RD-fl36 896 ROUGHNESS EFFECTS ON COMPRESSOR BLADE PERFORMANCE IN 1/2

CASCADE AT HIGH REYN..(U) AIR FORCE INST OF TECHWJRIGHT-PATTERSON APR OH SCHOOL OF ENGI. F J TANIS

UNCL ASSIFIED NOV 83 AFIT/GAE/AA/83D-23 F/G 20/4 N

1.0 L-' a' .jW

1j.6

136

IIIII L±.

MICROCOPY RESOLUTION TEST CHART

NATKONAL BUREAU OF SIANDARDS-1963-A

I !:

S-..

4..

'S.

#4v

- . ~ .. *.**~*-~.-~ . ..* .- :*..... .

9 ~OF

.

ROUGHNESS EFFECTS ON COMPRESSORBLADE PJEFORMANCE IN CASCADE

AT HIGH REYNOLDS NUMBER

THESIS

FIT/GAE/AA/83D-23 Frederick J. Tanis Jr.2Lt. U.S. Air Force

0.. iiC5-

DEPARTMENT OF THE AIR FORCE

AIR FORCE INSTITUTE OF TECHNOLOGY

Wright-Patterson Air Force Base, Ohio

ow pftlli 84 01 17 071 Xalfwn tMN

i7 S-. %- 1

AFIT/GAE/AA/83D-23

THESIS

lqq

4-%

~ROUGHNESS EFFECTS ON COMPRESSOR

BLADE P FORMANCE IN CASCADEAT HIGH REYNOLDS NUMBER

THESIS

AFIT/GAE/AA/83D-23 Frederick J. Tanis Jr.

2Lt. U.S. Air Force

.14;

Approved for public release; distribution unlimited ,

.94 -!

AFIT/GAE/AA/83D- 23

ROUGHNESS EFFECTS ON COIMPRESSOR

BLADE PERFORMANCE IN CASCADE

"* AT HIGH REYNOLDS NUMBER

THESIS

Presented to the Faculty of the School of Engineering

of the Air Force Institute of Technology

4Air University

in Partial Fulfillment of the

Requirements for the Degree of

Master of Science in Aeronautical Engineering

Frederick J. Tanis Jr.,B.S.

Second Lieutenant, USAF

November 1983

4 Approved for public release; distribution unlimited

-. PREFACE

I would like to thank my thesis committee members, Capt

Wesley Cox, Dr. Milton Franke and particularly my principal

advisor Dr. William C. Elrod for all the time they spent

working with me in order to get this study compleated. I

would also like to thank Maj. John Vonada for all of his

help in getting acquainted with the Cascade Test Facility

-i and keeping it alive throughout the course of this study. I

-am also very gratful for the the support of the members of

the AFIT Fabrication Branch and the technicians in the

Department of Aeronautics and Astronautics, with special

thanks going to Mr. John Brohaus and Mr. Harley Linville. My

deepest thanks go to my wife Melinda for her understanding

and support during my long hours away from home during the

course of this study.

Frederick J. Tanis Jr.

Nils GRA&I,:O DTIC TAB

t Unannounced 0Justif icat lOr

By-Distribution/

Availability Codes

Avil and/or4 Dist Speciali---i -i-% i

ii '-'----'

CO4TENTS

LIST OF FIGURES ................................ iv

LIST OF SYMBOLS ................................ vii

LIST OF TABLES ................................. ix

ABSTRACT. .................. .. . * . * ............. * * x

I. INTRODUCTION.... ... .... ..... . . .. . 1

OBJECTIVES AND SCOPE .............. 2

II. APPERATUS AND PROCEDURE.................. 4CASCADE TEST FACILITY.............o.. 4TEST SECTION. . o. .. . o. ........ ...... 4INSTRUMENTATION ..................... . 5HOT FILM SENSOR CALIBRATION........... 7EXIT PLANE TRAVERSER ................. 10BLADE ROUGHNESS CONFIGURATIONS........ 11DATA ACQUISITION AND PROCESSING ....... 12

III. DATA REDUCTION. ... ... ................. . 14EXIT VELOCITY ...... .... ..... ......... 14TOTAL PRESSURE LOSS................... 16TURBULENCE INTENSITY................... 19

IV. RESULTS AND DISCUSSION................... 20LOCATION EFFECTS.............. ........ 21ROUGHNESS MAGNITUDE EFFECTS........... 37REPEATABILITY............ ............. 48

V. CONCLUSIONS AND RECOMMENDATIONS......... 50CONCLUSIONS. ........ ......... ..... .... 50RECOMMENDATIONS. .......... ........ 50

BIBLIOG RAPHY.o.. ........ ........... .. ... 52

APPENDIX A: ROUGHNESS DEFINITIONS.............. 54

APPENDIX B: VELOCITY AND TURBULENCE INTENSITYPROFILES....................... 57

VI TA........................................... 98

.. , iii

" _ . - -,. -. . . . , ., .- . -. ,. . . .. . . - . .. . . . . . . .. . .-.... . . . . .

WIN

LIST OF FIGURES

Figure Page

1. Test S c i n. . . . . . . . . . . . . . . . . .. 5

2. Hot Wire CalibrationCurve..................... 9

3. Standard Data Window ............... ............. 104. Blade Roughness Configurations ................... 12

5. Roughness Configurations 2 and 3 ................. 21

6. Roughness Configurations 5, 7 and 8 .............. 22

7. Velocity and Turbulence Intensity ProfileConf#1 Eval#6: Smooth Baseline blade Traverse#l.. 26

8. Velocity and Turbulence Intensity ProfileConf#1 Eval#6: Smooth Baseline blade Traverse#2.. 27

9. Velocity and Turbulence Intensity ProfileConf#1 Eval#6: Smooth Baseline blade Traverse#3.. 2b

10. Velocity and Turbulence Intensity ProfileConf#l Eval#6: Smooth Baseline blade Traverse#4.. 29

11. Velocity and Turbulence Intensity ProfileConf#1 Eval#6: Smooth Baseline blade Traverse#5.. 30

12. Velocity and Turbulence Intensity ProfileTraverse#1 Conf#5 Eval#2: 20 micro-meter Ra

Roughness from L.E.-i/4Chord ..................... 32

13. Velocity and Turbulence Intensity ProfileTraverse#l Conf#7 Eval#l: 20 micro-meter RaRoughness from i/8-3/SChord ...................... 33

14. Velocity and Turbulence Intensity ProfileTraverse#1 Conf#8 Eval#l: 20 micro-meter RaRoughness from ./4-1/2Chord...................... 34

15. Velocity and Turbulence Intensity ProfileTraverse#l Conf#2 Eval#l: 27 micro-meter RaRoughness from L.E.-l/4Chord ..................... 35

16. Velocity and Turbulence Intensity ProfileTraverse#l Conf#3 Eval#l: 44 micro-meter RaRoughness from l/2Chord-T.E...................... 36

iv

'".-' 17. Loss in Non-Dimensional Total Pressurewith Ra Roughness ........................... . 39

18. Loss in Non-Dimensional Total Pressurewith Rtm Roughness .............................. 40

19. Loss in Non-Dimensionalpressurewith Rpm Roughness ........................... 41

20. Velocity and Turbulence Intensity ProfileTraverse#l Conf#1 Eval#6: Smooth Baseline blade. 44

21. Velocity and Turbulence Intensity ProfileTraverse#1 Conf#6 Eval#2: 3 micro-meter RaRoughness from L.E.-i/4Chord .................... 45

22. Velocity and Turbulence Intensity ProfileTraverse#1 Conf#5 Eval#2: 20 micro-meter RaRoughness from L.E.-1/4Chord .................... 46

23. Velocity and Turbulence Intensity Profile

Traverse#l Conf#2 Eval#1: 27 micro-meter RaA Roughness from L.E.-1/4Chord .................... 47

24. Arithmetic Average Roughness ................... 54

25. Three different surfaces with equal Ra ......... 54

26. Derivation of Rtm roughness .................... 55

27. Derivation of Rpm roughness .................... 55

28. Velocity and Turbulence Intensity ProfileConf#l Eval#7: Smooth Baseline blade ........... 58

29. Velocity and Turbulence Intensity ProfileConf#2 Eval#l: 27 micro-meter RaRoughness from L.E.-i/4Chord ................... 63

30. Velocity and Turbulence Intensity ProfileConf#3 Eval#l: 44 micro-meter RaRoughness from i/2Chord-T.E .................... 68

31. Velocity and Turbulence Intensity ProfileConf#5 Eval#2: 20 micro-meter RaRoughness from L.E.-l/4Chord................... 73

32. Velocity and Turbulence Intensity ProfileConf#6 Eval#2: 3 micro-meter RaRoughness from L.E.-/4Chord................... 78

33. Velocity and Turbulence Intensity ProfileConf#7 Eval#l: 20 micro-meter RaRughness from I/8-3/8Chord.................... 83

V

-- - - - - --.'.:.,.,:'. 4-..:f:, . : ,-.-..... ..., ..-. . .. .". - . ". . -.. - .• . - . ." .- . .

- - --. - - - -

~D ~ 34. Velocity and Turbulence Intensity ProfileConf#8 Evalil: 20 micro-meter RaRoughness from l/-/Cod.......... 88

35. Velocity and Turbulence Intensity ProfileConf#8 Eval#2: 20 micro-meter RaRoughness from i/4-1/2Chord .................... 93

bvi

C.3. VlciyadTrulneItnst rfl

9.

:0

.4

.-..

a

.. .-. . . . . . . . . . . . . . . .. . . . .' " "% ", w ' %W ,. "" .'"" ,, "V ,," .". -" " "- , ' " - " -"" " -"" "-. " "." C. ." '. -'.," .- . - .- -. -,,' .' ' ."'w '- -':.kI ""

LIST OF SYMBOLS

_Symbol Name units

A Area in

C Chord in

Cp Specific Heat at ConstantPressure B/Ibm-R

dA DifferintialArea in

dY Differintial Length in

Ratio of Specific Heats NONE

L Sample Length micro-meters

m Mass Flow Rate ibm/sec

Nu Nusselt Number NONE

Non-Dimensional Total NONECPressure Loss

P Pressure lbf/in

P Mass Averaged Pressure lbf/in

Ra Arithmatic Average Roughness micro-meters

Re Reynolds Number NONE

Density ibm/c ub ic-f t

Rpm Average Maximum Peak toMean Roughness micro-meters

Rtm Average Maximum peak toValley Roughness micro-meters

s Scaling Factor NONE

T Temperature R

TI Turbulence Intensity percent

V Velocity ft/sec

V Corrected Velocity ft/sec

vii

-,

I Subscripts

1 Inlet

2 Exit

S Static

T Total

Abbreviations

AFIT Air Force Institute of Technology

CTF Cascade Test Facility

HP Hewlet Packard

vii

LIST OF TABLES

Table Page

I. PERFORMANCE DATA FOR ROUGHNESSLOCATION TSS.....................2 3

II. PERFORMANCE DATA FOR ROUGHNESSMAGNITUDE TESTS .................... 38

4-.x

ABSTRACT

"An experimental investigation of the effects of/~

roughness on compressor blade performance in cascade at high

Reynolds numbers was performed. A s-e-ve i'blade cascade of

NACA 64-A905 blades with a chord of 2 in.jand a-n aspect ratio

of 1 was tested in a linear cascade. The blades were mounted

with a stagger angle of 31 deg and arn angle of attack of 15

deg. acing between blades was 1.33 inehee which gave a

solidity of 1.5.

This study was divided into two parts; -first-to

determine the importance of the location of roughness on

blade performanceand .6cond to determine how the degree of

roughness affects blade performance. The velocity and

turbulence intensity profiles of the flow region downstream

of the center blade and the non-dimensional total pressure

loss parameterkC4 were used to evaluate blade performance.

Hot wire anemometery "a-used t& measureithe' velocity and

turbulence intensity profiles downstream of the blade. The

measured exit velocity and static pressure were used to

derive the exit total pressure.

Roughness located near the leading edge of the blade

caused the greatest impact on Oand the velocity and turb-

ulence intensity profiles. Changes in the magnitude of t-he'

roughness did not affect the blade performance until a

threshold was exceeded; then tih- performance decreased

rapidly.

A .x

I. INTRODUCTION

BACKGROUND

The wide operating range of todays military aircraft

engine has lead to a flow regime characterized by a high4%

Reynolds number in the last few stages of the high pressure44

ratio compressor. In this regime the loss in performance is

thought to be a function of roughness alone. This is in

comparison to the low Reynolds number flow where the

efficiency is a function of the Reynolds number and not as

heavily dependent on roughness. The lack of good cascade

data for use in specifing blade surface finishes inorder to

minimize the loss in total pressure across the blade row has

lead to increased research in this area. One performance

parameter that of particular use to the compressor designer

is the total pressure loss coeficient (Ref.15).

In recent years the Air Force Institute of Technology

in conjunction with the Air Force Wright Aeronautical

Labortories, AFWAL/POTC, has been studying the effects of

roughness on compressor blades at a high Reynolds number.

This study was initiated in order to characterize the

effects of roughness on the performance of axial flow

compressor blades in cascade. The ultimate goal would be to

find a technically quantifiable surface finish that will

optimize the thrust specific fuel consumption (TSFC) and the

__ maintenance time of the gas turbine engine. With this

accomplished the Air Force could save millions of dollars in

"4

".$ Z ,' ,,,. ' -"" . ,,; ,' . -", , , " .. - ., -"- . . . ..- "-"- .. -. -

." reduced operating costs.

The research began in iL? 8 1 with the design and

construction of a cascade test facility at the Air Force

Instute of Technology's School of Engineering (Ref.2). The

following year a study of the roughness effects of

compressor exit guide vanes at a high angle of attack was

conducted. This study showed that there was minimal effect

of roughness on the blades because of a large separated flow

region on the suction surface of the blade. The losses from

the separation greatly over shadowed any losses due to the

*roughness (Ref.6).

OBJECTIVE AND SCOPE

The basic objectives of this investigation are asJ.

follows:

* "1. Locate the region on the blade

where roughness affects the loss

in total pressure the greatest.

2. Evaluate how the total pressure

loss coeficient varies with

changes in the magnitude of the

roughness.

The approach used to evaluate the effects that

.k roughness has on blade performance was to build a cascade of

roughened blades then perform a wake survey to evaluate

2

, _ .;. - - . --"- -" • -" ;" . " q - . : " -w -- -- --- - . " -: * " . , " . "

blade total pressure loss in non-dimensional form. In

addition to the total pressure loss coeficient the velocity

and turbulence intensity profiles of the blade wake will be

compared.

The surface roughness parameters chosen for this

investigation are the arithmatic averege roughness, Ra, the

average maximum peak to valley roughness, Rtm, and the

average maximum peak to mean roughness, Rpm (see Appendix A

for roughness definitions).

',°

Fq

3

• *.. . . . .. ,,.., .. ,. . . L . .-.'"" . . . . . . . .*""" ' " "' '- '"' ' '"" '

S.%

- II. EXPERIMENTAL APPARATUS AND PROCEDURE

CASCADE TEST FACILITY

The cascade test facility (CTF) at the Air Force

Institute of Technology School of Engineering was used for

this investigation. This apparatus, which was described in

detail by Allison (Ref.2), consists of an air supply coupled

to a diffuser and stilling chamber with provisions for

mounting a 2 in by 8 in test section. The air supply for the

CTF consists of a 40 horsepower centrifugal blower with a

name plate rating of 3000 cubic feet per minute at 26 onces

of head. The blower brings cool air from outside and mixes

it with warm recirculated air in order to control the test

section temperature. The air is then filtered using an

electrostatic cleaner to remove particles that would be

.etrimental to a hot wire anemometer sensor. A flow

straightening system consisting of two fiber filters and a

honeycomb is located in the stilling chamber. The air is

then delivered to the test section with turbulence

intensities of less than 2%, and a Reynolds number per foot

in excess of two million.

TEST SECTION

The test section, as deplicted in Fig. 1, was a cascade

containing seven NACA 64-A905 blades with a chord of 2 in

and an aspect ratio of 1. The blades were mounted with a

stagger angle of 31 deg and an angle of attack of 15 deg.

4

k..

1,', .','-' , 4,,j",-,-,- -,; ,.. - ,- . , ., .. ,.- . - -,. . , . -,-. , . . . . . -. , - , - - , , - . -

-....

eq

S.

-'S.

'I-

~.S. a

iS~

.4 C1.) 0

.9-)>2. U

S.. ~ \..j\ $ U)4 a

0V.- a

U, .9-)

... a)'i-I .. I-

....... emV.- ~~ (F)

U ~ .. "-4

Cu U)

-~....... LV.- v~..***C ~ C)

~V.- N N~K

-Si

.~S.

U

0 0~f0~ 0 0 0

.9

9%

5

-q -

.9

*%*% , *%%V''''~' *%%S..\ ~ -. *%.**.'*.***'*'** .. e.~,. .. '..:.-.v - *..

This produced a nominal turning angle of 22 deg. The spacing

between the blades was 1.33 in which gave a solidity of 1.5.

The cascade of blades was followed by a 13 in long

adjustable channel that could be used either as a constant

area or a diffusing channel. The top and bottom end walls of

the channel are fitted with static pressure ports that are

used in aligning the exit channel with the free stream

direction. When the end walls were parallel to the free

stream direction, the static pressure at the respective

static ports on the top and bottom end walls were in

agreement and the inlet velocity profile was uniform. In

addition there are static pressure ports in the side walls

of the channel to measure the local static pressure at each

traverse location. The inlet duct to the cascade is also

fitted with static pressure ports that are used to calculate

A ,he inlet velocity and check the profile during alignment of

the exit channel.

INST RUMENTATION

The CTF was instrumented with four pressure

transducers, two thermocouples, a hot wire anemometer and

numerous manometers. The pressure transducers were used to

measure stilling chamber(test section inlet) total, inlet.0

static, exit static, and ambient pressures. The

thermocouples were used to measure stilling chamber total

and control room ambient temperatures. The hot wire

anemometer was used to measure both the X and Y components

6

10-

of the velocity and turbulence (fluxuating velocity) in the

region downstream from the blades. The hot wire system

consisted of a TSI Model 1241-10 hot film X-sensor in

conjuction with two TSI Model 1050 constant temperature

anemometers. Manometers were used to monitor the static

pressure on the top and bottom end walls of the exit channel

to insure that this channel was properly aligned with the

free stream.

HOT FILM SENSOR CALIBRATION

The hot film sensors were calibrated using a

temperature compensation procedure that is described briefly

as follows. First the operating resistance of the sensor was

0 determined for temperature range of intrest assuming an over

heat ratio of 1.5 as recommended by TSI (Ref.16). Once the

operating resistance was determined the anemometers were

set-up and calibration data were recorded at various

velocities and temperatures in the expected test range.

Usually four temperatures and ten velocities were recorded

in the expected range. The following data were recorded at

each calibration point: 1) calibrator total pressure, 2)

calibrator total temperature, 3) exit static pressure

(ambient), and 4) voltage outputs from each channel of the

anemometer. This data provided the necessary information to

determine the velocity, adiabatic wall temperature, heat

transfer rate and mean boundary layer properties of the hot

film sensor. From this information the Reynolds number-.

'."

(based on the diameter of the hot film sensor and mean

boundry layer properties) and the Nusselt number could be

determined. The calibration curve for the sensor is a plot

of the Reynolds number raised to the 0.51 power (Ref.4)

versus the product of the Nusselt number with the ratio of

the mean boundary layer temperature and free stream static

temperatures raised to an exponent that would collapse the

calabration data for the various velocities and temperatures

into a straight line as seen in Fig.2. The exponent used to

collapse the data, Xo, was found by assuming a value and

running the data through a least squares curve fitting

routine until a good fit of the data was obtained. The

exponent was usually found to be approximately 0.03 in the

. range of velocities for this study. Once the exponent, Xo,

was found, the data were plotted and the slope and Y

intercept of the linear calibration curve were found and

stored for future use. In order to use the calibration curve

the voltage from the bridge circuit of the anemometer must

-. be converted to a Nusselt number then multiplied by the

appropriate temperature ratio raised to the Xo power before

reading the Reynolds number from Fig.2. The velocity is then

determined from the Reynolds number obtained from Fig.2

using the appropriate mean boundary layer properties of the

flow in the test section where the hot film sensor is

located. The above procedure was programed into the HP 3052

Data Acquistion System in order to reduce the time required

to calibrate the hot film sensors.

8

IItUr

mm w

It, m mM

-I. I- St -

k -d m

LL. LL &'L -C

~Ew4

Lt - 44 -4+0

0 .

Sb)to

A-Iwi*O * -4st.+x

9

.-

" .. EXIT PLANE TRAVERSER

The CTF includes a computer controled -X-Y axis

traverser that positioned the hot film sensor in the desired

location in the region behind the blades to within 0.002 in

in the X and 0.001 in in the Y directions. The traverser

mount was integrated with the test section design so that

its X-axis could be aligned with the free stream direction

for the flow leaving the compressor blade cascade (See

Fiq.l). Once aliqned the entire flow field behind thecascade could be probed with the hot film sensor.

After some initial testing on a smooth blade, a

standard data window was established behind the center

blade. This data window is comprised of five traverses in

Cthe Y-direction at 1 inch intervals starting 0.25 in behind

the trailing edge of the center blade. Each traverse

,contained 133 points located at 0.01 in intervals beginning

0.6 in below the center blade (Fig.3). The length of each

tr3verse was 1.33 in which corresponded to a single blade

spacing.

4

CENTER

BLADE

Figure 3: Standard Data Window

10

5'. '" . :, - ' 'v ,\ '-' ' .. :, -- , . - ., . -, - -- .-- -. ,. - . - •. . - . . - , ,- .

BLADE POUG9NESS CONFIGURATIONS

The effects of surface roughness were studied in two

parts in order to determine the importance of location and

degree of roughness. In the first part a strip of roughness

was applied at discrete locations. The strip was 1/2 inch

N. wide and its leading edge was located at 1/4 inch intervals

along the chord (Fig.4) . The second part of the study used

blades with roughness applied from the leading edge to the

1/4 chord on the suction surface (Fig.4a). At this location

the magnitude of the roughness was varied from 0.5

micro-meters Ra for the unroughened blades to 27 micro-

meters Ra for the roughened blades. In both parts of the

study, roughness was only applied to the suction surface of

the blade where the boundary layer and wake region are

thought to be greatly impacted by small changes in surface

roughness.

The blades were cast from epoxy and then the surface

was roughened to the desired finish. The roughening was

accomplished by one of two methods. The first method was to

coat the blade, or a portion there of, with a very thin

layer of adhesive and blow on a fine grit. The other method

used was to sand blast the blade to produce cavities in the

surface. When applying the roughness extreme care had to be

excersized as not to alter the shape of the blade. The

majority of the blade configurations investigated were

o. roughened by the first method as a wider range of roughness

4. could be created.

U 11

0 ,, .%'% % % .- - -.. % .. .% . . .. % . o - . . .. . . . ,

.1*r

" .,...

(A) (B) (C)

Figure 4: Blade Roughness Configurations

To measure the Ra, Rtm and Rpm values for the roughened

blades a profilometer manufactured by Rank Taylor Hobson was

used with the parameter and recorder modules attached. The

blades were measured for Ra, Rtm and Rpm at various

locations on the blade and an average of these measurements

was used to characterize the roughness for this

investigation.0DATA ACQUISITION AND PROCESSING

The acquisition and processing of the data were

accomplished using a Hewlet-Packard(HP) 3052A Automatic Data

Acquisition System which is integral to the CTF. The data

acquisition system consisted of a HP 9845B Computer which

controlled a HP 3455A Digital Voltmeter, a HP 3437A System

Voltmeter, a HP 3495A Scanner, and two !P 9895 Disk Drives.

Also integral to the system were a HP 9874A Digitizer, a HP

9871A Daisy Wheel Printer, and a HP 9872S Plotter for input

and output of data. The various software packages designed

for the system allowed for the maximun utilization of the

. systems capability. The data acquisition software allow the

experimenter to specify the number and location of the data

12

b'.

points to be taken. The computer would then direct the

traverser to move the hot film sensor to the desired

location. Once the sensor was in place the scanner would

step the voltmeter throuqh the various channels and record

the voltage outputs from the various transducers. The

voltages were then stored on magnetic storage disk as "raw"

data. The "raw" data were later reduced into usable form as

pressures, temperatures, and velocities and recorded on

magnetic storage disk as "engineering" data for use in

evaluating various performance parameters.

:ez

4°

1

7I T-,V

III. DATA REDUCTION

The parameters used in evaluating the effect of

roughness on blade performance were the exit velocity

profile, non-dimensionalized total pressure loss coeficient,

and the turbulence intensity profile of the blade's wake.

The total pressure loss across the cascade was evaluated

from the pressure and velocity data obtained from the survey

across one blade spacing centered on the center blade in the

cascade.

EXIT VELOCITY

The exit velocity as measured by the hot wire

anemometer experienced small changes in magnitude with

changes in temperature and humidity of the outside air

-supply to the CTF. The changes were first noticed as shifts

in the velocity profiles of two tests of the same blade

configurations on different days. The shift was later

noticed when the exit total pressure calculated from the

measured exit velocity was greater than the measured total

pressure in the stilling chamber. The shift in velocity is

thought to come primarily from a change in resistance of the

hot wire probe support with a change in temperature, and a

change in heat transfer rate with a change in humidity.

Tests run using a single sensor probe support, which had a

lower resistance than the double sensor support primarily

used (0.05 vs. 0.16 ohms), did not show the same size shifts

14

V.::V .. L 7,.-'. , - / .- .i ¢ .-.: .,,? .- ?.',-.'. . . . . .-:. .,..., .. .. , . . .i. . . - . .. , .

that had previously occured when the double sensor probe

support had been used. The 40 deg temperature increase in

the probe support's environment that occured when the sensor

was moved from the calibration (control) room to the test

section would cause an increase in the resistance of the

probe support as the temperature increased. Preliminary

experiments run to define how the resistance of the probe

support changed with temperature were unsuccessful. The

humidity problem has been studied by Larsen and Busch

(Ref.7) with the conclusions that humidity increases the

heat transfer rate between 0.5% and 2%, corresponding to an

increase in apparent velocity between 1 and 5% but no

quanitative relationships were found.

0 In order to correct the hot film velocity data a

continuity check was performed. If the calculated mass flow

entering the cascade based on pressure data differed from

the mass flow leaving the cascade based on the hot wire

data, a correction factor was applied to the exit velocity

data to force continuity through the cascade. The mass flow

rate exiting the cascade was evaluated by numericaly

integrating the following relation:

=fp:z v dy(i

Where fn is the exit mass flow rate, e is the local density

.". evaluated using the ideal gas law, V is the exit velocity as

measured by the hot wire and dy is the differential exit

15

area assuming unit width. The inlet mass flow rate was

evaluated using:

M 1, VA, (2)

Where fi, is the inlet mass flow rate, e is the local inlet

density, A, is the inlet, area and V, is the average inlet

velocity evaluated as follows with the known inlet total and

static pressures and total temperature.

* = 2C'I-(tI (3)

If the derived exit mass flow rate did not equate to

the inlet mass flow rate a scale factor was applied to the

hot wire velocity data as follows to force the equality.

-(4

V2 S~ (4

Where V2 " is the corrected exit velocity, s is the scale

factor that equates the exit mass flow rate to the inlet

mass flow rate, and V2 is the exit velocity as measured by

the hot wire. The scale factor s was evaluated by a

numerical trial and error scheme and was found to vary from

0.90 to 0.95.

TOTAL PRESSURE LOSS

The total pressure loss across the cascade was

16

calculated from the measured inlet total pressure and the

exit total pressure as calculated from the exit static

pressure and velocity. The total pressure at each data point

behind the cascade was calculated using the following

isentropic compressibility relation:

. P + j; T (5)

where V2' is the corrected exit velocity, P is the static

pressure as measured by the wall static pressure ports, and

Ts is the total temperature as measured in the stilling

chamber. The total temperature of the flow was assumed

constant through the test section.

The overall exit total pressure was calculated from a

mass-averaging of the total pressure of the individual data

.points in a single traverse. The mass-averaging was

accomplished with the following relation:

22 ( 6 )

where is the mass-averaged value of the total pressure, P.,

is the total pressure at a single data point in the traverse

being evaluated, 10 is the local density evaluated using the

• -- ideal gas law, and dA is the differential exit area. The

double integrals over the exit area were reduced to a single

17

"% ., -% - q q q . % ". ." .% " " % . . ." .. . . .% " % " . •. .• ..

g°'I

S"integral by assuming no var iations in the span wise

direction of the blade. For unit width of the blade, the

equation then reduces to:

R - _ V_ _ _ (7)

'2

The integrals were numerically evaluated using the

trapezoidal rule with dy being the step size in the

Y-direction (0.01 in).

Having calculated the exit total pressure, the loss in

total pressure across the cascade can be evaluated using:

= (8)IN

The total pressure loss can be non-dimensionalized by the

inlet dynamic head to give (Z, the desired performance

parameter.

APT

V 2 . (9)

where/0 is the inlet density evaluated using the ideal gas

law and V is the inlet velocity evaluated in equation 3.

The total pressure loss was calculated for each traverse and

left an average of all five traverses were taken as the total

pressure loss parameter in this study. It is realized thatV]

18

S S * . . . . . . . ..

-.

mixing losses occur in the wake but the error of the

instrumentation was greater than any mixing losses and an

average loss parameter was thought to be a better

representation of the blade losses.

TURBULENCE INTENSITY

The definition of turbulence intensity for this

investigation is the ratio of the RMS velocity (AC output of

the anemometer as indicated by a TRUE RAS voltmeter) to the

stream wise component of the local mean velocity (DC output

from the anemometer).

TT v , (10)0

Turbulence intensity would not normally be negative but

Pecause of the configuration of the X-sensor probe used to

measure the components of the velocity the two sensors allow

the orientation of the fluctuations to be determined. As a

result, positive value of the turbulence intensity indicates

that the fluctuations are occuring in a plane rotated in a

positive direction about the Z axes from the positive XZ

plane, and ,conversely, a negative Y turbulence intensity

indecates a negative rotation from the positive XZ plane

(Ref.17).

19

- .oJ

F7IV. RESULTS AND DISCUSSION

The objectives of the study were to determine the

importance of position and magnitude of roughness on the

loss in total pressure across a cascade of compressor

blades. Eight blade configurations were tested in the AFIT

CTF with sand type roughness applied at various locations

along the chord and with varying magnitudes. The cascade

used for this study did not use boundary layer removal

therefore the cascade could not be considered two

- dimentional according to Erwin and Emery (Ref.5). The

results being presented are good for making comparisons

between smooth and rough blades but should not be considered

design cascade data.

A. ' The non-dimensional total pressure loss coeficient, C0,

as defined below, was used as the primary means of

evaluating the blade total pressure losses.

Pr -( iY2epv 2

In addition, the velocity and turbulence intensity profiles

N of the blade's wake were studied to evaluate the effect that

roughness had on the wake. The surface roughness was

%..4 evaluated using a profilometer to measure Ra, Rtm, and Rpm

. values of the surface roughness (see Appendix A for

- .. *.*definitions of the above roughness parameters)

20

I .I

,. .- '', LOCAMION EFFECTS

Five blade configurations with roughness on por.tions of

the suction surface were tested to determine the importance

of the roughness location on the loss of total pressure

across the blade. Initially two blades were tested; one with

roughness from the leading edge to the 1/4 chord (Conf#2)

with an Ra value of 26.9 micro-meter, the other

configuration with roughness from the mid-chord to the

trailing edge (Conf#3) with an Pa value of 44.1 micro-meters

Figs.5a&b. The loss in total pressure across the blade, as

R: Conf#2 B: Conf#3

Figure 5: Roughness Configurations 2 and 3

measured by o, increased from 0.0963 for the blade with the

roughness near the trailing edge to 0.1037 for the blade

with the roughness begining at the leading edge. The blade

with the roughness begining at the leading edge thus

experienced an 11% greater loss in Q based on 0 of the

smooth blades. The increase in 0 occurred even though the

magnitude of the rough'ness was less on the blade

configuration with the roughness beginning at the leading

edge compared to the blade with the roughness near the

21

4 . .. -

trailing edge. These tests showed that rouihness located

near the leading edge, on the suction surface, is critical

to the performance of the blade. Further testing showed that

by moving the roughness location away from the leading edge

even small percentages of the chord caused a decreases in a.

Table I shows the decrease in c that accompanies the

movement of the roughness away from the leading edge. For

these tests blade configurations 5,7,&8 were used which

*, .- consisted of a 1/2 in band of 20 micro-meter Ra roughness

applied to the suction surface of the blade. The leading

edge of the strip of roughness was initially at the blade's

leading edge (Conf.5) Fig.6a but was progressively moved

away from the blade's leading edge in 1/8 chord intervals

(Confs.7&8) Figs.6b&c.

%LU7

R: Conf#5 B: Conf#? C: Conf#8

Figure 6: Roughness Configurations 5, 7 and 8

Moving the strip 1/8 chord from the leading edge decreased

a from 0.0973 for the blade with the roughness begining at

the leading edge (Conf#5) to 0.0918. Moving the roughness

__ back another 1/8 chord decreased 0 to 0.0870.

A cause for the high loss in total pressure when

22

W 01) trE0

4 I) 0 3 04-403 W 0 w 0

E .04 I4 U 'E

C.) I0 lam 0a rLoL43 (D 141 - 0 0 > to-~-

0f I U 0' 0-

04 -4-4 0->O4 u-

4 0 Uco r- m -w %D I44-- 10 43 In n %0 r E--4 IEr_1

m- IX I

8 1 r- O O m. '0m i- 1.3 1ir- 4 0 0 D CD

m i4 U) a;4C~ ~ C;J C; C; C; C;

.4. Ci)

to) I- -0

o) F- 0o m rI l n L41 0 01 0 0 0

w I4

u AA

E m~to Lnr c V-4 -

%.0 D-q I- N *- 4

P -4 CN r n r 0Oo2

" .. roughness begins at the leading edge of the blade is that

the size of the roughness is large in comparison to the

boundary layer thickness. On this premise, when the peaks

of the roughness extend through the laininar sublayer of the

boundary layer, a significant increase in the boundary layer

growth results. As the boundary layer grows along the blade,

the peaks of the roughness become submerged in the boundary

layer and their effects on the boundary layer decrease.

Similar effects have been seen on a flat plates where the

coeficient of skin friction decreased as the roughness was

moved away from the leading edge (Ref.1l). Schlichting

points out that the reduction in the coeficient of skin

friction is due to a reduction in the size of the roughness

in comparison to the local boundary layer thickness.

The effects of the location of roughness on the wake

down stream of the blade can be seen by inspection of the

velocity and turbulence intensity profiles in this region.

The measured velocity and turbulence intensity data for the

X and Y directions have been resolved into vectors with

magnitude and direction. The vectors were then plotted using

the point at which the data were taken as the origin and

drawing a vector in the appropriate direction. The solid

line vectors are the velocity vectors and the dotted line

vectors are the turbulence intensity vectors. The magnitude

of the vectors can be determined by measuring the length of

* the vector using the down stream position scale as a measure

of length and applying the appropriate scale factor. The

24

-v o- i.-. * ..

scale factors for the velocity and turbulence intensity

vectors are 540 ft/sec/in and 20 percent/in respectively. As

can be seen in the set of the velocity and turbulence

intensity profiles for the smooth blade (Conf.l),

Figs. 7,8,9,10,&11, both the velocity and turbulence

intensity profiles are fairly uniform in the free stream but

exibit large changes in magnitude and direction in the blade

wake region. The velocity profile exibits a small narrow

decrement accompanied by high levels of turbulence as seen

-. in the traverse located 1/8 chord from the blade's trailing

Oedge (Fig.7). As the traverse location moves down stream

(Figs.8,9,10,&ll) the velocity decrement becomes wider and

shallower accompanied by decrease of the levels of

0turbulence but this is accomoanied by a spreading of the

turbulence into the free stream. All the blade

gonfigurations tested exibited the same trends as the smooth

blades but with changes in the relative sizes of the

velocity decrement and turbulence levels. The following

dicussion of the effects of the location and magnitude of

roughness on the blade's wake will only deal with the

traverse located 1/8 chord behind the trailing edge of the

blade with the understanding that events similar to those of

the smooth blades also occur downstream of the roughened

blades. In order to verify that these trends do occur the

reader is refered to Appendix B which contains a complete

set of the velocity and turbulence intensity profiles for

each configuration tested.

25

.....................*'2..,.-...0 e.. .-

.U-

cr U1 0w.M

>0

0~

r--4 r4tta) I.

JiTl I~

0l 00-1!o

w il 0JpI

ill iii

* I * I p I . p I * c

sra see sil* GSS11E 6V6 6 1(6WHONI NOIJ.IS~d WW3NIS SSONiD

26

* b j - -. u

*Joil

etoll

' .* C'

~ ~II

J~ii1u el

gci m

wEr,4w) rOJ

zu 01~ W

i'i H I--11 U) 3Li ~ ,I'I,,,II~/

cc I~fI~IJJI~jaiI Ii0

I~l ii H

I ~ H I i Mi.1Pl itI I I

66H(63HN ISdW&9SSO

27'

*.*4~~il --4- *. -.- ...-. I I* IP. I I r

r ~ ~ ~ ~ ~ j -1.I IV'-- - ~ .- - ~ - -

. . . . . . . . . . . . . . . . . . . . . .. . . . . . .

zS

M~ 'I> IIi4 4U)E-

4JTM

G)Q

Jill 'Il 1 -II " f m ) -

z~~ ~ ~ ~ If1 1 ;x ) -Ii 1m1mK cnM0 L I

Li i I ~ il'~l~I'1'I'qiiI~ 11111 I i~ f E-4ljj0 4 1I9 ; fJf

.1 i iI ij

Y iN

cc~I~~t1~f

(SHdI NO-4S~ WUr-6 SOJi j

( L

*%p *a:

> hr

A0

"--4

~~~~®M11111 .- %'.f.--i.il:'l 11112'fl 1.. . . . . . . . . . .1111

in.

>LI CL U

u >- (Lz 4j >

-4- (0

4Z00

a w

I: r-4

WI r,~Ia-4

ieii iI [ Ii I

0 44

MSHONZ) NOI.LISOd WU361S. SSONO

29

7-.-77

Uoil

till

LO ice 5

0 10S

z. -w

U; Z 4. Q

0 Cl TI

00i Irv

0 to

* I I if -I! Il O l 0

in' Z m(SHN)NI.Id iJ3.SSO

Y T1 04

U". 30

............................... 0%0

. . . . . . . . . . . . .. . .'. .-.. ..- I

The velocity and turbulence intensity profiles for the

blade configurations where the roughness started" at the

leading edge and was moved away in 1/8 chord intervals

(Conf#5,7,&8) showed a large change in the blade's wake from

that of the smooth blade but did not show any differance

between themselves. This is seen by the large increase in

the depth of the velocity decrement and the increase in the

magnitude of the turbulence levels when Figs.12,13,&14 are

compared to Fig.7. When Figs.12,13,&14 are compared to each

other no significant differance can be seen. The velocity

and turbulence intensity profiles for the blade

configuration that has roughness begining at the leading

edge and extending to the 1/4 chord (Conf#2) and the one

with roughness from the mid chord to the trailing edge

(Conf#3) , Figs. 15&16 respectively, show a large change in

the blade's wake. When these profiles are compared Conf#2

exibits a larger velocity decrement than conf#3 but a

smaller leval of turbulence. A possible cause for the

increase in turbulance intensity, when roughness is applied

from the mid-chord to the trailing edge of the blade, is the

position at which the boundary layer becomes separated. When

the boundary layer experiences roughness after the mid-chord

it may immediately separate due to a unstabilzzing effect of

the roughness on the boundary layer in a unfavorable

pressure region. The separated flow over the blade would

*account for the high turbulence intensity levels. If the

boundary layer experiences roughness before the mid-chord,

31

* *1.7

:AL

cuN

4 ~..J ~ *-

,~ ->

LLm

00 F

w 0d

z-

U..-

QI- )tn a)

.41Z li li 1c 0 4

i laI i iiiii I i*i j

m E

CD C.)$

lit ~~fit

ii4L'L(S3HOD NO1IlSOd WW3N1S SS080

41' ,**

32

4)

*> U

0C-

00

1.4 4J fn

Z 0

*l I i

(S3H~N) NOI I WU11T,4

~~t,

33

FE . w-o - - . -

aS 0 U)lL

0 j

toW C4I

Lii~ INQ)

* 0 >

c- cfG

Li..li

U.E- I II

cc 0i~ ~'I i~~ II I ~ ~ i il ~ ~i1CE'4

ii'jI r

>

* * * * * 4

(S3H3)NI) NOIJISOd W3NIS SSONID

* . 34

CD

Cuj

$0 0ONi

U).> 0d

-4J 0n

(*UZ 44 .~ ij

0 ..............

o

j~~~4 m1 II~ILJ0

00

r Q)

r I wz 0.

1ff E- Ii

JL

a:1

1%* 35

-~~~~~~- - r . . . . , - .

-- ~ r--

---.... ~* -. - - .>

cu

* L40

AL HI

LLI 44L) '

12.4 WI

~t.to1!04

i' il m~ U( ) 4Jt

.~~. (S3H!D111 NOI..LI~ U~L NS~' !-.. ..Ih

'V I36E-

*in

.: :! A - .1

in a region where the flow is accelerating around the blade,

the boundary layer may not separate immediately dud to the

favorable pressure gradient but remain attached a along the

blade. The closer to the leading edge that the boundary

layer becomes detached, the greater the separation region

thus the larger the turbulence levels particularly if it is

unstable.

ROUGHNESS MAGNITUDE EFFECTS

• . Four blade configurations with roughness applied to the

first 1/4 chord were tested to determine the effects of

varying the magnitude of roughness on the loss in total

pressure across the cascade of compressor blades. The

results of these tests are oresented in Table II but can be

seen best in graphical form. Figures 17,18,&19 are plots of

.the increase in non-dimensional total pressure loss that

accompanied an increase in roughness as we hr(? ILy 7 , .ts,

an,! 7 im roughness parameters. Even though the size of the

,- values for the roughness parameters change from one plot to

the next the relative trend remains the same as can be seen

in Figs.17,18,&19. The sand type roughness used in this

study created such a uniform randomness that the method of

evaluating roughness did not affect the trend. For this

reason the following discusion will only deal with Ra values

of roughness. Rtm and Rom values of roughness could have

just as easily been used and would have led to the same

conclusions.

3714

J0 4

410 1I %D N 0)

I 04

CI CD 0 0 0UI c

wz N-41 P- Nr n :

rL0 -4-4r. E

0 - r I r-4 t-o '0 r-4 cotw Q*-e 4j 1- c r) r 0m 0 10 04 C 4 L 0 0c w-

E4 I. a * * G

43 o O 0 0 0II z C;C)t

0 ~ .-I 0nC2L m 0e Ai-'.

43 0 (aWI44'_404 '4 94 M

ca I r-4 %0 t4cq-4 40) -

E4 0.I

- ~I V *0

IPA A r-~38

U) 0f010iLI JZ toa

o2 0

'4. CU

-Aini 0

-~ L

Li

LL 0CE -'

L4-.

4'.I

-V.,-

U) in

M" MG9 W3WO

39

Cu M utu

I'- m' N U)

(45 F- @

oG

L)~ L) (i

I IE

0C'L CuC

0LJ

Z Hd

4.4J

-'. Iji ZO

a. r

0

U)

0G

SUB U 53WO

4440

*~- I- 4 - C4

CY L

M p Q U4> 44 42 -

4~C i'- CV S 0%.

c) -,) v

0)

LiiIT- D -

000 E

.1i)

0u0

xx

xL

.44

The change in total pressure loss across the cascade as

measured by o can easily be seen in Fig.17. As the roughness

magnitude varies from 0.5 micro-meter Ra for the smooth

blade (Conf#l) to 26.9 micro--meter Ra for the roughest

leading edge configuration (Conf#2) c changed from 0.0775 to

0. 1037 . For small values of Ra the change in a was

negligible, as the Pa value of the roughness increased the

non-dimensional total pressure loss, z, showed an increase.

When the z values for the blade configuration with a leadinq

edge roughness of 2.7 micro-meters Ra (Conf#6) is compaired

to that of a blade configuration with a leading edge

roughness of 19.8 micro-meters Ra (Conf#5) a 27% increase in

Itotal pressure loss is experienced. The percent increase is

based on the difference between 4 for the two tests in

comparison to z for the smooth blades (Conf#l) . A further

increase in roughness to a Ra value of 26.9 micro-meters

brings an additional 8.3% increase in non-dimensional total

pressure loss.

The increase in z with roughness is thought to be

caused by the peaks of the roughness extending through the

boundary layer and causing a significant increase in the

growth of the boundary layer. On this premise, when the Ra

value of the roughness was small the peaks of the roughness

did not extend through the boundary layer, thus the

insignificant change in r with roughness as seen in Fig .17.

As the magnitude of the roughness on the blade increases,

the peaks begin to extend through the laminar sublayer of

42.4

.*... . 4., . . . . .. . . ... . .

LI.-. ! .:'. the boundary layer and the loss in total pressure starts to

J:1

increase. The data seems to indicate that there should be

some Ra value of roughness at which point the peaks just

start to penetrate the laminar sublayer of the boundary

layer. At this point the blade's surface transitions from a

hydrodynamicly smooth surface, where the roughness is

submerged in the boundary layer and the performance is not a

function of the roughness, to a hydrodynamicaly rough

surface, where the peaks of the roughness extend through the

boundary layer and large losses in performance are

experienced with changes of roughness. This transition point

could not be investigated due to the lack of sand of the

necessary size. Similar affects of roughness on the total

pressure loss of compressor blades was found in a study by

Schaffler. Schaffler found that the polytropic efficiency of

a compressor decreased as the magnitude of roughness

increased past a critical value. At this point the surface

of the blades transitioned from a hydrodynamically smooth to

a hydrodynamically rough surface (Ref.10).

When the velocity and turbulence intensity profiles are

examined for configurations 1,6,5,&2 (Figs. 20,21,22,&23), an

enlarging of the wake region is noticed. These

configurations correspond to blades with roughness located

from the leading edge to the 1/4 chord with Ra values of

0.5, 2.7, 19.8 and 26.9 micro-meters respectively. When the

• i'~e., profiles of the first traverse of Conf#l (Fig.20) are

compared to those of Conf#6 (Fig.21) no significant

43

0 $4

Ld U4

C.0 0

'34 co

* S * 0

z .... .....

-c 0

Q 0Jill~

0 a4J 0o) vI 155 L

1 '!(!..~II I

Jil $4 %D

'UJ 4jt

i.a:.

>4

* ~ ~ ~ ~ ~ ~ ~ ~ ~ I Jill*. j ~ ' *.* ~ 4 .1 . .

,~ ******- ~*%**** ~I U '.*

cuu

00

if E-4 -4

> 4 -4zz

0- W

0

CLi.

H 11I !J l II

i E-1*iq~ l

(63H~~~~NI)a 44.I~ U3.SSO

cu

cuu.7.O.

d *En

0 a -0

zOb 0i4

Zk 0 4

z cc

0

I f (~1. 0 41 J~

F - I- I ~i ~ j 0

CE WIX.'. .

-LL 4Jj jIC t

II I i4~

IjI . .

>I., .

.7...ii I . i' I4

j~i~~iiH III -**~~ i~ i ll I4' I 4

6 IE 0 IS Ip E WE W~E(SHOD NOI11SOd WW3NiS SSOMO

46

La W .

in1

w 1.4 a

M~t

Cl U)

z

L4C~j 04 L

00z T LL1 J

II, jj III IJIj 4-(if 'jf i/ U~ 1 f Iu

~ I, 0

ifWillii

47I

-E't ST VM ISI SGIaSV> ............

-... .. . . . ... - . .. . .. - . . ... . .

a _.

, differences can be noticed. The velocity decrement in both

cases have the same general shape, characterized, by the

steep gradient on the pressure side of the wake (upper

v portion of the velocity decrement) and the gentle gradient

on the suction side (lower portion of the velocity

decrement) . The depth of the velocity decrement and the

magnitude and direction of the turbulence intensity vectors

are also the same. When configurations 5 and 2 (Figs.22&23)

are compared to configurations 1 and 6 a large difference in

their wake regions is noticed. Configurations 5 and 2, with

their larger magnitudes of roughness (19.8 and 26.9

micro-meters Ra respectively as comoaried to 0.5 and 2.11

micro-meters Ra for configurations 1 and 6), exibit a deeper

velocity decrement with larger gradiants on the sides than

configurations 1 and 6. The turbulence intensity vectors for

configurations 5 and 2 also show an increase in magnitude

over configurations 1 and 6.

.V The velocity and turbulence intensity profiles suggest

that as the roughness level increases there is an

insignificant change in the flow over to the blade until a

threshold is acheived. Once this threshold is passed a

significant increase in the boundary layer growth occurs and

a loss of performance is experienced.

REPEATABILITY

During the testing of the various blade configurations

4 lgreat care was taken to insure constant test conditions

48

. . . . . . . . . . . . . . . . .... . .. .. : ..- .. ...

during a run. It was found practically impossible to

duplicate test conditions from run to run because of 'changes

in the weather that affected the air supply's humidity and

temperature. These changes in the air supply primarily

affected the hot wire system by increasing the heat transfer

from the sensor and the resistance of the probe support.

These changes caused a uniform shift in the measured

velocity data from 3 to 7 percent. This increase in the hot

wire velocity data caused the calculated exit mass flow rate

to be greater than the inlet mass flow rate as derived from

the inlet static and total pressure data. The pressure

transducers having been very stable over the course of the

study were assumed correct and the inlet mass flow

calculated from the pressure data was therefore assumed

correct. To correct the calculated exit mass flow a scaling

".w factor was applied to the hot wire velocity data that caused

the derived exit mass flow to equal that of the inlet.

To estimate the total error of the system duplicate

tests were run on configurations 1 & 8 and comparied for

repeatability. After the mass flow correction factor was

applied to the measured velocity the average performance

parameter, 0, was within 3% for diplicate tests of the same

blade configuration as was the velocity data. The turbulance

intensity data differed by no more than 5% between the

. duplicate tests.

49

V. CONCLUSIONS A'ND RECOMMENDATIONS

CONCLUSIONS

The effect of roughness on the performance of

compressor blades was studied in a seven blade cascade of

NACA 64-A905 blades. The cascade used did not use boundary

layer removal therefore could not be considered two

dimensional according to Erwin and Emery (Ref.5). The

results of this investigation showed that roughness has

significant affect on the loss of total pressure across the

.cascade of blades but the data presented in this report

should not be considered hard fast design data.

Roughness affects the loss in total pressure across the

compressor blades as follows:

1. The greatest loss in total pressure occured with

xoughness located near the leading edge of the blade.

2. The loss in total pressure across the blade is

dependent on the degree of roughness only after a threshold

level of roughness is exceeded.

RECOMMEN DATIONS

A better understanding of how roughness affects the

loss in total pressure across compressor blades could be

accomplished if the boundary layer thickness of the blade

were evaluated and compared to the size of the roughness.

One possible method for evaluating the boundary layer

thickness would be to physicaly measure it with a hot wire

50

, A . .. . .. ... . .. .. . . .

anemometer. Before evaluating the boundary layer thickness a

new test section should be made that employs continuous

boundary layer removal along the side walls of the cascade.

Without boundary layer removal the cascade can not be

-+considered two dimensional according to the NACA report 1016

(Ref.5). In addition to the continuation of the present

investigation, a study of the effects that humidity have on

the heat transfer rate of hot film sensors and the effects

of a temperature change on the resistance of the hot wire

probe support should be initiated.

1'

i,051

BIBLIOGRAPHY

1. Abbott, I.H., von Doenhoff, A.E., and Stivers, L.S.,Jr., Report No 824, National Advisory Committee forAeronautics, p. 258.

2. Allison, Dennis M. Design and Evaluation of a Cascade-' -2 Test Facility, Unpublished MS Thesis. Wright-Patterson

AFB, Ohio: Air Force Institute of Technology, June1982.

3. Bammert, K. and Sandstede, H. "Influences ofManufacturing Tolerances and Surface Roughness ofBlades on the Performance of Turbines" ,Journal ofEngineering for Power (January 1976).

4. Bradshaw, P. An Introduction to Turbulence and ItsMeasurement, New York; Pergamon Press, INC., 1971.

5. Erwin, John R. and Emery, James C.,"Effects of TunnelConfiguration and Testing Technique on CascadePerformance", Report No. 1016 National AdvisoryCommittee on Aeronautics, P. 263.

6. Genovese, David T. Roughness Effects on CopressorOutlet Guide Vanes at H igh R o and Tigh AnleBAttack, Unpublished MS Thesis. Wright-Patterson AFB,M'Tio: Air Force Institute of Technology, December1982.

7. Larsen, Soren E. and Busch, Niels E. "On the HumiditySensitivity of Hot Wire Measurements".DISA InformationNo. 25: 4-5 (February 1980).

8. Perry, A.E. and Schofield, W.H. 'Rough Wall Turbulent

Boundary Layers", Journal of Fluid Mechanics, Vol.37(1969).

9. Pope, Alan. Wind-Tunnel Testing, New York; John Wileyand Sons, INC., 1954.

10. Schaffler, A. "Experimental and AnalyticalInvestigation of the Effects of Reynolds Number andBlade Surface roughness on Multistage Axial FlowCompressors", Journal of Engineering For Power,102:5-13 (Januarg§ -- --"

11. Schlichting, Hermann. Boundary-Layer Theory, New york;i". McGraw-Hill Book Co.,1979.

5• 52

.4

12. Smith, A. and Clutter, D. "The Smallest Height ofRoughness Capable of Affecting Boundary-LayerTransition", Journal of the Aero/Space Sciences,Vol.26 (April lT - - - - - -_

13. Stiles, R.J. Roughness Effects in Axial Compressors-AnEmperical Model-. UnpuETishe Report. WrTight-PattersonAFB, Ohio: Wright Aeronautical Laboratories, 1980.

14. Taylor Hobson. Surtronic 3. Operating Instructions.Leicester, Engla.--nT-T'lor Hobson, (undated).

15. Thomas, M.W. and Evans, D.C. General Electric GasTurbine Fundalmentals Vol II, Ufp'uf61fT7he-report,GenerlElit-F7c Co. h-Na e--O-io, July 1973.

16. TSI,General System Information For 1050 SeriesAnemometry,TSI Incorpora tecT,50-6-Caig-a- , St. Pual,Minnesota, 55164.

17. Vonada, John A. and Elrod, William C. Wake MixingInvestigation of Crenelated Trailing Edge Blades inSCasca e, Unp=W-' led report. Wright-Patterson AFB,7.!!,5i0Mir Force institute of Technology, October 1983.

53

, -. - . - - - -'- -"- --

..-.

."

APPENDIX A: ROUGHNESS DEFFINITIONS

Surface roughness is defined as the finer

irregularities in the repetitive or random deviations from

an intended surface contour (Ref.13). These random

deviations which look like a series of peaks and valleys are

most commonly characterized in terms of "Arithmetic Average

Roughness, Ra" as defined below.

lo- L~~

RIy(12)

Figure 24: Arithmetic Average Roughness

krithmetic Average Roughness does not completly

%characterize the surface for there could be several surfaces

with equal Ra but have different appearances and different

hydrodynamic characteristics Fig.25.

(a) Ra

(b Ra b

(C) V -' obbV .I Ra C

Figure 25: Three different surfaces of equal Ra (Ref.13)

. 5

_n 54

.=- - t " -" " *. %, *

For this reason two other roughness parameters will be

defined Average Maximum Peak to Valley, Rtm; and Average

Maximum Peak to Mean, Rpm.

'4 The Rtm roughness parameter characterizes the surface

by measuring the maximum distance between peaks and valleys

4 1."over several measuring lengths and then averages these

maximums for the total traverse Fig.26.

Am...,, t Ram.., Rnmj.. IA... " -t a m.,

SI .i "" ...I

Rtm =Rmax, Rmax - Rmax 3 Rmax. - Rmax(

J 5

Figure 26: Derivation of Rtm Roughness (Ref.14)

The Rtm roughness parameter defines the maximum average

*peak to valley distance, which is important when studing the

hydrodynamic character of a surface, but does not

distinguish between a surface primarilly of valleys or peaks

as in Fig 25b&c. A third roughness parameter, Average

Maximum Peak to Mean Roughness, Rpm is defined in Fig.27

similar to Rtm but only measuring the peaks of the

roughness. A, f ,oll

p . Rp, + Rp. + Rp, + Rp. + Rp.

Figure 27: Derivation of Rpm roughness

55

'-'"With the three roughness parameters Ra, Rtmn,and Rpm the

surface's hydrodynamic characteristcs can be adequately

-' def ined.

..

*5*!

56

" ' -, . " .. ' . " - - - - , " . -"' - ' - . - . • . . - .' - - - .... -

APPENDIX B: Velocity and Turbulence Intensity Profiles

57

7777__7_72 4 ;7;1

*7 7

zr c

CD_ 0CE _ LI

..U)Ldi

"V -4 (UU) -

IN I.

Hillj ffi, Il mm Lo c1tilI>1.

a. . Ifl 0:

if II .1 pI:ii H! 1~JillHI

Li IiWiL I it ItL

a: I I J I II' i ~11~ J IJJ~I~iI rL

.Wil

(S31HONI NOUlISOd WW3N1S SSONID

58

SUMU

oocv

qii cuIzr

-00

* CU Z

01 OD l

{ i4fj~i 1

411

11II

59

.* ~c K.. K7 17 -11 V-7 0 17

4 44

5's

.5 CZ'm

0i uz C,)'5M

8~kI4I zB0

05 L od o5, .. . . . .. . . ... .. . .. ... ...

.4M Z

CU)it uhf 0~t

VL

li **** t **co

44.

(S31HONI NOI.LISOd WU38.LS SSOMO

60

* C1

Li

to

CEL >.~ UI; Lhi

() U

zz

MM RMR M111T I fi Iffifij I lLflif

I.'. j'd 6 II I f, 61 oJ~~~ Jfiipiiifi W E~ szo UI

MODNIIId WW3NIS SS08

(5..)J~j.I[!I~JI~I 1JI' 1 Iil~I'~'i 1p1J I':P~a

.%

co

> w

* zfilli~iCOZ

CL4)4Ld 2:

c )u

Z~L ___ __ -4)

z fillC 0 4.

I~O III, ifi

lit1 se I IF S IF W VE1O

MSHOD NOIIISOd WU3NJ±S SSONO

62

V~ (U

slot 1

'-44

.5--'

LL0I

w O44

C0

LL U' LL)

-7 iJ 1 -4 , j E p~'~ E c -40

iii fu4j IH

I~~~~~ NIIi .-

SEft 11 Th, ot E V kMI HON N 11~ WW8i Gi '-I

631

cu

I ~0.. a

-,)

-j cm ZZ

%.: I _ U

z Cl 0

C -4

CCD

* cu

MIN LU L I 0

69 IF s ' 61", W s-E W

MHN NI11~ WW81 SS

64Lj

CD

LdI u

0 a'

z 0

0 W >

E-0 C4

z I-

.1~~ 1 f U.

Lri I? o I I 1 '.1 S II 61 0 -W(SHOI)N11S~ W31SSS8

-~ ~Iic~065

fu

oc

> X3,

z

Cuj 0

I-,

LL La 0

0 46U.

Ili jt 4

X I II I

a:

'I 4

ragI f s If? 61 If 661e 6C W(63HONI) NOIIISOd WU381S SS080

66

fig

co

ot

cc VC,

> wj

z

0 U; z

o '-4kk

ji, to r-' th~0 uj >fH'' j

;i'wI'Lj-Y JiI'

cc o

u~cu69 F B IF 61 *f 66*E IVE SI

CS3HOD NOIIISOd WU38I1S SSOND

67

'.g'.

w 4' 0.WW

CE 0

z

w (L) -4

Zt t

-4 0

//. if 54

L) i 0, to 0 wif~~~~~~~~~~ E-4..~'i II;~i~~IiIII

Ij I 0 *3

I . ; I - .;) . .lilt) 0

* . * * .. ***&**Wit

*r I 11 %. k kg k

>. P it

tic')

Cc'

> 3a

00

cr Iz .'-

z~ oii iii

fti E-4111 llt6666S

(S3HJA~fil NO.oS~ W&64SbcSI69 _________ ____-t~j

1. Ri l 1! !- .i l

oo

CD

zC;-z

0

~ Ii

41 E

-. 4J

.'Its

1 ii!70

I !: 4;* ..... * ~. *****v -

-N

InCC r

~w

.p .. ......... ... ... .. .tI cg

zz

I '-

Ij ItI

CC d

ifrr~ ,A'..-..*.......................

(S3HONI) NOIJ.ISOd WU381S SSONID

71

0 1 CD

Ii0 inN

z

LA z

C2 U) Ln citii i ( v

of XoL

'Hil l tj

IL dI '' I

4. 1 1161 1*il:

VV.4-

(S3HOD NOIJ.ISOd WW361S SSONAD

72

ti

0

U') 44

Lflii l w.-4Zfi .... 4

inn

LL- 4 U

U) -4

.4'4

L73

.4,j

0 E%LL

a).

ft t -CDf.f.f..f... t t * f.... f

ttf~tf

ft >

.f-fcf-. t fCE

tf ~ ~ .. '.t-

>t ft ft f

-. to

*1

CU

V. 0

tEy

ilkJ

LI H CmIU

ISE t 61 F WE VE W-V.N) O1S~ U41 S8

L74

ic~p

'CC

'MM &Ii lil a 111hI 41 131 11100

mz*d X'

0

0 CD

I /I. C,,

(S3HND NIIISd WU81S SIN

75H-E .i . .; . .

cuJ.,

00

zz

zLfl 000

a.z rmlfisM ~ im .1,14'.1 di 1.)1)Z

0u

L: r cU~$ 0 j ,

aa: ''I) L) I0'i' Iuil > I~~I iu ..' WL) W.;E?.4II~~~~- fn~i ~:-

Z! (D

(SHNI O.LSd UL1SSO

HI

76

4 a

CO

a

z

00

o:w

S.in zI[ Li I

-YII 0 0 )C* c1'lII 0

X. G4

..... Ii- vir, i r tT

ff IT ME

I I , - j - -, - I I -icI tC 1t o 1f 6 c 6 l r

MHNX I011~ WI81 SSOI I MO'.

'~ 'I77

C;uz gill

-S0 1

> w

to 0zr ii4~0

0; U)Cz*-z

z $4V *

0 tO I4fIi =1

I~ Ii-4I- s4'

0 8 ~' 11J.~q 1tf~1 41 ' I I Ia

z >

1r PR ~ .*~ 1d'> Hq II

MHOINI) NOI.LISOd WW3N.6 66ONO

78

All" !. 9 .. .

P6 16

* .- ~ lE

z U)a0

J.1 (.0 U

9- 4 I-.

If w

II

qj

79

A...~~~ .* * . . .-

CuIu

0

zz0

z 0.-

LiIY !1!u FI~ lhI i i 1d1~~~~l

Cc1~~1~ t1~f ~lIIIIIi~tt1 111111 1

C L) * I **

zG

(63HONI) NOI.LI60d WU3N.9 S6ONO

80

ccJ

.4 0

C;;

4z-

It I

.....I .

0 ~ il I sm m I jI UIlIQtj~f

4 ra ra It II1I s o r

i*.sOND jO1I~ W1:1N* S

81(Y

i --". -. . -

, - tC D

!.7!.P.. cu~u

N 0z

(3: w

G:I :

0 -z ..._ .......

MIM 4j i I IA : :

" w 0

i 63H)NI) NOLLSOd U:.S SSO)

82

- - - Y' : -t "r'; - " ; ' ; - ." ' ." ' ' ' ' ' - , " - " " , .., " ' ' ., ; , .--.. . .. rq.. . . : . .. ,.. . .... . ., ,. . " ' ",'.',, '-''

cu

z

>I LI Ili1%4.J u >4

00Z 4 )0U 4JU

-z I Eal 1* 141 1. 16 kk

II cc w~L LL U4 Q ) 0* >i $4

11 Jb) $4E-

CC 4,,

6660 6~WLiI. SSOI

83w

CC'

>: w

Ldi

o 0

LCL

z.. I I I j '

lit t9' 1q i ~ 1:

Lir ca wiU:

-VY

84&

z w

CcJ. \ ~* *.

, AD-Ai36 896 ROUGHNESS EFFECTS ON COMPRESSOR BLADE PERFORMANCE IN 2/2CASCADE AT HIGH REYN.. (U) AIR FORCE INST OF TECHWRIGHT-PATTERSON AFB OH SCHOOL OF ENGI.. F J TANIS

UNCLASSIFIED NOY 83 AFIT/GAE/AA/83D-23 F/G 20/4 NLlEEllEEEEElIIIIImi

.... L 3.2

111111.25 .4~ j1.

MICROCOPY RESOLUTION TEST CHARTNATIONAL BUREAU OF STANDARDS- I1963-A

%I'

CE,

* 0o;

I- 0

00 u~.

I iu

I'll ~ 1111E-Kas~ M zl

.. L-:

CE * (

x r34*I *

6V 6LBL6 G(SHN)NILSz U~L S~

c85

> *

C;u

*~ :-

> w

00

.4 1tit1 1 1 1 1zz

0 L0 >I cuIi~~~~- m l ~.G.

Li.

me I I

........ ....

Fifmotfi i '11 ',il lM. - I - IiI: f:I

(SHOD NOIIISOd WW3N16 SSONO

86

top S.

z~

cr*

> I

z CD

ain z

LU

Zino If4:1i

)I

.62 ip l 61I .V SZO SGMHOND NLzISi WU3jj II1!f

'I IiiI H87

Cuj

>1z

UC

0 0

.1 >1.

Z 4Jc 40

* 00

I .j I . . .1 1 11 1 1 1 1

IL 1111 ~ !(.) ((up0 liLi

I~~U II I II z-. I 1 i.Iw

-z CCI a) m I~*-IL] ~I1~I~II 'I>

0l toI1I(-- 14 E- %A.

U) -4i~f'I~f'0 4d ''

00 Zt - c:4I;"~~ 4 MttI~ r I

Li I'II IIII

III 0,IIII4I $4 i

S.U

66E BCE BG(S3LNI NfLI c W43.L C)

aal88 C

'-a ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ a go*aU* *~ -' -~~ - .

P .' -70 .5

0 0

CU

z aw' >

IJil II(J

T Ia

iriI if II M -111 6

z6 EI 6E U1 911'.HON 0 IIIII ' I I S EU..

89

-:~-j

00

*$ 0

za rr *0 E-

zL 0 I o

a: Ii "

-~ 1111 1:1 f~!II ~I., Ij

iT .;j Mc i 1i

6901, se6Ip 6 61 E W 6GOEMSHONI) NOI.LISOd WU3N1S SSOO

90

4%.

-T V h* . .r. . . .

Jo

In

0

z

I., zz

0

u~~c *r11-

z a:'I ot 'I

Li alIll ~ I ~ 0

a: 111X IIiwI M II

lj 1 H- II iiiI

CC ' I lR> ' L.~f~W~I IJ i

1 .,il W I I mos~r.ai 6Eb' a 61*sot 666e 614 0 IMP

MSHONI) NOILISOd WW3U±S SSOO

e' P

'IL

*U z.71K.E30

C;~f

t010 jj'-4 U;C

E '

LLI

a L

La. I )-

MHON NIII~ WW)L)SOH

I II I 1i '92

It'

Ice-

re 0!IIz 4- . zUi4)S

I.C1

-44 -,

tt P-40 1*~~U LU.I 4 0

*~ zI(1-4 0~

0 iij I

-a:s IV) i~

-. ~ 0

LZ r~~: ~i1 l~I 4

.1 I I IJI II,

lii'$4u $4E- $-I.~~U E- * I *)I

61 z 66 6a(S3LNI NOII9 W3WSSSN,,

a:3

f'.0-

N

.5'li

5*~0

.5- Cz

5., ~c.

zz

a 0zz

0 0N 0zL w 4-

so cu z1s- G~.Z I "E

5.' iI~ il

Jill* ~i~I~iII i ii JJIJI 3~CD

Li~L~. ~~rII (_

. I * I N T

(S53HONI) NOI.LISOd Wki3I1S S08

94

w*1.'

~z

~.! z4 (1.10

c;S

z ra

(2 Ix .U i

0y -cc -

Jill cu0 m

ILil IIIIL I.-

Il I MR I1H M tl

ill 0 I 6e011 61ol 6 C S? E 6 1Ecu

(63ON O1~ WWNpJSH

95I

M (U

w%. W. -- 1 7

u-u

00

00

LL Jl!iIii U44z IjIj

ii ~ ~ C I..69* e001s l, s e W e II

(SHONI) NOIJ.ZSOd WU381S SSONiD

96

1TO

CCD

z0z

* 0coZ I. .... Vkl

gO 4 0 L

c0 W~I %Ji L

Ln Z

w.. Js r4I

'''i IN0 0CCII l''

* '~ 1 ~fdIOI I4..VX.~

II' ii :*LJ Iz i i I

I'A lI

>'

(SHONI) NOI.LIS~d WU3NIS SSONO

97

4 " • -

VITA

Frederick Johnson Tanis Jr. was born on 29 July 1960 in

Vi: Red Bank New Jersey. After graduating from Middletown High

School South in 1978 he attended Lafayette College in Easton

Pennsylvania. He graduated from Lafayette College Cure Laude

with a Bachelor of Science degree in Mechanical Engineering

in May of 1982. He received his commision in the United

States Air Force Reserve through the Reserve Officer

Training Corps at Lehigh University in June of 1982. As his

first assignment Lt Tanis entered the Air Force Institute of

Technology rt Wright Patterson AFB in June of 1982.

Permanent Address:

500 Newman Springs Rd.

Lincroft New Jersey 07738

98

- .. . - 4 . * 4 V

o,. -. . '

UNCLASSIFIEDSECURITY CLASSIFICATION OF THIS PAGE

"..REPORT DOCUMENTATION PAGE. & REPORT SECURITY CLASSiFiCATION 1lb. RESTRICTIVE MARKINGS

UNCLASS IFIED2&. SECURITY CLASSIFICATION AUTHORITY 3. DISTRIBUTION/AVAILABILITY OF REPORT

2b. DECLASSIFICATION/OOWNGRAOING SCHEDULE Approved for public release;distribution unlimited

4. PERFORMING ORGANIZATION REPORT NUMBER(S) 5. MONITORING ORGANIZATION REPORT NUMBERIS)

AFIT/GAE/AA/83D-23

68. NAME OF PERFORMING ORGANIZATION 6b. OFFICE SYMBOL 7a. NAME OF MONITORING ORGANIZATION(If applicabie)

School of Engineering AFIT/ENY

6c. ADDRESS (City. State and ZIP Code) 7b. ADDRESS (City. State and ZIP Code)

Air Force Instute of Technology

Wright-Patterson AFB, Ohio 45433

Be. NAME OF FUNDING/SPONSORING 18b. OFFICE SYMBOL 9. PROCUREMENT INSTRUMENT IDENTIFICATION NUMBERORGANIZATION (If applicable)

8c. ADDRESS (City. State 9nd ZIP Code) 10. SOURCE OF FUNDING NOS.

PROGRAM PROJECT TASK WORK UNITELEMENT NO. NO. NO. NO.

11. TITLE Ilnclude Security Claaification)See Box 19

- , 12. PERSONAL AUTHOR(S)

Frederick J. Tanis Jr., B.S., 2d Lt, USAF13a. TYPE Of REPORT 13b. TIME COVERED 14. DATE OF REPORT (Yr., Mo.. Day) 15. PAGE COUNT

MS Thesis FROM _ TO _ 1o_ 1983 December 10816. SUPPLEMENTARY NOTATION O: 11w 14211 111-17.

1?. COSATI CODES 6S. SUBJECT WSPUjSaCuaktnI~i&@,JI necezaary and identify by block numb.r)FIELD GROUP SUB. GR. Cascade Testing, Compressor Blades, Roughness Effects21 1 High Reynolds Number, Total Pressure Loss

1. ABSTRACT (Continue an reuerme if neceuary and identify by blocl number)

Title: Roughness Effects on Compressor BladesPerformance in Cascade at High Reynolds Number

Thesis Chairman: Dr. William C. Elrod

20. OISTRIBUTIONIAVAILABILITY OF ABSTRACT 21 ABSTRACT SECURITY CLASSIFICATION

-'JNCLASSIFIEO/UNLIMITEO 1 SAME AS RPT. C3 OTIC USERS 0 UNCLASSIFIED

221. NAME OP RESPONSIBLE INDIVIDUAL 22b TELEPHONE NUMBER 22c. OFFICE SYMBOL(include Arova Code)

Willia C. 'lrod 513-255-3517 AFIT/ENY

D FORM 1473, 83 APR EDITION OF I JAN 73 IS OBSOLETE. UNCLASSIFIEDSECURITY CLASSIFICATION OF THIS PAGE

......................".............".......".........."...............'...;.."..,...'-.......-........'-.....-..-.........,.................".......

.r ,.o':o j- . ,.. . ozt - , ,. . . . . . ... --' * --. ... - - - • . o , oo.

UNCASSIFIED

SECURITY CLASSIFICATION OF THIS PAGE

-'- An experimental investigation of the effects of roughness on compressorblade performance in cascade at high reynolds number was performed. A sevenblade cascade of NACA 64-A905 blades with a chord of 2 in and an aspect ratioof I was tested in a linear cascade. The blades were mounted with a stagger angleof 3t deg and an angle of attack of 15 deg. The spacing between the bladeswas t.33 in which gave a solidity of 1.5.

This study was divided into two parts, first to determine the importanceof the location of the roughness on the blade performance and second todetermine how the degree of roughness affects blade performance. The velocityand turbulence intensity profiles of the flow region down stream of the center

S.blade and the non-dimensional total pressure lossparameter, r, were usedto evaluate blade performance. Hot wire anemomete!y was used to measurethe velocity and turbulence intensity profiles down stream of the blade. The

* measured exit velocity and static pressure were used to derive the exittotal pressure.

Roughness located near the leading edge of the blade caused the greatestimpact on Q and the velocity and turbulence intensity profiles. Changesin the magnitude of the roughness did not affect the blade performanceuntil a threshold was exceeded then the performance decreased rapidily.

..-.-

-. 5-..

.5

S...5

.. UNCLASSIFIED

" SECURITY CLASSIFICATION OF THIS PAGEv ..> :-. .-.... " - a .. - . . . .- .. - . . . ..

I.4r.

4-j4

e 7. 7

94 '1F

* * 'a, 4