who determined
that the transition region in a two dimensional wall jet ﬂow develops from 3 to 24
slot widths downstream of the jet exit. Akﬁrat [1966] also presents evidence that
the transition region can begin upstream of the nozzle exit. No measurements of
the heat or mass transfer coefﬁcient were made above xlw = 0.0 in this study.
The transient measurements obtained with the optical infrared reﬂectance
technique do agree with the steady state results obtained in the same facility.
This indicates that the variation from the literature is the direct result of the facility
and that the optical technique is functioning as expected.

The benchmark data set has been obtained and supplied to Ford Motor
Company. A new technique to measure small aliquots of water (< 5 mg) was
developed to facilitate the development of a novel optical condensate
measurement tool. An optical infrared reﬂectance technique was developed to
measure condensate on a reﬂective surface. The technique proved to be highly
sensitive and provided a 0 - 5 pm condensate thickness measurement range.
The optical infrared reﬂectance technique was applied in the Ford-MSU test
facility flow to measure the mass transfer to the thermally active test surface.
These results concluded the project and indicated that the development of the

optical technique was a success.

163

8.1 FUTURE WORK

Six recommendations are made concerning future work with and
development of the optical infrared reﬂectance technique. These are primarily
concepts to expand the capabilities of the technique, improve the determination
of the Sherwood number and to apply the method to other ﬂows of interest
related to this project.

1. Development of a more accurate method to determine or control the
concentration difference between the ﬂow and the surface of interest. The
largest contribution to the experimental uncertainty is the concentration
difference. New concentration measurement techniques could greatly
enhance the capabilities of the technique. Utilizing a dry nitrogen
environment in a closed wind-tunnel system would allow a precise amount of
water to be delivered and evaporated into the environment. This technique
would allow much more precise determination of the concentration difference
between the test surface and the ﬂow ﬁeld and reduce the uncertainty
associated with the Sherwood number. Coupling this technique with more
accurate surface temperature measurements would further improve results.

2. Veriﬁcation of the assumptions regarding distance and angle on the
condensate thickness — reﬂectance correlation. This would be critical at short
viewing distances or with optics that would decrease the infrared camera pixel
area to an order of magnitude where droplet features and patterns on the

surface could affect the correlation result.

164

3. Application on glass. To apply the technique in an actual automotive
defrosterldemister test, determination of the condensate thickness -
reﬂectance correlation would be necessary. The Inframetrics IR camera used
in the present experiments is not sufﬁcient for this procedure due to the
limited (8 bit) range. The low reﬂectance ( ~ 10%)of automotive glass to
infrared radiation would necessitate the use of a higher resolution infrared
detector. Note that a higher resolution camera would also extend the
condensate thickness measurement range of the technique on more reﬂective
surfaces.

4. Defroster/demister performance testing. As detailed in the present work,
FMVSS103 speciﬁes measuring the ‘dry' areas of the windshield during
defroster performance tests. The test begins with 500 pm of ice on the
windshield which is much larger than the dynamic range of the optical infrared
reﬂectance technique. As the test proceeds, this layer melts and areas of the
windshield become dry. The optical infrared reﬂectance technique could
determine which areas of the windshield contained less than 5 pm of
condensate quantitatively and aid in engineering analysis.

5. Validation of the technique in a canonical ﬂow ﬁeld. Further testing of the
optical infrared reﬂectance technique in previously well characterized ﬂow
ﬁeld would further strengthen the results of the present work. A ﬂat plate
boundary layer ﬂow undergoing transition would provide an ideal test
environment, especially if the concentration difference in the ﬂow could be

resolved precisely.

165

6. The thin ﬁlm model indicates that the dynamic range of optical infrared
reﬂectance technique is dependent on the absorptivity in water of the
wavelengths utilized for the reﬂectance measurement. This assumption is
supported by the visible light spectrum results obtained by Bearman et. al.
[1998] and the near infrared work performed by Anderson et. al. [1996].
Preliminary tests utilizing a 3.2 - 5.1 pm detector were conducted as a portion
of this work. The effect on the dynamic range was not conclusively
discernible from the tests conducted however due to complications that arose
with the gas chromatography equipment and the fact that the absorptivity

coefﬁcient is only 15% different over the wavelengths utilized.

The future work proposed would expand the potential applications of the
technique developed here and further verify and deﬁne the accuracy and
resolution in various test environments. The strength of the technique is the
measurement of condensate thickness on the surface of interest. Application of
the technique in defrosterldemister test environments could greatly enhance the
data available regarding visibility and driver safety and result in improved testing

methodologies.

166

APPENDICES

167

 

APPENDIX A

COLD WIRE CALIBRATION METHODOLOGY

A cold-wire probe, shown schematically in Figure A.1. is constructed from
a thin metallic wire material (5 pm diameter tungsten wire was utilized at the
FMTF) whose resistance is a function of temperature. The temperature-
resistance relationship can then be obtained through a calibration so that the
sensor can be utilized to measure temperature ﬂuctuations in a ﬂow ﬁeld of
interest. The resistance of the sensor is measured by passing a ﬁxed, small (~
1mA) sensing current through the probe. The resistance of the cold-wire probe

can then be monitored by measurement of the voltage drop across the probe.

  
   

9 w... w

LI 0 5 micrometer
tungsten
length ~ 1 mm

/ surface area ~ 4 nm2

 

 

/
0 20-30 microns
copper cladding
. l length ~lmm
\
\ \ q / I
View 'A’ x15

 

 

k

Flgure A.1. Schematic of a cold-wire probe.

168

A cold-wire calibration system was constructed and validated to improve
measurement accuracy and to eliminate the need for external cold-wire
calibration or the purchase of a calibration system. Four design objectives were
decided upon and consisted of
1. An economical device that can be incorporated into the existing FMT F facility.
Allow for cold-wire calibrations in less than 1 hour.

Incorporate 4 calibration points with differences of at least 5 °C.

:59!“

Generate a spatially uniform, steady state temperature proﬁle for each
calibration point.

The design presented in F igure A.2 was completed to meet these objective
targets [M. Desjardins, 2001]. Common materials were utilized to obtain the
most economical device possible. Stainless steel screening was utilized to heat
two of the airﬂow paths and a coiled, copper heat exchanger cooled the airﬂow in
one chamber. The screen heaters and the copper heat exchanger in the
temperature conditioning section are shown in Figure A.3. Axisymmetric nozzles
cut from soda bottles were utilized to accelerate the ﬂow to good effect due to the
high contraction ration (20:1). The exit velocity proﬁle of this nozzle is very
uniform and had been previously characterized for hot-wire calibration [Hoke,

1998]. The cold-wire calibration device is presented in Figure A4.

169

('1'. 11’. 2'1! ’. LII'.‘1/'.LJ/ '.',’U '. z/ '3. I)
'. ~ 4

v}
'1 ’2 .4
,4 \ ,/ '1
.4 x / 4
'. J . I
I” ’ T I I r ‘ t
\ .
f 4" I . 1 (I ) ,1
I, ‘1 II I ~| :4
a t I t
x . . I '1
'2 L. "x, , a" I,
I ‘ I -7 I .
r‘ I.
. e- -x' '. ~ ;'
r I . 5
, .
/ -—. l I . t I‘ 'r
r‘ ‘ l l ’4 ';
r I‘ 2’ I ‘- XV 1' /_
I I \ I '
2" \ ./ ,'
7‘ I x /
I _ :l
{'77 // ll 11'. // nut/l 2".1
.
I

 

 

“7 l .7 --*_WI '1 ' I
1 .’ I l f I.

I I ,,______
I I“ Coke bottle nozzle

 

Temperature and velocity E . ‘1.

homogenization section J I I I Y. Eggggcsagggws wire

 

Packed steel wool

' Temperature control section \
28.75" (sec Figure4fordetails) I / IE I

 

I B \ l B
I l. fi

/\~——— ss-304 screening
‘ ; . 30x30 mesh, 00.0065 wire

\

Momentum conditioning

Q ,_ ,\
j _, Thmnlcplatc

 

Inlet airﬂow

Figure A.2. Schematic of the cold-wire calibration device.

170

 

(~ 35 °C )
15 n-s<:reen
(~ 45 °C )
Unbound screen ngr heat 'I
. _. o exc ngereoi
(ambient 22C) (~IS°C)

Flgure A.3. Temperature conditioning section of the cold-wire calibrator.

 

Figure AA. Cold-wire calibration device.

171

Temperature proﬁles at the exit of the four nozzles were then taken with a

mechanical traverse system and a thermocouple. The results of these

measurements are presented in Figure A5. The temperature scale in Figure A.5

has been adjusted by subtracting the ambient temperature since the offset

between each nozzle is independent of the day to day drift in the ambient

temperature.

 

 

 

 

 

 

you
X) xxx
15 \x x=0 . ,
IO Harm 14,». 1.4% 4L4“ JL—xxxxxxx JF—ZOmm
«D “ x 7 - V. k 7
FE ﬁx it + 15 0 screen ,//' K\\

I 5 f 4'“ 8 0 screen /' \\ Coke bottle
[— Ii as Screen ’ (/ nozzle
,7 x x

o {bx it __x x x L It I x T X x. x x X x I —-¥4 mg coil / \
5 IO 15 20 ,
-5 k u x
xtxxu .H.-.H4wu}“»up/
~10

 

 

 

Location (mm)

Figure A.5. Temperature proﬁles at the exit plane of the four calibration nozzles.

Although the temperature proﬁles are not perfectly ﬂat in the center portions
of the nozzles, the spatial variation was deemed acceptable since it was less
than 1 °C I mm in the central portion of each nozzle and the local temperature is
measured independently for reference during the calibration process.

The convergence of the temperature measurement was studied to assess the
limit of the calibration accuracy. The time-averaged (mean) temperature and

standard deviation as a function of time are presented in Figure AG.

172

 

1437
C
A 14.85 :g
0 .9.
a a
1433
3 E
C
N a
on 14.81 1:
2 r:
3
14.79 x Mean temperature (D
0 Standard deviation
1417
1475

0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0

Time (minutes)

Flgure A.6. Time-averaged mean temperature and standard deviation of the cold

wire measurement as a function of time.

The uncertainty in the mean temperature utilized for the cold-wire

calibration can be calculated as [Neter et al., 1993]

XﬂKW (A1)

173

The uncertainty in the mean temperature calculated using Equation (A.1)
converges to less than 0.1 °C when two minutes of temperature data collected
with the reference thermocouple are utilized. Thus the expected accuracy of the

cold-wire measurement is expected to be better than 1 °C.

Cold-wire calibration is conducted with the following procedure:
1. Electrical power to the cold-wire calibrator and the FMTF flow are initiated and
allowed to stabilize for 30 minutes.
2. The cold-wire is positioned over the ﬁrst nozzle and 2 minutes of data are
collected.
3. Step 2 is repeated for each nozzle.
4. Mean temperature from the reference thermocouple is plotted as a function of
the mean voltage output of the cold-wire in Excel” and a linear regression model
is ﬁt to the data.
5. Results of the linear regression model are inputted into the IFA 300 Therrnopro
software for use with the calibrated cold-wire.
6. The cold-wire is now ready for use in experiments.

The author would like to acknowledge and thank M. Desjardins for her

efforts during the cold wire calibrator development.

174

APPENDIX B

EQUIPMENT AND SETTINGS UTILIZED FOR REFLECTANCE CALIBRATION
EXPERIMENTS

The following is a listing of equipment and equipment settings utilized in
the development and application of the optical infrared reﬂectance condensate

measurement technique.

Inframetrics 600L infrared camera
Emissivity set to 1.0
Ambient temperature adjusted to match environment (typically ~22.5 °C)
Temperature range - 50 °C (typically 0 °C to 50 °C was utilized)
External optics: NONE

F or.A VTG-33 video timer
75 ohm switch- ON

Sony Trinitron Monitor

Panasonic AG-2400 Video cassette recorder

Image Pro Plus - v.4.1

Brightness Setting- 37

175

Contrast Setting- 55

Image averaging - acquire 9 images and divide by 9.

A 5.0 liter/minute airflow meter (Cole Farmer) is utilized to regulate the
chilled nitrogen airﬂow to the copper heat sink. The flow rate is set at 4
liters/minute. The nitrogen airflow is chilled by circulation in a copper tube heat
exchanger (manufactured in the laboratory) submerged in a dewar ﬁlled with
liquid nitrogen.

A Hewlett Packard HP-6645A power supply connected to an Omega
Engineering OmegaLux SRFG 305/10 heater. The heater is thermally bonded
(Omega OT-201 thermal paste) to a 0.030" thick aluminum plate that has been
painted black. The heater and the aluminum plate are 3" x 5".

A BK Precision 1730 DC supply connected to a Tellurex Corporation CZ-
1.0-127-1.27 Peltier device. The system is allowed to cool for 4 to 5 minutes with
only the nitrogen airﬂow. Then 0.5 amps were provided to the Peltier device.
The temperature of the aluminum target was monitored by a thermocouple
mounted to the surface of the mechanical holder. This temperature was
monitored to insure that the surface did not reach the freezing point (0 °C).

The aluminum calibration target was then dropped in to a 22 ml
headspace vial and saved for future analysis with the HS-GC-FID technique
outlined in Chapter 6.

176

APPENDIX C

CONTROL SETTINGS UTILIZED FOR HS—C-GC-FID.

A Perkin-Elmer Autosystem gas chromatograph (Perkin-Elmer
Corporation, Norwalk, Conn.) with an integrated flame ionization detector was
utilized for all gas chromatography measurements performed. The GC-F ID
system was interfaced with a computer running PE-Nelson Turbochrom 4.1
(Perkin-Elmer Corp.) chromatography data-acquisition and control software. All
peak area measurements were performed with this software.

The control settings and equipment configuration for the GC-FlD system

utilized for all measurements are presented below:

Sample temperature Ambient ~ 22 °C

Isothermal Time 1 4 min

Oven temperature 40 °C

Injection volume 25 uL

Injection port temperature 250 °C

FID detector temperature 250 °C

Column PE—624 - 30m x 0.53 mm megabore
column

H2 Supply Pressure 41.5 psi

Air supply pressure 60 psi

Carrier gas (N2) supply pressure 90 psi

Carrier gas (N2) ﬂow rate 10 mUmin

177

Manual injection was performed using a 100 pL Pressure-Lok" precision
analytical syringe (Precision sampling, Inc. Baton Rouge, LA.). The syringe and

an example of the 22 mL headspace vials utilized are presented in Figure C.1.

 

Flgure 0.1. Analytical syringe and 22 mL headspace vial and crimp cap utilized

for chromatography experiments.

Calcium carbide (CaC2) was added to every sample to convert the water
present to acetylene (C2H2) as detailed in Chapter 6. This gas in the headspace
of the vial was then sampled with the syringe and injected into the port on the PE
Autosystem gas chromatography machine. Sample volumes of 25 uL were

found to optimize the dynamic range of the system for the concentrations utilized.

178

APPENDIX D

COMPARISON OF THE HEAT TRANSFER ASSOCIATED WITH THE PHASE
CHANGE EVENT TO THE SENSIBLE HEAT TRANSFER

The following is a derivation of the heat transfer associated with
condensation phase change. While not technically difﬁcult, this derivation
underscores the importance of mass transfer phenomena to the ﬁeld of heat
transfer and is therefore included for completeness.

The mass transfer - heat transfer analogy dictates that the transport
phenomena in a convective ﬂow are equivalent [Bejan, 1995]. The governing
heat transfer relations can also be extended to the mass transfer event with
appropriate modiﬁcation of the terms in the equations. This substitution leads

directly from Newton's law of cooling,

q"= hAT. (0.1)

to F ick's law of mass diffusion,
j"= hmAC. (0.2)

. This equivalence leads to the following relationship between the heat transfer

coefﬁcient (h) and the mass transfer coefﬁcient (hm) [lncorpera and Dewitt, 1990]:

179

h k _, 1
h =D Len=pche' , n~§. (0.3)
AB

 

For a phase change event with water vapor in air, the Lewis Number (Le =
011D), thermal conductivity (k) and the diffusivity (DAB) are approximately equal to
0.82, 0.025 W/(mK) and 2.52x10‘5 mzls respectively. This results in a mass
transfer coefﬁcient approximately of 1/1000th of the numerical value of the heat
transfer coefﬁcient. That is, if the heat transfer coefﬁcient is approximately 40
W/(mzK), the mass transfer coefﬁcient would be 0.040 We. Note that the term on
the relation on the right hand side of Equation (D3) is not dimensionless.

The latent heat (L) released or absorbed during a mass transfer process
can then be utilized to calculate the heat transfer associated with the mass

transfer.

qm"= Lj"= Lh mAC (0,4)

The concentration difference in the ﬂow can be driven by several factors
including the temperature gradient, chemical absorption, or ﬂow stratiﬁcation.
The temperature gradient is responsible for the concentration difference in the
windshield defrosterldemister application and in the experiments conducted in

the Ford-MSU test facility. Therefore, the water concentration in the wall jet and

180

at the thermally active test surface must be calculated from the relative humidity
and the temperature gradient present in the ﬂow ﬁeld.

The concentration of the wall jet is calculated directly from the measured
jet temperature and relative humidity (%Rh or 0). The concentration at the
surface of the thermally active test section is inferred based on the assumption of
thermodynamic equilibrium and the measured surface temperature. Initially, the

saturation pressure is calculated utilizing the Clausius-Clapeyron relationship

[Cengal and Boles,1994]

hf8 1__1_
R Tad T2

P238t = Prefe . (0.5)

Once the saturation pressure is calculated, the speciﬁc humidity (01) is

calculated:

 

_ 0.622 %Rh P83,
w _ Pm -P (0.6)

sat

Note that the relative humidity for the jet is measured directly in the
experiment and assumed to be equal to 100% at the wall surface during active

condensation. The speciﬁc humidity provides the concentration in units of

181

kg(H20)/kg(air) and thus the concentration in the units kg/m3 can be obtained by
multiplying by the local density.
The concentration difference in the flow (AC) can then be written by

combining Equation (0.5) and Equation (D.6):

Ac 40191-00. =

 

 

 

/ IﬁgI;_III IhfgI I III I
R TRef TI TI: Tic—{Ts
_ pj%the pse
_O.622PRC,. [5% 1 4]] —P P IifIr-L‘TLII (0.7)
R T e Tj o _ c 6 Ref 3
\PTot_PRcfe Rf Tt Rf }

 

The density can be calculated from the local temperature utilizing the ideal
gas law. The Clausius-Clapeyron equation is not exact however and slight error
is included in its application. This error becomes evident when the ideal gas law
is combined with Equation (DJ) and the concentration difference is small (for
example, at the onset of condensation). In these circumstances, the average
density of the ﬂow was utilized to maintain consistently positive (>0)
concentration difference. The acceptability of this application is drawn from the
fact that often condensation was observed to form even when Equation (DJ) and

the ideal gas law indicated negative concentration difference.

182

APPENDIX E

CATALOG OF CONDENSATION AND MASS TRANSFER MEASUREMENTS
OBTAINED WITH THE OPTICAL INFRARED REF LECTANCE TECHNIQUE.

The following ﬁgures contain the condensate thickness and temperature
data and the resulting Sherwood number calculations for the transient
condensation experiments conducted in the Ford-MSU test facility. Note that the
speciﬁc conditions of each test varied slightly and certain experimental features
could not be reproduced between experiments (such as the temperature of the
thermally active test surface). In spite of this test to test variation, the
condensate thickness measurements between experiments present consistent
development patterns. The condensate thickness development rates differ by
less than 5%. for experiments with similar wall jet temperature and humidity
levels. This repeatability is further demonstrated when the Sherwood number
results for different tests are overlaid. Speciﬁcally Figure 7.9, which compares
Sherwood number results obtained on different (but overlapping) sections of the

thermally active test surface.

183

 

 

 

Thickness (pm)

 

 

 

 

 

 

Flgure E.1. Thickness results for Rew = 6560 test
(11 June 2001. Experiment #2).

 

 

 

 

 

 

 

 

 

5 .
100 150

Time (s)

Flgure E.2. Temperature log for Raw = 6560 test (11 June 2001. Experiment #2).

80

 

. l2-t2 = 94-75

60 ‘ I t2-t1 = 87-75

 

'3; t2—t1=94-62

 

 

 

 

Flgure E.3. Sherwood number results for Rew = 6560 test (11 June 2001.

Experiment #2).

 

Thickness (ll m)
N w A

_L
r

 

 

 

 

Figure 5.4. Thickness results for Raw = 8400 test

(15 June 2001. Experiment #4).

185

Temp (°C)

 

0 20 40 60 80 100
Tlme (s)

Figure E.5. Temperature log for Rew = 8400 test (15 June 2001. Experiment #4).

 

 

120
10° F a t2—t1 = 5740
30 _ s t2-t1= 5745
U
. t2-t1 = 63-51

 

 

 

 

 

 

 

O 2 4 6 8
xlw

Figure E.6. Sherwood number results for ReW = 8400 test
(15 June 2001. Experiment #4).

186

N

 

 

 

 

 

 

 

 

 

 

 

 

———‘L~
[ot=32l
6. - t=37l
. r
o .- ‘1‘. _ 'At=42l
5 o E. - y _‘ *I‘ ;.‘r###th-47
E :. ab ‘ ' - xt=53§
3+ -F _,, " °t=581
3‘ $5”: ' . -—- . -t=68l
3 l
.5 -t=73,
2_ 19’ .
1 .
O _
o 1 2 3 4 5 6 7 a
xlw
Flgure E.7. Thickness results for Raw = 10200 test
( 15 June 2001. Experiment #3).
30
o.
25‘ : 1.:
A 2.5
A _ o x 3.8
.3 2° ~. . 5
a _.-..
l- 15 a ————————————— -l- - Tdewpornt
10 ~
5 I I I I I
0 20 40 60 80 100

Tlme (s)

Figure E.8. Temperature log for Rew = 10200 test (15 June 2001. Experiment
#3).

187

 

 

 

 

 

, .t2-t1=58-47

so . E amuse-53
'i t241=5e42l
4.;

60 ‘
E5."

 

 

 

3
.c
U,

40

20

0 r f
0 2 4 6
xlw

 

Flgure E.9. ShenNood number results for Rew = 10200 test (15 June 2001.

Experiment #3).

 

Thickness(um)
'7’ °.° ‘7 ‘."
\‘i’s it“

i
l

 

 

O
—h
N

Flgure E.10. Thickness results for ReW = 13300 test
(15 June 2001. Experiment #2).

188

 

L,__A4

 

 

 

 

 

 

 

 

 

 

 

 

0 20 40 60 80 100
Tlme (s)

Flgure E.11. Temperature log for Rew = 13300 test
(15 June 2001. Experiment #2).

 

 

  
   
  

    

 

 

 

 

120 x
I. l .1241 = 63-52 a t2-t1 = 58-52 l
100 . l atZ-tl = 6346.5 xt2-t1 = 5246.5 1
1 x 1241 = 69-59 a t2-t1 = 5846.5
80 - +t2-t1 = 7363
a 60 T '<, . ' ~ i".\ It,
40 -
X
20 -
o I I I
0 2 4 6 8

xlw

Figure E.12. Shemood number results for Rew = 13300 test (15 June 2001.

Experiment #2).

189

Temp (°C)

 

1

 

 

 

 

 

 

 

 

 

o t=+50
4 ~ :- t=+62
3_5 - A t=+69
E x t=+75
:1. 3 —
v x t=+88
g 2.5 . t=+101
g 2 f + t=+108
r5 - t=+122
1.5 4
8

 

Figure E.13. Thickness results for Re” = 13300 test
(5 June 2001. Experiment #1 ).

 

 

 

 

 

 

 

 

 

 

 

0 xlw-0.5
25
II XIV/=15
A xlw-2.5
20 ‘ x xlw-3.8
‘5. I o xlw=6
s
15 T I xlw=9
1')»
10 _ ———————————————— — — Tdewpoht
5 r I
0 50 100 150

Time (s)

Flgure E.14. Temperature log for Rew = 13300 test
(5 June 2001. Experiment #1).

Thlclrness ( p m)

 

 

 

 

o t2-t1 = 101-88
- 1241 = 122-101

 

1241 = 108-88

 

 

Flgure E.15. Shemood number results for Rew = 13300 test

(5 June 2001. Experiment #1 ).

 

 

9
010:

.N s»
“(001%

a
01”

 

 

o t=40
n t=51
A t=63
x t=71

'- xt=83

=91
=1oa
=111

 

 

 

Flgure E.16. Thickness results for ReW = 13300 test

 

(20 June 2001. Experiment #1 ).

191

 

25

20—

Temp (°C)
a

80

201

Figure E.18. Sherwood number results for Rew = 13300 test (20 June 2001.

 

 

 

O

X

I

xlw=9
xIw=12
XM=2.5

,, l

1

I

x/w=3.8 ‘

xlw=5

 

TJBI

 

— — Tdewpoint ‘

 

 

 

50 100
Time (s)

Figure E.17. Temperature log for Rew = 13300 test

(20 June 2001. Experiment #1 ).

 

 
    

0 12-11 = 91-71

a 12-11 = 83—71

12-11 = 91 -63

 

 

0 (Va

9'" . 1 .l.
V ~41» A
o

  

 

 

 

Experiment #1 ).

192

 

 

 

 

 

 

 

 

o t=32
4 < a t=44
‘ . t=50
3'5 ‘ ‘ . x t=62
A _ ‘ =
E 3 I 74
1 o t:
2.5 + 1:95
2 . t=100
2
1‘3 1.5
1 .
0.5
0
4

 

Flgure E.19. Thickness results for Rew = 13300 test (20 June 2001.

Experiment #2).

 

 

 

 

 

 

 

 

Tlme (s)

Figure E.20. Temperature log for Rew = 13300 test (20 June 2001.

Experiment #2).

193

 

 

 

 

 

 

 

80 ‘,ﬁ
0 1241 = 84-62
I: tZ-II = 74-62
60 ‘ 1
1241 = 84-50 T
5’, 4o
20 “
O r
4 6 8 10 12
xlw

Flgure E.21. Sherwood number results for ReW = 13300 test (20 June 2001.

Experiment #2).

APPENDIX G

ERROR ANALYSIS IN THE SHERWOOD NUMBER DETERMINED WITH THE
OPTICAL INFRARED REFLECTANCE TECHNIQUE.

The following program written in Mathematica" was utilized to calculate
the uncertainty in the Shemood number. The uncertainty is the result obtained
with the optical infrared reﬂectance technique for the test conditions utilized on
15 June, 2001. That experiment was typical of the conditions utilized for tests
contained in this thesis. Statements have been added to clarify the input to and

output from the program.

Clear[Sh. w. Dif. Ptot. Psatj. Psatw. Wj, WW. Rair. Rwatcr, Tj. Tw. hm. Pref.

Trcf. th, Delt, T2, T1]
Psatlej_] = Pref: e"((hfg/ .461) s (1 [Tref- 1/Tj))i
Psatw[Tw_] = Prefa- e"((hfg/ .461) t (1 /Trcf- 1/Tw));
Wj[Psatj_, th_] = 0.622 s th a» Psatlej] / (Ptot - th t Psatlejll:
ww[Psatw_] = 0.622 s Psatw[Tw] / (Ptot - Psatw[Tw]);
Cj[Wj_. Psatj_. Tj_] = (lePsatj, thl m (Ptot - Psatlej])) / (Rain Tj):
CW[WW_, Psatw_. Tw_] = (WW[Psatw] s (Ptot — Psatw[Tw]))/(Rairau Tw);
hm[T1_, T2_, Dclt__. Cj__. Cw_.] =

(T2 — T1) :1: r/((Cj[Wj. Psatj. Tj] - Cw[wW. Psatw, Tw]) s Delt);

Shlhm_. Dif_. w_] =
hmIT1, T2, Dclt. Cj[Wj[Psatj, th]. PsathTj]. Tj].
Cw[WW[Psatw]. Psatw[Tw]. Tw]] a: w/Dif;
Shlhm, Dif. w]

195

The resulting form of the Sherwood number:

(
((—Tl+ T2) wp)/ lDelt Dif
I
( 2 1692hf - 1, + 1 2 l692hf _ 1,,.,._,1
1:0.622e g( T3 Tmf) Pref {—e g( T3 Tmf) Pref+ Ptot
l

 

_ 1w, .1
th)/ {Rair [Ptot_e2.1692hfg( Tj Tref) Pref th] T3.) _
1

1
_O._6_gg e2.1692hfg( Tref‘m) Pref H

 

Rair Tw ) )1

The various sensitivity coefficients are then calculated:

(u Error analysis in the Sherwood calculation n)

(*1: Sensitivity Coefﬁcients u)
STj[Tw_] = D[Sh[hm, Dif, w], Tj];
STw[Tw_] = D[Sh[hm, Dif, w], Tw];
Sth[Tw_] = D[Sh[hm, Dif, w], th];
Sw[Tw_] = D[Sh[hm, Dif, w], w];
SDif[Tw_] = D[Sh[hm, Dif, w], Dif];
ST1[Tw__] = D[Sh[hm, Dif, w], T1];
ST2[Tw_] = D[Sh[hm, Dif, w], T2];
SDelt[Tw_] = D[Sh[hm, Dif, w], Delt];

196

The following is the output for sensitivity coefficient for the jet temperature

 

 

 

(Tl):
,1
—; (-T1+T2)
!
K
4. 338391115 _ 1.71 2.1692111? 1 41
{1.3492412 g1 13 Tmf)hfg1;>ref?- {-e g1 13' W1Pref+1>tot1th2
. _ ___ I # 1g __. #1” ___}:1‘ -___,___ ‘7"*—. #—_*+ -
1 RairiPtot—e 2 l692hfg1 TU“ Tmf) 12112me1 T33
4. 33839hfg(— 1.+— 1 ) 2 .
1. 34924e T] W hfg Pref Rh]
.1 _ .1 ___.._____1_ W 11.1____ _._11__ +
1 1
Rair[Ptot-e 2 1692th( TJ 1191?) meRh3'1Tj3
2.1692hf 1 1 2.1692hf —1.+ 1--
1.34924e ‘31 T5 m)hngmf{—e (3113 ﬁef) Pref+Ptot) 1613'
'0'; ___-a- ___ “—1 6925f 151—1 ’ _____.._____ '
Rair Pt t-e2 1 g1Tj lief) Prethj} Tj3
2.1692hf 1 1 2.1692hf 1 1 l \
0.622e g1 T‘J‘ w)1>ref{_e 9113' TIP-f) Pref+Ptot1th
_ _____-__ﬂ ___.____. ___ .11.... __.___ __L .__1 ..pl/
1 + 1 ; '
Rair {PtOt-e 2 l692hfg( 'Ij TIEf) Prethj) TjZ )i )1
2.1692hf 1 1 2.1692hf 1 1
.{ {0. 622e g( 1‘3 M) Prefr—e 9113 111115) Pref+Ptot1th
‘DeltDiva —-- --——-—- -————— -——1——— 1—- 1— 1 —
9
1 1 Rair{Ptot- e2162hfg1 TJ' Tmf)Prethj] Tj

l i
1 3
0 622 ilfgfhfgief._ﬂ eel 2:

RairTw . :

2' J

197

The following is the output for sensitivity coefficient for the surface
temperature (T w):

 

 

 

 

I
-1 (- Tl+ T2)
1
11 1
2L 34924e 2. l692hfg1lief'lw1 hngref 1108.222 2.1692hf91nef'm) pref? w01/
1 12511er RairTwZ ) )
2.1692hf -1...- 1 2.1692hf - 1.. 1.
1 110.622e (1113 T11‘1f1Prefz1—e 9117 11951 Pref+Ptot3th
.Delt Dif 1 - -——{ 1- 1 1 -_ _
1 2.1692111?
1 1 Rair \Ptot- e g1 T3 11Ef-1PrethjjTj
2 1692hf 1 1 1 1
0 €22 _ 11111111 Pref .21
RairTw l 1
I I

The following Is the output for sensitivity coefficient for the Jet slot width
(%Rh):
1 ) 2.1692hfg(- 1

I I 4. 33839hf -—- ~ I 1.4—1 \
‘ 0.622eg11'1‘j Tref Pref?- re T3 Tr‘31)Pref1rPtot)th

! i
— (-T1+T2)' - —- -— 1 2 -+

- 1 I 2.1692hf 4» 1
1 1 Rair Pt ot-e 911 '13 Tref) Prethj)’ Tj

\ K

 

 

f‘

I 2.1692hfg(~ 1.

1 1
1 Pref i—e 1‘3 9f) Pref+ Ptot 1

1.
- _ 1WD 1/
e2.l692hfg( 1 1 f_) .\ ,
RairiPto ot-e Tj Tre'f Prethj)'Tj 1}

I

 

 

0.622e

2.1692hf ”1.4—- .. 2.1692hf _11. Trefl

.Delt Diff - -———_--_.- 12-1.,1__1___
1 , I e2.1692hfg1( 1 1f) ,1 ,
1 '\ Ralr \Ptot-e Tj Pref th) T3

1
Pref + PtOt) th

 

1
1 1
O. 622 e2"1692hfg1h“ef 1w) Pref I

11
_______ﬁ,__ .- $2,
1' 1'

 

198

The following Is the output for sensitivity coefficient for the Jet slot width
(W):

((-T1 + T2) m/

 

( ( 2.1692hf — 1.+ -1— ( 2.1692hf - 1..., 1,2 \

J 30.622e g( T3 T1915) Pref V8 g( TJ TM) Pref+Ptot) th
JDelt Dif; ---——— —-———— f ———-—-————-- — ””"T'i - \ — ——-—— —
- 2.1692hf _

\ J Rair \Ptot-e gJ T3 hefJ Prethj)‘ Tj

H
1 1
0 622 2 l692hngnef1wJ __Pref ‘

Rair'I‘w i*

J)

The following Is the output for sensitivity coefﬁcient for the Diffuslvity (le
or D):

-( (-T1 +T2) wm/

 

 

2.1692 1 1 2.1692 1
if JD 6229. hng T3 ”31) Pref -e hngl "U W) Pref+Ptot th
:Delt 01?} __i_-._.___._____ -———. —- .21.. .1. . -
6 - .+
L J Rair Ptot-e 219%ng T3 Tief) Prethj; Tj

 

)J
1
0.622e e21692hf9Jnef 'IwJ Rref g1

Rair Tw

 

J J
The following is the output for sensitivity coefﬁcient for the condensate

thickness at t1 (T 1):
— (wp)

1'+ l \|
T3 “933) Pref+ Ptot) th

LDelt Dif; _______________,__, .- _ 1 1 \ .. ___ _ _
J _ W) Pref th ) Tj

1
O 622}? 2.1692hfg( )m PFef

 

_Rair 'I‘w

199

The following Is the output for sensitivity coefficient for the condensate
thickness at t; (T2):

(m /

 

 

{0.622 e2'1692hfg111'j1 '11"th Pref J-e2'1692hfg(_1‘lj+ 1119f) Pref+ PtotJ th
JDelt le' ___ —— —7——- 2 152%:;(_1.1 ) J j
i J Rair KPtot— e ' Ti Tiref Prethj)‘ Tj
1 1 H
0.622 £1692“ng 1:9.wa Pref ; f
’ —' _ﬂRaiETwi—__ ' "V

J)

The following is the output for sensitivity coefficient for the time between
condensate thickness measurements (Delt):

-((—T1+ T2) wm/

 

 

 

2.1692 1. 1. .92 -1.._1
J (06229 hfg( T3 1191) iI-’ref{-e2 16 hng 1‘] Tref) Pref+ Ptot“ th
,DeltZDif f- Mia—gm»? ﬂ--- ___l 1 __m \ __ __ _
J J RairKPtot-e2'1692hng'Tj1hefJ prethj) Tj

l l \\
0.6223169211ng nef‘mJ Pref J !

RairTw ;i
J}

200

The following Is the input of the variable values and the uncertainty
associated with each):

(u Error estimates for the parameters u)
dT2 = .3/ 1000000;

dT‘l = .3/1000000;

chlt = 0.5,

dw = 0.0005;

dth = 0.02;

dDif = .0000002;

dTw = 0.5;

de = 0.5;

(u Parameter values u)

Rair = 0.287,

r = 1000;

Tref = 293.15;

PTCf = 2.339,

hfg = 2454.1;

Dif = 1.87 .10"(-7).(Tj + Tw)/2 - 2.96 4. 10*(— 5)
Tj = 273.15 + 24.6;

Tw = 273.15 + 11.1;

th = 0.46;
W = 0.02,
Dclt = 10;

T2 = 3.90 / 1000000;
T1 =1.98/1000000;
Ptot = 101;

201

The following is the output for the uncertainty in the Sherwood number and
the graphical representation of the result of the calculation:

(in: Potential Error in Sherwood number calculation 4*)
dSh =
((STj an de)"2 + (STw a: dTw) + (Sth :- dth)“2 + (Swan dw)"2 + (SDifau dDiﬁ"2 +
(8T1 a: dT1)"2 + (8T2 an dT2)"2 + (SDelt :- dDelt)"2) A0.5

The equation representing the uncertainty in the Sherwood number can be
calculated with:

5.89824x1025

. m.43(0.0034112—11’) )2

s\J0.00880709- 5._06919§-____m ——u _. i (-0.0000296+9.35x108(295.15+'IW))4
K )

 

 

+

2.44594 x 1018

523.0(000341122— 1 ) i4

I
000880709- 9069199 10 1T!— q (-0.0000296+ 9.35x108 (2953.15.06 )2
K J

3.6608x1013

( 5323.43 0.0034112 .1 12
{0.00880709— 1069190 1‘1” BL); (-0.0000Z96+ 9.35x 108 (295.15+TW) )2
K

+

 

 

5313 1
9.6x 107'(_ ~2€9€5059¥®31 _ ._ Wm“ + 5.069199 “430334112:w)*\ T
k ._ H‘— 1‘“ ' ' ' \.,__ 1- .-_,¢—________ I _____ i A0.5

5323.43(0.0034m— 1) i?—

r __
J0.00880709— 5.069199 m (~0.0000296+9.35x10‘8 (29515.00) J
\ J

 

 

 

202

The graphical representation of the uncertainty in the Sherwood number
follows:

1000

Uncertainty
ON
O
Y '9 w

 

276 278 280 282

Note that the variable names utilized in the Mathematica" program do not
necessarily match the variable names utilized in the remainder of this document.
A list of the Mathematical" variables and the counterpart in the document is

provided in Table F.1 below.

203

Table F.1. Mathematica’lb variable names and corresponding variables in

 

 

 

 

 

 

 

 

 

 

 

 

 

 

document.
Mathematic Standard
a variable Variable
name

Dif D

hm hm

Pref PM

Rhi th

Delt At

03w cow

Rair R93
Rwater Rw

T] T,-

Tw Ts

Cj Cj

Cw C.3

T1 T(t1)

T2 T(t2)

 

204

APPENDIX 6

PUBLICATIONS ASSOCIATED WITH THIS WORK

Hoke, P. B., Wang, Q., and McGrath J. J., AbdulNour, B. S., 1999, Experimental
and Numerical Study of a Condensing Flow in 8 Developing Wall Jet,
Proceedings of the First International Turbulence and Shear Flow
Phenomena Conference, September, pp. 391 - 396.

Hoke, P. B., Kinnen, A. R., McGrath, J. J., AbdulNour, B. S., 2000, Experimental
Measurement and Numerical Prediction of Condensation in 8 Developing
Wall Jet Flow, 9th (Millennium) International Symposium on F low
Visualization 2000, Edinburgh Scotland, pp.134-1 - 134-10.

Hoke, P. 8., Loconto, P. R., McGrath, J. J ., Calibration of an Optical Condensate
Measurement Technique Utilizing Indirect Headspace Gas
Chromatography, The Journal of Chromatographic Science - Submitted
March 2001.

Paul B. Hoke, John J. McGrath and Paul R. Loconto, Quantitative Analysis by
Indirect and Reactive Static Headspace GC: Determination of a Thin Film
of Water Deposited on an Aluminum Surface; Calibration Considerations,
American Laboratory, Submitted July 2001.

Other proposed publications:
Hoke, P.B., McGrath, J. J, Measurement of Condensate Film Thickness on

Reﬂective Surfaces with infrared Themrography, Measurement Science
and Technology, Proposed submission September 2001.

205

BIBLIOGRAPHY

206

Bibliography

AbdulNour, R. S., Willenborg, K., McGrath, J. J., Foss, J. F., and AbdulNour, B.
S., 1997, Measurements of the Convective Heat Transfer Coeﬁ'rcient for a
Two-Dimensional Wall Jet: Uniform Temperature and Uniform Heat Flux
Boundary Conditions, HTD-Vol. 353, Proceedings of the ASME Heat
Transfer Division, Volume 3, pp. 109-116.

AbdulNour, B. S., 1999, CFD Prediction of Automotive Windshield Defrost
Pattern, Paper No. 1999-01-1203, Society of Automotive Engineering,
lntemational Congress 8. Exposition, Detroit, Ml, p. 1-9.

Akﬁrat, J. C., 1966, Transfer of Heat From an isothermal Flat Plate to a Two-
Dimensional Wall Jet, Proceedings of the Third lntemational Heat Transfer
Conference, (2), pp. 274-279.

Alenezi, G. A., and Abdo, M. S. E., 1991 , Mass- Transfer Measurements Around
a Single Cylinder in a Cross Flow at Low Reynolds Numbers, Chemie
lngenieur Technik, Vol. 63, pp. 381-384.

Anderson, K.F., Carlson C., and Wells, H., 1996, Infrared Water-Film Monitor for
Hydroponic Systems. NASA Tech Briefs, Vol. 20, October, No. 10.

Bearman, G. H., Biswas, A., Chrien, T. 6., Green, P., and Green, R. 0., 1998,
Optical Remote Detection of Ice on Aircraft Surfaces, NASA Tech Briefs,
NPO19929, Vol. 22, No. 9, pp. i, 1-7.

Bejan, A., 1995, Convective Heat Transfer, John Wiley and Sons, New York, pp.
46-51, 403-412. 466-477.

Bradshaw, P. and Gee, M. T., 1960, Turbulent Wall Jets With and Without an
External Stream, Aero. Res. Council, RSM 3252.

Cengel, Y. A., and Boles, M. A., 1989, Thermgyngmigz An Engineering
Approach, McGraw—Hill Inc, New York, pp. 639, 874.

Chui, Boon Keat, 2000, Personal Communication.

207

Cloud, Gary, 1998, Ogtigl Methgs of Engineering Analysis, Cambridge
University Press, New York, pp. 180, 416.

Davies, T. W., Patrick, M.. and Tanyildizi, V., 1990, Local Evaporation Rates in
the lmpingement Zone of a Free Air Jet, Proc. 18th Australasian Chem.
Eng. Symp., Auckland, New Zealand, 1(13), pp. 408-414.

Davies, T. W., Patrick, M. A., Balint, T., Balint, T. S., and Tanyildizi, V., 1991,
Distribution of Local Mass Transfer Rates in Obstructed Duct Flows, Proc.
Symp. On Heat Transfer, lchemE, Nottingham, UK.

DeBecker, B., Duby, P., Modi, V., West, and A. C., 1995, Mass-Transfer to a Line
Electrode in Presence of a Periodic Fluid-Flow, Journal of the
ElectoChemical Society, Vol. 142, September, pp. 3001-3007.

Desjardins, M., 2001, Personal communication.

Eriksson, J. G., Karisson, R. l., and Persson, J., 1998, An Experimental Study of
a Two-Dimensional Plane Turbulent Wall Jet, Experiments in Fluids, Vol.
25. Pp. 50-60.

Fedorov, A. G., Vrskanta, R. and Mohamed, A. A., 1997, Turbulent Heat and
Mass Transfer in an Asymmetrically Heated, Vertical Parallel-Plate
Channel, international Journal of Heat and Fluid Flow, Vol. 18, No. 3,
June, pp. 307-315.

F Orthmann, E., Uber turbulente strahleusbreitung, lng. Arch., Vol. 5, 42, 1934.

General Eastern instruments, 1999, M - lu w- oin nit rO er or
Manual, Document No. M0212904, Rev. A.00, Massachusetts, p. 89.

Goldstein, R. J. And Cho, H. H., 1995, A Review of Mass-Transfer
Measurements Using Naphthalene Sublimation, Experimental Thermal
and Fluid Science, Vol. 10, lss. 4, pp. 416 — 434.

Goldstein, R. J., 1996, Fluid Mechanig Measurement- 2"“ Edition, Taylor and
Francis, Washington DC, pp. 120, 141-143.

208

Gregory, C. 0., (1997). MRI Artifact Gallery [Online], Available

http:/Ichickscope.beckman.uiuc.eduIroostslcarl/artifacts.htmL (December
21, 2000).

Hassan, M. B., Petitjean, C. Defﬁeux, J. C. and P. Gilotte, 1999, Windshield
Defogging Simulation with Comparison to Test Data, SAE 1999—01-1202
International Congress 8 Exposition, March 1999, Detroit, MI, USA,
Session: Climate Control Systems (Part C8D).

Ho, C.-M., and Huerre, P., Perturbed Free Shear Layers, Annual Review of Fluid
Mechanics, Vol. 16, pp. 365-424.

Hoke, P. B., 1998, Experimental Measurement of the Slit Response Function and
Corrected Themrographic Measurements, Masters Thesis, Michigan State
University, p. 43

Hoke, P. B., Qingtian, W., McGrath, J. J., AbdulNour, B. S., 1999, Experimental
and Numerical Study of a Condensing Flow in a Developing Wall Jet,
Turbulent Shear Flow Phenomena -1 Conference, Santa Barbara CA, pp.
393-395

Hoke, P.B., Kinnen, A. R., McGrath, J. J., AbdulNour, B. S., 2000, Experimental
Measurement and Numerical Prediction of Condensation in a DeveIOpr’ng
Wall Jet Flow, 9th lntemational Symposium on Flow Visualization 2000,
Edinburgh Scotland, pp. 134-1 - 134-11.

Hsiao, F ., -B., and Sheu, S. -S., 1996, Experimental Studies on Flow Transition
of a Plane Wall Jet, Aeronautical Journal, November, Vol. 100, pp.373-
380.

lncorpera, F. P., and Dewitt, D. P.,1990, Fundamentals of Heat and Mass
Transfer 3"1 Edition, John Wiley and Sons, New York, pp. 353-367, 608-
615, 872.

 

lncorpera, F. P., and Dewitt, D. P.,1996, Fundamentals of Heat and Mass

Transfer 4‘" Edition, John Wiley and Sons, New York, pp. 320-322, 665-
666.

 

lnframetrics, Inc., 1988, Model 600L Operators Manual, Document #05250-200
REV C, Massachusetts. pp. A1 -A12.

209

Kestin, J., Maeder, P. F ., Wang, H. E., 1961, Inﬂuence of Turbulence on the
Transfer of Heat From Plates With and Without a Pressure Gradient,
lntemational Journal of Heat and Mass Transfer, Vol. 3, pp. 133-154.

Kinnen, A., 2000, Experimental Investigation of Condensation for Small
Concentration Differences and Short Times, Studienarbeit, Rheinishch-
WestFalische Technsche Hochschule, Aachen Germany, pp. 23-38.

Launder, B. E. and Rodi, W., 1981, The Turbulent Wall Jet, Progress in
Aerospace Science, Vol. 19, Pergamon Press, pp. 81-128.

Legay-Desesquelles, F ., and Prunet-Foch, 8., 1986, Heat and Mass Transfer
with Condensation in Laminar and Turbulent Boundary Layers along a Flat
Plate, lntemational Journal of Heat and Mass Transfer, Vol. 29, No. 1, pp.
95-1 05.

Loconto, P. R., Pan, Y. L., Voice, T. C., 1996, Determination of Trace Water in
Soils by Automated Indirect Headspace Gas Chromatography. LC-GC-
Magazine of Separation Science, Vol. 14, issue 2, pp. 128-132.

Loeper, J. and R. Grob, 1988, indirect Method for the Determination of Water
Using Headspace Gas Chromatography, Journal of Chromatography,
1988.457, 247-256.

Mabuchi, l., and Kumada M., 1972, Studies on Heat Transfer to Turbulent Jets
with Adjacent Boundaries, Bulletin of the JSME, Vol. 15, no. 88, pp. 1236—
1245.

Mao, Y., Besant, R. W., and Rexkallah, K. S., 1992, Measurement and
Correlations of Frost Properties with Air Flow Over a Flat Plate, ASH RAE
Transactions, 98(2).

Matuszkiewicz, A. and Vernier, Ph., 1991, Two-Phase Structure of the
Condensation Boundary Layer with a Non-Condensing Gas and Liquid
Droplets, International Journal of Multiphase Flow, Vol. 17, No. 2, pp. 213-
225.

Modest, M. F ., 1993, Radiative Haat Transfer, McGraw-Hill, New York, pp.442.

210

Moffat, R. J., 1988, Describing the Uncertainties in Experimental Results,
Experimental Then'nal and Fluid Science, No. 1, pp. 3-17.

Morel, T., 1977, Design of Two-Dimensional Wind Tunnel Contractions, Journal
of Fluids Engineering, Vol. 99. PP. 371-378.

Neter, J., Wasserman, W., and Whitrnore, G. A, 1992, Applied Statistics, Allyn
and Bacon, Boston, pp. 269-271.

Noller, C., 1966, Chamistgy of Organic Compounds, pp. 170-172.

Noui-Mehidi, M. N., Salem, A., Legentilhomme, P., and Legrand, J., 1999,
Overall Mass Transfer in the Swiriing Flow Induced by a Tangential Inlet
Between Coaxial Cones, Journal of Applied Electrochemistry, Vol. 29,
November, pp. 1277-1284.

Patel, R. P., 1962, Self-preserving, Two-dimensional Turbulent Jets and Wall
Jets in a Moving Stream, Masters of Engineering Thesis, Department of
Mech. Eng., McGill University, Montreal.

Pekdemir, T., 1994, Convective Mass Transfer from Stah'onary and Rotating
Cylinders in a Jet Flow, Ph.D. Thesis, No. DX185786, Exeter University,
UK, pp. 24, 39, 49,134-140, 150, 157-160.

Pekdemir, T., and Davies, T. W., 1999, Establishment of a new mass (heat)
transfer measurement technique and its validation, Experimental Heat
Transfer, Vol. 12(1): pp. 33-51.

Peters, A. R., 1972, Interior Window Fogging - An Analysis of the Parameters
Involved, Paper No. 720503, Society of Automotive Engineers, National
Automotive Engineering Meeting, Detroit, pp. 1-4.

Schlesinger, M., De Kee, D. and M. N. Godo, 1986, Apparatus for Film-thickness
Measurements of Postwithdrawl Drainage, Review of Scientiﬁc
Instruments, Vol. 57 (10), pp. 2535-2537.

Siegel, R. and Howell, J. R., 1992, Themal Radiation Heat Transfer,
Hemisphere Publishing Corp. Washington, p. 176.

211

Skoog, D. A. and Leary, J. J., 1992, Principlea of lnatrument ﬂalysis, Fourth
Edition Saunders College Publishing, New York, p. 610.

 

Society of Automotive Engineers, 2000, §AE Handbmk, Society of Automotive
Engineers, Pennsylvania, Vol. 3, pp. 34.19 - 34.26.

Tailland, A., 1967, “Contribution a I'étude d’un jet plan dirigé tangentieIIement a
une paroi plane", These de Docteur des Sciences, Univ. Claude Bernard,
Lyon.

Tennekes, H., and LumLey, J. L., 1994, A Fig; quga In Tamaleng, The MIT
Press, Massachusetts, pp. 115.

Thole, K. A., 1992, High freestream turbulence effects on the transport of heat
and momentum, Ph.D. Dissertation, U.of Texas at Austin, pp. 50-61.

TSI Incorporated,1997, lFA n n T m An t r m
lnstrupp'pn Manual, Minnesota, pp. 1-1, 7-19

Utton, D. B., and Sheppard, M. A., 1985, The Detemrination of Convective Heat
Transfer Using a New Optical Technique, Proc. Int. Conf. For Heat and
Mass Transfer, Hemisphere, Washington DC. Vol.18, pp. 155-166

Zhao, W., and Tress, 0., 1997, Electrochemical Mass Transfer Measurements in
Rough Surface Pipe Flow: Geometricaliy Similar V-shaped Grooves,
lntemational Journal of Heat and Mass Transfer, Vol. 40, no. 12, pp. 2785-
2797.

212