A COMPUTATIONALLY EFFICIENT MODELLING OF ... - NTRS - NASA

0 downloads 0 Views 5MB Size Report
measurements of Hoheisel et al. [1984], o_= 0° ........ 136. (a) Comparison of predicted and measured pressure distribution for the Wortmann. FX 63-137 airfoil.
A COMPUTATIONALLY

EFFICIENT

LAMINAR

SEPARATION

FINAL

MODELLING BUBBLES

REPORT for

NASA

Paolo

Dini

Department The

Grant

and

Mark

D. Maughmer

of Aerospace

Pennsylvania 233 Hammond University

NAG-I-778

Park,

July

Engineering

State University Building PA

1990

16802

OF

ABSTRACT

In predicting the aerodynamic characteristics of airfoils operating at low Reynolds numbers,it is often important to accountfor the effects of laminar (transitional) separationbubbles. Previous approachesto the modelling of this viscous phenomenonrange from fast but sometimesunreliable empirical correlations for the length of the bubble and the associatedincreasein momentum thickness, to more accurate but significantly slower displacement-thicknessiteration methods employing inverse boundary-layer formulations in the separatedregions. Sincethe penalty in computational time associatedwith the more general methods is unacceptable for airfoil designapplications, useof an accurateyet computationally efficient model is highly desirable. To this end, a semi-empiricalbubble model has been developed and incorporated into the Eppler and Somersairfoil design and analysis program. The generality and the efficiency have been achievedby successfully approximating the local viscous/inviscid interaction, the transition location, and the turbulent reattachment processwithin the framework of an integral boundary-layer method. ---%

Comparisons surements numbers

of th%i_redicted for several from

2,000,000

airfoils down

aerodynamic show

excellent

to 100,000.

characteristics

with

and

agreement

consistent

experimental

mea-

for Reynolds

PRECEDING

PAGE

BLANK

NOT

FILMED iv

TABLE

OF

CONTENTS

LIST

OF

FIGURES

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

vi

LIST

OF

SYMBOLS

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

xi

Chapter 1

2

INTRODUCTION

.

°

o

o

.

°

....

°

..........

1

Previous

Models

7

A Model

for Airfoil

LOCAL

The THE

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

Local

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

9 13

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

Independent

13 15

Parameters

PART

OF

Distribution

THE

Controlling BUBBLE

the

Bubble

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

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

of the

TRANSITION Criteria

Approximate

Envelope

Parameter

Governing Closure

40 43

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

43

d'

Method

Flowchart

47 52

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

57

....................... PART

OF

THE

62 BUBBLE

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

Equations ....................... .............................

Supplementary

Functions

29

39

Criterion .................... .........................

Error

TURBULENT

Scaling

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

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

Eppler's Transition The e n Method Drela's

Singularity

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

Empirical

22

33

Separation

Angle

......

29

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

Separation

Model

.

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

LAMINAR

Removal

THE

.

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

Philosophy

Results

Closure

The

Design

PARAMETERS

Pressure

5

.

1 3

Early

4

.

Background and Definitions .................... Problem Statement ........................

Modelling

3



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

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

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

67 67 72 73 77 89

v

6

w

RESULTS

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

NACA

663-018

NASA

NLF(1)-I015

Airfoil

Eppler NACA

E387 Airfoil 65-213 Airfoil

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

Wortmann 7

Suggestions APPENDIX: REFERENCES

w

63-137

CONCLUSION Summary

r

FX

Airfoil

93 ......................

Airfoil

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

Conclusions

for Future MODEL

Work

SUMMARY

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

98 108 135

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

......................... and

93

140 141

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

141

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

142

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

144 150

vi LIST

OF

FIGURES

Page_ Fig. I-I

Schematic

of bubble

and

Fig. 2-1

Prediction

of bubble

pressure

and Fig.

2-2

Stratford

the

the

[1989],

using (52)7

(H32)n 2-3

Fig.

2-4

Comparison

3-1

3-2

Green's

Fig.

3-3

Comparison

Fig.

3-4

3-5

Green

4-1

=

Fig.

Fig.

4-2

4-3

of the and

Comparison

model

with

and

Mueller

[1988]

turbulent

boundary-layer

Transition

at laminar

turbulent

Reynolds

number.

. 20

boundary-layer Transition

at 24

part

of the

parameter

bubble

as function

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

velocity

shape-factor

dissipation

profile

correlation

32 .........

for the

34

Falkner-Skan

laminar

coefficient

laminar laminar

correlations

for the

Eppler

region

singularity

shear

boundary-layer inside

for the 37

demonstrating

the

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

41

derived

transition

layer

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

lengths E387

correlation

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

separation

Goldstein

development

36

profiles

of two empirically

Eppler

those

recovery,

McGhee

Drela

reversed

of the

layer

Drela

laminar

Green

separated

Modified

Stratford

are from

number.

and

Comparison

shear

data

bubble

with

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

Comparison

for the

U_r = Us,

gradient

of the

of the

obtained

23

in the

profiles

of the

airfoil

compared

19

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

Falkner-Skan Close-up

program

transition

and

Eppler

two-parameter

removal Fig.

(2.4),

Eppler

pressure

Fig.

E387

of Horton .....

by Schmidt

at an intermediate

recovery

Eppler

5

[1988]

the

of the

Pressure

McGhee

incorporating

Comparison

separation

from

Somers

Experimental

laminar

Comparison

Cp,

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

and

and

distribution

inviscid Eq.

of the

velocity

program

separation

of Gaster's

Fig.

from

are for the

at a low Reynolds

methods

Fig.

the

= 1.51.

methods

Data

Eppler

characteristics:

on the

distribution.

characteristics

original

obtained

Fig.

recoveries.

Aerodynamic with

of its effect

predicted airfoil

45 by the

two

at R = 300,000

development the

correlations

bubble

plot ...........

showing

.....

46

the 49

vii u

Fig.

4-4

Comparison

of Eppler's

amplification

surface

Fig.

4-5

Neutral

curves

Fig.

4-6

Amplification

Fig.

4-7

Growth

curves

5-i

Normalized

Fig.

5-2

Comparison

shape of the

and

the

present

5-3

Summary

of bubble

Fig.

5-4

Example

of iteration

distribution

Fig.

5-6

shape

recovery

pressure

of bubble

Fig.

5-8

Variation transition

of supplemetary .........................

NACA 6-2

NACA Fig.

6-3

those

interactive the

........

65 ......

70

of Eppler, 78 parameters with

obtained

in governing

......

81

inviscid

equations

affecting

with with

laminar

length

bubble

height

. 88 at 90

and

airfoil,

laminar

measured

a = 0 °. (b)

a = 0 °. Data

are

thickness

original

obtained

with

program

XFOIL

Viscous

of the Eppler the

Pressure

by Gault

present of Drela, Tunnel,

analysis [1955] the

at laminar by Gault

NASA and

. 91

distribution

inside

transition are

bubbles

for the

growth

a = 0 °. Data

the

separation pressure

development

by assuming

airfoil,

Low-Turbulence

velocity

87

parameters

characteristics with

on bubble

at transition

of momentum

Aerodynamic with

airfoil,

663-018

obtained

64

86

boundary-layer

663-018

that

.....

function

to merge

for modelling

663-018

and

Comparison with

method

scaling

number

of predicted

NACA

summary

factor

83

height

of procedure

Comparison

for the

63

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

Variation

(a)

........

development

correlation

and

terms

5-7

6-1

. 51

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

the

Fig.

shape

on undershoot

of different

Flowchart

of the

flow

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

geometry

Comparison

5-9

factor

recovery

function

Fig.

Fig.

factor

Effect of decreasing Reynolds distribution ........................

Fig.

shape

envelope

turbulent

line,

for separated

boundary-layer

erroneous factor

Fig.

5-5

values

the

Dashed

criteria

of the

for two

with

profiles.

attempted

for two values

to Drela's

Fig.

Fig.

of the

of n in a non-similar

compared

criterion

for self-similar

for H12 > 4, is one

Drela,

transition

..... bubble

separation.

[1955]

.....

NLF(1)-1015

Somers bubble

R = 500, 000. NASA

LaRC,

96

airfoil

program model

94

compared

and Data 1987

with are

the from ....

99

viii w

Fig.

6-4

(a)

Comparison

for the

NASA

is XFOIL

Tunnel,

summary

and

compared airfoil, 6-5

(a)

with

line is XFOIL

Viscous

summary

analysis

m

Fig.

6-7

NLF(1)-1015

the

Low-Turbulence

the

those

(a)

[1988].

development 6-9

(a)

Comparison

for the from

Eppler

McGhee

boundary-layer a = 1.5 ° Fig.

6-10

(a)

Comparison

for the from

et al.

Comparison

et al.

Fig.

with

Eppler McGhee

boundary-layer a = 7.4 °

the

and

upper

(b)

present

XFOIL

of Drela, .................

Eppler

of predicted

and

summary airfoil,

.........

106

airfoil

obtained

compared

with

and

the

with Data

are

[1988 I. (b) Viscous

pressure

distribution

are

McGhee

from

and

boundary-layer

a = -2 °

measured

et al.

pressure

Experimental

......

analysis

110

distribution

a -- 2 °. Data

Eppler

are

summary

E387

and

airfoil, 112

and

measured

E387

airfoil.

Experimental

et al.

[1988].

(b)

development ........................

1987.

development

model

measured

for the

LaRC,

109

airfoil.

of predicted

a = 6 °. Data

R = 300,000.

E387

development ........................

distribution

NASA

E387

a = -2 °. Data E387

104

pressure

program

bubble

at

..........

Eppler

[1988]

analysis

prediction

a = 4.8 °

XFOIL

and

boundary-layer

Tunnel,

Somers

are from 1987.

boundary-layer

of the

airfoil,

surface

Experimental

and

and

Data

LaRC,

airfoil

airfoil,

Viscous

for the

with

distribution a = 2 °.

NASA

Pressure

of predicted E387

pressure

at a = 2 °.

Tunnel,

summary

Eppler

program

Eppler

NLF(1)-1015

experimental

airfoil.

characteristics

McGhee

for the

measured

and measured

NLF(1)-1015

original

interactive

6-8

analysis

obtained

from

of predicted

NASA

NASA

102

NLF(1)-1015

NASA

Aerodynamic with

Fig.

NASA

Comparison

for the

analysis

development

for the

at a = 1° compared

Viscous

line

Low-Turbulence

(b) Viscous

prediction

(b)

are from

the

1987.

airfoil,

Pressure

(b)

from

and

Low-Turbulence

for the

a = -5 °. Dot-dashed

prediction

the

(a)

distribution

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

NLF(1)-1015

a = 2 ° for the

pressure

boundary-layer

of predicted

NASA

development

6-6

are

LaRC,

surface

XFOIL

Comparison

Dot-dashed

Fig.

Data

upper

measured

airfoil,

NASA

a = -5 °

for the

and

NLF(1)-1015

prediction.

Pressure

Fig.

of predicted

Viscous

for the

pressure

distribution

a = 9 °. Data analysis

Eppler

E387

summary

are and

airfoil, 114

ix

Fig.

6-11 Aerodynamic with

Fig.

6-12

the

characteristics

original

obtained

with

Data

are from

McGhee

Comparison

for the from

McGhee

6-13

(a) from

6-14

(a) from

Fig.

6-15

for the

of predicted

for the

and

Experimental

et al.

[1988].

(b) Viscous for the

from

McGhee

boundary-layer a = -1.6 ° Fig. 6-17 (a)

Comparison

for the from

Eppler

McGhee

boundary-layer a = -0.5 °

distribution

analysis

and

et al.

[1988].

(b) Viscous

are and

airfoil,

pressure

distribution

o_ = 7 °. Data analysis

Eppler

are

summary

E387

and

airfoil, 122

and

Eppler

Somers

present

of Drela,

[1988]

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

of predicted

and

and

with Data

with the are 124

airfoil.

Experimental

et al.

[1988].

(b)

of predicted

obtained

compared

model

measured Viscous

for the

airfoil

R = 100,000.

E387

development ........................

E387

program

bubble

XFOIL

pressure

distribution

a = -2 °. Data analysis

Eppler

are

summary

E387

and

airfoil, 125

and

measured

E387

airfoil.

Experimental

et al.

[1988].

(b)

development ........................

distribution

summary

E387

measured

Experimental

et al.

airfoil,

pressure

Eppler

airfoil.

the

and

a = 0 °. Data

E387

with

are

120

of predicted

Eppler

E387

measured

airfoil.

program

Eppler

pressure

Eppler

E387

original

Comparison

117

118

of the

(a)

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

analysis summary

characteristics

McGhee

R = 200,000.

(b) Viscous

Aerodynamic

from

with

[1988].

development ........................

obtained

model,

measured

for the

interactive

compared

et al.

Eppler

the

obtained

a = -2 °. Data

McGhee

those

6-16

and

boundary-layer development = 6° .........................

with

Fig.

[1988]

airfoil

Experimental

Comparison

for the

program

bubble

et al.

E387

airfoil.

Eppler

McGhee

Somers

present

development ........................

boundary-layer = -0.5 ° Fig.

the

Eppler

E387

Comparison

for the

and

of predicted

Eppler

boundary-layer a = -1.8 ° Fig.

Eppler

those

(a)

of the

Viscous

for the

pressure

distribution

a = 0 °. Data analysis

Eppler

E387

summary

are and

airfoil, 127

w

Fig.

6-18

(a)

Comparison

for the

m

prediction (b)

of predicted

Eppler

E387

Viscous

analysis

6-19

(a) from

Fig.

6-20

MeGhee

and

Experimental

et al.

[1988].

(b) Viscous

Experimental

et al.

[1988].

(b) Viscous

boundary-layer development c_ = 6 ° .........................

for the

(a)

and

Hoheisel

upper

surface

airfoil,

are

of Hoheisel

Comparison

for the from

summary Wortmann Doppler [1988],

[1984].

FX

Brendel and FX

and

upper 63-137

Velocimeter a = 7°

are

summary

E387

and

airfoil,

distribution

e = 7 °. Data analysis

Eppler

measured

Experimental

to the et al.

63-137 Mueller

surface airfoil,

and

are

summary

E387

and

airfoil,

pressure analysis

[1984],

o_ = 0 °

measured

........

136 distribution

a = 7 °. Data

(b) Viscous

of Brendel

65-213

Velocimeter

pressure

a = 5 ° compared

and

NACA

Experimental

boundary-layer

are

summary

for the Doppler

[1988].

distribution

c_ = 0 °. Data

Laser

airfoil.

measurements

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

distribution

pressure

development

of predicted

Wortmann

at 129

analysis

(b) Viscous

boundary-layer

measurements (a)

airfoil.

o_ = -1 ° compared

prediction

133

of predicted et al.

boundary-layer

pressure

measured

airfoil.

6,5-213

[1988].

c_ = 5 °. Data

Eppler

E387

NACA

XFOIL

is XFOIL

131

of predicted

Comparison

et al.

surface

measured

airfoil.

Eppler

distribution line

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

and

from

6-22

with

airfoil

(a)

McGhee

McGhee

E387

Comparison

pressure

upper

for the

for the

Fig.

and

boundary-layer development cr = 4 ° .........................

from

6-21

E387

of predicted

Eppler

for the

Fig.

from

summary

Eppler

Comparison

for the

are

at oL = 1 ° compared

o_ = 2 ° for the

measured

_ = -5 °. Dot-dashed

at a = 2 °. Data

development

Fig.

and

airfoil,

analysis

development to the and

for the Laser

Mueller 138

xi LIST

b

width

of separated

C

airfoil

chord

Cd

section

cf

skin-friction

Ct

section

lift coefficient

Crn

section

moment

f

disturbance

h

distance

h_r

height

gl

laminar

g2

turbulent

length

Y/2

exponent

of velocity

FZ

linear

stability

p

static

pressure

r

decay

rate

drag

OF

shear

SYMBOLS

layer

coefficient coefficient

coefficient

frequency

length

quarter-chord

point

in Hertz

of separated of dividing

about

shear

layer

streamline

of the

from

airfoil

surface

at transition

bubble

of the

bubble

distribution

theory

power

amplification

of dissipation

function

downstream

coordinate

It

streamwise

velocity

inside

the

tt 1

streamwise

velocity

above

turbulent

shear

layer

112

streamwise

velocity

below

turbulent

shear

layer

tt I

streamwise

fluctuating

u,

friction

v I

l

turbulent normal amplitude

velocity

velocity, Reynolds fluctuating

_0 stress velocity

of fluctuating

velocity

stagnation boundary

flows

In (_-0)

streamwise

--pttlv

the

for Falkner-Skan

factor,

coefficient from

law

point layer

of reattachment

xii

X

distance

along

airfoil

y

normal

Ao

disturbance

A1

amplitude

of turbulent

H32-distribution

upstream

A2

amplitude

of turbulent

Hs2-distribution

downstream

B

van

CD

dissipation

distance

chord

from

the

amplitude

Ingen's

surface

at neutral

separation

angle

of reattaehment of reattachment

= 17.5 o,, - pu'v')_-_ydy -_y

coefficient,

C_-

maximum

DU

inviscid

velocity

F

reduced

frequency,

G

amplitude

Gt

turbulent

H12

boundary-layer

shape

factor,

H3:

boundary-layer

shape

factor,

P

Gaster's

R

chord

turbulent

shear

stress

decrease

coefficient,

as/71

_

modified

pressure

wake shape

gradient

parameter,

length

Reynolds

number,

scaling

Tu

turbulence

U

velocity

U_

freestrearn

Ot

angle

thickness

factor

Reynolds

to match

in Green's

profiles

62

u_

SF

....

parameter

number,

momentum

U 2/

oe

function

Reynolds

R62

_-

U=

of Coles's

laminar

_,5cr p

AS

Ks_t_ v number,

curvature

u_Ax8 V

of H32-distribution

at reattachment

intensity at the

Falkner-Skan

edge

of the

boundary

layer/inviscid

velocity

velocity

of attack

wavenumber

#

constant, ] 3 fo(# 1/2ou

coefficient,

pressure

stability

relative

to the

of sinusoidal laminar

chord

line

disturbance

pressure

gradient

parameter,

_s

dU ds

w °,,

Xlll

t

Clauser

turbulent

pressure

gradient

7

separation

6

boundary-layer

thickness

61

boundary-layer

displacement

62

boundary-layer

momentum

63

boundary-layer

kinetic

6

spreading

of turbulent

angle

angle

nondimensionaI #

viscosity

1/

kinematic

parameter,

thickness,

energy

velocity

fo (1- _) dy fo _ (1- _) _y

thickness,

U

thickness, shear

ratio

1-

layer

of turbulent

shear

layer,

of air viscosity

of air

dimensionless

streamwise

(FS

Falkner-Skan

characteristic

P

density

Tw

wall

CO*

radian

A

turbulent

defect

h_

Horton's

universal

coordinate

inside

boundary-layer

the

bubble

thickness

of air

shear

stress

frequency

of sinusoidal displacement reattachment

Subscripts: OO

undisturbed

flow

condition

/

the

imaginary

T

the

real

S

laminar



turbulent

reattachment

T

transition

point

TE

trailing

part

part separation

edge

point point

dy

disturbance thickness, parameter,

fo

u-'_ dY U.

(_)_ dU

It 13VU2

Chapter I INTRODUCTION

This thesis is concernedwith a particular aspectof the overall task of estimating the aerodynamic performance of airfoils. Before a more precise statement of the problem can be made, it is necessaryto describeits physical setting and to define some important terms. Background The can

prediction

be obtained

air over speed

it.

aerodynamic

by letting

The

speeds

of sound As the

of the

the

airfoil

of interest

of the

number,

Definitions

performance remain

here

air to justify

Reynolds

and

stationary

are usually

while

airfoil

analyzing

sufficiently

an assumption

defined

of a two-dimensional the

smaller

than

of incompressible

flow of the

local

flow.

as U_

R-

c

(1.1) p,

falls

below

approximately

of momentum

before

outer

flow

verse

pressure

is, the it.

cannot

separation such

that

surface.

This

range.

transition

amount

pocket separation

Due

reach

to the

the

separates

is usually

a small

on the

can only

layer

and

Laminar ber

gradient

cases,

= 4 x l0 s, the

transition

readily

boundary

In such

R

leaving

of fluid

bubbles ability

near

a thin

region

free

shear

by reattachment remains

the

turbulent

layer the

surface, velocity

of reversed layer

the

separation

are observed shear

layer

run

imposed

of the

flow.

flow

adThat

underneath of laminar

boundary shear

out

of the

the

downstream

between

may

momentum

as a turbulent

trapped

a laminar

on airfoils

of the

Since

by a negative

in the

is called

boundary

occurs. fluid

be balanced

followed

of fluid

surface

stagnant

occurs

laminar

layer

layer, and

the

bubble. over

a large

Reynolds

num-

to reattach

to the

surface

2

of a streamlined imately first

two

orders

observed.

observed At

shape

this value,

sition

and

number and

Unlike

in Reynolds state,

large

may

downstream

extreme

forty

attention

may

depending bubbles

below

larger ease progress

been

one

near

on their has

recent

Piloted

form

the

length a fortunate

(although

they

of measuring toward

their

between/{

that

burst

it causes

layer

to reattach

at

high

is generally development

Vehicles)

has

shifted

million.

In this

one.

the

range,

causing

thickness. Since

The the

appear

to be very

pressure

or flow

full understanding

trailing

to a deedge.

in lift

numbers

upper

In

and

near

surface

is ascribed

bubbles

the

and

= 40,000

a the

at high

in this

case

observed

by

gradient.

separation

bursting

on the

Reynolds

due

decrease

Reynolds

present

of tran-

critical

of the

a large

at high

peak

has

upstream

bubbles

told-chord

downstream

is usually

a bubble

is

22 = 300,000.

but

pressure

years,

below

the

is

coet_cient

of attack

shear

leading-edge

in drag

immediately

reattach

laminar

jump

is approxseparation

airfoils,

suction

since

since

In more

to values

bubbles

years

laminar

approximately

also occur

adverse

at which

For

and

of the

of the

in 1933,

(Remotely

terest

great

first

of the

inability

values

stall.

aircraft

greater

The

Jones

abrupt

will usually

Bursting

to the

of the

number

number

results.

of a cylinder,

in drag.

of attack.

most

case

Reynolds

a large

reattach

angle

bubble

angles

Melville

and

a long

edge,

In the

cannot

it is called

increase

leading

tude

the

cylinder,

separation

on thickness

that

is decreased

bubble

laminar

critical

than

a circular

number

the

massive

200,000.

crease this

in fact,

the

smaller

flow over

Reynolds

depends

as an airfoil,

of magnitude

In the

as the

such

Reynolds believed of small

Reynolds due

to the

significant shift latter similar

variables

of this

were

from are

first

numbers

to be responsible remotely

range

delayed

leading-edge

in profile

an order

has

it possible This

drag,

to mid-chord

flow

phenomenon.

of in-

transition,

in internal made

for

controlled

number

increases

usually

received

of magni-

structure),

the

to make

understanding,

2-2

however,

is still

Whether crease

between

conditions,

occurs

by

have

to refer

to them

the

means the

immediate

bubbles

due

the viscous

however,

to its

to establish

gradient,

be obtained

ited

sufficient

bursting

in pressure

interaction not

not

a reliable

to a decrease

resulting and

the

traditionally

been

as "weakly

purpose

separation over

bubbles

airfoils

called

From

the

cid

point

study

of view

bubble,

distribution

recirculating

flow between

thinness sumed important

a model

lift

these

limiting

bubble

is lim-

coefficients.

Such

it may

for weakly

incompressible, for the

of boundary-layer

therefore,

In addition,

not

is to develop

account

inviseid

interaction.

are

from

can-

be more

accurate

Statement

interaction

to the

pressure

to assume

although

properties

by the

at low to moderate "short,"

in the

accurately

viscous/inviscid

modification

interacting

induced

strong

two increase

interacting

dimensional, in airfoil

laminar

viscous

profile

flow

drag

that

them. the

strong

can

section

in-

interacting."

developing

that

accompanies

and

of this

and

or to an

by a global

Away

interaction

Problem The

regions

approximations.

especially

number

is characterized

inviscid

viscous/inviscid

criterion.

in Reynolds

flowfield

of simple

vicinity,

bursting

that

the

of the

to be valid signals

but

a separation is elliptic the

very

however,

the

everywhere. are

first

Rather

communicated

on

distribution

is concerned

Given

bubble,

pressure

only

flowfield

applicable.

is made

is such

when

theory,

and slow

basis

the

the

speeds

upstream

point,

the

through

the

A weakly

global

invis-

a strong

local

of a region it seems

reeirculating

flow are

recirculating the

of

of

natural

approximations

approximations

within

weak

magnitude flow.

by

boundary-layer

of the

boundary-layer than

on

description

a reattachment that

viscous

characterized

with

between

of the

as its effect

is still

and

the

distinction

by the

insofar

confronted

the

interaction

and

the

usually

as-

flow,

in fact, with

the

outer

flow.

A parabolic

laminar

separation

at least

a local

and,

bubbles

w

the

with

In order

algorithm.

an integral

must

This

computational

is therefore to the

has

been

efficiency,

the

adequate outer

for analyzing

elliptic

employed

flow through

in the

present

boundary-layer

model

development

is

method.

to determine

development

method

as long as it is coupled

interaction

to maximize

calculated

boundary-layer

the

increase

be calculated

in drag

through

caused

the different

by a bubble, regions

the

shear

of the bubble

layer

as shown

-i

in Fig.

1-1.

by

laminar

the

The

formation boundary

boundary-layer design

tracked

and

work. the

in the

aration

and

bubble

transition

the

occurs

downstream

corresponding

turbulent

to the

is possible, does

not Rather

it is more

highly

usually

smoothly especially

reach

from

than

describing

informative

reaches below

its fully

developed

with

inviscid

before

the

the

effect

to explain

why

be

flow,

which

near

the

the

numbers, trailing of the such

that edge bubble

an effect

occurs

bubble

point,

which

inviscid

for

must

this

causes

the

leads

a plateau

to laminar turbulent

to the

pressure

1966],

relaxing the

accuracy

corresponding

[Green,

non-equilibrium,

sufficient

integral

development

T¢. As an additional point

at low Reynolds

equilibrium

region

Using

figure,

layer

separation

From

recovery

at point

shear

in the

surface.

with

the points

region.

"undershoots"

layer

the

8, shown

airfoil

to turbulent

figure,

of a reattachment

boundary

merges

layer

free

between

a pressure

shear

of reattachment

laminar

transition

the

be determined the

in the

at point

from

can

separated, from

of the

encompasses

turbulent

point

distribution

end

is initiated

separating

As shown

velocity

of the

gion

this

Once

T, predicted.

to form

the

layer

methods,

airfoil

point

of a bubble

seppart

reattaehment

recovery

always

the

velocity

distribution

boundary

layer

downstream

distribution. state

and

velocity the of the

Eventually, the

the

undershoot

distribution. turbulent

of

re-

Clearly

boundary

it

layer

airfoil.

on the

pressure

is observed.

This

distribution, can

be done

7" \ \

U

%, \

Uoo

\ \ \ % \ \

INVISCID

/

\ \

\\

DIVIDING STREAMLINEZERO-VELOCITY LINE

,-

Fig.

1-1

/__

_

-_I-

Schematic

of bubble

4 _2 g---

and

of its effect

on the

Velocity

distribution

6

by invoking

two

conservation tion

of linear

of mechanical

layer

the

stresses,

as this

Bernoulli's

rise

the

boundary

layer

negative

As the

results.

This

runs

the

airfoil

pressure

In the new

the

imposed

gradient

pressure

pressure

flow,

this

kinetic of kinetic

be recovered. law

such

by the

same

induces

inviscid

flow

matches

what

the

boundary

layer

is strictly

development,

the

it to

As soon

outer

a lesser at

that,

allowing

gradient.

felt

shear

is a scalar

of reverse

pressure

the

out

can

no diificulty

it, in turn,

boundary

of energy

it has adverse

the by

runs

is a vector

decreases

growth

of static

layer

the

conserva-

available

hand,

reverse

decreases,

a bounded

distribution

other

effectively

it the

and

the

dissipated

for

no more

of the

inside

conservation

inviscid

presence

being

boundary

of momentum,

surface,

limit,

the

flow,

While

by trading

since

on

energy

incompressible equation.

quantity,

out

the

however, of the

flow.

Once

of separation,

to balance

happens,

reverse

layer.

momentum,

of mechanical

unevenly

achieved

is a positive-definite of linear

a modification

and

is still

boundary

downstream

energy

gradient.

is simply

pressure

Conservation

become

energy

conservation

In an inviscid

static

shortly

the

momentum.

is continually

energy

once

the

pressure

inside

and

laws,

total

energy

law

conservation

flow as pressure

growth

of

constant

pressure

energy

equation

allows. While

the

momentum

on what

is "handed

transfer

of momentum

tion in the bubble the

reverse

present

down"

by the

by the

brings

in a laminar

the

Reynolds

stresses

momentum

of the

flow to be accelerated

reattachment

point.

as a rapid

boundary-layer

This

upstream

loss

and

context

of an ifltegral

method,

growth

in momentum

thickness

in the

the

flow

measured

momentum part lost

in this

part

the

This

to be recovered

of the

of the

of transi-

wall.

is observed,

by the

cross-stream

downstream

flow near pressure

turbulent

momentum

appear

outer

near-inviscid

of outer

growth

that

efficient

dependent

by

allows by the

definition,

bubble.

Within

the

flow results

in the

large

bubble

and

can

result

7

in a significant

increase

in airfoil

drag.

Previous

Previous empirical quent

attempts

correlations

:

for the

reattachment

Navier-Stokes =

over

empirical

models

proposed.

Von

tangent

sition

location

this

problem

Doenhoff airfoil

by

Reynolds

The angle

Crabtree

of the

bubble

distance

of the

than

shear

many

others

a coefficient

and

subse-

the

the

years,

streamline

several were

and

the

present

of pressure

with

transition

along

determined

using

model

followed rise

tran-

formed

of 15 ° measured

have

long

is straight

number,

is then

layer

by

characteristics

separation

that

crude

He determined

to reattaehment

Interestingly,

proposed

dividing

between

turbulent

streamline.

more [1957]

distance

early

bubble

transition

the

the

very

turbulent

caused

the

point.

streamline.

years.

In

separation

dividing

configuration

studied.

for particular

a constant

separation

is partly

at the

the

initial

been

from

Reynolds-averaged

surface

and

dividing

of the

that

at separation

of the

laminar

assumed

velocity

direction

has

range

[1938]

assuming

spreading

from

of approaches

correlations

the

a constant

phenomenon

length

variety

or empirical

to the

flow

to full simulations This

which

this

transition

behavior

equations.

time-span

and

at modelling

Models

resembles

in the

in the

from

next

turbulent

the this fifty part

m,

equal

to

- ;r This

coefficient

happen are

for values

discussed Gaster

He

has

been

of cr > 0.35.

extensively [1967]

characterizes

used

These

in review

investigated the

mainly

bubble

by

to monitor early

papers

bubble the

(1.2)

results, by Ward

bursting values

with

of the

bubble

bursting,

measurements, [1963]

and

decreasing momentum

which and

Tani

would

hypotheses

[1964].

Reynolds thickness

number. Reynolds

8

w

number

at separation,

Us( 2)s

(R6.)s = and

a pressure

gradient

(1.3)

LI

parameter,

p_

Au

(i.4)

/ks

where ent

the

velocity

between

instance

He found

that

bursting

a fundamental

and

Reynolds

can be traced

Although

in Gaster's

separation

decreasing

evolution eters.

the

gradient

role

modelled

in the

present

governing

mated

inside

the

bubble.

many

years.

distribution growth are

along

inability

not

approximately

which

1970,

detailed

McDonald

[1975]

is matched

study,

these

same

[1980,

the

stream

of the

bubble

and

Gault's

[1955]

measurements

of the

steady

in computer by

the

inviscid

to an acceptable

equations

degree

layer

methods being

an

prompted

a

flow field.

solution

flow.

shear

shortcoming

have

finite-difference

outer

for

pressure

These

technology

Navier-Stokes

boundary-layer

The

2.

problem

to the

process.

bubble

approxi-

in Chapter

bubble

approximated.

of the

plot. play

coarsely

detail

the

main

on this

a semi-empirical

approximation

their

bubble

parameters

are

studied

adequately

simulations

two

for

two param-

line

proposed

transition

or general,

a local

with

1986]

to the

same

in more

gradi-

conditions,

in these

equations

by a good

ellipticity

viscous

[1967]

will be discussed

advances

developed

with

in this

is not

local

variations the

Horton

inviscid

corresponding

along

and

accurate

for the

measure

mean

varying

always

et al.

however,

sufficiently

of more

axes

is distinguished

bubble,

With

of points

boundary-layer

region

to the

occur

model

Ingen

bubble

to account

Since

and

the

Van

refers points.

a locus

model.

integral This

model

in the

usually

number

and

Their

would

is not

the

[1975]

number,

bursting

where

Ingen

reattachment

on a plot whose

model

Van

parameter

methods. for the upstream

Their

of accuracy

Briley

bubble and

predictions but

region downmatch

comparisons

with

airfoil

teracting

drag method

distribution for

the

bubble

direction,

flow

inside

one

in the and

their the

[Gault,

1955], are

not

[1979]

method

interacts

of the

inefficient

there

one

in the

well

the

degree

outer

model

they

and

advantage

Giles

employ

of time,

for the

[1987],

of such

methods

about

next

characteristics

effect

of the

Cebeci

are

predicted

bubble

report

on

and

airfoil

Kwon

and

boundary-layer

limited

present

to the

in their

[1989],

the

vortices:

in between,

presented

local

by

time,

the are

generalmuch

too

routinely.

have

is that

for the

a finite-difference

approaches

[1989],

a source

different

data

in-

method

details

three

one

for the

by

accounting

methods

and,

that

flourished:

boundary-layer

Drela

the

where

flow

an

boundary-layer

bubble

[1983],

These

to be used

period

and

the

and

Schimke

flow.

of complexity

formulation

cases

and

inviscid

to be

part,

developed

interesting

appear

While

formulations,

the

rather

experimental

number

Cebeci

with

same

sense. the

outer

By properly

turbulent

with

similar

turbulence

termediate

Drela

instance,

eomputationally

Over

integral

For

opposite

[1984]

a finite-difference

to resolve

Reynolds

follow

and

is able

method

Carter

of the

layer.

considered.

Pletcher

and

boundary

compare low

Davis

bubble

and

in the

given.

a perturbation

the

part,

rotating

not

on

bubble.

laminar

method

ity

based

flowfield

by this

drag

are

representing

flow

wall

data

be

interactive

development.

and

they

can

Gleyzes

et al.

combine

a relative

considered methods

Crimi [1983]

and

of an that

use

Reeves

are of this

computational

inan

[1976],

type.

The

efficiency

m

with

a potential

for sufficient

accuracy

A Model In 40,000 tied

Reynolds

< R

many

number

< 4,000,000,

to a particular

In fact,

= F_

the

airfoil

different

they

range

generality.

for Airfoil over

assume

geometry types

and

Design

which many

laminar different

or to particular

of aircraft

with

separation sizes

and

conditions

different

mission

bubbles

form,

thicknesses,

each

on different requirements

airfoils. and

10

design

constraints

goals

conflict

bubbles. use

fly in this range.

with

For instance,

of natural

or 60%

airfoils verse

pressure

able

to recover

on forcing

seem

angles

however,

and,

only

if the

of pressure One bubble

therefore, effects

such

in drag, of the

of accounting used

in the

that

the the

effect

pressure

at different

Reynolds

for laminar design

and

large

the

lam-

such

that

the

advan-

of the

bubble

at first

ct's

are

starts

flight

changes

near

the

is offset,

regime.

Since

in bubble

struc-

problem

calculated

sight

obtained

distribution

of this

can be resolved

under

different

types

numbers.

separation

analysis

Reynolds

recovery

gradient

engineering

be accurately

This a "tran-

As the

higher

of pressure

can effect

is more

gradient,

would

In fact,

ad-

separating.

disadvantages

characteristic

"fine-tuning" can

the

the

layer

is between

pressure

type

encounters

of the

for thermalling

adverse

distribution

bubble

and

of such

to destabilize

therefore,

bubble.

of this

numbers

this

beginning

leads

performance

region.

of

to 50

necessarily

pressure

insufficient

ct required

in avoiding

Reynolds

distributions

analog

higher

speeds

distribution

without

gradient

trade-off,

flow at cruise

destabilizing lower

The

extensive

boundary

adverse

the

necessitates

layer

pressure

becomes

effects

high

turbulent

pressure

before

forms.

The

in the

way

transition

of attack

by the

ture

ramp

designer

The

changes

transition

laminar

the

edge.

favorable

bubble

energy

of moderately

of the

into

flight.

to help

leading

small

the

high-drag

at higher

end

layer

in thermalling

higher

design

detrimental

airfoils

boundary

trailing-edge

of a region

at the

of natural

the

flow The

the

airplane

pressure

laminar

aft-portion.

before

the

favorable

on

their

near-freestream

decreases,

a thick, tages

since

by means

boundary

over

particular

of sailplanes

The

employed

that

minimize

requirement

transition

gradient

ramp,"

number

generally

happens

would

technology.

recovery

the

that

low-drag

flow

chord

hinges

sition

the

pressure

is achieved

conditions

laminar

of the

to a steep

inar

flow

It often

program

bubbles

in airfoil

of Eppler

and

design Somers

is the [1980].

11

The

design

allows

method

great

of this

freedom

program

and

uses

flexibility

distribution

teristics

are calculated

by an integral

velocity

distribution.

Close

used

in the

distribution gram,

to achieve

the

designer

unacceptably tion

designing

airfoils

to have

predictions

laminar

separation

ing

computational

can

be used

the

drag

difficult

of modelling empirical

the

more

While cases inside

bubbles

the be the

the

recent

than

boundary-layer taken

into

bubble,

Reynolds

been

account all other

a local global

approach

has

fully

short

bubbles

approaches

of the

development in order characteristics

early

approach

rests

such This

to predict

accurately bubble

in

do

of

increascriterion

not

increase

increasingly

1984].

effectively,

a new

It combines years

with

on the

hypothesis

of laminar

useful

this

it becomes

as a laminar

upstream

of the

that

[Mueller,

thesis.

transi-

significantly

R = 500,000

decreases

might

for the presence

without

above

very

pro-

be advantageous

account

in this

information.

In this

proven

it would

developed

phenomenon

behavior.

velocity

that

more

This

inviscid

assuming

airfoils

methods.

is ac-

is predicted

increases

number

inviscid

which

number drag

charac-

bubbles

more

with

airfoil

of separation

while

airfoils

of the

development

separation

can be done

In fact,

semi-empirical

to model

which

that

by the

of the

applications,

this

detrimental

interactive

this

number

provided

has

which

Although

low Reynolds

bubble

rather

that

driven

modification and

method

aerodynamic

method

presence

properties

as the

to design

and

local

over

to design

airfoil,

be possible

through

drag

The

boundary-layer

in the

the

for low Reynolds

effectively

the

the

about

of section

of the

mapping

characteristics

performance.

transition

separation.

to eIiminate

In order

desired

the

the

boundary-layer

to aid

requirements.

of the

desired

monitoring

is warned

at laminar

the

process the

increase

occurs

should

design

conformal

in specifying

pressure

tively

to achieve

an inverse

the the

accuracy that

separation

bubble

been

separation

flowfield

speed

hypothesis

has

the

method

of it

confirmed. must

transition have

of

in some location

been

found

12

to depend

on a few local

chapters.

Furthermore,

a uniformly model unable

been

peak

and

caused

caused

the

by bubbles

In Chapter the

bubble

bubble

are

are

are compared model given

formulation.

the

data. results empirical

range

to the

3, the

large

calculation 4, possible

in the

suction

the

increase

to predict distribution

which

is of most

of the

of the

bubble

is described

In Chapter are

are

7, the range

summarized,

data

base

here.

and

necessary

modelling

In Chapter

part

and

specific

5,

Several

of the

of validity

to

of the

in detail.

analyzed

are

control

part

to the

complicated

airfoils

that

that

laminar

approaches

it is

interest.

parameters

employed

6, several

This

reduction

method

most

developed. Although

the

this

iteration,

program.

pressure

independent

following

a local

been

it is able

inviscid

operational

to model

In Chapter

important

airfoil

has

Somers

as the

in the

through

model

bubbles,

with part

interaction

and

such

In Chapter

turbulent

to experimental

enlarging

Eppler

leading

together

necessary

proposed.

is assessed, toward

in the

In Chapter

discussed of the

the

of the

in detail.

functions

flowfield

alteration

will be discussed

bubble

interactions

2, the reasoning

calculation

empirical

into

occurring

is described

for the

by leading-edge

is retraced.

of transition the

local

that

separation

for strong

sometimes

in drag

laminar

incorporated

to account

parameters

by accounting

accurate

has

scaling

airfoil

the

results

of the present suggestions

confirm

the

are

present

13 Chapter 2 LOCAL PARAMETERS

In this chapter, the approachfollowed in developingthe present laminar separation bubble model is justified. It is then shown how the shortcomings of previous similar

models

can be remedied

by letting

the bubble

flowfield

depend

on three

local

parameters.

Modelling

cess

The

development

that

is usually

variables acterize

need the

correct

iterative

conditions

tions,

which

be integrated varying

between

and

embodies

and

as for thousands Motivated

by

grees

of approximation

duced,

the modelling

about must

what

does

be supported

instance,

pressure

mass

to obtain

have

been

challenge

either

case

(2) by spatial

momentum

for the

fluid

the

of the

laminar

char-

time,

Navier-Stokes This

velocity

separation

the

flow,

variables,

conservation.

dependence

best

(3) determining

of incompressible

relationship,

dependent

variables

and

pro-

and bubble

(1) and

equa-

model

can

pressure

on

problem

as

flowfields.

computational

viscous/inviscid

the

flow conditions

what

depends,

field,

and

a three-step which

independent

phenomenon In the

entails

(1) identifying

which

(3) by a differential

of other

and

phenomenon

sequential:

two.

and

pointwise

numerically

geometry

the

the

velocity

conditions,

than

(2) identifying

on which

by the

boundary

of a physical

rather

to be modelled,

relationships

is comprised

well

of a model

Philosophy

does

eKiciency

requirements,

developed.

As soon

changes not

by analytical interaction

in nature

need

since

methods as approximations assumptions

to be approximated arguments

methods

of varying

and

have these

or by experimental rely

on the

are

boundary-layer

deintro-

to be made assumptions

evidence.

For

assump-

14 tions and on the extensiveanalytical and experimental evidencethat There

are

ber

and

two important types

methods (2)

of flowfields

is severely

by

the

shifted

from

arrived

at by allowing

the

and

them

valuable

then

The

Navier-Stokes

will be even

and

closer

imation

process

of the

phenomenon,

if such

a solution

more

limited

to the

integrated

solutions.

explicit

necessarily is not The are

in applicability.

yond

the

viscods/inviscid

case,

this

approach

has

interaction proven

at

Laplace's

the

their

is roles

relationship equation

is limited

flows

above

solution

the

information

in this

or

to the

way

knowledge

thesis,

physical

adds

that

the

it is

and may

to be a successful

limit

to this

have

brought

become

model That

is brought

between

the by

too

to a semi-empirical compromise

approx-

mathematical

of approximation, insight

to

pointwise

of its generality.

approximation

approach

it will have

differential

original

present

is necessary.

and

the

expense level

as in the

phenomenon

the original

to the

that the

stringent,

In essence,

of speed

by a model

In this

of either

more

from

at a particular

advantages

off-set

with

in applicability

solution

obtained

possible

iteration,

and

even

of the

removed

further.

however,

are

are further

an

final

are

equations.

that

is simply

The

viscous

equations;

conditions

to interchange

solution

over

original

flows

number

solution

approximation

balance

restrictive

the

a more model

the

Reynolds

requirements

drastic

integration

Although high

by the

at each

num-

boundary-layer

independent

to flow variables.

boundary-layer

numerical

about

the

by the

them.

(1) the

interactive

be done

problem,

processes

the

insight

on relationships

solution,

and

with

could

variables

computational

resulting

is simplified

outer

of approximation:

analyzed

what

spatial

approximating

pointwi_e,

If the casc,

by

be

or outer

equations.

integral

governed,

rely

given

surface,

can

independent

boundary-layer

airfoil

and the

of this type

from

inviscid

geometry

as dependent

that

restricted

integrating

between

consequences

supports

is,

model

an explicit

simplistic

and

one

be-

method. speed,

step

In this gener-

15

ality, the

and

understanding.

governing

integral

complemented bubble

eral.

This

adding

the

formulation only

prediction

course

of the

predict

the

effects

inviscid

flow began

[1967],

modified

Mueller

[1989],

in terms

of integral

suited

to the

method

with

into

that

integral

here,

and

of the

Somers

ds

with in the

of Thwaites obtained local the

appropriate

directly

non-similar

program.

interacts

weakly

empirical

model

[1980]

Because

bubble

models

method

differential

a method

and

they

employed

equations,

the

the

of Horton and

formulated

as these by

to

with

Schmidt are

such

able

are well

Eppler.

This

momentum

and

(H. + 2)U dV ds

2

-

Contrary

gradient

the

boundary-layer

d--_

relations

for

methods

governing

parameter.

CD -- 3-_

to simpler,

in two-equation from

(2.1)

53 dU

closure

Appendix. [1949],

pressure

which classical

analysis

d53

given

and

equations,

dh _ ds

together

models

necessary.

to develop

of Roberts

properties,

governing

absolutely

efforts

bubble

suggestions

boundary-layer

two coupled

it was

in gen-

Results

incorporation

Eppler

with

are

features

simpler

starting

separation

boundary-layer

integral

employs

energy

the

essential

the

by

reported

to the

the

and

at

establishing

according

capture

bubble

of its effects

after

the

that

in the

"guesses,"

and

arrived

of a laminar

on empirical

of its characteristics

was

research

reliance

are enforced

relationships

Early In the

the

equations

a set of simple

allowing

complexity

to minimize

boundary-layer

with

of the

In order

equations This

allows

developments

(2.2)

c$,

CD,

and

one-equation the

shape

factor

and

is therefore

such

methods

characteristic

H12

[Eppler,

1963],

such

as that

methods (H32,

in this

independent to analyze

case)

is

of the accurately

of aerodynamic

flows

16 provided

that

reasonably

empirical

by

along

assumed

family

of velocity

profiles

approximates

the

actual

models

bubble.

of rather pressure

above

method

Instead,

coarse

they

between

by the

do not

to calculate obtain

approximations

plateau

brought

noted

a two-equation

the

the

near-stagnant

Thus,

region

the

shear

bubble

assuming

and

leads

accuracy

of the

of 52 along

transition

in this

of the

development

equations.

and

fluid

advantage

the

growth

of these

separation

take

negligible

the

ments

momentum

[Schmidt

Mueller

and

[1989]

integral

equation.

Mueller

1986;

suggest

using,

by means a constant-

skin

friction

to

the

tum

thickness

Russell this

similarity

From

length

in Chapter the

transition

is approximated

Combining

these

_

(62)s o

for the

predicted

increases the

be discussed

ble

growth

[1978],

equation,

solution

with of the

by

laminar this

laminar

4, must

on

low Reynolds

1986;

Brendel,

by Horton

free

shear

equation, Reynolds part

the

leads

growth

of the

the

momentum

Iayer.

[1967],

The

was

number. bubble,

value

proposed In order

given

turbulent

and

to Truckenbrodt's

it appears

from

and

of momenearlier

by

to evaluate

by correlations

part

energy shape

experimental

=0

v

Schmidt

(2.4)

which

of 82 in the

ds Ira w

1988],

measure-

(1"1969)2(gl/c)

" 62 dU 62dH32 G - (H12 - I)H 2F-a-7 + As discussed

number

to

be known.

point,

equations

_/1 +

decreasing

by simplifying two

O'Meara,

(2.3)

instead,

(62) c from

Based

aflayer

(52)7- = (52)s from

flow

well.

The forded

the

of the

bub-

integral

equations.

parameter

equation,

e fH32

(2.5)

2

data

that

(2.6)

17

Using

this

tachment,

result

Horton

along

was

with

able

that

to reduce

Eq.

U--d-s-s _ The

assumption

turbulent

of a universal

part

Up

and

can

integrate

of the

H32

and the

leads

1)

recovery

profile value

from

equation

friction

at

a point

of reat-

n

velocity

to a constant

integral

skin

to

Ha2(H12-

pressure

energy

(2.5)

reattachment

bubble

a linear

of vanishing

and

a constant

for A_.

Assuming

transition

to obtain

the

CD in the constant

to reattaehment, momentum

one

thickness

at

reattachment,

(5:)n

w

Then,

=(52)z

eliminating

+C2

\U--_n)

(52)_

between

(4---'_32)(1+

Eqs.

(2.7)

£2 =

In order from

the

of Uz'.

segment

joining

the

this

happens,

This account tribution

(52)_

is started type for the in the

the

neither

Has

local and

relies

influence nor

the Eq.

value

obtains

on

of the CD

are

is decreased

and

(Ha2)=

therefore

cannot

empirical

upstream constant

whether

distribution.

turbulent

boundary-layer

reasons.

It does

not

interaction

on the

pressure

accurately

calculate

the

boundary-layer turbulent

criteria

conditions.

which

development. part

or

When

for initial

transition

in the

increments

it is checked

velocity the

in small

= 1.51

for several

viscous/inviscid

local

and

inviscid

(2.8)

and

is inadequate

part It

from this

Us

g2 is calculated

intersects

is calculated

laminar

able

step,

7" to _

of the

one

j

(2.9)

numerically,

At each

effects

region.

to sense

result

of formulation

(2.8)

\_-_-)

Gj?_

at are using

of 52 in this

actuality,

this

value

not

method

w

to implement

and

U_--_-) 1+

of the

properly disgrowth are

un-

Also, bubble.

in

N

18

The

turbulent

Finally, the

the

pressure

corresponding In

the

recovery

In the

at the

at

van

laminar

part

appreciably

point

approximating

functions

use

deviate

from

a straight

can be significantly

below

line.

or above

value.

better

Ingen's

can

reattachment

inviscid

an attempt

bubble,

distribution

were

is made U

-

Us

the

pressure

implemented

of the

empirical

distribution

[Ingen

and

induced

Boermans,

by 1986].

function

.978 + .022exp(-4.545_

- 2.5_ 2)

(2.10)

where _-ss

(2.11)

¢ = (R 2)s(6 )s which the

allows

for a slight

turbulent

assumed Fig.

part,

that

2-1

dashed

Using

the

to capture

gradient length bers

while for

but

to small

not

ble

of predicting

dictions.

empirical

in pressure

variations

beginning

based Fig.

the

in the

of the

achieve

only

model

prediction

accurate

chord

Reynolds

number.

models

responds

to variations

gradient.

The

governing

of the

the

drag

parameters, recovery.

bubble

to within

at separation

aerodynamic

by Eq.

the

parts

thirty

cannot

characteristics

the

although percent,

of the

of

of the

the

bubble

it was

that

in turn,

provide

(2.9).

adverse

in chord

Thus,

The

sensitivity

It is found

for instance,

pressure

recoveries.

increasing

prediction,

it is

distribution.

predictions,

with

For

again

inviseid

various

polar

bubble

transition. and

Stratford

noted, to

drag

the

as given

of the

on conditions

2-2 shows

and

vanishing

turbulent

features

drag

with

points

bubble

and

is employed

Horton

reattachment

and

to

distribution

the

separation

at a constant

such

the

models

of possible

In order

separation

at the intersection

between

development

explored.

necessary

occurs

empirical

boundary-layer was

locus

between

pressure

a comparison

is the

recovery

a Stratford

reattachment

shows

line

pressure

found

pressure

the

transition

Reynolds

num-

is very

sensitive

pressure

level

generally empirical

acceptable Eppler

at

capabubble

drag E387

preairfoil

]9

m

0 o F-q

_o

0 0

0 0

0

!

0

c_ 1

0

_,

II Q)

_n 0 o

tY)

0 o o o o

(0 v

Id n,

21

at R = 300,000 the

original

[1988]

obtained

Eppler

taken

with

and

in the

this

Somers

early

version

prediction

Low Turbulence

of the

and

Pressure

the

bubble

model

measurements

Tunnel

at the

compared

of McGhee

NASA-Langley

to et al.

Research

Center. Before tions,

beginning

two issues

due

to the

does

not

equal

to drive

the

of the

predicted

and

discrepancy

can

measured be

While

appropriate

are not

really

necessary

concentrated. the

same

experimental dictions

and

obtained

polar.

It may

mid-chord

at lower since

than

locally.

the

if the

same

employing

inviscid

on highly

the

the

near-stall

where

poses

validity

aft-loaded

inviscid

the

to a discrepancy

between

is prescribed.

This

interaction

algo-

methods

design

in angle

no obstacles

of the

airfoils

and

it will modify distribu-

are

distribution

it, leading

pressure

conditions,

difference

pressure

is present,

Firstly,

distribution

falls inside

of attack

characteristics

a, the

but

leads

angle

approxima-

pressure

a viscous/inviscid

lift coeffcients

same

that

Using

necessarily

lift coefficients

be expected

bubbles

bubble

above

be addressed.

experimental

interacting

aerodynamic

at the

inviscid with

by

the

should

of attack

therefore, results

model

to the

angle

for analyzing

In fact, ct rather

only

eliminated

rithm.

layer,

If a weakly

model,

alternatives

bubble

one at the same

distribution

the bubble

the

boundary

coeffcient.

pressure

of possible

affecting

the inviscid

lift

viscous

tion

indirectly

presence

to a smaller the

a discussion

effort

of this is most

often

usually

compared

at

of attack

between

the

drag

pre-

to comparing to the

present

type

experimental

model

will

for leading-edge

drag

decrease bubbles

for near

bursting.

and tween high

Secondly,

the

employed

in the

the

integral

Reynolds

turbulent program

variables

number

flows

boundary-layer is based

analysis on empirical

[Eppler,

1963].

without

bubbles,

While such

method

developed

equilibrium quite

relationships

appropriate

a method

by Eppler

cannot

be-

for analyzing correctly

ac-

22 w

count

for

the

reattachment, makes

use

Drela

relaxing,

nonequilibrium

especially

at lower

of the

[1986].

ods,

in Figs.

are

analyzed

nal

and

predicted

is consistently

It is stated

parameters,

may

be

the

to capture

it has the

flowfield

been local

along

exception

flowfield.

have

the

correct

lowest-order

The

failure

or success

response of previous

transition

NLF(1)-1015

airfoils

using

the

separation

for the The

must found

origipoint

low Reynolds

drag

coefficient,

be able

of the

of the

approaches

early but

other

of the bubble to these to predict

also

of the

the

bubble.

models inability In spite

of this aspects

bubble mean

inputs

accurate

present

three

of

model. flowfield. of the

characteristics: inviscid

pressure

In addition,

must

bubble

of

transition

to their

at transition.

the

by means

an

in the

to characterize all

hypothesis

empirical

prediction

at separation,

bubble

solely

without

abandoned

detailed

the

Although

help

two parts

to control

thickness

confirmed

of transition

a more

Bubble

process.

failure

not been

thickness

the

bubble

of much

in the

has

to achieve

the

meth-

turbulent

has

interactive

modelling

conditions

and

by

boundary-layer

number.

model

the

processes

been

therefore,

developed

point

Controlling

it is not Thus,

poor

momentum the bubble,

of the

effect,

parameters

that

boundary-layer

gradient

weakly

extended

separation

Parameters

the

on local

NASA

of

10 to 15 counts.

to model

physical

the

by Eppler

present

only to their

reliance

SpecificMly,

the

not

the principal

their

bubble

bubble

two

predicted

the

the

these

Reynolds

model,

method

The

higher

downstream

present

methods.

1 that

important

of the

can be traced

Rather,

with

most

by

Independent

be possible

locM

estimation

higher

in Chapter

it should

of that

to it at the

The

and

laminar

boundary-layer

is equal

Local

E387

at the

is downstream

layer

boundary-layer between

Eppler

transition

turbulent

The

turbulent

the

boundary

numbers.

a comparison 2-4

assuming

case but

however,

this,

and

by Drela

number

that

provide

2-3

Drela's

Reynolds

nonequilibrium

To

turbulent

be identified.

flowfield

and

its

23 w

t-0

C: "--'

0 X

C _,_

,,,, a;

>,-, 0

>-. C_

4_

_0 0

C.O 0 "0 0

I I C c:

,_.o

o

c_

cSB

_

0

_

C

o

_2 ,.o o

eY__ 1 O_LLI LI_;

p--.. (3O

_

0

0

_

r..)_

e'_ !

_b

z

I

i

i

i

|

i

I

0

X

o ,x: I1)

C

C

_

0

m o

C r-

g_

0 - C,I .o

0

0

e_

x

\

_"

_

"_.--......

/I

-0

o_ o

_

_._

o_

o

(1) ._

f--

_-

kJ_

.___ _'_ W

_._ -L_

i I !

I

....

I

....

I

'

25 w

effects

on airfoil

success

to account

The

first

location They the

performance properly

recognized

boundary

layer

at separation, one

would

has

universally

flowfield

is.

becomes

longer

a good

flow

could

In

need

as the

it bursts

correlating

with

the

at the

parameter

for the

that

long

from

stability

of the

boundary

Reynolds

layer an

number

a particular the

plausible,

separating

in

whereas

is decreased,

of the

bubble.

present

turbulence

It seemed

or

transition

bubble,

The

number

value.

size

stable

bubble. far

the

of instability

A very

failure

conditions.

the

by a fairly

Reynolds critical

flow

degree

surface.

of how

chord

local

to their

recognized

a shorter

as an indicator

addition, until

to estimate the

related

in determining

to be followed

associated

used

and

problem role

from

is likely be

independent

important the

to be directly

bubble

as it separates

in fact,

unstable

that

of the

a fundamentally

therefore

be shown

for these

investigators

plays

been

can

bubble

therefore,

boundary-layer

m

velocity and

be arrived and

at by forming

momentum

thickness

a Reynolds

number

at laminar

with

separation

the

value

of inviscid

as characteristic

velocity

length,

(R6.)s-

us(6 )s

(2.12)

/./

This

parameter

tion

of the

been

used

stability

of the

extensively

at laminar than

is certainly

flow

but,

unfortunately,

at laminar

in empirical

separation

a measure

useful

separation.

correlations

nondimensionalized

of the

stability

of the

only

Another

is the

with

value

respect

separating

provides

a coarse

indica-

parameter

that

of momentum

thickness

to the

airfoil

chord.

layer,

this

parameter

shear

has

Rather only

m

w

contains

information

ary layer

upon

to the stream

effect

on how much

reaching of the

the

pressure

boundary-layer

laminar

momentum separation

distribution

development.

has point.

on transition,

already These

namely

As will be discussed

===

is well

modelled

by the

e" method

of linear

stability

been

theory.

lost by the

criteria

lack

the

effect

in Chapter

bound-

sensitivity of the

4, this

up-

effect

26 Another

effect

interaction. since and

Since

the

of separated

that

another

to calculate

ber

flow

of the

laminar

separation

lence

cannot

to the

grid

within an

reasonable

adverse

lower

presence

effect

has

of the

been

conditions,

viscous/inviscid

of separated

by the severity

that

found

flow

pressure

and

gradient

to scale

Gaster's

unsteady

study

parameter laminar

a duct. between

well

pressure

with

gradient

Navier-Stokes

used

to keep

the

A suction

port

on

upper

for the

local

laminar

and

transition

layer

are num-

tunnel

the

to turbu-

can

computational

boundary

in yet

Reynolds

of the

wall

1989]

equations

unsteadiness

time-step

gradient

the

entrance

(computational

the

et al.,

to be important

Although the

[Pauley

of R = 50,000-300,000,

computed

limits.

a recent

Gaster's

is on the order

and

pressure

the

distance

accurately

resolution

point

flow through

on the

point

be

flow

shows

study, the

based

these

bubble

In this

discretized

this

at this

separation

respect.

that

strong

(1.4).

be mentioned

laminar

of the

flow is affected

surprising

by Eq.

for is the local

is a result

incorporates

given

It should

to be accounted

interaction

it is not

a parameter parameter,

needs

this

amount

by (52)s,

of the

that

occur)

due

requirements

of the

duct

provides

developing

along

the

wall. The

with

value

the

of the

velocity

parameter

gradient

P,,_a_ = .24,

representing

the

a modification maximum

to Gaster's

rather

than

parameter

the

average

in-

=--

viscid

gradient

boundary This long

between

boundary (steady,

by Gaster. unsteady process

between steady

correlates

and

short

of this

well

reattachment,

unsteady

the

short

the

large-scale

of the

bursting

line

(unsteady,

between

it is proposed by

is found

reattachment

to Gaster's

correlation

bubbles,

be governed

and

and

P_,,a_ < .24) from Because

may

separation

the

P,_a_ steady

such

to correspond

to the

laminar

layer.

that

shear

it separates

> .24) bubbles and

by Pauley

et al.

laminar

pressure

long that

bubbles the

field

the

measured and

the

reattachment rather

than

by

27

turbulent

transfer

certainty,

the

fact,

onset

as Pm,_

ration

increases.

laminar

shear

ably

occurs shear

the

Thus,

pressure

layer

This

by Russell

[1978].

When

of the

below

the

large

vortex

flow

that

is, the

and

is shed

growing

causing

again

Before

and

the

shear

above

bubbles,

too,

were

thought

that

the

transition

it should

not

necessarily

not

effect

be given

depend

mainly

therefore, so well may

with

why with

have Given

the

more

arrived

large

the

prob-

region

below

is discussed

without

suction

bounds;

becomes

small

the

As the

favorable

pressure

This

used

laminar

for a complete of approximating

it was

and

until, unstable

bubble

than

short-time

that and

then

understanding

the

final

model,

downstream

of transitional

if there

the

present

simulation

feed

that

that

starts

back

bubbles. of bubble

the viscous/inviscid

It was

also

upstream

and

is some

unsteadiness

turbulent

transfer

reattachment word

of than

on transition

it appears

conditions.

reported

transitional

mean.

in determining

Although

in the

thought

would

now

factor

is present.

measurements

scale

in nature

significant

method

at,

reattachment

be periodic

the

a means

flow.

bubbles

growing

in size.

It seems

rather

to wait

starts

at scpa-

reattachment

exceeds

flow at constant

near

on upstream

Gaster's

of reverse

separation

In

to recover

of a low-pressure

gradient

flow.

loss

is necessary

transfer,

to laminar

was

location.

unsteady

momentum

amount

formation

bubble

of laminar

with

repeats.

in the

if turbulence

sizable

or disproved

case

the

transfer

to collapse

unsteadiness

is a much

the

bubble

unsteady

the

Coanda's

layer,

cycle

and/or

inviscid

proved

in the

momentum

or the

adverse

description

influence

momentum

of such

of recirculating

the the

the

as it relates

the

be

be justified

gradient

effect,

effect

cannot

momentum

capable

to Coanda's

layer.

inviscid

to accelerate

is not

this

can

a greater

and/or

due

While

of unsteadiness

increases,

inviscid

the

of momentum.

at this It is still

by Pauley

et al.

Resolution

canpoint

not

clear,

correlates

of this

issue

bursting. interaction

to

in the laminar

28

part

of the bubble

can be estimated. bridges

the for

the

weak

function part

the

This

laminar

length

turbulent

and

latter.

transition

geometrical

and

the As

of Reynolds of the

bubble

location, characteristic

turbulent

the

the

number follows

of the

parts,

spreading and

providing

angle pressure

directly

height

and

of the

of the bubble the

is the correct

turbulent

gradient, the

bubble

model

in fact,

at transition keystone

that

characteristic shear

the

is thereby

layer

length closed.

is a of the

29

Chapter THE

The

Eppler

separation.

and

LAMINAR

Somers

It is based

PART

program

on

the

3

uses

value

OF THE

a very

of the

BUBBLE

reliable

energy

criterion

to detect

to momentum

laminar

thickness

shape

factor, (H3_)s---This

value

is approached

development and

of the

energy

from

separated

integral

above.

laminar

equations,

Upon shear

together

Instead

of implementing

this

usually

done,

the

development

laminar

part

of the

bubble

1.515095

of the and

function

bubble given

laminar

used

its calculation

by Eq.

(2.10).

separation,

the

of that

This

approaching

bubble

region

NLF(1)-1015

airfoil

in the

NASA-Langley

that,

as the

tends

(2.10)

pressure

to fall below

becomes is therefore

flatter,

checked

gradient

along

van

closer

postulated

was

Ingen's to Horton's that

Eq.

allows

available

the

in the

Ingen

a slight

several

Using

wind-tunnel

bubble while,

can

different decreases, as the

as it is in the

be improved

laminar

Boermans

above

[1986]

detailed tests

conditions. the

part

recovery

after

pressure

of the

Pressure

pressure

and

and

pressure

value.

from

approximation (2.10)

mode

mode.

Low-Turbulence

for

curve

by van

a limiting

in the

of Eq.

direct

below.

distributions

distribution

developed

distributions

accuracy

inverse

of pressure

pressure

distribution

quickly

momentum

to be discussed

in the

in the

the

the

Distribution

to approximate

is a generalization

family

separation,

using

relations

method

of a general allows

of laminar

is calculated

closure

boundary-layer

Pressure The

detection layer

with

(3.1)

NASA

Tunnel, It was

the found

pressure

distribution

gradient

steepens,

van

by relaxing

it

Ingen's

curve.

It

the

amount

of

w

30

pressure

recovery

between

separation

and

U____= (1 - DU) Us The

steeper

the

This

behavior

pressure

by use the

along

with

an inviscid

displacement

While

the

of Eq.

agreement (3.2),

pressure

+ DU exp(-4.545_

gradient

is consistent

ever-thickening

surface with

the

over

the

- 2.5_ 2)

bubble,

the

velocity

experimental

leading-edge

smaller

the

adverse

apparent bubbles.

value

of DU.

calculated

over

an

gradient.

distributions

became

(3.2)

distribution

in a steepening

an inconsistency

distribution

transition,

was

when Given

much

improved

attempting the

very

to predict

large

gradients

r

along

these

quite

flat,

of the and

bubbles, in contrast

inviscid

the boundary

layer

there

can only

a very

inviscid

hold

distribution

Although at

value

of (62)s

separation

determines layer,

that

imposed average are

is, the inviscid

pressure

included

that

small

amount

the

trend

the

by the

short

bubbles

of fluid

and

for mid-chord

of the

(1-

can

separation

of pressure

bubble.

recovered displacement

Near

6%c)

the

usually

therefore

were

perturbation

observed

amount

part

between

perturbation

entrained

is so thin

is a strong and

more

such

of fluid

a small

recovery

the

of the

only

laminar

only

leading observed

modify

the

slightly.

there

layer

that

magnitude

amount

show

with

it is realized

in the

pressure

contradiction

once

edge,

which

a significant

to the

is, to the

distributions

measurements

apparent

proportional that

pressure

with

be resolved

is inversely surface,

to the

This

can

predicted

distribution

transition.

bubbles

the

the

correlation amount

precisely

between

of fluid

reflects

the

the

caught

input

thickness

in the

to the

separated

the

greater

the

greater

the

amount

of separated

flow necessary

DU

therefore

gradient

in Gaster's

The along pressure

variable the

bubble gradient

already

should as well

as on

parameter,

lost

(62)s. Eq.

boundary region,

momentum

amount:

gradient.

momentum

of the

the

balance by

the

that

boundary

to counteract depend Both (1.4)

the

both

on

these

effects

which,

for

the

this

31 reason,

is thought

dimensionless

to be

inviscid

a better

velocity

independent

gradient

parameter

along

the

than

simply

the

average

bubble,

A(U/U_,)

(3.3)

/,(s/c) In dimensionless

variables

P becomes,

p =/_

(5_______ss

From sented

experimental

as a function

relationship, of DU

of the

shown

in Fig.

and

P directly

1015

[NASA

LaRC

1988].

The

solid

pressure

from LTPT,

distributions,

Gaster

pressure

3-1, was the

it is found gradient

developed

1987]

line

is a quadratic

DU

= _ 0.0610 [ 0.0152

the

Eppler

least-squares

DU

parameter,

pressure

and

that

P.

by extracting

experimental

June

(3.4)

x(s/c)

o

is well This

repre-

functional

corresponding

values

distributions

of the

E387

[McGhee

et al.,

included

in the

fit that

airfoils has

been

NLF(1)-

model,

The

value

the

variation

of DU

= 0.022

in DU

As pointed

out

shown by van

ensure

continuity

in the

method

is used

to determine

of Eq. one

(2.10)

in Fig. Ingen

is evaluated

Ingen

and

2

-P -P

< .3 > .3

Boermans

falls

(3.5) in the

middle

of

derived

to

3-1. [1989],

gradient

the

+ 0.5072P

laminar

at laminar

the

factor

when

4.545

in Eq.

Thwaites's

separation separation

laminar

point. and

the

(3.2)

In fact, variables

was

boundary-layer if the are

derivative rearranged,

obtains

which (3.1)), for

by van

velocity

[ r

used

+ 0.3048P

is Thwaites's however,

differing

-2_v dU] _s s =-(4.545)(.022)=-.10

laminar corresponds

upstream

separation to

slightly

developments.

criterion. different Furthermore,

(3.6)

Eppler's values

separation of Thwaites's

in the

present

criterion

(Eq.

parameter formulation

a

,"]2

z

0 r_

@ 0

40

c)

I:L

(::)

o []



0

g g _

II

o.

I1_

o

0

II

°

_.x ILl 0

_:> _

Z D

5-_

0

:3

o_

._

33

variable

DU

a continuous with

Eppler's

is used

in place

velocity

gradient

laminar

of 0.022.

Therefore,

at separation

boundary-layer

the

when

method

constant

Eq.

necessary

(3.2)

is used

to ensure

in conjunction

is

act C = -DU When

C is substituted

for 4.545

U Us where the

the

prime

exponent

of the

in Eq.

(3.2),

s the

_l_DU{l_exp[2uU_S(s_ss)]}

denotes

the

derivative

original

(3.7)

-

has

is

(3.8)

Us

with

expression

result

respect been

to s and

the

second

term

in

neglected.

Closure In order mode

to integrate

as driven

H12,

c I, and

Co

Falkner-Skan, Hartree,

by

must

or

profiles

of separation. seem

was

measurements [1966] a base.

[Schlichting,

1979]

axe used

from plitude

the

measurements the

therefore and

Mueller the

centerline of Coles's

two-parameter

in Fig.

3-2,

profiles. of the

wake

wake

function.

above,

natural

equations closure

choice the

to develop

may

is to use

the

the

correlations

be

for

reversed

Falkner-Skan,

Mueller

not

direct

correlations

attached

and

in the

the

or

upstream

[1990], best

however,

choice.

This

between

their

detail. obtained

forming

(h/b)

given

since

profile

the

integral

Fitzgerald

have

layer

energy

profiles

in some [1990]

shear

of the

by

Stewartson

examined

a turbulent

characteristics

most

profiles

that

As shown

The

[1954],

and

for

distribution

be provided.

Recent

Fitzgerald

pressure

and

Stewartson

to indicate

matter

the

the momentum

two

good

family a free

originally

ratio

developed

stagnation

parameters

is the

agreement

are of the

point

linked distance

to the

width

of the

shear

layer

Since

there

is slip

along

the

by Green

downstream

to the

geometrical

of the and

of

shear

G is the

centerline

layer am-

of such

a

34

0 r._ O ,--4 |

>-)

,I

0 c_ >-

b_

> )-(

c_ c_

I

f,_

,i,,m

°_ =

_

35

recirculating

base

By

the

applying

for the

flow,

these

definitions

dissipation

profiles

cannot

for the

integral

thicknesses

of the

following

relationships

are

coefficient,

the

H12 =

be used

to develop

a correlation boundary

for cl. layer

and

obtained:

1 + 2_

(3.9)

3a) - 2 (1 - 2a) H32- --

(2-9

7G + __G _) + 4-}(1 - 3G + 2G 2) (1 - 3_G)2-}(1 - 2G)

3

_2_3

In order profiles, values can

it is necessary used

of the and

then

repeated

for values

in order

to determine is shown

are

compared

solid

lines

abscissa

quite

different While

very

(h/b)

(h/b)

developed

separation

further

downstream.

thereby crucial.

where

these [1986]

of G and and

corresponds

seems

correlations

transition,

to moving

Stewartson

quite making It appears

from

downstream

sensitive the from

to

the

are and

changes

then

Mueller The

correlations profiles.

H12 and

inside

inside

parameters.

to greater

near

determination

calculations

Stewartson

moving

evaluated

station

two-parameter

As both

The

one bubble

by Fitzgerald

the

correlations

bubble.

are

downstream same

Stewartson

inside

to these

new

(h/b).

the

variables

used

of the

the

inside

measured

The

to those

by Drela

variations

vary

same

another.

similar

3-4

from

boundary-layer to the

one

sensitivity

to the

less

two parameters

three

(3.11)

obtained

to fit the profiles

against

similar

is not,

to those

corresponding

3-3 and

fitted

plots

H12(Ha2)

CD(Ha2,Ra2) on G and

the

the

on these are

The

of G and

between

both

point.

in Figs.

utilize

monotonically

Mueller

plotted

to those

how the

and

parameters

h (1 - 2G)]

relationships

to know

as a starting

bubble

result

these

by Fitzgerald

serve

at values the

to compare

(3.10)

The

H32 increase values

the

bubble.

separation

but

of the Thus, can

in

the

parameters,

of its

exact

dependence

measurements

that

the

back-flow,

be

36

o t_

0 6.

"0

_r_

t_ 0

°_

0

_0

o

_f

0 0 {13

0

o o

o_

37

0 .°°°.

r_

l i!i/i ! _.iiiii_. I ::::i ,,

(D i

,-ul

.°,°

I

7:::: o 1:111

i _;f

0 i

N

i -_i!

"1"

0 l.)

0

i:::::,,.,_ '-0



I

OE_

0

I

0

i :D :: I

,

I

,

I

,

I

I

0

_

0

0 e4

o

38

proportional

to G, may

be constant

ent

As shown

in Fig.

bubbles.

although stant

different value

between Ha2,

Eqs.

(3.9)

calculated

bubble

value,

(3.10)

the

easier

the

each values

follow

velocity.

and

from

appears

3-3,

in absolute

of back-flow

within

the

These

expressing

governing

equations to local

3(1 - G) -

although

of shape

factors

same

slope,

considerations

and

to correlate

bubble

the

for differ-

actually

thus justify

closure

and

different

measured,

confirming

a con-

eliminating

(h/b)

relationships

G, whose

in terms

behavior

within

each

flow conditions,

H32

(3.12)

H12 = (1 - G)(I - 2G)

Rs,

Although oped

_r2G 2 a [ 1 -- 3-_G -

--

very

encouraging,

to be implemented

of G on from not

CD

local

these well

from

flow

represented

them

profiles.

are

until

these

oped

by

by

the

issues Drela

better, have [1986]

equations

yield

in their

is unknown

contrast,

not far from

It seems

integral

By

results

model

conditions

profiles.

(4 - 5G)(1 4(1 -

these

in the

therefore, been

no

values

to keep

resolved.

using

been

implemented

the

value

of Ha2 directly,

of c/

of the

integrated

the

in the

can

with

be

obtained

profiles

the

fitted

the

factor

are

derived Green for now,

correlations Since

shape

devel-

dependence

profiles,

closure

model.

Drela's

the

parameters

Stewartson the

well

velocity

obtained

Accordingly,

have

In fact,

measure

the

(3.13)

]

sufficiently

form.

details

profiles,

(2 - 3G)Ha2 2H32

appear

present

the

corresponding

G)G)

do not

and

although

Stewartson the

of

develgoverning

correlation

is

inverted, l

= The

skin-friction

0.08

coefficient Cf

w

H32 1104008 [(H3 110 008 ]2 +

is found

.-_

/

-64.4

(3.14)

from

f

-0.067

+ 0"01977(7"4-H12)H12-1 2,

H12 < 7.4

R62 -_- = [

-0.067

+ 0.022

H12 >_ 7.4

[1

H12-6 1.4 J 2 ,

(3.15)

39

while

the

dissipation

coefficient

is given

CD

0.207

__

R62 H3---2

Removal Knowledge simple

technique

to allow inar

for some

direct

of the

pressure

effect

mode

the

of the

When

the

from

of the

separation

remain

before

starting

to grow

normally.

in the

following

sense.

aration

does

by

the

singularity

steeper

point

values

continuous much present again

steep

behaves

upstream

aration

of the

with

is therefore

its

the as the

calculated

laminar

separation

(cubic)

development

than slope

is believed

local

thereby

upstream, cannot

where

actual

laminar

again point

of H12(s)

separated

a few point

and

solving

point. chord

prescribing the

The

using

laminar

The

as,

layer

in fact,

growing

boundary

them

boundary-layer

layer

"inviscid" an

a

In the

starts

of the

sep-

gradient

of growth.

laminar

between

the

a pressure

rate

upstream

separation

when

layer

shear

of H32

skin-friction

of the

Therefore,

previous

to a much

distribution

boundary

laminar

percent by

the

the

separation

from

to this

by

affected

is equivalent the

chord

of sep-

is increasingly

causes

location.

maintain

the

but

of separation

it is felt

the

singularity,

upstream

to a prediction

observed

above,

of the

of separation.

leads

in

for a few percent

effect

and

upstream

downstream

which point

present

immediately

experimentally

value

This

and

is calculated

variables

gradient

to a

of lam-

described

a consequence

boundary-layer

pressure

is actually

and

values

led

upstream layer

function

separation

of the

shear

has separation

distribution

the

is approached.

incremented

gradient

model, is taken

the

similarly

are

gentler

just

gradient

a very

coefficient

growth

as separation

pressure

to exhibit

This

at laminar

laminar

using

at their

of separation

singularity

development

variables

reflect

downstream

Goldstein

(3.16)

- 4) 2

Sing_

on the pressure

boundary-layer

not

Separation

bubble

laminar

The

- 0.003(H12

distribution

for removing

separation.

the

m

of the

by

assumed equations

4O in the

inverse

point

mode.

growth

from

local

the

tion

and

as the

the

to the

than

the

is calculated

Using

this be

the

separation

to start

deviating

corresponding

the

velocity

assumptions

inviscid to Fig.

original,

pressure

this

new

As this

of the

separa-

behavior

to be correct.

as separation

upstream

separation

of DU

point.

is believed

down

distribution

old value

separation

distribution break

3-5, as the the

laminar

is approached,

of separation

point

is usually

is usually

too

great

is to be at a higher

and

a new

one

from

value

used

Using

and

the

current

= DUotd

new

to generate

the

tangent

value

a new

distribution

new

of the

by an empirical .

through

distribution

measurements,

-

(Us)old

velocity

at the

of momentum

angle

the

whose

tangent

Eq.

(3.8)

is continuous

Angle at the

streamline

proposed

at separation,

point.

thickness

separating

(3.17)

gradient

distribution

separation

value

relationship

-- (Us),e_ (Us)ola

for the velocity

Separation

the

layer

velocity of the

experimental

DU,,e_

with

boundary

the

upstream

boundary-layer

Referring

pressure

causes

distance

of the

a modification expected.

solution

some

matches

point

In fact,

inverse

inviscid

distribution

can

of the

results. The

u

Smooth

"viscous"

makes

by Wortmann

separation

with

the

surface

point, is given

[1974],

i

64P (3.18)

= - (-a6,)s A similar

relationship

proposed

by van

Ingen

et al.

[1980]

was

also

examined,

B

t=-y_ B is given

the

experimental

mean

value

(R6=)s of 17.5.

(3.19) The

model

did

not

perform

at all

N

well with

m

the

latter

relationship.

The

reason

is that

the

factor

64P

assumes

a wide

.,.

_. 0

o

N)

0 0

o 0

E X 0

°_

0

.e

w °_

0 o _

°_

0

0 Z w

I

,011 m

o

o

m

,-

0

D

0

_4

_

42

range

of values,

their

average

hand,

gives

from does

as low as 2 up

not

give the

to 30 for the

model

enough

range

flexibility.

of bubbles Eq.

examined,

(3.18),

and

on the

other

w

the

for /3 as given Navier-Stokes

r

correct by Eq.

simulation.

scaling (3.18) Eq.

for this have (3.18),

been

variable.

In

reported

therefore,

addition,

by Pauley has

been

very et al.

included

similar [1988] into

the

values in their model.

43 =

w

Chapter

4

TRANSITION m

w

The deal

prediction

of the

of attention

the

bubble

the

calculations

flow.

since

bubble

found

switch

must

been

criteria

well

are

inside

the

bubble

models.

In fact,

both

the

behavior

depend

strongly

first

bursting

or

claimed

be modelled very

location

in a more

it has

to work

transition

the very

as well as the

Although

the

transition

less that

accurately

in the

the

empirical bubble

since

bubbles

correlation

were

together

between

characteristics

such

first

with

a great

increment

on the

of the

due

location

laminar

to turbulent

transition

region

chapter,

method

to

where

a point-transition

In this the

drag

from

1989],

received

has

a few

employed

inside been

empirical

in the

model.

looked

for

Criteria

observed,

the

length

model.

Empirical Ever

way

[Walker,

present

discussed

gradual

has

many

distance

as conditions

from

researchers

separation

at separation.

have

to transition Since

it was

and

observed

an

local that

_l

_J

the

length

transition w

bubble

criterion,

number and

of the

based

is inversely

proposed

on

the

proportional

by Von Doenhoff

velocity

at

separation

to the [1938], and

the

Reynolds

number,

assumed

a constant

distance

between

the

first

Reynolds separation

transition, UJ: Rt, --

w

As the forces a value

Reynolds the

laminar

of 40,000.

number length This

increases, to decrease expression

in size. can

-

w

-- 50,000

Us usually

el

z

V

increases, Horton

be rearranged

[40,000

(4.1)

(6 )s

too, [1967],

such thirty

that years

this

criterion

later,

used

as

(4.2)

44 w

O'Meara

and

Mueller

[1986]

took

an experimental

fl

which

is a less general

these m

relationships

reported

value

[Crabtree,

aration

of about

750,

and

Mueller

describe

able

to capture

data

sets

this

used

which

how

the

precedes

correlation

by allowing

by O'Meara

and

thickness

transition

of data

as

Rtl.

Schmidt

and

[1989],

experimentally

Reynolds

by

a variable

Mueller

at the

laminar

developed

Mueller,

and

of the bubble

of momentum

over

effect

out by Schmidt

the vanishing

1957]

sets

(4.3)

As pointed

capture

many

C

relationship.

cannot

from

_ 155(52)s

c

m

mean

number

at sep-

separation.

Vincent

de Paul

Similarly, Mueller

Schmidt

based

[1972]

on the

propose

the

is

same

following

criterion, gl

such

that

this

the

criterion

onto

While sition w

is reflected

[ [267 -

0.3709(R_)s]

c , (_2)s C '

72 < (R62)s

< 720

length

bubbles,

vanishes

form

dependence the

and

the

coincidence.

J

< 72

that

therefore

better For

flows,

Larger

short.

bubbles

that

occur

longer.

Fig.

4-2

far

on large values

downstream

illustrates

this

coarse

values peaks

of the

are

usually

stagnation

where

the

the

this

traneffect

correlations thickness

due

to a spurious

correspond

to leading-

associated

isolated

The

stagnation

point

origin.

on the

bubbles,

bubbles

not

whose

the

influence

shows

does

of momentum

of (52)s near

4-1

(4.3)

through

way.

value

Fig.

as a plane

of separation

to the

of (52)s

line

(4.4)

Eq.

figure

a strong

of leading-edge

suction

Since

in the

has

length,

small

effect,

(4.3).

in a rather

length

> 720.

a straight

number

overall

in fact,

Eq.

is simply

is proportional the

of (Re2)s

it is shown

correlations

predict

form

quite

on the

length

which

usually

plane Reynolds

of these

airfoil

with

on (R62)s,

the

transition

at values

together

(52)s/c-g_/c

in many

at separation

edge

27 < (Re2)s

it is true

length,

for which

(_2)s

bubble

any

projection

3.7820(Re2)s]

in graphical

incorporate

m

_

c

u

f [513-

and points

point with are are

and

are

mid-chord

usually

much

leading-edge

45 w

SCHMIDT

0.I0

w

O'MEARA

AND

MUELLER

0 72.12

301 0.001

(6 )S .Z

•720

C

Fig.

4-1

Comparison for the

of two

separated

empirically

laminar

shear

derived layer.

transition

correlations

m

f-

46

;

i

C

0.20

0.10

w

72.12

301 0.001 463

72O

4-

O'Meara

D

Schmidt

and

Muelle:

w

L

_

Fig.

w

4-2

Comparison

of the

laminar

correlations

for the

Eppler

lengths E387

predicted airfoil

by the

two

at R = 300,000.

47

bubbles.

The

about

the

so far

no local

bubble

of (_2)s,

stability

developed ically,

value

with

of the

can

however,

separating

capture

criterion

the

can

increasing

effects

capture

angle

location tend

success

in attached

such

above at

the

seek

based bubble

local

models,

of separation, crement to better

explain

development

plot

Integration _2 and

known.

layer

edge

downstream =

these

two

w

variables, point

at

a value

location

of the

boundary

the

layer

that

empirical

criterion

distribution.

Specif-

upper-surface

mid-chord

criterion

the

it should

be possible

that

Whereas free

the

shear

the

here

the

in predicting

development

accurate

at

also

each

transition

is implemented,

and

conditions criterion

unlike

is calculated

in the

is early

downstream

downstream

prediction. Eppler's

to ex-

described

Eppler's

Since,

transition

criteria

layer

airfoil,

layer.

development

to a more

Eppler's

on

boundary

monitoring

lead

in-

In order

boundary-layer

be described. momentum

inviscid

that

station

subsequent

occurs

is at a different

downstream

velocity

nects

stagnation

in the

of the

Since

it was thought

transition

should

_3 at each

also

layers,

criterion

between

how

information

Criterion

layers.

naturally

any

no

pressure of the

boundary

it seemed

would

Transition transition

characteristics

the

In fact,

vanishing

to separated

which

on the

layer.

contain

of attack.

boundary

a correlation

separation,

in itself

of a variable

the

of Eppler's

a criterion

not

shear

Eppler's Given

does

can

drives

and

The

shape

velocity

along

the

the

(1-132,R62)-pairs

to the of/-/32

boundary-layer to obtain on

describing

trailing =

integral

station.

be grouped

thereby

energy

edge 1.62

and

a plot the

factor airfoil

H32

gives =

is taken

development, the

whose

development axes

concise

approximately

way.

the

values

of

_3/¢52

is therefore

as the

boundary-

U and

62 at each

of R62.

measure

boundary-layer

in a very R6_

equations

Eppler

con-

the

variation

in

development

from

the

The equal

stagnation to 10.

point Values

of

48 R_ 2 < 10 are

not

Following

upper

tion

the

dotted

until

bubble

mately

correlating

trend

since

the

nar

parameters matched

at transition

dependence > 875,

/-/32 to increase

turbulent

/-/32 = 1.51

(H32):r

with of very

growth

dependent

to 1.51.

toward

for smaller

does

as P and

is weakly

edge

Thus,

again

The

causes

of these

as shown generality.

local

attached the

flat-plate

which

investigations,

part

of

approxi-

of cubits

with trend

intro-

to disappear decreasing

in laminar

distribution

this

similar

length

in the

criteria

P.

lami-

served

to

bubble

conditions

incorporated

in

any one

particular

bubble

be

figure,

again

fully

again

transition

is seen

any

to a value

thickness

gradient

it was found

criterion

value

for

in the that

the

Eppler's

curve

tur-

number

of 1.77.

the

next

was

the

development

drives

at H32 = 1.46

could

at reattachment

Reynolds

developed

will be discussed

such

of transition,

boundary-layer

pressure

occurs

particular

bubble

as a

be constructed

pressure

in the

momentum

plot

laminar

as a family

and

the

R62 stays

The

a similar

separa-

growing

therefore

Downstream

turbulent

adverse

which

method,

while

H32 to decrease

on the

separation

for Drela's course

limited

the

to local

In fact,

a criterion

toward

is approached,

on the

unsuccessful,

(Re,)s.

the

(H32)7-increases

imply

correlates

the

on

inside

can

figure

while

values

Although

easily

trailing

on P.

laminar

starts

separation,

(4.4)

in this

line,

H32

to (R62)s.

shown

bubble.

layer

close

layers

depends

bulent

In the

shear

of H32

found

always

L-i

of Ha2

laminar

development.

is shown

point,

to Eq.

of growth

consistently

which

from

necessarily

such

shear

distance

solid

monotonically

value

poorly

quite

Ha2 grows

the

criterion

separation

not

of the how

laminar

boundary-layer

by

This

criterion

in (Ha2):r

illustrate

given

similar

of (R62)s

part

a typical

A very the

rate

the

downstream

an additional

This

shows

Ha2 = 1.515095.

While

for separated

at values

4-3

development

From

is met.

constant.

duces

r_

line.

with

criterion

L--

when

transition

the

Fig.

surface

is encountered

vertical

by

plotted.

boundary

and causes

As

the layer

method chapter. corresponding

and

49

@

D'I

@

w

@

L_

_k

L

,-_

ORIGINAL

PAGE

fS

OF POOR

QUALITY

"_

50 to attached where

transition

a bubble

number

be higher transition

aration

line

modified

at

the

This

transition

precedes

laminar

the

value

criterion

[Eppler,

Re 2 = 463.

criterion

Based

now

sition

is still

order

This

leads

This

apparent

mainly _=

predicted

million

transition

too

to a value

when

bubbles

at the

the

value

and

high.

Mueller. the

measurements,

cases

of Reynolds

be quite

intersected

coefficient

In fact, Eppler's

laminar

Eppler

sep-

[1989]

has

The

reason

numbers

why

chord

Reynolds

cause

any

significant

drag

increase;

thickness

that

is given

is due

rather

than

transition

is simply

that

in fact, by the

separation.

Reynolds

is replaced seems

to its having from

such

it usually attached

is near

which

tranIn

numbers with

190.

to work

well.

been

calibrated

a modelling

of the

instead

of a bubble

is predicted at

however,

(4.5)

= 875,

(4.5)

a value,

at chord

125 in Eq.

criterion

1.573) 2

layer

on airfoils

of ]_

data

-

For such

boundary

the

in Eppler's

drag

+ 125(H32

forming

intersection

high

momentum

shown,

additional

approximately,

coarseness

process.

not

at R62 = 704.

soon

to compare

airfoil

can

by Schmidt

+ 18.4Ha2

occurs

or greater,

from

the

in many

to

intersection

to be able

of one

_

the

because

separation

1963], on

to be predicted

happens

of 720 proposed

[in R_2]_- _> -21.74

where

transition

present.

than

original

natural

is actually

at which

it can

will cause

values

the

bubble

causes

the

same

turbulent

boundary

actual at

does

not

increase

in

layer

over

very

suc-

together

in

its length. In an attempt cessful Fig. w

transition 4-4.

to observed pler's ]:t62 vs.

In

to reconcile prediction

the

transition

boundary-layer 1-112 in order

Eppler's method,

e n method,

transition the

a value in many

development

plot,

to include

several

e" method, of r_ =

locations

criterion

different this

they

9 has

are

been

flows.

comparison

measurements

with

another plotted

found Rather

is done of these

to than here

variables

correspond using

Ep-

on a plot at observed

of

._

o_

\

-_v. U

e,I

o

o

0_

"_.,

_

_._

o

I

© © L _

_..I

el

I_

52

transition locations. This comparison method

was attempted

also because, since the en

can predict transitionsuccessfullyin a separated shear layer,it was hoped

that a suitable criterionfor use in the boundary-layer development

plot could be

inferred from it. In this figure,contours of constant n are given for the FalknerSkan self-similardevelopments

[Schlichting,1979]. For such developments, in fact,

a unique surface existsthat allows the determination of the value of n from the local ;

z

:

A

values of the momentum

thickness Reynolds

number

and shape factor alone. The

equation defining this surface was developed by Drela [1986] and will be discussed below. Shown

in this figure is Eppler's transition criterionfor attached boundary

layers as well as one of the attempted

criteriafor separated flow. It is interesting

that Eppler's curve fallsquite close to the n = 9 contour, for zero-pressure gradient flow (H12 = 2.59). As can be seen from the wide scatter in the experimental data, however, it seems dubious that a single curve could capture the correct transition location inside the bubble with any generality. In fact, the growth of n in a nonsimilar boundary shown

layer development

will not follow the surface whose

in the figure,regardless of whether

contours are

the flow is attached or separated. This

implies that transition is not dependent

solelyon the local boundary-layer charac-

teristics but depends also on the manner

in which the boundary

=

_=_

layer arrivesat such

values. The inclusion of path-dependency, or the effectof upstream development

boundary-layer

on transition,has captured the generality that the transition criteria

discussed so far lack In order to fully understand

why, the en method

will now be

discussed in detail.

The M

ago

The

e n semi-empirical

[van

In_;en,

to a variety

1956;

Smith

of aerodynamic

transition and flows,

e n Method prediction

Gamberoni, relies

method, 1956]

on a theory

developed

and

since

that

can

thirty

successfully explain

the

years applied

moderate

53 successof someof the correlations described above,that can correctly distinguish betweenleading-edgeand mid-chord bubbles, and that can correctly model the effccts of variations in pressuredistribution upstream of the bubble. Linear stability theory, in fact, directly models the growth of instabilities in a boundary layer while indirectly, through the boundary-layer development, accounting for the effects of Reynolds number, plitude take

at station place

when

transition to lie number

n reaches

9 by many

It is generally process

and

fully

flow

between 1984],

process

is usually

employed

be completed

in the

of the

the

use

the

e '_ method

actual

last

nonlinear

n = 9 -

14 and

the

amplitude

of the

method

oncoming

empirical remains

the

the

spots

usable.

to model surface

disturbance

conditions

corrections

[Mack,

engineering

transition

fully

are

stability

influence

of the

In spite

prediction

is taken

input

with

this

is unknown. turbulence

of transition of these

method

to

abruptly

associated

onset

the

empirical

at neutral

1977].

turbulent

started

significant

on the

defined

function

transition

concern

the

transi(n = 0)

intermittency

often,

cases,

Another

directly

an

the

to approximate

calculations

In both

Reynolds

is usually and

In order

region,

turbulent

on

stability

however,

to n = 8. Equally

of the

air and

distance.

reported

models

neutral

of turbulent

station.

method

correctly

region,"

been

1990].

between

amplification

streamwise the

theory

has

to

observed

to depend

et al.,

distance

of this

close

value

am-

is assumed

to experimentally

it appears

stability

of disturbance

Transition

this

Horstmann

appearance

ratio

so.

correlated

"transition

30%

at a value

to render

of further

linear

first

stability,

although

70% of the

The

of the

environments

1989;

that

or the

when

is that

inability

tensity

flow

the

corresponding

is necessary

The

similar

accepted

transition

method

previously

Vemuru,

flow.

logarithm

at neutral

researchers,

and

turbulent

[Arnal,

as the

a value

for approximately

region

at the

For

[Evangelista

tion

is defined

s to its amplitude

locations.

around

as the

n(s)

inforces

weaknesses, most

faithful

54

to the

actual

The

cn

equation

physical

process.

method

is based

which

dimensional

is derived

from

Navier-Stokes

of locally

parallel

mean

partial

differential

ficients

are

particular

the

the

numerical

are

flow is made.

only

solution

Navier-Stokes

equations

equations

functions

on

equations

perturbed

Subtracting

for the

method

and the

perturbation

of y, the

field

An

The

a system

is obtained.

Since

can

two-

assumption

flow,

modes

F

implicitly

y-direction. and

as follows.

linearized.

mean

of normal

Orr-Sommerfeld

of three the

be applied

coefin the

form

v'(x, which

of the

Similar

co* the

reference

assumes

length

a sinusoidal

expressions

radian

frequency, and

"1

=ae p(y)e

(4.6)

disturbance.

are

assumed

which

can

Here

for u I and

v I is the

disturbance

pl.

a* is the

be nondimensionalized

with

in the

wavenumber respect

to a

velocity, o_ = o_*Lref co*L_,/

(4.7)

co--

Substitution turbance

of these field

turbances.

expressions

results

in a set

Eliminating

into

the

of three

t__ and

partial

ordinary

/31 in favor

differential differential

of 9t

the

equations

for

the

dis-

equations

for

the

dis-

Orr-Sommerfeld

equation

is

obtained,

5,1,,+ [_in(_u, a are

and

¢o are,

performed

obtained Two

_),

in general, in the

by taking limiting

the cases

2_2]¢, + [iR(_u _ co)_2 +inu,, both

complex real

complex. plane,

part

of interest

Although it is understood

of any

of the

complex

derivation that

and

physical

variables

the

(4.8) analysis

quantities

are

employed.

are

{ o_ co = Wrco += icoi a = = o_r, a_ + iai, w_

w

the

+ _4]_ = 0

temporal instability spatial instability

(4.9)

55 To calculate the amplification of disturbances as the boundary layer develops, a spatial instability analysis is necessary.Thus, the Reynolds number and the frequency are specified and the wavenumberis found by solving numerically the OrrSommerfeld equation subject to the boundary conditions dv t v 1-

The

homogeneous

or eigenvalues -ai

boundary a.

is largest.

-0

dy

The

This

value

at

y=0

and

conditions

imply

of greatest

physical

can be seen

from

the

v' = _(y)e

as

y_ec

the existence interest

expression

(4.10)

of an infinity is the

for the

of solutions

eigenvalue

for which

fluctuation,

i[(_7 +ia,'. )_-w: tl

(4.11) = where the

_(y)

is the

bracketed

term

reference

point.

represents

the

layer

distribution

The

is the

amplitude

ratio

between

amplification

developing

of the

in zero

disturbance of the

this

term

amplification

factor,

amplification

In this the such

case,

absence

this

factor

-2

v( )e

n, is defined

curve

turbance

gradient

them.

Assuming

x from

some

x-locations a boundary

__;(,2_,,)

as the

(4.12)

logarithm

of this

ratio,

• equal

is a constant

of a pressure

as, for instance, The

is thus

at two different

and

I

v(y)e-,_,

o= The

at a distance

evaluated between

(eigenfunction)

gradient, ^

A2

The

disturbance

of a disturbance pressure

amplitude

to the

area

because

prevent

the

any

(4.13)

under

the

amplification

parallel

change

flow

in the

rate

curve.

approximation

local

Reynolds

and number

R62.

continuous

wavelength

travelling

downstream

modulation

encountered

in a developing

boundary

by

a fixed-frequency layer

is approximated

dis-

56 locally by a

series

Sommerfeld

equation

and

at different

and

H12.

plateaus,

for a sequence

pressure

As the

generalized

of constant-wavelength

of parallel

gradients,

streamwise

each

extent

obtained

mean

flows

characterizable

of the

plateaus

by

solving

of different

to zero,

value

Eq.

Orr-

thicknesses

by a different

tends

the

of R62

(4.13)

can

be

to ?2 _

--O_ i*

_2

dx

(4.14)

1

As

a measure

of how

calculate

the

the

branch

lower

total

far

from

transition

amplification of the

that

neutral

a boundary

has

occurred:

layer

the

is,

it is preferred

amplification,

that

to

is, from

curve,

=

(4.15)

o_$5

0

where front

the

independent

stagnation In general,

not

The

local

Reynolds

with

time,

with

the L _

may

the

come

layer

environment

close

distance

that

the

the

but,

along

the

in simpler

airfoil

from

of several

n = 9 indicating

it may

not

upstream

to n = 9 upstream

separated

of separation

of broad-band that,

the

does

or white at

a specific

dynamical

systems

is called

Reynolds

number

changes

are amplified

as the

different

frequencies

boundary

layer

at the

same

transition. shear

be necessary

of separation.

or in free flight

frequencies

boundary-layer

frequencies

in the

tunnel

rather,

to what local

amplified

layer

pulse

those

tracking

reaches

in a wind

amplifies

different

of separation,

boundary

simply

Since

necessitates first

pressure

correspond

distance,

frequencies

upstream laminar

disturbance

frequency.

downstream

As the those

the

number,

This

s, the

point.

boundary

natural

develops.

is again

of a single-frequency

noise.

the

m

consist

variable

The does

layer

may

be ditTerent

to monitor fact not

that

the some

necessarily

from

stability

of

frequencies imply

that

w

transition

v

will

occur

sooner

once

the

boundary

layer

separates.

If this

is true,

57

then

it should

information,

be

for instance

of all previous this

possible

such

to

devise

a transition

on conditions

criteria

would

criterion

at laminar

still not

based

separation.

be sufficient

solely

The

proof

on

consistent

of the

local failure

untenability

of

hypothesis.

Drela's While

the

questions

approximation

to the

in the

Eppler

duces

an error

it is now

and

Rather

above

e '_ method

in the

than

raised

Somers

discussed

Approximate need

further

developed

program.

study,

by Drela

In order

calculation

in some

e" Method for the

[1986]

has

to demonstrate

of n over

and

above

present been

that

time

implemented

this method

its declared

the

intro-

approximations,

detail.

performing

a linear

distribution

as can be obtained,

downstream

station,

the

characteristics

stability

analysis

for instance,

following

Gleyzes

from

et al.

of the boundary-layer

a finite-difference

[1983]

Drela

velocity

method

computes

at each

a data

base

of

L_. stability

a boundary-layer More

precisely,

value

of the

number

of the

calculation the

Falkner-Skan

using

the

nondimensional

local

shape

is divided

by

the

local

growth

factor

profiles shape rate,

of a Falkner-Skan

local

boundary-layer

that

factor -a

can

be "tapped"

as the

coupling

i, corresponding

profile

and

during

parameter.

to a particular

of the

characteristic

local

Reynolds

thickness,

i.e.

62,

=

to obtain the

the

correct

integral

physical

growth

characteristic

equation,

the

to be used

thickness manner

is of no consequence. self-similar

rate

The

developments

is obtained

in which

most

at constant

integral

independently,

the

convenient

in the

from

nondimensional way

H12 values

(4.15).

data

is to calculate and

increasing

the base

the

Given

momentum is generated

growth

R6=.

that

rates

Starting

for from

L m

the

Orr-Sommerfeld

different

values

spatial of/-/12,

instability

a set

analysis

of neutral

curves

of the

Falkner-Skan

is generated,

one

profiles for

each

at many value

L

shape

factor.

An example

of these

neutral

curves

is shown

in Fig.

4-5 for two values

of

58 of H12.

For

is evaluated

each

value

of shape

along

rays

of constant

factor,

the

reduced F-

to form

curves

development

-o_i(H12,

of each

R52,

F).

dimensionless

rate,

these

profile

(4.16) curves

is found

n( HI_ , Rs_, F) =

-ai,

frequency, 2rcfu U2

From

Falkner-Skan

amplification

jr8

the

amplification

factor

for the

from

-o_ ds

$ 0

r% _aT dR_ = fJ R6_o ds For

a Falkner-Skan

profile,

u where •

rn is constant

(4.17)

and

-

u

defined

1+

m U

f'(1

-

f')dq

as

t

s dU m

(4.19)

--

U ds Denoting

the

(4.18)

momentum-thickness

integral

(not

a function

of s) by I,

Z

R_

=

1 -Fm

u

I

(4.20)

w

Thus,

v

=gI

l +mus

__m

=I

l+m_,s

--

[(l+rnU)½_] _s

1+

---_s U

2

__= !

..

=

I22

12 62

(4.21)

59

Thus,

Equation

(4.17)

becomes,

n( H12 , R_2, F)

-(Wh)

1 fR6_ = -fi ,j r_,2° 1

f_6_

-[Z(H12)p J_,20 In

this

way,

Sommerfeld

wave

dR62/ds The

H12

Gleyzes surface

dimensionless

can

number assumes

curves

are

original

equation

mensional since

the

be and

the

obtained

shown

in Fig.

et al.

[1983].

for selSsimilar

used distance.

with

the

frequencies

form

this integral Drela

takes

This

leads

to the

of

which can

F)

dR_

the

dimensionless

is defined

be done

(4.22)

for

in

terms

Orrof a di-

a self-similar

profile

shown. for different

the

frequencies

envelope

following

and

as a straight expression

at a constant

line

for the

as done

by

amplification

developments,

dn

n(R_,H_)= where

n,

This

4-6.

-ai(H12,

eigenvalues

to find

particular

d/G

superscript

"e"

H

d--_ (_)

denotes

a value

[_-R_0(H_)] obtained

from

the

(4.23) envelope

of amplified

and

_(dn

H

_)

]

e = 0.01[{2.4H_2

- 3.7

+ 2.5 tanh[1.5(H_2

l°gl°[R62°(H'_)]

=

[[H_2:1 1.415

0.489]tanh[

3.295

+

-

3.1)]} 2 + 0.25]½

20 1 H12-:--

+ 0.440

(4.24)

12.9] (4.25)

H12 -- 1 ----:2

While

Gleyzes

et al.

then

evaluate

the

amplification

dn R, 2 o

H

integral

as

(4.26)

6O Drela

changes

the

variable

of integration

back

to s,

1• dR62

_(_)= j[ s [--(HI_), dn o [dn_,

At

this

point,

Drela

finds

d-'_

-

an expression

u

_

ds

for dR_2/ds

1 s dU U52

U d52 +----+ u ds

us

shows

that

ond

term

inside

differential

dR_2/ds the

dU

equation

equals

square

us

in a rather

roundabout

way.

1 52 dU

us

this

d52

bracket.

152

d--7+ _-2_

last

expression

If this

can be integrated

:.

(4.27)

2 u ds

=_ L_-_2+2_, He

d_

+----u ds

2 U ds 1 [s

j

with

is taken

_ a

_

1 in place

as a condition,

of the the

sec-

resulting

as follows,

(

U d52

--

1 5_

2 _--_-_2 _ d---_- + 2 u s d52

s dU

26_ d--T+ u d_ - 1 s

[ 2

lnS_)

+ _s(lnU

= 1

d

= 1

_

[ln(5_U)]

ds = --

d[ln(5_Y)]

3

ln(52U)

= Ins

+ C

5_U = Cs

52 = CIu

(4.29)

r

Thus,

the functional

the

constant

two

"empirical

form

is in general relations"

characteristic a function (given

of Falkner-Skan of the

in the

shape

factor.

profiles Drela

is recovered, now

introduces

where the

Appendix),

s dU U ds

u_ 118 u

-

re(H12)

-e(H_)

(4.30)

(4.31)

61

\'Vhat he

he means

is using

these

assumption

profiles

correlations

here

and

of a Falkner-Skan

substitute dR_2/ds

is "analytical

the , Eq.

Falkner-Skan

from

since

his

62-growth,

it

expression

for

the

form

Falkner-Skan

for

is not 62(s)

profiles."

dti62/ds

clear

why

directly

into

Since

indeed

implies

he

not

did

his

the

simply

expression

for

(4.28),

ds

-

2

_

+ 1

us

62

o_

1 +m

]2

62

IS (4.32)

62

which

is identical Having

to Eq.

established

(4.21).

Thus,

integral

only

had

to curve-fit

I(H12).

that

re(E12)

Drela's

Drela

for evaluating

2

+ lt(g12)

= [i(//12)]2

(H_2)

2

(4.33)

n(s),

n(s)=

62(s)

ds

(4.34)

0

.

°

can

be written,

using

Eqs.

(4.17),

(4.21),

(4.26),

and

(4.33),

(4.35)

"r;

L

where

62(s)

comes

by the

momentum

from

the

and

energy

non-similar integral

_-2

J

6-_ds

boundary-layer equations.

development

as calculated

Thus,

W

n(s)

m w

= fy " [-a*62]¢ds

(4.36)

=

(4.37)

o

o _

ds

62 Which

is identical

of an envelope precisely n.

from

to find

what

That

to the expression

enables

oq.

Using

the

method

is, the dimensionless

the

value

thickness at the

of H12

which same

52(s)

and

R,2

the

actual

straight the

envelope line,

is divided

by

from

known

to rely analytically

fictitious

two

following

that the

obtained three

(4.22)

the

reduced

Fig. factor

with 4-7

the

shows

occurring

The

three

the

argument.

along

momentum development

It is desired

shown

envelopes

jump

Fig.

4-6

given

n-growth

along

the

the

to the airfoil

Thus,

the

in H12 shows

by

curves

Eqs. the

the

error it is

are

not

to envision

a

with

corresponding

a to

of n obtained

at constant

frequency

to

curves

for

amplification plot

of

this,

lengths

growth

as calculated

(4.23)-(4.25) boundary

The

in question

the

a

envelope

To see

curves

used

it with

true

it is helpful

neutral

neutral

had

surface.

functions

to compare

the

on the

if Drela

up of two constant-H12

4-5 shows

envelope.

even

is non-similar.

since

made

through

the

of

station

of approximating

correspond

numerically.

Fig.

factor.

frequencies

not

that,

instead

development

development

Drela's

growth

downstream

in a self-similar

found

curves

frequency

obtained

development

value

is

Error

example,

in between.

using

together

of H12. shape

are

boundary-layer

use

equations

on the

the

author

would

a numerical

of shape

the

amplification

boundary-layer

but

jump

values

[1990],

at constant

boundary-layer

discontinuous the

on

history at each

local

for the

governing

obtained

the

Envelope

n(s)-development

the

by the

except

R6_.

Selig

of the

curves

whenever

necessary

date

his

amplification

arises

by

with

Degenhart

for upstream

-o_i

The In collaboration

and

as calculated

rate

different

of H12 and

by Stock

to account

growth

is in general

value

used

for

the

by Eq.

two

layer

with

the

that

serve

best

values

switch

in

at R_2 = 500.

frequencies

selected

Referring

represent

to Fig.

4-7,

limiting

cases

as /_6_ increases

n(F1)

grows

to eluciaccording

63

b.

0 0

i I I I

! I ! I I

N

0

k.

\

I ! I I I I I I ! I I I I I I ! I

,

I I ! I I ! I

D.,

0 0 0

_J

0 u3

N 'O

n,'

0

m

0 0 In

2;

I 00 0 0

I

I

0

64

0 u

0

N ',O

t_ ;> 0

u 0 u

w


O

3: 0 ..J

_4

0

/11:

I a3 r_

U'I

• 0

II ,..I

II

I

N

I m

...I

oo

y

_

0

0

m

c_

l"....

0

i° II (

....

t

....

J

....

J

....

J

LJ

_

m

&

_

96

o. .C_ o" 0 0

_z

R

o_

0 0

X °_

°_

tn 0

/011mz-J

0 L_

O

ml-o

o C_

!0

0

_b c_

0 0

o

o

Co

97

of the

amplification Fig.

6-1(b)

analysis. the

Reynolds

this

plot.

as the

total

drag

lower r

=

This and

come plot

point.

figure

serves

using

the

compared

with

very

with

ttl_

vs.

plot

shows

original

separation

give

locations

the

z/c.

similar

and

without

At the that

point.

bottom

laminar

of z/c,

turbulent

Goldstein's of the

this

the

singularity

can

and

the

upper

and

is shown all

five

denoting be

clearly

point. airfoil

boundary-layer

is symmetrical

surface

method

bubble

difference can

the

analysis

of the

airfoil

as

Finally,

boundary-layer

Eppler's

chord,

plot

asterisk

lower

The

extents

lift

the

the

singularity

for

with

the

exception

figure

on

process.

Since on

as expected.

separation

design

points.

the

to the The

and

a bubble the

development

separation

turbulent

With

left

lengths

the

Drela's

results,

bubble

on

Eppler

that

respect

the

model

the

includes

surfaces.

Goldstein

bubble

Eppler

with

of the

several

This

lower

as functions

through

and

n = 9.

the

by

with

is drawn

indicating

and

during

together

bubble out

label

boundary-layer

removal

the

model

and

Eppler

to illustrate

prediction

methods

next,

developments

at

upper

useful

The

suppressing

laminar

printed

the

boundary-layer

for reference

calculated

normalized

the

are plotted

0, by

the

from

is especially

variables

transition

of the

as printed

below

lengths

The

The

analysis

right,

with

outputs

is provided

viscous

arc

last.

and

of attack.

to the

are

smooth

c_ =

of the

coeMcients

coefficients

in the

angle

separated

drag

This

distribution

performed

reattachment

seen

inputs

is printed

boundary-layer the

relevant

summary

surfaces

below.

all the

and

and

well

separation.

number

been

of turbulent

laminar

velocity

program has

after

inviscid

The

analysis

!

contains

The

Somers

occurs

be

in separation from

boundary-layer occurs

at

can

be

starting

region,

seen

present

the

at

the

two point

plot

of

development

the

boundary

with

m

natural analysis,

transition however,

according indicates

to the that

modified

n at separation

transition is still

criterion quite

Eq. small.

(4.5). This

The apparent

en

98 inconsistency

of Eppler's

calibrated

using

surfaces

derived

As shown

The

more

separation.

In fact, appears

laminar

part

a few percent

as will

of the

chord

part

shown

in Fig.

start

before

any

data

the

by

in the

pressure

This

airfoil

was

Tunnel.

Drag

polars

for

and

drag,

also

R = 2,000,000, on

effects

low-drag

of low-drag data

Fig.

high

of the

(top

500,000

the

at the

(bottom

from

bucket).

since

6-3 shows

here

and RPV

design

effects

the aerodynamic

and

present of the

at

laminar

a mid-chord thickness

transition

plateau

in

point

in the

transition

may

indeed [1978].

Low-Turbulence for

lift,

The

is therefore It was also

was

designed

predictions bubble

characteristics

are

for

surveys

obtained

designed

characterized

of R = 2,000,000 R = 700,000

wake

were

profile

Pressure

at

for use

by

a large

to minimize

the

for

a high-speed

maximum

endurance

are

compared

most

clearly

with

the

R

L5

_=::=

by the original

Eppler

and

Somers

program.

=

seen.

of this

airfoil

on the

Shown

is the original

plot

as gen-

w

erated

is

laminar

by Russell

measurements

conditions

bucket)

to an increase

length

as maintained

Langley

500,000. and

of the

measurements

ct requirement.

The the

NASA

force

interpretation.

assumed

calculated

the

Airfoil

distribution

700,000,

bubbles

dash

in the

long-endurance

to achieve

of the

tested

inside

in momentum

experimental,

is observed,

pressure

1,000,000,

The

lower

deterioration

lead

been

growth

NLF(1)-1015

calculated

detailed

a high-altitude

aft-loading

w

recently

to the

this

a certain

As the length

close

NASA

The

any

it was

and

thickness

support

had

that

upper

to cause

necessarily

below

evident.

too far downstream.

change

not

if transition

is clearly

is quite

may

drag.

of the

of momentum [1980],

later,

by realizing

drag

not seem

Roberts

shown

airfoil

bubble

6-1(a)

does

the

values

than

be

be resolved

case,

6-2, a bubble

greater

to reduce

this

experimental

in Fig.

thickness

For

may

the bubble

Gault's

clearly

criterion

data.

Therefore,

from

in momentum

bubble

drag

performance.

bubble,

the

mainly

is identical.

in airfoil

transition

prediction

99

_ _c_ "_ _o

o

-

w

"\o

_ _.a _

I \ \\

I

o_

i

_ o"__,..._.._:_.

×

/ o/

!

II

r--

-< o..n ,_ -I.t_ o_

LL

I I *

I i I

_,ilil]i]_

....

I

....

I

'

100

obtained

by assuming

model,

Drela's

difference is due

transition

XFOIL

in drag

results,

In fact,

bucket transition

bubble at laminar

produce

the

of the

polar,

the

the

to a drag

data

onset

original

data

from

program

and

methods

reduction case,

excellent

of strong

prediction

global

of the present

LTPT. the

how

is calculated

the

present

formulation

interaction

upper

cannot

be

version

at the

what

At the

of the

as shown

over

accuracy.

Part

present

employed,

6-3, it can be inferred

In any

with

the

experimental

2-4 and

separation.

measured

separation,

boundary-layer

Figs.

leads

the

between

turbulent

comparing

the

and

prediction

to the different

2-4.

at laminar

in Fig.

top

by

of the

assuming

is able and

to re-

lower

neglected.

limits

At

these

m

=

conditions,

the

the

moment

lift and

separation the

W

the

z/c

program

tween

the

transition It can

as the

pressure

adverse.

This

of transition they

occur

also

be

for

curves how

the

given

as calculated

two analyses

represents

decreases

upstream

of laminar

present

plots.

:r/c, model.

the

they

laminar

becomes

appear

the

to occur

As will be shown

more the

version

of

difference length

eventually

Because

labels

between

point,

and

turbulent

a rather

original

the

contains

program

the

separation

in subsequent of actual

by the

in size

the

difference

Since

separation

also and

s_,,,.b/c,"

little

laminar

bubble

in terms

very

plot

transition

"1 -

of presentation.

shown

The

although

is really

is usually

at the

for the

will be clearly are

there

ease

transition

distribution

beof the

disappears increasingly XFOIL

tran-

downstream

below,

however,

upstream.

Matching prediction.

is used

poorer.

In actuality,

variable

that

are

of c_ as well as the

of x/c.

independent

be seen

locations

predictions

as functions

Given

assumes

bubble.

sition

the

variable.

variables,

drag

as functions

as "z/c,"

cumbersome

lift and curves

locations

axis

two

present

the The

compared

experimental pressure

drag

distribution

to experimental

polar and data.

does the

not

by itself

boundary-layer

Unfortunately,

guarantee

an accurate

development it is much

more

should difficult

101 to measure these quantities. Consequently,few data sets are available. For the NLF(1)-1015 airfoil, the details of the bubble flowfield can be checkedonly through the pressuredistribution, since no boundary-layer data were taken. Figs. 6-4 to 6-6 show typical comparisonswith the measuredpressure distributions together with the corresponding calculated developments. Fig. 6-4(a) shows the pressure distribution corresponding to the lowest value of XFOIL

prediction.

on the

upper

bubbles

upper

and

surface

lower.

to see

how

bubbles. of the

surface

very

different

the

other

are well approximated

on the the

Two

Fig.

6-4(b)

large

the

These, airfoil.

mid-chord

and

The

a very

the

case,

do not

XFOIL

seem

together

with

at the

mid-chord

one

on the

lower

shorter

on the

pressure

distribution

can

become

inside

sin(l/x)

function

be clearly

for CD,,,,,

are very

similar,

bubble on

It is interesting

much

can

the

Both

a slightly

development.

factors

surface.

gives

to contribute

of the values

edge

perturbation

shape

polar

be seen,

boundary-layer

of the

The

can

leading

model.

slight

values

shape

bubble.

at the

by the

shows

in any

bubbles

ct on the

leading-edge

to increasing

the

the

recognized

greater

drag in the

increase

in 52

of the upper-surface bubble being largely due to the longer extent of itsturbulent part.

This

edge

bubble.

variable

bubble, The

has

little

in fact,

is much

distribution impact

thicker

of cy seems on

the

at the

transition

plausible.

boundary-layer

In any

point

than

case,

in this

development

the

or on

leading-

region

the

this

pressure

distribution. Fig.

6-5(a)

to the

middle

along

the

mental. by

the

shows of the

bubble, It can

the airfoil

the

be seen,

large

aft-loading

simultaneously.

In any

pressure

distribution

polar.

In an attempt

inviscid

angle

of attack

however,

that

the

prevents

the

matching

case,

the

model

strong

reproduces

at an angle

of attack

at matching is one

degree

the less

trailing-edge of the the

gradients measured

corresponding

pressure than

gradient the

experi-

interaction

induced

on

surfaces

both

pressure

distribu-

102

C_ O C_ o

O

×

X

--_ _._...+,a

c_ 00

o Z_

/

m



'I......... i......... I......... i.......



o

CD I

I

k

o

CD I

g

_ I

.o C_ 0

'"I......... [......... I......... I......... I......... I' CD

_

(_

CD

CD

,-]

_ I

cD I

CD I

CD i



CD I

O

O I

Q_ .< C_

o

I II _

x -Tr_ ._J _C

0

0 II

LLJ = =

i •

o

0

m

103 0

\

" \\



o

0 °"_

O

va_

I ¢.9

_o'

w

/

!

m ,¢...-.

,__

, /

II

X

-'r¢'-1 .._J

o

@

!

o 0 O

O

O ¢o

,5 11 B

0_



v

W

w

I

O [

I Q..

0

i3

107

/ m

_

P

_

%.

"-

j

"f

I" I I I I

/

I I I •



J



,

a

J

i

I

,,,I

i

CO

....

I,,,

0

2,1,-,

-r-

l

= J

II fl II tt

i,J

°°

/

W

I

....

I

III"

/ y I ....

f l I

I

....

I

....

II

]

108 tion

remarkably

may

be

expected

to be thick.

tion

the

drag

to the

a little

well.

due

short

starting

although

turbulent

surface

6-6(a), bubble

total

a fifth

proportions. At

the

laminar

the

separation

leads

As far

as the

a destabilizing

of the

long

at this

and

condi-

by XFOIL

are

this

of attack,

n is

angle

where

point.

the

upper the

surface.

step

The

in _2 in the

how

the

correct the

until

transition

on transition

are

the

the

and

for

transition

condition,

rise

to the

is entirely

upper

surface. drag

while

point

is

preserving

bubble

has

been

corresponds point

suction

formation

to

precedes peak

again

of a leading-edge

therefore,

analogous

upstream

It is particularly

the

of the

to

a direct

lower-surface

transition

concerned,

to be

the

is shrinking

scaling

upper little

distribution

of n along

shows

the

contributes

is believed

pressure

bubble

this

distribution

case

where

of a 16%c long bubble.

upstream

before

bucket,

surface

distribution

of attack,

Beyond

travels

lower

adverse

upper-surface

angle

pressure

quite

predicted

on the

of the

in this

in spite

higher

effects

6-3,

At

6-5(b),

summary

drag the

separation

used.

the

by the

that

and

in Fig.

bubbles

top

length

indicates

separation

to laminar

bubble.

how

a slightly

and

effect

analysis

point

to the

transition

viscous

This

are

separation

in Fig.

to disappear

shown

to observe

seen

bubbles

is evident.

is clearly

the

be

The

of laminar

corresponds

short

surface

of n was

is shown

of the upper-surface

identified.

the

This

6-6(b),

interesting its

The

lower

as can

value

of the destabilizing

of separation. In Fig.

same

bubble

finally,

and

is highest.

upstream

is about

drag.

consequence

only

the

of the

upper In fact,

development

part

Fig.

the

bubble

to be amplified

boundary-layer

the

Both

it appears

to a rise

that

in Reynolds

number.

Eppler This on model

airfoil gliders.

was

designed

It also was

more recently

E387 than tested

Airfoil

twenty in the

years

ago

NASA

and Langley

is intended

for

Low-Turbulence

use

109

0

I

o_

0

m

o=

2_ c

_

-I

_i: :ii

_ _._ __

_

L

_ ,



I

_

. a._

_

_

I1)

_0

w

_D 0

p--(DO

i i

li

,i LLJ _

,

,

i

i

0

.

110

C_

/

x

¢.,.)

t--oo

X

cD

L

i

C_

0

u_

0

,-_

oo C_ ,--4

,-_

0

....I....... I...... 1....... r...... r .......r ...... r ...... i"' C)

• tF

J

0

0

I

0

I

i

C2_

I

I

I

I

I

I

I

I

I

I

¢...2 C_ 0 O4 ! U

_o

u II

x

"¢C -lEL.

L.('} *

,_J

, k.-I

o:::: ::z::::

0

r_

1'.'3

J I ....

"'',°,,

_JQ:_

a

112

C) O0

C)

x

0 c_

o

.

C:)

o

C) _

u_) C;)

_ I

_ I

O I

I

(221 I

O I

0)

O

I1_



0

0

m

,-_

II

U r

II

, X

"1O_

0)

ot,O

0 O

D'_

0 0

_b

0

_,i¸ ii_

°_

II

0 !

0

v

I_lJ

N

113 0 I

-S

I

/-

l'-

I I

I C_

I I

/

. c_

_o

0 _ I

,

,

,

I

_.

[

i

i

i

l

I ....

oo

] .......

r...

c.o

i._

cz) i*o

c:_ c_

c:)

c_ i,

-t"

w cl

•,

_"_ _ I._ 0 (/I o

,

c_

w

c_ •

0 ..o l,

i,

"-_

..._1

II _ II .,-J 1"_ C_ _._ c:) r.) I::Cl

c_o .,..-

Z

I

u'-) r,,o

|

I ¢/_

,

'-_

I

II _

I1

I

._

'.1-

"_

.

_

.-2

0

°_

1

/

_'_

.d 0

d_ I ....

= w

,

I

........

I''

' ' " I

°_

114

C> m ,,..._,

C_ oo

x

0

"_,

C_ IT7

I

II

I

I

II

I

I

I11

IFI

I

k

I

U

C_ 0

I I I III

I III

I I II

IIII

II

0 _

.

/ o°

o

p_ i1) U

U II

X

-r-

/

e_

0

"_

0

0 0 Oo

(

.,..i

Ct'C 0

0 0

=

=

0

0

'1 ....

i ....

I ....

I ....

I ....

I ....

I ....

I ....

I ....

I ....

o

,e--t

II

o v

e_ I

I

I

I

1

I

I

I

_

r,.) "_

115

o

I

c,,,l -'-I-

I I

( l

I

I

I

!

I

I

I

I

.... c'_ c:_l _-_

I

c,,,I N-_

o 0



P'C_ L_I_I =

11 e-

E,_

Ill ._.1

!

i J I i I

. L

i

i

i

i

I

i

II

118

CD

C_

O

_

x

X

_

t_ 00 ¢o

CD

CD C2) 00 00

_D C:)

.o --_

C2) ft..)

,,,i ......... I......... I......... I....

o °_

g2) iiiiHl_"

z

o

o

I =

=

[

C_ _

CD O

CD

CD

CD

I

I

_

CD

O

_

C)

o

I

O_ L.3

I

I

I

I

o

I

C3 O

O oO o

I

t6

O II

O

X

ZIZ

ro

_.-I

O

"_ O

O

o[/1 O O c_

t

O

O

O O O

O

if')

O

O

O

""

II (.o

m

I

I

& f_

r_ O

O_

N

119 0



I I

/

t_

I,

I # # p #

I

x

I'

/

I

"-x

!

I'

u_ I

" I I

/I

t3. 0 ,

,

I

I

Jllll|

....

I:_

o

o

o

o

o

u_ c_

c_ I.

'-l-

:k

'-tL_ r_

¢",J C) _

c_ DO 0

II t_ L.,.J O

0

_:::) --J

w

t-l

0 ..._1 t

"r-

I a:]

_



II

II

°_

_-,o"

I

=

I

t_ c_ u_

c)

• •

m



I

II -_n ._1

, ._ .............

°

0

• • • i ....

i

....

i

'

,

,

,

|



,

,

ii

,

,,b

_:_ Hi,

I

I

I

C_J

¢.,_c",I :> w

¢z)

(_ c_

i °_

II

124

I 0

z

_

m

o_

!

oo c_ X

X

¢M ,._

D

,---&" oo

Q

,--,

L .......................................... r,,_ •

.......i.........I.......

U

0 0

,'tJ

_

= ....

I

w

II

°_

_

-r.--.J

0

"_

o w

o o

0

o o o

°,.._

o o 0

w

II

O •,-I

0

O 1

_

i.i

_b

126

/ [ / I I

w

I

_m

j_ I

....

I

....

I Z)

....

I ....

I

I I I

II

I

II

t

II

I

II

I

II

127

C_ 00 ¢"o r..4

x gO 0 0

C)

k

00

C> r,aO o

/

. _,.,,t

u'b

4-m

_r-,D

I......... I......... I......... I......... I......... I........

0

0

I

0

I

t-_

I

0

I

° ..---4

U

0

,'el

I

g,;,

c_

0 l

'-0

• o

,.-o II

w

I tj

u II

X

-r13.. u'D C

-.-J

o

"_

UI

o

N

o

_

o

O"

0 0

r_'_

0

,5 II

0 I

i

!

!

'

I

'

'

'

_ '

I

,7....

e

.-.

@ I

'

'

'

I v

I

Lrb

c_

_

0

0

0

I

o

i,i

128

I I

I

Ii

_

I

.

c_

_:_

I

..I ........

I

c'_i

-t-

c)

_.

_

• . q'_ C_

,,_

I

_

--_

_Z_

C_ C_ • CZ_

.e_

,.0

II -J

II

(._ n'_

w

o m

0

0

I m m

0,1 _

. II II C_ .._ _

_

_

._

1 1

=

m

0

*_

1

, _

'

/

I1 ,_ -1-[__ _.1 ,
n

,.d , _ o

r.) i

_._ t

w

--

....

I

_

--

....

o 1

0

....

I

0

II

II

m

_

"_

133

00

C_)

o

x

0

0 0 Lt_

0o

_o00o_ o

0 C_

0

-_---F .... ,'",","','",'","'," C_

0

_

C_ •

I

1

I

C_

U

C_

C_D C_3 C_

i

I

1

I

I

I

"0 o

o

o

"_ o

I

C3.. C3 0

0 0

u II X

--1tn _._J

o

o °,-_

_o 0 C_ 0 0 C_ 0

II

0 0

v

i,i t_ c_ I

,_I

,-I C3_

0

0 I

0

134 0

-f-S,

I I

lyI

\

I

II

/\

I

t-,,,o_ X

I I I I

i

/I

I

L

I

1

11

B

!

0 I

,

,

• i

..l ....

I ....

I ....

,,,[,+J.

I

C3

c',,I

_

0

0

¢J

0

m

.TI

_J

i

,,..,

o

B B :

+

o

0 0

/+

If -r" r'l

o o

........

0

e,..-

0

,,.,

I

....

1

....

I ....

oi

F_

ii ,,
s_ s_

= SF

sin-l(1/A1

upon

until

(A.29)

e -300(h7/c)

the

(A.30)

undershoot

merges

with

the

inviscid

velocity

distribution.

The equations

Governing

above in the

shape

factor

inverse

distribution

is used

to drive

the integral

boundary-layer

mode.

equations:

-

d62 _ ds

(cs12)H32 - Co

[_ 3(cy/2)H32 [ H12+2

dC,C_- _

x -

4.2(C_,

+hdH_] u as j _2H32(H12- 1)

+CD--_52

dH32 ] H12 + 2 ds J H32(H12 -1)

(A.31) (A.32)

x -

C¢ )

(A.33)

148 where 3

0.015H32 --

6 = 62 Closure

(A.34)

H12

3.15 + H12 1.72-

1)

(A.35)

+ 61

correlations: CD = f x [CD]D,-el.

(A.36)

m

where /

\l_ f = { 1

+ 1 _-

r = 15 -

0 < _-=_- < 1

],

., -_(_-qa-_--1)

(CD,.n.,, (CDm..

(A.37) --v__: • > 1 t2

1000 hT"

(A.3S)

E

(A.39) [CD]O_a = ciU_. p + 2C_(1 - U_.p) _

/_2

(A.40) (A.41)

4

H,_ =

1 • rg_2-B.2o

*L

r.r 22120

cf>O

1 r_v '

°, J

-- r Ha2-H32_ l -I-

(A.42)

A_

Ht2o [

c2

¢8

J

cf