Experimental Investigation of Turbulent Flow Through

0 downloads 0 Views 7MB Size Report
display and the input voltage range of the IFA200 digitizer is +5 volts. Most wire and film sensors will have .... FIFO is full, in which case ACK is held low until a word is read out of FIFO and into memory). ACK ... the waveform by digitizing at very.
i

NASA

Technical

Memorandum

105210

Experimental Investigation of Turbulent Through a Circular-to-Rectangular Transition Duct

David

Flow

O. Davis .......

Lewis Research Center Cleveland, Ohio _ " ..... - -

_(_'ASA-T>4-1052!O) EXPERIMENTAL INV_STIC_ATION :JF TUR_.JLE-t-_T _=LL.-IW THRQUGH A CIF.,

0

,IF

_J (J ¢1

Z

c_

d lm

c_

w

o

c :>.J

_m

r

_4

0

0

23

L E

®

'_'1_ lilt

_o

I

0 ,,,-i

0

-,,4

i" ,,

Q QE) ®

f_

f-r-i

--3a

_

k---_ _

a u

l---i _ --t_ /_ _

Q mo _.J.r4

_

N

f

"1]tF_1i!3_ I1==- ll_z_,

_ L,J'_

' o.6=1

IIB uc 0.4-

0000000

O

000°000

U

0

0.45

O0

oo 0

o

0

0.2-

0.0

i

q

t

0.60

0.50

i

0.70

i

0.80

.00

0.90

r/R

Fig.

4.9.

Shear

stress

correlation

I

0.50

0.40

distributions

I

I

•-

Rob=88,000

o-

R%=390,000

at Station

1.

I

.x 0.30 ¢'4 II o 0.20 0.15 0.10

0

-

0

0

" 0

o

oo_OOO_8

0 •

• 0

0.00

I

0.50

Fig.

4.10.

0

0.60

Shear-energy

I

0.70

ratio

parameter

I

0.80

distributions

I

0.90

1.00

at Station

1.

CHAPTER

5

AND

DISCUSSION

RESULTS 5.1

Introduction The sequence

in which

data were collected

in the transition

duct is as follows.

The first transition duct was installed and the peripheral distributions at each data station were obtained. Data were

wall static pressure then accumulated at

Station

of Reynolds

5 to check

dependence

flow symmetry

of the

flow.

The

and

determine

transition

duct

data were accumulated at Station 6. installed and measurements were made data

were

taken

from hot-wire however, structure

all stations,

were

3 and

where

cm

data

along

near

about

wall

were

spacing)

in each

cm in the 400

data

core

region were

a direct were spline

interpolated

onto

interpolant

no overshoot

5.2

Pressure

Wall coordinate,

static refer

an evenly

duct

of the

half.

as such,

3.2.d-3.2.f.

holes

spacing

than

made

(see Figs.

between

in both

region

data

of large

based the

of the

2.54

here

that

obtained

plane

of the

measurement

in

duct

is

techniques

results

at all stations

of a monotonic

performs

cm

Approximately

on data

x-y

the measured

of interpolation

from

be emphasized

6 are

by means

1 and (0.254

varied

gradients.

about

to 0.127 symmetry

grid

points

It should

data

Quadrants

x 19 rectangular

flow and mesh

The

core

5 and

purposes,

spaced

method

of the

in Figs.

access

vortex

and,

assuming

between

of symmetry

quality

plane

1.27 cm in the

were

for Stations level

deduced

quantities,

6, rather

on a 63

cm in regions

For plotting

[55]. This

guarantees Static

the

of the

in this study.

from 5 and

spacing

in the

results

so that

indication

employed

to 0.254

were

anticipated

x-z

probe

so indicated.

points

The

taken

contour

quadrants,

are

measurements

quadrant.

the

2 shown

of the

and

transition duct was Stations 3 and 4. No velocity

the

number

installed

6. Turbulence

6 since

1 and/or

varied

at selected

mean

then

in Quadrant 2. For presentation purposes, to show an entire duct half. All subsequent

At Stations

z=O,

taken

points

all subsequent both

traverses

surface.

midplane

2. Data

imaged

was

and

about

arrangement

taken only the y-axis

been

individual

the the

have

and 3,4,5

assumed

in Quadrants

to the

were about

pressure

at Stations

was

extent

Finally, the second at the intermediate

at Stations 5 and until Station 5.

only

4, due

3.2.d and 3.2.e), data the data were imaged points

Total

made

symmetry

accumulated

At Stations

plots

2.

were only measured was not well defined

At data

at Station

measurements

the

extension

derivative

no smoothing

and

data.

Distribution

pressure to Figs.

P,,A__INTENTION_LLy

distributions 3.2.d

through

BLANK

were 3.2.f)

measured of the

along lower

the half

PRECEDIb,'G

periphery of the

PAGE

duct

BLANK

(sat

NOT

FILMED

48

Stations

3,4,5,

shown

and 6. Peripheral

for Reb

= 88,000

is 1/4 of the duct pressure

laxity

at all stations

with

the

along

more

the duct

and

in static

flow

is the layer

centerline

pressure.

a positive

pressure

peak.

Concave

in static decrease of the

convex

pressure

pressure occurs in cross-sectional walls

changes

relative

to the

inlet

is no curvature curvature

peaks

5.3

Effect Results

layer

at

the

ber

dependence

the

influence

lower

which

thus

far

operating

exists

over

and

some

limited

mean

velocity

conditions

results are shown here field behavior will be

notable

4 and

vector at the difference

upper

the

this

are

the

two operating is that

lower

associated

wall

sidewalls

in a net

is reflected

in the

(s/s,.eI=O)

induces

observed

pressure

in the

pressure

results

(S/Sre¢=:i:l)

minima.

induces

A net

decrease

is the result of a slight the radius of curvature

4 which

causes

maximum

only

to viscous

upstream

effects.

and

curvature

5) are still strongly and

Although

the

effects

present.

static

the

there

(the

wall

At Station

pressure

6,

is nominally

shown

that,

aside

number,

limited

range

no

turbulence numbers contours

shown

flow

data

were

of 88,000 and

in Figs.

and

transverse 5.2,

a thicker

appreciable

considered.

on downstream

from

5.3 and

largest

measured

conditions vortex

pair

are

vector observed

for the

lower

boundary

Reynolds

In order

num-

to determine

development,

mean

accumulated

at

390,000.

Total

velocity

vectors

in the

Station for

the

These

of the flowthe reference

plane.

to be very Reynolds

flow

pressure

5.4, respectively.

only for comparison; the physical significance discussed in the next section. In Fig. 5.4,

represents

the

Number

have

the

contours,

axial

plane.

numbers

for

losses

which

results

3 and

station,

number

Reynolds

flowfields

along

Reynolds

bulk

velocity

the

in the

is due

at this

5 for operating two operating

along

Stations

ReynOlds

of Reynolds

measurements

5.1,

(Cp=O)

> 0) which

curvature

Station

presented

regu-

z-z

Reynolds

is larger

viscous

to Fig.

value

change or curvature effects, the entire cross section.

of Varying

the

along the side and upper walls, respectively. At to the inlet value so that a net decrease in pressure

walls

between

there is no area constant across

inlet

between

occurs

of the

two

between Stations 3 and 4 which area and viscous losses. Also,

sign

spanwise

about

duct

8re!

represent

of cross-sectional area, wall curvature at an axial position where streamwise

resulting

minimum pressures to occur Station 5 the area has returned

dimension symbols

excellent

of increased

curvature

wall

the

are

growth.

(aP/Or

gradient

The to the

along

distributions

Solid

of symmetry

In reference

the

gradient

Conversely,

a negative

above

normalizing

The

station.

referred

result

is occurring.

pressure

5.1. data

drop

Local static pressure is a function viscous forces. Station 3 is located of the

coefficient

centerline.

results

boundary

pressure

assumption

coefficient

case

rapid

the

between

pressure

number

diffusion rise

net

in Fig.

at a given

supports

difference

the

Reynolds

and

measured

primary

is that

390,000

circumference

static The

and

wall static

The similar.

number

mean One is more

49 circular

and

observed both

centered

were

not

operating

significant

practical

applications,

duct

remaining

measurements

Mean

5.4.1

Flow

Mean Mean

and

6.

Figs. and

would

velocity

velocity

contours

results

5.6,

shown

duct.

induced

the

of the

At

relatively Between

boundary

layer

vicinity

thickening

pressure probe

pressure

the

variables at

Station

are

P, -

defined

5 was

are

divergence

station.

develops

centers

away

upper

boundary

by natural

wall are

vortices

further

respectively.

sidewall with

their

6, the

layer,

The

flow arises

of the

and

thickens

in the

which

vicinity

of the

sidewall

are

indicated

5.1) associated

centered

to the

from

(lower) layer

Between growth

have the wall causes

Stations and,

in the

convergence.

calculated

Pwhere

boundary

distribution and

and

of the

of hot-wires

secondary

Station

At

in the axial

similar

at that

in shape At

of the

everywhere

by lateral

to Fig.

duct.

a non-diffusing

flow pattern

oblong

5, lateral

convergence

layer

static

are

very

This

axial

vicinity

grows

vector

flow in the

(refer

sidewall.

circular

2 and of the

boundary sidewall,

static pressure butions:

duct

in the

observed

and

of the

by means

sidewalls.

near-wall

vortices

are more lateral

duct

gradients

the

Stations and

magnitude

the

of the

5, the near

of the

The

along

shape

velocity

in

probe)

and 6 are shown

flow through

measured

3,4,5

shown

pressure

is qualitatively

vectors

are (Pitot

3,4,5 total

the distortion

of the reference

pressure

extent,

sidewall.

distortion

vector

skewing

Station

in lateral

and

for

of the

at Stations

stations

to develop

that

flow is due to a secondary

pair

of lateral

6, the

most

contours

sectional

is seen

6 show

velocity

primary vortex

positioned

5 and

of the

maximum

by transverse

thinning

5 and

data

3, the

cross-

contours

plane

at Stations

Station

the

The magnitude

curvature. grown

Since,

condition,

pressure

et al. [39] for turbulent

Transverse

a discrete

as a result

of the

by Taylor

represents

into

of the

case.

to theses

probes) At

follow

at Stations

in Fig. 5.7.

distortion

for

bulk

for a circular-to-rectangular

transverse

Total

(hot-wire

respectively.

development

presented

S-shaped

the

condition.

operating

in the

applicable

generally

The

operating

= 390,000

respectively.

contours

and

The data

direction.

plots

3.2.f,

4, a distortion

sidewall.

Res

were measured

coordinates

through

5.5

Station

axe for the

all measurements

to restrict

condition higher

the differences

contours

Reference

axial

to one

to the

though,

repeating

therefore,

stations

closer

flow variables

in Figs.

to justify

decided,

the operating

be

Overall,

Results

flow

3.2.d

the wall.

enough It was

at the remaining

transition

$.4

away from

conditions.

measurements most

further

at Station from

the

total

pressure

directly

and

velocity

1 2 _Pa,,_bU_/(R,_i,T,,mb)

as in equation

possible

5 was measured

since

the

(3.1). total

with

a

distri-

(5.1) Measurement flow

angle

of the is less

than

static 10 °.

5O The

measured

static

and

pressure

shown

in Fig.

agrees

well with

5.1.b.

The

calculated

pressure

field

but

doesn't

show

layer

mean

divergence

along

the

layer.

On

midplanes the

were

y=0

The

wall tap

agrees

5.9.

midplane

Along

part,:conflned ther illustrate

the

z=0

duct

z=0

y=0

(side

to estimate (negative

(upper

evaluated

upper

wall),

the

strength

divergence

wall),

the

midway

the

duct

divergence

the

parameter

2.8 and strain

rate

wall,

which

5.4.3

Mean Total

that

flow

is defined

profiles,

These

layer

substantially

This

is due

periphery Beyond layer

from

along

both

to converge a large

jump

most

To furof the

jump

in a manner

similar

to the

is observed

between

x/R

is due to a reduction

in the

primary

result

effects

results thickens

the

at

of the

will

Stations

wall) traverses Similarly, axial

to the secondary

Station

is, for the

vortex

pair.

The

results

= at

be reflected

in the

local

turbulence

profiles

pressure

5.12.b.

divergence

as

(_ 0.5 cm). is shown in

boundary layer continues to converge well into the Based on the strength of the divergence observed,

divergence

(sidewall) and y3 (upper and 5.11.b, respectively. and

This

is a direct

z/R = 8.0 shows that the transition duct extension. it is anticipated structure.

but then

4 and 5).

OU/O_/2

the

boundary

section, peaking at around 20%. on the upper and lower surfaces

the flow begins

upper

4.0 (Stations

surface,

same

parame-

surface oil flow results by Reichert et al. [4] obtained duct at an operating condition of Recl,i = 1.57 × 10 6.

sidewalls,

on the

the

of the

is the

divergence

through

the divergence

(lower)

to the actual transition the degree of divergence

duct, Fig. 5.10 reproduces in an identical transition Along

with

as clearly.

(OW/Oz)/(OU/Oy2) evaluated at the first data point from the wall An estimate of the axial distribution of these divergence parameters Fig.

measured

measurements

distributions

used

and

midplane

as (OV/Oy)/(OU/Oyl)

the

the

5.8.

qualitatively

saddle-shaped

distributions

On

is defined

the

in Fig.

divergence

velocity

as convergence). ter

are shown

wall region

Boundary The

distributions

in the near

measurements, 5.4.2

calculated

vicinity

show

that,

along

1,3,4,5

between

traverses

thickens.

measured

Stations

the Y2 traverse

of the the

6,

along

the

(see Fig. 3.2) are shown in Figs. velocity profiles are shown in Figs.

flow transferring

5, where

and

thins

boundary

1/3 traverse

cross-sectional Along

and

1 and

boundary

the

1/3 traverse.

fluid along

to the

vicinity

shape

is constant,

the I/3 traverse,

Of the the

5.11.a 5.12.a

5, the

along

layer

Y2

the duct

!/2 traverse.

the

boundary

thickening

is due

to natural boundary layer growth, but along the 1/2 traverse, the thickening due also to the common outward flow associated with the vortex pair. At

Stations

5 and

traverse

exhibit

a double

inflection

and

axial

5.5.c

and

5.5.d)

the

6, the

total

pressure behavior.

velocity

contours

and

velocity

The total (Figs.

profiles pressure

5.6.c

and

along contours 5.6.d)

the

is

1/_

(Figs. indicate

51 that

along

behavior high

traverses

adjacent

and

is observed.

This

is a result

momentum

sidewalls,

fluid

from

re-energizing

separation.

the

A break

of low momentum

creating

a flat

5.4.4

spot

6 (Fig.

ridge

than

Streamwise

vorticity The

second

in non-

vortex

These

flat

along

probably

the

layer

toward

flow

due

the

of duct

preventing

(y2 traverse)

region

5 (Fig.

to a

centerline,

is seen to be much

larger

5.6.c).

ducts

by a blunt

occurs

can

be

by the

of the transition of total pressure

generated

lateral

obstruction

vorticity

Oz +

duct loss.

different

(by pressure

in the

Oy +

vorticity

through

is much

wO ,

than axial

(4)

and by the

non-circular

OU

ducts. vorticity.

vorticity

equation:

BU

OU

+ a,N + a, o--; (3)

(2)

02 02 Oz2)(__--_) +__(.2 OyOz

02 + ( Oy2

The

skew-induced mean

-

layer. stresses

generation

straight

weaker steady

Reynolds

(1)

0-_) Ou

two

boundary

by the

Streamwise flows

vorticity

are represented

O 0-_ +-_x ( Oz

by

deflection

in a 2-D

created

vorticity.

in turbulent

stress-induced

mechanisms

in the exit plane significant regions

generation

is streamwise

stresses

pairs cause

circular

to as stress-induced

Generally,

very

region

midplane

peaking

a "ridge"

with spanwise vorticity. This mechanism is often streamwise vorticity. An example of this type is the

generated

mechanism

Reynolds

to the

boundary

This

at Station

first is vorticity

of a shear layer to as skew-induced

is referred

and

a double

convecting

vortlclty

mechanisms.

horseshoe

the

pair

flow

at the

field.

The presence of the vortex is undesirable, inasmuch as they

gradients) referred

layer

from

Y2 traverse,

vortex

core

occurs

velocity

5.6.d)

Streamwise

potential

fluid

in the

to the

of the

boundary

in the

transfer at Station

the

parallel

(5.2)

_ w2) + vv_,

(5)

(6)

(7)

where, fl'= The

LHS

OV

Oy

Oz'

fl_=

of equation

(5.2)

represents

The

term

on

convection. vorticity

OW

first

by vortex

of vorticity).

The

line

stretching

second

and

the

OU

OW

Oz

Ox'

the RHS

increase

terms

Ox

0_1

production

acceleration

on the

OU

in streamwise

represents

(streamwise

third

fl'-

OV

RHS

vorticity

by

of streamwise

causes

represent

amplification the

increase

in

vorticity due to lateral skewing (by transverse pressure gradients) of vorticity in the transverse directions. These are the terms associated with skew-induced vorticity. the

primary

gradient.

The

fourth shear

The

fifth

term stresses and

represents and sixth

is often terms

the

production

neglected on

the

RHS

of streamwise since

it contains

represent

the

vorticity

by

a streamwise production

of

52 streamwise

vorticity

by inhomogeneity

of the transverse

normal

stress

auisotropy

and by the secondary shear stress, respectively. These terms are responsible stressed-induced vorticity. The last term on the RHS represents the diffusion streamwise

vorticity

by viscous

forces.

The significance

of equation

(5.2)

transition planes

duct

configuration.

of symmetry,

to Quadrant

Since

to avoid

I (refer

can now be discussed the

confusion,

to Fig.

3.2)

of axial

of the

3,4,5

velocity

components

onto

a uniform

grid

by central

difference

approximations.

at Stations by dashed location prised _z

3,4,5

and 6 are

contour

lines.

where only

being

The

the peak

filled

circles

axial

in these

cm) and

then evaluating

fly and

the

approximate

vorticity

_z,

the

wall curvature

field is com-

axial

induces

5.3 and illustrated of the flow occurs

area

in Quadrant

pressure quires which

Fig.

gradient

5.13.a

creates

(Station primarily

there to be a thin layer was not resolved in the

tive

vorticity

will occur

the pressure gradient half of transition, the transverse global

pressure

negative

straight

is primarily wall curvature

of the

(see flow.

5), show

intensifies. tion

that This

(vortex

the negative

stretching)

Fig.

vortices

intensify.

tion. _zR/Ub

Between

5 and

1 the

no-slip

duct

condition

without

of wall curvature. sign which causes Also,

transverse re-

the

that

since

In the second a reversal of the

contracting

be expected

swirl

area

the

causes

reversal

a

of the

migrates vorticity surface

towards is caused

curvature

the

6, the

reversed

pressure

the

midplane

by streamwise along

the

gradient

magnitude

of the

peak

sidewalls.

concave GSrtler

assists

the developing vortex pair. 6), the vorticity is diffused

and

accelera-

duct

instabilities, that Taylor-

z=0

the

curtype

inward

In the transition by turbulent ac-

vorticity

drops

from

= 2.4 to 0.09.

Miau et al. [9] evaluated of two CR transition ducts aspect

concave

flow between 5.13.d, Station

Stations

5.1).

in Fig. 5.1. due to the

in the very near-wall region The generation of nega-

not generated by centrifugal vorticity in the same manner

In addition,

(negative y direction) duct extension (Fig.

The

transverse

cancel the vorticity generated in the first half however, as Figs. 5.13.b and 5.13.c (Stations 4

of the

and are the

a function changes

vorticity

strengthening

Although the vortices vature will accentuate

vorticity.

CR transition

It might

pressure gradient would effectively of transition. This is not the case, and

that

of positive vorticity present measurements.

in any

gradient

acceleration

3) shows

component

the

pressure gradients which were discussed in Section In the first half of transition, a global deceleration expansion.

vorticity

is represented

mark

1, the

com-

transverse

of axial

vorticity

plots

across

vorticity the

Contours

At Station

2, the

The

Negative

components

at Station

changes

will be restricted

interpolating

x 0.508

5.13.

occurs.

vorticity

Beginning

by

(0.508

in Fig.

vorticity

of transverse

zero.

shown

calculated

to the present

vorticity

discussion

duct.

at Stations

derivatives

6 was

transition

ponent the

and

sign

the following

relative

for of

ratios

(AR)

of two,

all the terms in equation of constant cross- sectional but

the

transition

lengths

(5.2) area. differed:

at the Both

exit plane ducts had

L/D=0.54

and

53 L/D=I.08.

The

vorticity than

results

at the exit

turbulence

and the

of their

plane

induced

present quadrant,

observed

in the equivalent the

ducts.

One

of one,

and

ratio

duct the

of three.

vorticity duct.

had

the

ducts followed

aspect had

feature

exhibit

area

by a contraction

considerably

vorticity suggest

that

geometry

and

the condition

of the flowfield More

Reynolds

distinguishing

feature

in the vicinity

of the vortex

as they pairs with are

the

The these

data

applicable

of the

vortices

deformation vortices

detailed common

present

the

ferences studied duct

pair.

sets

than

plane

the path

the

expansion the

that

negative

These

is very

results

sensitive

the vortex

is amplified

inlet

length

for the

respectively.

vorticity

between

at these

In Section

to the

core takes

or attenuated.

flows

were mentioned. Eaton

[31,32] results

away

from

the

afford

to determine

the

flow.

is taken

station

the

Based 70 cm

reported Mehta

studies in the

should

as occurs based

the

at Stations and

Bradshaw

In addition be noted.

where

5 and Eaton

the origin the geometry

length

the

of their length, other

boundary

region

is three-dimensional.

vortices

results

for the

6, respectively.

in a two-dimensional of the

transition

The

was 97 cm downstream most

to development Whereas

vortex

on their

development

present

inasmuch

comparisons

purposes,

as the location

on this,

in the

to make

conclusions

in-

the studies

for embedded

For comparative

by Pauley and

of 135 cm.

embedded layer

2). 30 and

structure

are relevant,

opportunity

if modelling duct

(Station

surface,

6. A

experimental

In particular,

turbulence

and

duct

length

detailed

and

flow

5 and

is the turbulence

mean flow

and

stations

at Stations

2.3, several

in the transition

generators

vortices

was measured

to the transition begins

boundary

In contrast, an area

is the transition

[30] and Pauley

is approximately

a development

tensor

measurements

first measurement from

stress

vortex

and Bradshaw present

with

duct.

specifically,

AR=6

vorticity

For all ducts,

ducts 3.0,

of positive

for the

positive

had

feature

in the exit

if the initial

of the flowfield

of embedded

by Mehta

exhibited

vorticity

CR ratio

regions

vorticity

section.

vs. 1.5 and

[2,3]

of two

Results

The complete

vestigations

positive 1.08

will determine

Turbulence

transition

is

a length-to-diameter

small

plane

distinguishing

for the

plane

was constant.

exit

vorticity

a length-to-diameter

which

the duct

at the exit

and McCormick

of six and showed

in the

the

Another

L/D=0.54

the duct

through vorticity

of the transition.

through

and

rather results

in Quadrant

axial

exit

of negative

ducts

region

positive

Patrick

4),

Miau_s

is observed

in the

ratio

regions

the

through

area.

shorter

ducts:

large

between

negative

the exit

ducts.

of three

an aspect

and

corner

study,

of axial

3 and

between

in the

measurements

generation

(terms

axial vorticity

present

ratio

the

induced

of this investigation

duct

that

difference

vortieity

of their

from

an

cross-sectional

area equalled was

qm_drant

The results

which

axial of the

duct

for the AR=3

skew-

A notable

negative

5.13.c)

vorticity

other

A common

is that

5.5

Fig.

axial

the

Whereas

1 (upper calculated

5-7).

is that

sign.

showed

was primarily

(terms

study

plane is of opposite

analysis

layer,

results other

at dif-

investigators the

transition Also,

the

54 velocity deficit in the vicinity of the vortex cores is much larger in the transition duct than in the other studies. Mehta and Bradshaw placed their generators in the settling tential

chamber

so that

core flow velocity

and

Eaton,

test

section,

on the at the

were

made

Table

5.1.

(97

by the time

other

so that

somewhat

the velocity

hand,

cm).

Table

the vortices

5.1.

deficit

was initially

station in the

Velocity

deficit

percentage

at the

present, the

vortex

of the po-

the test section.

vortices

where

deficit

of the vortex

entered

their

data

Velocity

was a small

generated

a significant

downstream

deficit

Pauley

beginning

of the

but

had diminished

turbulence

measurements

core

region

is summarized

in

in the vicinity

cores.

x (cm) v/u,

Based

on

these

qualitatively Mehta and 5.5.1

results results

Pauley

& Eaton

Pauley

& Eaton

Pauley

& Eaton

[31]

in Fig. local

intensity

used

0.95

Mehta

& Bradshaw

[30]

90

0.95

Mehta

& Bradshaw

[30]

135

0.95

differences

alone,

similar

intensity

5.14.

Note

velocity

that

for reducing figures

contours

the

turbulence

the

and

present

Eaton

The

between

of the

the

results

than

will

to those

increase

Stations

intensity observed six

5.15-5.20.

intensity the

(u'/U)mat

5 and

Reynolds Negative

equations

(see

of

6 are

relative the

filled

circles

in Fig.

(peak

vorticity)

the

vortex

pair

thickness

evident.

local

Appendix

cores

,,_ 14% and

layer

6 is clearly

is defined Since

The

of the sidewall,

5 and

velocity.

vortex

in boundary

exhibits in the

here

bulk

data.

of the

intensity;

at Stations

response

hot-wire

location

measured

not

second-order

In the vicinity

6, respectively.

= 0, the turbulence the double inflection

and

all of the mark

5 and

in Figs.

that

of Pauley

u'/U

10% in places,

turbulence

Contours

results

component

of moderate

shown

it is expected

to the

contours

tongue

pair

0.75 0.90

an aid to interpretation.

the vortex

0.50

97 60

axial

subsequent

66

[31]

142

exceeds

C) were

[31]

[30]

turbulence

to the

0.40 0.50

& Bradshaw

be more Bradshaw.

shown

30 70

Mehta

Turbulence Axial

and

Present Present

5.14 as

extrudes

a

11% at Stations and

Along

distortion

by

the midplane

z

a double peak behavior which is a result of mean velocity profiles shown in Fig. 5.12.a. stress

components

contour

levels

at are

Stations

5 and

represented

by

6 are dashed

55 lines.

The

z/R

level

of symmetry

-- 0 is very

shown

in Fig.

between stress

5.20.

values

increase

in peak

axial

turbulence

axial

velocity

for the

are

higher

and _ Figs.

displaced

of the

intensity

(Fig.

5.14)

6 than

between

at 135 cm show u2

component

cm

show

the

plane

all the vortex

that

as the

a peak

in the

of peaks

near

in the

of symmetry.

The

vortex

present

region, outer

core

region

results

of the transverse

show

in the local

normal

of Mehta

along

the plane

results

of Pauley

also

show

of the

stress

a large

and They

on the

peak

also

about plane

to the

generators

component which

exhibits

two

peaks

were not observed

that

all the

normal

Anisotrop_y

(u 2 - v 2) is shown larger

core

at Station

nent

by as much the

& Eaton stresses

and

tical

normal

core

at Station

stress

of y/R

= 1.4,

Station

6, the are still

ticai

on

show

the

a

plane

of symmetry

and

components

=l:z/R

= 0.35,

vertical nearly

horizontal

stress

equal

are the

transverse

observed

vortex normal vorticity

midplane

at 97 cm suggest peak

axial

exceeds

similar

Anisotropy

of the

core stress

region.

5.22. equal.

the the

axial axial

Anisotro__py

components

in non-circular

axial

compolayer Pauley

of the the

axial

Near

is

vortex

boundary

behavior

in Fig.

than

transverse component

the

between

exceeds

less

behavior.

components.

to be nearly

stress

is everywhere

in the

of streamwise

report

core.

vertical

y--0

vicinity

transverse

(u 2 - w 2)) is shown

stresses

the

to a two-dimensional either

both

w 2 stress

the horizontal 5, the

component

than

vortex

the

data

In the

velocity

that

this double

At Station

transverse

about

and

to the

is present__directly

show

Eaton's

component.

is greater

of the

5, the

in the generation

the

& Bradshaw

vicinity

5.21.

This is in contrast

component Mehta

in the

in Fig.

and

exhibit component

horizontal

6, however,

axial

they

the

as 20%.

5. Pauley

stress

deficit

results located

will eventually

everywhere

velocity

symmetrically

bet._.ween the axial than

a large

6, the present

at Station

stresses

component

where

since

At Station

traced

at 97

(or lack of) in the vortex core re, on. It is likely that the vortices in the and Eaton study exhibited similar behavior to the present results at a the generators.

can be

the

Eaton

an additional located

peak

differences

and Bradshaw

of symmetry,

vortices.

symmetrically

These

The in peak

deficit Pauley

closer

pair.

sidewall. decrease

edge

behind

vortex

peak

roug____ half the peak magnitude the u 2 stress component differs

but edge

duct

an increase

The results The

sign

the Reynolds

5, and that

.to the

from

change

that

at the

location

of the

studies.

the

the

mldplane component

levels.

magnitude

increases.

near-wall

of symmetry

from

resulted

wall is approached

monotonically

show at Station

turbulence

the

components

is in contrast

which

6, the peak

stress

v 2 and w 2 are ne___r equal and are stress component u 2. Qualitatively,

significantly

about

U'_ stress

5.15-5.20

farther

stresses

of the high

5 and

component

at Station

and

magnitude

Stations

components of the axial

the _

distorted

in the vicinity

At both

stress

dit_cult-to-measure

1 and 2. In general,

are more

contour

for each

even

As expected,

Quadrants contours

pair

good,

normal and

the

In the

ver-

vortex vicinity

component.

At

component,

but

be..._tween the

ver-

(v 2 - w 2) is important ducts

(see Section

5.4.1).

56 Contours differences

of this quantity are plotted in Fig. 5.23. These plots show significant between Stations 5 and 6. It is well known that in two-dimensional

boundary

layers,

transverse

component

acts

normal

duct

flow,

both

Stations

layers

u-_ > u-'_ > u"_, where

surface

the above

component

6, the

at Station

Pauley

Eaton

4.39

and

of Ref. The

primary

stress

levels

where

which

transition

lower

anisotropy

to the

and

is pos-

intensified.

results

length

the

In the region

the auisotropy

has grown

similar

at

exceeds

anisotropy). where

region

wall boundary

reported

by

of 97 cm (see

Fig.

in Fig.

in the velocity mean

y and

this

stress

stress

data

mean

velocity

summit quantity

vorticity is of the

(see

same

Section

profile

plays

primary

as a result, shear

ridge

W_

is crossed.

an important

5.4.1).

In the

of magnitude

>

Along

the

and,

the primary

that

Section

(OU/Oy

flattens, zero

These

5.4.3).

of the velocity

order

cores.

rate-of-strain

(see

In the z-direction,

is another

to be positive

vortex

is nearly

sign as the

_

is observed of the

ridge

z directions

are depressed.

stress

5.18

vicinity

of the positive

axial

the

of streamwise

cores,

shown

role

vicinity

of

as the

primary

kinetic

energy.

stresses. The

These

normal

results

kinetic

are

energy

duction

shown

was

also

of kinetic

were

for

used

Stations

calculated.

energy

to calculate 5 and

is given

of this quantity

energy

are

associated

on

duct

upper

the

near-wall

region

is essentially 35%.

with and

high

lower

when these the

stations.

same,

but

production

walls

at

to the Between

the

The

production

derivatives,

of

the

pro-

OU

in Fig. 5.25.

less than

compared

5.24.

streamwise

- --uw

shown

is significantly

disproportionate between

are

turbulence

by:

P= Contours

the

6 in Fig.

Neglecting

OU

ergy

and

at a development

region

also changes

shear

production

shear

_

the

in both

in Fig. 5.19

vortex

For the

to the surface

exists

very

aforementioned

in this region

secondary

the

of positive

are a result

the

z=0

rate-of-strain

in the

pocket

qualitatively

stress

stress

with

midplane

shown

(negative

for a small

of negative

mean

The

shear

except

0) associated the

to the surface

pair

54).

to hold true in the near-wall tangent

for a vortex

u-_ is the

u_ is the component

of Ref.

component

region

component,

31).

everywhere regions

and

18.5

of the upper

5, a small

6 are

axial

region

the

acts normal

By Station

Fig.

is observed

pair at Station

The results

surface,

e.g.,

6. In the outer

5, however,

which

of the vortex

to the

(see,

inequality

5 and

at Station

itive.

tangential

to the

u-_ is the

peak

(5.3)

Generally,

rates.

A notable

Station

5.

that

Here,

observed

relatively Stations

kinetic

high

5 and

exception

increases

occurs in the

6, which

difference 6, the

of kinetic

production

at Station

small

energy

the

levels

peak by

seems

in kinetic production

approximately

en-

57

5.5.2

Turbulence

profiles

In this section, minor

axes

at Station stress

Reynolds

are presented 1.

In the

components

stress

profiles

and compared

following

profile

are relative

along

with

the duct

the

plots,

initial

the

to wall coordinates

by the bulk velocity

local

layer

boundary

thickness,

5._.

Normalized

of the

x, y, z laboratory

(b/R)

at Stations

1,5 and

Reb

in Fig. the

5.26.

are shown

Reb

1

0.31

0.31

0.29

0.29

5

-

0.21

0.58

0.18

6

-

-

0.65

0.25

sidewall

vortex

destabilizing flow away 5, the

axial

strong

attenuation

components levels. The between are

friction

normal

made

and

mean shear

the

lateral

in the

to scale in the

velocity

with duct

was

6. the

outer

region

components

show

region.

Near

In wall-bounded friction at

energy along

boundary

curvature and the developing the

boundary

wall

and

layer.

In

The

resulting

in the

5.12.a.

largest

common At

double Station

deviation

with

6, these

stress

but are still below the initial in the near-wall region decrease

velocity. are

6 along

kinetic

near

in Fig. the

shown

5 and

The

the wall at Station

shear

Stations

results

deficit

1 are

distributions

curvature.

shown

These

the

profiles.

of the

concave

of velocity

local

and

convergence

profiles

mid-planes

deduced.

that

a region stress

stress

initial

velocity

in the near-wall

5 and

at Stations

first to stabilizing convex second half of transition,

destabilizing

creates

at Station

shear

apparent

have increased to a certain extent, transverse normal stress components

Stations

known

were

wall

and

from

stabilizing

divergence

of the

ys-axis

measured

It is readily

experiences

the

behavior

390,000

measured

5.27

considerably

creates

flow

from

infection

5.28.

in Fig.

sidewalls is subjected convergence. In the pair

the

distributions

distributions

shown

deviate

lateral

addition,

stress

in Fig.

=

cm

stress are

axis

layer on the duct stabilizing lateral

6.

y2-axis

axis

semi-major

thickness

ys-axis

Normal

by the

5.2.

88,000

Reynolds

semi-major

profiles the

normal

is normalized

y2-axis

R = 10.214

-"

layer

Reynolds

is always directed along 3.2). All of the stresses

in Table

boundary

measured

designation

the wall coordinate

as summarized

Table

Station

Reference

and

and semi-

and not the

coordinates; that is, the v fluctuating velocity component the wall coordinate of interest, either y2 or Ys (see Fig. are normalized

semi-major distributions

layers,

the

Preston

1,5 and summarized

Reynolds tube

6 from

stresses

measurements which

in Table

the 5.3.

local

58 Table

5.8.

Normalized

drop

and

components velocity to

lag

Station

y2 =0

ys =0

1

4.06

4.06

5

2.70

4.89

6

3.23

4.38

behavior stress

vortex

pair

6.

normal

Several

stress,

5.3.

The

friction

in the near-wall

further

away

from

stresses

outweighs

transverse

the

to the

attenuation

First,

the

primary

of the

boundary

layer

axial

and

flow of the vortex

acts

to suppress

that

increase

as the in the

normal

stress

Stations

turbulence

(aU/ay2),

outward

appear

decrease

between

of the

stress friction

stresses

The

in the

is observed

rate-of-strain

by the common

the

it is presumed

sidewalls.

increase

contribute

shear

with

will eventually

duct

energy

and

normal

although

region

the

normal

6 correlates

velocity,

kinetic

is reduced

convergence

Reynolds 1 and

in turbulence

factors

wall region.

axial

Stations

in Table of the

components

a decrease

in the

between

shown

moves

transverse

rise

wall

development

all the

that

the

Reb = 390,000.

m/s

subsequent

near

the

velocity

x 100) at mid-planes,

Ub= 29.95 The

friction

(U,./U,

in the

hence

so

5 and near-

the wall shear

pair.

Second,

lateral

turbulence.

And

finally,

Fig. 5.12.a shows that between Stations 4 and 6, the near-wall flow to streamwise acceleration which suppresses turbulence generation.

is subjected The high

turbulence

are

levels

observed

of high primary

rates-of-strain

The history cantly

half

the flow along the boundary

destabilizing

divergence 6 are

shown

in Fig.

ited

by the

was

most

and

distributions shown

bevel

gear

serious

boundary

Wall

and

the

shear necessary and

Over

able to be measured, the net effect small. The trends show an increase decrease The

in the

combined

5, the

shear

is recognized further,

transverse effect

stress that

turbulence

normal

on the value

one

the

point

measurements

range

axis).

half

In the

kinetic

of the

of transition, curvature.

axis energy

at Stations

5 and

rotation.

surfaces

concave

convex

semi-minor

and

duct

is signifi-

at Stations profiles

are

6 were

lim-

This

limitation

at

Station

5 where

boundary

layer

that

was

of the above flow conditioning is relatively in the axial normal stress component and stress

components kinetic

the

is not

second

for probe

lower

turbulence

nearest

stress

axis)

to destabilizing

stabilizing duct

measurements

upper

is thinnest.

the

a result

deficit.

(semi-major

In the

experiences

layer

(semi-minor

is subjected

along

arrangement

along

layer

flow

proximity

velocity

z=O mld'pl_e

layer

divergence.

the

5.29

the

boundary

y=O mid-plane

measured

in Fig.

5.30.

of the with

the

lateral

decreases

stress

5 and

the

the

from

and

Normal

associated

of transition,

curvature the

mid-region

of the flow along

different

first

in the

wall the

is very

is observed

statistically along

between

energy

to decrease,

significant. y3 axis

were

Stations small.

1 and

6.

At Station although

To investigate made

a

at a bulk

it this

operat-

59 ing Reynolds

number

region

could

shown

in Fig. 5.31.

els and

the

or that

indicates

wall shear

speculate behavior

both

equilibrium

which

will be examined

more

closely

Considerations

The

calculations

by

performance

lating

predictions,

transition

must

also

duct

be included.

difficult

aspect

gradient

effects

Since

presence

of the

ity,

however,

relies

efforts

The first

group

prediction shown

for the

group

The present The

with

detailed

set should

be useful

simplest

turbulence

model

on the

to behave

concept like the

for incompressible

and

vt is the k is the

is the

dimensional important, are

not

flows inasmuch in agreement

where

flow is written

as:

only

as the with

viscosity,

even

diffusion

of vortic-

components.

into

two

groups.

parameters

and

et al. [33] has of embedded

length

models.

The

of particular

stress

(zero-order)

.

behavior.

the

stresses

Reynolds

Kronecker

(5.4)

shear stresses

flows.

Reynolds and

viscosity are stress

(5.4)

/_ii is the

normal

simple

the

eddy

2

Equation

Reynolds

Reynolds

will predict

stress

Reynolds

stresses

OUj

energy. the

models

the ability

and

of

via pressure

in the region

Here,

viscosity

eddy

vorticity

mixing

algebraic

of Boussinesq.

kinetic

in the neighborhood

work of Liandr_t simple

calcu-

of turbulence the most

performance

flow

molecular

turbulent

turbulence

when

for

of course,

of the

flowfield

even

for this group.

.OUi

where

that

required,

can be divided overall

mean

5. Near-wall

standpoint,

Reynolds

the task of demonstrating

to predict

to

effects

even simple

of the

well with

data

based

assumed

is concerned

models

the

prediction

of the mean

reasonably

of

be neglected

of streamwise

with

the benefit

4, it is difficult

shown

modelling

The computational

features

have

cannot

configuration

primarily

region

5.7.

is the flowfleld

Accurate

levstress

of the Preston

Without

From a computational

modelling

present

is concerned

the primary

turbulence

model

pairs.

are

energy

in shear

at Station

predictions

process,

a larger

in the near-wall 3 and

et al. [7,8]

generation

on accurate

can be predicted

second

tensor

vortex

Stations

of turbulence

an inviscid

of flow separation.

that

vortices

the wall.

configuration

the initial

is essentially

the

Modelling

level desired.

of the present

pair.

near

effects

kinetic

by means

recovers

For accurate

The

on the information

Burley viscous

flows.

depends the vortex

measm_

and

for this case

decrease

stress

in Section

Modelling

overall

The

lead to the flow condition

Turbulence inviscid

is thicker profiles

the unchanged

stress

the shear

energy

stress.

at the intermediate

on the factors

layer

kinetic

confirm

either

measurements

and

of the shear

the flow is not in local

turbulence

6.6

results

that

the boundary

stress

behavior

to the increased

This

where

Shear

These

decreasing

is in contrast tubes.

of 88,000

be resolved.

For

stresses

(u--7_',

predicted the

delta

is applicable

present

function to three-

i #

by this

j)

are

equation

configuration,

6O equation transition

(5.4) may be adequate for prediction purposes through the end of the section, but will not be applicable if the flow is allowed to develop in the

rectangular in the

duct

generation

Since

gradients

flow solvers

normal

flow.

equation

secondary

reduces

are important

flows. with

to examine

For the primary (5.4)

stresses

are implemented

it is worthwhile

present

neglected,

Reynolds

of corner-generated

models,

for the

are

transverse

design-oriented

turbulence

viscosity

the

and diffusion

many

viscosity eddy

where

the

shear

algebraic behavior

stresses,

eddy of the

if streamwise

to:

OU

=

(5.5) OU

= Rearranging

these

equations

to solve

(5.6)

for the eddy

viscosity

yields:

(5.7)

= v,,, = _ l(OV)Oz where

the

eddy

y and

z subscripts

viscosity.

equations these

(5.7)

plots

were

The

are

and

for

in Figs.

These

Station

5.34 solid

computed

only

5.35.

and

symmetry that the and

strain

the

eddy

regions in the

shown about results

results,

rates-of-strain

5 is shown along

plotted 5.33.

To avoid

value

in terms

5.32

and

cross 5.35

for

where

and

plane. show

5.32.

in

Also

in

5.34),

(vt/v), 6 are

eddy

and

Although similar

viscosities

strain-rates

the

(5.7)

indicated

ratio

Station

the

the

trends

viscosity

eddy with

is

is greater viscosity respect

to

the midplane z=O, the magnitude of the differences is enough so should be considered only qualitative, In areas where the stress

in both viscosity

transverse

directions

components

where large deviations component strain rates

component

plotting.

are shown vanish

(5.8)

are

are

observed

nominally

of the

to be nearly

same

equal.

magnitude, Most

in their

embedded

of the

occur can be traced to either a large difference or to inadequate resolution of a large velocity

gradient, e.g., in the vicinity of the velocity ridge near the vortex core. This is in contrast to the results of Mehta and Bradshaw, who reported vt,y

of the

(denominators eddy

of a viscosity

in equations

in the

(anisotropy)

component

results

points

in Figs.

5.33

the

Equivalent

denominator

in Figs.

in Fig.

which

singular

lines

the

5% of its maximum

distributions

dependence

primary

at Station traverses

dashed

where

directional

of the

5 in Fig.

and

(between than

(5.8))

horizontal

calculated.

shown

admit

behavior

(5.8)

vortices

was

so ill-behaved

this flow. that the

so as to preclude

61 It was mentioned strain

often

terms

which

this

exhibit

more

account

is that

which terms

in Chapter

the

2 that

spectacular

behavior

for the extra

extra

boundary

strain

rates-of-strain

layer flows

than

rates.

cause

what

with extra

is predicted

It is argued

large

changes

that

rates-of-

by explicit

the

reason

in hlgher-order

for

terms

appear in the Reynolds stress transport equations. These higher-order are modelled in terms of the Reynolds stresses and require the specification

of empirical constants. Dimensionless turbulence structure parameters the cross-correlation coefficient are related to constants in turbulence

such as models.

Often

flows

such

when

applied

the constants

two-dlmensional to more

are determined

boundary

complex

flowfields.

computed

for the present

Ruw

= _"w/u_w

' and

R,,_ the

5.36,

and duct

5.37

R,w

sidewalls due

5, in the

region

value

and

is close

R,,w

to the

parameter

secondary in the

range

is the

shear-energy

of the

resultant

turbulence

of the

peak

is observed.

ratio

been

et al. [56]).

energy.

observed

For the

with

al.

but

to be constant flow, aly

with

R,_

a value

decreased

lower

walls, value.

shows

The

is also depressed of interest as the

surface

to twice

layer,

this

(see,

e.g.,

of 0.15

the

values

is defined

boundary

is

Station constant

has

parameter

to a wall

two shear-energy

At

a near

value

this parameter

Near

parameter

stations

parameter

normal

correlations

two-dimensional at both

',

shown

of 0.45.

layer.

and

dimensionless

In a two-dimensional

present

upper

the

This

in a plane

the

6, the

the

= _"_/v'w'

Another

parameter

stress

occur,

Station

of the flowfield, core.

stress

boundary

Near

is in agreement

vortex

cores,

of the

= _'_/ulv

6 are

value

intensities By

value. Rvw

shear

as

parameters

R_v

5 and

vortex

turbulence

correlation

shear

kinetic

also

the

of 0.25 for most

region

of the

0.55

at Stations

two-dimensional

convergence

two-dimensional

stress

in the

vicinity

structure

correlations

The primary the

of simple

performance

stress

evaluated

with

lateral

generally

shear

dimensionless

The shear

respectively.

in the

where

of these

= _-'_/v'w'

5.38,

to strong

in poor

flow.

be compared

of approximately

and

has

and

may

depressed

Rvw

on the results

resulting

Some

were

in Figs.

based

layers,

parameters

were

ratio the

parameter Bradshaw

computed:

(5.9)

= x/_'fi 2 + _"_2 /2k

(5.10) which

are

walls,

respectively.

in Fig.

applicable

5.39

and

away The

the

from

the

corner

al v shear-ratio

al= parameter

region

parameter is shown

on the

vertical

at Stations

in Fig.

5.40.

high turbulence levels are observed, the al_ parameter dimensional value and, like the shear stress correlation,

and 5 and

In the

horizontal 6 is shown

regions

where

is higher than the twois well below the initial

value in the region of the vortex pair. Along the horizontal walls, the alz parameter is elevated at Station 5, but returns to the initial value at Station 6. These results structure,

show

that

more

the

transition

so at Station

duct 5 than

produces at Station

a distortion 6. The

largest

of the

turbulence

distortion

occurs

62 in the

vicinity

of the

vortex

core

where lateral

region

convergence

suppresses

turbulence. Further ining

the

insight

data

into the

in terms

structure

of the

of the

invariants

turbulence

of the

can be gained

anisotropic

stress

by exam-

tensor.

This

analysis is based on the concept of physical realizability limits of turbulence which has been used extensively to study the return to isotropy of homogeneous turbulence

(see,

by Rotta

e.g.,

Lumley

[58], is defined

[57]).

The

tensor

where

must

I,II

satisfy

and

III

stress

tensor,

first

proposed

as:

u, i-

bij = This

anisotropic

the

are

}k6,

2k

Cayley-Hamilton

theorem:

_3 _ I_2

+ II),

tensor

invariants:

the

(5.11)

-

III

(5.12)

= 0

I = bii 1 = --bi, 2

II

III The first invariant,

I, is identically

ergy.

Invariant

defines

stress

tensor.

A positive

component ponents

III

that are

applying the as the strain to infinity,

is large,

large.

and

Limits

J

zero by definition indicates

a negative the

of the turbulence

of the ellipsoid

of III

on

b,i

= _bi._b._kbk_

the shape value

d

associated

that

value

indicates

anisotropic

stress

--_ u goes

to zero

and

bll

goes

in any one direction If all the turbulence

to -1/3.

5.41.

recast

The

shaded

area

realizable turbulence is also labeled. The

in terms

bulence

of the

is primarily

III

invariant

Fig.

5.42.a.

is only

one

principle

that

two

principle

can

be

anisotropic contained

will always Following

on this plot

stress

be positive.

Pauley

tensor axial The

and Eaton,

level

of turbulence

invariants location

region

is bl 1

turbulence are are illustrated within

which

all

of the turbulence at the at Stations 1, 5 and 6 was

invariants.

component the

the

nature tensor

by

must go to zero as OU/Oz goes

on allowable These limits

represents

must lie. The Reynolds stress in the

largest

com-

defined

is 2k, which occurs when them turbulence is in the z direction, then u 2 = 2k and

is equal to 2/3. Following Lumley [571, the limits recast in terms of the tensor invariants II and III.

physically boundaries

Reynolds

tensor

The

en-

the

condition that the Reynolds stress in any direction rate in that direction goes to infinity. For example,

that can occur one-dimensional.

in Fig.

there

kinetic

with

At

Station

I, the

so it is expected at Station of each

1 are data

tur-

that

the

plotted

in

point

on the

63 invariant which

map

relative

has its origin

invariant

map.

location

This

in the

to positive

vector

layer

dimensional though, turbulence.

which

That

one component

high

of anisotropy.

level

tions

5 and

observed the

the

6 are

rates

layer

cluster

more

limit

indicating

5.7' Wall The

use

Since

flow

These

along 5.46.

two

suggests layer

which

tubes were

both

tion

velocities

used

pitot

terval

however,

The

and

are

plots

rates

are

strain

contraction dominant.

are

30 and

toward This

friction

on the

plots.

< 0.6, inagreement

downstream. obtained in Fig.

average

Bound-

At

Station

greater

than

implies

The

is satisfied

5 fric-

fines

individual

5, deviations 80.

in

at Stations 5.47.

of the

behavior that

shown

distribution

diameters

law-of-the-wall

in Fig. 5 show

Although

(n=1,2,5,6,7,8)

shown

outside

behavior

for s/s,.el good

5 and

measured

are shown at Station

relatively

Results

based

data

at 5.45.

velocity,

at Stations tube

behavior

are

analyzed in Fig.

at this station.

6, the law-of-the-wall 200.

was

local

Preston

y,, traverses

for y+ values

tend

flow

are shown the

exist

probes.

on these

apparent

At Station

with

6, the

coordinates,

non-dimensional

all profiles

between

the

hot-wire

superimposed

30 < y+ 4.125

Co

1.5463591

0.51544988

10.000000

Ca

0:0000000

0.00000000

0.0000000

Cz

0.0000000

0.00000000

0.0000000

C3 C4

0.0000000

0.00000000

0.0000000

0.0000000

0.00000000

0.0000000

C5

0.0000000

0.00000000

0.0000000

R = 10.214

cm

137 Table

A._.

Transition

Duct

Geometry

Variables

Data Station

x/R

air

b/R

1

-1.00

1.0000

1.0000

2.0000

1.0000

1.0000

2

1.00

1.0000

1.0000

2.0000

1.0000

1.0000

1.12

1.0003

0.9997

2.1047

1.0006

1,0192

1.24

1.0025

0.9978

2.2154

1.0047

1.0376

1.36

1.0078

0.9831

2.3331

1.0251

1.0555

1.48

1.0174

0.9746

2.4588

1.0439

1.0729

1.60

1.0316

0.9719

2.5934

1.0614

1.0897

1.72

1.0509

0.9548

2.7383

1.1006

1.1056

1.84

1.0752

0.9333

2.8947

1.1520

1.1198

3

4

5 6

AR

Ac,/Ac,.i

1.96

1.1041

0.9177

3.0645

1.2031

1.1319

2.08

1.1371

0.8784

3.2497

1.2945

1.1410

2.20

1.1734

0.8462

3.4528

1.3867

1.1464

2.32

1.2123

0.8117

3.6770

1.4935

1.1475

2.44

1.2527

0.7759

3.9261

1.6145

1.1441

2.56

1.2936

0.7396

4.2050

1.7491

1.1362

2.68

1.3341

0.7037

4.5199

1.8958

1.1241

2.80

1.3729

0.6693

4.879I

2.0512

1.1087

2.92

1.4093

0.6370

5.2932

2.2124

1.0911

3.04

1.4423

0.6078

5.7775

2.3730

1.0725

3.16

1.4712

0.5821

6.3524

2.5274

1.0544

3.28

1.4954

0.5606

7.0484

2.6675

1.0381

3.40

1.5147

0.5435

7.9106

2.7869

1.0248

3.52 3.64

1.5290 1.5385

0.5308 0.5224

9.0131 10.000

2.8806 2.9451

1.0152 1.0083

3.76

1.5439

0.5176

10.000

2.9828

1.0026

3.88

1.5460

0.5157

10.000

2.9979

1.0003

4.00

1.5464

0.5154

10.000

3.0000

1.0000

1.5464

0.5154

10.000

3.0000

1.0000

8.00 R -

_7

10.214

cm

APPENDIX HOT-WIRE An

all

new

present

integrated

study.

department output, the

signal.

ACQUISITION

data

from are

appendix

important

components

of the

was

ones

used

directly

performed

contains

of the more

system

previous

operating

operations

This

acquisition

differs than

SYSTEM

on the

acquired within

analog

on a digital

a description

operating

for the

the M.E.

anemometer

representation

of the

system

of

and

an

features.

Hardware

(TSI)

primary

IFA

converter

100 Intelligent and

Workstation B.1. The two

IFA

signal.

The

range

of the

a voltage

usable

span

The more

IFA

less than

signal

must

The

an RS-232 this

link

active

are

the

channels

generates

transducer must

number

controls

starting

a 16-bit

which

the

data voltage.

be stripped

to the

FIFO

first-in, output

to a DRQ3B

P_I_/_'_

(channel rate.

and DMA

high

(FIFO) are

sent

controller

INTENTJONALLY_!1

A/D

as possible,

The

a DMA

through through

in the

the

first),

the

the

A/D

binary

number

channel

converting

word

of

converter

representation

digitizer

data

(Direct

to

characters

before

data

The

of up

are controlled

the

word

or

ASCII

sample,

contain

rates

of one changing

contain

16-bit

conversions on rapidly

that

each

data

memory. card

voltage will have

data

to be sampled For

4-bits

16-bit

via

input

range

conversions

parameters

low 12-bits

voltage.

first-out

A/D

of collecting

The

address

the

the

the

output)

film sensors

of this

IFA 200 by sending

sampling The

(when

amplified.

Selectable

The

a DC offset and

and

a low-pass

(anemometer

and

as much

of a Model

contain

display wire

in Fig.

anemometers

input

simultaneous

link.

from

Most

and

consisting

of applying to the

158 digital

capacity

the

word.

to a hexadecimal

word-deep

the

unit

conditioners

To utilize

signals.

communication and

of the

this.

provide

gain

volts.

high-speed

output

computer

serial

Model

be offset

200 provides

signal

(A/D) Scientific

is illustrated

temperature

capable

and

Incorporated

VAXLab/VSII

of this equipment

The

is +5

much

Hz per channel

signals.

of the

digitizer

(DEC)

150 constant

mode)

Systems

IFA 200 analog-to-digital

is a two-channel

an amplifier

DC

range

anemometer

50,000

and

IFA200

transducer

study

two Model

is in the

a TSI

Corporation

conditioners.

filter

filter

are a Thermal

schematic

in this

housing

a high-pass

Analyzer,

A block

157 signal

high-pass

system

Equipment

100 used

cabinet

Model

filter,

Flow

a Digital computer.

158 slave

cable

system rather

reduction

of some

The

the

This

data

analog

B.1

hot-wire

in that,

overview

DATA

B

address the

binary

is loaded

into

a 64-

enter

FIFO

pass

that Memory

the

Access)

interface

computer.

PRECEDZNG

PAGE

BLANK

NOT

FILMED

140 The that

DRQ3B

allows

of up

Parallel

real-time

to 1.3 megaHz.

buffering,

that

ventional

single

can

the

time

The

I/O

DRQ3B

between

has

increase

DMA

but

module

of 16-bit

significantly

buffer

fills, it is dequeued, ing

DMA

collection

transfers.

requires

the

To improve

enqueued

before

automatically

any

result

B.2

Buffer-Swinging

in the

method

of data

of data

and

enqueued

to the

enqueued,

data

DRQ3B

is filling, the data the second buffer process again B.3

repeats

prior

begins

to any to the filling

second

when

can

the

buffer's

are

sent

the

next the will

the

to be

DRQ3B

intervention.

transfer

exceed

rates.

the

available

are forwarded

directly

of the

buffer-swinging

a continuous memory

the

last

the

and

are

buffer

buffer

fills,

second

to the hard transferred

to disk

stream buffers

the first

While

written directly data have been

Dur-

buffers

fills,

When

buffer.

buffer.

or more

or more

When

a buffer

must not exceed IFA200, or data

higher

transfer.

second

next

software

data

to con-

when

enqueueing

to capture

buffer.

data

as compared

quickly

Two

data

first

double

a buffer

by means

is used

to as

the

any

enabling

manner.

referred DMA,

two

the digitized

feature

rates

and

without

computer

in the first buffer are fills and first buffer's as the

receives

This

begins

automatically

this,

in the

transfer

buffers in the

allows

interface

at

rates

to enqueue

rates

memory

data

buffer

buffer

buffers,

following

DRQ3B transfer

call

Now,

sampling

disk

in the

single

buffer

between

acquisition.

operates

transfer

DRQ3B

next

To avoid

hard

data With

transferred.

at high

computer.

feature,

time between FIFO memory

the

the

time

data

71 megabyte

are

filling

is less down

Transferring to the

situation,

data

begins

The

memory

this

a unique

previous

buffer, no data are transferred. This time it takes to fill the 64-word-deep be lost.

digital

a software

dequeueing

is a high-speed

parallel

is the

buffer

disk. When to disk, the

the

first

buffer

data.

Handshaking Synchronization

of data

transfer

between

the

IFA

200 and

troller card is accomplished by an interlocked two-wire erly implemented, the handshake prevents data overrun from input

occurring. port

REQUEST t At the port plish

DMA this,

FUNCT

and

BUSY

time

the

transfers the

The

pin-to-pin

200

DMA

listed

The

are connected,

DMA

pins,

B.1.

0, STROBE

DMA

Interface

respectively,

on the

data

acquisition

from

the

IFA 200 to the

cable

had

to be modified

connections

Interface

in Table

OUT

via the

Cable

between and

the

equipment

the

ACK

pins

to the

IFA 200 output

DRQ3B

on

controller

card.

J2 output on the

with connectors

DRQ3B

the

DRQ3B CYCLE

port. TSI

con-

When propconditions

READY,

acquired,

for compatibility port

DRQ3B

handshake. and underrun

Cable,

was

J1 and

J2 input

and

the

j For data did

not

sup-

To accomthe

DRQ3B.

on the DMA

card

IFA are

141 transfer high.

to begin, When

the

strobe,

begins

QUEST

line

the

last

the

ACK

last

buffer

computer

low,

indicating low,

that

Once

the

digitizer

following

1. When

the

of the

has

IFA

200,

releases

to receive

been

activated

data

and

When sets

the the

data

CYCLE

are

3. When

IFA 200 receives not

the

of the

available,

DRQ3B

IFA

DRQ3B

REQUEST

200,

FIFO

places

the

it releases

it places starting B.4

IFA valid

the

high,

prepares the

the

DMA

the

the

value

the ACK value. the

ACK

line

case

is held

200 receives

BUSY

high

from

ACK

lines

and

that word.

the

asserts

the

is read

out

DRQ3B

line

CYCLE

line

(unless

a word

the

it

valid

REQUEST

it to go high low until that

data

data

CYCLE

to

DRQ3B,

DRQ3B

next

the

line

indicating

line of the

indicates

the

low,

to the

allowing

ACK

en-

in its 512-word-deep

ACK the

from

been

REQUEST

signal

indicating high

have

CYCLE

high

on

the

it to go high,

buffers

ACK

data

asserts

When

transfer:

memory).

cycle

REUntil

DRQ3B data.

line allowing

to transmit

STROBE

is full, in which

of FIFO and into for the next word. 4. When

and

receives

the

low from

line

CYCLE

available.

to receive

low from

and

BUSY

on the sampling the

are

computer,

for data

FIFO memory. The DRQ3B then asserts the IFA 200 that it has received the data 2.

data

ACK

occur

STROBE

the

asserts

DRQ3B

data.

of events

it reads

it turns

and

ready

the

0 pin on the

line, valid

by the

it is not

DRQ3B

receives

that

DRQ3B that

sequence

DRQ3B

OUT

memory

DRQ3B

on the

it is now ready

the the

FIFO

to the

indicating

is enqueued,

the FUNCT

this on the READY

its 64-word-deep

is enqueued

signal

indicating

asserts

IFA 200 senses

filling

buffer

queued,

the

of the

is ready DRQ3B,

REQUEST

low,

over.

Software Control

subroutine of the ing,

of the library.

DRQ3B This

experimental

plotting

main

study,

and

programs

accomplished

analysis.

written

in the

by sending

by means

B.5

Digital

rely

The hot-wire on an accurate

output variable

of the

The

through provides

real-time

data

subroutines

FORTRAN

ASCII

characters,

FORTRAN

WRITE

the

useful are

of DEC's

called

from

Control

an RS-232

serial

VAXlab

for all aspects

acquisition,

language. via

use

programs

signal

process-

user

supplied

of the

IFA200

is

communication

statement.

Sampling

signal and

package

including,

data

link,

velocity,

is accomplished

software

techniques evaluation

from

the

therefore

is made

here.

employed in the of the mean and

hot-wire the

If the

present variance

anemometer.

anemometer instantaneous

output voltage

The

study (see Appendix C) (E and e 2) of the analog usual

voltage,

assumption is an

is decomposed

ergodic

that

the

random

into a mean

and

142 D

fluctuating

component,

E = E @ e, the true

by a finite

average

of samples

N are taken:

of discrete

= rnmoo .,o

At

E. are

is the

T, the

the

time

number

are

required

the

waveform

increment

by

hot-wire

convergence

data

point

time

were

taken more

very

which

yields

and than

at

mean

it was adequate

For Since

high

and

observed

be estimated large

number

time

of 150,000 the

for all regions

above

total

not rates.

For set

samples was

transfer transition

time

time

statistical

and

duration quantities

necessary

were

(B.2)

sampling

sampling

single

values of the

(B.1)

= (E.(nAt)--_)2

a fixed only

sampling

mean-square that

-_

it was

duration

a total

and

can

a sufficiently

T is the

measurements, rate

of the

_

samples.

sampling

respectively,

The

that

samples,

is N = T/At.

digitizing

the

--_)2dt

between

average

= E.(.AO

digitized

of samples

measurements, seconds,

individual

for the

provided

E(t)dt

f,:°+r(E(0

--e_=limo¢_l where

samples,

temporal

to "capture" the

turbulence

at 10 kHz for each

data

monitored rate duct

and

15

point. for each

and

sampling

flowfield.

143 Table

B.I.

DMA

IFA Signal Name

Interface

Cable

200

Pin

Connections

DRQ3B

Port No.

Pin No.

Pin No.

Port No.

Signal Name

J1

02

17

J2

STROBE

READY GND

J1 J2

06 01

22 42

_

FUNCT OUT GND

BUSY GND

_

02 03

18 41

ACK GND

GND

04

40

GND

GND

07

39

GND

GND

11

38

GND

GND

12

37

GND

GND

13

36

GND

GND

15

35

GND

GND

17

34

GND

GND GND

19 2O

32 31

GND GND

GND

21

29

GND

GND

22

28

GND

GND

23

27

GND

GND

24

26

GND

CYCLE

REQUEST

DATA

OUT

00

39

01

DATA

IN 00

DATA

OUT

01

37

02

DATA

IN 01

DATA

OUT

02

35

03

DATA

IN 02

DATA

OUT

03

33

04

DATA

IN 03

DATA

OUT

04

31

05

DATA

IN 04

DATA

OUT

05

29

06

DATA

IN 05

DATA

OUT

06

27

07

DATA

IN 06

DATA

OUT

07

25

08

DATA

IN 07

DATA

OUT

08

26

09

DATA

IN 08

DATA

OUT

09

28

10

DATA

IN 09

DATA

OUT

10

30

11

DATA

IN 10

DATA

OUT

11

32

12

DATA

IN 11

DATA

OUT

12

34

13

DATA

IN 12

DATA

OUT

13

36

14

DATA

IN 13

DATA

OUT

14

38

15

DATA

IN 14

DATA

OUT

15

40

16

DATA

IN 15

Note:

Unlisted

32 pins

are not

connected.

32

0

144

%PA

100

Anemometer

Model Digital Display

158

AL Model 150 Constant Temperature Anemometer

Model 157 Signal Conditioner

Ch.l Ch.2

J

Hot-Wlre(a)

O U

Oh. i

e4

Ch. 2

+

N !

_

IFA

200

A/D

Converter DMA

Interfa¢

Cable

VAXLab/VSII VR260

Ip-

--

DRQ3B DMA Module

I/0

Display

w

m_

+

r

Processor

Hard Disk Drive

Fig.

CPU

BUS

i_.._

Main Memory

iii+_;+i.;+iiii Keyboard

B.I.

Schematic

e2

of hot-wire

data

acquisition

system.

APPENDIX HOT-WIRE C.1

TECHNIQUES

Introduction The

derivation

the Method mean

and

of these

of the mean

A technique

The

response

in this

equations,

for the

in situations

when

This,

wire technique flow

and

amounts used.

The

individual

use

required

are

being

has

the

velocity data

and

error minimization here

slant-wire

probes

in order

of the

hot-wires

velocity

relation

Before Method

continuing,

A that

present

skew

angles,

interference

effects

from

associated

cannot ever,

be the

with

angular by suitable

probe

configurations, with

Me_hod

Method

with

the the

probe

sweep

of this

design along A are

this

with

illustrated

A will be discussed

body

as the

can

be rotated

interference

effects

can be minimized probe

an estimate in Fig. response

itself.

of their

C.1.

The

equations

and

is required.

regions impact are

with

B, for zerolittle

region

or no of rota-

probe

supports

be expected; the

Two

The

associated

the

can

and

by an effective

Method

from

hot-wire

variables.

360 ° with

data

least-

normal

is a significant

no valid

hot-wire

using

procedure

with

be

to the

The

flow

a restriction

there

region,

region

of the

note

can

compensated

and

is governed

calibration

B. Whereas

supports,

,4 where

Within

to

is usually

to the

yaw large

of sensitivity

flow variables.

yaw flow angles

and

is necessary.

use of custom-made

a simple

Method

probe

Method

neglected.

shifted

and only

mean

when probes

positions

sensitivity

it is important

is not

for the

pitch

available

This

of

a hot-

the

prohibitive

range

field

axis

is not based

recalibration

the

sequential

to maximize to pitch

frequent

of rotational

to solve

on the

so that

to-moderate tion

relies

to deduce

technique

on extensive

commercially

components.

number

criteria

proposed

cooling

stress

probe

[61] proposed

is used

becomes

here

stress

to the

Their

limits

is proposed

Reynolds

Kool

relies

and that

probe

at a large

technique

rather

accumulated slant-wire

and

slant-wire

but

and

B)

development

[46].

which

normal

for of the

(Method

of the

[45] and

flow

conditions.

for calibration

advantage

a single

mean

rotatable

form

technique

is nominally

similar

equations

by A1-Beirutty

De Grande

law relation,

time

mean

response

flow

field under

cooling

for by obtaining squares

of the

primary

response The working

For details

to references

hot-wire

a single

of data technique

rotatable

in which

The

Their

is referred

is not unique.

on an empirical

developed

derivation.

in itself,

turbulence

calibration.

equations

measurement the

hot-wire

in this appendix.

without

reader

A single

is applicable rotation.

chapter

the

Method

and turbulence

are presented

turbulence

are presented

use

C

angular

possible

howposition

slant-wire

of interference of this

developed.

restriction

for on

146

C.2

Generalized

Empirical

The

starting

for the

equations

point

for both

Method

Cooling derivation

A and

is given

Relation mean

and

turbulence

B is the empirical

cooling

response

law relation.

on whether a linearized or non-linearized For a non-linearized system, the relation

by: (ENL)

where

of the

Method

The form of this equation is dependent hot-wire anemometer system is used.

Law

E is the instantaneous

2

E20 -- BNLUe

bridge

voltage

(C.l.a)

n

and [7, is the instantaneous

velocity; E0_ and B are the intercept and slope of the wire the calibration stream is normal to the wire. Conversely, output

is linearized,

the cooling

relation EL

is given

calibration when the

cooling

curve when anemometer

by:

= BLUe

(C.l.b)

These equations can be presented in a general form applicable to both linear]zed and

non-linearized

signals

as: (E 2 - A) 2/''

where,

for a linearized

Eo) 2,

following

oped

for the

extracted C.3

The

mean

Method

velocity

A, this

to Positions

to Positions there is the from

one

discuss transverse

mean case,

vector into

and

can

how

the

probe

flow angle

and

from

-

BNL,

turbulence which

the

can be evaluated

four

angular

rn -- n

response

equations

linearized

system

2, and

supports situation in the

x-z

by

in magnitude

positions,

such

mean

bridge

corresponding

be accomplished

1 and

this

B

are

devel-

equations

are

Development

3 and 4. Without risk that the wire

of the

later

the

Equation

C.2,

m = 2

case.

is rotated

in Fig.

B = B2L,

A = E_,

non-linearized

Flow

a hot-wire shown

analysis,

as a special

Mean

A = O,

system: E2NL,

E 2 =

axis

-

for a non-linearized

In the

(C.l.c)

system:

E 2 = (EL and

= B21mUe2

by

rotating

rotating

will occur,

wire

rendering

7, defined

direction

the

if beforehand, as shown

when

1,2,3

and

is recorded. wire

about

mean velocity into a position

can be avoided, plane,

voltage

an inclined

a normal

knowing the will be rotated

and

as Positions

about the

4 For

the

same

yaxis

direction, however, where interference

data

invalid. an estimate

in Fig.

We

will

of the

C.2, is known.

147 The reader

should

a restriction For

inherent

Method

by rotating to-moderate present

this

in mind in the

B, the

with

that

this is strictly

response

four

wire

a single inclined skewness levels,

itself

direction direction

for

keep

equations positions

wire about the problem

Method

B,

and,

section,

mean

flow response

will be presented. Except both of these methods. When position

the shown

Combining

where

therefore,

no

that U.

time-averaged in Fig.

equations

C.2,

form the

not

in Fig.

C.2

can

be achieved

the

flow

following

of equation four

of the

in the

A and

discussion

(C.l.a)

flow x-

B is in the expressions In the remainder of

Method

Method

B

is applicable

is applied

response

mean

is nominally

Method (C.l.a).

for both

the

following

knowledge

equations

at each

to

wire

result:

(.E,_- ._)_/,,, = BW"(U2), (E_2_ ._)2/,,, = B_/,,,(U2)_

(C.2.a)

(E_ _ ._)2/,,,= B2/,,,(U_)_ (E, 2_ _02)2/m= B2/",(U.2),

(c.2.c)

(C.2.a)

(,_12

equations

noted,

and

the x-axis. Unlike Method A, for zeroof probe support interference does not

difference between Method A and cooling velocity term in equation

the

limitation

themselves. shown

is required apriori, other than and that V and W are less than

The primary the effective

a physical

&: (C.2.b)

-- F__2o)2/m/(E22

and

equations

-- Eo2)2/m

=

(C.2.b)

(C.2.d)

(C.2.c)

& (C.2.d)

yields: (C.3.a)

-(Ue2)l/(Ue2)2

(z__- _,_, )_/" /( E,_- ._)_/" = (uJ)_/(uJ),, Rearranging

where,

equations

for convenience,

By eliminating (C.4.a)

(C.3.a)

and

(C.4.b),

the

and

(C.3.b)

gives:

(uJ)_-s,_(u,2)2= 0

(C.4.a)

(U._)3 - _s4(U,2)4 = 0

(C.4.b)

following

definitions

are employed:

s,2 = (E,2- _0_)_/m/(E2 _- _o_)_/_

(C.5.a)

_34= (E3_ - _,_,)_/"/(E,2- _,)_/"_

(c.s.b)

the dependence we have

placed

on the wire the

calibration

restriction

that

slope the

same

(B)

in equations

hot-wire

must

148 be

used

in Positions

require, same

however, as the

intercept

one used

values

In order ponents, mean

the

used

equations cooling

components

in terms

the

(C.5.a)

(C.4.a)

(U, V, W),

and and

(C.4.b)

the

E34,

expressed

do

not

2 be

the

calibration

terms

velocity

(Ei_2, Es4)

of the must

be

(e 2) voltages.

(C.4.a)

and

by substituting

(C.4.b)

E = E + e into

in a binomial

com-

in terms

mean-square

in equations

expression

We

appropriate

for the mean be

voltage

and

4. 1 and

(C.5.b):

must

(E)

voltages

resulting

3 and

in Positions

and

(U,)

mean

E12 and

in Positions

is used

4, so long as the

velocity

of measurable

E02)2/', expanding at second order:

that

3 and

of measurable terms,

likewise,

in equations

to solve

voltage

expressed

hot-wire

in Positions

are

in terms

•The

2, and,

the

effective

velocity

expressed

1 and that

series,

e/_

and

can

be

(E 2 -

truncating

Ce2 + _2__o2

(E'-F_,g)2/'=(.E2-F_,g)2/'[l+(41m)__,2__g

]

(C.6.a)

where: C=2/m+(4/m)( Time-averaging

this

expression

-

(C.6.b)

E_ _ _02 )

1)(_2

yields: Ce 2

(E=- gg)_/-_= ($_ - &_)_/=[1+ _ _ Eo _] Substituting

equation

(C.6.c)

into

equations

(C.5.a)

and

(C.6.c)

(C.5.b)

gives:

r_12 = r_i/_2

(C.7.a)

(c.7.b) where: el

el

2

:_, = (&__ &_)_/,-[_+ _ - _1 C2

c2 2

:_ = (_ - ,&_)_/,-[_ + E_- _1

(C.7.c)

v

I

03e32

_:_= (_ - &_)_/,,,[_+ ._ - E_] 04e42

S4 = ('E'_ -

The (C.4.a)

exact and

expression

(C.4.b)

-Eo2)2/"[ 1+

for the effective

can be expressed

cooling

in terms

_42 _ _021

velocity of the

mean

term,

Ue 2, in equations

velocity

components

149 (U, V, shown

W)

and

in Fig.

Method

the

A and

[Method

Reynolds

C.2.

stresses

as follows

Method

(u-7_,

(recalling

i,j

that

= 1,2,3)

different

at the four expressions

wire

positions

are required

for

B):

A I (see

Appendix

D for derivation) w

__

V2

(U,2), =Ko,iU2{K,,,[I

+

_= [(v)' + -O-g] _-{]+

(C.8.a)

+ K_,,sN[(-ff) + _1 +

v2 (-U-j2)2 =KoaU2{KI,2[1

+ _-g] + [(_)

+ gzah_g[(W)

. ('Ue2)3

2 + U2

U)+

21

=Ko,sU

+ _--_]

_-_

{_(KI,a

+ h2ngKs,s)[1

+ -_1

_ + 1_(K,,a + h2BNKa,a)[("_ + [( V )2 + _-'_l -

(K,,3

_-_1}

2 W - hBNK3,z)[("-_-)

W)2

w2 + -U-'_]

(C.S.c)

+ _-'_1}

m

1/2

(U¢_),

=Ko,4U2{_(K1,4

+ h2BNK3,4)[1

+ -_]

"_ + 1_( K .,, + h2BNKa,4)t(W) + [(v)= + _1 + (K,,4 "

]Method

B[ (see

(Ue2)I

.

=

-- hBNI{.3,I)[(-'O-

Appendix

=I'(o,IU2{[1

W

) Jr

A, reference

_

_- _--_]-_ KI,I[(

+K3,,h2BN[(--_

IV)2

2 + _1 w2

(C.S.d)

_-_

_---_']}

[47] for derivation)

U)2 + _-_] v_

w2 +_--_1+2K2,,[(

V

(C.9.a) _'_ )+_-_]}

150

132

t,2 + K1,2[(V)2+ -U-_] (U+2)2=go,2U2{[1+ -_1

(c.9.b)

2 w 2 + _-_]_ + K3,2hBN[(_-) 2K2,2[(u) + _]} v2 m

m

u2 + (U,2)3 =g0,3U 2{[1 + _-_]

V 2 + _._ 132]

K3,3h_N[(-_)

+ K1,_[(w)2+ ,1,2] u2 + 2K2,3[(W) (U,2)4

u2 re" h 2 r,'V32 2 {[1 + _-'_] + ,,3,4 BNtt _':

=Ko,,tU

(C.9.c)

+ _-_] }

v2 + _"_] (C.9.d)

W)2

+ K1,4[(-_where

K0, K1, K2

tions the

of the tangential

cient.

The

equations

and//'3

wire

cooling functional

(C.8.a)

stituted

into

equations,

(the

inclination

second

angle

through

(C.8.d)

I(ov + Klv(V/U)

and

hnN

variables (C.4.b),

Method

wire

cooling is the

are

are

through

func-

(h_¢)

cooling

in Appendix

following

Method

position) coefficient

binormal

given

(C.9.a) the

A and

+ _']}

denotes

normal

or equations and

to both

_-)

the

(k);

for these

(C.4.a)

2K2,,[(W

subscript

(a),

coefficient forms

equations

applicable

_

+ _--_]-

and

coeffi-

D. When

(C.9.d)

non-linear

are

sub-

algebraic

B, result:

+ K2v(V/U) 2 + K3,,(W/U) 2 + Cv = o

(C.10.a)

Ko_ + K,_(W/U) + K_(W/U) 2+ K_(V/U) _+ C_ = 0

(C.10.b)

where u2

_

v 2

w 2

(C.11.a)

c_ = Ko_(_--_) + g,,,(_-_) + K2,,(_-_) + K3,,(_--_) m

.

U2 .

m

.

U_

c= = I;0w(_--_) + x;,=(_-_) + g2=(_-_) and

where,

for convenience,

the

w

W 2

following

l) 2

(C.ll.b)

+ g3w(_-_)

definitions

have

been

employed:

_ethod AJ Kot, = Ko,l Kl,1

-

Ko,2 K1,2 E12

Klo =2K0,1K2,1 + 2K0,2K2,2E12 h'2v =K0,_ - K0,2E12 K3_ =hBN(Iio,lI;3,1

-- Ko,2K3,2E12)

(C.12.a)

151 and, I

r

p"

2

Kow=(_)[Ko,3(_,_ + hBNN3,3) 2

- Ko,_(KI,, + hBNK3,41:34] (C.12.b) + Ko,4(K1,4 K_w

--

hBNK3,4_34 2

]

=Ko,,,

I':3,,,=Koa - Ko,, _.34 [Method

B]

=Ko,a - Ko,2_12 Klv =2Ko,1K_,l + 2KoaK2,2E]2 K2. = Ko,1KI,a - Ko,_K1,2E]2 2 ," K3,, =hBN(I_o,IK3,1 -- Ko,2K3,2Ea2) KOt)

(C.13.a)

and, --/(0,3

_w _w

terms

which

Cv and

would

to mean

result

which

nents

(V/U)

(C.14.a) first-order

+ KI,,,(W/U)

(C. 14.a)

tions

can and

and

and (W/U).

accurate

are may

and

fluctuating

the

terms

without

introducing

represent

two coupled for the

be obtained

solutions directly

(v/U)o = -KodK_. (tWU)0 = -X;ow/K_.,

mean

velocity and

Cw

serious

Under

these

(C.14.a)

non-linear

for (V/U) as:

(W/U).

Cv and

2 = 0

relative

mean

relative

2 = 0

2 + K3w(V/U)

second-order

neglected,

and

errors

velocity.

correction

2 + K3,,(W/U)

simultaneously

for the

components

cooling

ratios (V/U) reduce to:

+ K2w(W/U)

account

velocity

effective

(< 10%),

+ K_,,(V/U)

_34)

(C.10.b)

can be neglected

velocity equations

When

]x_0,4K3,4

for the

field

(C.10.b)

(C.14.b)

be solved

(C.14.b)

the

expression

intensity and

Ko,, + K1,,(V/U)

Equations

(C.10.a)

neglecting

errors in the computed mean conditions, the mean response

Ko,,

-/(0,4K1,4_34

-- h 2BN( Ko,3 K3,3 -

in the

(C.10.a)

(C.13.a)

= Ko,3 K1,3

from

In a low turbulence

Y]34 q- 2Ko,4K2,4_34

Cw in equations

components

in equations

Ko,4

--21(0,3I(2,3

_w The

--

(C.14.b) algebraic velocity

terms (I¥/U)

equacompo-

in equations which

are

152 which

are appropriate

For the more non-linear

case,

algebraic

"_-i/U2,

equations

at least

which, become

(C.14.a)

and

unknowns of the

solved

for the

values

the

mean

(U)

at

of the

each

effective

can

until

ratios

positions

_, to

the coefficient difficulty, ratios

ma-

equations (V/U)

and

the

1 is obtained,

Reynolds stress tensor. The calculated mean velocity components by simulta-

the

The

Reynolds

improved

stress

and

(W/U)

by

averaging

its

calculated

in Fig.

C.2.

When

after

1 (equation equation

some

tensor.

are known,

shown

following

mean This

velocity iteration

is achieved.

(V/U)

at Position

(C.2.a),

at Position

then this

2, v2/V

are required normal-wire

component

(C.10.b).

convergence

velocity

velocity

and

to update

be determined

four

cooling

equation

(C.10.a) used

is repeated

velocity

into

equations

are subsequently

Once

arbitrarily,

two

initial estimates are made using equations (C.15.a) and (C.15.b). results are then used to solve the turbulence response equations

solving

proce.dure

ui/U

and/or

flows.

represent

W/U,

slant

mean

skewed

(C.10.b) equations

to overcome

(to be developed shortly) for the complete stresses can then be used to correct the neously

(V/U,

rotation In order

in slightly and

independent

If this is done

are first

(W/U)

(C.10.a)

five more

requires

ill-conditioned.

(C.14.b)

(W/U) after The calculated

in seven

positions.

and

equations

Therefore,

in turn,

five different

may

(V/U)

however,

_-6/U 2, _'C/U2).

for closure, trix

for evaluating

general

(C.8.a) for the

the axial

value the

or (C.9.a)) axial

with

expression

velocity

mean

the

wire

for

the

is substituted with

the

wire

rearrangement:

[- t odAI u, ={(g2 _ _,o_)2/m [1+ C,F,_I(_,[- g2,)]I(Bo)_/'}'/_ {K,,,B + _] + V + 2,r_., 1 [(_.)

[Method

v_._ _ ]

_ [(-_--)_ + g_,,h _N

+ w'_

(c.16)

_"-_11_1/2 + _.,

B1

u, ={(&_ - _7)_/-'[a+ c,7,_/(_,_,- t,2,)]/(B