L. R.

ANDERSON Axisymmetric One Dimensional Compressible

Assistant Project Engineer.

[Link]
Flow—Theory and Applications
Senior Project Engineer.
Several generations of engineers have found that the one-dimensional approach to com-
pressible fluid flows in ducts is extremely informative and effective. This paper extends
[Link] these methods to axisymmetric flows so that they may be used, to understand and estimate
Analytical Engineer. the behavior of azimuthally unrestrained, swirling flows. The already-established
understanding of duct flows makes an elaborate investigation of the new results unneces-
Turbine Aerodynamic Research Group, sary. The paper contains the underlying assumptions and resulting equations, an
Pratt & Whitney Aircraft,
East Hartford, Conn. outline of the solution technique, a comprehensive table of influence coefficients, and a
variety of interesting examples peculiar to swirling flows drawn from the field of rotating
machinery.

Introduction amples which are based upon the authors' experience in the jet

w, HEN dealing with practical flow problems involving

compressible fluids, even in the face of the combined effects of
engine industiy are included in the third section of this paper.

area change, mass and heat addition, friction, and many other Theoretical Development
phenomena, a one-dimensional model is oftentimes found to yield The primary goal here is to derive expressions for the incre-
useful answers. Through the extensive circulation of the text by mental changes in all important flow properties when the flow is
Shapiro [ l ] , 1 the solution of this problem for ducted flows by Sha- subject to imposed perturbations. The difference equations
piro and Hawthorne [2] is probably the best known. The thus arrived at represent a complete nonlinear solution to the
beauty of their analysis lies not only in the quantitative results, problem and can, in principle, be integrated between suitable
which could have been obtained in a number of alternate ways, boundary conditions in order to determine finite changes, but
but also in the fact that their formulation affords one a clear their greatest practical use is to indicate the directions and rates
insight into the roles t h a t the various mechanisms play in de- of change.
termining the nature of the flow process. This understanding is Assumptions. The analysis is carried out assuming the following:
at least as important as the ability to accurately predict the flow
behavior. 1 The flow is axisymmetric.
In the turbomachinery industry, and particularly in jet engine 2 Property variations normal to the axisymmetric stream
turbine applications, flows are often found to be primarily cir- surface are not important. Fig. 1 contains schematic drawings
cumferential. Frequent encounters with complex problems of the axisymmetric stream surface coordinate system used in this
involving swirling flows have prompted the extension of the one- paper.
dimensional approach to axisymmetric flows. I t should not be 3 The flow is steady.
overlooked that the resulting method of solution is applicable As a result of assumptions 1-3, all properties depend only upon
without regard to the relative sizes of the axial and radial com- either the streamwise position or the meridional position. I t
ponents of velocity. In this manner, many problems, whose is more convenient to base the coordinate system upon x, the
solutions were previously the subject of speculation or were meridional position, in the analysis that follows.
discovered only after many hours of tedious calculations, now 4 Stream properties change in a continuous manner.
become transparent. In order to illustrate this point, a few ex- 5 Any injected flow is the same species gas as the main flow.
Consistent with assumption 2, the component of the injected
1
Numbers in brackets designate References at end of paper. flow velocity normal to the axisymmetric stream surface is
Presented at the Gas Turbine Conference and Products Show, neglected.
Brussels, Belgium, May 24-28, 1970. 6 The gas is perfect; i.e., it obeys both Charles' and Boyle's
Discussion of this paper should be addressed to the Editorial De-
partment, ASME, United Engineering Center, 345 East 47th Street, laws and possesses constant specific heats.
New York, N. Y. 10017, and will be accepted until January 20, 1971. 7 Any friction force acts directly opposite to the local stream-
Discussion received after the closing date will be returned. Manu- wise direction.
script received at ASME Headquarters, February 2, 1970; final
revision, July 10, 1970. Paper No. 70-GT-82. Independent Variables. T h e analysis presented here was de-

Journal of Applied Mechanics DECEMBER 1970 / 917

Copyright © 1970 by ASME

Downloaded From: [Link] on 02/01/2016 Terms of Use: [Link]

 STREAMLINE

Fig. 1(a) Axisymmetric stream surface

Fig. 1(b) Coordinate system for asixymmetric flow

•Nomenclature-
A = stream-tube area, normal to merid- T = absolute temperature ft = angle between V and V
ional direction u = meridional component of velocity yx = meridional injection velocity ratio,
Aw — wetted area M' = meridional component of velocity
u'/u
of injected flow
cp = specific heat at constant pressure y$ = azimuthal injection velocity ratio,
v = azimuthal component of velocity
AA v' = azimuthal component of velocity v'/v
D = mean hydraulic diameter, — 9 = applied azimuthal drag force
of injected flow
dAw/dx
V = total velocity p = density
f = skin-friction coefficient
V = total velocity of injected flow, 0 = angle between meridional and axial
k = ratio of specific heats
[(M')2+(«')21V2 directions
M = Mach number = mass rate of flow
p = absolute pressure = meridional distance Subscripts
r = radial location of streamline = applied meridional drag force 0 = total or stagnation condition
= flow angle in stream surface,
s = entropy per unit mass x = meridional direction
measured from azimuthal direc-
S = total stveamwise applied drag force, tion 6 = azimuthal direction
X sin a + 9 cos a

918 / DECEMBER 1 970 Transactions of the ASME

Downloaded From: [Link] on 02/01/2016 Terms of Use: [Link]

 CONTROL VOLUME (EXTENDS d 9 INTO PAGE)
veloped for use in estimating machine performance. The in-
dependent variables were chosen as those which are ordinarily
specified in advance, in particular the entire geometry of the
axisymmetric stream surface. The independent variables ac-
counted for are as follows:
STREAM SURFACES
1 Stream-tube area.
2 Streamline radial location.
3 Wall friction.
4 External forces. S„EA»TUBE

5 Energy addition.
6 Mass injection.
The foregoing selection is arbitrary in the sense that it reflects
only one set of zieeds. I t is certainly possible to choose other
combinations of independent variables in accordance with the
problem at hand. For example, it might be desirable for the
purpose of design to prescribe the meridional pressure distribu-
tion and have the stream-tube area variation result from the cal-
culations. The approach described later is applicable to any
such situation, even though the final results are restricted to the
stated choice of independent variables.
I t should also be noted t h a t all differential equations will be Fig. 2 Stream-tube control volume
found in the convenient dimensionless logarithmic form, with
dependent variables appearing to the left of the equal sign and
independent variables to the right.
Conservation Equations. Four of the six equations necessary to equations are also linear in both dependent and independent
determine differential property changes are obtained by applying terms so t h a t superposition may be used to greatly simplify the
the four relevant conservation laws to an arbitrary control solution. As a result, it is convenient to maintain the groups of
volume, such as that shown in Fig. 2. The results are as follows: independent terms intact exactly as they appear in equations
Conservation of Mass (1)~(4) to the conclusion of the analysis.
Once the four simultaneous equations have been solved, and
dp du dw dA the differential changes of the six dependent variables have been
(1) obtained, then other desirable difference quantities may be ob-
p u w A
tained from their definition and direct solution, such as:
Conservation of Energy
Mach Number
= (1+^iM.) (2) dM2 dV dT
2 (7)
1A2 ~ y" T
Conservation of Meridional Momentum
Meridional Component of M
fji/Y) fill ChV dw
(- kM2 sin 2 a — = &M2 cos2 a — — &M2 sin 2 a ( l — yx) — dM2 du dT
p u r w = 2— - (8)
dX A;M2 4/cfa M, ! u T
2 ^kpAM2
m 2 D^ v ;
Flow Angle
Conservation of Azimuthal Moment of Momentum (du dV\
da \ ^ - ~ ) U n a (9)
dv dr dw s 1 dQ 1 4fdx
— = (1 - ye) — sin 2a j^kpAM2 2 sin a D
v r w Total Pressure
(4)
kM2
Auxiliary Equations. The remaining two differential equations
dpo dp 2 dM2
do not require the use of a control volume, and are as follows: (10)
Po P + 1 +
k
_^l M2 M2
Perfect Cas Law
dp dp dT Entropy
= 0 (5)
V P dT 1 dp
ds
(11)
Velocity Diagram T k V

dV du 2
dv Influence Coefficients. The results of these computations, valid
sm' a — — cos" /Va =z (6) for any arbitrary angle of stream-tube inclination, (j>, are sum-
marized in the table of influence coefficients in Fig. 3. Notice
Solution Procedure. Algebraic solution of these six linear simul- that the four columns are the natural groupings of independent
taneous differential equations yields the relations governing the terms which were the right-hand sides of the conservation equa-
differential changes of the six dependent variables, p, T, p, u, v, tions. The (1 — M 2 sin 2 a ) terms were attached to the dependent
and V. Note t h a t equations (4) and (6) may be used to eliminate terms merely to simplify the table.
du and dv directly from the remaining four equations, and un- By way of example, the table of influence coefficients shows
couple them from the simultaneous solution for the remaining t h a t the differential change of total pressure is given by the ex-
four unknowns. Furthermore, the four resulting simultaneous pression :

Journal of Applied Mechanics DECEMBER 1970 / 919

Downloaded From: [Link] on 02/01/2016 Terms of Use: [Link]

 CONSERVATION OF
CONSERVATION CONSERVATION AZIMUTHAL MOMENT CONSERVATION OF
ORIGIN OF SOURCE TERMS OF MASS OF ENERGY OF MOMENTUM MERIDIONAL MOMENTUM
dw dA k-1 , dT_ dr ,. •-, , dw
> v SOURCE
(1+-M2)—° - k M 2 c o s 2 a — + k M 2 s i n 2 a ( l - x' 7 v ) i l
w A + (1-Tfl)— + r w
1
^"""V. TERM 'o r w
DEPENDENT — , . 1 de 1 4fdx kM2 dX kM2 4fdx
VARIABLE ^ ^ ^ +
Sinto'/ikpAM 2 2sina D 2 VikpAM2 +
2 Slna
T
2 2 k-1 , , k-1
£!l(l-M sin a) 2(1 + M 2 )sin 2 a 2 2
(1 + kM ) sin a-cos a 2
-2(1 + —M2)cos2a(l-kM2sih2a) 2(1 + —M 2 )sin 2 a
M2

dM x 2 , , k-1 , ,
k-1 , , 2(1 + — M 2 s i n 2 a )
—- l, (1-M 2 sin 2 a) 2(1 + — M 2 s i n 2 a ) 1 + kM 2 sin 2 a (k-l)M 2 cos 2 a(l + kM 2 sin 2 a)
M

kM4 , _{l+(k-l)M 2 sin 2 a}

-^•(1-M 2 sin 2 a) -kM 2 sin 2 a -kM 2 sin 2 a - ( k - 1 ) — sin2 2a
4

—•(l-M'sin'a) -(k-l)M 2 sin 2 a l-kM 2 sin 2 a (k-l)M 2 cos 2 a(l-kM 2 sin 2 a) -<k-l)M 2 sin 2 a

dp
-Kl-M 2 sin 2 a) -M 2 sin 2 a -1 -<k-l)M 2 cos 2 a -1
P
dV
sin2 a sin 2 a -<l-kM 2 sin 2 a)cos 2 a sin2 a
—(1-M 2 sin 2 a)

dv
0 0 -1 0
V

du
— (1-M 2 sin 2 a) 1 1 (k-l)M 2 cos 2 a 1
u

da(l-M 2 sin 2 a) i sin2a •y- sin 2a | l +(kcos 2 a-l)M 2 (I-sin2a -Lsin2a

2
2
kM
0 2 —kM 2 cos 2 a -1
Po 2
1+i^M

ds k-l
0 1 (k-l)M 2 cos 2 a

Fig. 3 Table of influence coefficients

kM2 examples which reveal behavior peculiar to swirling flows and

which show that there are many different ways to make use of the
2 7 i - l „ \ dT0 tabled influence coefficients.
k - 1 The following examples all arose in connection with jet engine
1 + M 2
^ applications, and the origin of each is made clear.
dr dw Behavior of Total Pressure. The most important commodity in
-kM2 cos2 a - + (1 - 7«) — fluid machinery is total pressure and designers take every pre-
r w caution to prevent its dissipation. The one-dimensional ap-
1 d6 1 4Jdx proach can be used to particularly good advantage here because
+ sin 2a \k-pAM2 2 sin a D it informs the designer about the relative importance of each loss
mechanism. Rearranging equation (12), it is found t h a t
dr , dw
-kM2 cos2 a h kM2 sin 2 a ( l - 7 J — dpo V cos 0 dw
r w = -kM2 1 -
2 2 po V w
kM dX kM . 4:fdx
+ (12) dS 21 dTo
^ikpAM2 +
^ S m a
^ . + (13)
kpAM2 sin a ' D \sm a J + 2r„
Examples The interpretation of equation (13) is straightforward and worth-
The influence coefficients may now be used to obtain a complete while. To begin with, imposed changes in r and A have no effect
one-dimensional solution for any given flow by allowing inte- upon total pressure, which is an entirely expected result. Next,
gration between suitable boundary conditions. At this point it is it can be seen t h a t total pressure losses are generally proportional
entirety possible to repeat the steps already taken by Shapiro and to kM2, which implies t h a t low streamwise Mach number opera-
Hawthorne in order to show t h a t the new method leads to cor- tion is conducive to low total pressure losses. However, since
rect and interpretable results. However, the strong similarities the power produced in a rotating machine also tends to be pro-
between the two methods, as well as the thoroughness of the portional to kM2, it is therefore entirely possible to design high
previous examination, make it unnecessary to construct an Mach number machines that are efficient.
elaborate verification of the effectiveness of the new method. I t The terms inside the bracket of equation (13) can be inspected
is more interesting and instructive, instead, to examine several individually.

920 / D E C E M B E R 1970 Transactions of the ASME

Downloaded From: [Link] on 02/01/2016 Terms of Use: [Link]

 (a) External Forces interesting, then, to examine the potential flow behavior of
swirling flows in which only A and r are allowed to vary. For
dS
this case the influence coefficients reveal that
kpAM1 sin a
dp kMr dA fctvV dr
It is interesting to note that only those portions of the external + (15)
V 1 -M,< A 1 - MT*
forces which are locally parallel to the stream can affect the total
pressure. This is directly related to the notion t h a t a stream can An interpretation of equation (15) is that meridional velocity
be isentropioally deflected only by a force locally normal to the may be exchanged for pressure largely through stream-tube area
direction of flow. The trigonometric term reflects the result of variation, while azimuthal velocity may be exchanged for pres-
extending the residence time within dx. sure largely through variation of streamline radial location.
This reasoning has already found application in the exhaust ducts
(b) Mass Injection of Moody turbines [3].
The table of influence coefficients indicates that for swirling
dw flows, unlike purely axial flows, all quantities pass without diffi-
( - ^ w culty through M = 1. Instead, a singularity is now found to
exist for most quantities when the meridional component of Mach
Here, again, it appears that the total pressure is only affected by number reaches one. For example,
the component of momentum locally parallel to the stream.
Total pressure can, in fact, be best preserved by injecting the
flow in the direction of the mainstream.
dMx*
M»
»K-iX 1 - M»
' dA
A
(Jc + l)Mo2 dr
1 -M »
(16)
(c) Energy Addition
This concept is not new nor is it inconsistent with its axial flow
dT0 counterpart in that one does not expect the forms of the physical
2To governing equations to change as long as azimuthally symmetric
pulses, moving at the speed of sound, can make themselves felt
The direction of the flow does not affect total pressure losses due upstream. The conclusions concerning transition through the
to burning or heat addition. choking point drawn by Shapiro and Hawthorne [2] remain es-
sentially unaltered. However, their arguments must now be
(d) Wall Friction
applied about M x equal to one.
Equation (16) indicates that, in an isentropic swirling flow, it is
possible to pass through the choking condition by variations in A
D \ s m a/
and/or r. For example, when only r can vary, equation (16)
The trigonometric term appears here because da;/sin a is the in derivative form is
actual distance t h a t the flow must cover in traversing dx.
dMJ (k + ljM/Mo 2 1 dr
An interesting application of equation (13) is associated with (17)
the possibility of using swirl to improve the performance of dx 1 - M2 r dx
annular combustion chambers. This concept appears attractive
Equation (17) shows that for variations in r to cause choking,
because it can reduce total pressure losses in a number of ways,
some swirl must be present and, furthermore, when M , = 1,
including:
dr/dx must equal zero for dMx2/dx to be finite. Further exami-
1 Lessening the turning requirements on the last compressor nation would lead to usual conclusions concerning the presence of
and the first turbine guide vanes. a minimum r at the choking point.
2 The possibility of increasing the burner diffuser total pres- On the other hand, when only A can vary, a similar argument
sure recovery, as suggested by [3]. would show that choking can only occur at the minimum stream-
3 The possibility of substantially decreasing the length of tube area A when the flow is isentropic. An interesting point
burner required for complete combustion, as suggested by [4], may be made for the case of constant radius isentropic flow,
so that the total wetted area for skin friction is reduced. where the continuity equation reveals that the flow per unit of
area normal to the flow direction, w/(A sin a ) , must still reach a
Since an important fraction of the combustion chamber total
maximum when M = 1. This implies t h a t the area normal to the
pressure loss is due to heat addition, it is important to know what
flow direction must begin to increase as M increases beyond one
effect the swirl will have on this phenomenon. Equation (13)
even though it has already been shown that the stream-tube area
directly reveals that
continues to decrease up until the point where M , is equal to one.
dpo kM1 dTo The expression that describes this phenomenon, derived at con-
(14) stant streamline radius r, from the table of influence coefficients,
2 T\
is
and, since M 2 necessarily increases with swirl while dTo/To is
M2
fixed, then it follows that swirl increases the total pressure loss due
to combustion. The one-dimensional approach shows, without
any further sophisticated analysis, that the detrimental effects of
d(A sin a)
A sin a = sin' a (a -
M,
dA
A
(18)

Therefore, as postulated, in the regime where M > 1 but

burning must be accounted for in any fair analytical or experi-
M , < 1, the flow swings toward meridional rapidly enough to
mental study of such devices.
allow the area normal to the flow to increase while the stream-tube
Potential Flow and Choking. I t is quite usual in the design of area continues to decrease. Finally, the reduction in maximum
compressible flow devices to make preliminary estimates of the flow rate per unit area due to swirl which may be found by taking
potential flow pressure distribution and of the choking condition. a ratio of choking flow parameters for fixed k, p0, To, r, and A, is
The former is useful in determining whether the boundary layer given by the relation:
will remain attached, and the latter may be used to find the weight
flow capacity of the device. ft-1
Variations in stream-tube area A and streamline radius r have KW90° '{k + l j s i n ' a n 2 ^ - 1 )
(19)
already been shown to have no effect on total pressure. I t is ™ a =90° k — cos 2 a J

Journal of Applied Mechanics DECEMBER 197 0 / 921

Downloaded From: [Link] on 02/01/2016 Terms of Use: [Link]

 GENERAL PROPERTIES

CASCADE PITCH, 1.20"

EXIT ANGLE. 20.8°

TRAI LING-EDGE THICKNESS. 0.06"

INJECTED COOLANT FLOW SLOT WIDTH, 0.0375"

cp = 0.24BTU/lbm

7 = 1.4

INJECTED COOLANT FLOW

INJECTED COOLANT
MAINSTREAM FLOW PROPERTIES
FLOW PROPERTIES

w = 1.0 lb /sec
P P
" MAINSTREAM
T - 2920 O n
o T Q - 2920°R, CASE 1
1460°R,CASE2
M - 0.34

Fig. 4 Trailing-edge ejection airfoil cascade

Estimating Plane Cascade Accuracy. When measuring compressor For modern compressor and turbine cascades, all of the co-
or turbine plane cascade performance, the experimenter ordinarily efficients found in equation (20) are of the order of one. Further-
prefers to measure close to the trailing-edge plane (near field) of more, the actual fractional total pressure loss across a cascade
the airfoils in order to obtain the details of the loss pattern as well is only of the order of a few percent. Therefore, one must con-
as to minimize tunnel wall effects. These requirements conflict clude that fractional errors in the near-field integral quantities
with his desire to also obtain a measurement of the ultimate uni- must be kept much less than a few percent in order to allow this
form (far-field) state of the flow after all mixing and shock losses method of analysis to be applied..
have taken place. This is so because highly cooled and high Convectively Cooled Turbine Airfoils With Trailing-Edge Coolant
Mach number airfoils m a y undergo so much additional loss in Ejection. After the coolant air in a convectively cooled turbine,
passing from the near field to the far field t h a t a mass-average of airfoil has served its purpose, it is oftentimes disposed of through
total pressure in the near field is not an accurate measure of the trailing-edge slots or holes. Assuming that the injection of the
ultimate value reached in the uniform state. coolant flow has no effect on the mainstream flow conditions at the
A means for circumventing the use of simple mass-averaging trailing-edge plane, their mass, momenta, and energy fluxes may
techniques in the near field is to integrate the mass, momenta, be added in the usual manner, and the conservation equations can
and energy fluxes in the near field and to apply the conservation be employed to ascertain the uniform conditions which would
equations in order to determine, uniquely, the uniform, far-field ultimately be reached after mixing.
state of the flow. Now the error in the computed far-field total The method of influence coefficients offers a simple, y e t fairly
pressure is only a function of the errors of each of the four near- accurate, substitute for this more complex exact solution. For
field integral quantities, each of which can be estimated from a simplicity, in the following example, the effects of airfoil surface
knowledge of the measuring devices employed. boundary layers and trailing-edge width are omitted and only the
For this purpose, the influence coefficients can be effectively effects of cooling air ejection, parallel to and in the direction of
used if the imposed changes are taken to represent the estimated the mainstream flow, are considered. The cascade used in this
errors of near-field integral quantities. Equation (12) can be example, along with pertinent data for the example cases, is
easily rearranged to yield shown in Fig. 4.
As always, the difficult part of the analysis relates to the
A(p + pV2sm1a) dissipative phenomena while accountings of mass, momenta,
= - ( l + &M 2 sin 2 a) X
p -f- pV sin 2 a
2
etc., are straightforward. Knowing the far-field total pressure
is equivalent to having determined all the effects of dissipation,
Fractional far-field Fractional error in near- and hence, for this case, the relevant form of equation (13) is
total pressure error field meridional momentum
integral
dpo
A(pF 2 sin a cos a ) = -kM2 1 + (21)
— &M2 cos2 a X
pV2 sin a cos a
* ) w 2r„
: ( '~~- ~Vt
V is determined from the knowledge of the injected flow's slot
Fractional error in width, total temperature, mass-flow rate, and static pressure,
near-field azimuthal
moment of momentum which is assumed the same as the mainstream static pressure.
integral dTv/Tti is simply obtained by an energy balance between the
mainstream and injected mass flows.
A ( p F sin a) kM2 Calculations were done using equation (21) and the exact
-kM2 X X (20)
pV sin a 2 theory for two cases:

Fractional error in near- Fractional error in near- Case 1: Mass injection a t the mainstream total temperature.
field mass-flow integral field total energy integral Case 2: Mass injection a t one half of the mainstream total
temperature.
where Motional and applied force effects can be taken to be zero
downstream of the cascade. The results of the comparative calculations for Cases 1 and 2

922 / D E C E M B E R 1970 Transactions of the ASME

Downloaded From: [Link] on 02/01/2016 Terms of Use: [Link]

 PERCENT INJECTED MASS FLOW PERCENT INJECTED MASS FLOW

Fig. 5(a) Theoretical far-field total pressure loss as a function of cooling Fig. 5(b) Theoretical far-field pressure loss as a function of cooling
airflow rate, Case 1 airflow rate, Case 2

are shown in Figs. 5(a) and 5(6), respectively. Notice that the should be considered when first attacking any axisymmetric flow
range of injected flow rates and temperature ratios more than problem. This is due, in no small part, to the similarity of the
cover the combinations of these two parameters that will be put results and the interpretations with those of fully clarified axial
into practice in the near future. Fig. 5 demonstrates the feasi- flows.
bility of using this approximate technique for generating pre-
liminary estimates of the losses associated with this type of
cooling scheme. References
1 Shapiro, A. H,, The Dynamics and Thermodynamics of Comjiress-
ible Fluid Flow, The Ronald Press Company, New York, 1953.
Concluding Remarks 2 Shapiro, A. H., and Hawthorne, W. R., "The Mechanics and
Past experience has shown that many flow situations can be Thermodynamics of Steady, One-Dimensional Gas Flow," JOURNAL
OF APPLIED MECHANICS, Vol. 14, No. 4, TKANS. ASME, Vol. 69,
understood and rapid quantitative estimates can be made by
Dec. 1947, pp. A-317-336.
means of the techniques presented in this paper. I t has also
3 Patterson, G. N., "Modern Diffuser Design," Aircraft Engineer-
been found that these methods can be applied in a great variety ing, Vol. 10, No. 15, Sept., 1938.
of ways, the application being limited only by the imagination of 4 Chervinsky, A., "Turbulent Swirling Jet Diffusion Flames,"
the user. For these reasons, the method of influence coefficients AIAA Journal, Vol. 7, No. 10. Oct. 1969.

Journal of Applied Mechanics DECEMBER 1970 / 923

Downloaded From: [Link] on 02/01/2016 Terms of Use: [Link]