Vous êtes sur la page 1sur 202

L.

t>

P187335

Nl V/
AGARD-AG-163

<
I I

< <

AGARDograph No 163
on

Supersonic Ejectors
Edited by J.J.Ginoux

DISTRIBUTION A N D AVAILABILITY O N BACK COVER

AGARD-AG-163

NORTH ATLANTIC TREATY ORGANIZATION ADVISORY GROUP FOR AEROSPACE RESEARCH AND DEVELOPMENT (ORGANISATION DU TRAITE DE L'ATLANTIQUE NORD)

AGARDograph No. 163 SUPERSONIC EJECTORS Edited by J.J.Ginoux Von Karman Institute for Fluid Dynamics 72 Chaussee de Waterloo Rhode-St-Genese Belgium

The material in this book is an updated version of the Lectures given during Short Courses organized at the Von Karman Institute in April 1968 and March 1969.

THE MISSION OF AGARD

The mission of AGARD is to bring together the leading personalities of the NATO nations in the fields of science and technology relating to aerospace for the following purposes: - Exchanging of scientific and technical information; - Continuously stimulating advances in the aerospace sciences relevant to strengthening the common defence posture; - Improving the co-operation among member nations in aerospace research and development; - Providing scientific and technical advice and assistance to the North Atlantic Military Committee in the field of aerospace research and development; - Rendering scientific and technical assistance, as requested, to other NATO bodies and to member nations in connection with research and development problems in the aerospace field. - Providing assistance to member nations for the purpose of increasing their scientific and technical potential; - Recommending effective ways for the member nations to use their research and development capabilities for the common benefit of the NATO community. The highest authority within AGARD is the National Delegates Board consisting of officially appointed senior representatives from each Member Nation. The mission of AGARD is carried out through the Panels which are composed for experts appointed by the National Delegates, the Consultant and Exchange Program and the Aerospace Applications Studies Program. The results of AGARD work are reported to the Member Nations and the NATO Authorities through the AGARD series of publications of which this is one. Participation in AGARD activities is by invitation only and is normally limited to citizens of the NATO nations.

Published November 1972

629.7.047.2:533.6.011.5

$ frinted by Technical Editing and Reproduction Ltd Harford House, 7 - 9 Charlotte St. London. W1P 1HD

PREFACE Two Short Courses were organized on "Ejectors" at the von Karman Institute for Fluid Dynamics in April 1968 and March 1969, respectively, in which 72 scientists and engineers from seven NATO nations participated. They were partly supported by AGARD and directed by Dr H.Uebelhack, Assistant Professor at VKI, now with Domier System in Friedrichshafen, Germany. The objective of these Short Courses was to present a state-of-the-art review of the significant progress which has been made in the past few years in the design of high performance ejectors. AGARD has felt it appropriate to publish the Short Courses as an AGARDograph in an updated version, with emphasis on supersonic ejectors. A small number of copies were originally printed at the time of these courses, but numerous subsequent requests for these notes were received and the demand could not be satisfied. The two first parts, by Dr H.Uebelhack, cover the classical one-dimensional inviscid analysis and design methods for ejector systems with second throat diffusers. In the third part, Professor A.L.Addy presents an ejector flow model, developed at the University of Illinois, which significantly departs from the one-dimensional analyses. Mr D.Taylor, Assistant Manager, Facility Support Branch, Engine Test Facility of ARO, Inc., discusses in the fourth part the ejector design for a variety of applications. Finally, Dr C.E.Peters, Research Engineer at the Engine Test Facility of ARO, Inc. analyses ducted mixing and burning of co-axial streams. The editor is most grateful to these lecturers who, in spite of a very heavy workload, have kindly agreed to revise and update their original notes. Jean J.Ginoux Professor at VKI and Brussels University. Editor

iii

Lecturers Lecture Series Director Dr H.Uebelhack Dornier System Friedrichshafen Germany. Professor A.L.Addy Mechanical and Industrial Engineering Department University of Illinois at Urbana-Champaign Urbana, Illinois 61801 USA Mr D.Taylor Assistant Manager Facility Support Branch Engine Test Facility ARO, Inc. Tennessee 37389 USA Dr C.Peters Research Engineer Engine Test Facility ARO, Inc. Tennessee 37389 USA

Editor Dr J.J.Ginoux Professor at VKI and Brussels University Head of VKI High Speed Department

CONTENTS

Page PREFACE LECTURERS ONE-DIMENSIONAL INVISCID ANALYSIS OF SUPERSONIC EJECTORS by H.T.Uebelhack ANALYSIS AND DESIGN METHOD FOR EJECTOR SYSTEMS WITH SECOND THROAT DIFFUSERS by H.T.Uebelhack THE ANALYSIS OF SUPERSONIC EJECTOR SYSTEMS by A.LAddy EJECTOR DESIGN FOR A VARIETY OF APPLICATIONS by D.Taylor ANALYSIS OF DUCTED MIXING AND BURNING OF COAXIAL STREAMS by C.E.Peters iii iv

17

31

103

165

ONE-DIMENSIONAL INVISCID ANALYSIS OF SUPERSONIC EJECTORS by H.T.Uebelhack

Lecture Series Director Dornier System Friedrichshafen, Germany

CONTENTS Page NOTATION 1. INVISCID ONE-DIMENSIONAL EJECTOR ANALYSIS 1.1 Conservation Laws 1.2 The Supersonic-Saturated Regime 1.3 The Supersonic Regime 1.4 The Mixed Flow Regime 2. 3. 4. 5. 6. FRICTION DIFFUSERS MIXING CHAMBER LENGTH EFFECT OF TEMPERATURE EJECTOR DESIGN AND OPTIMIZATION 4 5 S 6 6 7 8 8 8 9 9 10 11-16

REFERENCES FIGURES

NOTATION

a A Cf D f,q,z F L
HI

speed of sound cross section friction coefficient diameter dimensionless functions of M thrust mixing chamber length mass flow Mach number pressure gas constant temperature velocity specific heat ratio mass flow rate density

M P R T u 7 P

Subscripts the reference stations are shown in Figure 1 refers to the critical condition (M = 1)

Superscripts a prime (') is used for the primary jet a double prime (") is used for the secondary jet

ONE-DIMENSIONAL INVISCID ANALYSIS OF SUPERSONIC EJECTORS H.T.Uebelhack 1. INVISCID ONE-DIMENSIONAL EJECTOR ANALYSIS

Experiments on ejectors have shown that a wide range of operating conditions of an ejector can be described by fundamental conservation laws. In this analysis the geometry of the ejector is considered known (Fig. 1) and the calculation yields a set of equations for pj,/p4 , p'd/p4 and the mass flow rate n . One distinguishes in general two regimes of ejector operations which are termed by Fabri1 as the supersonic regime and the mixed flow regime. The flow pattern in the ejector for these regimes is shown in Figure 2. The supersonic regime is characterized by a certain part of the flow in the mixing chamber (which is acting at the same time as a diffuser) being supersonic over the whole cross section. In this regime the exit pressure p 4 does not influence the performance characteristics (p'0,Po,i"). This regime can be subdivided into: (i) the supersonic-saturated regime, where Mj is sonic, (ii) the actual supersonic regime, 1 > M2' > 0.3 , (iii) the supersonic regime with low secondary flow 0.3 > M'2' > 0 (or n * 0). In this regime viscous effects cannot be neglected and the present analysis gives poor results when n approaches zero. The limit between (ii) and (iii) which is indicated by M2' ~ 0.3 is rather arbitrary. This value has been found as an average in a series of experiments. In the mixed flow regimes there exists a subsonic region ofthe flow pattern between the secondary flow settling chamber and the exit of the diffuser (mixing chamber). The exit pressure p 4 thus influences the pressure pJJ in the secondary settling chamber. The value of pJJ increases when p4 is increased and vice versa (for p. = constant). Above a certain value of p'2' the flow in the primary nozzle will separate and thus cause the thrust of the nozzle to decrease and influence the whole characteristic performance. This regime, however, is of little practical interest and will not be discussed here. In the following analysis, stagnation enthalpies of the primary and secondary jet are assumed to be equal. 1.1 Conservation Laws Continuity Equations The mass flow in a duct is given by

(1)

11

, aOPo FT* P* . ,w a

- MT7PTM(M*)'

where

is a dimensionless function of the Mach number and the specific heat ratio only (Fig.3). It has been found useful for numerical calculation to introduce dimensionless functions of M# . Momentum Equation The momentum equation can be represented by the balance of jet thrusts in different sections of the ejector.

6 The vacuum thrust is defined by F = pA+pu2A or where ,1/(7-1) +M M h (1 1 f(M*) (1 + T - 1M MJ . '*) = = (i i ) ( - .-1 ^ i) The quantity F can also be expressed in terms of the mass flow m , using Equation (2). A A M . . F = Po T - - V ( M * )
A*

= p 0 A ( l + 7M 2 ) Po

(3)

F =

p 0 Af(M*)

(4)

. f(M*) = p 0 A* = - . q(M*)

(5)

The ratio f(MJ / 2 \!/(7-l)

where

Z(M,) =

M* + TJM

(7)
p

* represents

and

U + \)

.1/(7-1) - 1

Po

Using these relations in the above equation yields F = Po A*Z(M # ) Po


IN,

7+1 F = 27 1.2 The Supersonic-Saturated Regime a^mZ(M 4 )

The performance characteristics in this regime can be determined from the condition of sonic velocity at Station 1 for the primary jet and Station 2 for the secondary jet only. The relation between the mass flow rate <J and the stagnation pressure ratio Po/P 0
is

found from Equation (1).

. = ^ =^m 1.3 The Supersonic Regime The calculation of this regime is based on the following assumptions, first used in Reference 1. (i) (ii) The primary and the secondary jets are isentropically accelerated between Stations 2 and e (Fig.4). The mixing of the two jets is neglected. The secondary jet reaches sonic speed at Station e. A, p 0

(9)

The performance characteristics of the supersonic regime can thus be determined by applying the conservation laws between Stations 2 and e. For the particular case of equal stagnation temperature To
=

TQ

and therefore

a^ = a .

The momentum equation yields (Equation (8)) m'Z 2 + m"Z 2 ' = rii'Z; + m"Z;' . (10)

Introducing the mass flow rate n = m"/m' and the condition M^e = 1 , Zj = 2 , Equation (10) takes the form

/-2 The mass flow rate n can also be expressed by Equation (1), rh" m Solving this equation for
PQ/P 0

p"A'' q(M") p 0 A, q(M*,)

and noting that q(M^,) = 1 yields 1 A"


"
A

Po
Po

i PQ/PJ,

Equations (11) and (13) represent the solution for the supersonic regime. They relate the common variable M^e .

and (i through

In Equation (11) Z 2 is determined by a chosen M^e (0 < M^2 < I). The primary nozzle exit Mach number determines Z'2 . The secondary jet cross section at e is given by Ae' = A2q(M2) (continuity) and thus Ag = A3 Ajl is known. The area ration Ae/A', determines the primary Mach number Ml^e and Ze . From Equation (13) p0'/p0 i s found directly. The whole supersonic regime can be determined by repeating this procedure for all subsonic Ml^2 . In the diagram in Figure 5 both the supersonic and supersonic-saturated regime are represented by straight lines passing through the origin of the two coordinates (p0/p4,Po7p4), i-e. there is no dependence on p 4 . The slope of these straight lines, pjj/pj,, is a function of pi and the actual geometry, A2/A! and A3/A, , only. 1.4 The Mixed Flow Regime Increasing the exit pressure p 4 of the ejector reduces the length of the fully supersonic part of the flow field in the mixing chamber until it breaks down (Fig.2.3). Then the flow in the exit of the primary nozzle and in the center of the mixing chamber is still supersonic whereas the flow near the mixing chamber walls is subsonic. For most cases, the supersonic jet attaches to one side of the wall (Coanda effect). The secondary flow stagnation pressure in this regime is a function of the exit pressure p 4 . The characteristics of this mixed flow regime can be determined by the continuity and momentum equations applied between Stations 2 and 4. The momentum equation (no friction) yields p 0 A' 2 f(M; 2 ) + p X f ( M ; 2 ) = PonA4f(M*4) . The continuity equation reads m' + m" = m4 or p 0 A' 1 q(M; i )(l + p) = p 04 A 4 q(M* 4 ) . (15) (14)

Together with the definition of /u ,

" = S" = r 7 ^ q ( M ^ '

02a)

m p 0 A, the Equations (14) and (15) display the relation between p 0 /p 4 and PQ/P 4 for a certain mass flow rate. Here also M+2 is the parameter relating p'o/p'0 and n . To solve the above equations an arbitrary value for pj, is assumed, p'o then is found from Equation (12a). Equation (14) can be solved for p 04 f(M* 4 ) (introducing the same value of M^j). In the same manner p 04 q(M A4 ) is found from Equation (15). The exit Mach number MA4 is then determined by P 0 4 f(M,-/ 2 \l/(7-DZ(M
p04 q ( M ,

j-(xh)

"'-

The function Z(M#) is double valued. The subsonic solution has to be chosen in this case. The resulting procedure to determine p 0 4 , pJJ/p4 and p 0 /p 4 is straightforward.

8 If this calculation is repeated for various M^2 the whole characteristic line for the mixed flow regime and for a constant ii can be determined (Fig.5).

2.

FRICTION

In the supersonic regimes, friction does not greatly influence the ejector performance since the distance over which the secondary flow is accelerated to supersonic speed is rather short (of the order of two diameters). In the mixed flow regime, however, friction at the mixing chamber walls influences the characteristic line of this regime. If friction is considered in the theoretical calculation. Equation (14) should read F'2 + , F ^ ' - F F = F4 , (17)

where F F can be represented approximately by pL FF = M2 dx Cf7P 2 AD

In many cases it is justifiable to simplify Equation (18) by introducing the exit Mach number M4 . Figure 7 shows the effect of friction on the characteristics of the mixed flow regime. In other words, the pressure losses due to friction reduce p 4 and therefore increase both pressure ratios p 0 / p 4 and P0/P4 The characteristic line of the mixed flow regime is shifted as shown in Figure 7.

3.

DIFFUSERS

In the preceding calculations the effect of the subsonic diffuser was not taken into account. The best possible influence of a subsonic diffuser following Station 4 has already been given in the calculations in Section 1.4, where p 4 and p 0 4 were determined through M^ 4 . In the ideal case, p s would equal p 0 4 (100% diffuser efficiency). In most practical applications M A4 varies in the narrow range between 0.45 and 0.65 only. Assuming a mean exit velocity of M # 4 = 0.55 and a diffuser efficiency of 75%, we would get an improvement in pressure ratios of 15%. This value was found as a mean value in many experiments. The improvement of the mixed flow characteristic is shown in Figure 8 for a p = constant line. The pressure recovery in the mixing chamber can still be increased by using a second throat (Fig. 1). Here, the important parameters are the contraction ratio A3/A3< and the position of the second throat with respect to the primary nozzle exit, X/D 3 . It was found 3 that an additional improvement in pressure recovery of the order of 30% (p 0 /p 4 ,p 0 /p 4 ) can be realized in the transitional regime between mixed flow regime and supersonic regime (Fig.8) by a second throat. The influence of a second throat will be discussed in detail and a calculation method will be presented in the following chapter of this Agardograph.

4.

MIXING CHAMBER LENGTH

The effect of the mixing chamber length on the ejector characteristics is shown in the experimental curve in Figure 9. The influence of friction has already been shown in Figure 7. Using schlieren photography, one can observe that at transition from the supersonic to the mixed flow regime a minimum length of the mixing chamber is required in order to get approximately a uniform flow and the exit pressure p 4 . Figure 10 shows the typical pressure distribution in the mixed flow regime, very close to transition to the supersonic regime. From Figure 10 it can be concluded that for this particular ejector configuration and mass flow rate the optimum mixing chamber length is ( L / D ) o p t = 12 . For larger values of L/D , the exit pressure decreases due to friction. At lower L/D than (L/D) o p t it is not possible to reach ( p 4 ) o p t . The supersonic regime breaks down sooner. This effect is more drastic than the effect of friction (Fig.9). It is therefore advisable to choose a L/D > ( L / D ) o p t . The main parameters which determine the optimum length of an ejector are: (i) the area ratio A 3 /A, which determines the initial Mach number. At higher Mach numbers (or higher A 3 /A,) the shock waves angles are smaller and the whole supersonic flow pattern (i.e., at transition to the mixed flow regime) is stretched out. A larger L/D is required.

(ii) the mass flow rate p . Larger mass flow rates smooth the above-mentioned effect. This is, however, an upper limit of pi for the addition of a second throat. (iii) the nozzle outlet angle. Larger nozzle outlet angles provoke a stronger interaction of the primary jet with the secondary jet and the wall. The resulting shock waves are stronger and steeper. A lower (L/D) o p t is therefore to be expected.

5.

EFFECT OF TEMPERATURE

The ejector performance characteristics were derived in Section 1 for equal stagnation enthalpies of both jets. In the case of different stagnation enthalpies the energy equation has to be used in order to determine T 0 4 at the exit. In the calculation for the mixed flow regime a complete mixture of the two jets must be assumed. In calculating the characteristics of the supersonic regime the additional assumption of no heat exchange between Stations 2 and e is necessary. The simplest way to show the temperature influence is to look at the supersonic-saturated regime. From Equation (I) it follows that oA . . in = f= x constant . VTo Thus, for different T 0 , Equation (9) yields
u
p

= ^

lo
A2'

A; P ^ T ; ;

fn

pi

(19)

The characteristics derived for equal stagnation temperatures TJ, = TJJ can be used when the parameter p is replaced by pTjJ/Tj, . The calculation of the supersonic regime yields the same result. In References 4 and 5 the theoretical calculations and experimental verification are presented. Figure 11 shows the influence of TjJ/Tj, on the supersonic regimes.

6.

EJECTOR DESIGN AND OPTIMIZATION

The given data for the design of an ejector are mostly the total pressure of the secondary gas p , its mass flow rh" and the ambient pressure p s . The design problem is to determine and to optimize the geometry of the ejector and the mass flow rate n . The form of the equations in Section 1 shows that an optimization of the ejector geometry, i.e. to find the most economic geometry for the data given above, cannot be found easily. In many cases, however, the mass flow rate is limited or fixed to a rather narrow range. The calculations therefore have to be performed for one or two values of n only. Figure 5 shows that the optimum operational conditions are at transition from the supersonic to the mixed flow regime. The point of transition should be considered as the design point. It has to be recalled that it is displaced when a second throat or a subsonic diffuser is added. It can also be concluded from the equations in Section I that the parameter A 4 /A', (respectively A 3 /A',) has the greatest influence on the position of the transition point. An easy approach to the desired optimization is to choose two limits for p and to calculate the characteristics ju = constant. If the best combination of p and A4/A", is found, the geometry can still be improved by varying Aj/A', and by adding a supersonic-subsonic diffuser. The primary mass flow rh' is determined by p . The total pressure p'0 at transition then determines the primary throat cross section A', and thus the whole ejector geometry.

10 The problem of the optimization of ejectors is discussed in detail in Reference 6. In the case of very low required pressure ratios PQ/P 5 a two stage configuration might be considered. The characteristics for a two stage ejector are developed and compared with the single stage ejector characteristics in Reference 7. Finally Reference 8 should be mentioned; this contains a list of almost all papers on ejectors published before 1965.

REFERENCES
1. Fabri, J. Paulon, J. Lukasiewicz, J. Uebelhack, H. Leistner, G. Theorie et experimentation des ejecteurs supersoniques air-air. ONERA NT 36, 1956.

2. 3. 4.

Supersonic Diffusers.

ARC R & M 2501, 1946. VKI TN 28.

Supersonic Air-Air Ejectors with Second Throat Diffuser.

Experimented und theoretische Untersuchungen an einem Hochtemperatur-Uberschallejektor mit zylindrischer Mischkammer. Ph.D. Thesis, T.H.Darmstadt, 1966. Divers regimes de melange de deux flux d'enthalpies d'arret diffirentes. T P 4 1 1 , 1966. Performances d'ejecteurs supersoniques air-air a milangeur cylindrique. de Thermique, Jan. 1965. ONERA

5.

Le Grives, E. Fabri, J. Calvet, P.

6.

Revue Generale

7.

Loser, H.

Untersuchungen an ein- und zweitstufigen Uberschallejektoren mit zylindrischer Mischkammer. Ph.D. Thesis, T.H.Darmstadt, Germany, 1965. Ejectors and Mixing of Streams. RAE Library Bibliography No. 252.

8.

Seddon, J. Dyke, M.

II

NOZZLE

3'

AIR SETTLING CHAMBER

SUPERSONIC DIFFUSER (SECOND THROAT)

MIXING

CHAMBER

Fig. 1 Diagram of a supersonic ejector and reference stations

FIG.2.1

SUPERSONIC REGIME NORMAL SHOCK IN THE DIFFUSER

WAVE

PC PPPx=* -i
A

SUPERSONIC PORTION FIG. 2.2 SUPERSONIC REGIME CLOSE TO TRANSITION

FIG. 2 3

MIXED

FLOW

REGIME

/0>^>c
FIG. 2 4 MIXED FLOW REGIME SUPERSONIC ATTACHED TO ONE SIDE JET

FIG. 2 5 MIXED FLOW REGIME PRIMARY NOZZLE

WITH

SEPARATION

IN

Fig.2

Flow patterns in a supersonic ejector. Rising exit pressure from top to bottom

12
f

(M*)
1.2

"(Mm)

Jtoti
8
q

(Mm)

.4

.8

1.2

1.6

2.0

2.4

/V.

5.0 Z(M) 4.0

3.0 Z(M*)

2.0

1.0

.4

.6 Fig.3

12

1.6

2.0

2.4

M+

Dimensionless functions of M#

Fig.4

Supersonic regime

13

Fig.5

Ejector performance characteristics

*2

EXPERIMENTAL 1.0
A
/J.

1 A4 Aj '
1

.3
6 2 5

/V

*2

Am

* = 2.78 o
G
>a^

A] 0.8

>REF (1)

0.6 o sf/r
0

v^ i

^ S ^

0.4

02 o o 0 & 0.01
A

o 0.02

0.03

0.04

0.05

0.06

0.07

006

0.09

Po Po

'

Fig. 6 Supersonic regime

14

P"

.3

.2 EXPERIMENTS A a L/D = 13 L/D = 17

10 P_
P

Fig.7

Effect of friction influence

Pc p* 5

Po 1 p~ WITH s SUBSONIC DIFFUSER

. \ \ \

AM

4',

6.25

A2 \ 4 SUBSONIC \ \ DIFFUSER \ } ANL3 \ SECON,D \ THRO


A1

2.76 ^

tf

.3

y^

.1

P<
Fig.8 Influence of a subsonic diffuser and a second throat

15

A-,. -TrAi 6
A'l

=6.25
A

A'I

=2.78

*'S

./*

UPb*A

S^/y*
/fP ^ EXPERIMENTAL 0 L/D * 6 A L/D= 7 0 L/D = 6 V L/D= 9

10

12

Pa'

Fig.9

Influence of diffuser length

10

12

14

76

<-/r

20

Fig. 10 Pressure distribution in the mixing chamber, mixed flow regime

16

M
0.5 - 1 0.4 0.3
Mjm.2*

02 01 0 0.0
PQ_

Fig. 11

The influence of

TQ/TJ,

on the supersonic and supersonic-saturated regime, from Reference 4

17

ANALYSIS AND DESIGN METHOD FOR EJECTOR SYSTEMS WITH SECOND THROAT DIFFUSERS by H.T.Uebelhack

Lecture Series Director Dornier System Friedrichshafen, Germany

18

19 CONTENTS Page NOTATION 1. 2. INTRODUCTION THE EJECTOR CHARACTERISTICS DIAGRAM AND THE INFLUENCE OF A SECOND THROAT DIFFUSER 3. 4. 5. 6. 7. 8. FLOW MODEL ANALYSIS THE SECOND THROAT CONTRACTION RATIO THE EJECTOR STARTING PRESSURES DESIGN PROCEDURE COMPARISON WITH EXPERIMENTS 21 22 22 24 24 25 25 26 26-30 20 21

REFERENCES FIGURES

20

NOTATION a A D
h

speed of sound cross-section diameter, drag step height diffuser length mass flow Mach number pressure temperature longitudinal coordinate lateral coordinate or step coordinate specific heat ratio separation angle mass flow ratio density

L m M P T x
y 7

e
M p

Subscripts stagnation conditions before separation behind separation behind reattachment diffuser outlet section, Figure 3 * throat section second throat section nozzle outlet section separation, starting (ejector) Superscripts a prime refers to primary jet a double prime refers to secondary jet

21

ANALYSIS AND DESIGN METHOD FOR EJECTOR SYSTEMS WITH SECOND THROAT DIFFUSERS H.T.Uebelhack

1.

INTRODUCTION

A second throat diffuser reduces the driving pressure as well as the suction pressure of an ejector system by up to 30% as compared to the diffuser system with a constant cross-section diffuser. In order-to benefit from this improvement in performance the improved starting characteristics of a second throat ejector must be known. Only then can the whole ejector geometry, in particular the diameters in the various sections, be adapted to certain requirements. The performance characteristics (i.e. Pp/p 4 as a function of pj,/p 4 and M) can be determined by inviscid onedimensional theories 1,2 and in the lower secondary mass flow regime by additional considerations of viscous interactions 3 . The difficulty in calculating the improved performance characteristics of a second throat ejector system consists in determining the pressure integral over the contraction part of the supersonic (second throat) diffuser which is required in the momentum balance. Efforts have been made in the past decade at AEDC, Tullahoma, Tennessee, USA, to develop prediction methods for the zero secondary flow ejector systems 4 , 5 . Extensive experimental studies at VKI, Rhode-Saint-Genese, Belgium, on this problem have revealed two features of the second throat ejector system which can be used in developing a prediction method for these systems: (i) The ramp angle of the contraction portion of the supersonic diffuser is of negligible importance to the overall ejector characteristics. Angles up to 90 were investigated. The 90 step type contraction, therefore, can be used in defining a flow model for the second throat system. Results of studies of the supersonic separated flow field in front of s t e p s 6 - 8 , in particular the drag of the step, can be introduced into the analysis. The starting pressure ratio (p' 0 /p 4 ) s is not affected by small secondary mass flows. The starting pressure ratio which is at the same time the pressure ratio for the design and operation of a second throat ejector system, can therefore be determined for zero secondary mass flow. This ratio remains constant for secondary mass flow ratios up to n ~ 0.25, which is also the limit for second throat diffuser operations.

(ii)

In the following paragraphs a flow model based on the above observations will be defined which permits the determination of the starting and operating characteristics of a second throat ejector system. The working equations will be derived briefly and the design procedure is described. The limits of application are discussed and comparisons with experiments are made.

2.

THE EJECTOR CHARACTERISTICS DIAGRAM AND THE INFLUENCE OF A SECOND THROAT DIFFUSER

The representation of the ejector performance characteristics will be the same as the one utilized in the onedimensional analysis in Chapter 1 of this Agardograph. For a better understanding of the final results, the influence of different diffuser shapes should be briefly demonstrated in a schematic performance characteristic diagram. In Figure 1 the performance characteristics for an arbitrary ejector system and one mass flow ratio, n , is shown. The branch marked 0 represents the supersonic regime. The exit pressure has no infleunce on the secondary pressure pj| in this operational mode. The mixed flow regime, where the secondary pressure is usually highly influenced by the exit pressure is represented by the branches 1, 2 and 3 for various diffuser shapes. The curve 1 would be typical for a constant cross-section diffuser without, and the curve 2 with, a subsonic (divergent) diffuser part. A contraction of the diffuser pipe will act as a supersonic diffuser and will deliver a characteristic line shown as curve 3. The branch marked 4 would be the mixed flow regime with separation in the driving nozzle, an operational mode of merely academic interest. In the following representation only the intersection points between the mixed flow and the supersonic regime will be shown for various mass flow ratios, n . Those starting points give at the same time design and operational condition. For given geometry and mass flow they represent an optimum (minimum pJJ) operation.

22 3. FLOW MODEL

As mentioned earlier the flow model used in the following calculation is based on two features of second throat diffusers which have mainly been found experimentally: (i) the angle of the contraction part of a second throat diffuser has no influence on the overall starting and operational conditions of the system. The experimental results of flows over rectangular front steps can thus be introduced in the momentum balance during the analysis. (ii) the starting pressure ratio of the driving gas is for second throat systems, in contrast to constant crosssection diffusers, independent of the secondary mass flow. The analysis to determine the driving pressure of a second throat system carried out for zero secondary mass flow, therefore, is also valid for operations with secondary mass flow up to n ca 0.25 . In addition the following assumption and methods will be introduced for the calculation of the starting characteristics of a second throat system:

The calculations of the base pressure p^ , the initial pressure p, , initial Mach number M, , and finally the step pressure integral are carried out successively. (iii) The base pressure pJJ can be determined in several different ways: (a) By means of Korst's base pressure theory, i.e. the combination of the shear layer characteristics with an appropriate reattachment criterion (Ref.3). (b) Higher secondary mass flows are treated by the one-dimensional ejector theory presented in Chapter 1 of this Agardograph. (c) Conical nozzles of 10 to 20 half-angle produce a base pressure (with zero mass bleed) which is very close to the static pressure of an isentropic expansion from stagnation conditions to the diffuser area ratio, A d /A* (Fig.2). (d) Contoured nozzles of zero outlet angle produce an appreciably lower base pressure which can be taken from Figure 2 for various geometries (experimental results). (iv) The jet boundary Mach number is determined by the isentropic two-dimensional expansion from stagnation to base pressure at zero secondary flow. (v) The flow angle of the jet boundary is given by a two-dimensional Prandtl-Meyer expansion at the exit of the nozzle. (vi) The initial pressure p, and the initial Mach number \S\, are then calculated by the two-dimensional oblique shock equations. The geometrical parameter, A,j/Ae , which determines the relative length of the jet boundary, must be less than four in order to avoid curvature of the jet boundary due to the axisymmetric nature of the flow. In practical applications A(j/Ae is not larger than two. The step pressure integral will be calculated by the experimentally obtained relation7'8

. ft*
J Pi h 4. ANALYSIS

= 1.1 M .

(1)

In the following analysis the major aim is to determine the one-dimensional supersonic Mach number in the section after the contraction. The pressure recovery then can be determined by normal shock equations. Apart from two-dimensional considerations in the region between the nozzle exit plane and the contraction plane, the calculation is strictly one-dimensional.

23 Conservation Laws The momentum equation applied over a control volume (Fig.3) between the nozzle exit (subscript e) and the second throat contraction (subscript **) yields

pj^(1
Po A*

+7 M | . ) + 4
Po

Ad Ae

'

-fm%
stJp

Ad

** = f S S - ^ ^ S d + T M l U ) .
Po** Po A*

(2)

A*

p,h p0

A*

the equation having been normalized by p' 0 A*. The calculation will be carried out as stated in Section 3 for zero secondary mass flow. Equating the mass flows in the primary nozzle throat section and in the second throat section yields
u

** *

P* A+ P** A**

(3)

The ratio u ^ / a # in this equation represents the "star" Mach number in the second throat section, ( M A ) ^ . Equation (3) can be further transformed into P* Po Po** A*

(M*)** = Po ? Po** P** A** '

(4)

where p' 0 is the density in the primary settling chamber (pressure pj,) and p 0 * * P 0 **/RT 0 . The density ratios can be expressed by the local Mach number through the energy equation and isentropic flow relations 2 PjP \7- pn * * P**

' & $

-O* ^-*-)

1/(7-0
(5)

The state equation yields for adiabatic flow, T 0 = T j , , Po**/Po


=

Po**/Po

(6)

Introducing Equations (5) and (6) into Equation (4), we get

2 2

y/( 7 -D/ / y/w-u

7 7_ -

i 1

Mi

y/(7-D v

(7)

A** Po**

The relation between the Mach number M AA in the second throat and the "star" Mach number (M^) + + based on the critical speed of sound, a*, is

(M*)

7+ 1 7-1.
+

11/2 (8)

2 I 7 - 1 M^j yields 2 ( 7 - DMJ, 1/2 (9)

Introducing Equation (8) into Equation (7) and solving for A ^ p J , ^ / A # p J ,

1/(7-0 7 - . / 2 \l/(7-D/ T - l 1 + Po** A ^ _ ' 1+ M** 7 + 1 V" Po A # \y +1 When Equation (9) is introduced into Equation (2) and P**/p 0 ** is replaced by

**. Po**

fi-f MJ (.^n.) V
7
2

\(T-D/7

(10)

the right-hand side (RHS) of Equation (2) becomes a function of the Mach number in the second throat section only: / RHS = _ _ , 1 + \(7-0/7/ 2 \*/(7-l)/
7

- l

\l K f l ) y - 1

1/2 1 +

7+ 1

(7-DMJ

(l+7Mi*)-(2a)

This form of Equation (2a) enables one to determine the Mach number M*A , which itself permits the calculation of the ratio of the total pressures, P 0 **/P o T h e M a c h number M4 in the exit section and the pressure ratio p 0 * * / P 4

24

can then be calculated by normal shock equations. This finally enables one to calculate the starting pressure ratio of the second throat ejector system, (PQ/P 4 ) S by (Po/PA =
Pot

Po P4 P 0 * *

(11)

5.

THE SECOND THROAT CONTRACTION RATIO

Experiments have shown that the decrease of the ejector starting pressure, (pj,/p4)s of a second throat system is approximately proportional to the second throat contraction ratio A ^ / A j (Refs.9,10). It is therefore desirable to know and to work with the lowest value of J ^ ^ / A J for which starting is possible. The limit of the contraction ratio, A ^ / A j , can be calculated by one-dimensional flow equations and the following argument: When starting the ejector system (increasing pj,/p4) a normal shock wave moves downstream.

M rf >/
/ / / / / / / / l / / / / / J J /

M<l
y / / / / / / / / ? / - /

The mass flow in the system must be swallowed by the second throat. The limiting contraction of the second throat is thus dictated by the necessary re-acceleration of the subsonic flow behind the shock wave to sonic velocity. A further decrease of the second throat section would cause choking and would make it impossible to start the second throat system. The contraction limit turns out to be a function of the supersonic Mach number before the second throat only: (12) 27 2 7 M2J For Mach numbers Mj approaching infinity the contraction ratio approaches asymptotically to A#A/A,j = 0.6 for 7 = 1.4 . - i V 2 / 27 V / C l -i)/ 2 1 \ "2 / 7 - 1 1 \i/(7-D *** = (i^iy (ji_\^- r ) (,, 2 _Lv A 7-1 IV 1+ r 1 f l - 7-~ \ .
/ 2 2

Ad

7+ // \ \, 7 +1 l V77+ + l1//

V V

7 yy 7- l1M M

V" V

The lower limit of the contraction ratio /V^/A,- was also determined experimentally9'11. It is shown in Figure 4 as a function of the nozzle exit area ratio A e /A # . Because of the three-dimensionality of the flow during the starting process (flow separation, oblique shock waves), this limit was found to be lower than that obtained from a one-dimensional calculation. For high expansion rates (A e /A* > 10) it is practically constant and takes the value A ^ / A d = 0.47 . 6. THE EJECTOR STARTING PRESSURES

The ejector starting pressures pj,/p4 and (pj,/p4)s were computed by the one-dimensional theory and by the present method (outlined in Section 3) for two second throat contraction ratios: that given by Equation (12) and the experimental limit of Figure 4. The starting secondary pressure ratio pJJ/p4 has been determined by the onedimensional theory. In Figure 5, the starting pressures of the constant cross-section diffuser and the second throat diffusers are plotted as a function of the diffuser inlet area ratio Aj/.A* and compared. It is evident from this figure that it is not worthwhile to use a second throat diffuser at area ratios smaller than A d /A A = 8 . At area ratios above Ad/A* = 20 the benefit (i.e. the reductions of the starting pressure) obtained by a second throat diffuser remains approximately constant. Figures 6 and 7 show the ejector starting pressures for various geometries in the ejector characteristics diagram and Figure 8 represents a composite of these diagrams. They enable the choice of a tentative ejector geometry (Ad/A*, A e /j\* and A ^ / A j ) which meets certain prescribed requirements (pj,/p4, PQ/P 4 , /-") The exact calculation of the ejector characteristics of the chosen geometry can then be carried out as indicated in Sections 3 and 4.

25 7. DESIGN PROCEDURE

After fixing the ejector geometry (Ag/fi^, ^/.A*, A^/A^) which meets the presented requirements by the use of Figure 8 and the calculation procedure of Sections 3 and 4, and after fixing the mass flow ratio /i , the size of the throat section A* will be determined. The other geometric parameters are then given by the above area ratios. Restrictions: It is evident that the prediction method developed (Sections 3 and 4) is only valid for area ratios Aj/A e < 4 . Larger area ratios yield a flow field which can no longer be treated with two-dimensional equations as suggested in Section 3. However, in practical ejector design problems area ratios A,j/Ae > 4 are not required. The distance between the ejector driving nozzle and the location of the second throat contraction part should be designed to be adjustable. A too long or too short distance bteween them might produce hazards in ejector operations by either breakdown of the flow field or shock wave interference. 8. COMPARISON WITH EXPERIMENTS

Comparisons of the results of the present theory with experimental data are made in Figures 6, 7 and 9. Figure 6 shows excellent agreement of the experimentally obtained starting pressure ratios with the predicted ones at a contraction ratio of .A^/Ad = 0.68 . The experimental starting pressure ratios of the constant cross-section diffuser system of Figure 6 are by about 5% higher than the values predicted by the one-dimensional theory. The contraction of the second throat was realized by a step in this case. Figure 7 shows a comparison between calculated and measured starting pressure ratios for an ejector system with a nozzle of 15 half-angle and a 12 ramp as diffuser contraction. Here again excellent agreement between the predicted and the measured second throat starting pressure ratios can be observed. The experimental starting pressure ratios of the constant cross-section diffuser system are again 5% higher than the predictions of the one-dimensional theory. The difference between the experimental starting pressure ratios and those obtained by the inviscid onedimensional theory can be attributed to the effect of friction, which has been ignored in the theory. The presence of friction shifts the characteristics of the mixed flow regime and thus the starting pressure ratio to higher values
of PQ/P4

In the case of a second throat diffuser system the diffuser inlet Mach number is immediately reduced and friction becomes less important. The experimental values, therefore, correspond with those predicted by the present theory, which also ignores wall friction. The starting pressures of the second throat systems are compared with the starting pressure of the constant area diffuser system in Figure 9. The lines indicate the theoretical predictions of the reduction in starting pressure ratio, i.e. the starting pressure ratios of the second throat system divided by the starting pressure ratios of the constant cross-section ejector. Both circles represent experimental values obtained in the present study (experimental starting pressure ratio of the second throat system over theoretical starting pressure ratio of the constant cross-section ejector. The theoretical starting pressure ratios of the constant cross-section ejector have been taken as a common reference here since the experimental starting pressure ratios of the constant cross-section ejector were influenced by wall friction and are 5% too high). The two AEDC data are the ratios of measured starting pressure ratios (squares) and the lines are again the ratio of the two calculated values (second throat over constant cross-section). In both the experiments and the theoretical calculation p4 is the total pressure at the diffuser exit (a subsonic diffuser was used in the experiments). The geometric parameters are indicated at each data point. The very good agreement between the experiments and the present theory confirms again that all important factors have been taken into account.

26 REFERENCES

1. Fabri, J., Siebstrunck, R. 2. Uebelhack, H.T. 3. Chow, W.L., Addy, A.L. 4. Panesci, J.H., German, R.C. 5. German, R.C., Panesci, J.H. 6. Zukoski, E.E.

Supersonic Air Ejectors, Advances in Applied Mechanics. Vol.5, 1958.

One-Dimensional Inviscid Analysis of Supersonic Ejectors. Chapter 1 of this Agardograph. Interaction between Primary and Secondary Streams of Supersonic Ejector Systems and Their Performance Characteristics. AIAA Journal, Vol.2, No.4, 1964. An Analysis of Second Throat Diffuser Performance for Zero Secondary Flow Ejector Systems. AEDC TR 63-249. Improved Method for Determining Second Throat Diffuser Performance of Zero Secondary Flow Ejector Systems. AEDC TR 65-124. Turbulent Boundary Layer Separation in Front of a Forward Facing Step. AIAA Journal, Vol.5, No. 10, 1967. Theoretical and Experimental Investigation of Turbulent Supersonic Separated Flows over Front Steps. 4. Jahrestagung DGLR, Nr. 71-076, October 1971. Turbulent Flow Separation ahead of Forward Facing Steps in Supersonic Two-Dimensional and Axisymmetric Flows. VKI, TN54, 1969. Ejector Design for a Variety of Applications, von KarmSn Institute for Fluid Dynamics, Rhode-Saint-Genese, Belgium, Short Course on Ejectors, March 17-21, 1969. Supersonic Air-Air Ejectors with Second Throat Diffuser. VKI TN 28, August 1965.

7. Uebelhack, H.T.

8. Uebelhack, H.T.

9. Taylor

10. Uebelhack, H.T. 11. Oiknine, C , et al.

Etude des Ejecteurs Supersoniques. Congrs de I'Association Francaise des Ingdnieurs et Techniciens de I'Aeionautique et de 1'Espace, November 1969.

P-xit

P'~

STARTING PRESSURE RATIOS

/Petit

Fig. 1

Ejector characteristics diagram. Influence of subsonic and supersonic diffuser parts

27

1r-r
ONE-DIM EXPANSION P/R. EXPERIMENTS CONTOURED NOZZLES

AEDC -DATA, REF CONICAL NOZZLES

-10

Fig. 2

Ejector base pressures

*'(**-*.)

SpdA

(**%,*)*.

fox*

**A.U'**)A*

(t* 4<K

L.

..J
Fig.3 Control volume for diffuser analysis

28

1.0 A*0 Ad

.75.

ONE DIMENSIONAL STARTING LIMIT

.50

STARTING STARTING

POS/BLE

IMPOSSIBLE

.30to 20 30 40 Ad/A ^

Fig.4

Experimental starting limit

ONE-DIM. THEORY A d / A e = 1.5 NORMAL SHOCK RECOVERY 2ND THROAT ONE-DIM. CONTRACTION UMIT.A<*/A.~15 2ND THROAT EXPERIMENTAL CONTRACTION LIMIT A d /A^ ~ 1.5

20

10

20 Fig. 5 Starting pressure ratios

30

Ad/A*

29

EXPERIMENTS

/u. = IS
M - -10 4 L - OS

I Pjs

--ET- p-

' A d / A m M 2S

- v-_ o A e /Am, = 16

THEORY

.3

NO SEC. THROAT A /A d . .68

A^lAd'rif

;o
Fig.6 Ejector starting characteristics

fsv-u

(f).
<

EXPERIMENTS

AA.'.20 A*.'.15 'M.-.10 /A.'.05 M." 0

i
-.-ET>t(//j4<>^75
A / A * . 12.5 A * * / A d * .67 15* NOZZLE

--V-

o-

THEORY NO SEC. THROAT A**/Ad*67

10

(Po'/P*),

Fig.7

Ejector starting characteristics

30

(Po) P4S

1 1
Ad/A * . 1 6 Ae/A * ' 9

I I I I
A d / A * =25 A ' / A * *16 A * * / A d =1 .63 7

Ad/A. * 9 Ae/A*.6.2S .4 A*/Ad = l .68 .47 .3

Ad/Arrr.36 At/An -25 A**/Ad = 1 .64 47

A*./Ad 1 .66 .47

tj^\ \

-?

I0y

.15^

iS*-"""

_Jg^
_J75

.05^,

.;
0 0
1

05^-

\_^5__^
05 0
1

Li
10

WJSO 75

20

25

(Po/P+ln

Fig.8

Composite of ejector starting characteristics

1
(P0lp4)**s
AEDC EXPERIMENTS, REF. PRESENT EXPERIMENTS
r

fWs
1.0

H0RY

sJOZZLE HALF ANGLE 8

15

A . / A m = I2.S

O
,8

AKKfAdm

67

AmjAm 16 A x x f A d ' .68


A e l A * AXKIAd= .5

A e /Am = 25 A**/*d=
S

18

,i

to

20

30

40

SO

AdJAA

Fig.9

Experimental and theoretical ejector starting pressure reduction

31

THE ANALYSIS OF SUPERSONIC EJECTOR SYSTEMS by A.L.Addy

Mechanical and Industrial Engineering Department University of Illinois at Urbana-Champaign Urbana, Illinois 61801, USA

32

FOREWORD

These notes served as the basis for a series of lectures presented under NATO sponsorship at the von Karman Institute for Fluid Dynamics, Brussels, Belgium. April 1968. The important contributions of Professors H.H.Korst and W.L.Chow, University of Illinois at Urbana-Champaign, to the understanding and analysis of supersonic ejector systems are herewith acknowledged.

33 CONTENTS Page NOTATION INTRODUCTION 1. SUPERSONIC EJECTOR SYSTEM CHARACTERISTICS 1.1 Performance Characteristics 1.2 Flow Regimes 1.3 Recompression Within Ejector Systems 1.4 Methodology of Ejector System Performance Analysis "ZERO" FLOW REGIME 2.1 Flow Model 2.2 Primary Flow Field 2.3 Mixing Component 2.4 Recompression Criteria "SMALL" SECONDARY FLOW REGIME 3.1 Flow Model 3.2 Primary-Secondary Flow Fields 3.3 Mixing Component 3.4 Solution Criterion "MODERATE TO HIGH" SECONDARY FLOW REGIME 4.1 Row Model 4.2 Inviscid Flow-Fields Analysis 4.3 Solution Criteria 4.4 "Downstream" Flow-Field Analysis 4.5 Two-Stream Mixing Correction 4.6 Shroud-Wall Boundary Layer 5. 6. 7. EJECTOR THRUST EVALUATION EJECTOR FLOW MODEL IMPLEMENTATION AREAS WARRANTING FURTHER INVESTIGATION 35-37 39 39 40 41 42 43 43 43 44 44 45 47 47 48 48 49 49 50 50 52 54 56 57 57 58 58 60 61-77 78-84

2.

3.

4.

TABLE I FIGURES APPENDIX I - A Literature Review of Ejector Systems and Related Topics

APPENDIX II - Method of Characteristics Analysis for the Supersonic Axisymmetric Primary Flow Field

1. 2. 3. 4.

BASIC EQUATIONS FIELD POINTS AXIS POINTS BOUNDARY POINTS 4.1 Constant Pressure Boundary 4.2 Non-Constant Pressure Boundary PRIMARY FLOW-FIELD ANALYSIS 5.1 Calculation Sequence 5.2 Wave Coalescence

85 86 87 88 88 88 89 89 89

5.

34 Page 6 INITIAL PRIMARY NOZZLE CHARACTERISTIC 6.1 Sonic Nozzle 6.2 Uniform Supersonic Nozzle 6.3 Conical Supersonic Nozzle 6.4 A Compression at the Nozzle Exit 89 90 90 90 90 91-94

FIGURES FOR APPENDIX II APPENDIX III - Constant-Pressure Turbulent Mixing Analysis FIGURES FOR APPENDIX III REFERENCES

98-99 100-101

35 NOTATION

A
- *

A A.B.C C
C

vectorial area in ejector control volume analysis coefficients in equation defining the shroud wall contour U / U m a x = {1 + [ 2 / ( 7 - DM 2 ]}- 1 ' 2 , Crocco number generalized bleed coefficient. Reference 16 function of () momentum flux, Reference 16 iteration indices Mach number { [ ( 7 + l)M 2 /2]/[l + ( 7 - 1)M 2 /2]) I / 2 , Mach star absolute pressure mass flow rate, Reference 16 gas constant or radius absolute temperature gross thrust force velocity components velocity vector mass flow rate mixing length intrinsic coordinate system reference coordinate system cylindrical coordinate system

f() i i.n M M*

P
q

R T 'gross u,v,Vx -

w
x

x,y

X, Y X, R

Barred Symbols P. Pb Pi. Pos Pw X P a /Pop Pb/Po P Pis/Pop P0s/P0p P w /P0p X/Rlp Ws^r'RsTosl^/Wp^p'RpTop]1

w
Greek Symbols a 7

sin _ 1 (l/M), Mach angle C p /C v , ratio of specific heats increment

turning angle

36 6* e T) 7j m 6 A p a sp \p two-stream mixing region displacement thickness arbitrarily small positive quantity ay/x , homogeneous coordinate dimensionless displacement of the mixing region streamline angle (positive, ccw) Tn/o a - stagnation temperature ratio density or radius of curvature mixing region similarity parameter U / U a , velocity ratio angle of reattachment. Reference 16

Subscripts a c d e i j / m ma mtn p Op lp s Os Is shk x,y w BO IE OE i u conditions along the jet boundary choked condition in secondary stream discriminating streamline ejector exit section inviscid flow jet boundary streamline limiting streamline or limiting initial secondary flow Mach number turbulent mixing conditions for minimum secondary flow area conditions at the minimum secondary flow area primary stream primary stream stagnation state primary stream conditions at initial ejector section secondary stream secondary stream stagnation state secondary stream conditions at initial ejector section shock location conditions upstream and downstream of a normal Shockwave conditions along the shroud wall break-off point ideally-expanded flow over-expanded flow single-stream mixing two-stream mixing

37

Functions P *
=- (7.M*) =

l-<lZi>M*2
( 7 + 1) (7 + 1)
1/2

7/(7-0

p(7.M*) F(7,M*)
W

1 -

( 7 - 1)

M*:

-I/(7-D

*T

1/2 I PA = M" ( 7 + 1)

, - (^ lV2 7 + 1)

1/(7-0

= two-stream mixing region displacement thickness function. See Appendix III. l,(rj) = mixing integral function. See Appendix III.

erf (T?) = 4 " T ^

38

34

THE ANALYSIS OF SUPERSONIC EJECTOR SYSTEMS A.L.Addy

INTRODUCTION In ejector systems, "pumping action" is achieved through the controlled interaction and mixing of a high-velocity and high-energy stream with a lower-velocity and lower-energy stream within a duct; the simplicity of such systems has resulted in their wide-spread usage. Unfortunately, this simplicity does not carry over to the design and analysis of the flow phenomena within these systems. As a consequence of the problems involved, the design and performance evaluation of ejector systems has developed as a combination of scale-model studies, empiricism, and theoretical analyses applicable only to simplified configurations. This approach, although useful for the design and performance evaluation of certain configurations, has the disadvantage of failing to provide a more general insight into the significance and influence of system parameters and operating conditions on overall ejector performance. As a result of more sophisticated applications of ejector systems along with stringent performance requirements, the development of more general analytical methods for predicting their detailed performance characteristics is essential. This objective, clearly hampered by the complexity of the flow phenomena and the breadth of potential configurations, can be accomplished only by a detailed study and modeling of the flow phenomena throughout the operating regimes of an ejector system. The objective herein is to present and discuss the ejector flow model and its implementation 1 - 4 * that was developed at the University of Illinois at Urbana-Champaign for the performance analysis of "supersonic"ejector systems. This flow model along with its current implementation represents a significant departure from onedimensional analyses 5 - 9 . In this model the flow phenomena within an ejector system are delineated on the bases of the predominant flow mechanisms which occur within the various operating regimes. In this framework, it is then possible to establish overall ejector performance characteristics and to represent these characteristics in the unified form of "characteristic performance surfaces". These surfaces serve as a qualitative basis for understanding the overall performance of an ejector system and as a quantitative basis for making judgments regarding the significance and influence of system parameters and variations in the operating conditions. In the sections that follow, a general topical organization has been selected which will present: 1. 2. 3. 4. 5. An overview of the operating characteristics of ejector systems. Qualitative aspects of the ejector flow model in relation to the various flow regimes. The quantitative implementation of the ejector flow model. The performance evaluation of ejector systems. A brief discussion of problem areas, their significance and possible future investigations.

The material presented in the following sections is based principally on investigations-)- conducted at the University of Illinois at Urbana-Champaign and is supplemented, where appropriate, by brief discussions of the related work of others. Since only selected references are cited, a more complete compilation of recent work and other approaches in the analysis of ejector systems is included in Appendix I.

1.

SUPERSONIC EJECTOR SYSTEM CHARACTERISTICS

To establish a basis for the detailed modeling and performance analysis of supersonic ejector systems, a qualitative discussion of the performance and nature of such systems is given in this section. Emphasis has been placed on defining the general functional relationships describing the performance of these systems and how their form is dependent on the internal flow phenomena.

* The numbered References will be found on p.121 (i.e. after Appendix III of this paper). t Partially supported by NASA Research Grants NSG-13-59 and NGRI4-00S-032. Computer studies in connection with these investigations were conducted using the IBM 7094 and 360/75 Systems, Department of Computer Science, University of Illinois at Urbana-Champaign, Urbana, Illinois.

40 A representative ejector configuration and the associated notation is shown in Figure 1. The primary stream is assumed to be supplied from the stagnation state (Pop-Tn p ) through a sonic or supersonic nozzle, and the secondary stream is from the stagnation state (POS-TQS)- The secondary and primary streams begin their mutual interaction at their point of confluence at the primary nozzle exit. This interaction, as well as the mixing between the streams, continues to the shroud exit where they are discharged to the ambient pressure level P a . 1.1 Performance Characteristics

The objective of any ejector analysis is to establish, for a given configuration and working media, the performance characteristics of the system. In general, the mass-flow characteristics of an ejector system can be represented functionally by: W = f(P0s.Pa). (LI)

where the "reduced mass flow ratio" (W) is chosen as the dependent variable so that the influence of the stagnation temperature ratio is essentially removed from the functional relationship. An alternate formulation oJ[ the pumping characteristics in terms of the initial secondary-stream Mach number (M, s ), the static pressureratio (Pj s ) of the secondary stream at the point of confluence of the two streams, and the ambient pressure ratio (P a ) is given in functional form by; Mis = f ( P i s . P a ) (1-2)

This selection of variables, although less obvious, is convenient for performing the numerical calculations involved in the theoretical ejector analysis to be described. In addition to establishing the functional form of the pumping characteristics, another variable of interest is the shroud wall pressure distribution given by: Pw = f ( W , P 0 s . P a , X ) . (1.3)

After establishing the above functional relationships, the thrust characteristics of a system can then be determined. In practice, this is accomplished by considering the contributions in the axial direction of the entering momentum fluxes of the primary and secondary streams and the integrated shroud-wall pressure distribution. /. 1.1 Three-Dimensional Performance Surfaces The functional relations, (1.1) and (1.2), characterize^the "pumping" characteristics of an ejector system and represent surfaces in the spaces described by the triples (W, PQ S , P a ) and (Mi s , Pj s , P a ). The pumping characteristics of a typical ejector system in terms of the variables (W, P u s , P a ) are shown in Figure 2. This surface clearly delineates the flow regimes wherein the mass-flow characteristics are independent or dependent on the ambient-pressure level. These flow regimes merge together along the "break-off curve" and in principle, this condition serves to uniquely define this curve. To the left of the "break-off curve", the_mass-flow characteristics are independent of P a and the surface is cylindrical with its generator parallel to the P a -axis . For this regime, the mass-flow characteristics can be represented by: W = f(Pos) (1-4)

when P a < (P a )go To the right of the "break-off curve", the surface is three-dimensional in nature and extends from the spatial "break-off curve" to the plane where W = 0 ; hence, W = f(P 0 s ,P a ) when P a > (P a ) B o In principle, the "break-off curve" represents a simultaneous solution of the functional relationships (1.4) and (1.5). However, the "break-off curve" also has a phenomenological interpretation based on the recompression of the flow within the ejector shroud. Points on the "break-off curve" are determined by the condition that transition from independence to dependence, and vice versa, on the ambient pressure ratio will occur at the maximum values of P a to which the flow can recompress. Obviously, the locus of the recompression states defining the "break-off curve" is strongly dependent on the degree of mixing between the streams and the ejector geometry, principally the shroud length-to-diameter ratio. An alternative representation of the pumping characteristics,^ terms of the variables (Mi,s, Pi s , P a ) is given in Figure 3. For this surface, there are direct counterparts to the P a -independent and P a -dependent flow regimes of the W-surface. (1.5)

41

1.1.2 Two-Dimensional Parametric Curves The three-dimensional performance surfaces of Figures 2 and 3 have their principal value in presenting an overview of the performance characteristics of typical ejector systems. In theoretical analyses or experimental programs, it is generally more convenient to consider two-dimensional parametric representations of these operating surfaces. These parametric curves usually represent nothing more than intersections of the performance surfaces with various planes corresponding to constant values of the respective variables. The more useful of the possible parametric representations of the mass-flow characteristics, from an experimental standpoint, are obtained by intersecting the W-surface by planes of constant Pa , Figure 4(a), and planes of constant P 0s , Figure 4(b). Another mteresting and useful parametric curve can be obtained by intersecting the W-surface by a plane where P u s = Pa , Figure 4(c). The latter situation corresponds to inducting the secondary fluid at ambient conditions and then discharging the ejector to the same ambient conditions. More convenient, from the standpoint of the theoretical jinalysis, are intersections of the M]s-surface by planes of constant P. s , Figure 5(a), and planes of constant Pa , Figure 5(b). In actuality, the theoretical analysis establishes the ejector performance characteristics first in planes of constant P[S by varying the value of M| s until the appropriate solution criteria are satisfied for the various flow conditions that can exist in this plane. The next step in establishing the overall system performance is to select another value of P ] s and repeat the calculations; this procedure is repeated until the overall ejector performance surfaces are established. 1.2 Flow Regimes The flow regimes occurring withinan ejector system can be categorized according to whether the mass-flow characteristics are Pa-indcpendent or Pa-dependent. In addition, a further subdivision within these flow regimes can be made on the basis of the predominant governing flow mechanisms. These subdivisions are not as clearly defined as in the former case since the change in the predominant flow mechanisms is gradual and in essence continuous. A brief discussion of the general flow phenomena and the basic flow regimes within ejector systems follows. When the mass-flow characteristics are independent of Pa , flow conditions are established within the ejector system which effectively "seal-off the secondary flow from the ambient conditions. For "small" flow rates, this is accomplished by an oblique shock system within the primary stream that is located at the shroud wall. As the secondary flow rate is increased, an operating condition will be reached where the oblique shock system can no longer be sustained at the shroud wall and as a consequence,_the primary stream "breaks" away from the wall. For the mass-flow characteristics to remain independent of Pa , the secondary stream must_then accelerate until it "chokes" inside the ejector shroud. After this condition occurs, increasing the value of Pi s or P u s results in the progressive upstream movement of the "choking" point until the secondary stream "chokes" at the point of confluence of the two streams, i.e., Mjs = 1 . When W depends on both (Prjs'Pa)* '' l e secondary stream does not "choke" inside the ejector shroud. For this condition, the secondary flow is initially accelerated and then decelerated and diffused as a result of the boundary condition imposed by the ambient pressure at the ejector shroud exit. The "choked" and "unchoked" regimes correspond respectively to the Pa-independent and Pa-dependent regimes previously discussed. Thus, the "break-off curve" separates regimes of significantly different flow phenomena within the ejector shroud. 1.2.1 "Zero" Secondary Flow The typical variation of P us with PP , when W =_0 , is shown in Figure 6; this curve represents the intersection of the W-surface or MIs-surface with the plane W = 0 . This flow condition corresponds to the well-known "internal base-pressure problem" which has been the subject of extensive and continuing investigations10"26. For the portion of the curve where Prjs is a constant, the flow phenomena, Figure 7, within the ejector is essentially governed by the entrainment of fluid due to mixing along the primary boundary and the recompressionshock system resulting from the local interaction between the primary stream and the shroud wall. The recompression system establishes the level of mechanical^nergy required for the entrained fluid topass through the recompression zone. Thus, the requirement that W = 0 uniquely establishes the value of P 0 s and correspondingly the flow field to the terminus of the recompression zone. Although variations in Pa do not influence P 0s until the "break-off point is reached, such variations do influence the shroud-wall pressure distribution. The Pa-independent portion of the flow regime is sustained until the value of Pa is sufficiently large so that the recompression mechanism is modified. When this occurs, the primary stream "breaks away" from the wall and the system then operates in the Pa-dependent mode. For this mode, the primary stream is recompressed to the

42

ambient pressure at the shroud exit, and the condition that W = 0 is maintained by a combined mechanism of entrainment and backflow within the ejector. 1.2.2 "Small" Secondary Flows For P a -independcnt operation, this regime effectively spans that part of the operating characteristics where transition of the governing flow phenomena is from an entrainment-recompression-shock mechanism to essentially an inviscid-interaction mechanism between two-streams that are distinct. This transition occurs when the recompression-shock mechanism can no longer be sustained at the shroud wall. As would be expected, the extent of this regime and point of transition are strongly dependent on the ejector geometry. The flow phenomena within this regime are essentially the same as described for the zero-flow case with the following exceptions. In this case, the secondary stream has a non-zero, though small, component of velocity in the axial direction. This component of velocity has a modifying effect on: (i) the primary stream, (ii) the entrainment at the primary-secondary boundary, and (iii) the recompression-shock mechanism at the shroud wall. 1.2.3 "Moderate to High" Secondary Flows This regime is characterized by the primary and secondary streams remaining essentially distinct although mixing locally along their mutual boundary. In contrast to the other flow regimes, the secondary stream is no longer "sheltered" by the primary stream and as a consequence must interact with the primary stream so that the operating conditions, e.g., (Pr- S iP a ). imposed on the ejector are satisfied. The interaction can satisfy the prescribed operating conditions in either of two ways. In one case, Figure 8(a), the secondary stream is "choked" within the shroud as a result of interacting with the primary stream thus establishing the mass-flow characteristics of the system. Adjustment to the prescribed ambient pressure level is then made downstream of the "choking" point without effecting the mass-flow characteristics. In the other case. Figure 8(b), the secondary-primary streams interact so that "choking" does not occur within the shroud but rather the prescribed ambient-pressure conditions are satisfied by the secondary stream at the shroud exit. The mass-flow characteristics are then determined as a consequence of this interaction. 1.3 Recompression Within Ejector Systems

As an example of the recompression phenomena, consider two ejector systems with shrouds of significantly different lengths that are operating at the same values of (W, P 0 s ) and discharging into a region where (P a ) is very small. For the long-shroud ejector, the mixing between the streams will be nearly complete and the flow at the exit will be approximately uniform and at a supersonic Mach number. In contrast, the mixing between the streams in the short-shroud ejector will necessarily be incomplete and the two streams will remain essentially distinct throughout the ejector. Consequently, the flow at the exit plane will be highly non-uniform and both streams will discharge supersonically. If P a is increased, the flow will initially recompress external to the shroud; however, with further increases in P a a recompression shock system will move inside the ejector shroud and be located such that the boundary condition imposed at the exit by the ambient pressure level will be satisfied. The recompression in the ejector with the longer shroud is essentially accomplished by a normal shock in the mixed stream. In the short-shroud ejector, the same recompression must be accomplished by a combined mechanism consisting of a normal shock in the secondary flow and an oblique shock in the primary stream. Clearly, the latter mechanism is less effective in recompressing the flow than the former. With further increases in P a , the recompression shock system moves further upstream in the ejector shroud. This movement in the short-shroud ejector system can only continue, without influencing the mass-flow characteristics, until the recompression shock in the secondary stream is located at the secondary stream's "choking" point. This situation essentially corresponds to a reversible subsonic recompression of the secondary stream to the ambient pressure at the shroud exit; any increase in P a above this value, the "break-off point, will necessarily result in a reduction of the mass-flow rate and the secondary stream becoming subsonic throughout the duct. The ejector system then operates in the ambient-pressure dependent regime. On the other hand, the long-shroud ejector system can recompress to higher values of P a as a result of the more complete mixing between the two streams. Eventually,however, increasing P a will move the recompression shock system upstream and out of the well-mixed flow and into the region where the two streams are essentially distinct as was the case for the short-shroud ejector. A part of the recompression now takes place at this location with final recompression to the ambient pressure level occurring in the remainder of the ejector shroud. Again the "break-off point occurs at the value of P a corresponding to the recompression shock being located at the "choking" point of the secondary stream. As for the short-shroud ejector, this value of P a defines the "break-off point since an increase in the ambient-pressure ratio above this value would result in a readjustment of the system's mass-flow characteristics.

43

The qualitative aspects of this recompression phenomena* for these systems are depicted in Figures 9(a) and 9(b); the influence of the recompression on the location of the "break-off curve is shown in Figure 9(c). It should be noted from these figures, that in the flow regime where the mass-flow_characteristics are unaffected by variations in the ambient pressure ratio, that the shroud wall pressure distribution P w can have significant variations with P a . In contrast, a unique shroud wall pressure distribution corresponds to each set of values of (W, P u s ,P a ) in the P a -dependent flow regime. In thrust augmentation applications, the ejector shroud-wall pressure distribution must be determined in addition to the "pumping" characteristics. The strong linking between the shroud-wall pressure distribution and the recompression mechanism necessitates that this mechanism bo included as an integral part of the performance analysis of ejector systems. Due to the complexity of the flow phenomena, this is a difficult, if not impossible, task for ejectors with long nonconstant-area shrouds. Fortunately, the analysis of short-shroud ejector systems, which are more practical for this particular application, is tenable. 1.4 Methodology of Ejector System Performance Analysis

A single-model approach to the analysis of ejector systems is not possible because of the dependence of the governing flow mechanisms on the system geometry and the operating conditions. Instead, their analysis must be based on a multiple-component flow model which adequately identifies and describes the predominant flow mechanisms within the various regimes. Although posing no conceptual problems, it is not known a priori when these regimes occur, as well as when transition between regimes occurs; in fact, these factors must be included as an integral part of the overall ejector analysis. The flow regimes (Sections 1.2.1-3) are modeled according to their respective governing flow mechanisms. In principle, these models are then applied, subject to the imposed boundary conditions and the applicable internal flow solution criteria, to determine the detailed ejector performance. In practice, a direct method of analysis for specified operating conditions is impractical for the following reasons: (i) The flow regime corresponding to the imposed operating conditions is not known a priori and, hence, must be determined. (ii) The applicable solution criteria and means for satisfying the imposed boundary conditions must also be determined. As a result, the performance analysis is based on an indirect method which simply consists of evaluating a system's performance in the various regimes, subject to the applicable solution criteria. This approach, in essence, establishes the overall ejector performance surfaces for a given system. Specific operating conditions are then located on this surface. The ejector flow model, as specialized for each of the flow regimes, and it's implementation will now be considered.

2.

"ZERO" FLOW REGIME

For this regime, the flow mechanism consists of: the flow entrained along the primary stream boundary due to the viscous mixing between the stream and an essentially quiescent fluid, the interaction of the "nearly" inviscid supersonic primary stream with the shroud wall, and their interdependence. Korst, Chow, et a l . 1 0 " 1 5 , have studied this problem in detail in the course of their basic investigation of separated flow problems. Carriere, Sirieix, Delery, and H a r d y 1 6 - 2 3 have considered modifications to the basic analysis proposed by Korst, et al., in an attempt to improve the agreement between theory and experiment for the axisymmetric case. Their work, as well as the work of others 2 5 , 2 7 , has generally been concerned with modifying the way in which the recompression of the entrained fluid is treated. This is a logical approach to take since the recompression mechanism is a predominant factor that is not well understood. In general, however, these investigations have not resulted in significant changes in the flow model originally proposed, but rather, have resulted in modifications to the "recompression criterion" and some details in the method of analysis. A discussion of current approaches and analyses related to this problem is given by Korst 15 . 2.1 Flow Model

The general flow situation and applicable notation is shown in Figure 10(a). The axisymmetric supersonic primary stream is assumed to expand into a region at constant pressure P-,. This constant pressure region is assumed to exist up to the point where the jet impinges on the shroud wall; at this point, an oblique shock wave exists so that the boundary condition imposed by the local wall slope is satisfied. This flow field is defined as the "corresponding inviscid jet", (Ref. 13). * An analogy, which is more apparent for the short-shroud ejector, can be drawn between the behavior of an ejector system and a converging-diverging nozzle operated under varying back-pressure conditions.

44 The entrainment as a result of the mixing between the primary stream and the quiescent fluid along their mutual boundary is assumed to take place at constant pressure. For the present analysis, the mixing component 1 3 , is assumed to be represented by the two-dimensional turbulent mixing of a uniform stream and a quiescent fluid. On this basis, the mixing region is considered to be defined by the flow conditions along the jet boundary and within an intrinsic coordinate system which is displaced relative to the jet boundary. The mixing region, thus defined, is then localized relative to the "corresponding inviscid jet boundary" by superimposing the mixing region on the jet boundary, in a two-dimensional sense, while satisfying the integral continuity and momentum relationships. The linking between the "corresponding inviscid j e t " and the mixing region is accomplished by the recompression mechanism and the conservation of mass and energy within the "wake" region. The "recompression criterion" identifies streamlines within the mixing region which have sufficient mechanical energy to recompress to the high-pressure region downstream of the jet-wall interaction. The implementation of this flow model as it applies to the analysis of ejector systems will now be discussed. 2.2 Primary Flow Field The primary flow field is analyzed by the Method of Characteristics for irrotational axisymmetric flow. The primary nozzle flow conditions are assumed to be known along the_initial left-running characteristic emanating from the nozzle corner ( X ] p , R i p ) , and the flow expands to a pressure, P-,, which is constant along the jet boundary. These conditions are then sufficient to determine the resulting flow field. Since the general interior primary flow-field analysis is applicable here, as well as for the other flow regimes, the flow-field analysis will not be discussed in detail here but rather in Appendix II. The only specializing condition for this flow regime is that the jet boundary is maintained at constant pressure. 2.3 Mixing Component The "restricted" two-dimensional constant-pressure turbulent jet mixing theory of Korst, et al., is the basis for the analysis of the mixing component. Since a general treatment of single-stream and two-stream mixing* can be given, the general aspects of this analysis are given in Appendix III while the conditions specifically applicable to this flow regime follow. For this case, Figure 10(b), the dimensionless velocity profile, <p, within the mixing zone is given by V " rr U P id + erf(o.y/x)] , (2.1)

where (x,y) are the intrinsic coordinates and a, is the single-stream mixing parameter commonly evaluated by the empirical correlation: a, = 12 + 2.76M p . (2.2)

The mass flow per unit depth between any two streamlines, 1 and 2, within the mixing zone is W12 = J ^ p U d y .

(2.3)

One streamline with physical significance can immediately be identified within the mixing region. This is the "jet-boundary streamline", j , that separates the flow originally in the jet from the flow entrained due to the mixing see Appendix III. The entrained flow then is given by: We =

r
[
j
-oo -oo

pUdy .

(2.4)

Thus, the mass flow rate between a streamline, d , and the j-streamline is given by

wdj =
Note that Wdj W \ 0 when y- ^ y d

fVj pU d y - J " y d p U d y
oo oo

(2.5)

* The former is applicable to the problem at hand while the latter is used in the "moderate to high" flow regime analysis.

45 Similar relationships 13 can also be written for the energy transfer within the mixing zone for the non-isoenergetic case, i.e., when T b # Trj p . However, this significantly complicates the analysis of the problem under consideration and experience has shown 1 3 , 1 4 that over a relatively wide range of temperatures, T b = Trj p , the influence on the theoretical "base-pressure ratio" is small. Consequently, only the isoenergetic case, Tj, = T Q P , will be considered for this regime. If the overall wake region defined by the "corresponding inviscid jet" and the shroud wall is now considered, the conservation of mass within this region requires that the net flow of the entrained fluid out of this region must be zero. If the d-streamline* has just sufficient mechanical energy to "escape" from the wake but yj = y: , the problem then becomes one of finding the value of P-, such that the d-streamline coincides with the j-streamline. 2.4 Recompression Criteria

The j-streamline can be identified solely on the basis of applying the conservation equations to the mixing zone; however, the determination of the d-streamline can only be accomplished by linking the mixing phenomena and the "corresponding inviscid jet". This is done by means of the recompression criterion. 2.4.1 Korst's Recompression Criterion The recompression criterion of Korst, et al. 1 3 identifies the d-streamline as the streamline which possesses just sufficient mechanical energy to recompresst from the wake pressure, P-, , to the high-pressure region downstream of the shock, P ^ k . The recompression pressure rise for the d-streamline, based on this recompression criterion is found from the oblique shock relations for the local conditions at the impingement point, Figure 10(c). The turning angle is given by: 6 = -(flp - 0 W ) (2.6)

and the jet surface Mach number, Mp , is known for the assumed value of Pj, . Hence the d-streamline pressure rise is found from:

EsM = ^0d = f(5,M D )** .


Pb Pd
P

(2.7)

The d-streamline "Mach s t a r " t t is then given by:

MJ

= (Jf4-\) U 2 [ 1 - (Pod/Pd)^-^ 7 "] ll2

(2-8)

For the isoenergetic case, the d-streamline velocity ratio is:

The solution value of Pj, is then found when 0 < for e arbitrarily small. In practice, the solution value of P b is found by assuming a value (P b )- and then computing the difference
(^d
_

l^-^jl

< e.

(2.10)

^j)i

If

(Vd-Vj)j t l 0 then the next value ( P b ) i + i is correspondingly assumed to be: (P b )i+1 (Pb)i

(2.11)

(2'12>

Then convergence to the solution is both well behaved and rapid. * t Defined as the "discriminating streamline". This compression, though irreversible and diabatic, produces the same results as an isentropic compression from the static pressure, P b , to what would be the stagnation pressure, P shk , of the d-streamline.

** Convenient graphical or functional forms for numerical calculations are found in Reference 28. t t The variable M* = U/C* has been introduced and is used throughout the analysis because its finite range, 0 < M* < [(y + l)l(y - I))" 3 is convenient for the computer analysis.

46 It should be noted that the intermediate results obtained in the course of the zero-flow solution can be interpreted as solutions for "mass bleed at negligible velocities" into or out of the wake region 13 . For this case, the mass flow is evaluated from (2.5) or expressed in more conventional form (from Appendix III): 2 R w X m (A/A*) (yp.MJp) , . . , . ,
W = -r { l i *?;) I I ( I ? H )

... . , ,
(2.13)

a, R l p R l p ( A / A * ) ( y p , M * )

'J

Although the overall recompression pressure rise is reasonably well represented by the pressure rise corresponding to the oblique shock at the impingement point, it was well known at the time that the "discriminating streamline" does not recompress to this level. Rather, it recompresses to an intermediate value between P b and P s ^ . This situation is shown in Figure 10(d). Nash 2 3 defined a recompression coefficient for the two-dimensional case in an effort to better correlate the theoretical analysis with the experimental data; this correlation is not sufficiently universal and as a result has not been well accepted. Page, et al. 2 7 has proposed another recompression correlation based on the discriminating streamline velocity ratio and an "effective" entrainment mixing length. This correlation is shown to be applicable over a wide range of experimental conditions for plane two-dimensional supersonic flow. This approach yields results which are similar to those obtained by investigators at ONERA*. Since much of the work at ONERA was directed toward the analysis of axisymmetric ejector systems operating in the zero and "small" flow regimes, it will now be discussed in some detail. 2.4.2 The Recompression Criterion of Carriere, Sirieix, et al. An empirical relationship called "the critical angle of reattachment" that a stream must satisfy when reattaching to a wall has been deduced from a series of experiments 1 6 " 1 9 . This relationship was established for a uniform stream entraining a quiescent fluid at constant pressure and then subsequently reattaching to a wall to form a wake. Figure 11(a). This relationship is shown graphically in Figure 12; it has the functional form *o " >MMi) + * (2-14)

Carriere and Sirieix have shown 1 6 , 1 7 that the influence on the base pressure of: (1) the stream's initial boundary layer momentum thickness (5 ) and (2) mass addition (q) to the wake with a finite momentum (i) can be expressed in terms of a dimensionless generalized "bleed coefficient"**, C q . Where C q is defined 1 7 ' 2 0 as
(

.,

p,U,x

p,U2x

'

+ 5**

(2.15)

The above factors are then considered as perturbations to the case where they are absent. The reattachment angle for the perturbed case is given by *(MC <1 ) where 1 6 * 2 0 a,
(W

*0(M,) + O W 3 C q ) M l C q ,

(2.16)

1M2(1 - tf2) 1 ] l / 2
2

3Cq)Mi

= - - J - e x p ^

) '

'<(1 V ^

(2.17)

and the "limiting streamline"tt. / . is found from P"' P -tfdT,


J-ooP.

Oi = --Ca+
X 4

P + P J-oc Pi

ififl - sfi) drj .

(2.18)

It should be noted, in passing, that (2.18) can be expressed in terms of the j-streamline as * Office National d'Etudes et Recherches Aerospatiales, Chatillon-sous-Bagneux (Seine), France. t In their notation, used throughout this section, \p0 is the reattachment angle at the wall for this case and M, is the Mach number along the jet boundary. \jtQ is given graphically in Reference 21; a curve fit of these data which is more convenient for numerical computations is given by: i/r0 = [2.33 + 12.84M, - (85.46M2 - 381.2M, + 5 2 6 ) " T . ** This concept is similar in nature to the "equivalent bleed concept" (Ref.13). t t Similar to the "discriminating streamline" in Korst's analysis.

47

Z- sfi dTJ =

- - 1 CQ

(2.19)

in which form the analogy to "bleed" is obvious. The solution procedure to determine P b given: (1) the flow geometry, (2) the initial flow conditions, and (3) C q = 0*, proceeds in the following fashion. If a value of P b is assumed, then M, can be calculated from the isentropic relations, and the inviscid flow field can be constructed by the Method of Characteristics (Appendix II) subject to the condition of constant pressure along the boundary. Corresponding to the value of M, along the boundary, values of a., n-,, sp,, r) m , and the mixing integrals (Appendix III) - can be determined. Using (2.19) the dimensionless coordinate, )?/, locating the /-streamline in the intrinsic coordinate system can be found. Assuming a displacement of the /-streamline normal to the local inviscid boundary, the location of the /-streamline, Figure 11(b), in the inviscid flow field coordinates (X, R) is found from Xp + (77m - 17/) sin 8. (2.20a)

R, =

P + (Vm ~ Vl) cos

(2.20b)

The next step is to find where the /-streamline intersects the wall; corresponding to this location, the impingement angle of the /-streamline relative to the wall is given by

= -(*-*)

(2.21)

where 0 p is the inviscid boundary flow angle corresponding to the /-streamline impingement point and 0 W is the corresponding local wall angle. From the angular reattachment criterion, (2.16), the value of \b can be found. The question is - does i/-/ = i/-? If nott, then another value of P b is assumed and the foregoing calculations repeated until 0 < M i ) n - (i//)nl < e . (2.22)

Typically if l(i/'/) n (i//)nI ^ 0 , the next value (P b ) n + j is chosen correspondingly such that (P b ) n +i ^ (Pb'n

3.

SMALL" SECONDARY FLOW REGIME

The secondary flow, although small in this regime, has a significant modifying effect on the primary and secondary streams within the ejector shroud. The circumstances are still such that an oblique shock wave forms in the primary flow field near the shroud wall as a result of the interaction between the primary stream and the wall. Consequently, the flow mechanism retains essentially the same characteristics as for the "zero" flow regime. The analyses for the "zero" and "small" flow regimes differ only in the determination of the primary flow fields and their respective solution criterion. In contrast to the method of analysis to be discussed here, investigators at ONERA 2 0 " 2 2 have treated this flow regime by the methods outlined in Section 2. Their method does not consider the interaction between the streams but rather accounts for the effects of the finite mass and momentum addition of the secondary stream within the framework of their recompression criterion (see Section 2.4.2). The performance characteristics predicted for this regime by this method of analysis is shown 2 1 , 2 2 to be in excellent agreement with experiment. 3.1 Flow Model

At the point of confluence of the primary and secondary streams, Figure 13(a), the primary stream expands to the prevailing secondary stream static pressure. After this point, the primary and secondary streams must co-exist within the available flow area while satisfying the condition of continuity of pressure at their mutual boundary. As the primary stream expands in the shroud, the interaction between the streams accelerates the secondary flow resulting in a decrease in the static pressure along the streams' boundary. However, in this regime, the primary stream's expansion is such that the compression waves within the primary flow field are insufficient to turn the primary boundary before it reaches the proximity of the shroud wall; as a consequence, the primary stream has a finite streamline angle relative to the wall. An oblique shock recompression system then occurs due to the interaction between the primary stream and the shroud wall. For the "zero" flow regime, Cq q,i*0. 6** since q, i = 0 . In the "small" flow regime, Cq is evaluated according to (2.15) for

t These cases could be interpreted as non-zero flow solutions; however in a programmed numerical approach, it is more convenient to follow the overall calculation sequence described.

48 In addition to the interaction between the primary and secondary streams, they are also mixing along their mutual boundary and in effect, the primary stream entrains the secondary flow. In analogy to the "zero" flow regime, the secondary flow entrained by the primary stream can only pass through the recompression shock if it possesses sufficient mechanical energy to recomprcss from the local static pressure to the high-pressure region downstream of the oblique shock. The solution criterion for this regime is then based on satisfying the requirement of conservation of mass for the secondary flow, i.e., the entering secondary flow must equal the entrained flow which "escapes". 3.2 Primary-Secondary Flow Fields

The mixing between the primary and secondary streams is assumed to have a negligible effect on their interaction between their point of confluence and the proximity of the wall. These flow fields, "the corresponding inviscid flow fields", are then established on the basis of the inviscid interaction between the streams. The primary flow field is analyzed by the Method of Characteristics for irrotational axisymmetric flow while the secondary stream is assumed to be isentropic and one-dimensional in nature. The flow fields are linked by the condition that the two streams must coexist within the available flow area while at the same time satisfying continuity of pressure at their mutual boundary. This flow field analysis is the same as for the "moderate to high" flow regime where the two streams remain distinct; a detailed discussion of this analysis, along with the pertinent relations, is given in Appendix II and Section 4.2. Specialized to this regime, however, is the calculation of the primary stream's boundary in the proximity of the wall and the resulting recompression shock system. As the inviscid flow field calculations proceed downstream and the primary boundary approaches the wall, the mixing region is then assumed to reduce significantly the effects of the inviscid interaction between the streams and as a result, the primary boundary is assumed to smoothly approach and interact with the wall in a region of approximately constant local static pressure. The inviscid flow field calculations are made in the following way. For a given value of P j s and an assumed value of M l s , the flow field calculations (Section 4.2) are made, starting with the initial flow conditions, by a step-by-step procedure in the downstream direction until the primary boundary is in the proximity of the wall. Due to the influence of the mixing region, further interaction between the streams is assumed not to occur. The conditions for the oblique shock system, Figure 13(b), are then given by the local primary flow conditions at the boundary. The recompression pressure rise can be found from ^ P where the local turning angle, 5 , is 6 = -(0p-0
w

= f(6,M p )* ,

(3.1)

(3.2)

and M p and P p are respectively the local Mach number and static pressure at the primary boundary. As for the "zero" flow regime, the recompression pressure rise impressed by the primary stream is used to discriminate between the entrained flow having sufficient mechanical energy to pass through this region and that which has not. The secondary mass-flow ratio, for which the inviscid flow field calculations are made, is evaluated at the pointt of confluence of the streams for the given value of P j s and the assumed value of M | s . The non-dimensional secondary-to-primary mass flow ratio is given by _
w.

Pi s A i , F(7Mi,,) = -II _Li Ws' " '


p

(3.3)

lp

l p F(Tp.Mip)

where the conditions at the primary nozzle exit are known and F(7,M*) is the "weight flow function". 3.3 Mixing Component

The estimation of the flow entrained by the primary stream is based, in direct analogy to the "zero" flow regime, on the simplifying assumption that the actual process can be represented approximately by the entrainment of a quiescent fluid by an equivalent two-dimensional stream at constant pressure. In this case, the equivalent stream is evaluated on the basis of the local primary flow conditions near the wall and the corresponding mixing length, x m , up to this point. The two-dimensional constant pressure mixing analysis, Appendix III, is used to determine the entrained flow (see Section 2.2.2) which has sufficient mechanical energy to penetrate the recompression pressure rise. The flow "escaping" is given by * Functional or graphical forms are given in Reference 28.

49

wI
or, in non-dimensional form, Wd

2TTR

Vj

pU

dy

(3.4)

"

x m (A/A*) (7p.M7 p ) R s T 0s ~p*"; T.77* "l Rip Rip (A/A*) ( 7 p , M j ) R p T o p


A,

2 Rp
D.

D.

II.(T}j) - I.(l} d )l ,

(3.5)

where 1,(7?) denotes the mixing integral. Appendix III. The j-streamline separates the primary and the entrained flows, and the d-streamline is determined from the recompression criterion
P

shk

0d ,

"PT
*

" TT
M

P d

M*.

'

(3.6)

The discriminating streamline velocity ratio for the isoenergetic approximation is


d

(3.7)

and Md is found from (3.6) by M* Tp-1 3.4 Solution Criterion (P 0 d/ p d) (Tp ~ 1) (3.8)

1-

The secondary mass flow ratio corresponding to a given value of P . s , in this regime, is established by applying the continuity equation to the secondary flow region. Thus for a solution, the continuity requirement becomes W- = Wdj . (3.9)

The solution sequence proceeds in the following manner. For a given value of P . s , a value (M* s ) n is assumed; the inviscid mass flow ratio, (Wj)n and the corresponding inviscid flow fields are then calculated. Corresponding to the inviscid flow field analysis, the entrained mass flow ratio, (W d ;) n is then determined from the mixing analysis. If (W d j) n ^ (Wj)n , the assumed value of M l s is varied according to (Mi s ) n + 1 ^ (M* s ) n . This calculation sequence is continued until 0 < KWdj) - (W-)n| < e . (3.10)

When this criterion is satisfied, the solution values of (M- s ,P [ s ) and (W,Pi s ) are then known. The corresponding stagnation pressure ratio is then given by Pos = P l s y ( 7 P M * $ ) , where (P 0 /P)(7,M ) is the two-dimensional isentropic pressure ratio function. (3.11)

4.

"MODERATE TO HIGH" SECONDARY FLOW REGIME

In contrast to the other flow regimes where the mixing component constituted an integral part of the analysis, the mixing between the streams in this regime is considered to be a secondary effect in the sense of boundary layer concepts. As a result, the performance characteristics are essentially established by the mutual interaction between the now distinct primary and secondary streams, the "corresponding inviscid flow fields". After the inviscid flow fields have been established as a result of the mutual interaction between the streams, the effects of mixing are approximately accounted for by locally superimposing the mixing region on the inviscid boundary between the streams. The net effect of the mixing is usually reflected as an increase in the secondary stream's mass flow over that given by the inviscid-interaction solution. This technique assumes, of course, that the mixing between the streams does not significantly modify the inviscid flow fields. This method of analysis*, which was proposed in References 1, 2 and 3, gives solutions for the ejector performance characteristics which are in excellent agreement with experiment. As a result, this method has been well accepted for analyzing the performance characteristics of ejector systems. In addition, the same basic approach * A similar method of analysis was first proposed by Fabri et al. 7 ' 8 for analyzing ejectors with constant area shrouds. Their analysis is based on a one-dimensional technique where the streams remain distinct; however, their model does not satisfy the condition of continuity of pressure across the streams' boundary. In addition, the effects of mixing between the streams was neglected.

50 has been followed in a more generalized treatment recently given by Peters 2 9 ; his analysis includes chemistry effects, the simultaneous build-up of the mixing region and the inviscid flow fields, and a more generalized solution criterion for the "choked" flow regime. 4.1 Flow Model The basic flow m o d e l 1 ' 2 , 3 consists of two parts, viz, the corresponding inviscid flow fields and the two-stream mixing component. The primary and secondary streams. Figure 14(a), begin to interact at their point of confluence because of the requirement that both streams must co-exist in the available flow area while at the same time satisfying the requirement that the local static pressure must be equal at the streams' boundary. This interaction continues downstream and the secondary flow field can have the following distinguishing characteristics for an arbitrarily assumed pair of values (Mi s ,P I s ) within this flow regime. These characteristics are: (i) The secondary stream remains "unchoked" while the secondary flow area goes through a minimum value, (ii) The secondary stream "chokes" but at the "choking" point the secondary flow area is not a minimum, (iii) The secondary stream "chokes" and the secondary flow area is simultaneously a minimum. These situations all represent physically realizable operating conditions and the occurrence of any particular one depends solely on the imposed boundary conditions. The "solution criteria" that distinguishes between these potential flow conditions constitutes an important part of the flow model. For the first case to occur, the boundary conditions imposed by the shroud wall geometry and ambient pressure must be such that the secondary stream (or mixed stream in the long-shroud ejector) satisfies the boundary condition imposed at the ejector exit by the ambient pressure level. In the second case, the ejector shroud length must be such that the "choking" point and the shroud exit coincide and in addition, the ambient pressure must be less than or equal to the secondary pressure at this point. The secondary stream can then adjust to the ambient flow conditions by a supersonic expansion external to the shroud. The third case is of more general interest since it constitutes the basis for analyzing the majority of the P a -independent flow regime and for locating the "break-ofF* point in this regime. For this case to occur, the boundary conditions must be such that the secondary flow after "choking" in the shroud can be discharged to the ambient pressure level. After the appropriate inviscid flow field has been established for values of (Mi s ,P| s ,P a ), the influence of the mixing between the two-streams is evaluated by locally superimposing the two-dimensional mixing region on the inviscid boundary on a "quasi-constant pressure" basis. The influence of the mixing region is interpreted in the boundary layer sense as having an equivalent displacement thickness which is evaluated at the "choked" and/or minimum secondary flow area. The resulting change in the secondary flow is then evaluated at this point based on the local flow conditions and the change in available flow area due to the mixing region's displacement thickness. Inherent in this method is the tacit assumption that the increase in mass flow due to mixing will not significantly alter the inviscid flow field. 4.2 Inviscid Flow-Fields Analysis

The primary stream's internal flow field is analyzed by the Method of Characteristics, Appendix II, for irrotational axisymmetric flow. For this analysis, it is assumed that the primary flow field is known along the initial characteristic emanating from the nozzle lip, at ( X i p , R j p ) , to the nozzle axis. The primary stream is then assumed to expand*, through a centered expansion at this point, to match the initial secondary stream's static pressure. The secondary flowt is assumed to be isentropic and one-dimensional in nature. At the point of confluence of the streams, the secondary flow properties are uniform and specified by Mj s and P] s . The general conditions that must be satisfied at the streams' boundary are: (i) the local static pressure must be equal for both streams at their boundary, i.e., Pp = P s > (4-D

* Under certain circumstances, the primary stream must be recompressed by an oblique shock wave at this point. For this case, the compression is approximately taken into account when the initial characteristic is determined (see Appendix II). Hence, the flow field calculation technique is unaffected. t The assumption of uniform one-dimensional flow yields results which are in agreement with experiments for many cases where this assumption is only approximately satisfied. However, for ejector geometries with significant wall curvature and/or applications where the initial secondary flow is highly non-uniform, a more generalized treatment of the secondary flow is not only desirable but necessary. See, for example. Reference 29.

51 (ii) the flows must co-exist within the available area. These conditions provide the link between the Method of Characteristics analysis for the primary flow field and the isentropic one-dimensional analysis for the secondary flow field. For a given pair of values, (M] S , Pj s ), the inviscid secondary mass flow ratio at the initial section, Figure 14(a), is given by:
p

ls

ls

(7s. M ts)
s

W - = Ji
P A

* ,

(4.2)

1 P l p F(7p,M l p ) where F(7, M*) is the "weight flow function". If the primary nozzle is restricted to having either uniform parallel or conical outflow, the primary nozzle flow area, A i p , is given by: A l p = 27rR 2 p /(l -t-cosOip), where 0 l p = 0 for uniform and parallel outflow. is specified by M j p ( o r M ] p ) . (4.3)

At this area, the fully expanded primary nozzle Mach number

The secondary flow area, for shrouds with a small but finite radius of curvature, is assumed to be equal to the lateral surface area of the truncated cone whose generator is locally perpendicular to the shroud wall and extends from the wall to the primary boundary, i.e., segment a-a in Figure 14(b). The secondary flow area is then given by: As = TrtR2, - R 2 )/cos 0 W . (4.4)

The initial secondary flow area, A l s , is determined using (4.4) and the condition that the normal to the shroud wall, at the location (X W ,R W ), must pass through the point of confluence of the streams (Xjp.Rip). The condition for co-existence between the flows within the prescribed shroud area can be formulated for a general boundary point, Figure 14(c), by applying the continuity equation between_the initial secondary flow area, A [ s , and the general location where As is to be found subject to (4.1); hence, W ]s = Ws and for isentropic flow between these locations: p(7s.M*) = ^L^(7l,Mt$). (4.5)

Thus, for a known value of Aj , the local value of M* can be found from (4.5). However, to find As the secondary and primary flow fields must be linked. Consider two points (a) and (d). Figure 14(c), on the primary boundary corresponding to the right-running characteristics (n) and (n + 1). Assuming for the time being that the primary and secondary properties are known at Point (a) on the ( n ) t h R-R characteristic*, the conditions at Point (d) on (n + l ) t h R-R characteristic can be found in the following way. If the calculations are made along the (n + 1 )tl* R-R characteristic, then conditions at the interior point (c) are found from the characteristic relationships and the previously found data at Points (a) and (b) - see Appendix II for "field point" calculations. The data at (c) and the boundary conditions along (a-d) specified by (4.1 and 4.5) are then used in an iterative procedure to determine point (d). Assuming as a first approximation that the flow properties at (d) are simply the average values at (a) and (c), the characteristic relations can be used to find, to a first approximation, the coordinates (X p , R p ) of the point (d). The secondary flow area As aMhisjpoint is then found from (4.4). Corresponding to this flow area, M* is found from (4.5). Now since P s = P p , M p at point (d) is found for isentropic flow from: ^(7P,M*) = r o

(p/p 0 H 7 s X) (P/P0)(7s,Mts) P

<4-6-

where (P/P 0 )( 7 ,M*) is the isentropic pressure ratio function. With these values of (X p ,R p ,Mp), the local flow angle 0 p is then evaluated using the characteristic relations. By a typical method of successive approximations based on always using the previous approximation, (X p ,Rp,M p ,0 p )j , to evaluate the next (X p , R p ,M p ,0 p ) i + i , convergence to a stable set of values is rapid; the iteration is terminated when: 0 < | ( 0 p ) i + 1 - (0 p )-| < e . (4-7)

* Right-running or Family 1 Characteristic.

52 While calculating the downstream flow fields for a pair of values (M>-s,PIs), a boundary point calculation could occur where the secondary stream's area ratio, according to (4.5), would be less than one, i.e., (A/A )s < 1 . This of course is not physically possible since (A/A*)s > 1 is required. In this case, an iteration must be made within the flow fields to locate the point where 1 < (A/A*), < (1 + e ) . This calculation then determines the "choking" point corresponding to (Mi s ,P. s ); however, whether or not this represents a valid inviscid solution depends upon the boundary conditions imposed on the ejector system. Another important characteristic of the secondary flow field to be observed during the flow-fields calculation is whether or not the secondary flow area ratio, (A/A*) s , has a minimum value and if so, where does it occur. Physically, the secondary flow area goes through a minimum value because the strength of the compression waves within the primary flow field is sufficient to nullify the effects of the secondary stream's expansion on the primary stream's boundary. After the location where the minimum secondary flow area occurs, the secondary flow area ratio begins to increase. Depending on the flow conditions at the minimum point and the ambient pressure level, the secondary stream will start to diffuse or continue to expand, cf., the performance of a converging-diverging nozzle after the throat. To detect whether or not the secondary has a minimum value, the following procedure is carried-out concurrently with the inviscid flow field calculations. For a minimum value of (A/A*)s ,
(4.9)

(4.8)

dx at X , and

T* IA-' *
Rn
i
_ t a n

(4.10)

for X ^ X m i n . Condition (4.10) can be expressed in terms of the local boundary and shroud wall properties as tan (0W)
*-*p

Rw + H K ~ R^)RwCos20w

(4.11)

where R^, = d 2 R w /dX 2 . For a shroud wall profile of the form: Rw = AX2 + BX + C , Equation (4.11) becomes: tan (0 w ) m a tan 0 r
Rr-

(4.12)

Rw + (R , - R2)A cos 2 0 w

(4.13)

As the inviscid flow fields are established step-by-step in the downstream direction, the value of (0 w ) m a is found from (4.11) or (4.13) corresponding to the local values of (R p ,0 p ,R w ). For the secondary stream to have a minimum area, the following condition between the local shroud wall angle, 0W , and (0 w ) m a must be satisfied. I*w - (ew)mal 0
(4.14)

when X ^ X min The minimum area criterion, (4.14), is checked at each downstream location as the flow fields are established; when the difference [0W (0 w ) ma l changes sign, an iteration is then made within the flow field to determine the location where X = X m i n . The solution criteria for the inviscid analysis will now be discussed for the possible modes of ejector operation within this regime. 4.3 Solution Criteria The solution criteria are based on the internal conditions that the secondary stream must satisfy while undergoing changes in its flow area, Mach number, and static pressure as a result of interacting with the primary stream. These conditions are enumerated for the one-dimensional treatment of the secondary flow.

53 4.3.1 "Unchoked" Operation If the shroud is sufficiently long, the secondary stream, Figure 15(a), will accelerate and attain a minimum flow area within the shroud. After the minimum flow area, the secondary flow must then diffuse to the ambient pressure level at the shroud exit. These conditions are; (A/A*)s = minimum and'for * < X min , Then for Xmin < X < Xe , such that at X = Xe , (P s ) e = (P a ) . M* < QAt)m,n < 1 . Mfs < M" < 1 .

If the mixing between the secondary and primary stream is small, the flow field for, X m - n < X < X e , can be determined by the inviscid-interaction analysis described in the previous section. Since these calculations are normally for given values of (M*s, P Is ), the flow field calculations are made to the shroud exit where the^value of (P s ) e is then determined; for this to be a valid operating point (the indirect solution method), Pa = (P s ) e is required. For short-shroud ejector systems operating in this regime, Figure 15(b), it is possible that a minimum area is not attained within the shroud. If this is the case, the flow fields are established to the shroud exit where (P a ) = (P s ) e must be satisfied. When the shroud configuration is such that considerable mixing takes place between the streams for X > X m j n , completely mixed flow at the shroud exit can be assumed, and an overall one-dimensional analysis of the system made if the shroud has a constant cross-sectional area. However, for nonconstant area shrouds, this problem is difficult since the shroud-wall pressure distribution must be considered in such an analysis. 4.3.2 "Choked" Operation For the secondary stream to be "choked" within the shroud, Figure 16(a), the secondary flow must simultaneously satisfy the condition that: (A/A*)s = minimum and (Mf) min = 1 . (4.15)

These conditions then define the "choking" point as well as its location. For this mode of operation, Figure 16(b), the ambient pressure level must be such that the secondary flowcan be recompressed or expanded, either internally or externally, downstream of Xm-n to the prevailing value of Pa . The adjustment of the secondary flow to the ambient pressure can be accomplished by: 1. Subsonic diffusion to the exit, i.e., (M*) < 1 2. for X min < X < Xe and (P s ) e = Pa . (4.16a)

Supersonic expansion to the exit with the final adjustment to the ambient conditions external to the shroud. (M*) > 1 for X mjn < X < Xe . (4.16b) Supersonic-shock-subsonic flow within the ejector shroud to satisfy the ambient pressure conditions.
i* (M*) > 1

3.

for for

Xmin < X < X s h k , X shk < X < Xe and

and (P s ) e = Pa . (4.16c)

(Mf) < 1 4.

Internal flow conditions similar to (1), (2), or (3) coupled with significant mixing occurring between the primary and secondary streams so that the secondary stream essentially loses its identity.

For flow conditions (1), (2), and (3), above, the streams are assumed to remain essentially distinct within the ejector shroud and as a result, the flow fields can be analyzed as a continuation of the inviscid interaction between the streams (Section 4.4). Due to the complexity of the flow within the ejector, the analysis for the last case, item 4, must necessarily be based on typical one-dimensional methods and simplifying assumptions concerning the shroudwall geometry and the mixed-flow conditions at the shroud exit. A significant problem in applying the "choked" flow criterion is the determination of the conditions, (P<S,M*S), for which (A/A*)s is a minimum and M* is simultaneously equal to one. Since this point is a saddle point in the mathematical sense, its determination can, at best, be only approximate as a result of the necessary trade-off between computation time and an acceptable degree of accuracy in determining its location.

54 However, these approximations have a negligible effect on the determination of the inviscid mass flow characteristics and a more significant, although acceptable, effect on the determination of the shroud wall pressure distribution. The "choking" location within the shroud is determined for a given value of P i s by assuming an initial value of M*s and then calculating the corresponding inviscid flow fields until a point is found where in general either: 1.* (A/A*) s = minimum or It M* " ( 1 - e ) and and (Mf)min < ( 1 - e ) (4.17a) (4.17b)

(A/A*) * minimum.

For the first case, the value of M*s would be increased above its initially assumed value, now (M* s ) m a . Conversely in the second case, the value of M*s would be decreased below its initially assumed value, now (M] s ) c . The value of M*s is then varied according to this procedure until M*s is "bracketed", i.e., (M*s)ma < Mil < (M*s)c(4.18)

After the above condition is encountered, an interval halving technique, which always keeps M*s "bracketed", will rapidly establish a stable value for the mass flow characteristics, i.e., KM|s)ma - (M* s ) c | < e, KWj)ma - (W-)c| < e2 . (4.19a) (4.19b)

Corresponding to conditions (4.19), the inviscid mass flow characteristics are then essentially established; however even in this limited range of variation in M* s , the value of (M*) m i n can still be somewhat less than one. The assumption is then made that the inviscid flow fields corresponding to M*s = (M* s ) m a , after satisfying condition (4.19), are approximately equivalent to the flow fields for the true "choking" solution. The value of (M*) m i n is then set equal to: C<)min = (1 + ) (4-20)

and correspondingly, an improved estimate of the shroud pressure distribution for X, < X < X m - n is made. Since for this assumption (A s ) m - n = ( A * ) , then the improved estimate of the flow conditions at any interior point (n) is obtained from:

hr,<Mttl = ( ^ y After solving (4.21) for (M*) n , the improved estimate, the shroud wall pressure distribution is found from:

(4-2D

(-Vn = JT [T,.(M*)y J l7s,MtslPls .

(4.22)

For 6] sufficiently small, (4.19), corrections represented by (4.21) and (4.22) are also small; however, a more important aspect of this approximation is in establishing, although approximate, well-defined flow conditions at the "choking" location for use in the later "downstream" flow-field calculations. In summary, the inviscid flow-field calculations up to the "choking" location establishes: (i) the inviscid reduced mass-flow ratio Ws , (ii) the inviscid flow fields up to this location, and (iii) the conditions necessary for continuing the flow-field calculations downstream.

4.4

"Downstream" Flow-Field Analysis

The flow-field calculations downstream of the "choking" and/or minimum area location are restricted here to those cases where the streams remain essentially distinct throughout the ejector. The basic calculation technique is a continuation of the inviscid flow-fields analysis (Section 4.2); however in the "choked" case, the analysis must now include the possible ways in which the secondary stream can satisfy the boundary condition imposed at the ejector exit by the ambient pressure.

* This case corresponds to a possible "unchoked" solution. t This would be a possible solution only if this location coincides with the ejector exit and (P s ) e > Pa .

55 4.4.1 "Unchoked" Flow For the unchoked case, the secondary flow area has a minimum value but (Mf) m i n < I . The inviscidinteraction flow-field calculations are simply continued downstream from the minimum area location to the ejector exit. At the exit^the secondary stream static pressure ratio, (P s ) e , must then be equal to the prevailing ambient pressure ratio, P a , for this to be an inviscid solution. In the indirect method of solution, the initial values of (M*s, P[S) establish the corresponding solution values of (Wj and P a ). These conditions are illustrated in Figure 15(a). The only complications generally encountered in these calculations arise from compression waves coalescing within the primary flow field. This problem becomes more severe as the calculations proceed further downstream due to the local compression of the secondary stream, as well as the increasing strength of the compression waves in the primary flow field. 4.4.2 "Choked" Flow Between the "choking" location and the exit, the secondary stream, depending on the ambient pressure level, can be; (i) entirely subsonic, (ii) entirely supersonic, or (iii) supersonic-shock-subsonic. Each of these flow conditions can be used to define, in analogy to the converging-diverging nozzle, various "limiting" operating conditions for the ejector system. When the secondary flow is entirely subsonic to the exit, Figure 16(b), then P a = (P s ) e is required. For the values of (M*s, Pi s ), this flow condition thus defines the break-off point, i.e.,
(VBO = (Pi). ^ 1 > Mf > (Ms*)e . (4.23)

If the secondary flow is entirely supersonic to the exit, Figure 16(b), the "ideally expanded" secondary flow conditions, in analogy to the converging-diverging nozzle, occur when: (P a )iE = (P s )e and (M?) e > 1. (4.24)

Similarly, the degree to which the secondary flow can be "over-expanded" is determined for a normal shock standing in the secondary stream at the shroud exit. In the usual shock notation, the ambient pressure ratio for this condition is:
OVOE (Ps)e5*l7,.(Mf)el . (4.25a)

where P y / P x is the static pressure ratio function for a normal shock wave. Then for this operating condition, the ambient pressure ratio must be in the range: 0 < P a < (P a ) O E . (4.25b)

For (Pa)oE < Pa "^ (Pa^BO Figure 16(b), a shock system occurs inside the shroud and the secondary stream then goes supersonic-shock-subsonic to the exit. For the indirect method of analysis, the shock wave is assumed to occur in the secondary stream at various locations
^min < Xjhk * ^ e

i * within the shroud. Corresponding to each shock location, the values of (M S ,P S ) are known; consequently, the flow conditions in the secondary stream after the shock can be determined from the one-dimensional normal shock relations as:

(M*) y

= [(M s *) x l-'

(4.26a) (4.26b)

(P s )y = ( P s ) x ^ (7S,(MS*)X] "x (Af) y /(Af) x = ^[7S,(MS*)X] . H)y

(4.26c)

At the primary stream boundary, Figure 17(b), the values of the flow variables are determined from the condition that (P p )y = (Pj)y Then from the oblique shock relationships: (M*) y = f[7p,(M*)x,(Ps)y/(Ps)x]t (4.26d)

* x and y are locations respectively before and after the shock. t See footnote on next page.

56 and 6 - [(0 p ) x - (0p) y l = f[7p,(M*)x,(Ps)y/(Ps)xl* . (4.26e)

Thus the flow variables [X p , R p , M p , 0 p ] y , are determined for the primary stream after the shock system. The data determined from (4.26) are sufficient to continue the inviscid-interacting calculations to the ejector exit where P a = (P s ) e is required. The "downstream" flow field calculations for the various conditions described above follow, except as noted, the typical inviscid-interaction technique of Section 4.2. 4.5 Two-Stream Mixing Correction The effects of mixing along the boundary between the primary and secondary streams are approximately evaluated by locally superimposing the mixing region on the inviscid flow fields at the minimum secondary flow area. Although the mixing between the streams actually takes place along a non-constant pressure boundary, the assumption is made that the actual velocity profile at the minimum secondary flow area. Figure 18(b), can be approximated by the velocity profile which would result if two uniform streams at the local flow conditions, Figure 18(a), were hypothetically mixed at constant pressure through a distance equal to the actual mixing length along the non-constant pressure boundary. On this basis, the constant-pressure two-stream mixing analysis (Appendix III) is used to estimate the influence on the secondary flow due to entrainment by the primary stream. The effects of the entrainment of the secondary flow by the primary stream are interpreted, in the boundary layer sense, as being equivalent to a displacement in the secondary stream's boundary. If (6*) denotes a displacement of the secondary boundary corresponding to an increase in the secondary flow area. Figure 18(c), the change in secondary flow area at X = X m - n is approximately AAS = - 2 i r ( R p ) m i n 6 * or, introducing the mixing length and the similarity parameter, (4.27a)

=^ = , . ( 2 : / ) ,

,4.

An expression for (a II 5*/x m ) based on the two-stream constant-pressure mixing model is derived in Appendix III, and a graphical representation of this function is given in Figure AIII.3 for the isoenergetic case. For this change in the secondary flow area, the corresponding change in the mass-flow ratio is AW = -^2On Rip
(R

P)min RlP

(I

Vrm" (Pip)

F[y

* ' (M *>minl / Z l l ^ \ F(7p,Mfp)

where [x m '(Rp)min'(Ps)min'(- v -*)mini 1 are determined from the inviscid flow-field analysis, and [o n ,{ a,.6*/x m )] are evaluated from the mixing analysis for the local flow conditions at the minimum secondary flow area. The "corrected" mass-flow ratio is then given by W = (Wj + AW) . (4.29)

If the inviscid flow fields are assumed to be representative of the flow conditions in the presence of mixing, an improved approximation to the secondary flow conditions based on the "corrected" mass flow conditions can be made. In particular, the initial secondary flow Mach number (or M*s) corresponding to the "corrected" mass-flow ratio is found from F[ 7 s ,(Mt s )/l = ^^F(7p,M|p)W. A l s "ip (4.30)

The value of M | S found from (4.30) has been designated as the "limiting value", (M*s)y , for the given static pressure ratio P j s . For these conditions, the stagnation pressure ratio is Pos = P l s y iTs.O-fs),] (4.31)

The "pumping" characteristics of the ejector including the effects of mixing are thengiven by [W, P 0 s ] from Equations (4.29) and (4.31). Corresponding to these values, the ambient pressure ratio, P a , is determined from the inviscid "downstream" flow-field analysis (Section 4.4). t See Reference 28. For relatively weak compressions of the primary stream, the oblique shock can be approximated by a reversible compression thus simplifying the calculations.

57 The theoretical analysis using this approximate treatment of the mixing between the streams is shown 1 ' 2 , 3 to be in good agreement with experiments for the geometries investigated. Peters 29 obtained similar results by considering the mixing component concurrently with the inviscid analysis. 4.6 Shroud-Wall Boundary Layer

The secondary stream's flow area is effectively reduced by the boundary layer which develops along the shroud wall. The effects of the boundary layer on the ejector mass-flow characteristics can be estimated by superposition of the boundary layer's displacement thickness on the inviscid flow fields as was done for the mixing region. To do this the boundary layer development along the shroud wall is estimated*, to a first approximation, from the secondary stream's Mach number distribution determined by the inviscid analysis. The overall effects of the boundary layer, however, would be expected to be small since: (i) the secondary stream is accelerating, (ii) the boundary-layer development lengths are normally small, and (iii) the two-stream mixing region tends to locally cancel-out the displacement effect of the boundary layer.

5.

EJECTOR THRUST EVALUATION

The analysis of the internal flow within an ejector establishes the pumping characteristics of the system as well as the corresponding shroud wall pressure distribution. The results of this analysis are in general W = f(P 0 s ,P a ) and Pw = f(W,P0s,Pa,X). or Mls/ = f(P l s ,P a )

From these data, the gross (static) thrust of the ejector system can be evaluated by applying the momentum equation in the x-direction to the overall system (sec Figure I). The gross thrust is given by Tgross = J (P-P)dAx+J Vx(pV-dA),
(5.1)

where P ^ has been introduced as a reference ambient pressure level. To evaluate (5.1), the integrals over the non-uniform flow conditions at the ejector exit can be expressed in terms of the initial ejector flow conditions and the integrated shroud wall pressure distribution. For a control volume coinciding with the initial section, the shroud wall, and the ejector exit, the integrals over the ejector exit section can be expressed as: f (P - P . J dA x + | v (pV dA) = f J % Ae X [Ss (P - P) dA x f V x (pV dA) Ss

f
/Alp

(P - ?) dAx - J
'Alp

Vx(pV dA) +

/
A
.

(P - Poo) dA x .

(5.2)

Combining Equations (5.1) and (5.2), the equation for the gross thrust becomes f (P-P.JdAx Ss / Vx(pV-dA) Ss (P - P^) dA x f V x (pV dA) SP

gross

SP

+ J
Aw

(P-P<JdAx.

(5.3)

For the assumed flow conditions at section (1) of the ejector, the expression for the gross thrust can be simplified to:

r
IfilPii. = c o s 0 P 0pAlp
l w

^ - [1+7SM]S/] + P , p [ l + 7 p M ] p l - P 0 0 A lp

1 + COS 0[v
Ip

+ f (P-P.JdAx.(5.4) A l p -Aw

The ejector analysis thus determines [M l s / , P l s , P w ] and consequently the gross thrust can be found from (5.4). * A computer program suitable for the boundary layer analysis is contained in Reference 15.

58 For thrust-augmentation applications, Figure 19, the operating point of the ejector system is determined by matching the intake and afterbody characteristics to those of the ejector system4. The intake and ejector mass flow-pressure characteristics are matched at the ejector's entrance section while at the ejector exit section the matching, if necessary, is between the afterbody and the secondary stream static pressures. The intake's mass flow-pressure characteristics can be expressed as Wimake = MM...?......) intake = f2(M, P^,...) , (5.5a) (5.5b)

where fW^fg^g, P- nta k e l are respectively the dimensionless mass flow and static pressure ratios at the intake's exit section (ejector's entrance section, see Figure 19). Along the surface joining the intake and ejector, the matching requires that _ Wjntake " W (5.6a) Pintake = ?ls (5.6b)

In actuality, the intake-ejector matching represents in general the intersection of their performance surfaces when both are expressed in the same coordinates. At the ejector exit, the static pressure of the secondary stream can be

(PS)e 5 Pab -

(5P)

depending upon the flow regime and the conditions within the ejector. For the Pa-dependent regime, (P s ) e = P ab is required and consequently, the ejector pumping characteristics are influenced by the condition^ at the terminus of the afterbody. In the Pa-independent regime, condition (5.7) holds; however for this case, P a b only influences the shroud-wall pressure-area integral in Equation (5.4). The coupling between the intake, ejector, and afterbody subsystems is thus a significant factor to be considered in the net inflight thrust analysis of such systems. For the "matched" ejector operating point, the gross thrust is then evaluated from (5.4). The net inflight thrust performance of the system must then take into account the reduction in the thrust level due to the afterbody drag, the "additive" intake drag, and the intake momentum flux and pressure-area integral. The "matching" technique is in principle straight-for ward; however difficulties are encountered in practice as a result of the lack of information on intake performance characteristics and the influence of the intake on the flow over the system's afterbody. 6. EJECTOR FLOW MODEL IMPLEMENTATION

The ejector flow models described in the foregoing sections have been incorporated into a single computer program written in FORTRAN IV language. The salient features of this program are outlined in Table I. Comparisons between the theoretical analyses and experiments have been presented in the literature 2 ' 3,4 for several ejector configurations. However, for completeness, somejjf these comparisons have been included here. Figure 20(a) (References 2 and 3) is a comparison between the Pa-independent pumping characteristics of an ejector with a constant-area shroud and an ejector with a parabolic shroud with the same minimum flow area. The slight wall curvature of the parabolic shroud is seen to significantly alter the systems pumping characteristics. Experimental data for both systems is included in this figure. Figure 20(b) (Ref.2) is a comparison between the experimental data and theoretical curve for the "limiting initial secondary Mach number, Mi s /" for an ejector with a constant area shroud. The divergence between the theoretical and experimental data at the higher values of Pis is attributed to the small wake formed at the base of the primary nozzle due to its finite thickness. The third and last figure in these comparisons, Figure 20(c), is a comparison between the experimental and the inviscid theoretical shroud-wall pressure distributions up to the "choking" point for various values of P[ s . The latter figure tends to validate, at least for this configuration, the superposition of the mixing region on the inviscid flow fields. Peters29 has made a comparison with a similar pressure distribution given in Reference 3 for the parabolicshroud ejector system; his analysis also tends to validate the approach taken here. 7. AREAS WARRANTING FURTHER INVESTIGATION

The ejector analysis presented in the foregoing sections appears to adequately describe and treat the flow phenomena within ejector systems whose configurations are reasonably consistent with the various simplifying assumptions. Even so, there still exist many questions of a basic nature which must be resolved, or at least better understood before:

59 (i) an improvement between theory and experiment can be achieved, (ii) the limits of applicability of this analysis can be established, and (iii) the analysis can be extended to more general configurations and flow conditions. The principal problem areas are related to the development and recompression of the mixing region, the extent to which the primary-secondary flow-field analyses can be uncoupled from the mixing analysis and the need for an adequate configuration-oriented secondary flow-field analysis. More specifically, some of the problem areas are listed below. (i) Only limited information is currently available on the detailed development and associated transport properties for turbulent mixing at constant pressure or in the presence of favorable or adverse pressure gradients. Additional information from experimental and theoretical investigations is needed to define more clearly the limits of applicability of uncoupling the flow-field and mixing analyses. This information would also provide guidance in formulating a more general ejector analysis and could serve, partially at least, as a basis for assessing the merits of certain proposed thrust-augmentation systems. (ii) The interaction between the "nearly" inviscid flows and the mixing region in the vicinity of the wall and the subsequent recompression of the mixing region is not well understood. An investigation of this problem would be applicable to the "zero" and "small" flow regimes as well as to a class of base pressure problems. (iii) For configurations where the mixing and flow-field analyses can be uncoupled, a more generalized treatment of the secondary flow field and the secondary-primary streams interaction is required: (a) (b) for non-uniform "boundary layer like" secondary flow, and when the primary stream and shroud wall curvatures are significant thus invalidating the assumption of uniform one-dimensional flow.

(iv) When the secondary flow area and the flow area due to the mixing region's displacement thickness are on the same order of magnitude, a concurrent mixing and flow-field analysis must be carried-out. In addition, the various solution criteria must be reconsidered in relation to this analysis. The work of Peters 29 is a significant step toward solving this problem. (v) The influence of the mixing region and boundary layer on the "downstream" flow-field, the recompression processes, possible flow separation at the shroud wall, and the resulting shroud-wall pressure distribution are important factors in thrust performance evaluations, which require additional study. Many of these problem areas are basic and obviously go beyond the ejector analysis in their applicability. As a result, many of the problems should be investigated in simplified component studies freed from the necessarily complicating context of the ejector system analysis. A recent study of factors which influence the design and analysis of ejector nozzles has been conducted and reported in the literature 38 ; this study has particular importance because the findings have a direct relationship and bearing on the ejector flow model discussed in the preceding sections.

60

TABLE I Essential Characteristics of the Ejector Computer Program (FORTRAN IV, IBM 7094, 360/75).

Input Data 1. Ejector Shroud Wall Geometry: Specified by two equations of the form: Rw = AX2 + BX + C where 0 < X < Xe . 2. Primary Stream Properties and Variables: 7 p , Rp, T 0 p , X l p , R l p , M l p , 0 l p . 3. Secondary Stream Properties and Variables 7 S . Rs . T 0s 4. Options: (a) (b) (c) (d) (e) "unchoked" solutions and downstream flow field calculations. "choked" solutions. "choked" solutions and downstream flow field calculations. complete set of calculations, i.e., Pj, < Pi s < (P]s)max specified point calculation only.

Solution Output Data "Unchoked" Operation: Ks/^ls'Pal' [ W * V P a l ' Pw(X)' for P b < P l s < ( P l s ) m a x 2. "Choked" Operation: [M^.Pisl, (W,P 0s l for P b < P l s < ( P u ) m a x and, if the option is selected, P W (X), T g r o s s /PopAi p for the range of permissible P a .
3nd T

gross/ A lp P 0 P

61

OPI-OP

rP

Rw -AXw + BXw + C
Fig. 1 Ejector configuration and notation

Break-off Curve

Fig.2

Ejector mass flow characteristic surface, W = f(Prj s , Pa)

62

>i

Break-off Curve

Base pressure ratio

Fig.3

Ejector characteristic surface M>-s = f ( P l s , P a )

Pa = constant plane (Pa-independent regime)

Pa= constant (Pa-dependent regime)

Break-off Curve

Fig.4(a)

Intersection of the W-surface with planes of constant P a

63

w
A
Planes Pos = constant Break-off Curve

Fig.4(b)

Intersection of the W-surface with planes of constant PQS

Break-off Curve

Plane Pos= Pa

Fig.4(c)

Intersection of the W-surface with a plane Pa = P 0 s

64

Planes

of

P = constant

Break-off Curve

Fig.5(a)

Intersection of the Mis-surface with planes of constant Pa

Break-off Curve

a
Fig.5(b)

Constant

Pno is

Intersection of the M|s-surface with planes of constant P] s

65

CO

o 10to

IQT

/ Bre ak-off poii-it "B n"


i

W, M 1 S = 0

S
0
i i

R ase pre ssure


i

a
Fig.6 Intersection of the Mls-(or W-)surface with the base plane

<L
Mixing zone /Recompression shock Wake region

Primary boundary
Fig.7

Recompression zone

Flow field for "zero" flow regime

66

M. , P
Pos Is ' Is e

Increasing

P i Decreasing P Q Recompression

(Ps).= Pa

(a)

Pressure distributions for "choked" secondary flow

P,(x)
Increasing
Is
/

M,
l s

Decreasing (PS'e=Pa

Expansion - - Choked Recompression

(b) Pressure distributions for "unchoked" secondary flow

Fig.8

"Moderate to high" secondary flow regime

67

1. (M 1 S ,P 1 S ) same for both systems 2. "Choked" operation for both systems

. Break" off

Pa range

Exit
(a) Short-shroud ejector

Exit
(b) Long-shroud ejector

Fig.9

Qualitative influence of ejector shroud length on recompression to Pa ("choked" regime)

1.0 (P. ) > (P. ) 9


Is 1 Is 2

"BO"

Decreasing Shroud Length M


Is

Is 2

P
a

Pn

/ P

nn

op

Fig.9(c)

Influence of shroud length on the break-off (BO) point

68

CP
Pop'1
Pi I Ip I / /
v

(L
\

\ P A1P!X i

Recompression Shock

Wake Region

See Figs 10b 8 10c Fig. 10(a) "Corresponding inviscid flow field" for the "zero" flow regime

Inviscid boundary Recompression Shock

ew

\6i*)

*-w[i-*q
Fig. 10(b)

r^7777'

V77777777777yV7Zr Fig. 10(c) Recompression shock system at the reattachment point of the "corresponding inviscid boundary"

Superposition of the mixing region on the inviscid primary boundary

>

p?^W
Pw(x)
P

shk"

Korsf

A*
a ^ l

Experiment

Pw (X)

Fig. 10(d)

Actual recompression pressure rise in the vicinity of the reattachment point

64

Secondary

Inviscid boundary
%*..

Primary rd^
Fig. 11(a) Schematic of theoretical reattachment (ONERA, Reference 22)

1
* = i=H 1+er -]

Inviscid boundary

Fig. 11(b)

Superposition of the mixing region and determination of the displacement of the /-streamline

70

Fig. 12

Empirical law of angular reattachment (ONERA, Reference 22)

Inviscid Two-Stream Boundary


(a) Flow model for "small" flow regime

Two- Stream Inviscid Boundary

W;

(b) Recompression region for the entrained secondary flow

Fig. 13

Flow model for "small" flow regime

71

POP = I

Xp

* H ^ p (Xip.Rip)-^ I / A V
p

T O P ^ ^ s a ^ V ' x ' A7
is Has

n
Tos

wis A1S

, ^-(X,,R,)
(a) Two-stream inviscid flow model

*""

(XpiRp) Inviscid Primary Boundry

^ \ \ \ \ ^

(Xw,Rw)
(b) Secondary flow area determination

(n+l)tt>r(b)
nth

,''

\(c)
m' .\
/ :

(xp, Rp) Is
s

p = p

(Xw,Rw/
(c) Linking of secondary and primary flow fields Fig. 14 Inviscid two-stream flow fields

72

Primary

Secondary \

( M

s * >L mm i n

< 1

( A / A ) = minimum s Increasing M. , W,
9

Is'

<?).= Pa
(Po^BO

Fig. 15(a)

"Unchoked" secondary flow

[(A/A*)s = minimum. (Mf) min < 1, (P s ) e = Pa]

MIP

J ^

dAs # 0 dX

^<0

PW(X)

Fig. 15(b)

"Unchoked" secondary flow

[(A/A*)s = minimum, (Mf)e < 1 ]

73

\
\

\S

V^-

lnviscid

Boundary

<o
min

>0

(M;)=I
(a) "Choking" criterion

Is

M*
Is

W = constants

-r-(Pa'BO -P.-.'" 5 . "I?>0E -+1


x

PM)

(M )

=1

s min

-<Pc)ll

-. ) . Pa
min

(b) Shroud-wall pressure distribution for "choked" operation

Fig. 16

"Choked" two-stream inviscid solution

74

Primary
/

/
\

Shock // System

At Boundary (Ps)x (P s )y MPp)x =(Pp)y

Boundary \

(pp)\ /(Pp^

Secondary

(M*) x

^^NW^WW^
dX
(a) Supersonic-shock-subsonic "downstream" flow

tXn,RD ] p-"x,y

(Mg)x
(x) (y)

(9p D )^x

(b) Shock system configuration

Fig. 17

"Downstream" flow field calculations

75

Inviscid Boundary Secondary

^^^\v^\K\<\^^^
I

X min
where

dA s. _ dX =

(M S> *)m.n 'min < 1

( A / A * )Is c = minimum
(a) Inviscid velocity profile at the "choking" point

Primary Inviscid Boundary Secondary Two-St ream Mixing Region Xmin


(b) Viscid approximation at the "choking" point

fc*

==-S=JL_

Displaced Boundary ====. Due to Mixing "* Inviscid Boundary


i

.1*1
i-J

^\V^\1k\\\\VsV?\
Xmin
(c) Approximate displacement of the streams' boundary at the minimum secondary flow area due to mixing

Fig. I 8 Approximate two-stream viscid correction

76

Freestream M

Intake Afterbody

Ejector

Control

Volume

Fig. 19 Thrust augmentation system

0.6

Theory
0.5

Inviscid solutions Viscous effects included

0.4

Parabolic shroud, Ref.3


0.3 -

R w /Rip = - 1 . 7 4 - ^ ( X / R i p )

Exp, data o
A

0.2

Ref. 3 2

0.1
0

Constant area shroud, Ref. 2 Rw/Rlp=-1.74

18 Pos XIO'
Fig.20(a) Comparison of theory and experiment for the pumping characteristics of two ejector systems. (After References 2 and 3)

77

1
1 o l.VJ

Experimental
0.8

Ao A H

A
9r

71

0.6 -

0.4 -

0.2

Pis
n
^ i ^ \ i

- Pip J
i i \! i

10

12

14

16

Pis
Fig.20(b)

10 2

Comparison of theory and experiment for the "limiting initial secondary Mach number" for a constant shroud area ejector system. (M.p = 2, A w /A [ p = 3.06 (Ref.2))

P1S Table Inviscid Theory Value Line 0.06574 0.090 0.110 1.0 Experimental Symbol Value o
A

0.0658 0.092 0.111

0.5
_L

_L 3 X/Rip

Fig.20(c)

Comparison of theory and experiment for the pressure distribution up to the "choking" point. (M lp = 2, A w /A l p = 3.06 (Ref.2))

78 APPENDIX I

A LITERATURE REVIEW OF EJECTOR SYSTEMS AND RELATED TOPICS Ejectors Keenan, J.H., Newmann, E.P. and Lustwerk, F. "An Investigation of Ejector Design by Analysis and Experiment." ASME Transactions, Vol.72, 1950. Kochendorfer, F.D. and Rousso, M.D. "Performance Characteristics of Aircraft Cooling Ejectors having Short Cylindrical Shrouds." NACA RM E51E01, May 1951. Knoerschild, E.M. "The Design and the Performance Calculation of Ejectors and Aspirators." AF Technical Report No.6673, Wright Air Development Center, 1951. Ellis, C.W., Hollister, D.P. and Sargent A.F., Jr. "Preliminary Investigation of Cooling-Air Ejector Performance at Pressure Ratio from 1 to 10." NACA RM E51H21, October 1951. Greathouse, W.K. and Hollister, D.P. "Air-Flow and Thrust Characteristics of Several Cylindrical Cooling-Air Ejectors with a Primary to Secondary Temperature Ratio of 1.0." RM F 52L24, NACA, 1953. Fabri, J., Le Grives, F. and Siestrunck, R. "Etude Aerodynamique des Trompes Supersoniques." Jahrbuch der Wissenschaftlichen Gesellschaft fuer Luftfahrt, Braunschweig 11, pp.101-110, 1954. Kochendorfer, F.D. "Effect of Properties of Primary Fluid on Performance of Cylindrical Shroud Ejectors." NACA RM E53L24a, March 1954. Kochendorfer, F.D. "Note on Performance of Aircraft Ejector Nozzles at High Secondary Flows." RM E54F17a, NACA, 1954. Greathouse, W.K. "Preliminary Investigation of Pumping and Thrust Characteristics of Full-Size Cooling-Air Ejectors at Several Exhaust-Gas Temperatures." NACA RM E54A18, 1954. Hearth, D.P. and Valerino, A.S. "Thrust and Pumping Characteristics of a Series of Ejector-Type Exhaust Nozzles at Subsonic and Supersonic Flight Speeds." NACA RM E54H19, November 1954. Huntley, S.C. and Yanowitz, H. "Pumping and Thrust Characteristics of Several Divergent Cooling-Air Ejectors and Comparison of Performance with Conical and Cylindrical Ejectors." NACA RM E53J13, January 1954. Useller, J.W., Huntley, S.C. and Fenn, D.B. "Combined Compressor Coolant Injection and Afterburning for Turbojet Thrust Augmentation." N63-12531 NACA, 16th September 1954. Greathouse, W.K. "Performance Characteristics of Several Full-Scale Double-Shroud Ejector Configurations over a Range of Primary Gas Temperatures." RM E54F07, NACA, 1954. Allen, J.L. "Pumping Characteristics for Several Simulated Variable-Geometry Ejectors with Hot and Cold Primary Flow." RM E54G15, NACA, 1954. Ciopluch, C.C. and Fenn, D.B. "Experimental Data for Four Full-Scale Conical Cooling-Air Ejectors." RM E54F02, NACA, 1954. Kochendorfer, F.D. "Effect of Properties of Primary Fluid on Performance of Cylindrical Shroud Ejectors." RM E53L24a, NACA, 1954. Rolls, L. Stewart and Havill, C. Dewey. "An Evaluation of Two Cooling-Air Ejectors in Flight at Transonic Speeds." RM A54A05, NACA 1954. Stett, L.E. and Valerino, A.S. "Effect of Freestream Mach Number on Gross-Force and Pumping Characteristics of Several Ejectors." NACA RM E54K23a, March 1955. Fabri, J. and Siestrunck, R. "Etude des divers regimes d'ecoulement dans 1'elargissement brusque d'une veine supersonique." Rev. Gen. Sci. Appl. Brussels 114, pp.229-237, 1955. Vogel, R. "Practical Application of Air Ejectors." X65-17214, NASA, 1956.

79 Vogel, R. "Theoretical and Experimental Investigation of Air Ejectors." X65-17215, NASA, 1956. Cheers, F. "Development of an Air Injector." Report No. MT.31, National Research Council of Canada, 1956. Falanga, R.A. and Leiss, A. "Free-Flight Investigation at Transonic Speeds of Drag Coefficients of a Boattail Body of Revolution with a Simulated Turbojet Exhaust Issuing at the Base from Conical Short-Length Ejectors." RM L56H23, NACA, 1956. Carriere, P. "Interaction de 1'ecoulement externe et de I'ecoulement interne a la sortie d'un reacteur, aux vitesses transsoniques, et supersoniques." Compt. Rend. J. Intern. Sci. Aeronaut, May 1957. Fabri, J. and Paulon J. "Theory and Experiments on Supersonic Air-To-Air Ejectors." NACA TM 1410, September 1958. Fabri, J. and Siestrunck, R. "Supersonic Air Ejectors." Advances in Applied Mechanics, Vol.V, Academic Press, 1958. Crocco, L. "One-Dimensional Treatment of Steady Gas Dynamics." Fundamentals of Gas Dynamics, edited by H.W.Emmons (Princeton University Press, Princeton, N.J., 1958), Vol.III, pp.272-293. Rethorst, Scott and Royce, Winston. "The Annular Jet and Thrust Augmentation." N63-86009 Vehicle Research Corp., Pasadena, California, 1st December 1958. Mitchell, J.W. "Design Parameters for Subsonic Air-Air Ejectors." Report No.40, Department of Mechanical Engineering, Stanford University, 1959. Howell, R.R. "Experimental Operating Performance of a Single-State Annular Air Ejector." TN D-23, NASA, 1959. Lewis, W.G.E. and Drabble J.S. "Ejector Experiments." Report No. R 151, National Gas Turbine Establishment, Armed Services Technical Information Agency, 1960. Gertsma, L.W. and Yeager, R.A. "Preliminary Investigation of Off-Design Performance of Divergent-Ejector-Type Rocket Nozzles." N66-33346, National Aeronautics and Space Administration, Lewis Research Center, Cleveland, Ohio, March 1960. Chisolm, R.G.A. "Design and Calibration of an Air Ejector to Operate Against Various Back Pressures." N65-83059, Toronto University Institute of Aerophysics, September 1960. Corson, B.W. Jr, and Mercer, C.E. "Static Thrust of an Annual Nozzle with a Concave Central Base." NASA Technical Note D-418, N62-70992 NASA, Langley Research Center, Langley Station, Va., September 1960. German, R.C. and Bauer, R.C. "Effects of Diffuser Length on the Performance of Ejectors without Induced Flow." AEDC-TN-61-89, August 1961. Bauer, R.C. and German, R.C. "The Effect of Sound Throat Geometry on the Performance of Ejectors without Induced Flow. AEDC-TN-61-133, November 1961. Johnson, J.K., Jr., Snumpert, P.K. and Sutton J.F. "Steady Flow Ejector Research Program." Final Report N62-16458, Lockheed-Georgia Co., Marietta, September 1961. Bonnington, S.T. "The Design of Ejectors Driven by and Entraining Compressible Fluids." The British Hydromechanics Research Association. N65-80898, 1962. Chow, W.L. and Korst, H.H. "Influence on Base Pressures by Heat and Mass Additions." ARS J. 32, pp. 1094-1095, 1962. Eastman, R.H. "Effects of a Particle-Laden Driving Stream on Ejector Efficiency." Report 52, Joseph Kaye Co., Inc., 1963. Francis, W.E. "A Generalized Procedure for Optimum Design of Injectors, Ejectors and Jet Pumps." 66 11912, Midlands Research Station, The Gas Council, 1963. Traksel, J. "Simplified Ejector and Jet Thrust Augmentor Theory"' Lockheed Aircraft Corporation, 1963. Hale, J.W. "Auxiliary Ejector Effects on Rocket-Driven Diffuser Performance during Thrust Variation." N63-22022, Arnold Engr. Development Center, 1963.

80 DeHaan, R.E. "Supersonic Ejectors with Mixing at Constant Cross-Section." Section A, Vol.14, 1963. Hale, J.W. "Comparison of Diffuser-Ejector Performance with Five Different Driving Fluids." Arnold Engr. Development Center, 1963. Addy, A.L. "On the Steady State and Transient Operating Characteristics of Long Cylindrical Shroud Supersonic Ejectors (With Emphasis on the Viscous Interaction Between the Primary and Secondary Streams)." Ph.D. Thesis, Department of Mechanical Engineering, University of Illinois, June 1963. Sirieix, M. "Contribution a 1'etude des ejecteurs supersoniques." T.P. 32, Association Technique Maritime et Aeronautique, May 1963. Scott, W.J. "Experimental Thrust Augmentation of a Variable Geometry, Two-Dimensional Coanda Wall Jet Ejector." Aeronautical Report LR-394, National Research Council of Canada, 1964. Chia-An-Wan. "A Study of Jet Ejector Phenomena." AD 613485, Aerophysics Department, Mississippi State University, 1964. Fitch, R.E. "Analysis of the Inviscid Interaction Between an Underexpanded Rocket Exhaust and a Confined Secondary Stream." D2-36211-1 Boeing, 1964. Dyke, M.C. and Seddon, J. "Ejectors and Mixing of Streams." X66-15575 Royal Aircraft Establishment, Farnborough, England. Paris, Agard, 1964. Harper, D.J. "The Thrust Augmentation of Rocket Motors at High Altitudes." A65-16437, Ministry of Aviation, Royal Aircraft Establishment, Farnborough, Hants, England. Royal Aeronautical Society Journal, Vol.68, December 1964. Elliott, G.A. "An Experimental Study of Static Thrust Augmentation Using a Rocket-Ejector System." N64-84287, Southern Methodist University, Dallas, Texas, School of Engineering (M.S. Thesis), 15th January 1964. Chow, W.L. and Addy, A.L. "Interaction between Primary and Secondary Streams of Supersonic Ejector Systems and their Performance Characteristics." AIAA Journal 2, 1964. Geiger, F.W. "A One-Dimensional Thrust Augmentation Calculation." X65-18078, Brown Engineering Co., Inc., Huntsville, Alabama, Research Laboratories, June 1964. Bauer, R.C. "Theoretical Base Pressure Analysis of Axisymmetric Ejectors without Induced Flow." N64-15521 ARO, Inc., Arnold Air Force Station, Tenn., January 1964. Polak, P. "Notes on Jet Thrust Augmentation." A64-15527, Sheffield University, Department of Mechanical Engineering, Sheffield, England. The Engineer, Vol.217, 28th February, 1964. Strand, G.E. "A Study of a Supersonic Ejector Mixing Chamber." GAM 65B/ME/65-8, Wright-Patterson Air Force Base, Ohio, 1965, (M.S. Thesis). Middleton, D. "The Noise of Ejectors." N66-15539, Ministry of Aviation, London, 1965. Uebelhack, H. "Supersonic Air-Air Ejectors with Second Throat Diffuser." von Karman Institute for Fluid Dynamics, 1965. Chow, W.L. and Yeh, P.S. "Characteristics of Supersonic Ejector Systems with Non-Constant Area Shroud." AIAA Journal 3, pp.526-527, 1965. Hardy, J.M. and Delery, J. "Possibilities actuelles d'etude theorique d'une tuyere supersonique a double flux." Office Nationale d'Etudes et de Recherches Aeronautiques, Chatillon-sous-Bagneux (Seine), France, Technical Paper 287, 1965. Delery, J. "Recollement d'un Jet Supersonique de Revolution sur une Paroi Cylindrique Coaxiale." La Recherche Aerospatiale No. 104, 1965. Scott, J.E., Jr. "Mixing and Combustion of a Supersonic Fuel Jet and a Subsonic, Coaxial Gas Stream." N65-34264, Virginia University, Charlottesville. Presented at the AIAA 2nd Annual Meeting, San Francisco, 26th-29th July 1965. Nelson, J.R. "Investigation of the Mixing Region Between the Primary and Secondary Streams of a Two-Dimensional Supersonic Air-Air Ejector System." M.S. Thesis, N66-12816, Air Force Institute of Technology, Wright-Patterson Air Force Base, Ohio. School of Engr., 92 P Refs., June 1965.

81 Hoge, H.J. and Segars, R.A. "Choked Flow: A Generalization of the Concept and Some Experimental Data." AIAA Journal, Vol.13, No. 12, 1965. Korst, H.H., Addy, A.L. and Chow, W.L. "Installed Performance of Air-Augmented Nozzles Based on Analytical Determination of Internal Ejector Characteristics." Journal of Aircraft, Vol.3, No.6, 1966. Robinson, W.C. and Nelson, J.R. "Comments on 'Choked Flow: A Generalization of the Concept and Some Experimental Data'." AIAA Journal, Vol.4, No.7, 1966. Hoge, H.J. and Segars, R.A. "An Equation for the Entrainment of Ejectors." AIAA Journal, Vol.4, No.9, 1966. Nelson, J.R. "Investigation of the Mixing Region Between the Primary and Secondary Streams of a Two-Dimensional Supersonic Air-Air Ejector System." GAM 65B/ME/65-6, Wright-Patterson Air Force Base, Ohio, 1966, (M.S. Thesis). Yarborough, P.P. "Investigation of the Performance of an Ejector-Afterburner for Thrust Augmentation of a Small Rocket Engine." GAM/ME/66A-10, Wright-Patterson Air Force Base, Ohio, 1966, (M.S. Thesis). Fabri, J., Le Grives, E. and Michard, J. "Sur le Melange A Section Constante de Deux Flux A Enthalpies Differentes." France, 1966. Krenkel, A.R. and Kipowsky, H.H. "Design Analysis of Central and Annular-Jet Ejectors." Project No.7065, Aerospace Research Laboratories, 1966. Heiser, W.H. "Thrust Augmentation." A66-27008, MIT, Department of Mech. Engr., Cambridge, Mass. American Society of Mechanical Engineers, Gas Turbine Conference and Products Show, Zurich, Switzerland, I3th-17th March, 1966. Huang, K.P. and Kisielowski, E. "An Investigation of the Thrust Augmentation Characteristics of Jet Ejectors." AD 651-946, Dynasciences Corporation, 1967. Adamas, D.M. "Performance Analysis of Hypersonic Air-Augmented Supersonic Combustion Propulsion Vehicles." AIAA Paper No.67-226, 1967. Harris, G.L. "The Self-Preserving Turbulent Jet Ejector." AIAA Paper No.67-127, 1967. Bernstein, A., Heiser, W.H. and Hevenor, C. "Compound-Compressible Nozzle Flow." Trans, of ASME, Journal of Applied Mechanics, September 1967. Huang, K.P. and Kisielowski, E. "An Investigation of the Thrust Augmentation Characteristics of Jet Ejectors Final Report." Dynasciences Corporation, Blue Bell, Pa., April 1967. Peters, C.E. "Turbulent Mixing and Burning of Coaxial Streams inside a Duct of Arbitrary Shape." Doctoral Thesis presented to the University of Brussels, January 1968. Ahren, B. "Analysis of Ejectors used as Supersonic Propelling Nozzles for Jet Engines." Doctoral Thesis, Stockholm, 1968. Hardy, J.M. and La Combe, H. "Supersonic Bypass Nozzles - Computing Methods." Revue Francaise de Mechanique, pp.49-59, 1968. Hanbury, W.T. "The Performance of an Air-to-Air Ejector According to a Quasi- One-Dimensional Theory." ARC-29341, Aeronautical Research Council, Great Britain, 1968. Quinn, B. "The Characteristics of a Small Variable Geometry Ejector - Final Report." Aerospace Research Laboratories, Wright-Patterson Air Force Base, Ohio, ARL-70-0071, 1970. Anderson, B.H. "Factors which Influence the Analysis and Design of Ejector Nozzles." AIAA Paper No.72-46, 1972. Turbulent Mixing and the Base Pressure Problem Korst, H.H. "Compressible Two-Dimensional Jet Mixing at Constant Pressure." ME-TN-392-1, Office of Scientific Research TN-54-82, University of Illinois, reproduced as Office of Technical Services Collection, Publ. 132044, Department of Commerce, April 1954.

82 Page, R.H. and Korst, H.H. "Non-Isoenergetic Turbulent Compressible Jet Mixing with Consideration of its Influences on the Base Pressure Problem." Proc. Midwestern Conference., Fluid Mech. 11, pp.45-68, 1955. Korst, H.H., Page, R.H. and Childs, M.E. "A Theory for Base Pressures in Transonic and Supersonic Flow." University of Illinois, ME-TN-392-2, Office of Scientific Research TN-55-99, Contract AF 18 (600)-392, March 1955. Korst, H.H. "A Theory for Base Pressures in Transonic and Supersonic Flow." Journal of Applied Mechanics 23, pp.593-600, 1956. Chrisman, C.C. "Evaluation of the Free Jet Spreading Rate Parameter for Axisymmetric Flow of Air at Mach Number Three. AD 283 075, M.S. Thesis, Oklahoma State University, 1956. Wu, C.Y. "The Influence of Finite Bleed Velocities on the Effectiveness of Base Bleed in the Two-Dimensional Supersonic Base Pressure Problem." Ph.D. Thesis, Mechanical Engineering, University of Illinois, 1957. Bailey, H.E. and Kuethe, A.M. "Supersonic Mixing of Jets and Turbulent Boundary Layers. 57-402, Wright Air Development Center, 1957. Yakovlevskiy, O.V. "Thickness of the Turbulent Mixing Zone on the Boundary of Two Streams of Gases of Different Velocity and Density." N66-25920, Air Force Systems Command, Wright-Patterson Air Force Base, Ohio. Foreign Technical Division 8P Ref., Translated into English from Izv. Akad. Nauk SSSR, OTD. Nauk-Energ. 1 Transp. Moscow, No. 10, 1958. Korst, H.H., Chow, W.L. and Zumwalt, G.W. "Research on Transonic and Supersonic Flow of a Real Fluid at Abrupt Increases in Cross Section (With Special Consideration of Base Drag Problems)." Final Report, ME-TR-392-5, University of Illinois, Office of Scientific Research TR-60-74, Contract AF 18(600)-392, 1959. Chow, W.L. "On the Base Pressure Resulting from the Interaction of a Supersonic External Stream with a Sonic or Subsonic Jet." J. Aerospace Sci. 26, pp.176-180, 1959. Carriere, P. and Siriex, M. "Facteurs d'influencc du recollement d'un ecoulement supersonique." Intern. Congr. Appl. Mech. (September 1960); also Tech. Memo. 20, Office Nationale d'Etudes et de Recherches Aeronautiques, France, 1961. Nash, J.F. "A Review of Research on Two-Dimensional Base Flow." R & M 3323, Her Majesty's Stationery Office, London, 1963. Nash, J.F. "An Analysis of Two-Dimensional Turbulent Base Flow, Including the Effect of the Approaching Boundary Layer." R & M 3344, Her Majesty's Stationery Office, London, 1963. Carriere, P. "Current Research on the Problem of Base Pressure (Recherches Recentes Sur le Probleme de la Pression de Culot)." N64-11886 France. Office Nationale d'Etudes et de Recherches Aeronautiques, Chatillon-sousBagneux, 23P Refs, Presented at the Conference on Fluid Mechanics, Paris, 18th February 1963. Carriere, P. and Sirieix, M. "Resultats recents dans 1'etude des problemes de melange et de recollement." Congres dc Mecanique Appliquee, Munich, 1964. Hill, J.A.F. and Nicholson, J.E. "Compressibility Effects on Fluid Entrainment by Turbulent Mixing Layers." N65-11044, Mithras, Inc., Cambridge, Mass. Washington, NASA, November 1964. Roshko, A. and Thomke, G.J. "Flow Separation and Reattachment Behind a Downstream-Facing Step." SM-43056-1, Douglas Aircraft Co., Inc., 1964. Gogish, L.V. and Stepanov, G. lu. "On the Analysis of the Base Pressure in Two-Dimensional Supersonic Flows." N67-189-1, Stepanov, NASA, 1965. Korst, H.H. and Chow, W.L. "Compressible Non-Isoenergetic Turbulent (Pr = 1) Jet Mixing Between Two Compressible Streams at Constant Pressure." ME-TN-393-2, Engineering Experimental Sta., University of Illinois, Report for Research Grant NASA NsG-13-59, 1965. Carriere, P. "Rccherches recentes effectuees a 1'ONERA sur les problemes de recollement." Presented at the VII Symposium of Fluid Mechanics, 1965. Also ONERA TP No.275, 1965. Page, R.H. Kessler, T.J. and Hills, W.G., Jr. "Reattachment of Two-Dimensional Supersonic Turbulent Flows." ASME, 1967.

83 Sirieix, M., Mirande, J. and Delery, J. "Experiences fondamentales sur le recollement turbulent d'un jet supersonique." ONERA, 1966. Gailly, A. and Jacques, R. "Supersonic Turbulent Mixing and its Application to Problems of Reattachment Melange Supersonique Turbulent et Application aux Problems de Recollement." N66-39216 Ecole Royale Militaire, Brussels, Belgium. AGARD Separated Flows, Part 1, pp.271-301, References in French, Engish Summary (see N66-32906 19-12/CFST1), May 1966. Eggers, J.M. "Velocity Profiles and Eddy Viscosity Distributions Downstream of a Mach 2.22 Nozzle Exhausting to Quiescent Air." N66-37033, National Aeronautics and Space Administration, Langley Research Center, Langley Station, V. Washington, 93P Refs, NASA, September 1966. Sirieix, M. and Solignac, J.L, "Contribution to the Experimental Study of the Turbulent Mixed Layer Isobar of a Supersonic Flow (Contribution a I'Etude Experimentale de la Couche de Melange Turbulent Isobare d'un Ecoulement Supersonique)." N67-11979, France. Office Nationale d'Etudes et de Recherches Aerospatiales, Chatillon-sous-Bagneux, 30P Refs. In French presented at the Meeting of AGARD Specialists on Detached Flows, Rhode St.Genese, France, 10th-13th May 1966. Maise, G. and McDonald, H. "Mixing Length and Kinematic Eddy Viscosity in a Compressible Boundary Layer." AIAA Paper No.67-199, 1967. Schetz, J.A. and Jannone, J. "Planar Free Turbulent Mixing with an Axial Pressure Gradient." 67-FE-9, ASME, 1967. Korst, H.H. "Turbulent Separated Flows." von Karman Institute for Fluid Dynamics, Course Note 66b, Rhode St.Genese, Belgium, April 1967. Peters, C.E. "Turbulent Mixing and Burning of Coaxial Streams Inside a Duct of Arbitrary Shape - Final Report." AEDC-TR-68-270, ARO, Inc., Arnold Air Force Station, 1968.

Method of Characteristics Cronvich, L.L. "Numerical-Graphical Methods of Characteristics for Plane Potential Shock-Free Flow Problems." 629-105, Journal of the Aero Sciences, 1947. Isenberg, J.S. "The Method of Characteristics in Compressible Flow, Part 1A." A9-M-II/1, Brown University, 1947. Isenberg, J.S. "The Method of Characteristics in Compressible Flow, Part IB." F-TR-1173C-ND, Brown University, 1947. Cronvich, L.L. "A Numerical-Graphical Method of Characteristics for Axially Symmetric Isentropic Flow." 629-105, Journal of the Aero Sciences, 1948. Bower, J.M., Jr. "Determination of the Envelopes and the Lines of Constant Mach Number for an Axially Symmetric Free Jet." ZJ-7-054, Convair-Astronautics, 1958. Eastman, D.W. "Two-Dimensional or Axially Symmetric Real Gas Flows by the Method of Characteristics, Part 1: Formulation of the Equations." D2-10597, Boeing, 1961. Eastman, D.W. and Radtke, L.P. "Two-Dimensional or Axially Symmetric Real Gas Flows by the Method of Characteristics, Part HI: A Summary of Results from the IBM 7090 Program for Calculating the Flow Field of a Supersonic Jet." D2-10599, Boeing, 1962. Kackova, O.N. and Cuskin, P.I. Translated by H.N.Browne, "A Scheme for the Numerical Method of Characteristics." A64-19445, 1963. Moger, W.C. and Ramsay, D.B. "Supersonic Axisymmetric Nozzle Design by Mass Flow Techniques Utilizing a Digital Computer." AD 601589, ARO, 1964. Andrews, E.H., Jr., Vick, A.R. and Craidon, C.B. "Theoretical Boundaries and Internal Characteristics of Exhaust Plumes from Three Different Supersonic Nozzles." TN D-2650, NASA, 1965. Dushin, V.K. "Application of the Method of Characteristics to Compute Supersonic External Gas Flows in the Presence of Non-Equilibrium Processes." N 67-19170, Lockheed, 1966.

84 Naumova, I.N. "Method of Characteristics for Equilibrium Flows of an Imperfect Gas." AD 636 590, US Department of Commerce, 1966. Callis, L.B. "An Analysis of Supersonic Flow Phenomena in Conical Nozzles by a Method of Characteristics." TN D-3550, NASA, 1966. Volkonskaya, T.G. "Calculation of Supersonic Axisymmetric Jets." A collection of papers of the Computational Center of the Moscow State University. Israel Program for Scientific Translations, Jerusalem, N66 26631, 1966. Ratliff, A.W. "Comparisons of Experimental Supersonic Flow Fields with Results Obtained by Using a Method of Characteristics Solution." LMSC/0082-9, Lockheed, 1966. Butler, H.W. "Description of a Digital Computer Program for Nozzle and Plume Analysis by the Method of Characteristics." TM 54/20-108, Lockheed, 1966. Prozan, R.J. "Development of a Method of Characteristics Solution for Supersonic Flow of an Ideal, Frozen, or Equilibrium Reacting Gas Mixture." HREC/0082-8, Lockheed, 1966.

85 APPENDIX II

METHOD OF CHARACTERISTICS ANALYSIS FOR THE SUPERSONIC AXISYMMETRIC PRIMARY FLOW FIELD

1.

BASIC EQUATIONS

For steady irrotational axisymmetric flow 30 *, the differential equation for the complete velocity potential in terms of the cylindrical coordinates (X,R) is

V
where

cvax2

2 5

c axaR

\1

c 2 7aR 2

R::'

(1)

7 1 C2 = C 2 - ! (u 2 + v 2 ) ; and
II

V2 = ( u 2 + v 2 )

(2)

a<j> .

a*
3R

ax'

(3)

The condition that the derivatives of: a* and 3*

ax

aR

may be discontinuous along curves on the solution surface to (1) and hence, the values of 32<I>/aX2 , 32<I>/3R2 , and a 2 4>/aX3R are indeterminant along such curves - the characteristic curves yields: (i) The physical characteristics: /dR\

(dxju,
(ii)

= ,a (

" '+a)sin 6 sin a (dR). ,. n ( g T t t / R^

(4)

The hodograph characteristics: cot a

m** ^ u ^

= o.

(5)

The Family I (Right-Running) and Family II (Left-Running) characteristics and the applicable notation are shown in Figure All. 1. Introducing the velocity of sound at sonic conditions, C* and M* = V/C* , into Equation (5) yields (dM*), .I 1 sin 6 sin a (dfl). .I A '" + (dR)u, = 0 . IM " (M* tan a) R sin (0 + a) (8) (7) = 27 R l (6)

7 +

The problem of solving Equation (1) subject to the imposed boundary conditions then becomes equivalent to solving the set of simultaneous ordinary differential equations (4) and (8) under the same conditions. Various "unit processes" encountered in the numerical solution of (4) and (8) will now be discussed.

* See Appendix I for additional references.

86 2. FIELD POINTS

Consider two points (1) and (2), Figure AII.2(a), on the Family 1 and II characteristics, respectively, where the flow properties are known; then for the unknown point (3) at the intersection of these characteristics, the location and flow variables can be determined using the characteristic relationships, Equations (4) and (8), in finite-difference form. To a first approximation, the coordinates of point (3) are X3 and R3 = R, + (X 3 - X,) tan (0 - 0(),3 , (10) = [(R 2 - R,) + X, tan (0 - a ) , 3 - X 2 tan (0 + a ) 2 3 l [tan (0 - a ) I 3 - tan (0 + a ) 2 3 l (9)

where (0 + ( D'al-3 is defined as the average value of the quantity between the points (i = 1,2) and (3). Correspondingly, the flow variables at (3) arc given by: [ P , 3 0 , + P 2 3 0 2 + ( M * - M * ) + Q, 3 (R 3 - R , ) - Q 2 3 ( R 3 - R 2 ) |
03
=

(H)

I P | 3 + P 23 1

and M* = [ M * - P , 3 ( 0 3 - f l . J + Q.jCRj - R , ) l , (12)

where Qj3 and Pj3 (i = 1 or 2) are coefficients based on the average values between the points (i) and (3). These coefficients are defined as: P i 3 = (M* tan o0 j 3 and sin 0 sin a Qi3 = Pi3 R sin (0 -I- (-!)*] 13 (14) (13)

The values of [X,R,M*,0) 3 are determined initially by assuming that the flow variables at point (3) are simply the average of those at points (1) and (2); hence

0, -

i(fl, +02)

M* = i(M* + M*) .
Then, using Equations (9)-(12) in sequential order, an approximation to the values [ X , R , M * , 0 ] 3 ' ' , can be determined. These values are then used to determine the average quantities and subsequently the next approximation [ X , R , M * , 0 ] 3 2 ^ . If this procedure of successive approximations is repeated, values for the variables at point (3) for two successive approximations, (n 1) and (n), will be obtained, i.e., [X,R.M*,01 "-- and [X,R,M*,013n) .

If the problem areas discussed below are not encountered, the values calculated by this procedure stabilize rapidly and the iteraction is terminated when

[}')-*-]/#C>|<.I
or

(15)

(16) Typically, for values of e , , e 2 = 10" 4 , the iteration stabilizes for 5 < 1 1 < 10 . Difficulties encountered in the course of this iterative procedure have definite physical significance that can be traced to: (i) Either of the characteristics being oriented such that the quantity: [0 + ( - l ) ' a l - 3 0 (17a)

87 or, in other words, a characteristic is horizontal in the flow field for which Q-3-00 (ii) Compressions developed in the flow field due to wave coalescence such that M*<n> < In the first case, the quantity Qi3(R3-Ri) must be reconsidered when [+(-lW
i 3

(17b)

1.

(18)

(19a)

-0,

(19b)

The incremental length, A / , along either characteristic is given by (R 3 - R:) l * i sin[0 + ( - l ) * a ] , 3

A/ ~ and as

(19c)

[fl-M-iy-wliS-i-D,

A / - (X3 - X-) .

Hence, the quantity in the characteristics equations (11) and (12) must then be replaced by the limiting value, i.e.. Qi3(R3-R-)^P-3 when [0 + ( - l ) - a ] i 3 - 0 . (19e) sin a sin a 13 (X3-X-), (19d)

The second problem area usually results from "foldback" of the characteristics network as a result of the coalescence of the same family waves. To treat this problem, provisions must be made in the overall flow field calculation sequence so that the shock formed in the flow field as a result of this coalescence can be treated and thus, "foldback" of the characteristics network avoided.

3.

AXIS POINTS If any one of the points (1, 2 or 3) is located on the axis of symmetry, then Rj, 0j = 0 , j = 1 , 2 , or 3, (20)

and the calculation procedure of Section 2 is modified accordingly for the axis point calculations. No particular problems are encountered when these conditions are imposed on the field-point calculation sequence since the term, [R~'l, only appears in the characteristic equations as an average value in the coefficients Q- 3 . The two axis point calculations encountered in the primary flow-field analysis are shown in Figure All.2(b). For the one case, the unknown point (3) is on the axis and consequently R3 , 0 3 = 0 . Then the remaining values [X,M*] 3 are found from: X3 3

(21a)

X22

-^ tan (0 + a ) 2 3

(21b)

M* = M * - P 2 3 f l 2 - Q P 2 . To start the calculation sequence, assume that M* = M* .

(21c)

88 Then a first approximation, [ X , M * ] 3 , can be found from (21b) and (21c). Using the successive approximation technique, the calculations are continued until:

[(M*)^ \i'

(M*J n ~ l) ]M*^

< e .

(22)

Thus the conditions at (3) are determined as; [X,0,M*,01y 1 - 1 . In the other case, the known point (2) is on the axis where R2 , 0 2 0.

The calculation sequence for determining the values at point (3) are the same as outlined for the field point analysis, Section 2.

4.

BOUNDARY POINTS

For the ejector analysis, only two types of boundary-point calculations occur, viz, the constant-pressure condition for the "zero" flow regime and the non-constant pressure condition in the inviscid-interaction analysis. 4.1 Constant Pressure Boundary Along the boundary. Figure All.2(c), the condition of constant pressure for isentropic flow is: M3 = M2 where M* is found from P * l7,M*l 1 o = P b = constant. (23b) = M*. = constant , (23a)

A first approximation to the spatial location of the unknown point (3) is found from the streamline condition, (2) to (3), where (R 3 - R 2 ) = (X 3 - Xj) tan 0 2 3 and along the [-characteristic, (1) to (3), where (R3-R,) Then X 3 is found from: [R, - R2 -I- X2 tan 0 2 3 - X, tan (0 - a ) , 3 ] X3 = TTT-^ . Ar, ... . (tan 0 2 3 - tan (0 - ) 1 3 ] The local flow direction is then given by:
0i = 6 i

(23c)

= (X 3 - X , ) tan ( 0 - a ) , 3 .

(23d)

(23e)

_ [(M*-M,)-Q.3(R3-R.)] P'3

(23f)

Using Equations (23), the estimates for [ X , R , 0 ] 3 can be improved by successive approximations until

|[ e (n)_ 0 (n-l)]( 0 (n)|

<

Thus, the variables at point (3) are established as [ X , R , 0 ] ^ n ) and [M* = M*] . 4.2 Non-Constant Pressure Boundary The pressure along the primary boundary varies in this case due to the inviscid interaction between the primary and secondary streams (see Section 4.2, "Inviscid Flow-Fields Analysis"). The variation in pressure, and consequently M?, along the primary boundary is specified by the inviscid interaction in terms of the initial and local secondary flow variables as:

(7s. Mf) ir(7 P ' p) =


1

p ]7 (Ts- M *s) . o

Pis

(24)

89 The coordinates of point (3) are determined, as in the previous section, by Equations (23e) and (23c). The values of (X,R] 3 are then used to determine the secondary flow area and M* . Using these values, M* = M* is found from Equation (24), and then Equation (230 is used to determine 0 3 . These calculations are initialized by assuming that:
03
=

M* = M* . These values are used to determine the first estimate of [X,R,M*,0] 3 '- and [M*](''. By successive approxims tions, the values at point (3) are determined when: "(M*)(3n) - ( M * ) , " " 1 ) ] ! ^ " 5 PRIMARY FLOW-FIELD ANALYSIS
< e.

The unit processes described above must be organized into a sequential program which can be used to calculate the flow field subject to the imposed boundary conditions. This organization is principally one of "bookkeeping" and is normally not difficult. The basic sequence can take any of several different forms; the preference here has been to calculate along Family I characteristics toward the boundary. 5.1 Calculation Sequence The boundary conditions imposed on the flow-field calculations, Figure AII.3, are: (i) the flow variables are specified along the initial primary nozzle characteristic, and (ii) the conditions along the boundary are specified by the flow regime in which the ejector is operating. These data along with the conditions along the axis are sufficient to determine the primary flow field. At the nozzle corner, (X| p , Ri p ), a centered expansion (or compression), can occur as a result of the need for the primary stream to expand (or compress) at the nozzle exit to satisfy the imposed boundary conditions. The general flow-field calculation sequence, Figure AII.3, proceeds from the initial nozzle characteristic, along Family I characteristics, to the boundary. 5.2 Wave Coalescence The Family II characteristics from the boundary tend to steepen as the calculations proceed in the downstream direction. These characteristics eventually coalesce and form a shock wave within the flow field. This condition is detected by the crossing of waves of the same family thus giving rise to the "foldback" of the characteristics network. Although flow-field calculations where "foldback" occurs still yield results which are in reasonable agreement with experiment31 the flow-field calculations must invariably be terminated as a result of errors directly attributable to the unrealistic characteristic network that develops. An exact treatment of this problem is given in Reference 32, and an approximate treatment is given in Reference 33. However, since the calculated boundary is relatively insensitive to the method used to treat the coalescence problem, a simplified approach will be described here which is in excellent agreement with References 32 and 33. Referring to Figure All.3(b), wave coalescence has occurred within the flow field. Allowing a single "foldback" at this point, the conditions on the "upstream" and "downstream" sides of the coalescence point are determined by linear interpolation between the points (2) and (3) and the points (2)' and (3)' respectively. The "foldback" points (3) and (3)' are then dropped and the flow-field calculations are continued using the flow variables determined at the wave-coalescent point, Figure All.3(c). In actuality, the oblique shock wave formed in the flow field due to the wave coalescence propagates downstream where it becomes curved as a result of the continuous interaction between the shock and the waves in the flow field. As a consequence, the flow is rotational downstream of the internal shock wave. In the approximation described above, the flow is assumed to remain irrotational. This assumption yields, in most cases, results which are acceptable and consistent with the overall analysis. 6. INITIAL PRIMARY NOZZLE CHARACTERISTIC

The primary nozzle geometries are restricted to those configurations which produce sonic, uniform, or conical supersonic flow. The objective is to determine the flow variables along the initial nozzle characteristic for each configuration.

90 6.1 Sonic Nozzle

The sonic nozzle can be treated approximately as a nozzle which produces uniform flow at the nozzle exit that is slightly supersonic, i.e., M ] p = 1.01 . Brown and C h o w 3 4 ' 3 5 have recently developed a transonic flow analysis for establishing the initial Family I characteristic for typical sonic nozzles. 6.2 Uniform Supersonic Nozzle

For uniform supersonic flow at the nozzle exit, Figure AII.4, the initial I-characteristic is straight and the flow variables are known as [ M | p , G j p = 0) along this characteristic. The case where an initial compression must exist at the nozzle exit to satisfy the imposed boundary condition will be considered, in an approximate way, in Section 6.4. 6.3 Conical Supersonic Nozzle The flow in an ideal conical nozzle, Figure All.5(a), can be specified as being at a uniform Mach number (or M | p ) along the zone of the spherical sector that coincides with the nozzle. The initial I-characteristic can be determined exactly 3 6 or numerically from the flow conditions specified on the non-characteristic spherical surface. The numerical approach while yielding values for the initial 1-characteristic that are in excellent agreement with the exact theoretical results of Reference 36 has the advantage of being easily extended to treat approximately the problem of an initial compression at the nozzle exit (for both the uniform and conical nozzles). The flow conditions are specified along the non-characteristic curve, Figure AII.5(b), as being at a uniform value of M. p and that the velocity vector is always perpendicular to this curve. Any point on this curve is defined by l X , R , M | p , 0 ] N C where O > 0 > 0 l p and X = X l p + Rcone[cos 0 - c o s 0.p]
R = R

cone

sin 6

where

Rcone

R l p /sin 0 l p .

The non-characteristic curve is then subdivided and a characteristics network, shown in Figure AII.5(b), can be used to numerically determine the flow variables and corresponding locations on the initial conical nozzle characteristic. This general calculation sequence is important since the calculations are made from a non-characteristic curve to the corresponding characteristic curve. 6.4 A Compression at the Nozzle Exit

If the boundary conditions are such that the nozzle cannot discharge at its supersonic design Mach number, a compression wave is generated in the primary flow field. If this compression is assumed to be relatively weak, the oblique shock can be treated approximately as a reversible compression. For either the uniform or conical nozzle, the flow variables on the non-characteristic curve and the imposed boundary conditions are used to establish a single reversible compression wave at the nozzle exit location ( X i p , R ] p ) which satisfies the boundary conditions. With these data, the remainder of the initial I-characteristic can be established using the calculation sequence from the non-characteristic curve to the corresponding characteristic curve (see Figure AH.6).

-l

Family II

Y-i^ss. Family I

a = sin" 1 ( i )

K
Fig.AII.l Physical characteristic curves

9+ - x
Fig.AII.2(a) General field point

(R,0) = O

Fig.AII.2(b)

Axis points

92

Pb = Constant

Fig.AlI.2(c)

Constant-pressure boundary point

Initial II char. Axis points

Nozzle

Centered expansion at ( X i p , R l p ) Field point Boundary point


Fig.AII.3(a) Primary flow-field analysis

Boundary

93

Downstream
^-

U2
Fig.AII.3(b)

II

Some family wave coalescence. [(3) is calculated first; (3)' is calculated next]

12

Fig.AI1.3(c)

Modification of the calculation sequence for treating wave coalescence

X A

9\

Nonchar. Curve
R

(0 =0, M * * M J 5 )

V
M lp
/ - < \

v
a = s,n (

wP}

Initial II Characteristic

L (Xip.Rip)
Fig.AII.4 Uniform supersonic flow nozzle

94

Nonchar. Curve (axis)

Initial II Characteristic

a
(Xlp.-^lp)
Fig.AII.S(a)

'p=

sin 1(

" sfc )

Conical supersonic flow nozzle

Nonchar. Curve (axis)

I n i t i a l I I Characteristic

(xip>Rip)
Fig.AII.S(b) Characteristics network for numerical calculation of the initial Il-characteristic for a conical nozzle

Axis

-Char.

(Compression)

Char. (Design)

<Xi

'

'

ft

Ip

Ip

IP

> P

Ip

Fig.AI 1.6 Approximate analysis for a weak compression at the nozzle exit. (Uniform or conical supersonic flow nozzles)

05

APPENDIX

III

CONSTANT-PRESSURE TURBULENT MIXING ANALYSIS The flow model for mixing at constant pressure of two uniform streams is shown in Figure AIII. I. The mixing between the streams is assumed to take place over a length X and results in a fully-developed velocity profile at this section which has the form
^ =

!fb+i^!fberf(^)

(1)

where \p = U/Ua and sp^ = U b /U a . This velocity profile is assumed to hold within an "intrinsic" coordinate system (x,y) which is displaced relative to the reference system (X,Y) by an amount y m . The two coordinate systems are then related by x x X y = Y + y m (X) , where y m (X = 0) = 0 . Applying the continuity and momentum equations to the control volume indicated in Figure AIII. 1, the results Continuity Equation p b U b Y R b + P a U a Y Ra + J X p b V b dX =
J

(2a) (2b)

f (Y+y"*) p u dy
J

(3)

(YRb+ym)

Momentum Equation -p b U b Y R b + P.U2YR. + J X p b V b U b d x


0

= J ( Y R a + y m ) p U 2 dy
J Y

(4)

( Rb+ym)

Introducing the reference conditions [p a ,U a ] into Equations (3) and (4), the continuity and momentum equations become: (5) and

*^[m'*-izz(&*(YRa + Vm)

(6)

YRa and Y Rb are respectively large positive and negative values such that: | U a - U ( Y R a ) | < ea |T 0a - T 0 (Y Ra )l < el and |Ub - U(Y Rb )| < e b IT0b - T 0 (Y Rb )l < e b ,. where ea , ea , e b , and e b are arbitrarily small positive quantities. Combining Equations (5) and (6) and imposing the conditions (7), the expression for the displacement of the intrinsic coordinate system is found as
1
ym =
Y

(7a)

(7b)

Ra

sp dy
,
Y

(8)

(1 - * b )

Rb

'

Rb

'

After introducing into Equation (8) the mixing region similarity parameter a n (plotted in Figure AIII.2), the mixing length x , and the homogeneous coordinate T? = aMy/x , the result is

96 1 Vm ~ -?Ra (1 ~ * b )
p-^Ra
<P2 dr? - -Pb

p'-Ra iRa p <p dt?


Rb

(9)

For fluids of unity turbulent Prandtl number, Crocco's energy integral relationship is applicable and the stagnation temperature distribution can be uniquely related to the velocity field throughout the mixing region. Accordingly, the stagnation temperature profile is given by
T

A =

0 0a

Job l - t f i T 0a 1 - V b

+ r ~"^b
1 ~<Pb

(10)

The density ratio, for constant pressure mixing, can be expressed as P


Pa

ji $ m
= A - CW , T -M- = 'Oa 1 - Ca

(1 la)

From the energy equation, the temperature ratios are expressed in terms of the Crocco number, C a , and the velocity ratio as T T0 (llb.c)

The expression for the density ratio then becomes P pa d-c


a

(A-CjV-)'

(lid)

The integrals in the continuity and momentum equations can then be expressed as:
\P a \ (--Cj|) a

(12a)

and (1 - C | )

(\-C\sp1)

sp2 &r\

(12b)

To shorten the notation, the following integral functions have been defined

!,(?)

(i -ci)sphnRh
KTob/Toa) - CJ*61 ( i - c . ) ^

J^ R b p" L "*7Rb

p" d-cl^dT?
(A-Ca^) (1-Ca.ydt, (A-Ca^)

(12c)

[(T0b/T0a)-C I a ^D * J

(12d)

In this notation, the dimensionless coordinate shift, Equation (9) becomes 1 Vm = -?Ra (1 - * b ) [I 2 (T} R a )-l,(TJ R a )] . (13)

The jet boundary streamline (j-streamline) is defined as the streamline which separates the two flows. Applying the continuity equation, the condition for the j-streamline is PaUaYRa J(YRa+ym)pUdy. y, (14)

Expanding this expression and introducing the dimensionless variables, the integral equation to be solved for the j-streamline is >-.Ra p p''Ra sp drj = [j? Ra - tj m ] (15)

07

The fluid entrained as a result of the mixing can be expressed as We = fyj


-.Rb + Vn.

pU dy - p b U b ( - Y R b )

(16a)

We

PaUax
J

-..
-.Rb

. Pb a Pfa sp dl? - spyyVm + Pa Pa Pa

^ R b

(16b)

In terms of the integral notation of (12a), Equation (16b) becomes W. Pauax (16c)

M-7j) - 1 m 7 * b Pa

A mixing region displacement thickness, Figure AIII. 1, can be defined, in analogy to boundary layer theory, as

-Pb-V = w

P U ax ' 'a " a

i(-?j) - Vm ^ <ft, Pa

(16d)

Substituting (1 Ic) into (16d) and rearranging yield the mixing region displacement thickness function:

==)

[(Tob/T0a)-CaVb]
1(-7j)--?ii

(16e)

This function is plotted in Figure AIII.3 for isoenergetic mixing, i.e., T u b / T o a = 1 . The mass flow between the d- and j-streamlines in the mixing region is given by

Wdj = J V j pUdy or
Wdj = PaUa [ 1 , ( ^ - 1 , ^ ) 1

(17a)

(17b)

The mass flow per unit area, p a U a , can be expressed by "7 ( Pa U a 2 \(T+1)/(T-1)
1/2

R \ 7 + 1/ r* (T.MJ)

POaToa' 1 ' 2 (17c)

The foregoing results, although derived for two-stream constant pressure mixing, can be applied to the single stream case by simply letting sp^ = 0 . The single stream similarity parameter must be used and is given by the empirical correlation a, = 12 + 2.76M a , (18)

Further details of this analysis, as well as graphical and tabular data for the mixing integrals are given in Reference 36.

08

Fig.AIII. 1 Model for two-stream turbulent mixing at constant pressure

5.0

4.0

3.0
b tt |b~

2.0

I Single stream mixing II Two stream mixing

1.0
0

"ktWo.^-^^-'iWa)
where Cia = Cia W>b,Cna) 0

10

.20

.30
^b

.40
Ua

50

.60

.70

Fig.AlII.2

Similarity parameter for two-stream mixing region

99

tX5|x

Fig.AHI.3

Displacement thickness due to two-stream mixing region

100

REFERENCES 1. Addy, A.L. On the Steady State and Transient Operating Characteristics of Long Cylindrical Shroud Supersonic Ejectors (With Emphasis on the Viscous Interaction Between the Primary and Secondary Streams). Ph.D. Thesis, Department of Mechanical Engineering, University of Illinois, June 1963. Interaction Between Primary and Secondary Streams of Supersonic Ejector Systems and Their Performance Characteristics. AIAA Journal 2, 1964. Characteristics of Supersonic Ejector Systems with Non-Constant Area Shroud. AIAA Journal 3, 1965, pp.526-527. Installed Performance of Air-Augmented Nozzles Based on Analytical Determination of Internal Ejector Characteristics. Journal of Aircraft, Vol.3, No.6, 1966. An Investigation of Ejector Design by Analysis and Experiment. ASME Transactions, Vol.72, 1950. Performance Characteristics of Aircraft Cooling Ejectors Having Short Cylindrical Shrouds. NACA RM E51E01, May 1951. Etude des divers regimes d'ecoulement dans I'elargissement brusque d'une veine supersonique. Rev. Gen. Sci. Appl. Brussels 114, 1955, pp.229-237. Theory and Experiments on Supersonic Air-To-Air Ejectors. NACA TM 1410, September 1958. One-Dimensional Treatment of Steady Gas Dynamics. Fundamentals of Gas Dynamics, edited by H.W.Emmons (Princeton University Press, Princeton, N.J., 1958), Vol.Ill, pp.272-293. A Theory for Base Pressures in Transonic and Supersonic Flow. University of Illinois, ME-TN-392-2, Office of Scientific Research TN-55-99, Contract AF 18 (600K392, March 1955. A Theory for Base Pressures in Transonic and Supersonic Flow. J. Appl. Mech., Vol.23, 1956, pp.593-600. On the Base Pressure Resulting from the Interaction of a Supersonic External Stream with a Sonic or Subsonic Jet. J. Aerospace Sci., Vol.26, 1959, pp. 176-180. Research on Transonic and Supersonic Flow of a Real Fluid at Abrupt Increases in Cross Section (With Special Consideration of Base Drag Problems). Final Report, ME-TR-392-5, University of Illinois, Office of Scientific Research TR-60-74, Contract AF 18(600)-392, 1959. Influence on Base Pressures by Heat and Mass Additions. ARS Journal 32, 1962, pp. 1094-109 5. Turbulent Separated Flows, von Karman Institute for Fluid Dynamics, Court Note 66b, Rhode-Saint-Genese, Belgium, April 1967. Facteurs d'influence du recollement d'un ecoulement supersonique. Intern. Congr. Appl. Mech. (September 1960); also Tech. Memo. 20, Office Nationale d'Etudes et de Recherches Aeronautiques, France, 1961. Current Research on the Problem of Base Pressure (Recherches Recentes Sur le Probleme de la Pression de Culot). N64-11886 France. Office Nationale d'Etudes et de Recherches Aeronautiques, Chatillon-sous-Bagneux. 23P Refs presented at the Conference on Fluid Mechanics, Paris, 18th February 1963. Resultats Recents dans I'Etude des Problemes de Melange et de Recollement. Congres de Mecanique Appliquee, Munich, 1964.

2. Chow, W.L., Addy, A.L. 3. Chow, W.L., Yeh, P.S. 4. Korst, H.H., et al. 5. Keenan, J.H., et al. 6. Kochendorfer, F.D., Rousso, M.D. 7. Fabri, J., Siestrunck, R. 8. Fabri, J., Paulon, J. 9. Crocco, L.

10. Korst, H.H., et al. 11. Korst, H.H. 12. Chow, W.L. 13. Korst, H.H., et al.

14. Chow, W.L., Korst, H.H. 1 5. Korst, H.H. 16. Carriere, P., Sirieix, M. 17. Carriere, P.

18. Carriere, P., Sirieix, M.

101 19. Carriere, P. Recherches Recentes Effectuees a TONERA sur les Problemes de Recollement. Presented at the VII Symposium of Fluid Mechanics 1965, also ONERA TP No.275, 1965. Contribution a I'elude des ejecteurs supersoniques. TP32, Association Technique Maritime et Aeronautique, May 1963. Recollement d'un Jet Supersonic de Revolution sur une Paroi Cylindrique Coaxiale. La Recherche Aerospatiale No. 104, 1965. Possibilities actuelles de-etude theorique d'une tuyere supersonique a double flux. Office Nationale d'Etudes et de Recherches Aeronautiques, Chatillon-sous-Bagneux (Seine), France, TP 287, 1965. Experiences Fondamentales sur le Recollement Turbulent d'un Jet Supersonique. ONERA, 1966. A Review of Research on Two-Dimensional Base Flow. R & M 3323, Her Majesty's Stationery Office, London, 1963. An Analysis of Two-Dimensional Turbulent Base Flow, Including the Effect of the Approaching Boundary Layer. R & M 3344, Her Majesty's Stationery Office, London, 1963. Theoretical Base Pressure Analysis of Axisymmetric Ejectors Without Induced Flow. N64-15521 ARO, Inc., Arnold Air Force Station, Tenn., January 1964. Reattachment of Two-Dimensional Supersonic Turbulent Flows. ASME, 1967.

20. Sirieix, M.

21.

Delery, J.

22. Hardy, J.M., Delery, J.

23.

Sirieix, M., et al.

24. Nash, J.F.

25. Nash, J.F.

26. Bauer, R.C.

27. Page, R.H., et al. 28. Ames Research Staff 29. Peters, C.E.

Equations, Tables and Charts for Compressible Flow. NACA, Report 1135, 1953. Turbulent Mixing and Burning of Coaxial Streams Inside a Duct of Arbitrary Shape. Doctoral Thesis presented to the University of Brussels, January 1968. The Dynamics and Thermodynamics of Compressible Fluid Flow. New York, 1953, Vol.1, pp.73-87; Vol.11, pp.676-682. Ronald Press,

30. Shapiro, A.H.

31.

Love, E.S., et al.

Experimental and Theoretical Studies of Axisymmetric Free Jets. NASA Technical Report R-6, 1969. Calculation of Supersonic Axisymmetric Jets. A Collection of papers of the Computational Center of the Moscow State University Israel Program for Scientific Translations, Jerusalem, N66 26631, 1966. Theoretical Boundaries and Internal Characteristics of Exhaust Plumes from Three Different Supersonic Nozzles. TN D-2650, NASA, 1965. Compressible Flow Through Convergent Conical Nozzles With Emphasis on the Transonic Region. Ph.D. Thesis, Mechanical Engineering Department, University of Illinois at Urbana-Champaign, June 1968. Supercritical Flow Through Convergent Conical Nozzles. Flow Measurements SymposiumASME, AIP, ISA. Pittsburgh, Pa., May 1971. An Analysis of Supersonic Flow Phenomena in Conical Nozzles by a Method of Characteristics. TN D-3550, NASA, 1966. Compressible Non-Isoenergetic Turbulent (Pr t = I) Jet Mixing Between Two Compressible Streams at Constant Pressure. ME-Tn-393-2, Engineering Experiment Station, University of Illinois, Report for Research Grant NASA NsG-13-59, 1965. Factors Which Influence the Analysis and Design of Ejector Nozzles. AIAA Paper No. 72-46, 1972.

32. Volkonskaya, T.B.

33.

Andrews, E.H. Jr et al. Brown, E.F.

34.

35.

Brown, E.F., Chow, W.L.

36. Callis, L.B.

37.

Korst, H.H., Chow, W.L.

38.

Anderson, B.H.

102

103

EJECTOR DESIGN FOR A VARIETY OF APPLICATIONS* by Delbert Taylor

ARO Inc. Engine Test Facility Tullahoma, Tennessee, USA

The research reported in this paper was sponsored by Arnold Engineering Development Center, Air Force Systems Command, Arnold Air Force Station, Tennessee, under Contract F40600-69-C-0001 with ARO Inc. Further reproduction is authorized to satisfy the needs of the US Government.

104

105 CONTENTS Page NOTATION 1. 2. INTRODUCTION SIMPLE EJECTOR DESIGN METHODS 2.1 Constant-Pressure Mixing Theory 2.2 Constant-Area Mixing Theory EJECTORS USING PROPULSION SYSTEM NOZZLES FOR MOTIVE ENERGY 3.1 Ejector-Diffuser Experiments 3.2 Influence of Ejector-Diffuser Inlet Geometry 3.3 Effects of Ejector-Diffuser Mixing Length 3.4 Effects of Second Throats on Ejector-Diffusers 3.5 Influence of Second-Throat Diffusers on Minimum Cell Pressure 3.6 Reynolds Number Effect on Ejector Performance 3.7 Influence of Pertinent Parameters in Ejector-Diffuser Performance 3.8 Effects of Different Driving Fluids on Ejector-Diffuser Performance 106 109 109 110 113 114 114 116 117 119 121 122 122 122 128 130-131 132-163

3.

REFERENCES TABLES FIGURES

106 NOTATION

A Cf Cp D F gg H I K,, K', K j , Kj

area coefficient of friction specific heat at constant pressure duct diameter impulse function gravitational constant height impingement ratio of experimental starting or operating pressure ratio to normal shock total pressure ratio ratio of experimental starting or operating pressure ratio to normal shock static to total pressure ratio duct length Mach number mass flow rate pressure specific gas constant Reynolds number per unit length temperature velocity axial position ratio of specific heats efficiency half-angle dynamic viscosity fluid density

I. M
nil

P R Re//

T V
X 7 IJ 0 V p

Subscripts
0 , 1, 2

etc.

station locations free-jet boundary test cell, condensation duct exit, entrained exhaust

b c

d
e ex

107 1 n ne ns r s st t w wf x y impingement section nozzle nozzle exit normal shock radial, rejected subsonic, supersaturation second throat total wall wall friction upstream of normal shock downstream of normal shock

Superscripts primary secondary nozzle throat

108

109

EJECTOR DESIGN FOR A VARIETY OF APPLICATIONS Delbert Taylor

1.

INTRODUCTION

Ejectors, exhausters, jet pumps, or ejector-diffusers as they are called, have been used in a very wide variety of applications during the past 40 to 50 years. The chemical industries have used them for many years for such operations as exhausting fumes; exhausting air from condensers; vacuum evaporation, distillation, and crystallization; refrigeration; filtration; drying; air conditioning; and pumping large volumes of vapor and gas at low pressures 1 . Ejector-diffusers may be used for fixed-geometry, constant-pressure applications or for variable-geometry, variable pressure applications. For fixed-geometry, constant-pressure applications, the ejector is usually designed to operate at fixed values of primary or driving gas flow rate, stagnation pressure, and stagnation temperature. Therefore, it has no moving parts, is simple in design, operates with cheap, readily available fluids, and is reliable in service. Such an ejector may be used as an aerothermopressor by employing a small flow rate of gas at a high density level to entrain and pump a larger flow rate of a similar gas from a low density level at the ejector inlet to a higher level at its exit. This application is useful in situations where an installed compressor has an excessive compression ratio but insufficient flow rate for a specific requirement. The variable-area, variable pressure ejector can be used in aircraft propulsion installations to augment the net thrust or enhance the aircraft lift-drag characteristics. A shroud is placed downstream of the propulsive nozzle which serves as the ejector driving nozzle. The exhaust gas from the nozzle entrains and accelerates either the air used to cool the propulsion system accessories or the boundary layer from the aircraft lifting surfaces, both of which result in decreased drag values 2 , 3 . Since the advent of the turbojet engine, aircraft propulsion systems have been developed and/or qualified in ground test facilities under conditions simulating atmospheric flight (Reference 4 and Figure 1). Ejectors are used to pump the propulsion system exhaust products from the test cells and thus provide the simulated altitude pressure or to augment the altitude performance capability of the exhauster plant. In both cases, the first-stage ejector was comprised of the propulsive nozzle and a mixing duct. The pressure ratio of the ejector is a function of the kinetic energy in the exhaust jet and the amount of secondary (cell cooling air) flow. This application is representative of that employed in the developmental testing of rockets. The ejector principle is also used in ground effect machines and in V/STOL aircraft 5 . large mass ratios at small pressure ratios are employed to produce the desired force. In these applications,

Methods and techniques have been developed to improve the performance of ground test facilities in order to test propulsion systems for atmospheric and space applications. The use of the ejector principle as a pumping device, as a diffuser, and as a combination of both the ejector and diffuser has resulted. Much experimental data have been acquired from the study programs associated with these developments. Various applications of ejector design are discussed in the sections to follow.

2.

SIMPLE EJECTOR DESIGN METHODS

Because the very complex nature of the flow in an ejector renders a purely theoretical design approach unfeasible, semi-empirical methods have been developed for designing ejectors for various applications. Flugel6 in 1939 published a theoretical design method which forms the basis for ejector design. To design an ejector, it is necessary to make the following assumptions: (i) The mixing of the two streams is complete, and the Mach number at the exit plane of the constant-area mixing channel is equal to or less than one. (ii) The flow at stations 2, 3, 4 and 5 is one-dimensional and steady (see schematics in Sections 2.1 and 2.2). (iii) No heat is transferred at the duct wall.

110

2.1 Constant-Pressure Mixing Theory A schematic of a constant-pressure mixing ejector is shown below:

Primary Flow

Flugel's design method6 makes use of the equation of state and the conservation of mass, enerigy, and momentum at constant static pressure between stations 2 and 3. Between stations 3 and 4 (constant-area flow channel), the conservation equations are again used, and the flow is assumed to decelerate from supersonic to subsonic velocity through a channel shock system. The flow decelerates subsonically from station 4 to station 5. The constant-pressure mixing theory7 for gases having different state conditions is presented in the following: Primary plus Secondary Mass Flow Rate rh' + rh" = 1+ m m Constant Static Pressure Mixing
P2 P2 Ps

(1)

(2)

Conservation of Momentum (ihV)J + (AV); = (rfiV), . Dividing Equation (3) by nV gives


v

(3)

^7

1+

^/)v'-

rh \

I4l

m \ m / Equation (4) may be written in terms of Mach number (M): M

' V H XT), + ^ j( 7RTt Tt\ = 0 + $ M W H %)m ' (5>


+

The state condition of the mixture is then calculated at station 3 by using the Conservation of Energy equation: CpTt' m / m \ ^C;Tt" = ^ l + ^ j c p j T t 3

and the values for R3 , y 3 , and T t j : R3 = R' + (m7m')R" 1 + (m"/m') 7 + (m /m )7 1 + (lii'VnV) CpTt' + (m7m')C;Tt" C p3 (l + rh'Vm') By assuming a value for Mj , and knowing Tt'2, T* , R'2 , R'2', y'2 , y'2', C p , Cp , p' tj , p"2 = p 0 , m', m", then

Tj

P;/Po

= mi.-ft

Pi' = P 0 ( P > o ) = P.

Ill

Pj/pJ is calculated and M2 = f ( p - > ; , 7 2 ) .

Now Equation (5) can be solved for M3 . The ratio (A3/A'*) can be calculated by using the Continuity Equations as follows:

,), =

(pAV)*

= ^2_AMV^RT)

At the nozzle throat (A*), where M' = 1.0 = M'* ,

m' = p'*A'*M'* , \ (

7gc

(6)

and

rh, = (pAM J
3

^-)

(7)

V R T t ( T / T ty 3

Now, dividing Equation (7) by Equation (6) gives p3A3M3 |R'*T t '*(T/T t )'*7 3 p'*A'*M'*^ R 3 T tj (T/T t ) 3 7'
l * a l * . . . *

m'

(8)

Since m3 = 1 + m"/nV, M'* = 1.0 , and p 3 = pi' pi , where p2' = p p = = ({A' i /A" k ),y , ,p[ i ). therefore Ai_ m (1 + m"/m')p ti -f ( p ' * / ^ R 3 T t3 (T/T t ) 3 7'* A'* " Pi, x pj/pl, x M3 ^R'*T t '*(T/T t )'*-r 3 The pressure rise from station 2 to 3 may be calculated by using the impulse function; thus, F 3 = p 3 A 3 (l + 7 3 M 2 ). Then by conservation of momentum, the impulse function at station 4 may be determined by F = F 3 - F w f j _ 4 . However, F w f j _4 = 7rDLCfwq3 = 7 r D L C f w p 3 | M 2 . Now, by substituting Equation (10) and Equation (12) into Equation (11), F 4 can be obtained. The Mach number at station 4 may be calculated by graphical means as follows. From the continuity equation and impulse function, m = pAV = pAM F = pA(l + 7M2) and 7-1 . T,/T = 1 + M2 , (12) (11) (10) (9)

RT t (T/T t )

m becomes

m = f. / j M ^ f e - / ' \1 + ym*) y RTt

+ i(7 ~ DM2

(13)

112 Rearranging Equation (13) gives m /RT, 8c


M

N/TV- + i(7~ 1 + M2

DM2

(14)

A graphical presentation of this equation with 7 as a parameter is presented in Figure 2 (Ref.8). All of the parameters at the left of the equality are known; therefore, the graph is entered at the value rh/Fv'RT t /g c , and the corresponding subsonic or supersonic Mach number is read from the graph at the desired value of the ratio of specific heats (7). Equation (14) can be solved explicitly by rearranging in a quadratic form. First, rearrange such that m RT, 7gc Now let G = so that M 2 (l + i ( 7 - 1)M2 (1 + 7M 2 ) 2 A rearrangement gives 7-I-2C j + M 2 (l - 2 7 O - G = 0 . (15) My/\ + ^(7 1 + 7M2 1)M2

Now let K = 1 - 27G

and make substitution in Equation (15) to get the quadratic equation 7(K - 1)M 4 + 2KM2 - 2G = 0 . Therefore, the solution for M is M K v^K - 2 G \ " 1 - 7K 7 " (16)

The static pressure at station 4 may be determined by use of the impulse function (17)

P4

A 4 (l + 7 4 M 2 )

since A 3 (A 3 is known) 74 F4 = 7 3 (assume no change) = Equation (11)

M4 is obtained from Equation (14). The pressure ratio available across the subsonic diffuser depends on the value of M4 and the diffuser efficiency (17). The nonuniformity of the flow at station 4 and the thick boundary layer tend to cause separation at the inlet to the subsonic diffuser which results in poor compression efficiency. For this reason, the recommended values of 1 7 and A s /A 4 are 60 percent and 4, respectively. Then, Ps P4 where f(M 4 ,7 4 ) P4

1 +

-'GH

(18)

113

The overall pressure ratio of the ejector (p 5 /p 0 ) may be calculated by the following equation:

Ei = h
Po Pa

h
P3

h
Po

(19)

where p 3 / p 0 is a function of the flow rate of the secondary gas, the stagnation pressure of the primary gas in the nozzle plenum, the primary nozzle expansion ratio, and the geometry of the installation at station 2. 2.2 Constant-j\rea Mixing Theory Constant-area mixing theory for ejectors is presented in the following :

Primary Flow

UO
The analysis makes use of the equation of state and the conservation of mass, energy, and momentum between stations 3 and 4. A channel shock system decelerates the flow, as in the constant-pressure mixing case, to subsonic velocity at station 4. The flow decelerates subsonically from station 4 to station 5. The equations are obtained in the following way. By the conservation of momentum, the impulse function at station 4 becomes F4 since F'3 = F3' = and Fwf 3 . 4 " [pA(l +
7M 2

= F 3 + F3' - F w f j _ 4 ,

(20)

)]3

[pA(l + 7 M 2 ) ] 3 ' *DLCf w P 3 h M2

The Mach number and static pressure at station 4 may be determined as for the constant-pressure mixing case; thus m RT t

FVT
P4 =

f(M,7) 4

(21)

Solve graphically for M4 , and then substitute M4 into the impulse equation to determine p 4 : (22)

A 4 (l + 7 M 2 ) 4

The static pressure rise through the subsonic diffuser is calculated as for the constant-pressure mixing case. Thus, Ps = P4 where ^P = f(M, 7 ) 4 1 + 0.6

(H

(23)

114 The overall pressure ratio of the ejector may be calculated by *


Po

= *x^x^,
?4 P3 Po

(24)

where, again, p 3 /p 0 is a function of the selected design parameters: rh"/m', p}/p0 . and A3/A* . 3. EJECTORS USING PROPULSION SYSTEM NOZZLES FOR MOTIVE ENERGY

The development of ejectors to serve the dual purpose of evacuating the test cells and performing as supersonic diffusers in propulsion system test installations has been in progress during the last ten years by many organizations. Although the programs have been similar in general, each has been designed to study in detail the configuration required to produce the specific requirements of the organization conducting the studies. Therefore, much experimental information resulting from studies of the influence of various parameters on ejector performance with and without secondary flow is available. Also, a rigorous analytical model for the calculation of the base (test cell or vacuum vessel) pressure behind a two-dimensional model developed by Korst9 was modified by Bauer10 to solve the axisymmetric base pressure problem. These methods are very complicated and require the use of a high-speed computer operated by personnel familiar with the analysis in order to produce design parameters. Therefore, the empirical design methods developed from experimental results are used for practical applications. These experimental data and a discussion of their use in ejector-diffuser design for propulsion system test application are presented in the following sections. 3.1 Ejector-Diffuser Experiments An investigation" was conducted to determine the effects of nozzle area ratio and nozzle throat-to-diffuser area ratio on performance of ejectors without induced flow. Five nozzles having constant exit diameters and different throat diameters were tested in three diffusers having different diameters. All nozzles were conical with a 36-degree total exit angle, and all diffusers had length-to-diameter ratios of 3.0. Test cell pressure, ejector nozzle driving pressure, exhaust pressure, and diffuser static pressure profiles were measured. Data from all configurations were analyzed to determine the effects of ejector geometry on performance. In defining the characteristics of ejectors without induced flow, certain terms and flow phenomena should first be discussed. Three terms of primary interest are (see Figure 3): (i) The driving pressure ratio of the ejector (nozzle driving pressure/ejector exhaust pressure, p t /p e x ) (ii) The nozzle pressure ratio (nozzle driving pressure/test cell pressure, p t /p c ) (iii) The ejector rise ratio (ejector exhaust pressure/test cell pressure, P ex /P c )A thorough understanding of these three ratios and their significance is mandatory for all ejector work. Generalized performance characteristics of ejectors without induced flow are presented in two different ways to better illustrate ejector flow processes. In Figure 4(a), the reciprocal of the driving pressure ratio is plotted versus the reciprocal of the nozzle pressure ratio; in Figure 4(b), the driving pressure ratio is plotted versus the ejector rise ratio. Since the factors p t and p e x in the abscissa of both curves are the same, it is possible to produce these characteristic curves by varying either factor of the abscissa ratio. In the subsequent discussion, ejector exhaust pressure is the variable with nozzle driving pressure held constant for all conditions. Points A (Fig.4) correspond to some maximum value of ejector exhaust pressure and, consequently, give minimum driving pressure ratio. At this condition, the ejector is said to be ''unstarted". In the unstarted condition, the supersonic stream from the nozzle is not attached to the diffuser walls but forms a supersonic core in the diffuser for some distance downstream of the nozzle. The supersonic core does some small amount of pumping by entrainment, causing the ejector rise ratio to be only slightly greater than unity. As the driving pressure ratio is increased (decreasing p e x at a constant p t ), operating conditions are depicted by moving along the curve toward points B. In this region, test cell pressure decreases faster than ejector exhaust pressure, resulting in an increased, but still relatively low, ejector rise ratio. The supersonic core becomes larger in diameter and longer but does not fill the supersonic diffuser. A discontinuity begins, as shown at points B, resulting from attachment of the supersonic core to the downstream end of the supersonic diffuser. With the supersonic stream filling the diffuser, pressure waves can no longer be transmitted upstream against the now solid cross section of supersonic flow. The resulting isolation of test cell pressure from ejector exhaust pressure causes the discontinuity in performance (points B to C). The magnitude and position of the discontinuity vary for different ejector configurations but have always been found to repeat for the same configuration. In most cases, the magnitude of these discontinuities may be decreased by increasing the length-to-diameter ratio of the supersonic diffuser. Many ejector configurations produce discontinuities which extend to conditions shown by points D (Fig.4), having no continuous operation between points C and D.

115 The highly sloped portion of the curve between points C and D shows that test cell pressure decreases quite rapidly for small decreases in ejector exhaust pressure. In terms of the performance ratios, a slight increase in driving pressure ratio results in large increases in both nozzle pressure ratio and ejector rise ratio. This region of high sensitivity and occasional instability is probably caused by the rapid increase in the length of diffuser wall washed by the supersonic stream as the flow attachment point moves toward the nozzle. At point D, the driving pressure ratio is adequate for full expansion of the supersonic stream to the diffuser walls; the stream attachment point is close to the nozzle and stable. This condition is defined as "peak" or "start", and the ejector is said to be "started" as opposed to the previously defined unstarted condition. This peak condition gives the maximum pressure rise ratio attainable for the configuration. At any point beyond peak (driving pressure ratio greater than that required for start), the test cell pressure will remain constant regardless of a further decrease in ejector exhaust pressure (Pt/p ex "* )- ' n this started region (points D to E, Figure 4), test cell pressure is a function of nozzle driving pressure only. In terms of pressure ratio, the nozzle pressure ratio is insensitive to diffuser exit pressure for all started conditions (see points D to E, Figure 4(a)). In full-scale ejector usage, it is often impractical to optimize all performance parameters. It is, therefore, necessary to select a configuration suited to the particular application for which it is to be used. As an example, two different types of ejectors are used at the Rocket Test Facility (RTF), Arnold Engineering Development Center (AEDC), in connection with high altitude rocket testing. One of these is the auxiliary ejector used for pumpdown of the test cell before rocket firing. This ejector is driven by steam which may be throttled to the desired nozzle driving pressure; however, the ejector must exhaust to a given pressure, determined by the capability of the rotating exhaust machinery. Since the desired result is minimum test cell pressure, the auxiliary ejector must produce maximum ejector rise ratio to attain minimum test cell pressure. The second application is the rocket itself driving the ejector during firing. This differs from the auxiliary ejector in that the nozzle driving pressure, which in this case is the rocket chamber pressure, cannot be optimized from the standpoint of the ejector. It is, therefore, necessary to have the diffuser so sized with respect to the rocket motor as to give maximum nozzle pressure ratio (p,/p c ) for the ejector. 3.1.1 Effects of Ejector Geometry Peak nozzle pressure ratio variation caused by changing Aj/A* is shown in Figure 5. The curves are lines of constant A ne /A* . Figure 5 again shows that, for a constant A n e /A*, the nozzle pressure ratio increases with increased Aj/A* . The close grouping of the lines of constant A ne /A* again demonstrates that A ne /A* is secondary in importance to Aj/A* as a controlling parameter. The spread between air and steam data was partially caused by the difference in the specific heat ratios of the two gases; however, the total spread between air and steam data is considerably larger than the spread between the theoretical lines of constant specific heat ratio (dashed lines, Figure 5). This additional spread is probably caused by the use of a two-phase stream (wet steam) as the driving gas. At large values of Aj/A* , the air data deviated from the trend established at small values. For any small value of A(j/A*, nozzle pressure ratio (p,/p c ) was increased by increasing A ne /A* of the nozzle; however, at large values of Aj/A* , this trend was reversed. Data from a full-scale rocket test (Fig.6) with high values of A(j/A* show the same pattern as shown by the model data at low values of Aj/A* . The rocket data show no performance decay for large values of Aj/A*, and the nozzle pressure ratio (p t /p c ) increases with increased A ne /A* (10, 18 and 25) for a constant A,j/A* = 310 . Model data in the high Aj/A* region (Fig.6) is probably invalid because of Reynolds number effects, the effects of condensation shocks, or possibly air liquefaction at these low static temperatures. Since all supersonic nozzles used in this investigation had the same exit diameter, the nozzles which gave the largest values of A,j/A* were also those having the largest nozzle expansion ratios (Ane/A*). The peak driving pressure ratios for the 15 ejector configurations are plotted versus Aj/A in Figure 7. Data from all configurations form a single straight line for air and a single straight line for steam. These lines of constant specific heat ratio are slightly convergent toward decreasing values of Aj/A , as would be expected from theory. Peak driving pressure ratio for a given ejector geometry and a given specific heat ratio is a function of A,j/A* only. A comparison of Figure 7 data with corresponding theoretical one-dimensional total pressure loss ratio across a normal shock is shown in Figure 8. Theoretical total pressure loss ratio was determined from the corresponding A^/A* and the proper specific heat ratio (air or steam) for each point. Reciprocal values of the peak driving pressure ratios are used for easier correlation with theoretical values. When plotted in this manner, the air and steam curves are now parallel and very nearly coincide. The slight separation of the curves is probably caused by an error in the value of the specific heat ratio used for the wet steam. Both curves lie very close and parallel to the theoretical values shown by the dashed line, which indicates that one-dimensional theory may be used to design these systems for practical application. 3.1.2 Full-Scale Correlation A number of full-scale exhaust gas ejectors have been used in the RTF in tests of full-scale rocket engines, and transient recordings of rocket combustor pressure and test cell pressure have been obtained. Peak nozzle pressure ratios (p t /p c ) determined from these data are presented in Figure 6 as a function of the ratio of supersonic diffuser flow area to the rocket nozzle throat area and the rocket nozzle expansion ratio.

116 Performance of the full-scale ejectors during tests of solid-propellant rockets equipped with single exhaust nozzles compares favorably with performance of the model ejectors. The scatter of the full-scale ejector data is attributed to variation in test cell inleakage (different test cells and numbers of service leads through test cell walls), variation in rocket operating time, differences in the ratio of the specific heat of the exhaust gases, differences in the contours of the rocket exhaust nozzle, and the presence of solid particles in the products of combustion of the solid fuels. 3.2 Influence of Ejector-Diffuser Inlet Geometry An experimental investigation12 was conducted to determine the effect of truncated conical inlets on the performance of cylindrical ejectors without induced flow. Five conical supersonic nozzles having exit diameters of 4.16 in, exit half-angles of 18 deg, and various throat diameters were tested in 6-, 8-, and lO-in.-diameter straight cylindrical supersonic diffusers equipped with diverging conical inlets having half-angles of 12, 18, 24, 30 and 36 deg. These cylindrical diffusers had length-to-diameter ratios of 3 and discharged into conical subsonic diffusers. Nozzle total pressure, test cell pressure, and subsonic diffuser exit pressure were measured. The truncated conical inlets in the cylindrical diffusers (Fig.9) produced as much as 600-percent improvement in diffuser performance. Figure 10 shows the performance of a typical cylindrical ejector with and without a conical inlet. Without the conical inlet, the ejector rise ratio (p e x /p c ) increased as the driving pressure ratio (p ex /Pt) decreased until the ejector started, that is, until the diffuser cross section was filled with supersonic flow. Thereafter, the pressure ratio (Pc/Pt) across the driving nozzle remained constant, and the ejector rise ratio decreased as the exhaust pressure was decreased. When the conical inlet was placed in the duct at such an axial position that the free-jet boundary interacted with the inclined surface, the minimum value of the pressure ratio (p c /Pt) across the nozzle decreased. The driving pressure ratio (p e x /p t ) required for ejectors equipped with conical inlets did not change appreciably from that required for ejectors without such inlets (Fig. 11); therefore, the ejector rise ratio increased in proportion to the decrease in the pressure ratio across the driving nozzle. 3.2.1 Effect of Conical Inlet on Diffuser Performance The results of testing conical inlets having various divergence angles in cylindrical diffusers to determine the influence of the divergence angle on the diffuser performance are summarized in Figure 12. Optimum performance improvement lp c /p c (opt)- where p c is cell pressure without conical inlet and p c ( o p t) ' s t n e lowest value of cell pressure with conical inlet at optimum axial position) is plotted versus the half-angle of the conical inlet (0COne) with the nozzle area ratio (A ne /A*) as a parameter. The lines connecting the data points obtained for each configuration serve only to simplify the reading of the graph. Since the exit areas of all nozzles were equal (exit diameter of 4.16 in), the Mach number of the free-jet boundary between the nozzle exit and the duct increased as the nozzle area ratio increased: Mj, = f l ^g- x j = f( j 1 (one-dimensional, isentropic theory).

The half-angle of the conical inlet required for optimum diffuser performance decreased as the nozzle area ratio increased. The decrease in performance improvement noted in the case of the large area ratio nozzles may have resulted in part from the very low Reynolds numbers of the flow in these nozzles. Also the poor performance of the expansion ratio 5.00 nozzle in the 6 in duct may have been caused by leakage of the high pressure air which bypassed the nozzle seal from the nozzle plenum chamber into the evacuated region (test cell), since such inleakage of air thickens the viscous mixing layer along the free-jet boundary and results in performance decay. 3.2.2 Effect of Axial Location on Conical Inlet The effect of conical inlet angle and axial position on diffuser performance is indicated in Figure 13, in which the pressure ratio across the nozzle is plotted versus the axial position (X) of the conical inlet in relation to the nozzle exit plane. (The steam configurations are not included in these plots; however, the data are presented in Table I.) The minimum values of p c /pt with conical inlets retracted varied slightly; this variation is attributed to variations of inleakage around the positioning rods (see Figure 9). It is apparent that the axial position of the conical section became more critical as the angle of the conical inlet was increased toward its optimum value, that is, toward the angle which produced the greatest improvement in diffuser performance. The effect of the radial height (Hr) of the conical inlet on diffuser performance and conical inlet position was investigated using a 10.8 area ratio nozzle in the 6 in duct with a 24 deg half-angle conical inlet. Figure 14 shows the pressure ratio across the nozzle versus the axial position of the conical inlet. Diffuser performance decreased as the radial height was decreased; however, the axial position of the inlet required for optimum performance did not change appreciably.

117 The effect of a three-step conical inlet tested with the expansion ratio 5.0 nozzle in the 10 in duct is presented in Figure 15, which also shows a sketch of the inlet. The pressure ratio across the driving nozzle is plotted versus the axial position of the inlet. In this case, the minimum value of the pressure ratio across the nozzle was 2.0 x I0~ 4 at X = 2.81 in; the performance for the 36 deg conical inlet (Fig. 15) is also presented for comparison. The optimum position of the three-step inlet was slightly more critical than that of the 36 deg inlet. The performance of this diffuser configuration was 600 percent better than the performance of the diffuser without a conical inlet. The optimum values of p c / p t obtained for each nozzle-duct configuration and for each nozzle-duct-best-cone configuration are presented versus A^/A in Figure 16 for comparison. The nozzles having area ratios of 3.56, 5.0 and 10.8 in ducts without conical inlets produced p c / p t values which correspond to one-dimensional values based on Aj/A . Those nozzles in ducts equipped with conical inlets produced p c / p t values which were as much as one-eighth of isentropic values. 3.3 Effects of Ejector-Diffuser Mixing Length

An investigation 13 of ejectors without induced flow was made to determine the effects of varying diffuser lengths on ejector performance. Four 18 deg half-angle conical nozzles having constant exit diameters and different throat diameters and two contoured nozzles having zero-degree half-angles at the exit were used as the ejector driving nozzles. Unheated air was used for all tests. The diffuser length-to-diameter ratios were varied between 0.7 and 21.5, and three cylindrical ducts of different diameters were used both with and without a subsonic diffuser. An empirical method was developed to estimate the starting and operating pressure ratios of such ejector configurations using simply determined one-dimensional normal shock relationships. Typical ejector configurations are shown in Figure 17. The fundamental ejector starting phenomenon is illustrated in Figure 18. In region (1) of the performance curve, both the nozzle and the ejector were unstarted. As the ratio P e x /Pt w a s decreased, the nozzle became started (minimum nozzle exit pressure) at point A in region (2). However, the ejector did not start (minimum cell pressure) until point B in region (3) was reached. Point B occurred at the maximum starting pressure ratio. When the ratio pex/Pt_ was increased after the ejector started, the reverse of the described phenomenon occurred, and the ejector became unstarted when the operating pressure ratio was exceeded. To demonstrate the effect of diffuser length on the starting and operating pressure ratios, some typical data are shown in Figure 19. For a diffuser L/D of 8.1, the starting and operating pressure ratios, points A and B , respectively, were essentially identical. For diffuser L/D's near 1.0, a significant difference existed between the starting pressure ratio, point A , and the operating pressure ratio, point B , which resulted in significant hysteresis. 3.3.1 Effect of Subsonic Diffuser Ejector starting characteristics were determined both with and without a subsonic diffuser. Figure 20 shows the static pressure distribution in a long cylindrical diffuser (L/D = 9) at p e x /Pt values slightly less than the operating pressure ratio, This wall static pressure distribution in the cylindrical diffuser remained the same, whether a subsonic diffuser was or was not used. The subsonic diffuser gave a slight increase in pressure recovery which resulted in an increased ejector starting pressure ratio although the major portion of the pressure rise occurred in the cylindrical diffuser section. The experimental starting and operating pressure ratio was improved approximately 12 to 18 percent for long diffusers (L/D > 8) when a subsonic diffuser was used (Figures 21 and 22). As diffuser length was decreased (L/D < 8), the subsonic diffuser was found to improve the starting pressure ratio from 10 to 68 percent (Fig.22). The improvement in starting pressure ratio produced by the subsonic diffuser can be very closely predicted for long diffusers by comparing the one-dimensional relationships for static-to-total and total-to-total pressure ratios across a normal shock. For a given cylindrical duct-to-nozzle throat area ratio (Aj/A ), a one-dimensional isentropic Mach number was determined. Then the corresponding normal shock static-to-total pressure ratio which occurred when no subsonic diffuser was used (Fig.21(b)) was compared with the total pressure ratio across a normal shock which occurred when a subsonic diffuser was used (Fig.21(a)). This theoretical comparison predicted an approximate 12 to 15 percent improvement in the starting pressure ratio when a subsonic diffuser was used. 3.3.2 Effect of Supersonic Diffuser Length The starting pressure ratio reached a maximum at a diffuser L/D = 3 for ejectors equipped with a subsonic diffuser and a conical nozzle (Figures 22(a) through (c)). It is also significant that the starting pressure ratio equaled the operating pressure ratio above an L/D = 3 for conical nozzles. For the contoured nozzles, the starting pressure ratio reached a maximum at an approximate L/D = 8 (Fig.22). It was noted that a small difference existed between the starting and operating pressure ratios at L/D > 8 for the contoured nozzles. As the length of the cylindrical supersonic diffuser section was decreased below the optimum values of L/D , the starting pressure ratios also decreased (Fig.22); however, the rate of decrease was less for configurations having contoured nozzles than for those having conical nozzles. No significant change in the operating pressure ratio was

118 noted for diffuser lengths as low as L/D = 1.6 for ejectors equipped with conical nozzles and a subsonic diffuser (Figures 22(a) through (c)). The ejectors equipped with the contoured nozzles, however, had a decreasing operating pressure ratio below approximately L/D = 8 (Fig.22(d)). Figure 22 also shows that the difference between the starting and operating pressure ratios (the ejector hysteresis) increased as the L/D was decreased. When no subsonic diffuser was used, the optimum L/D for conical nozzles increased to approximately 5. For contoured nozzles, no major change was indicated in the optimum L/D when no subsonic diffuser was used; however, insufficient data were obtained to evaluate this adequately. There was no variation in the starting pressure ratio for diffuser lengths above these optimum values except for a slight decrease in p e x /Pt resulting from the frictional effects of very long diffusers. As the length of the cylindrical supersonic diffuser section was decreased below the optimum values of L/D when no subsonic diffuser was used, the starting hysteresis was noticeably increased for the conical nozzles as compared with that obtained using a subsonic diffuser (Figures 22(a) through (c)). 3.3.3 Estimating Ejector Starting and Operating Performance Although the compression shock system in a long duct is a series of lambda shocks resulting from an interaction between the boundary shock and the boundary layer on the duct walls, Shapiro 1 4 states that one-dimensional normal shock relationship used with the duct inlet Mach number will predict the pressure rise across the shock system within approximately 6 percent. This good agreement is explained by the fact that the wall shearing forces in the region of separation caused by shock-boundary layer interaction are extremely small. Although Shapiro's results were obtained for uniform duct inlet flow, the experimental results for ejectors in which the flow was not expected to be uniform still showed good agreement with one-dimensional normal shock relationships. As shown in Figure 21, the experimental results for diffuser L/D > 8 were approximately 88 percent of the theoretical normal shock values (p t / p t ) n s for conical and contoured nozzles when a subsonic diffuser was used. When no subsonic diffuser was used, the experimental results were approximately 90 percent of the theoretical normal shock values (p 2 /p t ) n s for conical nozzles and 80 percent of those for contoured nozzles. A more accurate prediction of the starting and operating pressure ratios can be obtained by correcting the theoretical normal shock value by a parameter which is a function of nozzle geometry and the diffuser length-todiameter ratio. This correction parameter is expressed as a ratio of the experimental starting or operating pressure ratios to the theoretical normal shock value. For ejectors having subsonic diffusers which essentially diffuse the air to a zero velocity condition, the total pressure recovery across a normal shock for the one-dimensional isentropic Mach number corresponding to the area ratio of the cylindrical diffuser in the region of jet impingement to the nozzle throat area (Aj/A*) was used. Thus,
K

= (Pex/Pt)exper/(Pt 2 /Pt,)ns

For ejectors having no subsonic diffuser, it was assumed that the ejector system diffused to the exhaust conditions by the static to total pressure ratio across a normal shock for the isentropic Mach number corresponding to A d / A * . Thus,
K

"

(Pex/Pt)exper/(P2/Pt,)ns

The correction constant was calculated for each test configuration and was plotted versus the diffuser lengthto-diameter ratio (Fig.23). For diffuser lengths above optimum, there was a small change in the correction constants with increasing diffuser length, which varied primarily as a function of nozzle geometry and whether or not a subsonic diffuser was used. Below the optimum L/D , the correction factor also varied considerably as a function of whether or not a subsonic diffuser was used. The decrease in K2 for diffusers having L/D > 10 (Fig.23) reflected the pressure loss resulting from the frictional effects discussed in Reference 14. For configurations having diffuser lengths greater than optimum and either contoured or conical nozzles, the values of the correction constants varied less than 5 percent at any given length for all configurations tested. For diffuser lengths less than optimum, this variation increased to 10 percent (Fig.23). By using the values of K from Figure 23, the starting pressure ratios from Reference 15 can be estimated within 10 percent for L/D's greater than 4 (Fig.24). Although the described method predicts the starting pressure ratio for ejector configurations using nozzle shapes similar to those tested, its use for ejectors having nozzles with other shapes may require modifications to the presented methods. As diffuser length approaches the jet impingement distance, the diffuser length parameter (L/D X/D) would provide a better correlation parameter than L/D . This would appear to be a more realistic approach because it more nearly reflects the effective length of duct as far as the pressure recovery phenomena in the diffuser are concerned. Figure 25 indicates an improvement in correlation of as much as 15 percent for some configurations equipped with short diffusers when this parameter was used. However, as the diffuser L/D approaches X/D, the

119

values of K', and K2 are expected to become unreliable because the minimum cell pressure ratio may change and thus shift the region of jet impingement. The following jet impingement distances were used in determining the length parameter for the experimental data plotted in Figure 25:
D - 6.09 in A/A* X 3.63 5.07 10.85 25.00 23.68 2.05 2.14 2.3 3.85 3.13 X/D 0.336 0.351 0.378 0.632 0.513 X 4.7 5.0 6.46 8.8 X/D 0.462 0.492 0.633 0.863 D = 10.19 in

These impingement distances were calculated using Latvala's method of calculating the jet boundary shape l6 for a total pressure of 45 lb/in2 and the experimental minimum cell pressure ratio (p c /p t ) which can be calculated using Reference 1 7. 3.3.4 Total Pressure Effect The ejector starting and operating characteristics remained unchanged by a variation in nozzle plenum total pressure in the case of an L/D = 9 , as shown in Figure 26. However, as diffuser length was decreased to values at which the jet impingement distance became an important parameter, the ejector starting and operating pressure ratios varied with total pressure level, as shown in Figure 26(b). This variation is in order since jet impingement distance is a function of minimum cell pressure ratio, which is also a function of nozzle total pressure. Figure 27 shows a typical variation in the minimum cell pressure ratio (p c /p t ) with total pressure. Jet impingement studies show that as the cell pressure ratio increases, the jet impingement distance increases. This gives the decrease in the length parameter term (L/D X/D) in Figure 25 and a corresponding decrease in the value of K , which results in a decrease in the starting or operating pressure ratio for short diffuser lengths. Thus, the influence of total pressure level in the case of very short diffusers (L/D < 3) indicates that the accuracy with which the starting and operating pressure ratios can be predicted will depend on the accuracy of the calculation of the jet impingement distance. Reference 17 presents a good method of estimating the cell pressure for a given configuration, which can be used in determining the impingement distance. 3.3.5 Effect on Ejector Minimum Cell Pressure Ratio The effect of diffuser length on ejector minimum cell pressure ratio (p c /p t ) is shown in Figure 22. The cell pressure ratio varied only slightly as a result of diffuser length or the method of subsonic diffusion. The experimental results show that the minimum cell pressure ratio was not affected appreciably at diffuser lengths as low as L/D = 0.7 . It is believed that the cell pressure ratio would not be affected at diffuser lengths shorter than this, provided the diffuser length was maintained greater than the impingement distance. 3.4 Effects of Second Throats on Ejector-Diffusers An investigation18 of ejectors without induced flow was made to determine the effects of second-throat geometry and position on the starting and operating pressure ratios. Twenty-seven ejector configurations were tested using three 18 deg half-angle conical nozzles and one contoured nozzle in combination with six secondthroat configurations. Unheated air was used for all tests. The starting and operating pressure ratios were improved by the presence of a second throat. Second-throat contraction ratio and length of minimum area had the greatest influence on the starting and operating pressure ratios. The limiting second-throat contraction ratio determined in this investigation agrees with published NASA results, although the ejector geometries were considerably different. The data presented in Reference 15 were obtained from ejector systems having the cylindrical duct diameter nearly equal to the nozzle exit diameter. In this investigation the ejector systems (Fig.28) studied were made up of an axisymmetric nozzle located concentric with a cylindrical duct having a diameter significantly greater than the nozzle exit diameter. 3.4.1 Effect of Second-Throat Location A typical variation of minimum cell pressure ratio and starting pressure ratio with second-throat location is shown in Figure 29. The range of second-throat locations which do not influence the minimum cell pressure ratio is bounded at the upstream end l(X/D d ) min l by the increase in minimum cell pressure ratio and at the downstream end l(X/D d ) max ] by the inability to start the ejector system.

120 The initial increase in minimum cell pressure ratio as the second throat was moved upstream of ( X / D d ) m i n is caused by the free jet impinging on the contracting portion of the second throat. This increases the static pressure rise through the impingement zone and results in an increase in the minimum cell pressure ratio 1 9 . The result is consistent with the decrease in minimum cell pressure ratio produced by a conical inlet which causes the static pressure rise through the impingement zone to decrease 12 . The value of ( ( X / D d ) m a x of each second-throat configuration was experimentally determined, based on ability to start the ejector system. Figure 29 presents the duct length l ( X / D d ) m a x ( X / D d ) m i n ] within which a second throat may be located without influencing the minimum cell pressure ratio. This duct length decreases sharply as the second-throat contraction ratio is decreased from 0.65 to 0.4. The optimum location of a second throat must lie within the range of duct length l ( X / D d ) m a x (X/Dj-Jj^jpl since the second throat must not influence the minimum cell pressure ratio. It is also necessary to locate the second throat at a position at which the starting pressure ratio ( p e x / P t ) I s a maximum. Figure 29 shows the starting pressure ratio to be nearly constant for all second-throat locations within the duct length range [ ( X / D d ) m a x (X/D(j) m j n |, with the maximum starting pressure occurring at or slightly downstream of the ( X / D d ) m i n location. However, for some configurations, a second optimum location existed further downstream. 3.4.2 Effect of Nozzle Total Pressure Level In Reference 17, the effect of nozzle total pressure on the starting and operating pressure ratios of ejectors without second throats is shown to be negligible although the minimum cell pressure ratio varies considerably. This is also true for ejectors using second throats, as shown in Figure 29, if the second throat is located where it does not affect the minimum cell pressure ratio. Since the free-jet impingement point is a function of the minimum cell pressure ratio, the optimum second-throat location is, therefore, a function of the nozzle total pressure level. 3.4.3 Effect of 0 s t on Starting and Operating Pressure Ratios Figure 30 shows the starting and operating pressure ratios obtained from ejectors equipped with second throats having inlet angles of 6, 12 and 18 deg and unequal minimum area lengths of less than one throat diameter. Based on the data presented in Reference 15, the inequality of the minimum area lengths is not a factor in this comparisoa The starting and operating pressure ratios were essentially independent of the second-throat inlet angle for inlet angles within the range of 6 to 18 deg. However, a small second-throat inlet angle is necessary to prevent boundarylayer separation, which has a strong influence on the optimum location of the second throat. 3.4.4 Effect of Second-Throat Contraction Ratio on Ejector Performance The variation of ejector performance with second-throat contraction ratio ( A s t / A d ) is best assessed by comparing the performance of the same ejector system without a second throat. Figure 31 shows the variation in the relative starting pressure ratio and the relative minimum cell pressure ratio with second-throat contraction ratio. All second-throat configurations improved (increased) the ejector starting and operating pressure ratios, and the improvement increased as the contraction ratio was decreased. A further increase of approximately 30 to 40 percent in the starting and operating pressure ratios can be accomplished by increasing the length of the minimum area of the second throat, as shown by the closed symbols in Figure 31(a). Only one second-throat ejector system was tested using a contoured nozzle. The improvement in the starting and operating pressure ratios shown in Figure 31(a) for this ejector system indicates that a second throat will produce a greater improvement in the starting and operating pressure ratios of an ejector system having a contoured nozzle than of one having an 18 deg conical nozzle. The criteria for determining the limiting second-throat contraction ratio for an ejector system can be defined as the minimum contraction ratio which causes no increase in minimum cell pressure ratio when the second throat is at the optimum location. The variation of the relative minimum cell pressure ratio with the second-throat contraction ratio shown in Figure 31(b) can be used to estimate the limiting contraction. In Reference 15, a limiting second-throat contraction ratio curve is presented for ejector systems having cylindrical duct diameters slightly larger than the nozzle exit diameter and long minimum area second-throat configurations. The data from the present investigation obtained using ejector systems having a cylindrical duct diameter much greater than the nozzle exit diameter agree very well with the limiting curve from Reference 15, as shown in Figure 32. However, the limiting curve from Reference 15 is not necessarily valid for second-throat configurations having very short minimum area lengths, as shown by comparing ejector configurations (contraction ratio 0.5) in Figure 32. Included in Figure 32 is the limiting second-throat contraction ratio determined by the well-known normal shock method. This limiting contraction ratio curve is shown to be very conservative for ejector systems.

121

3.4.5 Theoretical Analysis A method of estimating the operating pressure ratio of second-throat ejector systems is presented in Reference 20. The assumptions used in this method were: (i) The ratio of specific heats of the driving fluid is constant. (ii) All losses occur in the second throat. (iii) The Mach number at the entrance to the second throat is defined by the ratio of duct-to-nozzle throat area and isentropic one-dimensional flow. (iv) Sonic Mach number exists at the minimum area of the second throat. The operating pressure ratio is then the total pressure loss obtained by applying the continuity relationship for adiabatic one-dimensional flow between the nozzle throat and the minimum area of the second throat. The equation is
Pt ex /Pt "
A

*/Ast

A comparison of the calculated operating pressure ratio obtained by this method with the experimental results is presented in Figure 33. The maximum error of this method as the second-throat contraction ratio approaches the limiting value is about 30 percent except for second-throat configurations having a long .minimum area section, in which case the error is much smaller. A most unusual result of the investigation is the decrease in minimum cell pressure ratio caused by the presence of a second throat. The magnitude of this decrease is shown in Figure 31(b) to be a function of nozzle exit flow conditions, second-throat contraction ratio, and the length of the second throat. 3.5 Influence of Second-Throat Diffusers on Minimum Cell Pressure An analysis of the influences of second-throat diffusers on minimum cell pressures is presented by Panesci and German in Reference 21. According to Crocco in Reference 22, it is possible for the "dead water" region (or cell region) not to be isolated from downstream effects. Schlieren pictures in Reference 22 show clearly that shocks in the region of free-jet impingement do not reach the wall. Crocco points out that this is because of the presence of subsonic flows in the boundary layer along the diffuser wall and in the zone where the jet mixes with the gases recirculating in the cell region. Crocco further states that these subsonic regions are thin, and if their longitudinal extent is sufficient, a disturbance downstream would influence the pressure in the cell region. A disturbance by a second throat could cause either an increase or a decrease in cell pressure if the mixing zone is initially laminar; however, if the mixing zone is initially turbulent, only an increase in cell pressure would be anticipated. The increase in cell pressure can be explained when the mixing zone is initially laminar by considering a relation of the mass flows entrained and rejected in the cell region with the pressure differential (pj p c ) discussed in References 19 and 23. When the point of transition from laminar to turbulent flow in the subsonic region is downstream of impingement as shown in the following illustration, an initial increase in p- p c results in a temporary increase in the mass flow rejected into the cell region. This causes an increase in cell pressure to establish equilibrium.
Transition Point

'///Sc

////.
Stagnating Streamline Dividing Streamline Initial Effect of Transition Upstream of Impingement m >

Stagnating Inner Edge and Dividin * " of Mixing Streamline Zone Transition Downstream of Impingement (Steady State) me - m r

Mass Flow Transit! (m) Downstream of "I"

Transition Upstream of "I"

Pi - Pc

122 An increase in cell pressure will also occur when the peak static pressure (ps) at the jet impingement point is increased because of an increase in the gas turning angle as a result of a downstream disturbance. When the point of transition from laminar to turbulent flow moves upstream of free-jet impingement, turbulence moves upstream into the lower velocity portion of the mixing layer. If it is assumed that the mixing zone is laminar, this increases the velocity between the dividing streamline and the outer edge of the mixing zone such that the dividing streamline must move downstream of the stagnating streamline to satisfy continuity relations (see Reference 19). Initially, this causes only a slight increase in the mass flow entrained since the length of the mixing layer along the dividing streamline is not decreased appreciably. However, the rejected mass flow would be decreased because of the turbulence which energizes the fluid particles in the low velocity portion of the mixing layer and enables more of them to overcome the pressure rise through the impingement zone. Thus, the amount of gas reversed is reduced for a given p- p c , when the transition point moves upstream of jet impingement, and cell pressure must decrease to establish equilibrium, with a resulting decrease in the entrained mass flow. Although the manner in which transition is influenced by a second throat is not understood, the effect of transition on cell pressure provides an explanation of the decrease in cell pressure noted in Reference 18 for the low Reynolds number data. 3.6 Reynolds Number Effect on Ejector Performance The variation17 of minimum cell pressure ratio with nozzle plenum total pressure is of primary interest and is graphically presented in Figure 34. From these data, the following ejector characteristics may be noted: (i) The rate of change of minimum cell pressure ratio with nozzle plenum total pressure level became very small as the nozzle plenum total pressure level was increased. (ii) For each conical nozzle having a half-cone angle of 18 deg, one nozzle plenum total pressure existed at which the minimum cell pressure ratio was less than the minimum cell pressure ratio corresponding to any other nozzle plenum total pressure level. For ejector configurations equipped with contoured nozzles having 0 deg exit half-angles, experimental data were not obtained at nozzle plenum total pressure levels required to prove or disprove this phenomenon. The significant parameter involving nozzle total pressure level was found to be the unit Reynolds number at the nozzle exit times the nozzle throat diameter (Fig.35). For values of this parameter of less than one million, significant variations in the minimum cell pressure ratio occurred. 3.7 Influence of Pertinent Parameters in Ejector-Diffuser Performance Data from various diffuser model studies were analyzed24 to determine the influence of pertinent parameters on ejector-diffuser performance with and without ejected mass from the test cell. The parameters varied were conical nozzle area ratio, mass ejected from the test cell, number of nozzles, and the ratio of diffuser-to-nozzle exit area. The influence of these parameters on the contraction ratio and the starting and operating pressure ratios were investigated. The p c /p t increased with decreasing nozzle area ratio (for a given diffuser-to-nozzle throat area ratio). Mass ejection from the test cell resulted in a small variation in p c /pt with variations in Ad/A . The limiting secondthroat contraction ratio (A st /A d ) increased with increasing A d /A n e for the nozzle configurations tested. When A(-/Ane was equal to or greater than 6, the spacing of the second throat from the nozzle exit became very critical. The start and breakdown pressure ratio can be affected by only a small change in the spacing of the second throat from the nozzle exit. The position of the second throat for such configurations with a large value of A d /A n e (greater than 6) determines the amount of improvement in start and breakdown pressure ratio. The effect of nozzle area ratio is shown in Figure 36. The effect of boundary removal or parallel pumping and the number of nozzles on performance is indicated in Figures 37 and 38. The influence of unsymmetrical nozzles is presented in Figure 39. 3.8 Effects of Different Driving Fluids on Ejector-Diffuser Performance An investigation25 was conducted to determine if there is a difference in ejector-diffuser performance when using different driving fluids having the same average ratio of specific heats, and if the performance of the ejectordiffuser for one particular driving gas can be used to accurately predict the ejector-diffuser performance for another driving gas, both with the same and with different average ratios of specific heats. Some fluids have invariant ratios of specific heats with temperature over a wide temperature range, whereas other fluids have a very large variation in the ratio of specific heats with temperature. For most real gases, this variation of the ratio of specific heats with temperature is well defined. The five different gases used as the ejector-diffuser driving fluids were air, nitrogen (N 2 ), hydrogen (H2), argon (A), and helium (He). Two ejector-diffuser configurations consisting of different nozzles and different diameter diffusers were used in this investigation. Correlation between the data obtained and one-dimensional isentropic

123 relationship as given in Reference 26 for diffuser-to-nozzle throat area ratio, cell-tc-nozzle total pressure ratio, and ratio of specific heats, is shown. The primary purpose of this experimental investigation was to determine if, by knowing the ratio of specific heats of a particular driving gas, the performance of an ejector-diffuser could be predicted from the one-dimensional isentropic relationship as given in Reference 26. Only two ejector-diffuser configurations were used in the test. The important parameters of the configurations are:

Configuration

Diffuser Diameter D, in. 4.026 1.373

Ratios L/D 5.71 7.97 AneM* 18.00 10.76 A d /A* 73.06 20.40

en
18 7.58

3.8.1 Properties of the Selected Driving Fluids The average thermodynamic properties (R, 7, C p , and /i) at 1 atm pressure and ambient temperature for the five driving fluids used were taken from References 27, 28 and 29 and are presented in Table II. It was desirable to have a constant property gas (a gas with properties not variant with temperature and pressure, especially the ratio of specific heats). Taken from References 28 and 29 and presented in Figure 40 are the effects of pressure and temperature on the ratio of specific heats (7) for the driving gases: air, argon, helium, hydrogen and nitrogen. Air, nitrogen and argon at 0.1 atm pressure and helium at 1 atm pressure have ratios of specific heats which are practically independent of temperature even down to very low temperatures (between 100 and 200R), but 7 does vary with temperature for air and nitrogen at temperatures above 600R. For hydrogen gas at temperatures below 600R, the variation of 7 with temperature is quite large: from 1.40 at 600R to 1.66 at 120R for 1 atm pressure. Only a slight change in the temperature-ratio of specific heats relationship was noticeable in the low-temperature range when the pressure was reduced to zero. As much as 40 atm pressure alters the relationship of temperature and 7 for hydrogen or the other gases but little (see References 28 and 29). The hydrogen curve in Figure 40 indicates that, for stagnation temperatures around 600R, hydrogen does not behave as a constant property fluid when expanded through a convergent-divergent nozzle. The ratio of specific heats for hydrogen is a strong function of temperature. The static temperature of a constant property fluid drops as the area increases downstream of the nozzle throat in the divergent portion of a nozzle according to the isentropic relationships given in Reference 14, which are

AA
and

' * " 5
1+

(^TK ^-)
7-1

4(T+I)/(7-l)

T t /T -

M2

and are tabulated in Reference 26. When 7 is dependent on the local static temperature as in hydrogen gas expansion for stagnation temperature around 600R, the above equations are still valid, but 7 becomes a variable. Presented in Figure 41 are the T/T t versus A/A* isentropic expansions for gases with constant and variable ratios of specific heats (for thermal equilibrium flow). The 7 = 1.40 curve represents air and nitrogen, and the 7 = 1.67 curve represents helium and argon. Actually there is a small difference in 7 for helium and argon as presented in Figure 40. The dashed curve which starts approximately at the 7 = 1 . 4 0 isentrope increases with decreasing temperature ratio (T/T t ) for an increasing area ratio (A/A*) to the 7 = 1.67 isentrope and represents the isentropic thermal equilibrium expansion of hydrogen for a stagnation temperature of 500R. The dashed vertical lines represent the nozzle area ratio ( A n e / A * = 10.76) and duct-to-nozzle throat area ratio (A d /A* = 20.40). 3.8.2 Start and Breakdown Pressure Ratio Presented in Figure 42 for the two ejector-diffuser configurations are the start and breakdown pressure ratios for the different driving gases in relation to their respective theoretical one-dimensional, normal shock total pressure ratios ( p t y / P t x where x is upstream of normal shock and y is downstream). For the small configuration (configuration 2) all five gasses - air, argon, helium, hydrogen and nitrogen - were used as driving fluids. Air and nitrogen, which have almost the same ratio of specific heats (Fig.40) differed in p e x / p t at breakdown by approximately 10 percent. Only three (air, nitrogen and argon) of the five gases were used for driving fluids in configuration 1. The breakdown pressure ratio (p e x /Pt) for argon was 73.58 percent of Pt y /Pt x fo r 7 = 1.67 .

124 The remaining driving gases used in configuration 2 (argon, helium and hydrogen) gave p e x / p t ratios in percent of normal shock total pressure ratio as 94.25 percent for argon, 83.43 percent for helium, and 97.26 percent for hydrogen. The ratio for hydrogen was computed at the nozzle exit, based on a y = 1.67 as shown in Figure 4 1 . Actually the static temperature rises as the stream shocks down and approaches the diffuser exit. This rise in static temperature causes a decrease in y toward 1.40 for hydrogen. The normal shock total pressure ratio for y 1.40 is higher than for 7 = 1.67; therefore, for 7 = 1.40 the p e x / p t in percent of normal shock Pty/Pt x is 88.54 percent. This value is inline with nitrogen (90.22 percent) and is believed to be the more reliable value. The variation in the breakdown pressure ratios is considered to be the result of 7 variation with pressure and temperature and data scatter. An insufficient number of breakdown points were obtained for determining a mean value of Kp e x /Pt)act/(Pty/Ptx)l f o r each driving gas. 3.8.3 Cell Pressure Ratio Variation with Nozzle Plenum Total Pressure Presented in Figure 43 for the different driving gases is the performance variation in the cell pressure ratio (P c /Pt) * r various nozzle driving total pressures. Air and nitrogen were used for locating the optimum driving pressure from which driving pressures for argon, helium and hydrogen were determined. The driving pressures for the various driving fluids were related such that the nozzle throat Reynolds number per unit length for the different gases would be equal. To have dynamic similarity in test results with different fluids, Reynolds number continuity must be achieved. Reynolds number is defined as follows: Re = where p = fluid density, lbm/ft 3 V = fluid velocity, ft/sec u = dynamic viscosity, Ibm/ft-sec / = characteristic length, ft. Since, from the definition of Mach number V MvAygcRT , (26) Pv'
rl

(25)

and from the equation of state of a perfect gas P then Equation (25) becomes Re _ ( p / R T ) M y ^ R T . / M By arrangement, Equation (28) becomes Re _ / But p[Mv/(7gc/R)y/(Tt/T)]v/(l/Tt) M
7

i .

(27)

(zo)

f(m) = M J ^ (1 + i (

- 1)M 2 ] 1 / 2

(30)

and
which gives for Equation (30)

5- = 1 + y - ^ - M2 ,
T 2

(31)

f(ifa) = M j ^ .

It.

(32)

125 Substituting Equation (32) in Equation (29) gives Re pf(rn) (33) Rearrangement of Equation (33) gives Re Kp/pt)f(m)]pt

^VTT
The mass flow equation is ih

(34)

X*
Re

[(p/p,)f(m)] p t

7T7
m 1.

(35)

When Equation (35) is substituted in Equation (34), the result is (36)

A * y-7

For a particular fluid flowing through a choked nozzle, the nozzle throat Reynolds number per unit length becomes, from Equation (34), Re " Kpt , (37)

where

K =

M*0*-1 ,

ft/lbf.

(38)

where, for a particular gas, K is a constant as long as the total temperature remains constant and the nozzle remains choked. Then based on equal nozzle throat Reynolds number per unit length for the gases used in this investigation, the following equation is satisfied, and the stagnation pressure required to give similarity may be calculated as follows: (Kp t ) a i r = ( K P t ) N i = (Kp t ) A = (Kpt) H a = ( K p t ) H e . (39)

The nozzle throat static temperature was chosen for the basis of determining the Reynolds number per unit length continuity for the various gases. Based on the optimum driving pressure of approximately 300 lb/in 2 , as presented in Figure 43 for air and nitrogen, the desired driving pressures for argon, helium and hydrogen from the above equation are 300 lb/in 2 for argon, 859 lb/in 2 for helium, and 568 lb/in 2 for hydrogen based on a nozzle stagnation temperature of 470R for each gas. It is not known if these driving pressures are optimum for helium and hydrogen because in Figure 43 the cell pressure ratio was still decreasing at the 866 lb/in 2 pressure level. No pressure higher than 866 lb/in 2 was investigated for helium. Only one pressure level was investigated for hydrogen (651 lb/in 2 ). Since air and nitrogen have almost the same ratio of specific heats and are almost invariant with temperature below 600R (see Figure 40), the same minimum values of p c /Pt were expected for a given ejector-diffuser. For the same reasons, argon and helium should give the same minimum value of p c /Pt Presented in Figure 44 are the minimum values of p c /Pt obtained for the various driving gases in the two ejector-diffuser configurations in relation to their respective isentropic ratios. A difference in (p c /Pt)act/(Pc/Pt)isen for the two configurations was expected because they differed in (i) nozzle half-angle, (ii) nozzle area ratio, and (iii) duct-to-nozzle throat area ratio. Also for a given gas, any variation in 7 was expected to produce a variation in p c / p t . 3.8.4 Expansion Beyond the Saturation Limit For a fluid in a thermal equilibrium, a definite pressure-temperature relationship exists between phases, such as gas, liquid and solid. The relationship of pressure and temperature between the gaseous and liquid or solid phases is known as the saturation curve or phase boundary (the condition at which different phases of a fluid will exist in thermal equilibrium). Figures 45 and 46 present the phase boundaries (saturation curves) taken from References 28, 29 and 30 for the various fluids used in this investigation. The triple point is shown on the phase boundary curves (Figures 45 and 46) for the fluids except air and helium. (A state of a fluid at which the liquid, gaseous and solid phases can exist in thermal equilibrium is referred to as the triple point.) The superheated state (single gaseous phase) of a fluid exists on the right of the phase boundary curve, whereas a two-phase state (gaseous-liquid or gaseous-solid) exists on the left. This phase boundary curve can be thought of as a thermal equilibrium saturated expansion. The theoretical expansion of a fluid through a convergent-divergent nozzle gives a different curve known

126 as an isentropic thermal equilibrium expansion, constant entropy process. This isentropic expansion curve is presented in Figures 45 and 46 for the two nozzles used in this investigation. The isentropic expansions presented in Figure 45 were for a fluid having an invariant ratio of specific heats (7 = 1.40) and a fluid having variable ratio of specific heats (7 = 1.40 to 1.67) with temperature; the stagnation conditions were p t = 300 lb/in2 and T t = 490R . The end of the A n e /A* = 10.76 nozzle shown on the 7 = 1.40 isentropic expansion curve was on the right (gaseous phase) of the nitrogen phase boundary curve, whereas the A ne /A* = 18.00 nozzle extended to the left (gaseous-solid phase). As seen in Figure 40, 7 for both air and nitrogen is essentially constant with temperature. If thermal equilibrium could be maintained throughout the isentropic expansion of air or nitrogen from the stagnation pressure of 300 lb/in2 and temperature of 490R through the A ne /A* = 18.00 nozzle, an increasing two-phase gaseous-solid fluid would extend from the point where the isentropic expansion curve crosses the phase boundary curve as shown in Figure 45. If the 7 for hydrogen were constant, then, for the stagnation condition given above and in Figure 45, the fluid would remain in the single gaseous phase (superheated gas) throughout either nozzle. For a stagnation temperature no greater than 490R, Figure 40 shows that 7 for hydrogen is not constant in an isentropic expansion down to a temperature of approximately 95R; 7 ratio increases from approximately 1.40 to 1.67. The 7 for helium and argon is essentially constant with temperature. A constant 7 = 1.67 isentropic expansion from stagnation conditions of 228 lb/in2 and 452R is presented in Figure 46 with the helium, hydrogen and argon phase boundary curves. This isentropic expansion for both nozzles (A ne /A = 10.76 and 18.00) extends far into the thermal equilibrium, two-phase, gaseous-solid region for argon fluid. Helium is in the single gaseous phase throughout the expansion. If hydrogen is considered a constant 7 = 1.67 fluid and isentropically expanded from the p t = 228 lb/in2 and Tt = 452R stagnation condition, again the fluid expansion never reaches the hydrogen phase boundary curve. The hydrogen would remain in the single gaseous phase throughout the nozzle (A ne /A* = 18.00) as shown in Figure 46. If the stagnation pressure was increased from the p t = 228 while the stagnation temperature remained constant at T t = 452R , the isentropic expansion curve would move upward. If p t were increased to a sufficiently high value with T t held constant, the hydrogen isentropic expansion curve would approach and eventually cross the hydrogen phase boundary curve. If the stagnation temperature is increased from the Tt = 452R value while p t remains constant at 228 lb/in2, the isentropic expansion curves shift to the right delaying the onset of condensation for argon and allowing hydrogen and helium to be in a higher superheated state when leaving the nozzle (Fig.46). A higher hydrogen stagnation temperature (Fig.40) decreases the average 7 , or the 7 at the end of the nozzle, toward 1.40. This decrease in 7 would allow the isentropic expansion curve to move still farther to the right of the hydrogen phase boundary curve as shown in Figure 45. Air, nitrogen and argon undergo a phase change during the selected isentropic expansions from the stagnation conditions used. The fluid-phase-change position in the nozzle can be shifted downstream by either lowering p t or increasing T t from the values used. Helium has such a low saturation temperature (Figures 45 and 46) that the isentropic expansions from the above stagnation conditions result in a single-phase fluid throughout the expansion. The slope of a variable 7 (7 = 1.40 to 1.67) isentropic expansion curve from a specific stagnation condition is different from a constant 7 isentropic expansion as shown in Figures 45 and 46. Such an isentropic expansion curve results for hydrogen at the stagnation conditions investigated. 3.8.5 Delayed Condensation in a Rapid Expansion Flow with condensation has long been a subject of primary interest. J.L.French of RTF has made a study of literature concerning flow with condensation. When an isentropic expansion of a fluid through a nozzle results in the fluid properties leaving a single-phase (superheated gas) and entering a two-phase (gaseous-liquid or gaseous-solid) thermal equilibrium region, a strange phenomenon occurs. This phenomenon is known as "condensation" (phase change) as reported in References 14, 31, 32, 33 and others for condensable fluids such as steam. It has been observed 14 ' 33 that, in a rapid expansion of steam through a convergent-divergent nozzle, condensation is delayed to some point below the thermal equilibrium saturation line. This delay in condensation results in a metastable state known as a supersaturated state 32 . The supersaturation limit for steam is referred to as the Wilson Line and lies approximately 60 Btu/lmb below the line of saturation; air saturated with water vapor will expand isentropically to a volume 25 percent greater before condensation occurs33. Shown below are a sketch of a nozzle, the pressure distribution through the nozzle with condensation, and a temperature versus entropy diagram. An isentropic expansion from the superheated state through the thermal equilibrium saturation limit to a metastable state is depicted. As shown in the nozzle sketch, condensation does not occur when the fluid properties reach the saturation condition in the isentropic expansion but continues to expand apparently isentropically until the supersaturation limit is reached. At the supersaturation limit, a discontinuity in pressure similar to that of a normal shock occurs. This discontinuity is known as a condensation shock. The fluid from the condensation shock through the remaining portion of the nozzle is two phase.

127

Isentropic Expansion

Supersaturation Region Pi
Or

Region h Ps

u
3

a) H

ei U C D Da E

/
/ **'

' ' ' \


X

(Condensation
I Region

Entropy

X Distance
Typical characteristic for steam

Gases such as the ones used in this investigation have supersaturation limits similar to the Wilson Line for steam. Such supersaturation limits for air, nitrogen and helium are presented in References 34 through 45. References 36, 38 and 41 show that the supersaturation of nitrogen can be decreased by the addition of impurities such as C0 2 , water vapor, argon and oxygen. A fraction of a percent of C0 2 or water vapor will eliminate completely all supersaturation. Unlike steam, the onset of condensation in air35'37,40,42,43,44 a n d njtrogen34,36,37,42,4s does not produce condensation shocks. A gradual rather than an instantaneous pressure increase due to condensation was evident. Reference 35 indicates that a maximum of 55F of supersaturation may be obtained for air. The static pressure rise after the onset of condensation is due to the heat released by the condensing fluid37. This static pressure can be predicted by the saturation expansion theory presented in Reference 39. The theory is based on the assumption that the expansion through the nozzle follows the isentrope in the pressure-temperature plane until the fluid saturation curve is intersected. Further expansion then follows along the fluid saturation curve. Since there is a static pressure rise after the onset of condensation, it is desirable to have the condensation limit delayed, such as when the fluid expands to a high degree of supersaturation. A limiting static pressure range (about 0.077 lb/in2) above which air supersaturation will not occur is reported in Reference 35. The Mach number is lower and static pressure is higher at the exit of a nozzle for a condensing fluid (two phase) than for a noncondensing fluid (single phase). Since an ejector-diffuser duct is larger than the nozzle exit, there is an additional expansion from the nozzle exit to the diffuser duct. Even if a single-phase fluid leaves the nozzle, fluid condensation could start in the free-jet upstream of the impingement on the duct which would result in a larger value of p c /p t than that desired. The thermal equilibrium saturation curves for hydrogen and helium were not crossed by the isentropic expansion curve for the stagnation conditions investigated. Since hydrogen and helium pumped essentially the same value of p c /p t (Fig.44) in the same configuration and remained in the gaseous phase throughout the expansion, 7 is, for practical purposes, the only gas state condition on which free-jet diffuser performance depends. The isentropic expansion curve for argon crossed the thermal equilibrium curve and extended far into the twophase fluid region (Fig.46). This two-phase fluid expansion resulted in a high degree of condensation even when a highly supersaturated state had been reached. The high degree of condensation resulted in a lower Mach number and higher static pressure at the nozzle exit than would have existed in the absence of condensation. Therefore, the minimum values of p c /p t increased as a result of liquefaction. This result is shown in Figure 44.

128 REFERENCES

1. Kroll, A.E. 2. Payne, P.R.

The Design of Jet Pumps. Chemical Engineering Progress, Vol.1, No.2, 1947. Steady State Thrust Augmentors and Jet Pumps. Peter R.Payne Inc., Rockvale, Maryland, March 1966. Jet Ejectors and Augmentation. NACA Advanced Report, 1942.

3.

Morrison, R.

4. Goethert, B.H., Taylor, D. 5. Harris, G.L.

High Altitude Testing of Propulsion Systems. (Zeitschrift fur Flugwissenschaften 8, Jahrgang, July 1960, Heft 7), Tullahoma, Tennessee, USA. Steady-State Ejector Thrust Augmentation. 22nd-26th April, 1968. VKI-LR-7 Short Course on Ejectors,

6.

Flugel, G.

The Design of Jet Pumps. NACA TM 982, (Berechnung von Strahlapparaten) VDI-Forschungsheft 395, March-April 1939, pp. 1-21. Ejector Experiments. National Gas Turbine Establishment, Pyestock, Hants (Great Britain), Report No. R.151, 1954. Computation Curves for Compressible Flow Problems. J.Wiley and Sons, New York.

7. Lewis, G.W.E., Drabble, J.S. 8. Daily, C.L., Wood, F.C. 9. Korst, H.H., Chow, W.L.

Compressible Non-Isoenergetic Two-Dimensional Turbulent (p r = 1) Jet Mixing at Constant Pressure. University of Illinois Engineering Experiment Station, ME-TN392-4 (AD211328), January 1959. Theoretical Base Pressure Analysis of Axisymmetric Ejectors without Induced Flow. AEDC-TDR-64-3 (AD428533), January 1964. An Investigation of Ejectors without Induced Flow, Phase I. (AD229860), December 1959. AEDC-TN-59-145

10. Bauer, R.C.

11.

Barton, D.L., Taylor, D.

12. Taylor, D., et al. 13. German, R.C., Bauer, R.C. 14. Shapiro, A.H.

An Investigation of Cylindrical Ejectors Equipped with Truncated Conical Inlets, Phase ll. AEDC-TN-60-224 (AD252634), March 1961. Effects of Diffuser Length on the Performance of Ejectors without Induced Flow. AEDC-TN-61-89 (AD262888), August 1961. The Dynamics and Thermodynamics of Compressible Fluid Flow, Vol. 1. The Ronald Press Company, New York, 1958. Experimental Study of Zero-Flow Ejectors Using Gaseous Nitrogen. March 1960. NASA-TN-D-203,

15. Jones, W.L., et al. 16. Latvala, E.K.

Spreading of Rocket Exhaust Jets at High Altitudes. AEDC-TR-59-11 (AD215866), June 1959. Some Reynolds Number Effects on the Performance of Ejectors without Induced Flow. AEDC-TN-61-87 (AD262734), August 1961. TTie Effect of Second Throat Geometry on the Performance of Ejectors without Induced Flow. AEDC-TN-61-133 (AD267263), November 1961. Research on Transonic and Supersonic Flow of a Real Fluid at Abrupt Increases in Cross Section (with Special Considerations of Base Drag Problems). Final Report. University of Illinois, ME-TR-392-5, December 1959. Phenomena in Supersonic Diffusers. No.43, November 1947. TR No.F-TR-2175-ND GS-AAF-Wright Field

17. Bauer, R.C., German, R.C. 18. Bauer, R.C., German, R.C. 19. Korst, H.H., et al.

20.

Ramm, H.

21. Panesci, J.H., German, R.C.

An Analysis of Second-Throat Diffuser Performance for Zero-Secondary-Flow Ejector Systems. AEDC-TDR-63-249 (AD426336), December 1963.

129 22. Emmons, H.W. Fundamentals of Gas Dynamics, High Speed Aerodynamics and Jet Propulsion. Vol.111, Princeton University Press, 1958. Section B, "One-Dimensional Treatment of Steady Gas Dynamics", by L.Crocco, p.291. Investigation of Separated Flows in Supersonic and Subsonic Streams with Emphasis on the Effect of Transition. NACA Report 1356, 1958. Influence of Pertinent Parameters on Ejector-Diffuser Performance with and without Ejected Mass. AEDC-TDR-64-134 (AD602770), July 1964. Comparison of Diffuser-Ejector Performance with Five Different Driving Fluids. AEDC-TDR-63-207 (AD420813), October 1963. Gas Flow Tables. GM-TR-154, March 1957. Kent's Mechanical Engineers' Handbook, Power Volume. John Wiley and Sons Inc., New York. Tables of Thermal Properties of Gases. NBS Circular 564, US Department of Commerce, 1955. Hydrogen Handbook. AFFTC-TR-60-19, April 1960.

23. Chapman, D.R., et al. 24. Hale, J.W. 25. Hale, J.W.

26. Wang, C.J., et al. 27. Salisbury, J.K. 28. Hilsenrath, J., et al. 29. Arthur D.Little Inc. 30. Jorgensen, L.H., Baum, G.M. 31. Obert, E.F. 32. Kiefer, P.J., et al. 33. Kennan, J.H.

Charts for Equilibrium Flow Properties of Air in Hypervelocity Nozzle. NASA TN D-l333, September 1962. Thermodynamics. McGraw-Hill Book Company Inc., New York.

Principles of Engineering Thermodynamics. John Wiley and Sons Inc., New York (Second Edition). Thermodynamics. John Wiley and Sons Inc., New York. Experimental Determination of Limit of Supersaturation of Nitrogen Vapor Expanding in a Nozzle. Journal of Heat Transfer, Vol.83, SeriesC, No.l, February 1961, pp.27-32. Air Condensation in a Hypersonic Wind Tunnel. AIAA Journal, Vol.1, No.5, May 1963, pp. 1043-1046. Experimental Saturation of Gases in Hypersonic Wind Tunnels. Memorandum No. 10, California Institute of Technology, Contract No.DA-04-495-Ord-19, 15th July, 1952. The Condensation of Nitrogen in a Hypersonic Nozzle. Journal of Applied Physics, Vol.23, No.10, October 1952, pp.1089-1095. Effects of Impurities on the Supersaturation of Nitrogen in a Hypersonic Nozzle. Memorandum No.7, California Institute of Technology, Contract No.DA-04-495-Ord-19, 1st March, 1952. Condensation of Nitrogen in a Hypersonic Nozzle. Memorandum No.6, California Institute of Technology, Contract No.DA-04-495-Ord-19, 15th January, 1952. Condensation of the Components of Air in Supersonic Wind Tunnels. Princeton University, Aeronautical Engineering Laboratory, Report No. 127, March 1948. Effects of Impurities in the Supersaturation of Nitrogen in a Hypersonic Nozzle. Heat Transfer and Fluid Mechanics Institute, 1952, pp. 125-137. The Effects of Air Condensation of Properties of Flow and Their Measurement in Hypersonic Wind Tunnels. Memorandum No.8, California Institute of Technology, Contract No.DA-04-495-Ord-19, 15th June, 1952. Theoretical and Experimental Investigation of Condensation of Air in Hypersonic Wind Tunnels. NACA TN 2559, November 1951. Results of Recent Hypersonic and Unsteady Flow Research at the Langley Aeronautical Laboratory. Journal of Applied Physics, Vol.21, No.7, July 1950, pp.619-628. The Supersaturation of Nitrogen in a Hypersonic Wind Tunnel. Journal of Applied Physics, Vol.23, No.l, January 1952, pp.40-43.

34. Goglia, G.L., Van Wylen, G.J. 35. Daum, F.L.

36. Arthur, P.D., Nagamatsu, H.T. 37. Willmarth, W.W., Nagamatsu, H.T. 38. Arthur, P.D., Nagamatsu, H.T. 39. Nagamatsu, H.T., Willmarth, W.W. 40. Charyk, J.V., Lees, L. 41. Arthur, P.D., Nagamatsu, H.T. 42. Grey, J., Nagamatsu, H.T.

43. Stever, H.G., Rathbun, K.C. 44. Becker. J.V. Faro, T., et al.

45.

130 TABLE I Summary of Configurations and Experimental Results

ne

Pc

tifotrJT ,.
Air Steam

NOTE: Nozzle Exit Diameter = 4.16, d n = 18 deg

A*

Duel Diameter
D A in i-*(l> " i -

Ad A*

Cone Angle,
"conc>

Cone
Prisitinn

deg 12 18 24 18 24 30 24 30 36 12 18 24 18 24 30 24 30 36 12 18 24 18 24 30 24 30 36 12 18 24 18 24 30 36 12 18 24 18 24 24 24 24 24 24

X,in. 1.4375 1.250 1.1875 2.3750 2.250 2.1875 3.000 2.375 2.875 1.625 1.500 1.500 2.5625 2.3125 2.1875 3.125 2.625 3.000 1.75 1.6875 1.625 3.25 2.8125 2.750 4.125 4.000 2.000 2.0625 1.750 3.5625 3.4375 2.5625 3.5000 2.500 2.4375 2.0625 3.8750 3.500 5.125 1.625" 1.625** 1.560** 1.500**

Total Pressure P lb/in2


38.5 38.6 39.2 36.8 36.8 36.4 40.0 40.5 40.0 34.0 34.0 34.4 40.5 43.3 40.9 36.7 43.0 42.0 43.8 43.4 43.3 42.8 44.5 42.4 45.0 45.0 43.7 42.6 45.2 45.6 44.5 44.9 45.1 45.8 44.9 44.0 44.9 44.4 44.9 43.4 43.3 43.5 43.4

(h)
\Pt7opt
.00402 .00296 .00252 .001790 .00137 .00165 .000680 .000529 .000485 .003953 .00355 .00332 .00110 .00103 .00113 .000474 .000370 .000318 .000994 .000758 .000646 .000607 .000630 .000981 .000172 .000560 .000531 .000545 .000360 .000382 .000674 .000797 .000184 .0005078 .000474 .000813 .000237 .000588 .0001507 .000648 .000850 .001160 .001338

Pc
(Pc)opt

Cone Position
X.in.

Total
PfSSUf

(h)
_
.00229 .002234 .002883 .00708 .00663 .00524

Pc
(Pc)opt

Pt, lb/in2 \Pt/opt

3.56

5 . 0

10.8

18.0

25.0

10.8

6 6 6 8 8 8 10 10 10 6 6 6 8 8 8 10 10 10 6 6 6 8 8 8 10 10 10 6 6 6 8 8 8 10 6 6 6 8 8 10 6 6 6 6

7 , 6
13.18

20.75

10.68

18.49

29.15

23.0

39.82

62.80

2.560 3.4797 4.087 3.0335 3.964 3.290 4.0735 5.226 5.653 1.481 1.648 1.762 2.790 2.981 2.7168 3.404 4.324 5.000 1.828 2.382 2.785 1.770 1.714 1.102 3.294 1.000 2.283 2.200 3.361 1.7368 .9851 .8250 1.9440 1.922 2.085 1.2098 2.208 .8983 1.80 2.870 2.153 1.578 1.360


2.125 2.000 2.500 1.375 1.375 1.250

_
22.0 23.0 23.0 28.0 28.0 28.0

2.7074 2.7623 2.2664 1.6667 1.7798 2.3282

2.750 2.125 2.625

32.0 31.5 33.0

.00160 .001478 .00238

2.450 2.6049 1.7282

1.125

114.7

.00129

2.7907

2.875 2.5000 3.000

68.0 69.0 67.0

.000570 .000941 .001043

2.5649 1.6876 1.4717

38.60

66.40


2.625

104.6 53.30

92.40 145.70 23.0

2.500

202 124.6

.000240

2.8042

.000203

2.4975

24 deg cone with different radial height.

TABLE II Gas Constants

Driving Gas

Specific Gas Constant ft - Ibf sIbm - R 53.321 38.679 386.057 55.147 766.369

RAir Argon (A) Helium (He) Nitrogen (N 2 ) Hydrogen (H 2 )

Average Ratio of Specific Heats above 0F, 7

Average Specific Heat at Constant Pressure above 0F Btu P ' Ibm - R 0.242 0.124 1.240 0.246 3.395

Dynamic Viscosity at 70F Ibm ft sec 12.222 x Ifr* 15.184 x i r * 13.017 x 10"* 11.820 x ifr* 5.945 x 10"*

Mach Function Based on Average 7 above 0F

[tA
\Pt
/M=1

1.401 1.667 1.66 1.40 1.41

0.5319 0.6624 0.2094 0.5224 0.1406 0.1486 for 7 = 1.66

\(7 + l)/(7-D

(IHM=I

1T(T4T)'

132 Or 14.7

20. 000

40,000

<
01

60 000

80, 000

100, 000

100 200 300 Test Cell Airflow, lb/sec

(a)

Basic plant

r
20, 000 -

14.7

10 000

I
60 000 on
SA1

80, 000 -

100,000

Exhauster with Ejector 1:10

100

200

300

400

Test Cell Airflow, lb/sec

(b) Plant plus exhaust gas diffusers

Fig. 1 Typical engine and facility exhaust performance map

133

0.2

0.5

1.0
Mach Number, IVI Fig. 2 Mach number in a constant area channel

134

Subsonic DiffuserSupersonic Diffuser/ Simulated Rocket Nozzle,;

^High Pressure Air and Steam Supply Line


Fig.3 Typical ejector configuration

0.100 0.060
Pc/Pt 0.020
i t

H
R Jj
0 , i i

(a) p c /p t versus p ex /Pt


c

0.010 0.006 0.01 0.02

b
0.06 0.10 Pex/Pt

_j

0.2

18.0
D

16.0 14.0 12.0


-^
a

PeA

10.0 8.0 6.0


4.0 2.0 0

C
i

(b) Pex/Pc

versus

Pt/Pex

E -

t
4

12

16

20

24

28

Pt'Pex
Fig.4 Characteristic ejector performance curves

135

60

100

200

Fig.5

p c /p t versus A,j/A

A rt

ne'M

/A*

Propellant Type Solid Liquid Solid Solid Solid Liquid Solid

o
a * o o
A

5.5 8.0

10.0 11.0 (4 Nozzle Cluster) 18.0 20.0 25.0 Model Data with Steam Model Data with Air Isentropic One-Dimensional Compressible Flow Theory, y - 1.2

10

20

60 100
A d /A'

200

600

1000

Fig.6

Comparison of full-scale and model ejector performance

136

Duct Size, in. A M 200.0 100.0 60.0 /A* m ne 8

10

3.56 o -o 5.0 a -a +a 10.8 o -o +o 18.0 a - 25.0 A w. +A Open Symbols, Air Liosea bymoois , o i c a n i

Pt'Pex 20.0 10.0 6.0 4.0

10

20 Ad/A*

60

100

200

Fig.7

pt/Pex versus A d /A*

1.000 . _
A

Duct Size, in ne * 3.56 5.00 10.8


18.0
/A

0.600

6
D

8 - - -o
-*>

10 * I -*

,
/

J"
\ ^ A

\.

'

/ J/

0.200 0.100 0.060 Pex'Pt 0.020 0.010 0.006

4
&

25.0

-^

Open Symbols, Air Closed Symbols, Steam


r !

0.002 0.001 0.0010.002


_

0.0060.010.02

0.06 0.1

0.2

0.6 1.0

Fig.8

Pex/Pt

versus

isentropic one-dimensional normal shock total pressure loss ratio

137

Subsonic D i f l u s e r Super sonic Oilluser Simulated Rockel Nozzle-

^B
Positioning Rod -High Pressure A i r or Steam Supply l i n e

m^V/

WT
To Exhaust Machines

Conical Inlet

-Elleclive Dill user L e n g t h -

Fig.9

Typical ejector configuration

100 x
*

l u

10 * *

1 1 1

A d /A*-23.0
fift

Ane/A"-10.8 Dd 6 in.
9

40

cone"24d/ / /

-7- - Z -9 Z L
/

20

Open Symbols, without Conical Closed Symbols, with Conical Inlet

10 6 4 Pc'Pt / 2 / / 1.0 0.6 0.4 / / t

y _,.
y

TffT Pex'Pc '

1.0

*'

/ / /
y

/
0 t.
7

J \
;

y
/ /

4/ / /

-/
1

0.2

PeA" 30_ ^f
./

90 / /

n 1

0.1

0.2

0.4

0.6

1.0 Pex'Pt

10 x

-2

Fig. 10 Performance of typical diffuser configuration with and without conical inlet

138 200 x 10"*-

"1.
v
HL \ \

100

6U 40
Duct S ze, i n .
A

II

V J

r*s\
-*>

ne/A*

10

>jk
\ ',

Pex'Pt

20

10 6 ~ _ _

3.56 o a cf 5.0 o cf cr 10.8 o c o 18.0 > a a 25.0 c. a a Open Symbol, Smallest Conica i Inlet Angle Half Closed Symbol, Intermed ate Conical Inlet Angle Closed Symbol, Largest Conic al Inlet Angle
-ex'i-i rui ouiniyuidiiuns vmnuui

> l

Conical Inlets (Ref. 11)


i i i i i 1 1_ i

10

20 A d /A-

40

60

100

200

Fig. 11

Comparison of ejector starting pressure ratio with Aj/A* for configurations tested with and without conical inlets (cones located at optimum axial location)

7.5
O^_,

c-f c o
Q-

7.0 6.5 6.0 5.5 5.0 4.5


4.0 3.5

a>

.2 x

<
3

e c o A-*

e "S.
o a)
x:

_) - o o
mm 3 a>

,, . 8

*-r

a>
A-*

o
C 1

3.0 2.5 2.0 1.5 1.0


12 18 24 18 24 30 24 30 36 Cone Half-Angle, 8 c o n e . deg

r O O

a>

s
CO

-:
1_

s =3
SA, iAi

OLC

CJJ

a.
cu c_)
^ J

yi

!i)

o B

Fig. 12

Summary of optimum improvement in diffuser performance resulting from use of conical inlets

-
20 x in'*-*
w

w*-

3L.
9

cone18 24 30 Dd - 8 in. Ane/A'-3.56 A d /A' 13.18


de

i
10
9

0 0
de

20 x 10 cone9

12 18 24 Dd 6 in. Pc'Pt

6 : 4 e-r-o 3 >-o-=:

Ad/A* - 7 . 6 A /A* 3.56

1 1
k
^

r
2

1
0 0.5 1.0 Distance from Nozzle Exit to Cone Exit, X/D (a) "cone = !2. 18 and 24 deg

ne

0.5 l.o Distance from Nozzle Exit to Cone Exit, X/D n e <b) 'cone =
18

1.5

> 24 and 30 deg

Fig. 13 Effect of axial position of conical inlets on minimum test cell pressure

m
X iu

' ^ "

. ^ .

40
=a-f ^C-T=

1
20
Pc'Pt

i* = = = = m

1
-*
1
24

/ 7

\f

9
- 10 i n .

10

A /A* -3.56 H nem - 20 75 "d A^/A*


e

cc,ne24 30 36

de

o
9

0.5 1.0

1.5
ne
Fig. 14

0.25

0.50

0.75

1.00

Distance from Nozzle Exit to Cone Exit, X/D (c> "cone = Fig. 13 > 30 and 36 deg

Distance from Nozzle Exit to Cone Exit, X/D n R

Effect of radial height of conical inlet on axial position and minimum test cell pressure

Concluded

100 x 10

400 x 10'4

Pc'Pt

0.5 1.0 Distance from Nozzle Exit to Cone Exit, X/D n e

40.0 VA'

100.0

200.0

Fig. 15

Results obtained using a three-step conical inlet in the 10 in duct with the A n e /A*= 5.0 nozzle

Fig. 16 Summary of minimum values of P c /p t for configurations tested without conical inlets and with best conical inlets

142

c,"JI

J j -A h ^

Pt D'-i Pne

D-i

7CT

4 deg

(a) With subsonic diffuser Pex D-l

Pt D'-, Pne

IJ

irT
Pc

D
Typical ejector configurations

30 i n .

(b) Without subsonic diffuser Fig. 17

^cT

ID
1

Q-

13
ISS Ul CO 1

Pne'Pf
"o
=J

t= _cu Dcu

(3) Nozzle and Ejector Started

(2) Nozzle Started Elector Unstarted Exhaust Pressure ^ex Nozzle Plenum Total Pressure ' pt

Fig. 18 Typical ejector starting phenomena for constant nozzle plenum total pressure

143

A - Start ngp ex'Pt B - Operating Pex/Pt

Pc'Pt
CL

'
to
X

8::
O)

B
Or

c
'l/l

1,
A

to
[,i

8
ba

u
c

L/D~ 1.0

AB L/D - 8.1

PeA
Fig. 19 Typical effect of L/D on ejector starting characteristics of contoured nozzles at constant nozzle plenum total pressure

0.0400 * 0.0200 A n e /A' - 25, 8 n 18 deg, A d '/A*-53.7 Jet mpingemeiit 0.0100 0.0060 Pw'Pt o 0.0040 / ) J
PPV r
CA

ex
Pt r

**- r t

tU

-1

"-|

With Subsonic Diffuser

0.0020 Pc 0.0010 - K J 0.0006

Pt

^t^J
* U - L - ^

Pex

PtZ

Without Subsonic Diffuser

c)

12

15

23

24

28

L/D
Fig.20 Comparison of duct wall static pressure ratio distribution just prior to ejector unstart with and without subsonic diffuser, L/D = 9, p t = 45 lb/in2

144

0.400
O D

6 n , deg 18 0

0.200 ^

0.100
0.060 0.040 Pex'Pt 0.020

\ s
X"

.
Q3y

fheoretical Normal c ,hock

Pex'Pt*" V V n s

1
Unstarted

K- 0 . 8 8 ^

Reg on

\5"
MS LN

0.010 0.006 0.004 0.002 6

Start oH
Rpninn

\ i \

s\

10

20

40

60
Ad'A
6

100

200

400 600

(a) With subsonic diffuser

10

20

40

60
Ad/A<

100

200

400 600

(b) Without subsonic diffuser Fig.21 Ejector pressure ratio required for starting; L/D > 8

l.UU

0.60 0.40

With Subsonic Diffuser o


D

Without Subsonic Diffuser

0.08 Parametei PcPt Pne'Pt


;

0.06

^i
/

V
A ^

T
^

A 0 20 A^-"V

V-Vopei

0.04

MVslar t
>

a *

~fy

0.02

With Subsonic Diffuser o

Without Subsonic Diffuser

Parameter Pc'Pt Pne'Pt

p/pt

0.10

s
P/Pt

V-Voper
Pex'Pt'start
* -

0.01 0.006

0.06 0.04

fl_
0.02

rf

0.004
0

0.002
0.01

0-

rv
2

g
4

i> 6

o 8

10

0.001 0

Diffuser Length-to-Diameter Ratio, L/D


(a) A ne /A* = 3.63 , Ad/A* = 7.7 , 0 n = 18 deg Fig.22

4 2 ( 3 Diffuser Length-to-Diameter Ratio, L/D


(b) A ne /A* = 10.8 , Ad/A* - 23.3 , 0 n = 18 deg

10

Effect of diffuser length on ejector performance characteristics (p t = 45 lb/in2)

0.0400

0.0200 0.04

0.02

-r
/ /
With Subsonic Diffuser o Without Subsonic Diffuser

sv+
Parameter Pc'Pt

0.0100

0.0060 0.0040 P/Pt 0.0020 With Subsonic Diffuser o o A 0.0006 0.0004 V Without Subsonic Diffuser A

0.01
0.006 P/Pt 0.004

Parameter Pc'Pt Pne'Pt 'PexHper


'Pex'-Vstart

D A V

W-H

foex'-Voper
(

0.0010

Pex/Pt,start

0.002

0.001 0.0006 21
Diffuser Length-to-Diameter Ratio, L/D (c) A ne /A* = 25.0, A d /A* = 53.7, 0 n = 18 deg Fig.22 Continued

22

0.0002 L 0

9 '

-V 20

21

Diffuser Length-to-Diameter Ratio, L/D (d) A ne /A* = 23.68, A d /A* - 44.7 , 0 n = 0 deg Fig.22 Concluded

147

1.00
ac

0.80
SAS

cu
JAC

0.60

0.40 -

s
a o.2o
o

AIU

Operate o D
K

Start

n-

de

0 18 2 lns

l- ( Pex / Pt ) exper / ( Pt ? / (lV

"5
o

8 12 Diffuser Length-to-Diameter Ratio, L/D (a) With subsonic diffuser

16

3.

V
*y.

I
.JL. o

r
v

.00

o"

= 0.80
O-

/ r 0.60

il
/
/ i1

/s

< I
3 0.40
B
V-

I'

I fm
j

ft
/ /
' Operate o D
K

JZ F

s
o

k
jr

1 1
t

Start

n- de*0 18

0.20

2"VPl,exper/,p2/pl1,ns

V s
i

O C_)

8 12 16 20 Diffuser Length-to-Diameter Ratio, 17D (b) Without subsonic diffuser

24

Fig.23

Variation of starting and operating pressure ratio correction factor with diffuser length-to-diameter ratio, L/D

148

50

40

1T3

30

3
SJl

20
cn C
ro
A~r

Nozzle

Configuration

US

10

NASA Data I o (Ref. 15) )

15-deg Conical Without Subsonic Diffuser 15-deg Conical With Subsonic Diffuser Values with Subsonic Diffuser Based on K 1- Fig. 23a Values without Subsonic Diffuser Based on K2, Fig. 23b

10

12

14

16

Ejector Overall Length Parameter, (L/D + LJD) Fig.24 Comparison of estimated starting pressure ratio and NASA data1

fT 1.
ra 0. a>
i

SSS tys

0.80 -

cu
V-

Q.
SA.

SZ CO

0.60

l= ra 0.40
o

s
3
o

0.20

o o Diffuser Length Parameter, L/D - X/D (a) With subsonic diffuser Fig.25 Variation of starting and operating pressure ratio correction factor with diffuser length parameter, L/D X/D

"1
*

HXK
.*

r
h

149

.00
ro

a.
0}
CO 1

"CD

G 1

-> <

J w
o

5 0.80
CL

-*. 0.60

/ /

If- /

XI

11
/
f

cn

1
I
0. 40 u

1 1
i

r
/ / T

/
/

1
Operate o Start
e

g
o
c o

n-

de

0. 20 -

o 18

V ( Pex / Pt , exper /(p 2 / Pt 1 , ns "


L

o o

12

16

20

Diffuser Length Parameter, L/D -X/D


(b) Without subsonic diffuser Fig.25 Concluded

0.40

0.20

A n e /A-3.63-

0.10
A n e / A - 10.8

Pex/Pt - %
0.04

-D

Ane/A*-25.0

-o0.02

0.01
10 20 Pt- P
(a) Fig.26 L/D = 9.0
sia

30

40

50

Starting and operating pressure ratio variation with total pressure

150

0.40

(b) L/D = 1.6 Fig. 26 Concluded

U.LKtU

A n e /A e - 3.63 0.020 Ad/A*-7.7 6 n 18 deg Pc'Pt 0.010

0.006

n. no4

10

20 Pt. Psia

30

40

Fig.27

Typical cell pressure ratio variation with nozzle plenum total pressure

151

Ejector Driving Nozzle

Pex^ ' { * To Exhaust Machines High Pressure Air Supply Line Packing Gland Fig.28 Typical ejector configuration

c
"^ ro

^
X cu

08

A d /A' 22.8 A /A* H n e m 10.85 e st 18 deg ^ = 40 ps l ^


i,
{)

0.2
CJ.

g. . o. 07
n

6.02 in.
(X/DHU^V"

a nj C Q
-

<u U.
3
Q) U.

06 05
rvi

^ /~Pt~ ^J_t
i*

d max
JSI

C
^

i/5 o.

</*Vmin_

"(X ' u d 'ma X

j
1

(a) Variation of starting and operating pressure ratio * 0.006 < ; ,


a
I

\B 0.005
on
3
iy>
sS> CO

ITJ

r
-(X/l^
0-.
I

tp^ ~ 20 psi
- -

XI J d'max
1 V

0.004

f 1 k1
(S-rc

1
-min 1

1 l j

Q- 0.003
CL)

(X/D d 'max

o
3

0.002

i.

0.001 0

i 0.5

n (X/

o-c

r
,i d ' mir i

P t ~ ' Opsi

1 1

n u

F =

P4 ^ 0

***

^ - -

xi
4.0 4.5

1.0 1.5 2.0 2.5 3.0 3.5 Second-Throat Location, X/Dd

(b) Variation of minimum cell pressure ratio Fig.29 Effect of second-throat location and nozzle plenum total pressure level on ejector performance

2.2 . i> 2.0 0.40 - -..

rt

0
.

A /A* nem 3.627


i n si/is

Ad'A* 7.66 S 3 ?n

9 n . deg 18
is IR

Dd - 6.02 in. A d /A n e -2.08 A s t /A d -0.654 Second Throat at Optimum Location O A ne /A- - 3.627

1.8
CD Q,

?s nm

tn 1.6

__,. o 23.684 45.79 ,_ DH -6.02 in.


c

2r
3 1.4 1.2 1.0

Flagged Symbols, 9st -18 deg


Liu*>cu oymuuii, VULM,* - o.u

X o>

0.20

ly,

cS
i_ Caen

(a)

0.10
1-6 0.06

Variation of relative starting pressure ratio with second-throat contraction ratio

1
s.
o
o

e/A-

10.848 n
i_>

^1.4
1/1

ro

$1.1
0.04 1.0
A I

La
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4

A n g/A*- 25.000

0.8 Second-Throat Contraction Ratio, A st /A d

0.02 5.0 10.0 15.0 20.0 (b) Second-Throat Inlet Angle, 9 st , deg Fig.30 Effect of second-throat inlet angle on ejector starting and operating pressure ratio

Variation of relative minimum cell pressure ratio with second-throat contraction ratio

Fig.31 Comparison of ejector performance with a second throat at optimum location with performance without a second throat

153

Ane/A o A o 3.627 10.848 25.000 23.684 Dd -6.02

Ad/A* 7.66 23.30 53.70 45.79 in.

0 n , deg 18 18 18 0

1.4
o

<T 1.2

Flagged Symbols, (L/D) st -8.0 No Second Throat-

k
2 5

LU

^ ^

"1 0.8
0.6

SVN.
k.

t/1
^
V

ClOSe^ sv/mtink ^ L _

't

Inrroacorl n In.

TH

\\

o
*N

1 L
/
1
^V1

,,

L l f l l i i i n y v^un l a c n u n r\auu hv "shrvk Method uy Nfii"mal i

rt

r' f t r .

S | 0.4
"O

0.2
to

7 M.imiting Contractor
.atio from Ref. 15 F

0
1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 One-Dimensional Duct Mach Number, i\.d, Based on A d /A'
Fig.32 Comparison of experimental results with limiting second-throat contraction ratio curve

1.4 1.2

^nf

<1
73
X

1.0

LJ

#CL 0.8
X

/ . n J lV r)
ci t. <>

A H //>

W jeq

x a>

0.6 0.4

3.627 7.66 18 10.848 23.30 18 25.000 53.70 18 23.684 45.79 0 Flagged Symbols, ( L/D)st - 8.0
r*i
L

.. A no i n

0.2 0.4

1
0.5
0.6 0.7

'd
0.8

IN.

1
0.9

1.0

1.1

Contraction Ratio, A st /A d
Fig.33 Comparison of calculated operating pressure ratio with experimental results

0.006

0.00002 0.00006 0.00004 20 40 60 100 200 400 600 1000 Nozzle Plenum Total Pressure, pj, psi

0.00001
400 Nozzle Plenum Total Pressure, pj, psi (b) D d = 10.19 in 600 1000

(a) Dd = 6.09 in Fig.34

Variation of minimum cell pressure ratio (p c /p t ) with nozzle plenum total pressure level (p t )

155 0.04
1

I I A 0 /A* o c> < > 3.627


5.Ui-U

AJA 7.66 10.81 23.30 53.70

e,

1 deq

0.02

10.848 25.000
'

18 18 18 18

0.01
0.006 0.004
ro
CrC

C-

0.002
OJ

CJ

E
3

0.001 , 0.0006 0.0004 ^ ^,


"l

E
rr

0.1

0.2

0.4

0.6

10 xlO 6

Reynolds Number, Re

(O

(a)

D d = 6.09 in

0.01
0.006
^ ^ " " ^ a

0.004

^ , ^ ^ ^ -o^ *
VL

ar
B
dr-

K .,o?
3_

0.002

>

y ut
I

n(

,/A*

A , / , ,"
U

fi

- R '

rifln

""

3. 627
< - 5. 070 o 10 848 < * > 25. 300 .o in OV-,9 TOC
1U.

0.001
ew O

21.45 30.27 65.25 150.36


lO. JJ

18 18 18 18
1

TO

E
'ci

0.0006 0.0004

fi

in

E W \

0.0002

0.0001 0.1 0.2 0.4 0.6 1 2 4 Reynolds Number, Re ( - ^ - \

10 xlO 6

(b) Fig.35

Dd = 10.19 in

Comparison of empirical results with experimental data

156

-Ik
i r i 1iirr

t yp,
H

24.00

L_
4.026

iir

t
psia
O A

}'
in.

A /Anem

d/Ane

312 2.375 18.00 348 0.918 5.00

Pc'Pt

0.001 -rfr-A6 -B-

0.001

J. 6 Pex'Pt

0.01

(a) 7 = 1-40, A d /A* = 73.07

'LJ

fca

J-

K-

31.438
8.092

^rT 1ii 1iii

T
i i i 1 i i r

/ / /

Pex'Pc" 175 200

"

250

0.0001

Pt' psia

z. in.

ne / A *

VAne -

359 2.375 18.00 342 0.918 5.00


4
a _

16.40 59.04
_

0.001
Pex'Pt

0.01

(b) 7 = 1-40, A d /A* = 295.22 Fig.36 Ejector-diffuser performance for different average nozzle area ratios

t
i_

25.50

L_
4.680

2.375

-I
psia
A

0.01

1iir

rnim

Driving Fluid

d/A

Open Symbols, No Ejected Mass Flagged Symbols, with Ejected Mass (One Jet Pump) Closed Symbols, with Ejected Mass (Two Jet Pumps)

Pc'Pt 0.001

Steam (y = 1.30) 192 98.74 o Air (y - 1 . 4 0 ) 307 98.74 0 Open Symbols, No Ejected Mass Flagged Symbols, with Ejected Mass (One Jet Pump) Closed Symbols, with Ejected Mass (Two Jet Pumps)

/A*

ex't-c

ft

ne'ft

4.12 (Four Nozzles) Pc'Pt 0.001

c*-_.

A n e /A = 5.00 (Single Nozzle) Isentropic Curve (y-1.40)

40.0001 0.001
j . J

A n e /A

4I

18.00 (Single Nozzle)


I L-L -L_LJ_

12.01 (Two Nozzles)

0.0001 6
Pex'Pt

I \l

I I

0.01

10
A/A* Fig.38

100

Fig. 37

Ejector-diffuser performance for different driving fluids

Effect of parallel pumping (boundary removal) on diffuser performance

158

25.50 4.680

L_
3.561 [
0.01
8 6

T
i

"

Pt> psia

Ane'A* 3.147 3.914

VAne 11.15 8.97

l ir Test Configuration Scarf Symmetrical

I M

i l l "

o
A

1 Pc'Pl
4

305 309

Pex'Pc -

2 -

B 's
/ 7 7
/

-c

18

15

JL^ / 2 5 -

/
I I I I

0.001 0.001

'

I . . . 0.01 Pex'Pt

' '/ / -'C<.<mJ

(a) 7 = 1.40, A d /A* = 35.09


24.00

L
H=a
0.56 iii r
1 i I I I I

6.141

o
A

/w * A /Apsia "ne VAne 309 3.147 19.20 302 3.914 15.44

Configuration Scarf Symmetrical

Pc'Pl

0.001
6

0.001

(b) 7 = 1.40, A d /A* = 60.43 Fig.39 Comparison of ejector-diffuser performance for scarf and symmetrical nozzles

159 1100

Alr_Ni -

1000

^
/

Hydrogen

< w o

soo
7 0 0

1
l|
Helium (1 aim) 1 aim) 1
i

.600

1
i

.500

|
1 I * 1Lc=Argo Ultf n) 1f

400

300 Nitrogen (0.1 atmlv Nitrogen

II

^Air

V.
1 1 1

I
\

n
100

INv_Air
(0.1 atm)

\ \ : ^
"ydrogen*5> ss. Oatml

irgon 0.1 atm)

l^^J
|
1,7 I 8

1.3

1 i

1.5

1.6

Ratio ol Specific Heat 1

Fig.40

Variation of y with temperature and pressure

1.0
x 0.6 -V

0.4

v\>\ \ "*-^"^ V \s \ * * .

/ A d / A 20.40

~^\
^ ^ S - y 1.3 )
"-"--1.40

I e
a

0.2

Msent ropic Expan sion (or Hydrc gen at T , - 5 00R

**"s

0.10 A ne /A - 1 0 . 7 6 \ V 0.06 ^1.67

S O.Oi1
CO

0.02

0.01 6 10 Area Ratio, A/A* 20 40 60

100

Fig.41

Static-to-total temperature ratio variation with area ratio for isentropic expansion for different constant 7's and variable 7

s
0.010
Driving Gas 0.006
0.10

Exit Pressure Kir-txnaust RTF-Exhaust RTF-Exhaust Atmospheric Atmospheric RTF-Exhaust Atmospheric

0.004

o 0 A A o a C

Hir Argon Helium Helium Hydrogen Nitrogen Nitrogen

0.002

I 0.0010
(/
\y>
OJ s-

gj^

i
0.0006 0.0004
& -

a.
CZ

CO

1
1

/A* 18

0.02
0.0002

18 deg

i
1
I

0.0001
20 4) 60 Ratio of Duct-to-NoziMe Throat Area, A ri A

80

100

J O

60 1

100

200

W0

600

1000

Nozzle Plenum Total Pressure, p^ psia Fig.43 Variation of minimum cell pressure ratio, pc/Pt, with nozzle plenum total pressure, p t , for configuration 2

Fig.42

Diffuser-ejector average pressure ratio required for starting

161

0.020 Ad/A-

0.471 0.304

73.06 20.40

Gas Exit Pressure RTF-Exhaust RTF-Exhaust RTF-Exhaust Atmospheric Atmospheric RTF-Exhaust Atmospheric

u d
cr:

OJ XLi

0.0004 20 30 40 60 80 100 Duct-to-Nozzle Throat Area Ratio, A^/A

Fig.44

Diffuser cell-to-driving pressure ratio compared with one-dimensional isentropic pressure ratio

u
8.0
6.0 4.0

!
1

1 - j 1 - - - ' / /
mW
m - '

i/
1/ i /

/ / \ * ^ / - ^ A

i i 1
A

2.0

5/ *''

^A.

''

Ise ntropic Expansion


= Vm nui-l T. = *)f|OR
x
" -

i/
CCA^X
A ^ l

r =1.40 End of A /A* '18.00 Nozzlene


i

I /
*

/ /
r>^

V ^ E n d of A../A* 10.76 \ozzie


' 1

1.0 0.8 0.6 0.4

i.-g
loo

fl

' ^--Isentropic Expansiiin psia, T( t190R y Variedfrom 1.40 to 1.67 A 10.76 Nozzle

# /
/ /

' .

"

/
/ /

V *

0.2

/ /^Hydrogen Saturation Curve / for a Plane Surface (Ref. 28)

Nitrogen S atu ration C urve for a Plane Si rface (Ref. 28)

/ /

'

1 1

- A i r Saturation Curve (Ref. 301 I

0.1 0.08 1 0.06


i

l i
i i i
i

0.04

0.02
t

0.01

I
10 20 30 40 50 60 70 80 Temperature, T, R 90 100

I
110
1 2 0

IW

140

150

Fig.45

Constant and variable 7 isentropic expansion in relation to air, nitrogen and hydrogen saturation curve

163

100 125 Temperature, T, R

Fig.46

Constant and variable 7 isentropic expansion in relation to argon, hydrogen and helium saturation curve

164

165

ANALYSIS OF DUCTED MIXING AND BURNING OF COAXIAL STREAMS* by C.E.Peters

ARO Inc. Engine Test Facility Tullahoma, Tennessee, USA

The research reported in this paper was sponsored by the Arnold Engineering Development Center, Mr Force Systems Command, Arnold Air Force Station, Tennessee, under Contract F40600-69-C-0001 with ARO Inc. Further reproduction is authorized to meet the needs of the US Government.

166

167 CONTENTS

Page NOTATION 1. 2. INTRODUCTION DEVELOPMENT OF I D CORE THEORY 2.1 Principal Assumptions 2.2 Basic Integral Equations 2.3 Mixing Zone Profiles 2.4 Turbulent Eddy Viscosity 2.5 Duct Wall Equations 2.6 Transformation of Integral Equations 2.7 Correlation of 1-D Core Theory with Low Speed Mixing Experiments DEVELOPMENT OF 2-D CORE THEORY 3.1 Basic Integral Equations 3.2 Mixing Zone Profiles 3.3 Transformation of Integral Equations 3.4 Method of Characteristics Solution 3.5 Correlation of 2-D Core Theory with j\ir-Air Ejector Experiments 3.6 Correlation of 2-D Core Theory with Rocket-j\ir Experiments 4. CONCLUDING REMARKS 168 171 172 172 172 173 174 174 174 175 175 175 176 176 176 176 177 177 178 179-187

3.

REFERENCES FIGURES

168

NOTATION

b |B| Cfc C C, , C 2 , . . . . D |D| F , , F 2 , ....

mixing zone width determinant in equation for mass fraction of element k mass fraction of elements from central stream constants in duct wall equation diameter coefficient determinant for system of equations coefficients in system of differential equations db/dx

Fj Fn G,, G2 ,....
H0 H , , H 2 ,.... k k0 L M O/F p pj, p0 |P| Q r rn |R| u umax umin v V wa

mixing duct thrust vacuum thrust of primary nozzle coefficients in system of differential equations
total, or stagnation, enthalpy, including chemical heats of formation coefficients in system of differential equations constant in eddy viscosity equation incompressible eddy viscosity constant length of mixing duct Mach number rocket oxidizer-fuel mass ratio static pressure back pressure total, or stagnation, pressure determinant in equation for dp w /dx

species conservation parameter radial coordinate radius of primary nozzle exit determinant in equation for dr-/dx axial velocity component maximum velocity in mixing layer minimum velocity in mixing layer transverse or radial velocity component magnitude of total velocity vector initial mass flow of secondary stream

169
Wj
X

initial mass flow of primary stream axial coordinate length parameter in duct wall equation integration step size turbulent eddy viscosity flow angle density turbulent shear stress

*e Ax e

e
p
T

Subscripts mixing duct inlet inviscid secondary flow centerline inner mixing zone boundary inviscid primary flow half radius control surface in mixing zone primary nozzle exit duct wall

170

171

ANALYSIS OF DUCTED MIXING AND BURNING OF COAXIAL STREAMS C.E.Peters 1. INTRODUCTION

Ducted turbulent mixing of coaxial streams occurs in many devices of practical interest. Typical examples are the jet pump (or air-air ejector) and composite propulsion systems such as the air augmented rocket 1 ' 2 . For propulsive applications one must consider the possibility that exothermic chemical reactions will occur in the mixing layer. An extensive theoretical and experimental investigation of ducted mixing has been accomplished at the AEDC 3,4,S . The basic objective of this research has been to develop an adequate engineering theory to describe the ducted mixing process, including chemical reactions. Emphasis has been placed on relatively long mixing systems in which the mixing layer may extend over most or all of the duct cross section at the exit plane. The duct pressure distribution will be strongly influenced by the thick mixing layers, and will be very different from the inviscid pressure distribution. In other words, this may be considered a "strong viscous interaction" problem. Consider the mixing system shown schematically in Figure 1, in which the primary and secondary fluids are specified. The specific objective here is to predict the secondary mass flow, wa , and the duct wall pressure distribution if the following parameters are described: (i) Geometry. (ii) Primary (central) stream initial conditions. (iii) Secondary stream stagnation pressure pga and stagnation temperature, T u a (iv) Back pressure p-, . Three distinct flow regimes are shown in the mixing flow field of Figure 1. In the first regime, turbulent mixing occurs between the secondary flow and the core of inviscid primary flow. In the second regime, the inviscid core has been dissipated, but a region of inviscid secondary flow exists near the duct wall. The third regime occurs after the mixing layer has spread to the wall, and the flow is entirely turbulent. Several modes of operation are possible for a ducted mixing system when the primary stream is initially supersonic and the secondary stream is initially subsonic. These modes are distinguished by the factor which limits the secondary mass flow rate, wa . The "upstream choking" mode (Fig.2(a)) occurs when wa is limited by choking of the secondary flow near the entrance of the duct. The primary stream has expanded, causing the secondary flow area at the choke point to be less than the initial flow area. Fabri et al. 6 ' 7 called this mode the "supersonic regime". The "downstream choking" mode (Fig.2(b)) occurs when wa is limited by choking of the flow at the duct exit. This mode will probably not occur in a cylindrical or divergent mixing duct unless the primary and secondary fluids have greatly different densities, or unless chemical reactions occur in the mixing process. The downstream choking mode is important for certain propulsive applications. The "back pressure dependent" mode of operation (Fig.2(c)) occurs when the back pressure pj, is sufficiently high to unchoke the duct flow. The secondary flow is subsonic throughout the duct, and the duct exit pressure exactly matches the back pressure. Fabri called this mode the "mixed regime", and it is commonly encountered in jet pump applications. If the mixing duct is cylindrical, sufficiently long and operating in the back pressure dependent mode, the flow may be analyzed by application of one-dimensional theoretical concepts. The flow is assumed to be completely mixed at the exit plane, and overall conservation equations for momentum, mass flow and energy are solved for the flow between the entrance and exit sections. The one-dimensional theory gives no information about the duct length required to achieve complete mixing, and is not applicable to ducts of variable cross section because the unknown axial wall pressure forces cannot be predicted. The detailed mixing duct flow must be analyzed in order to compute the unknown wall pressure integral in variable area configurations.

172 For cylindrical mixing ducts operated in the upstream choking mode, Fabri et al. derived a "quasi-one-dimensional" theoretical model 6 ' 7 . In this theory, it is assumed that the primary upstream undergoes an isentropic expansion and the subsonic secondary stream undergoes an isentropic contraction. The combined momentum equation is written between the entrance and the secondary stream choking section, thus eliminating the need to assume that the average primary and secondary pressures are equal at the choking section. The specification of Ma = 1 at the choking section then allows the system of equations to be solved for the mass flow ratio, wa/w: . The quasi-one-dimensional theory gives good results for many practical applications, but cannot be used for variable area mixing ducts. The detailed flow processes must be analyzed in order to predict the axial pressure force on the duct wall. The quasi-one-dimensional theory can be criticized because the viscous effects are neglected, although Chow and Addy8 have shown that viscous effects are small for a ducted mixing system operating in the upstream choking mode and having a relatively large secondary flow. Viscous effects, however, are quite important at zero and low secondary flow rates. Chow and Addy8 made a significant improvement on the quasi-one-dimensional analysis for a constant area mixing system operating in the upstream choking mode. The primary stream is computed with the method of characteristics and the inviscid secondary flow is computed simultaneously by use of the one-dimensional assumption. The inviscid solution is then corrected for viscous effects along the jet boundary separating the primary and secondary fluids. The Chow and Addy theory is applicable to variable area mixing ducts, and is useful for thrust augmentation devices. For zero and low secondary flows. Chow and Addy used a Korst-type base pressure analysis, including effects of base bleed. Chow and Addy show excellent correlation of their theory with experimental results for a constant area supersonic air-air ejector. Chow and Yeh9 show similar excellent agreement between the theory and experiments for variable area air ejector configurations. The Chow and Addy theory is valid for flows with relatively weak viscous interaction. Their technique of superimposing constant pressure two-dimensional mixing profiles on the jet boundary limits the technique to flows with thin mixing layers. Another limitation of the superposition technique in general is that the mixing profiles are computed with the assumption of negligible transverse pressure gradients, but then are superimposed onto a flow which has strong radial pressure gradients. As discussed earlier, the main interest in the AEDC investigation of ducted mixing has been on mixing systems with thick mixing layers. For this reason, it was necessary to develop an analytical model in which the mixing layer is computed simultaneously with the inviscid portions of the primary and secondary flows. Two analytical models will be described. In the first, the inviscid portions of the primary and secondary flows are assumed to be one-dimensional (1-D Core Theory). In the second model, the inviscid portion of the primary flow is computed with the method of characteristics (2-D Core Theory). In both models, the viscous layer is treated by use of the well-known von Karman integral method. 2. DEVELOPMENT OF 1-D CORE THEORY

2.1 Principal Assumptions 1. The flow is axisymmetric. 2. All gases obey the perfect gas law. 3. The usual boundary layer assumptions are applicable to the mixing layer. 4. Inviscid portions of primary and secondary flow are one-dimensional and isentropic. 5. The mixing layer is fully turbulent and boundary layers at initiation of mixing are negligible. 6. Viscous effects at duct wall are negligible. 7. The mixing zone velocity profiles are assumed to have similar shapes (cosine function). 8. The turbulent Prandtl and Lewis numbers are unity. 9. For mixing with simultaneous chemical reactions, the reactions are assumed to be in equilibrium. 10. In the third regime (Fig. 1) the free mixing concepts of shear and profile shape similarity are assumed to be applicable. 2.2 Basic Integral Equations The flow model is shown schematically in Figure 3. Three basic integral equations are used in the analysis; (i) a continuity equation for the entire flow, (ii) a momentum equation for the entire flow, and (iii) a momentum equation for the portion of the flow between the duct centerline and a control surface located half-way across the mixing layer.

173 Continuity Equation 1 dx Jo Overall Momentum Equation d ,r w dp w TL f pu 2 r dr = - - ^ . dx J 0 dx 2 Half-Radius Momentum Equation


J prm A ftm do

purdr = 0 .

(1)

(2) K '

5J. **-'- U. <-*---'--| ?Equation (3) contains a term, r m , representing the turbulent shear stress at the half-radius control surface. 2.3 Mixing Zone Profiles Relation between velocity, composition and enthalpy The axisymmetric boundary layer equations for momentum, energy and conservation of elemental species, with unity turbulent Prandtl and Lewis numbers, are 1 0 : Momentum Equation 3u 3u 3 / 3u\ 3p pur + pvr = per 1 - r . dx 3r 3r \ 3r/ dx Energy Equation 3H 0 3H n 3 / 3H0\ (4)

r2

Conservation of Elemental Species

" ?
~ua Uj - u a
u

ur

3C k

+P V T

3C k

3 /

5T

37r 3rj"
c =

3Ck\

(6)

For isobaric flow, the dp/dx term in Equation (4) is zero, and the following linear relation is obtained between u , H 0 and C k :
H =

0 ~ H 0a H0j - H 0 a

k ~ Cka Cjcj - C k a

(?)

Because the transport coefficients for all species are the same, the composition at any point in the mixing zone may be characterized as a simple two-component mixture. Thus 0 ~ H 0j H H H

"j

0a 0a

C ,

(7a)

where C is the mass fraction of elements from the central stream. Although derived for isobaric flow, Equation (7a) is also assumed to be valid for flows with axial pressure gradients. Note, however, that the inviscid reference velocities, u a and ui , are pressure dependent. The assumption that Equation (7a) is valid for flows with axial pressure gradients is made conditionally. If Equation (7a) is valid, then the computed total flux of elements from the central stream should be the same at all axial stations. In other words, the solution of the integrated momentum equation is also a satisfactory solution for the integrated species and energy equations. A species conservation parameter, Q , is defined:

Q = ^frwpUCrdr.
Wj -D

(8)

If Q remains unity, the use of Equation (7a) is justified for flows with axial pressure gradients. It has been found that Q does remain approximately unity for typical air augmented rocket systems 5 .

174

Mixing zone velocity profiles The mixing zone velocity profiles are represented by a cosine function: " ~ "min _ max umin *
1 + cos it

P?)

(9)

In the first and second regimes (Fig.3), u m i n = ua ; in the third regime u m i n = u w . In the first regime, u m a x = Uj ; in the second and third regimes, u m a x = uc . At the half-radius control surface
u

m " .("max

+ u

min)

(10)

Equilibrium mixing zone chemistry For rocket-air mixing, the mixing zone compositions and temperatures are determined by an approximate equilibrium chemistry computation. It is assumed that the static temperature distribution and the composition distribution in the mixing zone can be computed from the concepts of equilibrium chemistry with representative values of p w , u; and u a . The corresponding total temperature distribution is then assumed to be invariant throughout the flow field. This treatment of the mixing zone chemistry is similar to the "flame sheet" chemical model used by Libby, except that the mixing zone temperature and composition profiles are corrected to the equilibrium profiles at the reference condition. 2.4 Turbulent Eddy Viscosity The term r m in Equation (3) is determined by
7

3u m ' ' Pm em T~

(11)

where e m is the empirical eddy viscosity. The semi-empirical model for the eddy viscosity which is used in this work is the incompressible Prandtl model e = kb(u m a x - u m i n ) , with the empirical constant k of the correction suggested by density could be generalized if control surface. The constant (12)

corrected for the effect of variable density. The correction used is a modification Donaldson and Gray11. Donaldson and Gray found that the influence of variable k is taken to be a function of the local Mach number, Mm , at the half-radius k is related to the incompressible constant, k 0 , by the empirical equation

k = 0.66 + 0.34 exp (-3.42 M2,) . (13) ko The present theory has been correlated with low speed air-air mixing experiments, and it was found that k0 = 0.007 in the first regime, and k0 = 0.011 in the second and third regimes. These values have been used for all of the computations presented in Reference 5. It should be noted that the Prandtl eddy viscosity model (Equation (12)) is known to be deficient when u m - n /u m a x exceeds about 0.3. Also, Equation (13) has not been checked against a sufficient number of experiments to be considered completely reliable. Nevertheless, the present eddy viscosity model gives good results for the experimental rocket-air mixing configurations of References 3 and 4. In those experiments, u m j n /u m a x did not exceed 0.1. 2.5 Duct Wall Equations The duct wall is assumed to be represented by the following pair of equations, where the constants C,-C 7 and xe are specified: rw = C, + C2x + C3x2 + C4x3 + C4x4 , x < xe ,

2.6 Transformation of Integral Equations Sufficient information has been developed in the preceding sections so that the terms in Equations (1) to (3) can be related to three flow field variables. In the first regime (Fig.3), these variables are selected to be p w , rand b . In the second rejdme, the selected variables are p w , uc and b . In the third reigime the selected variables are p w , u c and u w .

175 First regime transformation - With t h e t e c h n i q u e described in Reference 5, E q u a t i o n s ( 1 ) t o ( 3 ) can be transformed into the following system of equations: Continuity Equation dPw ^ dn _ db - i + F, dx * + F , dx dx (14)

F,

Total Momentum Equation ^ dPw . A-. drdx Half-Radius M o m e n t u m E q u a t i o n dp, dr; db H, - P + H2 J - i + H3 J = 1 dx dx dx H, . (16)
2

.-, db
3

G, -pL + Gj - i + G3
dx dx

G4
'

(15)

Equations ( 1 4 ) - ( 16) can be solved for the derivatives d p w / d x , dr-/dx as long as the d e t e r m i n a n t of t h e coefficients is n o t zero: dpw dx where the d e t e r m i n a n t s are F, |D| =
G

and d b / d x

by use of C r a m e r ' s rule,

IPI D| '

drj dx

|R| |D| '

db dx

|B| |D|"

(17)

F2
G

F3
G

F4 , IPI = G4 H4

Fj Gj Hj

F3 G3 H3 etc.

H,

Hj

HJ

Equations (17) are numerically integrated by use of the well-known Runge-Kutta method. Second and third regime transformations - The transformation in the second and third regimes is similar to the first regime except that the derivatives in question are ( d p w / d x , d u c / d x , db/dx) and (dp w /dx, d u c / d x , d u w / d x ) , respectively.

2.7

Correlation of 1-D C o r e T h e o r y with L o w Speed Mixing E x p e r i m e n t s

Most of the correlations between theory and e x p e r i m e n t have been m a d e with t h e 2-D Core T h e o r y (Section 3 ) . T h e 1-D Core T h e o r y has been correlated with a variety of low speed air-air mixing e x p e r i m e n t s in o r d e r t o establish the values for the incompressible eddy viscosity c o n s t a n t s , and t o establish t h e validity of t h e t h e o r y for c o n s t a n t density flows. T h e agreement between t h e o r y a n d e x p e r i m e n t is generally satisfactory, a n d t h e correlations will n o t be shown here.

3.

D E V E L O P M E N T O F 2-D C O R E T H E O R Y

The ducted mixing theory for t h e first regime has been e x t e n d e d t o include c o m p u t a t i o n of t h e inviscid p o r t i o n of t h e primary flow w i t h t h e irrotational m e t h o d of characteristics. T h e mixing z o n e and secondary flow are t r e a t e d in essentially the same way as in Section 2, t h e major difference being in t h e m e t h o d of c o m p u t i n g the inviscid core flow. T h e basic integral e q u a t i o n s are derived w i t h the inner mixing zone radius, r- , as t h e lower limit (Fig.4). (It will be recalled t h a t t h e corresponding lower limit is zero for t h e 1-D Core T h e o r y integral equations.) A m e t h o d of characteristics solution of t h e inviscid core flow t h e n provides t h e b o u n d a r y c o n d i t i o n s ( a t r-) for t h e solution of the integral equations. It should be n o t e d that t h e 2-D Core T h e o r y is identical t o the 1-D Core T h e o r y in t h e second and third regimes.

3.1

Basic Integral E q u a t i o n s

By integrating the continuity equation and the boundary layer momentum equation, one obtains the following integral equations. Continuity Equation d .r w dr J pur dr = p-v-r- - p^-r- . (18)

176

Total Momentum Equation

=c** --'* fi-^MHalf-Radius Momentum Equation


r d_ P rr m 5 d frm / dr\ dp w / r m r.2\ ,. ; 2 dx J^ pu r dr - u m J pur dr = (u m - u-) I^Utf - ftT,ql + r m r m - ^ - J . (20)

3.2 Mixing Zone Profiles The velocity and density profiles in the mixing zone are computed with the same equations as were used in Section 2. In the velocity profile equation (Equation (9)), however, "max = "i =
V

i cos e i .

where Vs is the total velocity vector at r; and 0- is the flow angle at r- (V- is an isentropic function of p w ). The velocity at the control surface, u m , is
u

m = i("i + ua>

3.3 Transformation of Integral Equations The method of transforming the integral equations is the same as in Section 2. One obtains Equations (14)-(16), but the coefficients F . G , and H are of course different from the 1-D Core Theory coefficients. 3.4 Method of Characteristics Solution Many terms appearing in the coefficients F, G and H depend on the flow conditions at the inner mixing zone boundary. In order to solve the system of equations, one must be able to evaluate the following parameters at r-: 0-, 3u/3x, 3u/3r, 3p/3x and 3p/3r. To provide these parameters, the inviscid core flow is developed (with the irrotational method of characteristics) simultaneously with the numerical solution of the equations for the mixing layer and inviscid secondary flow. The flow near the inner mixing zone boundary at some downstream location, x , is shown in Figure 5. At x , the upstream solution has defined the values of p w and r; . The derivatives dp w /dx and dr-/dx at x are also specified. Since the flow is completely defined at x , a new Family I characteristic can be generated from the boundary point D. (The "interior" solution is thus extended outward by one Family I characteristic at each computation interval, Ax). With the conditions at x completely known, the Runge-Kutta method prescribes a series of intermediate points at which dp w /dx, drj/dx and db/dx are to be computed. These specified points (x,p w ,rj,b) do not necessarily correspond to points on the actual solution, but the solution depends on a weighted combination of the intermediate results. Such a Runge-Kutta point is shown as point E in Figure 5. With the position of E specified, along with the pressure, the only unknown at E is the angle, 0, . The angle, 0- , which is required for the succeeding RungeKutta points, is determined by the following procedure. A Family II characteristic EF is constructed which intersects the last Family I characteristic, along which the flow is completely defined from the interior solution. By an iterative solution of the compatibility equations for a Family II characteristic, the location of F and the value of 0j at E are determined. After the system of equations is solved for p w and r- at x + Ax (point G, Figure 5), the above procedure is repeated to define the physically correct value of 6i at x + Ax . The above procedure is the key technique by which the solutions for the core flow and mixing layer are coupled. 3.5 Correlation of 2-D Core Theory with Air-Air Ejector Experiments The 2-D Core Theory has been correlated with the air-air ejector experiments of Chow and Addy8 and Chow and Yeh9. All of the experimental cases considered here were operated in the upstream choking mode, i.e., with relatively low back pressure. The experimental configurations are shown in Figure 6. Constant area air-air ejectors Chow and Addy presented a series of experiments for the configuration of Figure 6(a). The experimental mass flow ratios, along with the results of the 2-D Core Theory, are shown in Figure 7. The agreement between theory and experiment is very satisfactory for the entire range of p ua /P0j over which the theory is applicable. The theoretical results of Chow and Addy are not shown in Figure 6; they are nearly identical to the 2-D Core Theory results.

177

The experimental and theoretical duct pressure distributions for the constant area air-air ejector are shown in Figure 8. Variable area air-air ejectors Chow and Yeh9 presented a series of experiments for an air-air ejector in which the wall shape was parabolic (Figures 6(b) and 6(c)). The mass flow ratio for the configuration of Figure 6(b) is shown in Figure 9. As was the case for the cylindrical duct, the 2-D Core Theory accurately predicts the experimental results over the entire range of applicability. The initial secondary Mach number, M a | , is plotted against the pressure ratio Pwi/poa in Figure 10 for the configurations of Figures 6(b) and 6(c). Because of the downstream displacement of the minimum duct area, Fabri's saturated supersonic regime is not encountered with the displaced parabolic duct. 3.6 Correlation of 2-D Core Theory with Rocket-Air Experiments The main objective of this work has been to predict the performance of ducted mixing systems in which chemical reactions occur. Peters et al.3 and Cunningham and Peters4 have presented experimental results for the mixing system shown in Figure 1.1. The hydrogen-oxygen rocket was operated fuel-rich (O/F = 3.2) at a combustion chamber pressure of approximately 20 atm. The rocket nozzle area ratio was approximately 5. The secondary flow consisted of room temperature air. In any propulsion system application, an important parameter is the mixing duct thrust, F,j , defined by
F

d =

2w

w r wf ^ j dx ,

where L is the length of the mixing duct. Mixing system operated in upstream choking mode The mixing configuration of Figure 11(a) was operated in the upstream choking mode3. A typical experimental wall pressure distribution is shown in Figure 12, along with the theoretical results for no mixing and for mixing with equilibrium chemistry. The theory shows that the inviscid pressure distribution is significantly altered by mixing and combustion, especially in the downstream portion of the duct. The theoretical pressure distribution (with equilibrium chemistry) predicts the experiment pressure distribution reasonably well. The differences between theory and experiment are attributed to shock waves in the supersonic portion of the flow; these shock waves are neglected in the theory. The mass flow ratio, wa/w: , is plotted against Pna/P0j -n Figure 13. The theory predicts a mass flow ratio which is 4-5% higher than the experimental ratio. The duct thrust ratio, F,j/Fn , is shown in Figure 14 (F n is the vacuum thrust of the rocket). The theory predicts a thrust ratio which agrees with experiment to within 10% over the range of Poa/POj considered. Mixing system operated in the downstream choking mode The mixing system of Figure 11(b) was operated in the downstream choking mode4. A typical experimental wall pressure distribution is shown in Figure 15, along with the theoretical result for equilibrium mixing zone chemistry. The theoretical wall pressures are higher than the experimental pressures, but the shape of the experimental pressure distribution is correctly predicted. As was the case with the upstream choking mode, the differences between theory and experiment are attributed to shock waves in the flow. The mass flow ratio, wa/wj , is shown in Figure 16. The theoretical mass flow ratios are 10 to 15% higher than the experimental values. The duct thrust ratio, F d /F n , is shown in Figure 17. The theoretical thrust ratio is approximately 6% larger than experiment at the lower values of Poa/POj a n d approximately 16% too large at the higher values of p ua /P0j 4. CONCLUDING REMARKS

The analytical model which has been presented for ducted mixing of coaxial streams includes the effects of (i) a non-uniform supersonic core flow and (ii) equilibrium chemical reactions in the mixing layer. Even though the current detailed knowledge about turbulent flows with chemical reactions is meager, the use of integral methods permits reasonably accurate computations of the flow in complex mixing systems. The main emphasis of this work has been placed on mixing systems which are strongly influenced by the mixing process. Of course, the theoretical model is also applicable to flows which are weakly influenced by mixing, such as supersonic air-air ejectors operating in the upstream choking mode. The present theory accurately predicts the experimental performance of such air-air ejectors, but the theory offers no quantitative improvement over the superposition technique of Chow and Addy.

178 The weakest aspects of the present theory are (i) the model for the turbulent eddy viscosity, and (ii) the assumption that the inviscid core flow is irrotational (without shock waves). Although the theoretical model has several deficiencies, the results are qualitatively correct. In addition, the theoretical results are sufficiently accurate so that the theory may be considered useful for engineering analysis of ducted mixing systems.

REFERENCES

1. Perini, L., et al. 2. Staff Report 3. Peters, C.E. et al. 4. Cunningham, T.H.M. Peters, C.E. 5. Peters, C.E.

Preliminary Study of Air Augmentation of Rocket Thrust. Journal of Spacecraft and Rockets, Vol.1, No.6, November-December 1964, pp.626-634. Composite Engines. Space/Aeronautics, Vol.48, No.3, August 1967, pp.83-90. Mixing and Burning of Bounded Coaxial Streams. AEDC-TR-65-4, March 1965, Arnold Engineering Development Center. Further Experiments on Mixing and Burning of Bounded Coaxial Streams. 68-136, October 1968, Arnold Engineering Development Center. AEDC-TR-

Turbulent Mixing and Burning of Coaxial Streams Inside a Duct of Arbitrary Shape. AEDC-TR-68-270, January 1969, Arnold Engineering Development Center. (see also)

Peters, C.E., et al. 6. Fabri, J., Paulon, J. 7. Fabri, J., Siestrunck, R. 8. Chow, W.L., Addy, A.L. 9. Chow, W.L., Yeh, P.S. 10. Libby, P.A.

Theoretical and Experimental Studies of Ducted Mixing and Burning of Coaxial Streams. Journal of Spacecraft and Rockets, Vol.6, No.12, December 1969, pp.1435-1441. Theorie et Experimentation des Ejecteurs Supersoniques Air-Air. ONERA Note Technique No.36, 1956. (English translation - TM1410, September 1958, NACA.) Supersonic Air Ejectors. In "Advances in Applied Mechanics", edited by H.L.Dryden and T. von Karman, Academic Press, New York, 1958, pp. 1-34. Interaction Between Primary and Secondary Streams of Supersonic Ejector Systems and Their Performance Characteristics. AIAA Journal, Vol.2, No.4, April 1964, pp.686-695. Characteristics of Supersonic Ejector Systems with Non-Constant Area Shroud. Journal, Vol.3, No.3, March 1965, pp.525-527. Theoretical Analysis of Turbulent Mixing of Reactive Gases with Application to Supersonic Combustion of Hydrogen. ARS Journal, Vol.32, No.3, March 1962, pp.388-396. Theoretical and Experimental Investigation of the Compressible Free Mixing of Two Dissimilar Gases. AIAA Journal, Vol.4, No.l I, November 1966, pp.2017-2025. AIAA

11.

Donaldson, C.duP., Gray, K.E.

179

I Secondary Stream Stagnation Conditions


_
IT.

Inviscid Secondary Flow

oa'

oa

Fig. 1 Schematic of ducted mixing system

Section Where Secondary Stream Chokes

1.0 r-

Duct Exit
0.5 oa

Wall Pressure Distribution

(a) Upstream choking mode Fig.2 Operational modes of a ducted mixing system

180

Flow Chokes at or Near the Duct Exit

l.Oi-

0.5 oa

Wall Pressure Distribution

Duct Exit

(b) Downstream choking mode

Duct Exit Pressure Matches Back Pressure

i.Ot-

Pw oa

0.5

Wall Pressure Distribution

Duct Exit

(c)

Back pressure dependent mode Fig. 2 Concluded

181

Fig.3

Nomenclature for integral analysis (1-D Core Theory)

I n v i s c i d Secondary Flow Pa-Ua f Mixing Zone

Fig.4

Nomenclature for 2-D Core Theory

* Fig.5

(x + Ax) Solution at a general boundary point

182

0
\f/ff^^
oj

/ Cylindrical Duct 75 * V r n - 1.7:

7_E
L
(a) Cylindrical duct of Chow and Addy 8

P a r a b o l i c Duct r w / r n - 1.75 + 0 . 0 2 ( x / r n ) '

(b)

Parabolic duct of Chow and Yeh 9

xxzzz* M,
(c)

r
- 2.0

D i s p l a c e d P a r a b o l i c Duct

r w / r n - 1.75 + 0 . 0 2 ( x / r n - 2 ) '

Displaced parabolic duct of Chow and Yeh 9 Fig.6 Air-air ejector configurations

0.6

1
O

1
8)

1 /
/o /o

0.5

Experiment (Ref. 2-D Core Theory

-*

0.4

/o
o
H

0.3

b.

in i
AX

0.2

0.1

1
0.02

o ?
0.04

1
0.14

1...
0.16 0.18

0.06 0.08 0.10 0.12 Total Pressure Ratio, P oa /P 0 ^

Fig.7

Mass flow ratio for a cylindrical air-air ejector

183

2-D Core Theory Experiment j Q ^ ) A 0.092 1.0


P w l / P r<M

{ a o.m

0.4

X
0 1 2 3 4 A x i a l D i s t a n c e ,' x / r _ n 5 Fig.8 Wall pressure distributions in a cylindrical air-air ejector

0.6

0.5 O
\
H

0.4

Experiment (Ref. 9) Parabolic Mixing Duct 2-D Core Theory

* a 0.3

8
0.2

Ha

0.1

0.02

0.04

0.06

0.08

0.10
Oo

0.12
OJ

0.14

0.16

0.18

Total Pressure Ratio, p_/p4 Fig. 9 Mass flow ratio for a variable area air-air ejector

184

Experiment (Ref. 9)

O Parabolic Duct (Fig. 6b) A Displaced Parabolic ( . Duct (Fig. 6c) 2-D Core Theory

1.0

tj

0.8 -

0.6 -

I
x>

c 0 u
CJ

0.4

n n 0.2 -

0.02

0.04

0.06 0.08 0.10 Pressure Ratio, p , / p . ' wl *oj

0.12

0.14

0.16

Fig. 10

Initial secondary Mach number in a variable area air-air ejector

r n /r wl 0.38, ri - 3 in. (a 7.5 cm.) Conical Rocket Nozzle (15 Half Angle)

H2-O2 Rocket -7

^n-. ?
(a) Conical mixing duct 3 5.17 r w l

-2r wl-

rw/rwl

= 1 + 0.052 x / r wl

rw/rwl -

1.269

(b)

Conical mixing duct with cylindrical extension 4 Fig. 11 Rocket-air mixing configurations

185

1.0

1 O

Experiment (Ref. 3) i 2-D Core Theory, Eq. Chemistry 2-D Core Theory, No Mixing

0.8

5 a
a*
*A

rt

0.6

Back Pressure O

m
in

s
Or

HI

0.4

jixit Plane

~k
1 I 6.0

0.2
p /p . - 0.033 *oa *oj

I
2.0

I 4.0

Axial Distance, x/r -

Fig. 12 Wall pressure distribution for ducted rocket-air mixing (upstream choking mode)

6.0

T"
X oo

5.5

5.0

AS

ra
X

%
-A

4.5

o
r
OD

4.0

O Experiment (Ref. 3) 2-D Core Theory Eq. Chemistry

3.5

1A

1
0.025

JL
0.030 Total Pressure Ratio, P o a / p o i 0.035

Fig. 13 Mass flow ratio for ducted rocket-air mixing (upstream choking mode)

186

0.24

0.22

Experiment (Ref. 3) 2-D Core Theory Eq. Chemistry


CD

o - 0.20
rt

I *18
0.16

(A

0.025

0.030 Total Pressure Ratio, p /p . oa o j

0.035

Fig. 14

Mixing duct thrust for ducted rocket-air mixing (upstream choking mode)

1.0

0.8

m f
Jj * 6
rt M

O
0.4

Experiment (Ref. 4) 2-D Core Theory Eq. Chemistry

First Second Regime Regime Back Pressure

rt

0.2

Duct Exitp /p , - 0.030 F oa r oj

Axlal Distance, x/r Fig. 15

wl

Wall pressure distribution for ducted rocket-air mixing (downstream choking mode)

187

0.024

0.032 0.028 Total Pressure Ratio, P o a / p o i

0.036

Fig. 16

Mass flow ratio for ducted rocket-air mixing (downstream choking mode)

0.40

a
IN

0.30

i
0.20 0 -^ 0 0.024

O Experiment (Ref. 2 - D Core Theory Eq. Chemistry

4)

0.028
'

0.032

0.036
oj

Total Pressure Ratio, p


oa

/p

Fig. 17

Mixing duct thrust for ducted rocket-air mixing (downstream choking mode)

AGARDograph No. 163 North Atlantic Treaty Organization, Advisory Group for Aerospace Research and Development SUPERSONIC EJECTORS Edited by J.J.Ginoux Published November 1972 194 pages

AGARD-AG-163 629.7.047.2:533.6.011.5

AGARDograph No. 163 North Atlantic Treaty Organization, Advisory Group for Aerospace Research and Development SUPERSONIC EJECTORS Edited by J.J.Ginoux Published November 1972 194 pages

AGARD-AG-163 629.7.047.2:533.6.011.5

Two Short Courses were organized on "Ejectors" at the von Karman Institute for Fluid Dynamics in April 1968 and March 1969, respectively, in which 72 scientists and engineers from seven NATO nations participated. P.T.O. AGARDograph No. 163 North Atlantic Treaty Organization, Advisory Group for Aerospace Research and Development SUPERSONIC EJECTORS Edited by J.J.Ginoux Published November 1972 194 pages AGARD-AG-163 629.7.047.2:533.6.011.5 -

Two Short Courses were organized on "Ejectors" at the von Karman Institute for Fluid Dynamics in April 1968 and March 1969, respectively, in which 72 scientists and engineers from seven NATO nations participated. P.T.O. AGARDograph No. 163 North Atlantic Treaty Organization, Advisory Group for Aerospace Research and Development SUPERSONIC EJECTORS Edited by J.J.Ginoux Published November 1972 194 pages AGARD-AG-163 629.7.047.2:533.6.011.5

Two Short Courses were organized on "Ejectors" at the von Karman Institute for Fluid Dynamics in April 1968 and March 1969, respectively, in which 72 scientists and engineers from seven NATO nations participated. P.T.O.

Two Short Courses were organized on "Ejectors" at the von Karman Institute for Fluid Dynamics in April 1968 and March 1969, respectively, in which 72 scientists and engineers from seven NATO nations participated. P.T.O.

They were partly supported by AGARD and directed by Dr H.Uebelhack, Assistant Professor at VKI, now with Dornier System in Friedrichshafen, Germany. The objective of these Short Courses was to present a state-of-the-art review of the significant progress which has been made in the past few years in the design of high performance ejectors. AGARD has felt it appropriate to publish the Short Courses as an AGARDograph in an updated version, with emphasis on supersonic ejectors. A small number of copies were originally printed at the time of these courses, but numerous subsequent requests for these notes were received and the demand could not be satisfied.

They were partly supported by AGARD and directed by Dr H.Uebelhack, Assistant Professor at VKI, now with Domier System in Friedrichshafen, Germany. The objective of these Short Courses was to present a state-of-the-art review of the significant progress which has been made in the past few years in the design of high performance ejectors. AGARD has felt it appropriate to publish the Short Courses as an AGARDograph in an updated version, with emphasis on supersonic ejectors. A small number of copies were originally printed at the time of these courses, but numerous subsequent requests for these notes were received and the demand could not be satisfied.

They were partly supported by AGARD and directed by Dr H.Uebelhack, Assistant Professor at VKI, now with Dornier System in Friedrichshafen, Germany. The objective of these Short Courses was to present a state-of-the-art review of the significant progress which has been made in the past few years in the design of high performance ejectors. AGARD has felt it appropriate to publish the Short Courses as an AGARDograph in an updated version, with emphasis on supersonic ejectors. A small number of copies were originally printed at the time of these courses, but numerous subsequent requests for these notes were received and the demand could not be satisfied.

They were partly supported by AGARD and directed by Dr H.Uebelhack, Assistant Professor at VKI, now with Domier System in Friedrichshafen, Germany. The objective of these Short Courses was to present a state-of-the-art review of the significant progress which has been made in the past few years in the design of high performance ejectors. AGARD has felt it appropriate to publish the Short Courses as an AGARDograph in an updated version, with emphasis on supersonic ejectors. A small number of copies were originally printed at the time of these courses, but numerous subsequent requests for these notes were received and the demand could not be satisfied.

AGARDograph No. 163 North Atlantic Treaty Organization, Advisory Group for Aerospace Research and Development SUPERSONIC EJECTORS Edited by J.J.Ginoux Published November 1972 194 pages

AGARD-AG-163 629.7.047.2:533.6.011.5

AGARDograph No. 163 North Atlantic Treaty Organization, Advisory Group for Aerospace Research and Development SUPERSONIC EJECTORS Edited by J.J.Ginoux Published November 1972 194 pages

AGARD-AG-163 629.7.047.2:533.6.011.5

Two Short Courses were organized on "Ejectors" at the von Karman Institute for Fluid Dynamics in April 1968 and March 1969, respectively, in which 72 scientists and engineers from seven NATO nations participated. P.T.O. AGARDograph No. 163 North Atlantic Treaty Organization, Advisory Group for Aerospace Research and Development SUPERSONIC EJECTORS Edited by J.J.Ginoux Published November 1972 194 pages AGARD-AG-163 629.7.047.2:533.6.011.5

Two Short Courses were organized on "Ejectors" at the von Karman Institute for Fluid Dynamics in April 1968 and March 1969, respectively, in which 72 scientists and engineers from seven NATO nations participated. P.T.O. AGARDograph No. 163 North Atlantic Treaty Organization, Advisory Group for Aerospace Research and Development SUPERSONIC EJECTORS Edited by J.J.Ginoux Published November 1972 194 pages AGARD-AG-163 629.7.047.2:533.6.011.5

Two Short Courses were organized on "Ejectors" at the von Karman Institute for Fluid Dynamics in April 1968 and March 1969, respectively, in which 72 scientists and engineers from seven NATO nations participated. P.T.O.

Two Short Courses were organized on "Ejectors" at the von Karman Institute for Fluid Dynamics in April 1968 and March 1969, respectively, in which 72 scientists and engineers from seven NATO nations participated. P.T.O.

They were partly supported by AGARD and directed by Dr H.Uebelhack, Assistant Professor at VKI, now with Dornier System in Friedrichshafen, Germany. The objective of these Short Courses was to present a state-of-the-art review of the significant progress which has been made in the past few years in the design of high performance ejectors. AGARD has felt it appropriate to publish the Short Courses as an AGARDograph in an updated version, with emphasis on supersonic ejectors. A small number of copies were originally printed at the time of these courses, but numerous subsequent requests for these notes were received and the demand could not be satisfied.

They were partly supported by AGARD and directed by Dr H.Uebelhack, Assistant Professor at VKI, now with Domier System in Friedrichshafen, Germany. The objective of these Short Courses was to present a state-of-the-art review of the significant progress which has been made in the past few years in the design of high performance ejectors. AGARD has felt it appropriate to publish the Short Courses as an AGARDograph in an updated version, with emphasis on supersonic ejectors. A small number of copies were originally printed at the time of these courses, but numerous subsequent requests for these notes were received and the demand could not be satisfied.

They were partly supported by AGARD and directed by Dr H.Uebelhack, Assistant Professor at VKI, now with Dornier System in Friedrichshafen, Germany. The objective of these Short Courses was to present a state-of-the-art review of the significant progress which has been made in the past few years in the design of high performance ejectors. AGARD has felt it appropriate to publish the Short Courses as an AGARDograph in an updated version, with emphasis on supersonic ejectors. A small number of copies were originally printed at the time of these courses, but numerous subsequent requests for these notes were received and the demand could not be satisfied.

They were partly supported by AGARD and directed by Dr H.Uebelhack, Assistant Professor at VKI, now with Domier System in Friedrichshafen, Germany. The objective of these Short Courses was to present a state-of-the-art review of the significant progress which has been made in the past few years in the design of high performance ejectors. AGARD has felt it appropriate to publish the Short Courses as an AGARDograph in an updated version, with emphasis on supersonic ejectors. A small number of copies were originally printed at the time of these courses, but numerous subsequent requests for these notes were received and the demand could not be satisfied.

NATIONAL DISTRIBUTION CENTRES FOR UNCLASSIFIED AGARD PUBLICATIONS Unclassified AGARD publications are distributed to NATO Member Nations through the unclassified National Distribution Centres listed below BELGIUM Coordinateur AGARD - V.S.L. I lat-Major Forces Aeriennes Caserne Prince Baudouin Place Dailly, Bruxelles 3 CANADA Director of Scientific Information Services Defence Research Board Department of National Defence - 'A' Building Ottawa, Ontario DENMARK Danish Defence Research Board Osterbrogades Kaserne Copenhagen 0 FRANCE O.N.E.R.A. (Direction) 29, Avenue de la Division Leclerc 92, Chatillon-sous-Bagneux GERMANY Zentralstelle fur Luftfahrtdokumentation und Information Maria-Theresia Str. 21 8 Munchen 27 GREECE Hellenic Armed Forces Command D Branch, Athens ICELAND Director of Aviation c/o Flugrad Reykjavik ITALY Aeronautica Militare Ufficio del Delegate Nazionale all'AGARD 3. Piazzale Adenauer Roma/EUR LUXEMBOURG Obtainable through BELGIUM NETHERLANDS Netherlands Delegation to AGARD National Aerospace Laboratory, NLR P.O. Box 126 Delft NORWAY Norwegian Defense Research Establishment Main Library. P.O. Box 25 N-2007 Kjeller PORTUGAL Direccao do Service de Material da Forca Aerea Rua de Escola Politecnica 42 Lisboa Attn of AGARD National Delegate TURKEY Turkish General Staff (ARGE) Ankara UNITED KINGDOM Defence Research Information Centre Station Square House St. Mary Cray Orpington, Kent BRS 3RE

UNITED STATES National Aeronautics and Space Administration (NASA) Langley Field, Virginia 23365 Attn: Report Distribution and Storage Unit If copies of the original publication are not available at these centres, the following may be purchased from: Microfiche or Photocopy National Technical Information Service (NTIS) 5285 Port Royal Road Springfield Virginia 22151, USA Microfiche KSRO/HLDO Space Documentation Service European Space Research Organization 114, Avenue Charles de Gaulle 92200, Neuilly sur Seine, France Microfiche Technology Reports Centre (DTI) Station Square House St. Mary Cray Orpington, Kent BRS 3RE England

The request for microfiche or photocopy of an AGARD document should include the AGARD serial number, title, author or editor, and publication date. Requests to NTIS should include the NASA accession report number. Full bibliographical references and abstracts of the newly issued AGARD publications are given in the following bi-monthly abstract journals with indexes: Scientific published Scientific P.O. Box Maryland and Technical Aerospace Reports (STAR) by NASA, and Technical Information Facility, 33, College Park, 20740, USA United States Government Research and Development Report Index (USGDRI), published by the Clearinghouse for Federal Scientific and Technical Information, Springfield, Virginia 22151, USA

*
Printed by Technical Editing and Reproduction Ltd Harford House. 7-9 Charlotte St. London. WIP IHD

Vous aimerez peut-être aussi