A Quasi-Three-Dimensional

Turbomachinery Blade Design

i. K. Jennions
System: Part I—Throughflow
Principal Theoretical Scientist.
Analysis
P. Stow The purpose of this work has been to develop a quasi-three-dimensional blade
Head of Theoretical Science. design and analysis system incorporating fully linked throughflow, blade-to-blade
Theoretical Science Group, and blade section stacking programs. In Part I of the paper, the throughflow
Rolls-Royce Limited, analysis is developed. This is based on a rigorous passage averaging technique to
Derby 0E2 8BJ, England derive throughflow equations valid inside a blade row. The advantages of this
approach are that the meridional streamsurface does not have to be of a prescribed
shape, and by introducing density weighted averages the continuity equation is of
an exact form. Included in the equations are the effects of blade blockage, blade
forces, blade-to-blade variations and loss. The solution of the equations is
developed for the well-known streamline curvature method, and the contributions
from these extra effects on the radial equilibrium equation are discussed. Part II of
the paper incorporates the analysis into a quasi-three-dimensional computing
system and demonstrates its operational feasibility.

Introduction
If one considers the flow in a compressor or turbine, there back to the general theory of Wu [1], In his theory, the steady
are rotating and stationary blade rows that change in their flow solution is calculated on two families of intersecting
orientation and cross sectional geometry from hub to tip. streamsurfaces: a family of blade-to-blade streamsurfaces (S{
There are boundary layers on the annulus walls and the blade surfaces) and a family of throughflow streamsurfaces (S2
surfaces, wakes from the trailing edges of blades, overtip surfaces). Calculations are peformed on each family of
leakage flows, cooling flows ejected over the surfaces of streamsurfaces, but contain linkage terms to the other family.
turbine blades etc. The flow is unsteady, compressible, three- The solution is obtained in an iterative manner and is fully
dimensional and viscous, and the solution of the full three-dimensional. Krimerman and Adler [2] have performed
equations of motion with the true boundary conditions such a calculation, but in general this procedure has not been
represents a formidable task. In the past few years, fully adopted. The most common technique is to use only one S2
three-dimensional methods have been developed and are now surface through the blade row, together with a number of S}
in use for turbomachinery blade design. However, even for surfaces. With this approach the S1 surfaces are taken as
steady flow it is likely that for some years to come methods surfaces of revolution and the method is referred to as quasi-
for solving three-dimensional problems will be of such a speed three-dimensional .
that only isolated blade rows, or at best single stages, can be The flow on the S, surfaces can be computed using a variety
computed in the design times available. In order to in- of techniques, e.g., streamline curvature [3], time marching
corporate three-dimensional effects early in the design cycle a [4] and finite element [5]. Information from the throughflow
quasi-three-dimensional design system has been developed. calculation is used to define the S[ blade-to-blade surfaces in
This is efficient and economical, both in man and machine terms of streamline radius and streamtube height and to
time, and is capable of describing some of the more fun- provide flow boundary conditions for blade design. In order
damental aspects of three-dimensional flows. As such it can to obtain accurate streamline shapes through the blade row,
make a significant contribution to a blade design system. intra blade computing stations are necessary, and so an
Most throughflow calculation procedures can be traced analysis to account for the blade force and geometry effects
has to be produced. It is this analysis that is central to all
quasi-three-dimensional methods. Four numerical solution
Contributed by the Gas Turbine Division of THE AMERICAN SOCIETY OF techniques have been used in quasi-three-dimensional
MECHANICAL ENGINEERS and presented at the 29th International Gas Turbine
Conference and Exhibit, Amsterdam, The Netherlands, June 4-7, 1984.
schemes; these are: the streamline curvature method, the
Manuscript received at ASME Headquarters December 19, 1983. Paper No. matrix solution procedure, and the more recent finite element
84-GT-26. and time marching methods.

Journal of Engineering for Gas Turbines and Power APRIL 1985, Vol. 107/301

Copyright © 1985
1984 by ASME
Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use
 The streamline curvature technique has been developed in averaging avoids any indeterminacy associated with defining
papers by Smith [6], Silvester and Hetherington [7], Novak [8] the S2 surface, a problem that has continually been the subject
and Frost [9], with quasi-three-dimensional applications of debate as illustrated by the various alternative approaches
reported by Senoo and Nakase [10] and Novak and Hearsey that have been adopted [11, 13]. The passage averaged
[11]. Senoo and Nakase defined their S2 surface to coincide throughflow equations contain the effects of the blade force,
with the blade mean camberline and demonstrated their blade geometry, and perturbations on the mean flow, which
method on a centrifugal impeller. For axial flow turbines, are supplied from blade-to-blade and section stacking
Novak and Hearsey showed that this definition could not be programs. The modular approach adopted in this quasi-three-
expected to give reliable results. Instead they chose their S2 dimensional design system is discussed fully in Part II of this
surface to be generated by S, streamlines that divide the paper [18].
passage into equal mass flows. The work includes the specific The first section of the current paper describes the
mass flow ratio (B), a quantity analogous to the local coef- streamline curvature method as it relates to the present work.
ficient of discharge along each 5] surface, which is used in the The governing equations are then passage averaged. Finally
throughflow continuity equation. B was shown to change the radial equilibrium equation is formed and the new terms
rapidly with meridional distance and was not unity im- appearing in the analysis simplified. The major definitions
mediately upstream and downstream of the blade, regions and consequences of passage averaging may be found in the
where the throughflow program must treat the flow as Appendix.
axisymmetric {B = 1).
In the matrix solution procedure, the throughflow Description of the Streamline Curvature Method
equations are written in terms of a stream function. The
resulting stream function equation is then put into finite The streamline curvature method has been described in
difference form and solved by a matrix inversion technique. some detail by Silvester and Hetherington [7] and Frost [9]. It
Marsh [12] presents the method for the throughflow problem, is based on successive refinements of streamline paths through
while Bosman and El Shaarawi [13] have incorporated the the flow domain by considering the radial equilibrium of the
technique into a quasi-three-dimensional solution. Their gas on each of the calculating stations shown in Fig. 1. This
method defines the S2 surface by S} mass averaged equilibrium can be described by a radial equilibrium equation
streamlines that do not coincide with any actual 5, (REE), which is a first order differential equation relating the
streamlines but are taken to be representative of them all. static pressure gradient to the streamline geometry, whirl
angle, etc. The streamlines provide the axial linking between
The finite element methods presented by Hirsch and
calculating stations and are approximated by a curve fit
Warzee [14] and Habashi and Youngson [15] are very similar
through points of equal mass flow. The solution proceeds by
to the matrix method, using a stream function as the
guessing the midspan static pressure from which the REE can
dependent variable but adopting a finite element rather than a
be solved and the mass flow through the station calculated.
finite difference solution technique. Hirsch and Warzee
The midspan pressure is then adjusted until the desired mass
employ a passage averaging technique similar to that used by
flow through the station is attained, and the streamlines
Smith [6], but extend it by using mass averaged quantities to
updated using the new flow variables. The whole procedure is
reduce the interaction terms between the S, surfaces and the
repeated on each calculating station from inlet to exit suc-
meridional flow. Their quasi-three-dimensional system is
cessively until changes in the flow variables are small.
integrated into a single program, and is therefore aimed at
analysis type calculations. Governing Equations
Denton has shown the feasibility of using a time marching
method for solving both the throughflow [16] and the blade- The steady three-dimensional flow through a turbomachine
to-blade [4] problems. A quasi-three-dimensional application is governed by the equations of mass conservation,
incorporating both solution procedures has been reported by momentum, energy, and state. In a coordinate system
Spurr [17]. This involved the calculation of the transonic flow rotating with the blade at a constant angular velocity fi the
(in the meridional plane) in an annular cascade, and hence continuity equation is
was ideally suited to a time marching approach. V(pW) = 0 (1)
The present paper describes a rigorous passage averaging
and the inviscid momentum equations are written as
technique similar to that used by Hirsch, which is used to
derive the throughflow equations valid inside a blade row for
(W.V)W= - - Vp + fi2R-2fixW + FT (2)
use with an existing streamline curvature method. Passage p

Nomenclature

B = blade blockage term t = static temperature

C„ = specific heat at constant \y = relative velocity vector / = perturbation about a passage
pressure x distance along machine axis average
FB = blade force relative whirl angle / / = perturbation about a density
dissipative body force 8 radial blade lean weighted average
/ = rothalpy A hade angle
m = direction along a meridional circumferential direction Subscripts
streamline P density c = mean cambersurface
m mass flow rate stream function H = hub
N number of blades Q angular velocity of the blade p = pressure surface
P static pressure row R = radial component
P perturbation term axial blade lean s = suction surface
R radius from machine axis; gas T = tip
constant Superscripts x = axial component
entropy; true streamwise = passage average d = circumferential component
direction = density weighted average 4> = along a streamline

302/Vol. 107, APRIL 1985 Transactions of the AS ME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 NOZZLE
GUIDE VANE Passage Averaged Equations
In this work, equations (3-6) are averaged across the blade
passage. Two types of average are defined: a passage average
TYPICAL denoted by an overbar and a density weighted average
STREAMLINE
denoted by a tilde. Their use for the terms encountered in the
analysis is developed in the Appendix. It is therefore sufficient
to quote the resulting averaged equations, starting with the
continuity equation

— {BRpWx) + — {BRpWR) = 0 (10)

ox oR
(a) THROUGH FLOW in which B is the blade blockage term defined by
N(ds-ep)
B= (11)
2w
Equation (10) is an exact form of the continuity equation with
a simple physical interpretation for B: it is the non-
(b) BLADE-TO BLADE dimensional circumferential spacing between the blades. This
can be contrasted with other approaches [11] where in-
tegrating factors are introduced to take account of the dif-
ferent means used. The form of equation (10) is important in
a streamline curvature method as we can now define a stream
function by
R + 5R
{ BRpWxdR (12)
R
TYPICAL
STREAMLINE from which the total mass flow through any radial calculating
station can be seen to be
Fig. 1 Geometry definitions T
m = 2%\ BRpWxdR (13)

where W is the relative velocity vector and F T is a dissipative The averaged momentum equations are
body force term. This F T term was discussed by Horlock [19]
Wx
and later by Wennerstrom [20]; it accounts for entropy in- Wv + Fltr+Frr+Pr (14)
creases in adiabatic flows and is related to the entropy field in bx p dx
a later section. In an (x, R, 6) cylindrical polar coordinate
(We+QR)2 1 dp
system, equations (1) and (2) are x +FBR+FTR+PR (15)
dx R ~P m.
yx(RpWx)+ ^(RpWR)+~(pWt)^Q (3) and
Wx d -
=Fm+Frfl+Pl1 (16)
~ (RPK)+ ^ («PW« Wx) + ~ (PW»WX) -jf.-HW.+WRl
in which differentiation along a streamline is defined by
dp 8
= -Rf+RpFTX (4) — +tan\ — (17)
ox dx dx dR
1
and the hade angle (A) is given by
l(RpWxWR)+~(RpWl)+~(PWflWR)-p(Wf) + aR)'
tanX= WRIWX 08)
The deviation from axisymmetric duct flow
= -Rd£+RPFTR (5) a ds
( n ,A

~ (RpWxWe)+ ^ {RPWRWe)+ ^ ( P W1) can be seen in the last three terms in each of equations
dx dR do
(14-16). The first of these is the blade force term and is related
to the pressure difference across the blade and the blade
+ pWR(We+2QR)=-^+RpFTe (6) geometry by

The energy equation is written in its customary form for

N / des aeB\
adiabatic flow as 2-KBP dx
W.V/=0 (7) N (/ ,Mdd,JL_ den
.dOjA
[Ps dR PP dR ) (20)
i.e., the rothalpy (I) defined by 2irBp
1 fi2R2 and
I=Cpt+-(W*+Wi + W%>- — (8)
N
FRH = • (P' -P's) (21)
is constant along a streamline. Finally the equation of state 2-wBpR p
for a perfect gas is written as
P=p/Rt (9) Rolls-Royce terminology

Journal of Engineering for Gas Turbines and Power APRIL 1985, Vol. 107/303

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 The second is the consistent loss term, while the third term component, which is determined directly from the change in
involves the perturbation terms which are given by angular momentum. A convenient way to do this is to select
the mean cambersurface and relate it to the suction and
1 [f —
9 (BRpW;
- — -^^,- 2 ) + —
d (BRpW' W\)] (22) pressure surface 6 coordinates by
Pr=- X
dp = dc + Tr/N(l-B) (27)
and
BRp 0s = 0c + ir/N(\+B) (28)
(23)
Differentiating these expressions, substituting them into
P equation (20) and simplifying the result using equation (21)
« = -EST \ ~ {BRpW;wn
BRp L 9x
+ ^ (BRpW'R WD
oR } produces
1 dB
WRWj (29)
(24) FBR = -Fmt&nb+ — — (ps+Pp)
+ R
in which the radial blade lean angle (<5) is defined by
The evaulation of the blade force, loss and perturbation terms
will be discussed after the REE has been derived. ddr
tan5 = .K- (30)
dR
Radial Equilibrium Equation Similarly for the x-component of the blade force
In early formulations of the streamline curvature method 1 dB
applied to duct flows, the REE was derived in such a way that FBx: -F S9 tan<j+ — — (Ps+P'p) (31)
the only axial derivatives in the equation involved streamline 2pB dx
hade and curvature. When flow inside a blade row is con- with the axial blade lean angle (o) defined by
sidered, this is not necessarily the most appropriate approach. ddc
In the work presented here, we have elected to form a REE tantr=R- (32)
from the averaged R and x momentum equations only, in dx
order to avoid numerical singularities and maintain both
simplicity and a clear physical interpretation of the terms Dissipative Force Term
involved.
A prescribed loss model is adopted in this work and we need
In general, the REE would be formulated in a local to relate the dissipative force terms in our momentum
coordinate system along a curved calculating station, but in equations to the entropy. Using the definition of rothalpy
order to clarify the analysis, the REE for purely radial (equation (8)) the Gibbs relation can be written as
stations is derived here. Defining the relative whirl angle (13)
by 1
r v 5 = v / - - vp- v( — ) + Q2R (33)
Xan$=WR/Wx (25) p V 2 /
we can combine equations (14) and (15) to give the desired from which Vp is eliminated by using equation (2), giving
REE as f V i = V / - W x ( V X W ) + 2!)xW-FT (34)
sec2A dp If we now take the scalar product of this equation with W and
use equation (7), we arrive at the entropy/dissipative force
relation
~ , 9tanX (EKttan]3 + nft) 2 tanX dp
fW.Vs=-W.F, (35)
w ~R "" p "dx
A Following [19], we assume that the dissipative force acts along
dx
I II III a streamline, i.e.,
(Fm - F H , t a n \ ) + (FTK -F, v tanX) + (PK -PvtanX) (26) Fr=Frs (36)
IV V VI where s is a unit vector in the streamwise direction. From
The terms contributing to the radial pressure gradient can be equations (35) and (36), assuming that the perturbation terms
identified as follows. are negligible when the resulting equation is averaged, we
have
I Streamline curvature in the meridional plane
- ds
II Centrifuging of the fluid Fr' (37)
III Axial pressure gradient
IV Blade force the three components of which can be written as
V Dissipative body force
t ds]
VI Perturbation (non axisymmetric) effects F„=- (38)
l+tan 2 X+tan 2 /3 dx U
In the next few sections we will examine how to evaluate the
= F„tanX (39)
last three of these terms.
and
Blade Force Terms FT# = F«tan ( 3 (40)
Expressions for the blade force components have been The loss and its gradient would in general come from a blade-
presented in equations (19-21). We could form these terms to-blade program or be prescribed from an experimental
from information obtained from the blade-to-blade and correlation. The loss would include the profile loss of the
geometry programs in the proposed system, but this would blade along with shock losses if these occurred. Any account
result in a redundant circumferential momentum equation. of annulus endwall boundary layers would be included by
Instead, we choose to relate the R and x components to the 6 modifying the loss at the throughflow calculation stage. In

304/Vol. 107, APRIL 1985 Transactions of the ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 adopting the above model, the dissipative force terms in the + F m (tanXtaner - tanS) + P (45)
REE (equation (26)) can be seen to be zero, but the F r f term in
the circumferential momentum equation remains. in which
'dtanX
P=W"
Perturbation Terms
The perturbation terms given by equations (22-24) involve
velocity perturbations in all three coordinate directions. While ) (46)
the Wg and W" terms can readily be found from a blade-to-
blade program, the Wg terms would be more difficult, Fm is found from the 6 momentum equation as
requiring an analysis such as that performed by Abdallah and Wx d
Hamed [21] on a series of cross passage surfaces. We, FD„= • [(WxUaiP + QK)Rl\ . -Fre-Pe (47)
however, have assumed that each «Sj surface is a surface of R dx
revolution so that there is no velocity component normal to in which
the surface. This can be expressed as t tan/3 ds
WR = H^tanX (41) (48)
l + t a n 2 X + tan 2 |8 dx
and
W"R = W'i tanX (42)
P
which considerably simplifies the perturbation terms. Their ^BWPTx'BRpWmU
form in the REE is given by

P,-P,tanX=-^9tanX ^
+ ^" ' ^ dR
+ 3 p t a n X
R
(49)
(43)
dx if R
The continuity equation takes the form
and in the 6 momentum equation by
p• R
R++ dR
SR
1 9 dtanX 5^= BRpWxdR (50)
P» =
BRp dx (BRpwpvz)\. +w:w: dR JR
with the energy equation
+ —^-^tanX (44) 1
R I=Cpt+ - ^ ( l + t a ^ X + t a n 2 / ? )
2
The perturbation terms in equations (43) and (44) are found
from the blade-to-blade solution flow variables that are in-
terpolated onto x=constant lines. The mean and perturbation
+ \W?H + tan2X) fwf\- ^
terms are then evaluated by numerical integration along these
lines using equation (53) et seq. = / ta (51)
The perturbation terms presented here are similar to the G and the equation of state
functions discussed by Smith [6]. In his paper, Smith assumed
that the major flow variables vary linearly across the blade-to- P=p/Rt (52)
blade passage. Here we need no such assumption as we can On internal calculating stations we need loss and turning in
evaluate these terms exactly. common with other calculating stations, but now information
By setting the velocity normal to our S^ surface to zero, we regarding perturbation terms and the blade geometry is also
have constrained the flow by neglecting any mixing in the required. The loss, turning, and perturbation terms are found
spanwise direction. It is possible to expand the analysis to from a blade-to-blade program and contribute directly to the
include a spanwise mixing model which contains the radial blade force. The three-dimensional blade geometry is found
perturbation terms and the effects of mixing in the energy from a stacking program, and includes the blade lean angles
equation. which appear as a multiplier of the blade force. The blade
blockage acts chiefly to modify the continuity equation and
Final Form of the Equations has relatively little influence on the REE. It is through these
interaction terms that a designer can examine some of the
We have derived the equations used in the throughflow major three-dimensional effects inside a blade row.
analysis under the following assumptions:

1 The flow is on surfaces of revolution and can be Acknowledgment

represented by mean and perturbation values with respect to
the 8 coordinate. The authors wish to thank Rolls-Royce Limited for per-
2 The loss terms can be modeled by a dissipative force mission to publish this paper.
which acts in the direction of the total flow using a prescribed
loss model.
References
3 There are no velocity perturbations normal to the St
streamsurfaces. 1 Wu, C. H., "A General Theory of Three-Dimensional Flow in Subsonic
4 The flow is adiabatic. and Supersonic Turbomachines of Axial-, Radial-, and Mixed-Flow Types,"
NACATN 2604, 1952.
2 Krimerman, Y., and Adler, D., "The Complete Three-Dimensional
With these assumptions the radial equilibrium equation is Calculation of the Compressible Flow Field in Turbo Impellers," Jnl. Mech.
sec2X dp Eng. ScL, Vol. 20, No. 3, 1978, pp. 149-158.
3 Wilkinson, D. H., "Calculation of Blade-to-Blade Flow in Tur-
p dR bomachines by Streamline Curvature," Aeronautical Research Council R & M
No. 3704, 1972.
I (iVxtan(3 + QR)2 tanX dp 4 Denton, J. D., " A Time Marching Method for Two- and Three-
= -** — + Dimensional Blade-to-BIade Flows," Aeronautical Research Council R & M
x U R p ~dx No. 3775, 1975.
dx

Journal of Engineering for Gas Turbines and Power APRIL 1985, Vol. 107/305

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 5 Deconinck, H., and Hirsch, C , "Finite Element Methods for Transonic
Blade-to-Blade Calculation in Turbomachines," ASME JOURNAL OF pWeWx = PWsWx = HWe+WiMWx+W£ (57)
ENGINEERING FOR POWER, Vol. 103,1981, pp. 665-677.
6 Smith, L. H . , Jr., "The Radial Equilibrium Equation of Tur- = -pWeWx + pW^Wx' (58)
bomachinery," ASME JOURNAL OF ENGINEERING FOR POWER, Jan. 1966, pp.
1-12. This splitting of terms into a mean and perturbation com-
7 Silvester, M. E., and Hetherington, R., "Three-Dimensional Com- ponent is adopted throughout the analysis. The mean com-
pressible Flow Through Axial Flow Turbomachines," Numerical Analysis—An ponent is the one we work with directly in the throughflow
Introduction, edited by J. Walsh, Ch. II, Pt. Ill, 1966, pp. 182-189. solution, while the perturbation component will be retrieved
8 Novak, R. A., "Streamline Curvature Computing Procedures for Fluid-
Flow Problems," ASME JOURNAL OF ENGINEERING FOR POWER, Oct. 1961, pp. from a blade-to-blade calculation.
478-490. If we now take the d(RpWx)/dx term in the continuity
9 Frost, D. H., " A Streamline Curvature Through-Flow Computer equation, passage averaging, using Leibnitz's theorem, gives
Program for Analysing the Flow Through Axial-Flow Turbomachines,"
Aeronautical Research Council, R & M 3687,1972. 1 res 9 I r a (•«*
10 Senoo, Y., andNakase, Y., "An Analysis of Flow Through a Mixed Flow -~(RpWx)dd=——\-\ RPwxde
Impeller," ASME JOURNAL OF ENGINEERING FOR POWER, Jan. 1972, pp. 43-50. J e„ dx 8, — o„ L dx J e„
11 Novak, R. A., and Hearsey, R. M., " A Nearly Three-Dimensional In-
trablade Computing System for Turbomachinery," Journal of Fluids
Engineering, Mar. 1977, pp. 154-166.
12 Marsh, H., " A Digital Computer Program for the Through-Flow Fluid
(RpWx)s^
ox
+ (RPFWx)p^
dx }
(59)
Mechanics in an Arbitrary Turbomachine using a Matrix Method," which, as we are working on streamsurfaces that are surfaces
Aeronautical Research Council R & M No. 3509,1968.
13 Bosnian, C , and El-Shaarawi, M. A. I., "Quasi-Three-Dimensional of revolution (i.e., R^R(6)), is
Numerical Solution of Flow in Turbomachines," Journal of Fluids 1 f's 3
Engineering, Mar. 1977, pp. 132-140. — ~(RpWx)de
14 Hirsch, C , and Warzee, G., "An Integrated Quasi-3D Finite Element Ad J ep dx
Calculation Program for Turbomachinery Flows," ASME JOURNAL OF
ENGINEERING FOR POWER, 1979, pp. 141-148. _ 1 d
15 Habashi, W. G., and Youngson, G. G., "ATransonic Quasi-3D Analysis {BRpWx^^\{RpWx)p^
for Gas Turbine Engines Including Split-Flow Capability for Turbofans," ~B ~dx
International Journal for Numerical Methods in Fluids, Vol. 3,1983, pp. 1-21.
16 Denton, J. D., "Throughflow Calculations for Transonic Axial Flow
(60)
Turbines," ASME JOURNAL OF ENGINEERING FOR POWER, Vol. 100, 1978, pp.
212-218. •«**>.£]
17 Spurr, A., "The Prediction of 3D Transonic Flow in Turbomachinery with the circumferential gap and blockage defined by
Using a Combined Through-flow and Blade-to-Blade Time Marching
Method," Int. J. Heat and Fluid Flow, Vol. 2, No. 4,1980, pp. 189-199. 2-KB
18 Jennions, I. K., and Stow, P . , " A Quasi-Three-Dimensional Tur- M=e,-e„ ~N
(61)
bomachinery Blade Design System. Part II Computerized System," Submitted
to the 29th ASME Gas Turbine Conference, Amsterdam, June, 1984.
19 Horlock, J. H., "On Entropy Production in Adiabatic Flow in Tur- in which N is the number of blades in the row. In forming the
bomachines," Journal of Basic Engineering, Dec. 1971, pp. 587-593. averaged continuity equation, the last two terms in equation
20 Wennerstrom, A. J., "On the Treatment of Body Forces in the Radial (60) can be combined with those from the remaining terms in
Equilibrium Equation of Turbomachinery," Aerospace Research Laboratories
Report No. 75-0052,1975.
equation (3) and eliminated using the condition of no flow
21 Abdallah, S., and Hamed, A., "The Elliptic Solution of the Secondary through the blade surfaces. This can be expressed as
Flow Problem," ASME Paper No. 82-GT-242,1982.
W.n = 0 (62)
when n is the vector normal to the blade surface and is used in
the form

W RW
^ >tx+RW«o% (63)

APPENDIX on both pressure and suction surfaces of the blade.

For a typical term in the momentum equations we have
Passage Averaging - Definitions and Consequences
The passage average of a quantity A is defined by
Ad h pTx(Rpw*w*)de
Ae>e
Add
(53)
A=]± . 30,
)p
dx
Qs~oP
with a density weighted average defined as
-(RpWxWe)s^ (64)
pAdO
pA
A= (54) i a
P (BRWxWe) +^ [(RpWxWe)p^
pd6 B dx
From these definitions we can see that
A=A+A', A?=0 (55)
(RpWxWe)s
dx } (65)

and
i — (BRpWxWB)+ 1 (BRpW^WD
A=A+A", A*=0 (56)
where the prime and double prime notation has been used to
denote a perturbation about the mean value. The per- + — [(RpWx W9)p -£ - (RPWX We)s-£\ (66)
turbations need not be small, and indeed have not been
assumed to be small in the analysis. Using these definitions, Again, the last two terms of the right-hand side can be
we can write the average of a simple product as eliminated when combined with other expanded terms in the 6

306/Vol. 107, APRIL 1985 Transactions of the ASME

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 m o m e n t u m equation by using the no-flow condition. Combining these resulting terms (ps =P+p's, etc.) gives
However, unlike equation (60), gradients of perturbation
terms appear.
The only other average that needs consideration here is that A6 J ep dR dR Ad I " dR " dR i
of the pressure gradient terms, viz., which is the average radial pressure gradient a n d the pressure
1 f» s dp ln R Y d [es , dd„ desl perturbations, the latter being referred to as a blade force
R~de= — \ ~ \ pdd+pD — - p , — (67) term, e.g., for the R m o m e n t u m equation
Ad.
I dp 1 dp
R
T d
( n 0 30„ 38s "1 :i;R^-dd =
RpAd Up dR p dR
+FR (71)

with the radial component of the blade force given by

R r dp dds ddn ddD dds-\ N
\p^-pp^] (72)
Ad litBp r dR yp dR J '

IMPORTANT NOTICE
Mandatory Excess Page Charges for Transactions
Effective July 1, 1981, all Transactions papers that exceed the standard (six
pages) length, will be assessed a mandatory page charge of $125 per page for
those pages exceeding the six page limit.

Journal of Engineering for Gas Turbines and Power APRIL 1985, Vol. 107/307

Downloaded From: http://gasturbinespower.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use