Aerodynamic field prediction by a streamline method in the ejector

NINTH EUROPEAN ROTORCRAFT FORUM

Paper No 85

AERODYNAMIC FJELD PREDICTION BY A STREAMLINE METHOD IN THE EJECTOR SYSTEM OF A

PASSIVE INFRARED SUPPRESSOR

P. SCRIPELLJTJ, P. MANNELLA

!.A.M. R.PIAGG!O Viale Rinaldo Piaggio Finale Ligure (SV), ITALY

and M. TROILO

Istituto di Macchine Universita di Genova, ITALY

September 13-15, 1983

STRESA, ITALY

ASSOCIAZ!ONE JNDUSTRJE AEROSPAZIALJ

,QQYNAMIC FIELD PREDICTION BY A STREAMLINE METHOD IN THE EJECTOR SYSTEM OF

1\

PASSIVE INFRARED SUPPRESSOR

P. SCR! PELL! TI, P.

Ml\

NNELLA

!.A.M. R.P!AGGIO

Viale Rinaldo Piaggio

Finale Ligure (SV), ITALY

and

M.

TRO!LO

Istituto di Macchine

Uni versitll di Genova, ITALY

Summary

Reported herein are some of

theoreti-cal and experimental

activities held

by

PIAGGIO

in the aerodynamic field

prediction

inside an

ejector system

which his !R suppressor concept is

ba-sed on.

This

report summarizes the study on

both the analytical model and the

com-puter code development joined to some

correlations to

experimental

results

on

typical

ejector systems

represen-ting a scaled down !R suppressor

desi-gned for a defined gas turbine engine.

A previous paper was written showing

PIAGGIO activity to simulate the

aero-thermodynamic phenomena through

mathe-matical models. This paper now is but

referring to a specific area of

simu-lation that is the complex aerodynamic

field inside an

!R

suppressor where

two flows are mixing together.

The diffusion process of

thermodynami-cal and mechanithermodynami-cal quantities has been

simulated

by

a method that can be

fit-ted on the classical streamline

sche-mes and is based on influence matrices

_under well defined boundary canditions.

This method appears rather flexible to

describe a wide class of ejector

sy-stems.

A wide

trade-off

both theoretical

study

is shown for

and experimental re-sults and also some information infer-red from 1iterature are accounted for.

The

experimental

results

have

been

used to test the mathematical model

and

satisfactory

agreement

has

been

obtai ned.

1. INTRODUCTION

The infrared technology of heat

see-king missile requires more efficient

IR signature suppression for

helicop-ters. The improvements on heat seeking

detectors have mandated the require-ment for an efficient hot metal and

plume radiation suppression. Lower

helicopter target signatures are re-quired to enable probable escape from

detection when operating in hostile zones.

Other devices addressing the radiation

sources such flare dispensers and jam-mers are in general unactractively'

large and heavy so they adversely im-pact the he 1 i capt e r m i s s i an perform an-ce.

The importance of a defense against weapons locking on the hot end of an

engine has been recognized and an

en-gagement to develop suitable technolo-gy and hardware has been made.

PIAGGIO has conceived and has under development an IR suppressor for· a gas turbine helicopter engine in house ma-nufactured.

A dedicated effort is made to improve the efficiency of the IR suppressor addressing both the engine hot metal and the exhaust plume radiant

emis-sions that are the most important

ra-diant sources in the helicopter. (see Fig. 1-1)

Studies predicted that these emissions can be satisfactorily lowered by dilut-ing the exhaust gas and film cooldilut-ing the hot metal walls.

This requires a device able t9 pump ambient air to achieve these tasks but

inducing low power penalties to the engine. Reduced weight and size are also important constraints. The shape itself must be able to hide the

hot-test parts of the engine exhaust

nozz-les from an afterend lock-on.

The concept envisaged by PIAGGIO is a tai 1 pipe two-stage ejector to dilute the hot exhaust plume and to cool the hot visible walls. No moving parts are foreseen and the characteristics of

mechanical integrity are enhanced. Lightness, small size and a self con-tained philosophy are joined to

featu-res of mechanical simplicity.

(See fig.l-2)

To deeply analyse the ejector aero--thermo performance many analytical models have been studied and

transla-ted in some computer programmes.

This software has. been proven to. be an excellent tool to desi!!n ejectors or to verify experimental results on tes.t models.

It is evident that a well defined me-thod of design will involve the know-ledge of the two main parameters; the signature level and the allowable en-gine power penalty which have to be achieved.

Trade-off studies on both ejector and

engine performance are an easy matter

only by a computerized way.

The heart of this design activity are the cross-correlation studies between

the ejector inner aerodynamic field and the heat transfer phenomena in two

fluids mixing together. The optimization of the

parameters is strictly

aerodynamic

function of temperature requirements on ,both plume and walls.

Unfortunately poor data are available from literature on this subject so PIAGGIO developed an aerodynamic model based on a streamline method with the aim to predict the aerodynamic field and mixing rate in an ejector system. Sets of data required for heat trans-fer calculation on both plume and walls emissions are produced.

2. ANALYTICAL MODEL

A number of methods exist for the cal-culation of a steady plane flow through a channel of a given shape

with defined boundary conditions.

Not so many reports on the mixing flow due to a primary hot jet entraining

into the channel a secondary stream of cold air, as required for the dilution process to take place, can be found. Even if the problem of the turbulent diffusion of transport properties has been extensively studied from a physi-cal point of view

/1/

and a lot of experimental tests have been made in diff&rent conditions, the need of a general computation code, able to as-sist the IR suppressor ejector design, was felt.

Instead of developing an analytical model accounting for the flow features in a rigorous physical way, it was preferred to choose a more flexible model based on a suitable simulation of the diffussion process, whose agre-ement to the actual flow could be ob-tained by calibration with experimen-tal results, derived both from the li-terature and from tests performed at Pli<GGIO facilities.

The basic idea was to fit the well--known streamline curvature computing procedure with a mixing model based on the cross-influence of adjacent streamlines with regard to some tran-sport properties.

The streamline curvature method for the calculation of a steady plane or axisymmetric inviscid flow is well--known, and only the main equations to be solved are reported here for con-venience of introducing the diffusion mode 1.

As far as the boundary layer i s con-cerned, the classical methods of cal-culation do not seem applicable to the case of a louvered w a 11 • Waiting for an extensive be performed the boundary been included cal model. experimental enquiry to in P!AGGJO facilities, layer effects have not in the present

analyti-Considering first a plane curved chan-nel, the Euler equation is written along directions that are quasi-ortho-gonal (q.o.) to the bulk flow (see fig. 2-1).

The final form of the equation here

used is;

e [de)=

.£Q.§..Q[J

de\ tg (a+a)

+

\d1p

s

Qel

\dXf1p

The continuity equation to be satisfiec is;

(eos<J eosa-sin<J sina)Qldq.(2;

1 t is channel worthwhile width could to note that be different thE f 0' different q.o., so allowing, within ar acceptable approximation, the treat· ment of a tapered channel with a plane model. Referring to an (fig.

2-2),

with no ty components, the are; axisymmetric case tangential veloci· two main equation5

efde~

=

eos<J [Jdc\ tg(a+<J) +

\d1pfs

2CQrn

\dx71/!

m~

JR

2c

0osa-sin<Jtg'

gr

dr.

( 4 ) 0

In this case the symmetry condition applied for r=O gives:

so overcoming the discontinuity other-wise appearing in eq. (3).

When the inlet flow exhibits large non-uniformities, a mixing process

ta-es place, with the main ,f a spreading across the .he flow properties*

consequence channel of

"he mixing process as basic phenomenon s due to some turbulence level, which s responsible of a lateral transport ,f physical quantities. Within a tur-•ulence characteristic length, the 'low cannot be considered (and it is 10t indeed) steady, so the above

equa-ions are not longer valid, neither treamlines exist i~ a true sense.

leverthe1ess, 1imiting the detail le-•el of the analysis to a lengtl\ scale 1reater than the characteristic mixing engths, a mixing flow can be yet tudied as a steady one, providing an lgorithm able to give the istribution of total pressure and nthalpy determined by the mixing recess itself.

his algorithm should match the follo-ing requirements;

to give for each q.o. inside the channel, the distribution of;

Pt('lp),ht('lp). which the partial derivatives

in (1) or (3) are depending on; to be camp at i b 1 e with the energy conservation law and with an assu-med total pressure loss along the flow;

to be enough flexible to be cali-brated with reference to experimen-tal ·data and to fit within the ge-neral calculation framework.

nc~ the total pressure and enthalpy istributions have been defined the ntropy can easilybe obtained from:

( 6)

or a general quantity g the distribu-ion at the k-th q.o. is obtained from he previous one applying an influence atrix laij I whose terms denote the

influence at the i-th streamline the value at the j-th streamline

of

(7)

The normalizing factor N is choosen as follows:

a) as far as ht is concerned, the

b)

energy conservation gives

N

by:

( 8)

as far as Pt is concerned, once ( 6) has been integrated, the entro-PY increment is checked against the total pressure loss i>PtiPt'

i

m im

s

d'lp -

s

d'liJ

=

m

tR- __

L>p

t •

k k-1 Pt

0 0

( 9)

For each q. 0. the influence matrix

I

a i j

I

is evaluated taking into ac-count the distance between the i-th and j-th streamline and the distance between the previous q.o.

3. COMPUTER COOE

The previous algorithm has been coded in a suitable computer program.

Besides the channel geometry, the in-puts are:

The Ute

is

the primary and secondary mass flow rate;

the static pressure in the mixing inlet plane, assumed uniform as cu-stomary;

the temperatures of the primary and secondary flow;

the total pressure loss distribu-tion.

result whole not yet

is the flow description in channel. Because the code implemented with the

ana-lysis of recirculation zones their ap-pearance is simply detected and moni-tored.

Among the results obtained, particular interest for the IR suppressor perfor-mance prediction have the following:

the temperature distributions along the channel walls and across the exhaust flow;

the static pressure distribution along the walls.

These results permit th~ evaluation of: the plume temperature level and the effectiven·ess of the desired

dilu-tion;

the wall suction capability of the ejector, for the calculation of the film cooling performance.

lhe influence of film cooling flow can be considered by mixing it with the bulk flow, so iterations are necessary ·to take into account the wall pressure variations due to the increased mass flow rate along the channel.

Moreover the computer code is based on specialized routines that perform:

the solution of the main equations, with the streamline curvature me-thod;

the simulation of the mixing pro-cesses;

the calculation of the gas

proper-ties.

tomputing time on IBM 4341 computer is of the order of a few seconds per ite-ration in the streamline curvature

procedure, while the total time

de-pends on the desired accuracy and on the channel geometry. In the worst

ca-se of highly curved channels, a fifty iterations run is necessary to meet relati~e error bel~w 0.001. a A good performance

been obtained by

damping factors as

4. TEST CASES

in this respect has the use of dynamic suggested in /2/.

The computer code has been tested on a wide number of cases for calibration

purposes. Three different situations are shown here.

a) Planar free jet

For this case the experimental re-ference is the rere-ference /3/.

The results are shown in fig. 4-1 where the axial velocity decay is plotted along the jet axis. In this case the model appears to be sensi-tive to the axial spacing_ of the calculation stations, and moreover an exact modelling of the actual experiment.al situation becomes im-possible because an external chan-nel wall is necessary even if pla-ced far apart.·

b) Axisymmetric jet

Reference made to the experimental values reported in /4/, /5/, both the center 1 ine velocity and tempe-rature decay has been calculated in two different cases. Fig. 4-2 shows the theoretical results compared with those reported in /4/, and fig. 4-3 with those reported in

/5/. The calculated decay of tempe-rature results steeper than that of the velocity in agreement with the referenced experimental data; how-ever the axial position of the cal-culated curves should have been di-splaced somehow downstream (tempe-rature) or upstream (velocity). This fact follows from the known difficulty to treat with the same model the potential core region and the mixing region.

Indeed the present model can do it with a proper different spacing of the q.o. in the two zones.

c) Curved ejector

The flow field obtained in this ca-se has been ·compared with the rimental results obtained at GIO facilities on suppressor led-down models.

ex

pe-p lAG-sc a-In the case of fi_g. 4-4 a ram pres-sure has been simulated in the ex-perimental faci 1 ity, and the vela~ city ratio between the secondary

and primary flow Cs/Cp at the inlet was around

o.s,

that ensures a quite stable operation of the ejector.

culated

The agreement between cal-and measured values is

fa-irly good and the suction capabili-ty of the system is enhanced.

In the case of fig. 4-5, the actual (test) velocity ratio Cs/Cp is around O.l7.The same figure in the analytical model creates unstable operation as indicated by the ap-pearance of recirculation zones. Additional calculations have been performed modifying the inlet area ratio in order to obtain a velocity ratio of about

o.s.

This means to suppose the mixing to start some q.o. dewstream.

The agreement with the experimental results is less satisfactory than the planar and axisymmetric cases. Fig. 4-6 and fig. 4-7 show the suc-tion pressure trends along the press ure surface both in the same boun-dary conditions as before: ram ef-fect and static.

Also on this side of the channel, the ram effect seems to offer a mo-re stable operation as mo-regards the suction effect whereas a static condition appears less favourable to stabilize the pressure field along the wall. The experimental results showed indeed some tendency of the primary hot jet to impinge the wall at the middle of the chan-nel.

In fig. 4-8 and fig. 4-9 the

tempe-rature distributions on the

stream-lines nearest the walls of the cur-ved ejector are shown for the

pre-vious two conditions: ram and

sta-tic. The good agreement between the simulation and the test is evident. However, other PIAGGIO computer programmes are devoted to the task of predicting the wall" temperatures Nith more accuracy.

In fig. 4-10 a typical streamline oattern is shown as plotted by the :omputer code.

:ONCLUDING REMARKS

~ral comments can be drawn from the >arisen activity that has been per-ned and reported above for a few

ca-~he model is quite flexible, and can >e adapted to different

experimen-tal situations;

b) as far as curved ejectors are con-cerned, additional experimental da-ta are required for a more accurate calibration;

c) the analytical model is satisfacto-rily reliable and is a good sup-port for the analysis of the com-plex inner aerodynamic field of a multistage ejector system.

6. FUTURE DEVELOPMENTS

PlAGGlO activity in the field of IR suppressors is oriented towards a dif-ferentiation of basic 1 ayouts and ar-rangements, based on the specific ap-plications. As a consequence, the code wi 11 be improved to d 1

e:

above computer properly han-a) general mixing for different loss correlations, geometrical shape of ejectors;

b) the recirculation zones, that could exist in the flow rate ranges typi-cal of lR suppresssor ejectors; c) the film cooling properties of

va-rious louvered walls, including their heat transfer coefficients. d) the impingement effect of the

primary hot jet onto the pressure surface of curved ducts.

7. ACKNOWLEDGEMENTS

The writers wish to thank Mr. Massone and Mr. Mordeglia of the aero-thermo and structural R & D team whose con-tribution made possible to perform the work herein reported.

8. REFERENCES

111 J.A. Schetz: 'Injection and Mixing in Turbulent Flow'.

Progress in Astronautics and

Aero-nautics. Vol. 68, Martin

Summer-field Ed.

/2/ M. Troilo: "Meridional Through--flow Calculation•.

VKI lecture Series No.l983/06, May 1983.

/31

A.S. Weinstein, J.F. Osterle,

W.

Forstall: "Momentum Diffusion from

a

Slot Jet into

a

Moving

Seconda-ry".

Journal of Applied Mechanics, Sept. 1956, pp. 437-443.

/4/ W. Forstall, A.H. Shapiro; "MOmen-tum and Mass Transfer in· Coaxial Gas Jets•.

Journal of Applied Mechanics, Dec. 1950, pp.399-406.

/51

F. Landis, A .H. Shapiro: "The Tur-bulent Mixing of Coaxial Gas Jets". Heat Transfer and Fluid Mechanics lnst., Preprint and Papers, Stan-ford University Press, Stanford,

Calif., 1951.

g. LIST

OF

SYMBOLS Symbols

a element of the influence matrix c flow velocity

D diameter of primary jet nozzle g general quantity

h enthalpy channel width m mass flow rat~

N normaliling factor

p pressure

q coordinate along q.o.'s r radial coordinate R radius

~

gas constant

s

entropy T absolute temperature x x-axis coordinate y y-axis coordinate

a

q.o. t i l t angle from y direction

coordinate normal to q.o.•s

Q density

a

streamline slope

~ stream function

Subscripts c centerline

i i-th row of a matrix. j j-th co 1 urn n of a matrix: k k-th q. 0. t t o.t a l p primary ps pressure side s secondary ss suction side

RADIANT HOT EXHAUST PLUME

ENGINE EXHAUST HOT PARTS

SECONDARY HOT METAL

PIPE HOT METAL

FIG.

1-1

TYPICAL RADIANT IR SOURCES

ENGINE ADAPTER

$

EXHAUST GAS CENGINE]

¢-DILUTION FLOW ~MIXED FLOW - W A L L MIXING FLOW FILM COOLED

_/··(

' HELICOPTER COWLING

--·

_

__.--··

BULK FLOW

ST~EAMLINE

GUASI-O~THOGONAL

L

FIG. 2-1 ST~EAMLINE METHOO

BULK FLOW

~

PLANE CHANNEL

FIG. 2-2 ST~EAMLINE METHOD

AXISYMMET~IC CHANNEL 1.0

o.s

0.4 0.2 1

I I I I I I I I

I

I

I

' ...!._

ST~E~ML.INE

METHOO FIEF. C3l:

•

VELOCITY ---EMPIRICAL. RELATION

•

0

•

~

...

"'S...~

_•••

~.

:---Cs/Cp=0.5

l

2 3 4 5 10 20 30 4050

'

....

X/0

FIG. 4-1 PLANAI=l .JET: CENTE~ -LINE

...

I

I

I I II

I

I I

I

I -STREAMLINE METHOO

---IH-+++1

REF'. [4): e VELOCITY

+----+--+ ___

EMPII=!ICAL I=!ELATION

'

1 e 3 4 s 10 eo 30 4050 100 X/C

FIG. 4•2 AXISYMMETRIC .JeT: CeNTER-LINe VALUES OF VeLOCITY

I

I

I I

I I

I I I I

- STI=!EAMLINE METHOO IVELOCITYl

TJ

-·--STI=!EAMLINE METHOO [TEMPERATUR El

REF'. [5l:

•

VELOCITY I

+

TEMPERATURE

l

---EMPIRICAL RELATION 1.0 C:c-

c:.

o.a

Cp-Ca 0.6 Tc -Ta T,.-Ts _o.4

::t:e~+

.

~~

~

·~

~~--\>

C:../Cp:0.46

1\

I'

Ps/PP :1.087

r.

'

'

1 e 3 4 s 10 eo 3o 4o so 100 X/0

FIG. 4·3 AXISYMMETRIC .JeT: CeNTER-LINe

PRESSURE TAPS POSITION

X

- T E S T

---THEORY !STREAMLINE 211

e TEST POINTS AT MEAN

SECTION OF THE CHANNEL

FIG. 4 - 4 TYI='ICAL AEROCYNAMIC FIELC ON SUCTION SURFACE IN SIR MOCELS CRAM EFFECTJ

Ul II :J 100 Ul Ul UJ II

c.

200

u

~

~ 300 IJ]

1-2

UJ

u

II UJ

c.

400-500 I o o o o POSITION I

.

I I I I X - T E S T ---THEORY !STREAMLINE 211

• TEST POINTS AT MEAN

SECTION OF THE CHANNEL

FIG. 4 · 5 TYI='ICAL AEROCYNAMIC FIELC ON SUCTION

w

II :J Ul Ul

w

II

c.

100

u

1-c( l--UI 1-2

w

200

u

II

w

c.

PRESSURE TAPS POSITION

X

- T E S T

---THEORY CSTREAMLINE 1)

• TEST POINTS AT MEAN

SECTION OF THE CHANNEL

FIG. 4 · 6 TYPICAL AERODYNAMIC FIELD ON PRESSURE SURFACE IN SIR MODELS CRAM EFFECT]

w

II :J Ul Ul

w

II

c.

u

100 1-c( l-UI 1-2

w

u

II 200

w

c.

PRESSURE TAPS POSITION

.4

,.---

/

/

/ / / - T E S T X ---THEORY (STREAMLINE 1 l

e TEST POINTS AT MEAN

SECTION OF THE CHANNEL

FIG. 4·7 TYPICAL AERODYNAMIC FIELD ON PRESSURE SURFACE IN SIR MOOELS

w

II :J 1-cl: II

w

a.

~

w

f-z

w

u

0:

w

a.

w

II :J f-cl: II

w

a.

~

w

f-z

w

u

II

w

a.

1SO 100

• ·TEST POINTS } MEAN SECTION

+

TEST POINTS !FlAM EFFECT) OF THE CHANNEL

- T E S T

---THEORY } STREAMLINE 21

-·--THEORY I FlAM EFFECT l

-=.

- - - - -

!,._ - - - :.2'2-• 0

-.~

--+

_._.:;l:.:.m'

-.-+

~+_..---·-­

:...-·-.

...

+

.

TEMPEFIATUFIE SENSORS I'IOSITION

X

FIG. 4•8 TYPICAL TEMPERATURE OISTRIBUTION ON SUCTION SURFACE OF SIR MOOELS

1SO

100

•

+

TEST POINTS } MEAN SECTION

TEST POINTS !FlAM EFFECTl OF THE CHANNEL

- T E S T

---THEORY }

STREAMLINE 1 -·-·-THEORY !FlAM EFFECT l

----·

.

_----·---

_{· - ·}

__...

---·

_{- - - - -}

---

·=;"·-+

- - - -

-==·-

:..-·-- :..-·-- :..-·-- :..-·-- _:._ :..-·--:..-·--_:_ . ..:=:+ ..

...-+

-·-·- ·-·

TEMPEFIATUFIE SENSORS I'IOSITION

X

FIG. 4-S TYPICAL TEMPERATURE OISTRISUTION ON PRESSURE SURFACE OF SIR MOOELS

STREAMLINE

STREAMLINE 1

FIG. 4•10 TYPICAL STREAMLINE PATTERN

IN SIR MOOELS