‘L. :‘=i 1' ‘3
, ‘ ‘F‘uepg'L'EYS:
, Mb!

"1‘: 3951‘ L' ’r ' 1 4 Jim-pith!
{Mtgﬁg‘gw ;»° ; ~ -

“@5934 ‘iw
23$?!

1‘!
L r‘
i

‘ .
“_. .. w
n .

o

‘

.-
~««—< ,.- .

:1:
i::::u:;;::::_

«c "J

u...—

i '1‘. ..
{$15345

2‘ 3:3“
‘ ’19:

,.
..

ASA

nu

.-. -, .
- -n-
.__ "‘- —— A .
- m-~-“~_ _.
Hr-..
I'-

.h-‘
_. C
743" 0 Al»!
9-,..._.‘...

1

=9“. :3

u u
ﬁ'
V 733‘

.I..-..‘_.

an ._
I"? .

.

.-
h; —I-—-

,.-.....A. -m,
..._..._ .v——-<-. ,
4—.4.
”w-

I

.. ‘—‘.:‘.' ‘7
..
In.-.

2"?

.fw; .r'
I.‘ MM-
» n I

~mﬁﬁ—LEI-L

Vaﬂr MHz-315' A

2-- _

.c 01-4-

 

 

w

, ‘. .4
#23353;
C " 5

>- v‘
1, x' V

o

K“W

15 .gijE':
"1% :I. :44.‘ , '21 ,

anaerq‘qm’ﬁig; 7
figafﬁjhl 4'
wiggﬁﬁy. .
5M
“. 9 'r

‘3 1214i
gtz’ﬁ

¢-:
til}; 'i

ra-

.1 1
,.z!;3§2
n w;

3

f3 x: 4-.
} .1:

.7;
.lf!
9 ‘f

 

llﬂiilﬂ'liﬂiliﬂimilﬂil'liliiﬂiiﬁﬂlTI'ITJTIiﬂiﬁﬁII

/
’I" 3 1293 01565 4415

This is to certify that the

thesis entitled

Experimental Investigation of an Aerodynamic
Shroud for Cooling Fan Applications

presented by
Scott C. Morris

has been accepted towards fulﬁllment
of the requirements for

MS degree in Mechanical Engineering

 

Wiggins

Major professor

Date 8M“? 9?

0-7639 MS U is an Afﬁrmative Action/Equal Opportunity Institution

v—Vv—v—nr—‘ﬁ‘ _

V A r "’-f 1 »_.—«._._.___
r r r. -—o—

 

leBRAIEfY

Mlchlgan State
Unlverslty

 

 

 

PLACE IN RETURN BOX to romovo this checkout from your record.
TO AVOID FINES return on or before date duo.

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

J J 'J
J J J

L__[__J
V—JV—T—J

MSU lo An Afﬁrmative Action/Emu! Opportunity Inotltmon

 

 

 

 

 

 

 

 

 

 

 

 

EXPERIMENTAL INVESTIGATION OF AN AERODYNAMIC
SHROUD FOR COOLING FAN APPLICATIONS
By

Scott C. Morris

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

MASTER OF SCIENCE

Department of Mechanical Engineering

1997

ABSTRACT

An experimental study was performed on an automotive cooling fan with a
unique aerodynamic shroud. The development of the aeroshroud device was motivated
by the relatively high (2.5cm) tip clearance required in engine driven fan applications.
The shroud consists of a pressurized plenum and a Coanda attachment surface to deliver
an annular jet of high momentum air into the tip clearance region. The present
experimental data were acquired in the recently developed Automotive Cooling Fan
Research and Development (ACFRD) facility. The ACFRD facility provides the
capability to measure the integral ﬂow characteristics: pressure rise, volume ﬂow rate,
and fan input power, as well as detailed velocity measurements in the wake of the fan.

The functional dependence of the relevant dimensionless parameters was
derived using a boundary value problem approach. Analysis of the performance data
presented indicate that two of the independent variables can be combined into a single
parameter .which represents the relative power input into the aerodynamic shroud. The
data show that the shroud decreased performance and efﬁciency for the lower ﬂow
rates, and increased values at higher ﬂow rates.

Time mean ﬂow directions were measured with a wool tuﬁ. These data indicate
that reversed ﬂow exists in the tip clearance area at higher ﬂow rates. However, when
the aerodynamic shroud was activated, positive axial momentum was maintained in the
tip clearance area. Velocity measurements were taken using an X-array hot-wire probe
whose angular orientation was aligned with the mean ﬂow vector measured by the tuﬁ.
Time averaged and phase averaged velocities were computed from the recorded time
series. These data are presented at several operating conditions in order to characterize

the ﬂow field in the wake of the fan.

Acknowledgments

I would like to thank my advisor Dr. John Foss, who has provided me with the
oportunities and challenges that have allowed me to grow. Additionally, I would like to
thank Dr. Foss for the personal attention that is not just uncommon, but unfound in
today’s academia.

Secondly I would like to thank Tom Veling of the Ford Motor Company. It is Tom’s
technical guidance and insight that has allowed this project to be successful. Thanks also
go to John Pakkala for all the extra work that went in to the design of the ACFRD
facility.

I would like to thank my friends Doug Neal and Klaus Willenborg for the great
coffee breaks to procrastinate with. Also I would like to thank Eric Partyjak for the
conversations that fueled my passion not only for ﬂuid mechanics, but for science in
general. I would also like to thank Yvonne Chang for being there to keep me healthy and
positive through the past few years.

Finally, I need to thank my parents, who have given me the perfect blend of

guidance, and freedom to achieve my dreams.

iii

TABLE OF CONTENTS

 

 

 

 

 

 

LIST OF FIGURES VI
NOMENCALTURE X
1. INTRODUCTION 1
1.1 AERODYNAMIC SHROUD CONCEPT ............................................................................................... 1
1.2 PREVIOUS WORK: AERODYNAMIC SHROUD ................................................................................. 4
1.3 PREVIOUS WORK: VELOCITY MEASUREMENTS ............................................................................ 5

2. ANALYSIS 8
3. AERODYNAMIC SI-IROUD AND TEST FAN 11
3.1 PLANAR COANDA MODEL .......................................................................................................... l l
3.2 PROTOTYPE DESIGN AND PRESSURIZATION SYSTEM .................................................................. 12
3.3 TEST FAN ................................................................................................................................... 14

4. EXPERIMENTAL EQUIPTMENT AND TECHNIQUES 21
4.1 THE ACFRD FACILITY ............................................................................................................... 21
4.1.1 Overview of facility ............................................................................................................ 21

4.1.2 Traverse System ................................................................................................................. 22

4. I .3 Flow Rate Measurement .................................................................................................... 23
4.1.3.1 Details of the Apparatus ............................................................................................................. 24

4. l .3.2 Airﬂow Meter Calibration ......................................................................................................... 26

4.1.3.3 Test of Flow Rate Measurement Device .................................................................................... 27

4.1.4 Power Drive and Measurement ......................................................................................... 28

4.2 DATA ACQUISITION AND PROCESSING SYSTEM .......................................................................... 30
4.3 PRESSURE TRANSDUCERS ........................................................................................................... 30
4.4 TUFT PROBE MEASUREMENTS .................................................................................................... 31
4.5 VELOCITY FIELD MEASUREMENTS ............................................................................................. 31

iv

4.5.] Measurement Procedure ...............................................................................
4.5.2 X-array calibration .......................................................................................
4.5.3 Evaluation of Q and y ...................................................................................
4. 5. 4 Uncertainty Considerations ..........................................................................

4.5.5 Time Averaged and Phase Averaged Calculations .......................................

..................... 32

..................... 33

..................... 35

..................... 36

..................... 37

48

 

5. RESULTS AND DISCUSSION — INTEGRAL PARAMETERS

5.1 EXPERIMENTAL PROGRAM ...................................................................................

5.2 PERFORMANCE DATA ............................................................................................

5.3 EFFICIENCY DATA .................................................................................................

5.4 SMALL TIP CLEARANCE STUDY .............................................................................

6. RESULTS AND DISCUSSION - VELOCITY MEASUREMENTS

6.1 TUFT RESULTS .......................................................................................................
6.2 HOT-WIRE RESULTS ..............................................................................................
6. 2.1 Time Averaged Velocity Results ....................................................................
6.2.2 Phase Averaged Velocity Results ..................................................................

6.2.3 Phase Averaged Vorticity Results .................................................................

7. CONCLUSIONS AND VIABILITY PREDICTION S

...................... 48

..................... 48

..................... 50

..................... 50

59

 

 

REFERENCES

 

APPENDIX A COAN DA JET VELOCITY MEASUREMENTS

 

APPENDIX B PHASE AVERAGED KINETIC ENERGY

 

APPENDIX C EXIT FLOW ANGLE CALCULATIONS

 

..................... 59

..................... 62

..................... 63

..................... 64

..................... 67

124

127

129

131

138

LIST OF FIGURES

Figure 1.1: Schematic of fan and aerodynamic shroud ........................................................ 7
Figure 3.1 Detailed drawing of the planar test geometry ................................................... 15
Figure 3.2 Schematic representation of the ﬂow directions observed .............................. 15
Figure 3.3 Detail of shroud and plenum ........................................................................... 16
Figure 3.4 Schematic of shroud pressurization system ..................................................... 17
Figure 3.5 FordHaus facility used to calibrate elbow meter .............................................. 18
Figure 3.6 Elbow meter calibration data and ﬁt ................................................................ 19
Figure 3.7 Shroud pressure vs. volume to check ﬂow symmetry ..................................... 19
Figure 3.8 Schematic representation of test fan ................................................................ 20
Figure 4.18chematic of the ACF RD Facility ..................................................................... 39
Figure 4.2 Schematic of Traverse Used in the ACF RD .................................................... 40
Figure 4.3 Schematic of Probe Holder ............................................................................... 41
Figure 4.4 Schematic of Flow Metering Device ................................................................ 42
Figure 4.5 Calibration and Fit for South Side Flow Meter ................................................ 43
Figure 4.6 Calibration and Fit for North Side Flow Meter ................................................ 43
Figure 4.7 Flow Meter Test Data ....................................................................................... 44
Figure 4.8 Schematic of Power Measurement Circuitry ................................................... 44
Figure 4.9 Power Measurement Test Data ......................................................................... 45
Figure 4.10 Drawing of an X-array Probe ......................................................................... 45
Figure 4.11 Histogram of ﬂow angles for a properly aligned probe .................................. 46
Figure 4.12 Histogram of ﬂow angles for an improperly aligned probe .......................... 46
Figure 4.13 1] versus 7 with 5th order polynomial curve ﬁt ................................................ 47

vi

Figure 4.14 Example of statistical convergence of phase averaged data ........................... 47

Figure 5.1 Pressure rise vs. ﬂow rate data for selected shroud conditions ....................... 52
Figure 5.2 Contour plot of system performance (w) .......................................................... 53
Figure 5.3 Data ﬁt for x>0.009 ......................................................................................... 54
Figure 5.4 Efﬁciency data for selected x values (115,131 .0) ............................................... 54
Figure 5.5 Contour plot of fan/shroud efﬁciency (nsh,=l .0) .............................................. 55
Figure 5.6 Efﬁciency data for nsh,=0.70 ............................................................................ 56
Figure 5.7 Contour plot of fan/shroud efﬁciency (nsh,=0.70) ............................................ 57
Figure 5.8 Performance curve for fan with 1.0 mm tip clearance ..................................... 58
Figure 5.9 Efﬁciency curve for fan with 1.0mm tip clearance .......................................... 58
Figure 6.1 Performance data with test conditions for detailed study labeled .................... 70
Figure 6.2a Schematic of tuﬁ survey locations ................................................................. 71
Figure 6.3 Representative digital image of tuﬁ ................................................................ 72
Figure 6.4 Tuﬁ results from condition ‘A’ ....................................................................... 73
Figure 6.5 Tuft results from condition ‘B’ ......................................................................... 74
Figure 6.6 Tuﬁ results from condition ‘C’ ......................................................................... 75
Figure 6.7 Tuft results from condition ‘D’ ....................................................................... 76
Figure 6.8 Tuf’t results ﬁom condition ‘E’ ......................................................................... 77
Figure 6.9 Tuﬁ results from condition ‘F’ ........................................................................ 77
Figure 6.10 Tuﬁ results from condition ‘G’ ..................................................................... 78
Figure 6.11 Tuft results from condition ‘H’ ..................................................................... 79
Figure 6.12 Tuﬁ results from condition ‘I’ ....................................................................... 79
Figure 6.13 Tuﬁ results from condition ‘J’ ....................................................................... 80

vii

Figure 6.14 Tuﬁ results from condition ‘K’ ..................................................................... 81

Figure 6.15 Mesh of experimental data locations ............................................................. 82
Figure 6.16a Time averaged mean axial velocity normalized by U(, ................................. 83
Figure 6.17a Time averaged mean tangential velocity normalized by U07 ........................ 84
Figure 6.18a Time averaged mean radial velocity normalized by UC, ................................ 85
Figure 6.1% Time averaged mean axial velocity normalized by Utip ............................... 86
Figure 6.20a Time averaged mean tangential velocity normalized by Utip ....................... 87
Figure 6.21a Time averaged mean radial velocity normalized by Utip .............................. 88
Figure 6.22 Phase averaged mean axial velocity for condition ‘G’ .................................. 89
Figure 6.23 Phase averaged mean axial velocity for condition ‘H’ ................................... 90
Figure 6.24 Phase averaged mean axial velocity for condition ‘I’ .................................... 91
Figure 6.25 Phase averaged mean axial velocity for condition ‘J’ .................................... 92
Figure 6.26 Phase averaged mean axial velocity for condition ‘K’ ................................... 93
Figure 6.27 Phase averaged mean tangential velocity for condition ‘G’ .......................... 94
Figure 6.28 Phase averaged mean tangential velocity for condition ‘H’ .......................... 95
Figure 6.29 Phase averaged mean tangential velocity for condition ‘I’ ........................... 96
Figure 6.30 Phase averaged mean tangential velocity for condition ‘J’ ............................ 97
Figure 6.31 Phase averaged mean tangential velocity for condition ‘K’ ........................... 98
Figure 6.32 Phase averaged mean radial velocity for condition ‘G’ ................................. 99
Figure 6.33 Phase averaged mean radial velocity for condition ‘H’ ................................ 100
Figure 6.34 Phase averaged mean radial velocity for condition ‘I’ ................................. 101
Figure 6.35 Phase averaged mean radial velocity for condition ‘J’ ................................. 102
Figure 6.36 Phase averaged mean radial velocity for condition ‘K’ ................................ 103

viii

Figure 6.37 Phase averaged RMS of axial velocity for condition ‘G’ ............................. 104

Figure 6.38 Phase averaged RMS of axial velocity for condition ‘H’ ............................. 105
Figure 6.39 Phase averaged RMS of axial velocity for condition ‘I’ .............................. 106
Figure 6.40 Phase averaged RMS of axial velocity for condition ‘J’ .............................. 107
Figure 6.41 Phase averaged RMS axial velocity for condition ‘K’ ................................. 108
Figure 6.42 Phase averaged RMS of tangential velocity for condition ‘G’ .................... 109
Figure 6.43 Phase averaged RMS of tangential velocity for condition ‘H’ .................... 110
Figure 6.44 Phase averaged RMS of tangential velocity for condition ‘I’ ..................... 111
Figure 6.45 Phase averaged RMS of tangential velocity for condition ‘J’ ..................... 112
Figure 6.46 Phase averaged RMS of tangential velocity for condition ‘K’ ..................... 113
Figure 6.47 Phase averaged RMS of radial velocity for condition ‘G’ ........................... 114
Figure 6.48 Phase averaged RMS of radial velocity for condition ‘H’ ........................... 115
Figure 6.49 Phase averaged RMS of radial velocity for condition ‘I’ ............................. 116
Figure 6.50 Phase averaged RMS of radial velocity for condition ‘J’ ............................. 117
Figure 6.51 Phase averaged RMS of radial velocity for condition ‘K’ ........................... 118
Figure 6.52 Phase averaged axial vorticity for condition ‘G’ ......................................... 119
Figure 6.53 Phase averaged axial vorticity for condition ‘H’ .......................................... 120
Figure 6.54 Phase averaged axial vorticity for condition ‘I’ ........................................... 121
Figure 6.55 Phase averaged axial vorticity for condition ‘J’ ........................................... 122
Figure 6.56 Phase averaged axial vorticity for condition ‘K’ ................................ 123

ix

NOMENCALTURE

English

A,B,n

Ac]
A

flow

A'nozzle

D hub

P'fan
Pishr

P

system

12

losses

Q

Qfan

Calibration coefﬁcient of Collis and Williams relationship; see (4-11)
Area bounded by the shroud and fan tip; see (6-1)

Area bounded by the fan hub and the shroud.

Exit area of calibration nozzles; see (4-2).

Discharge coefﬁcient of calibration nozzles; see (4-2).

Diameter of the fan hub; see Figure 1.1.

Diameter of fan; see Figure 1.1.

Voltage measured from hot-wire anemometer.
Force measured for ﬂow rate measurement.

Gap thickness of Coanda jet; see Figure 1.1.

Voltage measured in power measurement; see section 4.1.4.
Calibration constant for ﬂow rate measurement; see (4—1)
Pressure rise across the fan (Pmim-Pm)

Shroud plenum pressure.

Shaﬁ power input to the fan.

Power input to the shroud; see (2-6)

Drive motor power including drive system losses; see (4-6).
Drive system losses measured without cooling fan; see (4-6).
Magnitude of the velocity vector measured by the hot-wire probe; see
Section 4.5.2

Approach ﬂow rate through the fan.

Qshr

u,v,w

Ux’Ur’UB

-.<

Tl sys

Tl shr

Flow rate through the shroud.

Radial coordinate measured from the fan axis. See Figure 3.8.

Radius of curvature of the Coanda jet; see Figure 3.1

Radius of fan (R=D/2); used in presentation of velocity data

Mean ﬂow velocity through the fan (Q,,,,/Anow).

Linear velocity of the fan tip (Uﬁp=nD§2).

Velocity components in probe coordinates.

Velocity components in laboratory coordinates.

Voltage measured in power measurement; See Section 4.1.4

Axial coordinate measured from the trailing edge of the blades. See Figure

3.8

Angular coordinate ﬁxed to the rotating fan. See Figure 3.8.
Angle of the hot wire with respect to the probe axis.

Angle of the exit ﬂow of the Coanda jet. See Figure 3.1
Dimensionless power input to the shroud. See (2-9).

Upper receiver gage pressure.

Shroud Plenum gage pressure.

Dimensionless ﬂow rate through the fan. See (2-1)

Angle of the velocity vector with respect to the probe axis.
Efﬁciency of fan and shroud system. See (2-5)

Efﬁciency of the shroud delivery system.

xi

A Dimensionless ﬂow rate through the shroud. See (24)

p Density of the working ﬂuid (air)

q; Dimensionless pressure rise across the fan. See (2-2)

0 Angular coordinate of the laboratory reference frame. See Figure 3.8.

Q Rotational speed of the cooling fan.

11 Ratio of hot-wire voltage to the voltage of the same wire at y=0. See (4-14).
§,§ Angular position of probe support apparatus. See Figure 4.3

Subscripts

11 Flow rate metering device on north side of facility.

5 Flow rate metering device on south side of facility.

+ Reference to the hot wire with [3>0

- Reference to the hot wire with B<0

Symbols

For a quantity a:

<a> Phase averaged quantity

(7 Mean value (time or phase averaged).

'ci’ RMS value (time or phase averaged).

xii

1. INTRODUCTION

1.1 Aerodynamic Shroud Concept

Engine mounted automotive cooling fans (primarily truck applications) require a
large tip clearance between the fan tip and the shroud. A 2.50m tip clearance is, typical
for a 45cm fan. As a consequence, their typical operating efﬁciencies are less than 20%.
An innovative concept1 has been developed to modify the current shroud in order to
enhance the overall efﬁciency of the fan-shroud system. This ‘aerodynamic shroud’
utilizes an annular pressurized plenum to deliver a Coanda attachment jet that serves to
“ﬁll” the gap between the blade tip and the physical surface of the shroud; see Figure 1.1
Note that the jet velocity proﬁle shown in Figure 1.1 is a schematic representation of
what would exist if the fan were not spinning. The use of the convex wall jet as opposed
to a straight air delivery was to prevent the jet apparatus from obstructing the area
upstream of the fan. Also, the entrainment properties of a convex wall jet could aid in
efﬁciency. The goal of the device is to realize a power savings that is greater than the
power required to pressurize the shroud’s plenum.

The variables used to describe typical fan investigations are the fan diameter (D),
rotating speed characterized by Udp, ﬂuid viscosity (U), pressure rise (APm), and ﬂow rate
(QM). The aerodynamic shroud introduces the following variables: shroud jet-gap (g),

shroud ﬂow rate (Qsm), and shroud plenum pressure:

 

‘ The original concept was conceived by Professor John Foss, Department of
Mechanical Engineering, Michigan State University, September 1994. Patents of this and

a related concept are pending.

AP

shr = Rim—plenum — Pr_appmm (1-1)
The use of Puppmch permits the presence of the upstream loss elements (e.g. heat
exchangers) to be represented in the measurements. In their absence, this value is that of
the undisturbed atmosphere (i.e., the present case). The ﬂow ﬁeld shown in Figure 1.1
can be considered to be a boundary value problem whose solution (the pressure and
velocity ﬁeld) is of interest. The only parameter in the governing (Navier-Stokes)
equations is the Reynolds number. Chapter 2 uses this approach to identify the functional
dependence of the dimensionless variables of interest.

Prior to the prototype construction of the aerodynamic shroud, a planar model was
built to observe the Coanda attachment phenomena for a range of possible geometries. A
constant speed centrifugal blower and throttle were used to provide pressurized air to the
plenum. Visualization results aided in the selection of the geometry to be used for the
prototype. An axisymmetric prototype was then constructed for testing with an
automotive cooling fan. A 1994 Ford Explorer fan was selected for the experiments.
Details of the planar model results, prototype design, air delivery system, and test fan are
given in Chapter 3.

The Automotive Cooling Fan Research and Development (ACFRD) Facility has
recently been constructed at the Turbulent Shear Flows Laboratory (TSFL) at Michigan
State University. This facility was constructed for investigations of the integral
performance characteristics (pressure rise, volume air ﬂow, and input power) as well as
discrete velocity measurements in the wake of automotive cooling fans and auxiliary

components. The aerodynamic shroud was mounted in the test section of the facility for

the experiments. Details of the ACFRD facility and its measurement capabilities are

given in Chapter 4. Procedures for the tuft and hot-wire measurements will also be
discussed in Chapter 4. Hot-wire calibration and data collection procedures are explained
in detail. An iterative algorithm was used to convert the recorded voltages from the x-
array probe to velocity time series. Equations are then presented to calculate the time
averaged and phase averaged velocity ﬁelds.

A series of experiments were conducted to measure the performance and efﬁciency
of the fan and aerodynamic shroud combination. The ﬂow rate through the shroud was
measured independently from the total system ﬂow rate such that the approach ﬂow rate
could be used in performance calculations. A range of jet gap heights (g), shroud
pressure conditions, and fan rotation speeds were tested. The dimensionless groups of
interest and their functional dependencies (derived in Chapter 2) were used to present the
data. An additional “collapse” of the data in addition to that expected from the boundary
value problem analysis was found to be possible. The results showed an increase in
performance and efﬁciency values for higher ﬂow rates, and degraded values for the
lower ﬂow rates. Details of the experiment and results of these “integral” characteristics
are given in Chapter 5. Additionally, a brief study was conducted to measure the
performance of the fan with a small tip clearance. The results shown in Section 5.4
demonstrate that the aerodynamic shroud with a 2.50m tip clearance can increase
performance and efﬁciency when compared to a shroud with a near zero tip clearance.

Several conditions of interest were identiﬁed for further study based on the results of
the integral data. Speciﬁcally, tuﬁ visualizations and velocity data were acquired in the
fan wake to characterize the ﬂow ﬁeld for these conditions. The tuft observations were

useful in identifying the time mean velocity vector directions in the wake of the fan.

These data were used to characterize the general attributes of the ﬂow ﬁeld and for setting
the angular orientation of the hot-wire probes. Probes using hot-wires in an x-array were
used to acquire time series data from which time averaged as well as phase averaged
velocity components were computed; see Chapter 6. Phase averaged vorticity values are
also reported.

The mass air ﬂow of an automotive cooling system depends on the fan performance,
and the resistance to airﬂow caused by the heat exchangers, grill, and engine components.
The data acquired in the ACFRD facility indicate that the operating condition will greatly
affect how the aerodynamic shroud affects the system performance. A brief study to
estimate the system resistance of a Ford Explorer was performed. Discussion of this as

well as predictions of the in-vehicle performance are given in Chapter 7.

1.2 Previous Work: Aerodynamic Shroud

The ﬂuid mechanical attributes of curved wall jets have been studied extensively; see
for example: Wilson and Goldstein (1976), Shakouchi et. a1 (1989), and El-Taher (1983).
The added entrainment of the convex wall jet has motivated the use of annular Coanda
ejectors for many applications (Ameri and Dybbs (1993)). To the author’s knowledge,
this is the ﬁrst work based on the use of an annular jet as a physical boundary condition
for any type of rotating machinery.

The experiments described herein were designed speciﬁcally for the automotive
cooling fan application. A 1994 Ford Explorer cooling fan was selected as the test fan for
the experiments. The design of efﬁcient automotive cooling fans is typically based on a
‘trial and error’ approach. Speciﬁcally, different blade designs are tested for a given

vehicle prototype until a fan is selected as having suitable performance characteristics. In

general, the shroud design is selected based on packaging constraints (i.e. available space)
in the underhood compartment. It is not a typical approach to explore the use of
innovative shroud designs to enhance system performance. It is worth noting that the
aerodynamic shroud concept could be applied to other applications. The data to be
presented in Chapter 5 indicate that applications in which a small tip clearance is allowed

could beneﬁt from a larger tip clearance with the aerodynamic shroud.

1.3 Previous Work: Velocity Measurements

The majority of the past research involving automotive cooling fans has been
restricted to the measurement of the integral quantities such as pressure rise, volume air
ﬂow, and input power. Detailed measurements in the wake of cooling fans have been
acquired by Van Houten et. a1 (1993) using a one component LDV system. A number of
investigators have used hot-wire techniques to measure the ﬂow ﬁeld downstream of
other turbomachinery such as compressors, turbines, and axial fans; see e.g. Kuroumaru
et. a1 (1982), Hirsch et. al (1977), and Lakshminarayana (1981). The ﬂows of interest are
difﬁcult to measure because of the inherent unsteadiness and the relatively high
turbulence levels common to these devices. Most of the available literature describes
geometries in which an axial ﬂow machine is placed into a cylindrical wind tunnel for the
measurement program. It is typical in automotive cooling fan research Baranski (1974)
to provide the fan with quiescent inlet conditions, and a large receiver as the outlet.
These “open” boundary conditions allow the fan to create large radial velocity
components that would be bounded with the cylindrical passage of a wind trmnel. In
order to acquire velocity data in these types of ﬂows, the hot-wire probe must be oriented

such that the probe’s axis is aligned with the mean velocity vector. To the best of the

author’s knowledge the measurement techniques developed for this thesis have not

previously been used in the study of automotive cooling fans.

Inlet Flow Direction i Jet Gap (9)

\
\
\
\
\
\

\\\\\\\\\\\\

FL

Jet Thickness
\

 
   

 

Test Fan

l 4:] [4— Tip Clearance

>l D 4Dtip

hub

 

 

 

 

 

 

 

Figure 1.1: Schematic of fan and aerodynamic shroud

2. ANALYSIS

It is useful to consider the cooling fan and the aerodynamic shroud to be the internal
elements of a boundary value problem whose upstream boundary is the laboratory
“atmosphere” and whose downstream boundary is the receiver (see Chapter 4.1 for details
of the experimental facility). The governing parameters for the ﬂow ﬁeld can then be
reliably extracted as those quantities that would inﬂuence the solution: I7(x,r,0,t) and

P(X,r,9,t). Proper scaling quantities are required to characterize this ﬂow ﬁeld; these

are selected to be:
i) the tip speed of the fan blade (Udp) as the reference velocity in the problem,
ii) the fan diameter (D) as the length scale of the problem, and
iii) the kinematic viscosity (U) of the working ﬂuid (air).

The boundary value problem can therefore be speciﬁed in terms of its boundary
condition parameters and its Reynolds number (UﬁpD/U). It is typical in fan studies
(Baranski (1974)) to not include the Reynolds number as a signiﬁcant parameter. The
results shown in Chapter 5 conﬁrm the insensitivity of the present data to the magnitude
of the Reynolds number; hence it will not be considered in the set of governing
parameters given below.

The boundary value problem described above in general terms for the ﬂow ﬁeld of
Figure 1.1 will be made speciﬁc by identifying the boundary conditions, i.e. the
independent parameters. The approach ﬂow rate (Q,,,,) can by expressed as:

«I = .3"? (2“)

up

 

 

where U0 = 31:“ , and Aﬂow = %(Df,,, — 1);”) represents the annular area between the shroud

and fan hub. The pressure of the shroud plenum (APshJ, referenced to the tip speed as

AP

shr

pUip , characterizes the “driving potential” for the shroud ﬂow. The height (g) of the

Coanda jet opening characterizes the shroud (given its basic aerodynamic shape as shown

in Figure 1.1). Its parametric designation is (g/D).

The dependent integral parameters of interest are: fan pressure rise, shroud volume
ﬂow rate, and system efﬁciency. In dimensionless form, the typical expression for the

pressure rise across a conventional fan (APfan):

APfan __ (2'2)
‘é‘pUZ —W(¢)

tip

 

W E
is modiﬁed to:

AA," (2-3)
W = W (Ii 7,7727%)

The shroud ﬂow rate (Qshr) and system efﬁciency (115,5) can be respectively described as

 

 

_ 9.. M... g (2-4)
A = LIMP/:IIUW = A(¢, PUIIII’ ,3)
and
_ APanQan APE," g (2-5)
1105 = 'Pfan'PRI-hr — T1915 d), pUjp ’ D °

where the power required to drive the fan is Pm and that to pressurize the shroud is
deﬁned as

P E APrthslIr (2-6)

shr n5,”

 

10

and ﬁsh, is the efﬁciency of the shroud delivery system. The effect of nsh, on the overall
system efﬁciency depends on the relative size of P,,,, with respect to PM. Chapter 5 will
use two values nsh,=l .0 and 0.7 in the presentation of efﬁciency data.

Using equation ( 2-4), equations (2-3) and ( 2-5) can be rewritten in terms of their

functional dependence as

 

A2,, (2-7)
W = w (¢,A, pug”)
and
AA,” (2-8)
nsys : nsys (d) ’ A ’m)

respectively. Examination of the experimental data that will be presented Chapter 5 has

revealed a useful simpliﬁcation of equations (2-7) and (2-8). Speciﬁcally, the variables

 

AP , , . . . . . . .
—pU‘§ and ,1ij p can be comblned to a srngle dimensronless variable usrng therr product:
up an I!

x _=_ A. £1. = _A_P_9_ (2-9)
p Ullp DA [In W Uri];

3. AERODYNAMIC SHROUD AND TEST FAN

This chapter describes the geometric details of the prototype aerodynamic shroud and
how the design parameters were selected. The ﬁrst section (3.1) describes an initial study
using a planar Coanda jet to observe the qualitative attributes of the ﬂow ﬁeld. The
following section (3.2) describes the axisymmetric prototype and air delivery system

respectively. Section 3.3 shows the details of the test fan used in the experiments.

3.1. Planar Coanda Model

A planar Coanda jet was built for initial study due to the anticipated expense of the
aerodynamic shroud prototype (which employs an annular jet). The purpose of the planar
model was to test the ability of the jet to remain attached to the convex surface for a range
of jet Reynolds numbers (Rej,,=(UJ-w/v) where U]. is the reference velocity derived from
the plenum pressure of the shroud.) Given the relatively large ratio of the shroud radius to
local jet thickness (Rm/Own), it was decided that the planar model should yield
satisfactory qualitative results when compared to the prototype. A drawing of the
geometry is shown in Figure 3.1 . A Cincinnati centrifugal blower and throttle plate
were used to pressurize the plenum. The jet width (w) was varied by repositioning the
upper surface of the plenum chamber. Plexiglas was used for the side wall in the area of
the jet for ﬂow visualization.

The basic aerodynamic shape consists of a 180 degree convex wall, followed by a
Scm straight section; see Figure 3.1 . The intent of the straight section was to provide a
constant tip clearance for an axial distance equal to the chord of the blade. The end of the

straight section is characterized by a sharp 90 degree edge. Two radii of curvature (rjet =

11

12

4.60 and 3.81 cm) were tested for the convex wall section. Experimental data reported
Shakouchi et. a1 (1989) have shown 2D Coanda jets with similar radii of curvature will
remain attached for the range of Reynolds numbers (400<Reje,<10,000) of interest. The
upper surface of the jet consisted of a curved “lip” with a sharp edge to deﬁne the jet exit.
Two values of [i = O and 30 degrees were used to observe the effect of the location of the
jet exit.

A tuﬁ was used to observe the ﬂow ﬁeld. The tuft consisted of a short (zlSmm)
segment of white yarn held in the ﬂow using an aluminum shaft, sewing needle, and light
thread (details of the tuﬁ probe are given in Section 4.4) Tests were conducted for a wide
range of plenum pressures (20<(Pp,enum-Pm,,,,<1000 Pa) and jet widths (1.0 < w < 4.0
mm.) The visualization results indicated that the jet remained attached for the radii
values and the Reynolds numbers tested. The ﬂow ﬁeld presented did not visually appear

to be sensitive to Re. and the ﬂow remained attached to the convex wall for all the

jet’
conditions tested. Figure 3.2 Shows a schematic of the vector directions observed for
rje,=4.6cm, Reje, = 2500, and B=30°.

The dimensions of the Coanda surface shown in Figure 3.1 were selected for the
axisymmetric prototype based on the above visualizations and machining constraints.

Speciﬁcally, the planar observations indicate that the selected cross sectional geometry

should provide the cooling fan an attached, non-reversing ﬂow in the tip clearance area.

3.2 Prototype Design and Pressurization System

A detailed drawing of the prototype shroud and plenum is shown in Figure 3.3. The
inner diameter of the Shroud was 508mm (20”); cross sectional dimensions are shown in

Figure 3.1. The shroud was machined as two pieces which allowed the setting of the jet

13

gap. Both pieces were machined from RenShape #440. The upper surface of the plenum
was constructed from by a V2” stock plywood plate which was mounted to three segments
of ‘/2” threaded rod. The upper shroud piece was fastened to the under side of this plate.
The jet gap was then varied by adjusting the three support rods to position the upper piece
of the shroud; a drill bit was used to measure the gap size (g). The side walls of the
plenum were also constructed from V2” plywood. A 3%” plywood cover plate was
mounted to the test section of the ACFRD, which acted as the lower surface of the
plenum, and rigidly supported the lower piece of the shroud.

A schematic of the pressurization system used to supply air to the shroud is shown in
Figure 3.4. The Cincinnati blower (noted above) was used to move the air from the
laboratory, through the system of PVC pipes, and into the shroud plentun. The piping
system terminates at the four inlets to the shroud plenum in order to provide a symmetric
delivery of low momentum air into the plenum. A throttle plate upstream of the blower
and an auxiliary outlet allowed the ﬂow rate into the shroud to be controlled. The
purpose of the auxiliary output was to maintain a relatively high volume air ﬂow through
the blower to prevent viscous heating of the air before entering the shroud plenum.

The volume ﬂow rate into the shroud was measured by an “elbow meter”. As shown
in Figure 3.4, pressure measurements on the inner and outer surfaces of a bend in the
PVC delivery system allowed the ﬂow rate to be inferred. A calibration of the device
was conducted using the F ordHaus facility of the TSF L, see Figure 3-5. In this process
a calibrated venturi nozzle was used to measure the ﬂow rate through the system. Two

Validyne pressure transducers and a PC based A/D system (see Chapter 4) were used to

14

measure the pressure differences of both the elbow meter and the venturi nozzle. Figure

3.6 shows the calibration data and parabolic curve ﬁt. The resulting transfer function:

Q5,” = 2(5.348)AP°'S (3-1)

shr

was used to infer the volume air ﬂow rate through the shroud. Note the factor of 2 used
in equation (3-1) because the elbow meter is located downstream of a “split” in the
delivery system as shown in Figure 3.4. The equality of the ﬂow in the two branches was
checked by measuring APshr as a function of the ﬂow rate Qshr with the cooling fan
removed. This test was repeated with the elbow meter moved to the other “branch” of the
delivery system. These data are shown in Figure 3.7. These data indicate that the ﬂow
rates through both the ‘east’ and ‘west’ branches of the ﬂow system are equal to within
the precision of the measurement.

A brief study was performed to characterize the velocity proﬁle of the annular jet
with the cooling fan removed. The results are useful in characterizing the shroud, but do
not necessarily represent the boundary conditions of the cooling fan. These data are

given in Appendix A.

3.3 Test Fan

The selected cooling fan used in the experiments was that of a 1994 Ford Explorer.
A representation of the 457 mm diameter test fan is shown in Figure 3.8. The 10 blades
have an 81 mm chord with a 41° pitch (i.e. angle of the chord with respect to the plane of
rotation.) The hub to tip ratio was 0.472. The blade thickness was approximately 2mm
over the span. The solidity (chord divided by blade spacing, see Lakshminarayana, 1996)

at blade tip was 0.349.

 

15

 

33°

l':3.94

 

Cincinnati

From
Blower

9
6—
e.
e
<—

Plenum

 

 

 

 

l——4.eo—.l

Figure 3.1 Detailed drawing of the planar test geometry

    

 

  

 

 

 

 

 

 

'"> Entroined Flow

—>Moin Flow

  

Figure 3.2 Schematic representation of the ﬂow directions observed

l6

 

 

 

Flow

lnlets

Support Bolts

 

ﬂ

 

>9

Secﬁon
A—A

U

0

Flow

lnlets

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 3.3 Detail of shroud and plenum

 

><—

17

 

 

\Elbow Meter
Pressure Tops

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Auxiliary Exit

 

 

 

 

 

Inlet
Throttle

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Cincinnati Blower

Figure 3.4 Schematic of Shroud pressurization system

Spencer Blower

18

 

 

Control Valve
/

 

{l

 

Venturi Meter

l/

Elbow Meter

 

 

 

 

 

 

 

 

 

 

 

 

._l

 

 

 

 

\k_|_//

|
l
I
[1

/i>'
/‘D

in

Plenum

/

Flow Conditioners

,r/ Laboratory Air //

1m

 

 

 

GK
6‘“

<}~\

 

 

I
l
I
l
|
’]

 

Figure 3.5 FordHaus facility used to calibrate elbow meter

Flow Rate (M3/min)

700

l)shr (P3) -

600

500

400

300

200

100

19

 

 

 

 

 

 

 

7.0 , , , r , T T
6.0 - _
5.0 - -
4.0 - ‘1
3.0 — ' -

_ 0.5
2.0 _. - Q -5.348* (P0-Pi) _
‘l.0 _.
0.0 l l 'L L l l l I
0.00 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00
PO "Pi (cm H20)
Figure 3.6 Elbow meter calibration data and ﬁt
I I l l I l l I
9

Data Fit

0 East Meter

9 West Meter

1 . l
0.5 1.0 1.5 2.0 2.5 3.0 3.5

 

4.0

Qshr (m3/min)

Figure 3.7 Shroud pressure vs. volume to check ﬂow symmetry

20

 

 

 

/\/

 

// /

t

Figure 3.8 Schematic represe

 

 

of test fan

ntation

4. EXPERIMENTAL EQUIPTMENT AND TECHNIQUES

4.1 The ACFRD Facility

The recently constructed Automotive Cooling Fan Research and Development
(ACF RD) facility was used for the ﬂow system in this study. The ACFRD was designed
to measure both the integral ﬂow quantities (volume ﬂow, pressure rise, and input power)
and velocity vectors in the wake of “axial” ﬂow fans. The facility was designed by the
author, Professor John Foss of the TSFL, and Mr. John Pakkala2 of Troy Design Inc,

Lansing, Mi.

4.1.1 Overview of facility

The ACFRD facility is shown in Figure 4.1. The test fan (B) and prime mover (I) are
responsible for the ﬂow from the laboratory, through the facility, and back to the
laboratory. A pressure transducer (0-747 Pa) was used to measure the differential
pressure ﬁ'om the atmosphere to the upper receiver (i.e. the pressure rise across the fan:

AP=P ch) using the tap located at point (C). The air is moved from the

-P
upper receiver T-approa

upper receiver (F) through two ﬂow metering devices (L-M-P) to the lower receiver (T).
The prime mover (I, centrifugal fan: Chicago Blower Co. SQA-36.5) moved the air out of
the lower receiver and past a throttle plate. The inlet to the Chicago blower was located
at (S) in the lower receiver. The throttle plate was used to control the ﬂow rate through

the system.

 

2 Mr. John Pakkala is currently employed at Capital Machine Design, Lansing, Mi.

21

22

The test fan was driven by a vertical shaft, belt drive, and DC. motor (J) as shown
Figure 4.1. A Reliance Electric 11.2 kW (15hp) motor with feedback speed control was
used for this drive. The motor speed range was 250 - 1250 rpm. A 1:3 diameter ratio
was used for the pulley system to increase the fan speed range to the desired range: 750 -
3750 rpm. The test fan was cantilevered 61.0cm above the shaft supports (D) to avoid
aerodynamic blockage of the drive system in the fan wake. Note that the upper 30.5cm of
the shaft can be removed for the study of fans which are driven with a built in electric
motor. An optical encoder (G) was comprised of a notched disk (mounted on the drive
shaft) and an infra-red interrupt circuit; this provided an angular position reference of the
fan. The output signal was also used with a frequency counter to accurately set the fan
speed. The resolution of the frequency meter was 0.01 hz, corresponding to 0.6 rpm. The

feedback speed control typically maintained the speciﬁed speed to within 120.04 hz (2.4
rpm)-
4.l.2 Traverse System

A computer controlled traverse was designed and assembled for positioning of the
tuft and hot-wire probes in the wake of the fan. A schematic of the traverse is shown in
Figure 4.2. All positioning systems were driven by stepping motors controlled remotely
from an Intel 486-66 computer. Azimuthal positioning about the fan’s axis was provided
by a 1.52m diameter steel ring (A). The ring was driven by a stepping motor through a
6/1 pulley system and a 100/1 worm gear box (not shown in Figure 4.2). Two horizontal
linear slides (C), (Velmex Inc. #MB4024BJ-S4-20) were mounted to the ring (180°
spacing) to provide 51cm of horizontal travel. A 40.5cm vertical slide (E) (V elmex Inc. #

MB4024BJ-S4-16) was ﬁxed to each of the two horizontal slides. All four slides used a

23

20 thread per inch lead screw. The manufacturer quoted repeatability of the positioning
was 0.05mm/meter. A probe support ﬁxture which allowed the angular orientation of the
probe to be manually set was mounted to one of the aluminum support arms (H); see
Figure 4.3. These support arms were given additional vertical strength from linear
bearings (I) riding on steel rods (F). The vertical shafts (G) were used to hold the
horizontal rods (F) in place. This conﬁguration provided the two arms with linked
vertical motion and independent radial motion. All stepping motors were connected to
Velmex #NF90-3 controllers that provided the computer interface for the system. The
computer commands used to move the probes were called from subroutines written in the

‘C’ language. These were incorporated into the data acquisition program.

4.1.3 Flow Rate Measurement

A new and unique measurement device has been assembled and tested to measure the
volume air ﬂow rate through the ACFRD facility. The wide variety of fan and
component geometries to be tested in the facility can be expected to signiﬁcantly alter the
velocity proﬁle entering the metering nozzles. For this reason a point measurement of a
characteristic velocity in the nozzle outlet, such as that using Pitot and static pressure
measurements, would not provide a reliable experimental technique for the ﬂow rate
evaluation. Also, these ﬂow ﬁeld irregularities negate the option of using a simple nozzle
discharge coefﬁcient. The use of thick screens could have been used to “regularize” the
velocity proﬁle; however, the pressure drop caused by the screens would have limited the
operating capabilities of the facility.

An integral measurement technique was therefore implemented using the device

shown in Figure 4.4. Given that the downward ﬂow at the exit of the nozzle (L) is turned

24

into the horizontal direction at the exit of the turning vane (M), the resultant moment of
momentum ﬂux can be used as the quantity that is indicative of the mass ﬂow rate.
Speciﬁcally, this net ﬂux was balanced by the moment of the force at point (P) acting

with respect to the hinge at (R). The magnitude of this force: F P was measured using a
ring type full bridge force transducer. Given the expression for the moment of
momentum ﬂux for a control volume (see Figure 4.2) that spans the nozzle exit to the exit

of the turning passage:

7H x F, = M? x I7](I7'ii)dA a K(rR_P)(m2/pA) = 1m? (41)

one can compute the m given the measured force (F P) and the coefﬁcient k' . Note that
K is dimensionless and that k’ has dimensions (lengthz/mass). Evaluation of k' will be
discussed in Section 4.1.3.2. It is relevant to note that the inﬂow to the control volume
contributes to the counter clockwise moment-of-momentum ﬂux (since 17-17 < 0 and
F x J7 is clockwise.) The exit ﬂow also contributes to the net counter clockwise flux
(since 1742' > 0 and F x I7 is counter clockwise).

The relative insensitivity of the measurement technique to the inﬂow characteristics
from the upper receiver reﬂects the conditions that the integral, and not point-wise,
attributes of the velocity ﬁeld are represented in the transfer function. Also, the
acceleration of the ﬂow from the nozzle exit to the control volume exit will lead to a

velocity proﬁle at (P) that is relatively insensitive to the upstream conditions.

4.1.3.1 Details of the Apparatus

Two identical measurement devices exist in the fan testing facility as Shown in

Figure 4.1. This design allowed a symmetric delivery of air ﬁom the upper receiver to

25

the lower receiver, as well as to provide a redundant measurement of the ﬂow rate (see
section on calibration).

A set of three delivery nozzles directs the air vertically downward from the upper
receiver toward the turning vane for each of the two metering passages. The nozzle
assemblies consist of a screen entrance and a one sided contraction as shown in Figure
4.4. The former consists of a ‘/4 circle (r=0.406 m) of coarse wire mesh to support a
cheese cloth ﬁlter. Thin ﬁlters were used to prevent wind drafts of the upper receiver
from inﬂuencing the ﬂow in the nozzles. An elliptical shaped contraction with a 3/2 axis
ratio was used to provide a 4/1 contraction in ﬂow area. The exit ﬂow area for both sets
of nozzles was 0.310 m2.

The measurement device consists of an aluminum sheet rolled to a 0.406m radius.
The shape of the curved surface is rigidly held by plywood side walls. This assembly
(M) is supported on one side by two 0.20 m wide knife edge pivots (R); this design
provides less friction moment than a bearing support. The outﬂow side of the unit was
supported by a counter balance (N) and force transducer (P). The counter balance was
also mounted on a knife edge support as shown in Figure 4.4. Adjustable steel plates
were used to offset the weight of the turning vane assembly. A ring type full bridge force
transducer was connected between a rigid beam and the turning vane to measure the
reaction force of the air ﬂow. The moment was then inferred from this force. The ring
was constructed from aluminum with the dimensions: 51mm in diameter, and 1.0mm
thick. Four 350 ohm stain gages were fastened to the ring and connected to a bridge
circuit using an instrumentation ampliﬁer. Adjustment screws were used to set the

maximum deﬂection of the device to limit the maximum strain of the transducer.

26

4.1.3.2 Airﬂow Meter Calibration

Calibration of the force transducer outputs was necessary to compute the ﬂow rate
from the two measured voltages. A set of four metering nozzles were placed at the roof
level of the facility in place of the test fan. The mass air ﬂow rate through the system was

determined from the equation:

 

<4-2)

atm receiver )

2
m = p X N X Anozzle X Cd\/_p-(P

where N is the number of calibration nozzles used, and Cd is a known discharge
coefﬁcient of the nozzles. Initial investigations indicated that the interaction of the two
metering systems when operated together caused biased results when compared to data
taken with the independent operation of the metering systems. Therefore, both force
transducers were calibrated Simultaneously using the total ﬂow rate. This technique
assumes that the fraction of the total ﬂow rate in each of the measurement devices during
the calibration process is identical to that in an actual fan test. Data to be presented in
Section 4.1.3.3 Show this to be an acceptable assumption. As a result, two independent
estimates of the total ﬂow (Q5) and (On) rate are obtained (the subscripts ‘n’ and ‘5’ refer
to the metering device on the south and north side of the ACFRD facility). The actual

ﬂow rate is assumed to be the mean of these:

Q. + Q. (4-3)

Qmeas ured _ 2

A range of ﬂow rates was established using the auxiliary fan and throttle combination to
obtain the calibration data. The voltage output from the two force transducers was
measured using a PC based A/D system described in Section 4.2. Assuming that the

force transducers are linear (the output voltage is proportional to the applied force), it

27

follows ﬁom equation (4-1) that the ﬂow rate is proportional to the square root of the

voltage. Time series data acquired were then averaged and curve ﬁt to the functions:

Q" = 3.165./E,, + 0.0531 (4-4)
Q, = 2.9781/E3 — 0.0094 (4-5)

where E“ and Es represent the average voltages measured from the transducers on the
north and south sides of the facility respectively; Q is calculated in units of m3/s. The

calibration data and linear curve ﬁt of the data is Shown in Figure 4.5 and Figure 4.6.

4.1.3.3 Test of Flow Rate Measurement Device

A test fan with a well established characteristic curve (a cooling fan from a 1994
Ford Explorer) was used to validate the technique and calibration. The performance was
measured using the following procedure:

i) the fan rotational speed is set to a constant value,
ii) the back pressure created by the fan was varied with the auxiliary fan and throttle
plate,
iii) time series data for the force transducer outputs as well as the pressure of the upper
receiver were recorded using the A/D system.
The time series data were averaged for the T = 10 seconds of data taken at 1,000 Hz.
Equation (4-3) was then used to infer the ﬂow rate. Performance data were acquired at
several fan speeds and made dimensionless using equation (2.2). It was decided that
scaling performance data would provide a quantitative check on both the repeatability and
accuracy of the ﬂow metering device. Speciﬁcally, the data would not collapse properly

if the measurement device was not accurately measuring the ﬂow rate. Two performance

28

curves were acquired for speeds of 1400 and 1600 rpm. Data supplied by the fan
manufacturer was taken at 1500 rpm. These results are presented in Figure 4.7. The data
indicate that the metering device is working properly. The differences between the two
data sets obtained in the ACF RD are quite small (a 2% difference in ﬂow rate at a given
pressure rise was typical). Larger deviations observed over the middle ﬂow rate range
(0.15<¢<0.18) are suspected to be a result of the fan stall phenomena being sensitive to
the Reynolds number (UH-pran/D). The data taken in the ACFRD facility agreed in general
with the manufacturers data. However, larger deviations are observed. Typical
agreement for the larger ﬂow rates was within 7%. The large differences noted for the
smallest ﬂow rates are suspected to be a result of slight differences in geometrical
boundary conditions used and not an error in the ﬂow metering device. Speciﬁcally, it
has been observed that the performance characteristics of cooling fans are extremely
sensitive to the geometry and location of the shroud due to the radial outﬂow of the fan at
low ﬂow rates; see Chapter 6 and Baranski (1977).

The proper collapse of the scaled data indicates that ﬂow rate measurement device
should provide accurate results for the ﬂow range of interest. The uncertainty in the
performance data presented in Chapter 5 is estimated to be i2.0% of the ﬂow rate for a

given fan pressure.

4.1.4 Power Drive and Measurement

The power input to the fan was measured using the electrical circuitry shown
schematically in Figure 4.8. This device provides voltage signals ‘v’ and ‘i’ which are
proportional to the drive motor voltage and current respectively. The ‘v’ signal was made

proportional to the motor voltage signal by connecting a 100:1 voltage divider to the

29

motor’s power lines. The ‘i’ signal was made proportional to the motor current using a
hall effect current meter (F.W. Bell Model CL-lOO). To measure the input power, time
series data of signals ‘v’ and ‘i’ were acquired at 1000 hz for 10 seconds. The power

delivered to the fan was then computed from:

.. " 4-6
Pf“ = klﬁyﬂem —*)losses]: k[z(vi)i “2(140 1.01.], ( )
i=1

i=1

where n is the number of samples taken (10,000). The quantity (Plow) represents the
power losses of the drive system which are a function of the fan rotation speed. This was
recorded before and after each experiment by removing the cooling fan and recording the
motor power as above. These two estimates were averaged for use in equation (4-6).
Initial investigations indicated the presence of transient drive losses for approximately 15
minutes when the facility was “cold”; therefore, the drive system was operated for 20
minutes prior to an experiment to bring the drive motor and shaft bearings to thermal
equilibrium.

The constant k was determined by comparing data from the test fan’s manufacturer to the

data acquired in the ACFRD. Data were acquired for rotational speeds of Q=l400 and
1600 rpm. The manufacturer’s data were acquired at Q=1500 rpm. For comparison, the

data acquired at 1400 and 1600 rpm were scaled according to the equation:
3 -
ﬁan<0=1500)=e..(0)-(EP). (4 7)

where N represents the fan speed at which the data were acquired. Note that equation (4-
7) can be found in Baranski (1977). A least squares approach was then used to solve for
k. The value was found to be 0.26 to obtain units of kilowatts. Figure 4.9 shows the

data taken for this test plotted as a function of the ﬂow rate. The largest difference in

30

power between the ACFRD data and the manufacturers data is 3% for Qfan>0.9 m’/s.
Differences in power for the lower ﬂow rates range from 3 - 9%. The largest difference
realized between the two data sets acquired in the ACF RD facility was 3.2%. The larger
differences observed between the manufacturers data at the lower ﬂow rates is believed to
be a result of the sensitivity of the ﬂow ﬁeld to slight differences in boundary conditions.
The overall uncertainty of the measurement technique is estimated to be 3% of the actual

power.

4.2 Data Acquisition and Processing System

An Analogic Fast-16 plugin board and an Intel 486-66 were used to acquire data.
The maximum possible data rate was 1 Mhz. An 8 channel “sample-and-hold” was used
for simultaneous sampling. The resolution of the A/D conversion was 16 bits, with a
range of i10 volts. The electrical noise of the system (tested by connecting a short circuit
to the input) was typically 0.3 mv, or 1 bit. The A/D card contained 10" samples of
onboard memory. If more than 10" data samples in a given time series were required,
independent “bursts” of data were sequentially sampled. All data were transferred to a

DEC ALPHA orXP-l 50 computer system for processing.

4.3 Pressure Transducers

Three Validyne DP15-20 pressure transducers were used to measure the AP,,,,, APshr, and
Qshr (as described in Section 3.2). The linear calibration constants for these were 0.842,
2.91, and 0.068 respectively. These calibrations were checked before and after all

experimental data were acquired and found to be invariant to within 1%.

31

4.4 Tuft Probe Measurements

The tuft consisted of a 15mm segment of white yarn supported by a thin thread
which is held in place be a sewing needle. A small hole was drilled into a %” aluminum
shaft such that the needle could be press ﬁt into the shaft. This design was used to
minimize the effect of the tuft support on the local ﬂow characteristics.

Images of the tuft were acquired using a Dycam DC-10 digital camera to document
the time mean angle of the ﬂow. The camera had 24 bit color resolution, with a
maximrun of 640x480 pixel resolution. A black background was used to hold white
arrows for angular reference in the photographs. The angular excursion limits of the tuft
were measured using digital image analysis software. Speciﬁcally, the reference angles
that were placed in the background allowed the two angles, corresponding to the extremes
of the tuft excursion, to be measured with respect to the (r, 0, x) coordinate system. The
time mean ﬂow direction was assumed to be the angle centered between the excursions.

The ability of the tuft probe to provide an accurate estimate of the mean ﬂow angle
varies depending on the ﬂow ﬁeld. Speciﬁcally, extremely high turbulence levels
ii / (7 >10 and highly reversing ﬂows can cause unreliable results. However, for ﬂow
conditions where the velocity is nominally unidirectional, the tuft provides an accurate

measure of the time mean ﬂow angle often to within 5 degrees.

4.5 Velocity Field Measurements

This section presents the measurement procedure, calibration method, and data
processing algorithms used in the acquisition of quantitative velocity data in the wake of

the cooling fan and aerodynamic shroud. A probe using two hot-wires conﬁgured in an

32

X-array was used to acquire these data. The probes used were fabricated in the TSFL’.
The X-array consisted of two tungsten wires mounted 1mm apart and at a nominal angle
of 45° with respect to the probe axis. The wire consisted of a 3mm long 5pm diameter
wire with a 1mm active region. Copper plating was used to cover the 1mm of wire on
either side of the sensor. A schematic of the probe used is shown in Figure 4.10.

The data were taken using two DISA 55m10 anemometers. The electrical noise
levels were typically 2 mvolts RMS. The frequency response was always in excess of 30
Khz for a ﬂow speed of 20 m/s. This permitted ﬂow properties that are of the order
0.67mm to be evaluated. All hot-wire data were temperature compensated to account for
ambient changes in the laboratory air temperature. A thermistor with 2.03 K/Kohm
sensitivity was used to measure the air temperature during calibration and data acquisition

procedures.

4.5.1 Measurement Procedure

The angular orientation of the probe was aligned with the mean ﬂow directions
measured from the tuft observations. Histograms of the incident ﬂow angle (7) with
respect to the probe’s axis were calculated for each of the spatial locations measured. If
the probe was correctly aligned with the mean ﬂow, the histogram would reveal a
‘centered’ proﬁle as Shown in Figure 4-11. If the probe was improperly aligned, the
histogram would be shifted; see e.g. Figure 4.12. The magnitude of the shift was then

used to re-align the probe, and the data were retaken.

 

’ Turbulent Shear Flows Laboratory, Michigan State University.

33

Stochastic values for the three components of the velocity vector were acquired by
rotating the X-array probe about its axis for each measurement. This technique provides
a single measurement of V and W (velocity components normal to the probe axis) and a
redundant measurement of ii (velocity component aligned with the probe axis). The
angular orientation of the probe (i.e., the Q and g angles of Figure 4.3) were used to
compute U,, U,, and U6 from the u, v, and w components measured. This calculation was
simpliﬁed by the fact that §=0 was used for the measurements reported herein.
Speciﬁcally, the velocity vector expressed in the (r, 0, x) coordinates are related to the

probe coordinates by:

U,r = \lu2 +v2 sin(§ +y) (4‘8)
U6 = \lu2 + v2 cos(§ +7) (4‘9)
and

U, = w (4-10)

where y is the instantaneous ﬂow angle with respect to the probe.

4.5.2 X-array calibration

The air velocity magnitude and direction with respect to the probe axis are deﬁned as
Q and 7 respectively as shown in Figure 4.10. (Note that Q represents the air velocity at
the probe tip; it is not related to the ﬂow rate of the fan used in earlier sections. This
notation is used to follow what is typically used in hot-wire literature.) An iterative
processing algorithm known as the “speed wire - angle wire” technique (Foss et. a1
(1995)) was used to infer Q and 7 given simultaneous measurements of the anemometer
voltage outputs E+ and E_. Note the subscripts + and - refer to the +0 and -[1 sensors of
Figure 4.10. This technique requires calibration data to obtain E,(Q,y) and E,(Q,y).

These calibration data were acquired for 8 ﬂow speeds in the range 0.1Udp<Q<l.2U,,-p,

34

where Utip represents the maximum velocity that would be expected in the ﬂow ﬁeld. For
each speed setting, data were acquired for a range of -36<y<+36 degrees in 6 degree

steps. These data were ﬁt at each calibration angle to the modiﬁed Collis and Williams

equation:

E3(Q,r) = 4.0 ) + 3. (Y)Q"*“’ (4-11)
and,

E3(Q,Y) = 4-0 ) + 3-0 )Q"“”- (4-12)

The coefﬁcients AB and n were found for each of the 13 values of y by computing a least
squares linear ﬁt of the data as n varied from 0.2 to 0.7. The most linear data ﬁt was
determined by choosing the n value which provided the smallest value of the standard

deviation of the ﬁtted function:

1 N 2 %
StDeV = (W Z (Qcalculared — measured) ) '
l

A new variable was deﬁned to obtain a relationship between the wire voltages and the

(4-13)

ﬂow angle (y) for a given speed. Speciﬁcally, n(y) at a ﬁxed Q is deﬁned as

=M_1 (4—14)
E(Q,r = 0)

for each wire. Using equations (4-11) and (4-12), 11(7) was evaluated at even speed

11(r)

increments for a total of 51 speeds in the range of the calibration data. A ﬁfth order

polynomial was used to ﬁt the data computed with equation (4-14). Figure 4.13 shows an
example of the 11(7) curve ﬁt at a ﬂow speed of 20m/s.

A ﬁnal calculation was made using the calibration data to estimate B, and [i for the
two wires. It can be shown (Bohl (1996)) that B for each wire can be found by evaluating
the ﬁmction

[3 = tan—‘(L’dﬁz-AE2 — A)n) (4'15)

at 7:0. A second order ﬁt of E2 versus y was differentiated to evaluate equation (4-15)

for each calibration.

35

4.5.3 Evaluation of Q and y

The speed and angle of the air velocity were determined using the “speed wire/angle
wire” technique. This iterative technique assumes that the wire which is more
perpendicular to the ﬂow (i.e., lower value of 13-7) is more sensitive to the air speed (Q),
and the wire which is more parallel to the ﬂow (i.e., larger value of 13-7) is more sensitive
to the oncoming ﬂow angle (y). The algorithm requires that initial estimates of Q and y to
be made. These are computed using the cosine law. Speciﬁcally, the u and v

components were estimated by the functions:

 

 

Q. + = “008(13.)+vsin(l3.) (4-16)
and
21,—. =ucos(l3.)+vsin(l3.). (4-17)
The effective cooling velocity Q,” was determined for both wires from
Q B E: - A.<o> — “'18)
= co 3

eff + + B + (0)
and

e _ = co 3 _

ff B _ (0) J

where the coefﬁcients A, B and n used were evaluated at 7:0.

The initial values used in the algorithm are calculated from

Q)” = W (4-20)
and
Yard = tan“(t'-). (421)
Note that the wire which has the larger initial computation of Qcﬂr determines which wire
will be treated as the “speed wire” and which will be the “angle wire.”

A new estimate for the speed, Qm, was calculated from the A, B, and n values from
the speed wire calibration closest to yo“. The variable 11 was then evaluated for the angle

wire using equation (4-14). Two estimates of the new ﬂow angle were made using the

two 7 versus 1] curves closest to the ﬂow speed Q,,,,. The next value of yum, was

36

determined by linear interpolation between the two angles based on ﬂow speed. The
new estimates of Q and 7 were then used as the initial guess for the next iteration. For

this work, 15 iterations were used to ensure convergence.

4.5.4 Uncertainty Considerations

The hot-wires used were calibrated before and after each experiment. The maximum
time between calibrations was 3 hours to ensure that “drift” would not contaminate the
measurements. The standard deviation of the calibration as calculated from equation (4-
10) was typically less than 0.04m/S (0.14% of full scale). Typical maximum values for
the standard deviation were 0.3m/s (1.0% of full scale); these occurred at the extreme
angles of the calibration where the wire is nearly parallel to the velocity vector.

The calibration drift was estimated by processing the data sets with the “pre” and
“post” calibrations independently and comparing the results. If the difference between
the data processed with the two calibrations was greater than 2% of the measured value
then the data set was discarded and the experiment was repeated.

The ability of the X-array and processing routine to calculate the speed and angle of
the ﬂow was tested. Speciﬁcally, time series data from the calibrations were processed to
compute the speed and angle of the velocity. The results were compared to the known
speed (from the pressure measurement) and angular orientation of the probe in the

calibration. Typical agreement in ﬂow speed was within 1%. The deviation in angle was

usually within 0.15 degrees for -24<y<24 degrees. For 24 <ly | < 36 typical deviation

was within 3 degrees (See Foss et. a1 (1995)).
An additional source of error in the hot-wire measurements is the existence of the
velocity component normal to the plane of the X-array. The probe was aligned with the

mean ﬂow angle measured from the tuft measurements to minimize the time average of

37

the normal component. However, instantaneous component of velocity normal to the
plane of the X-array will cause a “binormal” cooling which will yield an overestimate of
the true velocity magnitude. No attempt was made to compensate for the binorrnal
cooling in this work. The error introduced by this can be estimated to be a “cosine”
effect. Speciﬁcally, the speed wire responds to the vector addition of the desired value
(Q), and the binorrnal component (w). For example, if w is of order 20% of the velocity
component in the plane of the X-array (i.e. w/Q=0.2) then the error in Q will be 1.9%.
The maximum average value of w/Q measured is about 0.30 corresponding to an
overestimate in Q of 5%. More typical values of w/Q<0.15 were recorded which would

cause an uncertainty of only 1%.

4.5.5 Time Averaged and Phase Averaged Calculations

Time series data were acquired in the wake of the fan were processed to provide
statistics based on time averaged and phase averaged calculations. The time mean of
velocity was deﬁned as:

_. I N 4-22

U(x,r,0)=—ZU}.. ( )
N H

The standard deviation of velocity is

~ 1 ~ _ 2)” <4-23)

U(x,r,0) = FIE“ 41))

Phase averaged mean and RMS of velocity are calculated from

_ N 4-24
<U >(x,r,0,or)=—1—lv-ZU}.(jT+ta) ( )

i=1

and

38

 

i[Uj(jT+ ta )- < (7 > (Jr.r.0,ot)]2 (4-25)

< (7 > (x,r,0,0t) = H

 

N—l

respectively, where or is an angular coordinate measured from a ﬁxed point on the
rotating fan. In these equations the time for one complete revolution of the fan is written
as T. The symbol t, is the phase lag for a given angle of 0 < a < 360° such that 0 S ta< T.
Past investigators of rotating machinery have used similar phase averaging techniques;
see Lakshminarayana (1981).

Another quantity used to analyze the velocity data is the phase averaged vorticity.
The phase averaged axial component of vorticity in a cylindrical coordinate system is

given by

 

lao<%>)a<a> (4%)
<c0x>(x,r,0,0t)=- 6r — 60 .
r

The partial derivative terms of equation ( 4-26) were approximated using a second order
ﬁnite difference scheme. The interior points were calculated using a central difference
(see Anderson (1984)):

6a) _ a a 1 (4-27)

i+l i-

n,’ ma’
and the boundary points were calculated with a forward difference:

2a) _ -3a,. +4a. , —a,.+2 (4'28)

1+

n,’ :Mh ’
where Ah is the distance between data points.

The data rate at which time series of velocity were acquired was 6.0kHz,
corresponding to every 1.0 degrees of fan rotation. A total of 324000 samples were
acquired to build the statistical population for 900 fan revolutions (54 seconds of data for
a fan speed of 1000 rpm) Statistical convergence of all phase averaged quantities was
checked. An example of the statistical convergence of the mean velocity computed from

equation ( 4-24) is shown in Figure 4.14.

 

/i

 

 

!
\
§

  

'IIIIIII/IIIIIIIIIf

 

,A /B SC

 

 

 

 

 

XM®W®W§$\

 

J

 

 

 

 

C) C en m <3 w >>

 

 

l—l

 

 

 

II”IIIII/IIIIIIIII/I/IIII/IIII/IIIIIIIIIIIII/II/IIIIIIIIIII III/IIIIIIIIIIIIIIIIIIJII.‘VIII/II.-

fwmm -.
F“

H—t
i—L-I
\Sxmmxxxkzmnkmxm
\\\\\\\\\\\\\\\\\\\\\ \\\\\\\\\\\‘.§\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\ ' \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\

[III II II III/I II III/IIIII/I I/II/IlI/IIIIII I IIIIII/IIIIIII/l/IWIIII III/IIIIIIII/IIIIIIIIIl/l/IIIII/IIIIII III/IIIIII/III/III“

-T

§.

AXISYMETRIC SHROUD II THROTTLE PLATE

TEST FAN PRIME MOVER
PRESSURE TAP DRIVE MOTOR

TRAVERSE NOZZLE

UPPER RECEIVER TURNING VANE

FORCE TRANSDUCER
INLET TO PRIME MOVER

LOWER RECEIVER

BEARING SHAFT SUPPORTS
OPTICAL ENCODER

emtzL-‘C—ir"

Figure 4.18chematic of the ACFRD Facility

40

 

 

 

 

 

 

PTCJOUJD

 

 

   

 

 

 

 

 

 

 

 

 

 

 

AerIHMN Troverse ng

Homzonm.8hde Bracket

80”

Steppmg Motor

16”

F
G
HorzontOlSmde H
l
G

Vertmxﬂ Shde

 

 

 

 

 

1‘ £1\
_ H E T D
/ \
I
F r J G

Cross Support Rod
Vertmxm Stopper Rod
Probe Mount

Lmeor Beormgs

Probe support Swmg

Figure 4.2 Schematic of Traverse Used in the ACFRD

41

A

:11

 

K/C

 

 

 

 

Figure 4.3 Schematic of Probe Holder

.__.__L

 

 

 

 

 

 

42

 

 

 

 

 

 

 

 

 

 

 

Inkat Pdter

[Umse Nozzm?
Iurnﬂm;\Vone
Counter BQMNKI?
Force Tronsducer
KmFe Edge PNot

Peoever PMXN‘
J lWoor o? Lower RecEver

E/IPUUZZl—X

Figure 4.4 Schematic of Flow Metering Device

43

 

Q(m /S)

0.9 T 3
0.8 r r
0.7 T “

0.6 _ ‘*
0.4 ‘- ‘

0.2 _ _

 

 

l

 

0 0 L l l l
0.00 0.05 0.10 0.15 0.20 0.25 0.30

S(ll-“ES

Figure 4.5 Calibration and Fit for South Side Flow Meter

 

 

 

 

000 l L r I I l r 1 l 1 I

0.00 0.05 0.10 0.15 0.20 0.25 0.30
s(lrt(En)

Figure 4.6 Calibration and Fit for North Side Flow Meter

44

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.50 f l I I I I
I;
0.40 _ A u. C 1400 rpm _
D A
0“ [:l 1500 rpm (Mfg Data)
0.30 - E] -
‘IG A 1600 rpm
0
05
0.20 — D —
A
o
. 0‘. .13.
0.10 —— ‘A “3% —
0‘
‘ v:
000 l l L l l J? l
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35
Figure 4.7 Flow Meter Test Data
From i 31 3
Power To DC Motor
Supply
(230v) l 101 J 6
\D \:S
Current Meter
i ——e
100k 0 output i
Voltage/ l 50k i output
Divider O

“t

Figure 4.8 Schematic of Power Measurement Circuitry

Power (kW)

45

 

 

 

 

 

 

 

 

 

 

 

0.7 — _
0.6 — ~
0.5 — A
0.4 — -

+ 1400 rpm .
0‘3 _ + Manufaturcrs Data: 1500 rpm -
0.2 _ + 1600 rpm _
0.1 — —
0‘0 1 l l l l l l l
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80
3
Flow Rate (m /S)
Figure 4.9 Power Measurement Test Data
Y
7. X $333"
V WIRE 2 —/ I
‘——__ L
a) FRONT VIEW b) SIDE VIEW

 

Figure 4.10 Drawing of an X-array Probe

Realizations

Realizations

46

 

 

 

 

 

 

 

 

 

12000.0ililililwwwwwwﬁr,
r- -I
10000.0 —- m\ ——
1' .
_ 1 ‘1 q
8000.0 — f 1, —
_ f ’1‘ -
6000.0 — / \ -
— I In A
40000 / i’»
u — f —
_ , \ -
2000.0 " —
0.0 J L l l 1 L L l l l l l l l l l l l
~36 -30 -24 ~18 -12 -6 0 6 12 18 24 30 36
V
Figure 4.11 Histogram of ﬂow angles for a properly aligned probe
10000-0 ' I ' I ' I ' l ' r ' l ' l ' I ' I ' I 1 '
.- ..oom‘ q
/
8000.0 - ,r -
I— U, D _,
a" ‘3
6000.0 - j 1‘! -—
P u
I ,# "~
4000.0 — fl ‘ _
r f! 1“ -
_ 5
2000.0 ’4' f
- if 11
.00.
00MM"?1.ILILI.IIIII 1.
-36 -30 -24 -18 -12 -6 0 6 12 18 24 30 36

Figure 4.12 Histogram of ﬂow angles for an improperly aligned probe

47

 

 

 

 

 

 

  

 

 

 

 

 

 

 

40 I I T I I I
30 - -
20 — -
10 ~ -
0 — J
-10 1— —
-20 ~ —
-30 ~ —
-40 l l l I l l
-0.20 -0.15 -010 -0.05 0.00 0.05 0.10
Ti
Figure 4.13 11 versus y with 5'h order polynomial curve ﬁt
45 .. I I r I I I1 60I " I I I I I I I I I -
. _llIl IIIIIIIIIIIlllllllfiilulll_-
”Ev?”— marl/"WI .1. :—
3.5 - 1.56 _ ' -_ -
v 3 0 ; 1.54 _— V _ _:
J? ° - 1.52 f j -
82.5 1.50 lllIlllIlllIlllIlllIlllIlllﬂlll —-I
T) 2.0 500 550 600 650 700 750 800 850 900.
> -
C.‘ 1.5
“3 j
a) 1.0 -
2 0.5 ‘
0.0 I l l I l l I I l l l I l l l I l

200 400 600 800
Number of Samples

Figure 4.14 Example of statistical convergence of phase averaged data

5. RESULTS AND DISCUSSION — INTEGRAL PARAMETERS

The efﬁcacy of the aerodynamic shroud concept has been demonstrated using
performance and efﬁciency data for the fan and shroud system. This chapter will present
and discuss these integral performance characteristics. A description of the dimensional
conditions tested are given in Section 5.1. System performance and efﬁciency will be
discussed in Sections 5.2 and 5.3 respectively. Experiments were also executed using the
test fan with a small tip clearance boundary condition for comparison with the

aerodynamic shroud results; these data are presented in Section 5.4.

5.1 Experimental Program

The system performance was established as the pressure rise across the fan as a

function of the ﬂow rate through the fan (i.e. Q10” 2 QM, — Q5," .) Typically 15 points

were acquired from 0<APf,,,,<APf,n_max to deﬁne this “characteristic curve.” These data
were acquired for fan speeds of 800, 1000, 1200, 1400, 1600, and 1800 rpm with the
shroud gap (g) set to 2mm. Data were also taken for g = 1.0, 1.5, 2.0, 2.5, and 3.0 mm
with a fan speed of 1000 rpm. For each of the above tests typically 5 shroud pressure
conditions were tested; these ranged from 0 to 800 Pa. This combination of variables led
to a total of 50 curves to fully characterize the shroud and fan performance. The above

dimensional conditions can be described in non-dimensional form as 0.000<x<0.068

where x is deﬁned in equation (2-9).

5.2 Performance Data

The system performance is deﬁned in terms of dimensionless variables by equation
(2-7). An empirically based simpliﬁcation of (2.7) has been identiﬁed by the process of

examining correlations of the data. Speciﬁcally

I! =v(¢,x) (5'1)

provides a two parameter representation of the pressure rise data for the present
system. This equation implies that the pressure created by the fan/shroud combination is
only a ﬁmction of the ﬂow rate through the fan, and the energy delivered to the shroud.

This simpliﬁcation of equation (2-7) can not be argued to be a necessary result of the

48

49

boundary value problem. However, the analysis and presentation of the experimental
data have been greatly simpliﬁed by the reduced (3 —->2) number of independent variables.

The system performance for several x values is shown in Figure 5-1. This
“characteristic curve” type of data presentation is common in fan studies. Note that x is
not a unique function of 111 and It) as seen by the crossing of the curves in Figure 5-1 at
lower ﬂow rates. Therefore a clearer and more descriptive method of data representation
is a contour plot of 111 on a 4) versus x plane allowing the ﬁinctional dependence of
equation (5-1) to be clearly shown; see Figure 5-2. Note that all the dimensional
conditions listed in Section 5.1 were used in this ﬁgure.

The data show increased performance for higher ﬂow rates, and decreased

performance for lower ﬂow rates in terms of the pressure rise at a given ﬂow rate. As
seen in Figure 5-1, the data for each x value tested “crosses” the x=0 curve. For ¢<¢mss,
the pressure rise at a given ﬂow rate is decreased for x>0 when compared to w(x=0).
However, the pressure rise is increased for ¢>¢mss. For x=0.0010, this crossing occurs at
¢cmss=0.l7. As x is increased, the value of Items, decreases to 0.10 for x=0.068. Figure 5.2
shows a line representing the locus of points representing the (bums for all x values.

At the higher ﬂow rates (¢>¢cmss) the pressure rise increased monotonically with
increasing x. Also, the ﬂow rate at zero pressure rise: ¢o=¢(w=0,x) increased
dramatically for 0<x<0.009 from (110:0.27 to 0.36. For 0.009<x<0.068, 4), increased more
gradually according to the empirical function:

1. = 0.900(x) + 0.346. (5-2)

An additional collapse of the performance data was observed to deﬁne the function
of equation (5-1). Speciﬁcally, for x>0.009, and ¢/¢m,,>0.21 the performance can be
described by the function:

11 = (1000. x )°'2 (— 0.1708(%) + 0.0365(%) + 0.1351) (53)
This data ﬁt along with performance data from several x values are shown in Figure

5.3. Equations (5-2) and (5-3) can be used to estimate the pressure rise across the fan

given the ﬂow rate and x value to within an estimated error of i5%.

50

5.3 Efficiency Data

The efﬁciency of the cooling fan and aerodynamic shroud combination is deﬁned in
equation (2-5) and given in terms of its functional dependence in equation (2-8).
Analysis of the data has shown that the system efﬁciency, like the performance, can be
described by:

n... =n...(I>,x). (5-4)

Figure 5.4 shows the efﬁciency for the same x values shown in Figure 5.1. Similar
to the performance data, the efﬁciency data can also be presented as a contour plot to
uniquely describe its functional dependence; see Figure 5.5. The data presented in
Figures 5.4 and 5.5. were calculated using nsh,=1.0. Figures 5.6 and 5.7 are
representations of the efﬁciency data using nsh,=0.7. Comparison of Figures 5.4 and 5.6
indicate that a less efﬁcient shroud delivery system would cause some decrease in overall
efﬁciency, however, the qualitative characteristics of the data remain the same.

The data shown on Figure 5.4 Show that the aerodynamic shroud increases
efﬁciency for high ﬂow rates, and decreases efﬁciency for lower ﬂow rates. The amount
of increase or decrease is extremely sensitive to the shroud energy for 0<x<0.020, and
less sensitive to this variable for 0.020<x<0.068. It is interesting to note that the ﬂow
rate at which the efﬁciency curves for x>0 crosses that of x=0 is approximately 0:017

for all x values.

5.4 Small Tip Clearance Study

The data presented in Sections 5.2 and 5.3 compare the performance of the fan and
shroud combination to the shroud off condition (i.e. x=0.000). A single experiment was
conducted to compare the performance and efﬁciency of the aerodynamic shroud with a
2.5cm tip clearance to a shroud with near zero tip clearance. A ring made from
Styrofoam was manufactured to ﬁll the annular area between the fan tip and aerodynamic
shroud. The inner diameter was build slightly undersized, and then “reamed” by the
rotating fan to create a small (zl .Omm) tip clearance.

Figure 5.8 and Figure 5.9 show the performance and efﬁciency respectively for the

small tip clearance along with data presented in Section 5.2 and 5.3. The results show

51

that both performance and efﬁciency are enhanced at higher ﬂow rates, and degraded for
lower ﬂow rates. These data indicate that fan applications where a small tip clearance is
allowed could beneﬁt from a larger tip clearance with the aerodynamic shroud.
Applications which operate in an environment which has a relatively small system
resistance would be the most likely to beneﬁt, i.e., high ﬂow rate--low pressure rise
conditions. It should be understood however, that the fan used in this test was not
designed to operate with a small tip clearance. Additional research will be able to

determine if the general trends observed in these data would apply to other fan designs.

52

 

0.50

0.40 —

0.30 '1‘

0.20 p-

 

 

 

 

0.00

0.00 0.50

 

Figure 5.1 Pressure rise vs. ﬂow rate data for selected shroud conditions

53

‘P....

. 6

 
  

20? 0970
: l
k 0

10‘. Qaqé’
5 xx , . / ,.
0.1 0.2
1’

Figure 5.2 Contour plot of system performance (11!)

 

 

033' ' ' '014

54

 

 

 

 

 

 

 

 

 

 

O-ls'l'l'l'l'l'l‘l'l'l‘
W 0.14 —
0-2 0.13 —
(1000*x) 3
0.12 — x*10
0.11 - . 9.2
0.10 —
0.09 — - '0'9
0.08 ~ A 18.7
0'06 _ t 43.9
0.05 ~
0.04 _ E 68.4
0.03 e —— Fit
0.02 —
0.01 -
000 r l I l
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
/¢o
Figure 5.3 Data ﬁt for x>0.009
25
n(%) _
20 1— ' ii 5. + 0.0
+ 1.0
+ 54
15 — + 92
_ 21.4
10 — 35.5
43.9
F 68.4
5 _
0 i 1
0.00 0.10 0.20 0.30 0.40 0.50

Figure 5.4 Efﬁciency data for selected x values (115,51 .0)

55

60-

5.2

3.3

50-

7.0

40-

11*103

30-

 

 

 

Figure 5.5 Contour plot of fan/shroud efﬁciency (nsh,=l .0)

56

 

25.0 I I
Tlsys(%)
3
x*10
20.0 —
+ 0.0
—I— 1.0
+ 5.4
15.0 _
_o_ 9.2
+ 21.4
10.0 —.{+— 35.5 ..
—x— 43.9
—E— 68.4

5.0

 

 

 

00 i 1 1 1 I 1 i u I “‘1 l
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50

(1

Figure 5.6 Efﬁciency data for nshr=0.70

 

 

 

 

 

 

 

 

 

111

3.3
5.2 .—
7.2
7
f
0 6

13.1

 

15.0
17.0
~13

 

 

 

 

 

 

Figure 5.7 Contour plot of fan/shroud efﬁciency (nsh,=0.70)

58

 

 

 

 

 

0.50 ' I ' l j I I I I I T I r I I
\ll . .
0.40 — —l— 1:000" —
_ + x=0.0213 1
0 30 I— --.— 1mm tip clearance _
\
0.20 ~ _.
\ x
0.10 — -
0 00 i l I l 1 l l l J l 1 l L l I
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40
Figure 5.8 Performance curve for fan with 1.0 mm tip clearance
30-0 I I I ' T I I j j I ﬁr I I T I
Tlsys + x=0.000
25.0 — + x=0.00213 -

+ 1mm tip clearance

20.0 — / _

15.0 P 4

 

 

 

10.0 — _
5.0 - u
00 I l I I I 1 1 l I l i l . l i

0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40

Figure 5.9 Efﬁciency curve for fan with 1.0mm tip clearance

6. RESULTS AND DISCUSSION — VELOCITY MEASUREMENTS

Tufts and hot-wire anemometry were used to quantify the velocity ﬁeld in the wake
of the fan. A single fan speed of 1000 rpm was chosen for the detailed experiments. It
was assumed that different fan speeds would not signiﬁcantly affect these results given
the collapse of the dimensionless variables shown in Chapter 5.

The tufts were used to gain a qualitative understanding of the ﬂow ﬁeld in addition
to the measurement of the time mean ﬂow angle for use in hot-wire positioning. Tuft
surveys were conducted for a wide range of operating conditions. The dimensionless
parameters as well as an identiﬁcation letter for these conditions are listed in Table 6-1.
These conditions are represented graphically with the performance data presented in
Chapter 5 on Figure 6.1.

Hot-wire surveys were conducted for conditions ‘G’ through ‘K’. For purposes of
comparison, it was decided that three points of constant 4) with x = 0.000 , 0.020, 0.068
would be investigated (points ‘G’, ‘H’, and ‘1’.) Additionally, three points of constant \1/
with x = 0.000 , 0.020, 0.068 (points ‘I’, ‘J’, and ‘K’) would be examined. This would
allow the details of the ﬂow ﬁeld to be compared as the shroud power is increased for

constant ﬂow rate and for constant pressure respectively.

6.1 Tuft Results

Tuft surveys were conducted for the conditions listed in Table 6-1. Ten radial
locations (0.5<r/R<l.0, Ar/R=0.056) were interrogated at a single axial distance

(x/R=0.l l) as well as several axial locations (-0.05<x/R<+0.1, Ax/R=0.056) at a constant

radius (r/R=1.17). The ﬂow was assumed to be axisymmetric; therefore a single arbitrary

59

60

0 position was used for the traverse. These locations are shown with a schematic of the
fan and shroud in Figure 6-2a. Figure 6-2b shows a view of the x-0 plane with a vector

representation of the outlet ﬂow angle ([32). Digital images in the x-r and r-0 planes were
acquired for each tuft position. An example of a digital tuft picture in the x-r plane is
shown in Figure 6-3.

Results from conditions ‘A’ through ‘K’ are shown in Figures 6-4 through 6-14
respectively. Each ﬁgure contains a partial schematic of the fan and shroud, as well as a
vector representation of the mean ﬂow angle measured from the tuft as shown by vector
arrows. The arrows which originate at an “X” show the angle of the vector in the x-r
plane. An arrow representing the x-0 view of the same location is drawn directly below
its x-r plane representation, and originates from a “+” Sign; note that this view is

presented only for the locations of constant x/R.. A vertical line is shown at each vector
origin for reference. The symbol ® represents a location where the tuft was extremely

unsteady and a single mean ﬂow direction was not observed.

Conditions ‘C’, ‘F’ and ‘I’ represent the data taken with the aerodynamic shroud
“off”, i.e., x= 0.000. Condition ‘1’, which is a relatively high ﬂow rate condition is
characterized by nearly uniform outlet ﬂow angle ([32) and low radial velocities. A large
region of reversed ﬂow is noted in the tip clearance area. As the ﬂow rate is decreased
(pressure is increased) the ﬂow near the hub separates from the suction side of the fan
blades as indicated by the reversed ﬂow directions shown near the hub in condition ‘F’.
This condition is referred to as “fan stall” in Baranski (1977). Past researchers (Dring et.
a1 (1982)) have observed reversed axial ﬂow and strongly radial ﬂow corresponding to

suction side boundary layer separation as indicated by ﬂow visualizations. As this

61

separation occurs, the ﬂow near the tip becomes predominantly radial. As the ﬂow is
further decreased, condition ‘C’ shows reversed ﬂow over the entire blade span, and
outwardly radial ﬂow in the tip clearance area.

Conditions ‘A’ and ‘B’ represent the ﬂow ﬁeld as x is increased while the ﬂow rate
is held constant from condition ‘C’. Note that the aerodynamic shroud decreases both
performance and efﬁciency at this ﬂow rate. The results show relatively little change in
the tuft directions. The entire blade span remains stalled and the radial ﬂow in the tip
clearance area appears unaffected by the aerodynamic shroud. Similarly, conditions ‘D’
and ‘E’ represent the ﬂow ﬁeld as x is increased for the same ﬂow rate as condition ‘F’.
These data also indicate that the ﬂow directions are somewhat insensitive to the
aerodynamic shroud. It is believed that the reduced performance of the fan/shroud
system at these operating conditions is a result of the radial outﬂow at the fan’s tip.
Speciﬁcally, the tip clearance ﬂow is the only positive net ﬂow through the blades, and is
therefore the only contributor to the net pressure rise across the fan. The aerodynamic
shroud was designed to maintain axial ﬂow throughout the tip clearance area, and
therefore it reduces the effectiveness of the fan to produce the radial outﬂow.

It is interesting to note that axial ﬂow turbomachinery typically shows low pressure
rise at the lower ﬂow rates as a result of fan stall. A peak in performance is then
observed. Automotive cooling fans do not exhibit this behavior because of the partially
shrouded boundary condition that allows the fan to create the radial outﬂow and thereby
maintaining a relatively high pressure rise at low ﬂow rates. The aerodynamic shroud
which “ﬁlls” the tip clearance gap with a Coanda jet of air seems to cause the fan to

behave as if it had a cylindrical boundary condition. This characteristic peak in

62

performance at the middle ﬂow rates is slightly visible in Figures 5.1 and 5.2 for
0.030<x<0.050.

Conditions ‘G’, ‘H’, and ‘I’ were taken at a ﬂow rate where the performance and
efﬁciency were enhanced by the aerodynamic shroud. The ﬂow ﬁeld directions in the
wake of the blades were relatively unaffected by the shroud jet. However the tip
clearance ﬂow in this area was reversed. Speciﬁcally, the negative axial ﬂow in the tip
region observed for condition ‘I’ was reversed by the presence of the Coanda jet. This
behavior was also observed as x was increased at constant fan pressure (i.e., conditions
‘I’, ‘J’, and ‘K’). Additional observations show that the mean ﬂow direction in the entire

clearance area:

Ad = %(Dszhr " Dian) (6-1)

is in the negative axial direction for condition ‘I’. The relatively large ratio of
”ﬁlm = 0.31 indicates that a negative axial velocity through the tip clearance area would
have a large effect on the net volume ﬂow rate. Speciﬁcally, given a uniform velocity
distribution, 31% of the ﬂow through the blades would be recirculated in the tip clearance
and would not contribute to the net Q“. The ability of the shroud to enhance the

performance of the fan for these ﬂow conditions is believed to be a result of the positive

mean axial ﬂow momentum cause by the shroud in the tip clearance region.

6.2 Hot-wire Results
Hot-wire surveys were conducted for conditions ‘G’, ‘H’, ‘I’, ‘J’, and ‘K’ listed in
Table 6-1. The hot-wire locations surveyed for these test conditions were 0.5<r/R< 1.1,

ArfR=0.55, with x/R=0.ll. The ﬂow was assumed to be axisymmetric. Time series

63

voltages were acquired from the two hot-wires as well as the optical encoder attached to
the drive shaft. Data were taken as 6000hz, corresponding to a single measurement for
every 1 degree of fan rotation (given a fan speed of 1000 rpm). Time averaged mean and
RMS of velocity are reported in section 6.2.1. Data collected from all radial locations
were organized to be presented on a grid as shown in Figure 6.15. Note that data were
acquired for the full 360 degrees of revolution. Only the ﬁrst 90 degrees of rotation will
be presented. Phase averaged mean and RMS of velocity, as well as vorticity, will be
reported in this chapter. Two additional quantities: the phase averaged kinetic energy and

exit ﬂow angle are reported in Appendices B and C respectively.

6.2.1 Time Averaged Velocity Results

Time averaged mean and RMS of velocity were calculated at a function of radius
from equation (4-23) and (4-24). Axial, Tangential, and Radial mean and RMS velocities
are shown in Figures 6.16 through 6.18 using U0 to normalize the data. Figures 6.19

through 6.21 show the same data scaled by U Note that Uo can be recomputed from

tip'
equation (2-1), given the blade tip speed Uﬁp=23.94m/s at 1000 rpm. The magnitudes of
the time averaged velocities were not greatly affected by the aerodynamic shroud.
However, qualitative changes in the proﬁles were observed. For example, Figure 61%
indicates that conditions ‘J’ and ‘K’ have considerably lower RMS levels for
0.65<r/R<0.9. Also, condition ‘G’ showed greater mean radial velocities for r/R>0.8
compared to conditions ‘H’ and ‘1’, whose mean radial velocities were nearly identical;
see Figure 6.21a. The stalling phenomenon clearly visible at the lower ﬂow rates from

the tuft observations appears to be present in condition ‘G’ as seen be the negative axial

velocities near the hub. Dring (1982) also noted large radial velocities in axial

64

turbomachinery when a “comer stall” was present. Additional discussion of these
velocity data will be given in the following section where in the phase averaged

representation of the results is presented.

6.2.2 Phase Averaged Velocity Results

Phase averaged data were calculated from the time series using equations given in
Chapter 4. These data are presented as contour plots for the ﬁrst 90 degrees of rotation.
The orthogonal axis indicate the radial position normalized by blade radius (R=Dfan/2).
Tick marks are shown at 15 degree increments. ISO-velocity contours are shown and
given a label number in the ﬁgures. A legend is shown which relates the contour number
to the velocity scaled by U0 and Udp.

Mean axial velocity results for conditions ‘G’ through ‘K’ are shown on Figures 6.22
through 6.26 respectively. The data show a periodic signal every 36 degrees indicating
the passing of the fan blades. A radial line of local minimum axial velocity is observed,
and believed to have been caused by the wake of the fan blades. This is most clearly seen
for conditions ‘J’ and ‘K’. For reference, tick marks with a (*) have been placed at 6.4,
42.4, and 78.4 degrees to indicate the center of the blade wake. These values were
selected from local minimums in axial velocity observed for condition ‘I’. The
qualitative aspects of the axial ﬂow ﬁeld between the blades are similar for all ﬁve test
conditions. Speciﬁcally, a local maximum is observed for all conditions near r/R=0.8.
The azimuthal position of the maximum did however vary for the different conditions. It
is interesting to note that the magnitudes of U,,/Utip were relatively unaffected by the
increase in ﬂow rate from ‘I’ to ‘K’. It is therefore believed, as discussed in Section 6.1,

that the increase in ﬂow rate is a result of the positive net ﬂow in the tip clearance area;

65

i.e., less ﬂow is recirculated in the tip thereby increasing the net positive ﬂow rate. As a
result the relative values of UX/Uo decreased for conditions ‘J ’ and ‘K’.

Tangential mean velocities are shown in Figures 6.27 through 6.31. These Show the
sharp gradients in velocity near the hub and tip areas showing the shearing of the ﬂow
exiting the fan with the non-rotating ﬂuid in the upper receiver. These results show
considerable variation in the qualitative character of the ﬂow ﬁeld between the different
ﬂow conditions. The magnitudes of UB/Utip are, however, relatively insensitive to the
ﬂow condition.

Radial mean velocities are shown in Figures 6.32 through 6.36. Large positive radial
velocities are observed for a range of or values slightly greater than that of the blade wake
(this can be interpreted as the “suction surface” region) for all test conditions. Also,
negative radial velocities are observed at values of or less than that of the blade wake (i.e.,
the “pressure surface”). This region of negative radial velocity is considerably larger for
conditions ‘J’ and ‘K’. There are several theories in the literature which attempt to
explain this pattern of radial velocities. Several researchers (Inoune and Kuroumaru
(1984), Dring eta] (1982)) have observed strong positive radial velocities near the suction
side of the blades. This has been attributed to centrifugal effects acting on the blade’s
boundary layer. It was also noted that a corner stall at the hub will signiﬁcantly increase
the positive radial velocities. Note that condition ‘G’, which appears to exhibit comer
stall as indicated by the small positive and negative axial velocities, shows the largest
positive radial components. Lakshminarayana (1996) provides a different reasoning for
these “secondary ﬂow features” which is also compatible with the presented data.

Speciﬁcally, the streamlines seen by an observer rotating with the blades are curved

66

corresponding to the pressure gradient created by the suction sides and the pressure sides
of the blades. The boundary layer on the fan hub can not support this pressure gradient,
and therefore a secondary ﬂow exists. This phenomenon is identical to the secondary
ﬂow which is observed in a stirred tea cub which causes the collection of particles at the
center of the bottom of the cup.

The phase averaged RMS results for the three velocity components are shown on
Figures 6.37 through 6.51. Note that these results indicate the “unsteadiness” of the ﬂow
ﬁeld and are not necessarily indicative of sheared or turbulent ﬂuid. Large cycle to cycle
variations have been observed in the time series, and are suspected to be a result of small
disturbances in the laboratory air. These can cause the RMS calculation to Show
relatively high values of unsteadiness in ﬂuid that has not been sheared by the fan. It is
difﬁcult (if not impossible) to separate cycle to cycle variations from turbulence caused
by the shearing of the ﬂuid by the fan blades.

The RMS data reported show high disturbance levels near the blade wake region, as
well as the hub and tip regions. The area interior to the blades shows relatively low RMS
levels. This central region of low RMS intensity is clearly seen for condition ‘K’. Figure
6.1 of Lakshminarayana (1996) is a similar picture showing shaded regions of high
unsteadiness. The shape of the “inviscid core region” in this reference is very similar to
contour number 6 shown on Figure 6.41. As the ﬂow rate is decreased (‘J ’ and ‘1’) this
region becomes smaller indicating that the boundary layer thickness on the surface of the
blades is increasing. As x is increased at constant ﬂow rate (‘H’ and ‘G’) this region
becomes even smaller indicating that the increased pressure rise across the fan caused by

the aerodynamic shroud has caused the boundary layer thickness to further increase.

67

6.2.3 Phase Averaged Vorticity Results

Axial vorticity was calculated from equation (4-29) and normalized by 20,“. These
results are shown in Figures 6.52 through 6.56. A large region of negative vorticity is
observed in the wake of the blades. This is a result of the positive radial velocity on the
suction surface and the negative radial velocity on the pressure surface which interact
downstream of the blades. A weak region of negative vorticity is also realized for all test
conditions near the hub as would be expected from the shearing of the rotating ﬂuid
through the fan with the ﬂuid in the wake of the hub. A large region of strong positive
vorticity is observed for all conditions in the tip clearance area. This is also expected
based on the above comments for the hub ﬂow. Relatively low magnitudes of (0, were
observed for conditions ‘J’ and ‘K’ for much of the region interior to the blade wakes. It
is interesting to note that this region becomes smaller for conditions ‘I’, ‘H’, and ‘G’.
Note that the contour of near zero 0)x is similar in shape to the contours of low RMS
values for all ﬁve test conditions, i.e. there is an obvious correlation between the areas of
low disturbance level and low vorticity level.

Large regions of concentrated positive vorticity are observed for conditions ‘J’ and
‘K’ near the tip of the fan. This is also observed for conditions ‘G’, ‘H’, and ‘I’, however
it is more difﬁcult to see. Many previous researchers (see, e.g. Lakshminarayana (1970),
Inoue (1986)) have measured similar behavior in axial fans. The literature often refers to
this as the “tip clearance vortex”. It is believed (Lakshminarayana (1996)) that this is
caused by a “rolling up” of the vorticity caused by the tip leakage ﬂow. Speciﬁcally, a

vortex sheet is created by ﬂow separation at the tip of the blades. There is a great deal of

68

literature which attempts to correlate tip clearance effects with the “losses” in the blade

row (Storer (1994), Lakshminarayana (1970), Adamczyk (1993)).

69

Table 1 Test conditions for detailed study

Test Condition 11: (1) X
‘A’ 0.30 0.05 0.068
‘B’ 0.24 0.05 0.020
‘C’ 0.38 0.05 0.000
‘D’ 0.30 0.10 0.068
‘E’ 0.23 0.10 0.020
‘F’ 0.30 0.10 0.000
‘G’ 0.28 0.17 0.068
‘H’ 0.20 0.17 0.020
‘I’ 0.10 0.17 0.000
‘J’ 0.10 0.30 0.020

‘K’ 0.10 0.34 0.068

70

 

0.50

0.40

0.30

0.20

0.10

 

0.00

 

 

 

 

Figure 6.1 Performance data with test conditions for detailed study labeled

0.50

7l

Shroud

Poni&odes

\‘fi :

 

 

 

 

 

XXXXXXX

I
I
TuFt Locoeons/I/

XXXXXXXXXXl '

Exﬁ:Flow Angus

RQQP“ shown Shown m Second
If) FIQUICE’S 6H4—614 ROW 01C VQCJCOICS

Note: Symbol 60 represents undetermined flow GngIe

Figure 6.2a Schematic of tuft survey locations

\\

/ /

/- 1
It
,r\

A

 

 

 

72

Figure 6.2b Representation of exit ﬂow angle 62.

 

Figure 6.3 Representative digital image of tuft

73

 

Pun Blode

 

 

_/. [fffiif

See Figure 6.1 For deﬁnition OF symbols

Figure 6.4 Tuft results from condition ‘A’

74

 

 

 

 

See Figure 6.1 for definition of symbols

Figure 6.5 Tuft results from condition ‘B’

\‘
§§§/Sbroud \
\
§ Fan Blade Q
E \\
E / ® 2 .2' f f f I I l
i

 

75

 

 

 

Sbroud
Fon Rode E§3SE
:; R
j®ax711711
1

See Figure 6.1 For deﬁnition of symbols

Figure 6.6 Tuft results from condition ‘C’

76

\V
I

/'

Sbroud

 

 

 

 

flilll‘1

See Figure 6.1 For deFinition of symbols

Figure 6.7 Tuft results from condition ‘D’

 

 

 

 

R
1

See Figure 6.1 For deFinitIon OF symbols

 

W

77

Figure 6.8 Tuft results from condition ‘E’

 

 

 

Sbroud

// Q
Fon Bode E§S

:: NR

:p/®\\1/711

See Figure 6.1 For deFinition 0? symbols

Figure 6.9 Tuft results from condition ‘F’

78

S

Sbroud
/

 

W

 

 

k. \'\‘\.*\.\'\

See Figure 6.1 For deFinition of symbols

Figure 6.10 Tuft results from condition ‘G’

R

 

 

R

 

 

l
l
r
i/
/
/
“”ffllllli\®
xxx

"i‘i’i‘i‘i‘i‘i’

llll

See Figure 6.1 For deFinition o? symbols

W

 

79

Figure 6.11 Tuft results from condition ‘H’

 

 

 

Sbroud
/ \\\
Fon Blode R
i \
1 k
f
f
f
I

ftillllll

H

+-—>+\,\

i‘i‘i‘i‘ii’i‘“

See Figure 6.1 For deFinition oF symbols

Figure 6.12 Tuft results from condition ‘I’

80

 

 

 

 

 

Q§§ Sbroud

S N

\\ Fon Blode \
I .\\\

liiiiiiiii

_1

See Figure 6.1 For deFinition oF symbols

Figure 6.13 Tuft results from condition ‘J’

W

 

81

 

 

 

Sbroud
/ \\
FonBlode \

l

1, RN
1

I

lllllllllli
XXXSXXXEX
IIIIIIIIIII

See Figure 6.1 For deFinition OF symbols

Figure 6.14 Tuft results from condition ‘K’

82

 

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Figure 6.15 Mesh of experimental data locations

3-5jl'l'l'l'I'I'l'IIIr1u
Ux30_

U025” +6

2.0- +H

83

 

15.. +1
10- . J

 

+K
0.5—
0.0-
1111 l 1 1.1.1. 1

 

 

 

 

 

 

 

 

1.00 I I r 1 I I I I I I

0.80:-

0.601

0.40-

0.20P

0.00- 1 1 1 1 I 1 1 1 l I
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

r/R
Figure 6.16b Time averaged RMS of axial velocity normalized by UC,

1.1

84

 

3.5 I I I I I I I I I I

U9 __
a) 3.0 ...

- + G ’ ‘
2.5 _

2.0- +1
15— . J

 

 

 

 

 

 

 

 

- __*_ K
1.0 — ‘1 / \ —i
0.5 T _
p .
0.0 1 l 1 l 1 l 1 1 1 l
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
r/R
Figure 6.17a Time averaged mean tangential velocity normalized by U0
1.20 I I I I I I I I I I
1.00 _
0.80 _
0.60 F
0.40 T
0.20 _
O 00 - l l l l
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1
r/R

Figure 6.17b Time averaged RMS of tangential velocity normalized by Uo

1.1

85

 

1.0 I

0.6 T

0.4 T

0.2 T

0.0 T

 

 

 

 

Figure 6.18a Time averaged mean radial velocity normalized by UC)

~ 1.0 I I I

 

l

{I

0.2 T T

0.8 T

0.6 T

0.4 T

 

 

 

0.0 1 1 I I l I l 1 I 1
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1

r/R

Figure 6.18b Time averaged RMS of radial velocity normalized by U()

86

 

 

 

 

 

 

0.8 _ v I I I . I I I . I I I I rm I ' r ' I ' .
— 0.7 :— .3
Ux : I
— 0.6 _T .1
Utip : I
0.5 : + G .-
0-4 i- + H —i
0.3 :- + I '3
" -I
0.0 :— _:
-01 ’ . 1 . I . 1 i l i L i l . 1 L 1 i 1 i 1 . ‘
0 0 0 l 0 2 0.3 0.4 0 5 0.6 0.7 0.8 0.9 1.0 1.1
r/R
Figure 6.1% Time averaged mean axial velocity normalized by Uﬁp
~ 0.16 I I I I I I I I I I
Ux
Utip 0.12 _
0.08 T
0.04 T

 

 

 

 

000 I I l l l l l l l l
i 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1

r/R

Figure 6.19 Time averaged RMS of axial velocity normalized by U

tip

87

 

0.8'I'I'I'III'ITI'I'I'
Utip0.6
0.5
0.4
0.3
0.2

0.1

IIIIIIIIIIIIITIIIIIIIIIIUIIITTj

 

 

llllllllllllllllllillllllll1141

 

 

00 L I l I L I l l l l l I l I I I l l l
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

p—l
O
p—A

 

~ 0.20 l I I l

0.12 _

0.08 —

0.04 _

 

 

 

 

000 l l l l l l l l l
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

r/R

Figure 6.20b Time averaged RMS of tangential velocity normalized by U

1.0

tip

1.1

88

 

I; 0.25
Etip 0.20 -
0.15 e
0.10 —

0.05 —

0.00 "

 

 

 

-0.05
0.0

 

Figure 6.21 a Time averaged mean radial velocity normalized by Un-p

 

N 0-25 I I I I I I I I I I I
1' _

I-

0.15 '—

O.10 —

0.05 *-

 

 

 

 

0.00 l l l l l l l I l l
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1

r/R
Figure 6.21b Time averaged RMS of radial velocity normalized by Utip

1.0—

0.9 -

0.8 -

0.7 —

0.6 -

0.5 --

0.4 —

0.3 -

0.2 —

0.1 —

89

 

 

 

0.0

Figure 6.22 Phase averaged mean axial velocity for condition ‘G’

   

Contour
Number

15
14
13
12
11
10

ANth’IODNmCD

U

I

U

0

2.878
2.671
2.465
2.259
2.052
1.846
1.640
1.434
1.227
1.021
0.815
0.608
0.402
0.196

 

-0.011

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0

U .
Utip
0.489
0.454
0.419
0.384
0.349
0.314
0.279
0.244
0.209
0.174
0.138
0.103
0.068
0.033
-0.002

 

.7 0.8 0.9

 

1.0

0.1 --

0.0

 

 

 

 

I

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Contour
Number

15
14
13
12
11
10

ANODAO'IOJVQCO

Figure 6.23 Phase averaged mean axial velocity for condition ‘H’

U

X

 

0

3.083
2.899
2.716
2.532
2.348
2.165
1.981
1.798
1.614
1.430
1.247
1.063
0.880
0.696
0.512

 

 

\

1.0

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0.0

 

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Contour
Number

15
14
13
12
11
10

ANOO-hUIOJVCDCO

Figure 6.24 Phase averaged mean axial velocity for condition ‘I’

91

U

X

U

0

3.039
2.841
2.644
2.447
2.250
2.053
1 .856
1 .659
1 .462
1 .265
1 .067
0.870
0.673
0.476
0.279

 

U

I

 

Utip

0.517
0.483
0.450
0.416
0.383
0.349
0.316
0.282
0.248
0.215
0.181
0.148
0.114
0.081
0.047

0.2

0.1 -

0.0

92

 

 

 

 

 

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

Contour
Number

15
14
13
12
11
10

41903-15010me

Figure 6.25 Phase averaged mean axial velocity for condition ‘J’

U

I

U

0

2.723
2.559
2.394
2.229
2.065
1 .900
1 .736
1 .571
1 .406
1 .242
1 .077
0.913
0.748
0.584
0.419

 

  

 

0.1

0.0

93

 

 

l

Contour
Number

15
14
13
12
11
10

ANOJ-hU'IODVQCO

Figure 6.26 Phase averaged mean axial velocity for condition ‘K’

U

X

U

0

2.171
2.027
1.883
1.740
1.596
1.453
1.309
1 .165
1 .022
0.878
0.735
0.591
0.447
0.304
0.160

 

 

l l
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

U .

Urip

0.738
0.689
0.640
0.592
0.543
0.494
0.445
0.396
0.347
0.299
0.250
0.201
0.152
0.103
0.054

 

94

 
   

 

 

/
1.0 —
0.9 —
0.8 —
0.7 —
0.6 -—
/
0.5 — u b
0.4 —
0.3 —
0.2 —
0.1 —
0.0 I10,
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0'9
Contour (7
Number E _9
U0 Utip
15 3.238 0.551
14 3.077 0.523
13 2.916 0.496
12 2.755 0.468
11 2.594 0.441
10 2.433 0.414
9 2.272 0.386
8 2.111 0.359
7 1.949 0.331
6 1.788 0.304
5 1.627 0.277
4 1.466 0.249
3 1.305 0.222
2 1.144 0.194
1 0.983 0.167

Figure 6.27 Phase averaged mean tangential velocity for condition ‘G’

0.1 —

 

Contour
Number

15
14
13
12
11
10

ANOObU'IODVCDCO

Figure 6.28 Phase averaged mean tangential velocity for condition ‘H’

95

U9
U

0

4.024
3.836
3.648
3.460
3.271
3.083
2.895
2.706
2.518
2.330
2.142
1.953
1.765
1.577
1.388

 

l
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

U6
U “p
0.684
0.652
0.620
0.588
0.556
0.524
0.492
0.460
0.428
0.396
0.364
0.332
0.300
0.268
0.236

 

I

rm

 

  
 
   
   

96

/

 

   

 

 

0.2 —
I
0.1 ~
0-0 I I I I I I
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Contour _ —
Number 91 $-
Uo U“);
15 3.062 0.521
14 2.917 0.496
13 2.773 0.471
12 2.628 0.447
11 2.483 0.422
10 2.338 0.397
9 2.193 0.373
8 2.048 0.348
7 1.903 0.324
6 1.758 0.299
5 1.613 0.274
4 1.468 0.250
3 1.324 0.225
2 1.179 0.200
1 1.034 0.176

Figure 6.29 Phase averaged mean tangential velocity for condition ‘I’

10

09

08

0.7

06

05

04

03

02

01

00

00 04 0.2 0.3 0.4 0.5

Contour
Number

15
14
13
12
11
10

9

ANCAJ-hU'IODVm

Figure 6.30 Phase averaged mean tangential velocity for condition ‘J’

97

®
*%
‘

E

U

0

1.814
1.702
1.589
1.477
1.365
1.253
1.141
1.029
0.917
0.805
0.693
0.581
0.469
0.357
0.245

I

b

 

0.7 08 09

U9
U up
0.544
0.510
0.477
0.443
0.410
0.376
0.342
0.309
0.275
0.242
0.208
0.174
0.141
0.107
0.073

  
    

10

0.4

0.3

0.2

0.1

0.0

.—A

—

4

—

 

98

  

  
   

 

   

   

 

l l I l I '1 l ,
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Contour ('7 (7
Number 9 0
U0 Utip
15 1.890 0.643
14 1.795 0.610
13 1.700 0.578
12 1.604 0.545
11 1.509 0.513
10 1.414 0.481
9 1.319 0.448
8 1.223 0.416
7 1.128 0.384
6 1.033 0.351
5 0.938 0.319
4 0.843 0.286
3 0.747 0.254
2 0.652 0.222
1 0.557 0.189

Figure 6.31 Phase averaged mean tangential velocity for condition ‘K’

   

 

     

 

l

l

 

1.0—

0.9 -

0.8 -

0.7 —

0.6 —

0.5 —

0.4 -

0.3 -

0.2 -

0.1 —

99

 

0.0

 

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Figure 6.32 Phase averaged mean radial velocity for condition ‘G’

l

l

Contour
Number

15
14
13
12
11
10

ANOO-h-U'IO'D‘IQCD

 

1.165
1.074
0.983
0.892
0.801
0.710
0.619
0.528
0.437
0.346
0.255
0.164
0.073
-0.019
-0.110

 

 

0.198
0.183
0.167
0.152
0.136
0.121
0.105
0.090
0.074
0.059
0.043
0.028
0.012
-0.003
-0.019

100

 

 

 

/
1.0 -
0.9 —
0.8 —
0.7 —*
0.6 —
0.5 —
0.4 —
0.3 -
0.2 —
0.1 -—
0.0 I r I I
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Contour (7 (7
Number ’ '
U0 U“,
15 0.974 0.165
14 0.832 0.142
13 0.691 0.118
12 0.550 0.094
11 0.409 0.070
10 0.268 0.046
9 0.127 0.022
8 -0.014 -0.002
7 -0.155 -0.026
6 -0.296 -0.050
5 -0.437 -0.074
4 -0.578 -0.098
3 -0.719 -0.122
2 -0.860 -0.146
1 -1.002 -0.170

Figure 6.33 Phase averaged mean radial velocity for condition ‘H’

 

 

 

101

 

 

 

 

 

1.0 H
0.9 —
0.8 -
0.7 —
0.6 —
0.5 —
0.4 —
0.3 —
0.2 -
4
0.1 -
0-0 I I I I I I I
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Contour — —
Number U’ U'
U0 U“),
15 0.868 0.148
14 0.793 0.135
13 0.718 0.122
12 0.643 0.109
11 0.568 0.097
10 0.493 0.084
9 0.418 0.071
8 0.344 0.058
7 0.269 0.046
6 0.194 0.033
5 0.119 0.020
4 0.044 0.007
3 -0.031 -0.005
2 -0.106 -0.018
1 -0.181 -0.031

Figure 6.34 Phase averaged mean radial velocity for condition ‘I’

1.0—

0.9 —

0.8 -

0.7 -—

0.6 —

0.5 —

0.4 -

0.3 —

0.2 —

0.1 —

102

 

 

 

0.0

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Figure 6.35 Phase averaged mean radial velocity for condition ‘J’

I

|

Contour
Number

15
14
13
12
11
10

-8N00-h0105\lm©

U

r

U

0

0.751
0.644
0.536
0.428
0.321
0.213
0.105
-0.002
-0.110
-0.218
-0.325
-0.433
-0.541
-0.648
-0.756

 

U ,
U’iP
0.225
0.193
0.161
0.128
0.096
0.064
0.032
-0.001
-0.033
-0.065
-0.098
-0.130
-0.162
-0.194
-0.227

 

 

1.0-

0.9 ~—

0.8 -

0.7 —

0.6 —

0.5 —-

0.4 —

0.3 —

0.2 -

0.1 —

0.0

 

l

l

I

l

103

 

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Contour
Number

15
14
13
12
11
10

9

8
7
6
5
4
3
2
1

I'

 

0

0.812
0.697
0.581
0.466
0.351
0.236
0.121
0.006
-0.110
-0.225
-0.340
-0.455
-0.570
-0.685
-0.801

U .
Uﬁp
0.276
0.237
0.198
0.159
0.119
0.080
0.041
0.002
-0.037
-0.076
-0.116
-0.155
-0.194
-0.233
-0.272

 

Figure 6.36 Phase averaged mean radial velocity for condition ‘K’

1.0 —
0.9 -
0.8 -—
0.7 -
0.6 n
0.5 —
0.4 -—
0.3 —
0.2 -

0.1 —

104

 

 

 

0.0

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Figure 6.37 Phase averaged RMS of axial velocity for condition ‘G’

l

Contour
Number

15
14
13
12
11
10

Awwtscncnwooco

 

0.978
0.925
0.871
0.818
0.764
0.711
0.657
0.604
0.550
0.497
0.443
0.390
0.336
0.283
0.229

 

0.166
0.157
0.148
0.139
0.130
0.121
0.112
0.103
0.094
0.084
0.075
0.066
0.057
0.048
0.039

 

1.0 _
0.9 —
0.8 —
0.7 —
0.6 —
0.5 —
0.4 —
0.3 —
0.2 ..

0.1 —

105

 

 

0.0

 

0.0 0.1 0.2 0.3 0.4 0

I

I

I

Contour
Number

15
14
13
12
11
10

ANOO-hUIODNmO

I

 

 

I |
.5 0.6 0
U. U.
U0 Utip
1.074 0.183
1.023 0.174
0.971 0.165
0.919 0.156
0.868 0.148
0.816 0.139
0.765 0.130
0.713 0.121
0.661 0.112
0.610 0.104
0.558 0.095
0.507 0.086
0.455 0.077
0.403 0.069
0.352 0.060

I I I
.7 0.8 0.9 1.0

Figure 6.38 Phase averaged RMS of axial velocity for condition ‘H’

0.0

 

 

106

 

 

 

 

0.0 0.1

Contour
Number

15
14
13
12
11
10

Amwhmmwmo

Figure 6.39 Phase averaged RMS of axial velocity for condition ‘I’

 

0.2 0.3 0.4 0.5 0

 

1 .032
0.982
0.931
0.880
0.829
0.779
0.728
0.677
0.627
0.576
0.525
0.475
0.424
0.373
0.323

I
.60

 

Utip

0.175
0.167
0.158
0.150
0.141
0.132
0.124
0.115
0.107
0.098
0.089
0.081
0.072
0.063
0.055

   

I I I
.7 0.8 0.9

I
1.0

 

107

 

 

 

 

1.0 -
0.9 —
0.8 -
0.7 -
0.6 —
0.5 -
0.4 —
0.3 —
0.2 —
0.1 —
0.0 I I I I I
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Contour (7 (7
Number ’ x
U0 Uﬁp
15 0.596 0.179
14 0.557 0.167
13 0.518 0.155
12 0.479 0.144
11 0.440 0.132
10 0.401 0.120
9 0.362 0.109
8 0.323 0.097
7 0.284 0.085
6 0.245 0.074
5 0.206 0.062
4 0.167 0.050
3 0.128 0.038
2 0.089 0.027
1 0.050 0.015

Figure 6.40 Phase averaged RMS of axial velocity for condition ‘J’

108

 

 

 

 

 

   

1.0

0.9

0.8 —

0.7 ——

0.6 —

0.5 ——

0.4 —

0.3 —

0.2 m

0.1 a

0.0 I

0.0 0.1 0.2 0.3 0.4
Contour
Number

15 0.523 0.178
14 0.489 0.166
13 0.454 0.154
12 0.419 0.142
11 0.384 0.131
10 0.350 0.119
9 0.315 0.107
8 0.280 0.095
7 0.245 0.083
6 0.211 0.072
5 0.176 0.060
4 0.141 0.048
3 0.106 0.036
2 0.072 0.024
1 0.037 0.013

Figure 6.41 Phase averaged RMS axial velocity for condition ‘K’

1.0—

0.9 —

0.8 —

0.3 —

0.2 —

0.1 -

109

 

 

0.0

 

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Figure 6.42 Phase averaged RMS of tangential velocity for condition ‘G’

I

Contour
Number

15
14
13
12
11
10

ANOO#UIO)NCD<O

I

I
U9

1.062
1.012
0.962
0.912
0.863
0.813
0.763
0.713
0.663
0.614
0.564
0.514
0.464
0.414
0.365

U9

Ulip

0.181
0.172
0.164
0.155
0.147
0.138
0.130
0.121
0.113
0.104
0.096
0.087
0.079
0.070
0.062

0.1 -

110

   
   

 

 

0.0 I I I

I

I

.... I ”mg

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Contour
Number

15
14
13
12
11
10

ANCD-h-U'IOJNQCD

U9

1.190
1.143
1.095
1.048
1.000
0.953
0.905
0.857
0.810
0.762
0.715
0.667
0.620
0.572
0.525

U9
Utip
0.202
0.194
0.186
0.178
0.170
0.162
0.154
0.146
0.138
0.130
0.122
0.113
0.105
0.097
0.089

 

Figure 6.43 Phase averaged RMS of tangential velocity for condition ‘H’

   

0.1 —

111

 

 

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Contour
Number

15
14
13
12
11
10

dNCﬂh-U'IODVGDCD

Figure 6.44 Phase averaged RMS of tangential velocity for condition ‘I’

I

U9

0

1 .136
1 .084
1 .031
0.979
0.926
0.874
0.822
0.769
0.717
0.665
0.612
0.560
0.508
0.455
0.403

 

U9

Utip

0.193
0.184
0.175
0.166
0.157
0.149
0.140
0.131
0.122
0.113
0.104
0.095
0.086
0.077
0.068

112

 

1.0—

0.9 -

0.8 —

0.7 —

0.6 -

0.5 —

0.4 —

0.3 -~

0.2 —

0.1 —

 

 

 

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

 

Contour ~

Number U9 —U—°—
Uo Utip

15 0.597 0.179

14 0.560 0.168

13 0.524 0.157

12 0.487 0.146

1 1 0.451 0.135

10 0.414 0.124

9 0.377 0.1 13

8 0.341 0.102

7 0.304 0.091

6 0.268 0.080

5 0.231 0.069

4 0.195 0.058

3 0.158 0.047

2 0.121 0.036

1 0.085 0.025

Figure 6.45 Phase averaged RMS of tangential velocity for condition ‘J’

113

 

 

 

 

0.2 —-
0.1 —
0-0 I I I I
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Contour ~ ~

Number U9 $—

Uo Ulip

15 0.405 0.138

14 0.381 0.130

13 0.358 0.122

12 0.335 0.114

11 0.311 0.106

10 0.288 0.098

9 0.264 0.090

8 0.241 0.082

7 0.217 0.074

6 0.194 0.066

5 0.171 0.058

4 0.147 0.050

3 0.124 0.042

2 0.100 0.034

1 0.077 0.026

Figure 6.46 Phase averaged RMS of tangential velocity for condition ‘K’

0.3 —

0.2 -

0.1 -

0.0

   

114

 

 

 

I I I

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Contour
Number

15
14
13
12
11
10

ANWAU‘IODVQCD

Figure 6.47 Phase averaged RMS of radial velocity for condition ‘G’

U

I'

U

0

0.965
0.911
0.857
0.803
0.749
0.695
0.642
0.588
0.534
0.480
0.426
0.373
0.319
0.265
0.211

 

I

~

U.-

Ufip

0.164
0.155
0.146
0.137
0.127
0.118
0.109
0.100
0.091
0.082
0.072
0.063
0.054
0.045
0.036

 

115

 

 

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

Contour
Number

15
14
13
12
11
10

Amwhmmumo

U

r

U

0

1 .090
1 .038
0.985
0.932
0.880
0.827
0.775
0.722
0.669
0.617
0.564
0.512
0.459
0.407
0.354

 

 

r

 

Uﬁp
0.185
0.176
0.167
0.159
0.150
0.141
0.132
0.123
0.114
0.105
0.096
0.087
0.078
0.069
0.060

1.0i

Figure 6.48 Phase averaged RMS of radial velocity for condition ‘H’

1.0A

0.9 -

0.8 —

0.7 -*

0.6 —

0.5 —

0.4 —

0.3 —

0.2 -

0.1 —

116

 

0.0

 

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Figure 6.49 Phase averaged RMS of radial velocity for condition ‘I’

Contour
Number

15
14
13
12
11
10

ANOO-bU’ICDNCDO

U

r

U

0

1.047
0.992
0.938
0.883
0.829
0.775
0.720
0.666
0.612
0.557
0.503
0.449
0.394
0.340
0.286

 

r

 

Utip
0.178
0.169
0.159
0.150
0.141
0.132
0.122
0.113
0.104
0.095
0.085
0.076
0.067
0.058
0.049

 

I

0.1

0.0

_

117

 

 

 

| I I
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 I0.8 0.9 1.0

Contour
Number

15
14
13
12
11
10

ANOJ-bU'IO‘JNCDCO

Figure 6.50 Phase averaged RMS of radial velocity for condition ‘J’

U

r

U

0

0.536
0.507
0.478
0.449
0.420
0.391
0.363
0.334
0.305
0.276
0.247
0.218
0.189
0.160
0.131

 

U ,

Utip

0.161
0.152
0.143
0.135
0.126
0.117
0.109
0.100
0.091
0.083
0.074
0.065
0.057
0.048
0.039

 

 

118

 

 

 

 

1.0
0.9 —
0.8 -—
0.7 —
0.6 —
0.5 -
0.4 —~
0.3 -—
0.2 —
0.1 —
0-0 I I I I I
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Contour (7 ﬂ
Number ' '
U0 Utip
15 0.491 0.167
14 0.465 0.158
13 0.438 0.149
12 0.412 0.140
11 0.385 0.131
10 0.359 0.122
9 0.332 0.113
8 0.306 0.104
7 0.279 0.095
6 0.253 0.086
5 0.226 0.077
4 0.200 0.068
3 0.173 0.059
2 0.147 0.050
1 0.120 0.041

Figure 6.51 Phase averaged RMS of radial velocity for condition ‘K’

 

0.0

119

 

 

I I I I I I I |
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

 

Contour (Ex
Number 29f“
15 1.89
14 1.51
13 1.14
12 0.760
1 1 0.377
10 0.000
9 -0.377
8 -0.760
7 -1.14
6 1.51
5 -1.89
4 -2.27
3 -2.65
2 -3.03
1 -3.41

Figure 6.52 Phase averaged axial vorticity for condition ‘G’

\

120

 

 

 

 

1.0-A
0.9—1
0.8—
0.7—
0.6#
0.5—
0.4-
0.3—
0.2—
0.1 —
0-0 I I I I
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Contour (Ex
Number 20!“
15 2.36
14 1.73
13 1.10
12 0.471
11 -0.157
10 -0.785
9 -1.41
8 2.04
7 -2.67
6 -3.30
5 -3.93
4 -4.56
3 -5.19
2 5.82
1 -6.45

Figure 6.53 Phase averaged axial vorticity for condition ‘H’

1.0—

0.9 —

0.8 -+

0.7 -

0.6 —

0.5 -

0.4 —

0.3 —

0.2 -

0.1 —

 

121

 

 

0.0

 

I I | I
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Contour (F:
Number 20]”
15 0.835
14 0.465
13 0.100
12 -0.270
11 -0.634
10 -0.999
9 -1.37
8 -1.73
7 -2.10
6 -2.47
5 -2.83
4 -3.20
3 -3.57
2 -3.94
1 -4.30

Figure 6.54 Phase averaged axial vorticity for condition ‘I’

 

1.0—

0.9 -

0.8 -

0.7 —

0.6 -

0.5 -

0.4 -

0.3 —

0.2 —

0.1 -

 

122

 

 

0.0

I I I

 

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Contour
Number

15
14
13
12
11
10

9

ANOJ-hUIODNCD

(.0

x

20m

 

2.02

1 .60

1 .18
0.754
0.333
-0.088
-0.509
-0.930
-1 .36
-1 .78
-2.20
-2.62
-3.04
-3.28
-3.89

Figure 6.55 Phase averaged axial vorticity for condition ‘J’

123

 

 

 

 

‘ /
1.0 —?>
0.9 2
0.8 --
0.7 —
0.6 —
0.5 —
0.4 —
0.3 -
0.2 —
0.1 —
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Contour (I):
Number 20f”
15 2.59
14 2.21
13 1.82
12 1.43
11 1.05
10 0.666
9 0.283
8 -0.100
7 -0.483
6 -0.867
5 -1.25
4 -1.64
3 -2.02
2 -2.41
1 -2.79

Figure 6.56 Phase averaged axial vorticity for condition ‘K’

7. CONCLUSIONS AND VIABILITY PREDICTIONS

The ACFRD facility was constructed in 1994 for the study of automotive cooling
fans. The experimental equipment and measurement techniques described in Chapter 4
have been developed to aid in the understanding of the ﬂuid mechanical attributes of fan
ﬂows. In addition to this thesis, the data acquired using the ACFRD facility have aided
the Ford Motor Company in the validation of computational models of underhood
cooling systems.

The aerodynamic shroud was conceived in September 1994 by Professor John Foss
of Michigan State University. To the best of the author’s knowledge, the use of an
annular jet to aid in the efﬁciency of fan ﬂows is new and original. A prototype shroud
was assembled, and performance data were acquired. The ability of equation (5-2) and
(5-3) to represent the functional dependence of the data indicate that for a given geometry
both the system performance and efﬁciency depend only on the ﬂow rate through the fan,
and the power input into the shroud.

The system efﬁciency will clearly depend on the design of the shroud delivery
system. Speciﬁcally, losses in the prime mover and ducting system which provides the
aerodynamic shroud with pressurized air will result in lower system efﬁciency. The
losses in the shroud ducting will be proportional to the square of the air velocity. The
aerodynamic shroud should be designed with a relatively small gap height such that a
relatively low ﬂow rate (i.e., low V2) and high pressure can be used to obtain an
equivalent shroud power (x). A prime mover should then be selected which operates

efﬁciently in a high pressure rise - low volume ﬂow rate environment.

124

125

The performance and efﬁciency data were presented in Chapter 5. The results show
enhanced performance and higher ﬂow rates, and degraded performance at lower ﬂow
rates. The ability of the aerodynamic shroud to enhance the efﬁciency of an automobile
cooling system greatly depends on the system resistance (i.e., the operating point is the
intersection of the performance curve with a parabola which deﬁnes the system
resistance). This resistance is typically created by the presence of the grill, air
conditioning condenser, radiator, and downstream blockage elements such as the engine.
The data show that a large system resistance would cause the aerodynamic shroud to
lessen the efﬁciency, whereas a smaller system resistance would allow the aerodynamic
shroud to increase the system efﬁciency.

A 1994 Ford Explorer was investigated to obtain an understanding of the ﬂow ﬁeld
in an actual vehicle. The vehicle was loaned to Michigan State University for a brief
period. As a result very little quantitative were was acquired. However, tuft
visualizations were used to gain a general understanding of the ﬂow ﬁeld in the
underhood compartment. The author estimates that the cooling fan operates in the range
0.10<¢<O.17 at the idle condition. This estimate is based on qualitative comparison of
the ﬂow ﬁeld observed in the underhood compartment using a tuﬁ with results shown in
Chapter 6. It is difﬁcult to determine if the aerodynamic shroud will increase or decrease
the performance based on this information and the data presented in Chapter 5. It is
encouraging to note however, that the idle condition represents the “worst case” operating
condition. A “ram air” effect caused by the relative motion of the air with respect to a
moving vehicle will increase the (I) value of the operating condition. Additionally, many

vehicles exist which are expected to have much less resistance to the cooling air ﬂow.

126

For example, the larger engine compartments found in the heavy truck industry might be
more suitable for the aerodynamic shroud.

For the experiments presented in this thesis a single design was selected for the
prototype aerodynamic shroud. Future work in this area should concentrate on
geometrical optimization of the fan and shroud. The data reported on the velocity ﬁeld
indicate that the strong positive radial ﬂow at the tip of the fan at low ﬂow rates is
causing interacting with the axial Coanda jet and decreasing the system efﬁciency. It is
believed that a proper system design which considers both the fan and shroud geometry
may signiﬁcantly increase the efﬁcacy of the aerodynamic shroud.

In conclusion, the data acquired show the aerodynamic shroud to be a concept which
clearly has potential for automotive, as well as other applications. The data presented
should be used to understand the ﬂow ﬁeld created by the fan/shroud combination.
Continued work should be directed toward the optimization of the system in conjunction

with the application environment.

REFERENCES
Adamczyk, J ., Celestina, M., Greitzer, E., “The Role of Tip Clearance in High-Speed
Fan Stall,” Journal of Turbomachinery, Vol. 115, January 1993.

Ameri, M., and Dybbs, A., “Theoretical Modeling of Coanda Ejectors,” ASME FED-
Vol. 163, Fluid Machinery, 1993

Anderson, D., Tannehill, J ., Pletcher, R., Computational Fluid Mechanics and Heat
Transfer Hemisphere Publishing Corp., New York, 1984.

 

Baranski, B., “Designing the Engine Cooling Fan,” SAE #740691, 1997

Bob], D., “An Experimental Study of the Near Field Region of a Free Jet With
Passive Mixing Tabs,” MS. Thesis, Dept. of Mechanical Engineering, Michigan State
University, 1996.

Dring, R., Joslyn, H., Hardin, L., “An Investigation of Axial Compressor Rotor
Aerodynamics,” Journal of Engineering for Power, Vol. 104, January 1982.

El-Taher, R., “Experimental Investigation of Curvature Effects on Ventilated Wall
Jets,” AIAA Journal, Vol 21, No 11 November 1983.

Foss, J ., Wallace, J ., Wark, C., “Vorticity Measurements ” Instrumentation for Fluid
Dmamics, Joseph Schetz and Allen Fuhs ed., Wiley and Sons, Inc., pp. 1066-1067, 1995.

 

Hirsch, C., and K001, P., “Measurement of the Three-Dimensional Flow Field

Behind an Axial Compressor Stage,” Journal of Engineering for Power, April 1977, p168

Inoue, M., Kumoumaru, M., Fukuhara, M., “Behavior of Tip Leakage Flow Behind
an Axial Compressor Rotor,” Journal of Engineering for Power, Vol. 108, January 1986.

127

128

Inoue, M., Kuroumaru, M., “Three Dimensional Structure and Decay of Vorticies Behind
an Axial Flow Rotating Blade Row,” Journal of Engineering for Power, Vol. 106, July
1984.

Kuroumaru, M., Inque M., Higaki, T.,“Measurement of Three Dimensional Flow Field
behind an Impeller by Means of Periodic Multo-sampling with a Slanted Hot Wire,”
Bulletin of the J SME, Vol. 25, No. 209, November 1982.

Lakshminarauana, B., “Techniques for Aerodynamic and Turbulence Measurements in
Turbomachinery Rotors,” Journal of Engineering for Power, Vol. 103, April 1981.

Lakshminarayana, B., “Methods of Predicting the Tip Clearance Effects in Axial Flow

Turbomachinery,” Journal of Basic Engineering, September 1970.

Lakshminarayana, B., Fluid Mechanics and Heat Transfer in Turbomachinery, John
Wiley and Sons, 1996.

Shakouchi, T., Onohara, Y., and Kato, S., “Analysis of a Two-Dimensional, Turbulent
Wall Jet Along a Circular Cylinder,” J SME Inemational Journal Volume32 No 3, p332,
1989.

Storer, J ., Cumpsty, N., “An Approximate Analysis and Prediction Method for Tip
Clearance Loss in Axial Compressors,” Journal of Turbomachinery, Vol. 116, October

1994.

Van Houten, R., Hickey, R., Kobayashi, 8., “Fan Wake Measurements By Five-Hole
Yaw Tube and Laser Anemometer,” Airﬂow Research and Manufacturing Corporation,

Document 160, September 1993.

Wilson, D., and Goldsteinm, R., “Turbulent Wall Jets With Cylindrical Streamwise

Surface Curvature,” Journal of Fluids Engineering, September 1976.

Appendix A Coanda Jet Velocity Measurements

A single hot wire was used to acquire velocity measurements in the axisymmetric
Coanda jet with the cooling fan removed. This was executed prior to the measurements
presented in Chapters 5 and 6. The purpose of the experiment was to provide the author
with a general understanding of the boundary conditions imposed on the fan by the
aerodynamic shroud. It should be noted that the measured velocity ﬁeld would likely be
greatly altered by the presence of the rotating fan.

Time series voltages from a single anemometer output were acquired for 10 seconds
at 5,000 hz. These data were converted to velocities by the modiﬁed Collis and Williams
equation as discussed in Chpater 4. These data were acquired at a single streamwise
location of x/g=98, where x is the curvlinear distance from the jet exit, and g is the width
of the jet exit. A value of g=1.0mm was used for the data presented. The resultant value
of x=98mm corresponds to the inlet plane of the fan blades. Four values for PIIImm-PIIm
(82, 351, 543, and 817 Pa) were used to characterize the jet. The velocities derived from
this presure reading were used to scale the velocity values.

The proﬁles for the mean and RMS of velocity are shown in Figure A.1. All four
test conditions indicate that the Coanda jet is attached at this streamwise location. The
thickness of the wall jet (y=y(u=0.5um,x)) for all tested conditions is less than the fan tip
clearance of 25mm. It is interesting to note that the proﬁle corresponding to the lowest
plenum pressure is signiﬁcantly different in shape compared to the three higher values. It
is believed that this is a result of turbulent transition of the boundary layers the jet

contraction at larger shroud pressure.

129

0.40
0.35
0.30
0.25

0.20

Velocity

0.15
0.10
0.05

0.00

 

 

 

 

 

' I I r r ' I ' I I
U/Uo + Uo=l I62 m/s _
A " U/U d
""""" T“ o + Uo=24.00 m/s _
A \ -

\
__ + Uo=29.84 m/s __
\
_. + Uo=36.60 m/s -—
I \ J
I— \ _
- "' " .. . '—
‘h‘ ' “ z:- _ __ _..__ __ -
l I I l I 1 l ”“7““ :W-" 7“" '" """
0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.

y (mm

Figure A1 Mean and RMS velocity proﬁles for x=98mm (x/g=98)

 

 

Appendix B Phase Averaged Kinetic Energy

The kinetic energy of the ﬂow was calculated from the mean velocity values presented in
Chapter 6. This quantity was felt to be of interest to those who might consider modeling
or simulating this type of flow ﬁeld numericaly. The time averaged kinetic energy was
calculated from

Emma) = ((7,,2 + (7,2 + (7,2)/2(B-1)
Figures BI and B.2 represent the data calculated from B-l scaled by Utip and U0
respectively. The phase averaged kinetic energy was calculated from
<KE>(x,r,0,a)=(< Ux >2 +<U, >2 +<l7e >2)/2 (B-2)

where the brackets <> represent a phase averaged quantity. These data are shown in

Figures B3 through B7 for conditions ‘G’ through ‘K’, as described in Chapter 6.

131

 

132

 

 

 

 

 

 

 

 

 

 

— 0.70 I I I I I I I l I I I I I I l T
2KE I I 1 l I
U . 2
up 0.60 — G
+
0.50 — + H
+ I
0.40 -
+ J
0.30 — + K
0.20 —
0.10 —
0.00 l l l I l I I I l
0.0 0.2 0.4
Figure B] Time averaged kinetic energy normalized by Uﬁp
_20.0*I'I'I'I'I'I'Irﬁ't'l
2KE
Uo2 -
15.0 - + G ..
—--— H
. I I
+ J
10.0 "
" K I
I
5.0 ‘—
0.O.I.II1II. .III
0.0 0.11 0.2 0.3 0.4 0.9 1 0

 

Figure B.2 Time averaged kinetic energy normalized by U0

133

 

 

1.0 1
0.9 4

0.8 -

 

 

 

 

0.7

 

 

0.6 -

0.5 —

0.4 —

0.3 H

 

 

 

 

 

0.2—
0.1 -
0-0 I I I I I I I T I I
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Contour 2KE 2KE
Number 7:— U;

15 18.2 0.52
14 17.0 0.49
13 15.8 0.46
12 14.7 0.42
11 13.5 0.39
10 12.3 0.36

9 11.2 0.32
8 10.0 0.29
7 8.8 0.26
6 7.7 0.22
5 6.5 0.19
4 5.3 0.15
3 4.2 0.12
2 3.0 0.09
1 1.8 0.05

Figure B.3 Phase averaged kinetic energy for condition ‘G’

134

1.05
0.9 —
0.8 -
0.7 —
0.6 —
0.5 -
0.4 -
0.3 —
0.2 —

0.1 —

 

 

 

0-0 I I I I I
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Contour 2KE 2KE
Number U 02 U;

15 25.2 0.73
14 23.6 0.68
13 22.0 0.64
12 20.4 0.59
11 18.9 0.54
10 17.3 0.50

 

9 15.7 0.45
8 14.1 0.41
7 12.6 0.36
6 11.0 0.32
5 9.4 0.27
4 7.8 0.23
3 6.3 0.18
2 4.7 0.14
1 3.1 0.09

Figure B.4 Phase averaged kinetic energy for condition ‘H’

135

 

0.1 —

 

 

0.0

 

 

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Contour 2KE 2KE
Number U2 U;

15 15.8 0.46
14 14.8 0.43
13 13.8 0.40
12 12.9 0.37
11 11.9 0.34
10 10.9 0.31

 

9 9.9 0.29
8 8.9 0.26
7 7.9 0.23
6 6.9 0.20
5 5.9 0.17
4 4.9 0.14
3 3.9 0.11
2 3.0 0.09
1 2.0 0.06

Figure B.5 Phase averaged kinetic energy for condition ‘I’

136

1.0-

0.9 —

     
    

0.8 —
/

2

0.7—
06—
05—
04—

/\ 4 .
03x

I I I I I I I I I
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

 

 

 

 

Contour 2KE 2KE
Number U: U;
15 1 1 .7 1 .05

14 11.0 0.99
13 10.2 0.92

12 9.5 0.85
11 8.7 0.78
10 8.0 0.72
9 7.2 0.65
8 6.4 0.58
7 5.7 0.51
6 4.9 0.44
5 4.2 0.38
4 3.4 0.31
3 2.7 0.24
2 1.9 0.17
1 1.2 0.11

Figure B.6 Phase averaged kinetic energy for condition ‘J’

137

  
 
     

 

 

 

 

0.1 —
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Contour 2KE 2KE

Number 302— U;
15 8.6 0.99
14 8.0 0.92
13 7.5 0.86
12 6.9 0.80
11 6.4 0.74
10 5.8 0.67
9 5.3 0.61
8 4.7 0.55
7 4.2 0.48
6 3.6 0.42
5 3.1 0.36
4 2.5 0.29
3 2.0 0.23
2 1.5 0.17
1 0.9 0.10

Figure B.7 Phase averaged kinetic energy for condition ‘K’

Appendix C Exit Flow Angle Calculations

The phase averaged ﬂow angle of the air leaving the fan ([32) was calculated from the
equation:

 

<a>)
<I7x>

B, = tan"(

Simple analysis of fans is oﬁen performed using velocity triangles and the relative inﬂow
and exit ﬂow angles. Performance and efﬁciency data can be estimated based on radial
integration this information. These data for the test conditions ‘G’ through ‘K’ (see
Chapter 6) are given in Figures C .1 through C.5 respectively.

138

139

 

 

 

 

/
1.0 —
0.9 —
0.8 —
0.7 ‘1
5
0.6 —
0.5 —
0.4 --
0.3 — ,
0.2 -—
0.1 -
0.0 I
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Contour [32
Number
15 98.9
14 94.9
13 90.9
12 86.9
11 82.9
10 78.9
9 74.9
8 70.9
7 66.9
6 62.9
5 59.0
4 55.0
3 51.0
2 47.0
1 43.0

Figure C1 Phase averaged exit ﬂow angle for condition ‘G’

1.0—

0.9 —

0.8 A

0.7 —

0.6 —

0.5 -

0.4 —

0.3 -

0.2 —

0.1 -

\

140

 

 

0.0

0.0 0.1

Figure C2 Phase averaged exit ﬂow angle for condition ‘H’

I

I I
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

Contour

Number
1 5
1 4
1 3
1 2
‘I 1
1 0

Amwbmmwooco

 

I32

72.4
69.8
67.2
64.5
61.9
59.3
56.7
54.1
51.4
48.8
46.2
43.6
40.9
38.3
35.7

141

 

 

0.0

 

0.0 0.1 0.2 0.3 0.4 0.5 .6 0.7 0.8 0.9 1.0

Contour [32
Number
15 84.2
14 80.5
13 76.8
12 73.1
11 69.5
10 65.8
9 62.1
8 58.4
7 54.7
6 51.0
5 47.3
4 43.6
3 39.9
2 36.2
1 32.6

Figure C3 Phase averaged exit ﬂow angle for condition ‘I’

142

1.0 —1 i
0.9 '—
0.8 —
0.7 -*
0.6 -
0.5 -*
0.4 -—
0.3 A
0.2 -—

0.1 —

 

 

 

 

 

0-0 I I I I I
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Contour I32

Number

15 71.2

14 67.1

13 63.0

12 59.0

11 54.9

10 50.9

9 46.8

8 42.7

7 38.7

6 34.6

5 30.6

4 26.5

3 22.4

2 18.4

1 14.3

Figure C4 Phase averaged exit ﬂow angle for condition ‘J’

143

 

 

 

 

1.0

0.9-

0.8 -

0.7 —

0.6 A

0.5 —

0.4 —

0.3 —

0.2 —

0.1 —

0-0 I I I I I I r I '

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Contour
Number B2

15 84.4
14 79.9
13 75.4
12 70.9
11 66.4
10 61.8
9 57.3
8 52.8
7 48.3
6 43.8
5 39.3
4 34.8
3 30.3
2 25.8
1 21.2

Figure C5 Phase averaged exit ﬂow angle for condition ‘J’

HICHIGAN STATE UNIV. LIBRQRIES
IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII
31293015654415