W. R.

Hawthorne
Theory of Blade Design for Large
Cambridge University Engineering
Laboratory, Deflections: Part I—Two-
Cambridge, England

C. Wang
Dimensional Cascade
Tsing Hua University, As a step in the development of an analytical method for designing highly loaded,
People's Republic of China three-dimensional blade profiles for axial compressors and turbines, a simple two-
dimensional method was first investigated. The fluid is assumed to be in-
compressible and inviscid, the blades of negligible thickness, and the mean
C S. Tan tangential velocity is prescribed. The blades are represented by a distributed bound
vorticity whose strength is determined by the prescribed tangential velocity. The
velocity induced by the bound vortices is obtained by a conventional Biot-Savart
J. E. McCune method assuming a first approximation to the blade profile. Using the blade surface
boundary condition, the profile is then obtained by iteration. It is shown that this
6as Turbine and Plasma procedure is successful even for large pitch-chord ratios and large deflections. In
Dynamics Laboratory, order to develop a method for use in three dimensions, the velocity is divided into a
Massachusetts institute of pitchwise mean value and a value varying periodically in the pitchwise direction. By
Technology, using generalized functions to represent the bound vorticity and a Clebsch for-
Cambridge, Mass. 021.39 mulation for the periodic velocity, series expressions are obtained which can be
adapted to three-dimensional problems. Several numerical results were obtained
using both approaches.

1 Introduction
This paper and its companion (Part II) [1] describe the first The expression for the blade profile obtained from the
stage in the development of a method for designing three- boundary condition at the blade surface is then solved
dimensional blade profiles for axial compressors and turbines iteratively.
to produce large deflections and specified whirl or tangential The foregoing procedure which we called the Biot-Savart
velocities. Several authors, [2,3, 4,5] have described methods method is geometrically complicated for use in the three-
of designing three-dimensional blade profiles for small dimensional case, so we have adopted the Clebsch for-
deflections in which the blades introduce only small velocity mulation of the equations of motion as in previous work by
perturbations and a linearized theory may be used. The large the authors and others on the nonaxisymmetric flows entering
deflection nonlinear theory has been developed here in the axial compressor blade rows, as well as in blade wake flows
first instance for the two-dimensional flow through a linear and secondary flows [6-10].
cascade, primarily in order to explore and test methods which When the problem is formulated in terms of the Clebsch
would be suitable for use in the design of annular cascades. transformation, it enables the flow field, periodic in the
The fluid is assumed to be incompressible and inviscid and the pitchwise direction, to be expressed in the form of a series. As
blades are assumed to have zero thickness and incidence so before, the equations and the nonlinear boundary condition
that there are no stagnation points at the leading edges. The on the blade surface can be solved for the blade shape in a
overall deflection and the schedule of pitchwise averaged simple iterative manner. The resulting solution not only gives
tangential velocity Vy is prescribed at all points within the the blade shape but also the pressure distribution—and the
blade row. entire flow field if desired.
The paper presents first the development of a theory in It should be emphasized that it is not the intention in this
which the effect of the blades is represented by a distribution paper to present a two-dimensional design method in com-
of bound vorticity whose strength is determined from the petition with those of Costello [11], Wilkinson [13], or of
prescribed tangential velocity schedule and which is Novak and Haymann-Haber [14], the latter of which is
distributed along assumed blade profiles. The velocity in- similar in spirit to the method based on the Clebsch for-
duced by bound vorticity is determined by familiar methods. mulation.
2 Two-Dimensional Cascade Theory
2.1 The Biot-Savart Approach. A common con-
Contributed by the Gas Turbine Division of T S E AMERICAN SOCIETY OF
MECHANICAL ENGINEERS and presented at the 28th International Gas Turbine ventional analysis of the potential flow of an incompressible
Conference and Exhibit, Phoenix, Arizona, March 27-31, 1983. Manuscript fluid through a two-dimensional cascade assumes that thin
received at ASME Headquarters December 22, 1982. Paper No. 83-GT-124. blades may be represented by bound vortices of strength y(x)

346 /Vol. 106, APRIL 1984 Transactions of the ASME

Downloaded From: https://gasturbinespower.asmedigitalcollection.asme.org on 06/28/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use
Copyright © 1984 by ASME
 K„, + Vv-
V= exVx+ey +v (3)

where e x and e^ are unit vectors in the x- and .^-direction,

respectively, the subscripts x and y denote axial and tangential
directions, respectively, 1 and 2 are far upstream and
downstream values, and a bar superscript denotes a pitchwise
mean value, i.e.,

1
V=
JoV* (4)

The velocity v in (3) induced by the bound vorticity is given

by
dw „ / f1 it
= vx-ivy = - ~ \ y(t) coth— (z-z0)dt
dz 2s Jo s

2ir 2ir
sinh— (x-t) —i sin— \y— f(t)]
Fig. 1 Two-dimensional cascade: notation s s

distributed along the camber lines, Fig. 1. The complex

-irllw 2%
cosh— (x-t) -cos—
2%
\y-f(t)\
dt (5)

s s
potential for the flow induced by such an array of vortices
may be written The boundary conditions upstream and downstream are
satisfied by (3) and (5) provided that
— | y(t)lnsinh — (z- -z0)dt (1)
2ir Jo 1 f1
yl'
V
yi =
= — y(x)dx (6)
Here s Jo
w = <t> + i\p, and, defining a mean boundary vorticity
z=x+iy, dVy
y(x)=y(x)/s = —z (7)
Z0 = t + if(t) dx
where z 0 defines the location of the vorticity of strength 7 ( 0 we may also write
on the blade with the leading edge at the origin. The blade
spacing or pitch is s, the leading edge of each blade is at x = 0, Vy = Vyl + dt (8)
!o*'>
and the trailing edge at x = 1, so that all dimensions are
scaled by the axial width. We use a right-hand rule to define The remaining boundary condition is on the thin blades.
the sign of the vorticity which is here opposite to that used in The condition for a zero velocity normal to the blade surface
conventional two-dimensional airfoil theory. This is done to may be written
avoid confusion in the three-dimensional analysis. Vw«Va = 0 (?)
We define the blade surfaces by
where
a=y-f(x)=0, ±ns (2)
Vw=(V++V-)/2 (10)
where n is an integer.
i.e., the mean of the velocities on the upper (pressure) surface,
In conventional cascade theory the flow velocity, v, induced
V + , and the lower (suction surface, V " , Fig. 1, respectively.
by the vortices given by (1), is superimposed on a mean flow
We note that
which is the vector mean of the upstream and downstream
flow so that |V"-V+ \ = y(x) (11)

e = unit vector on blade with leading edge at

i = imaginary number \A-^1 the origin
Im = imaginary part of a = without subscript refers to x = in the x-direction or dif-
m = integer blade surface with subscript 1 ferentiation wrt x
n = integer or 2 refers to the blade angle y = in the ^-direction or dif-
P = pressure 8 = Dirac delta function ferentiation wrt y
Re = real part of 7 = strength of bound vorticity 1 = far upstream or upstream of
s = blade spacing or pitch along camber lines the blade leading edge
S = periodic saw tooth function X = scalar 2 = within the blade or far
V = velocity vector /* = scalar downstream
V = velocity induced by the bound <f>= real part of complex potential 3 = downstream of the blade
vorticity or velocity potential trailing edge
w = complex potential i = imaginary part of complex
Superscripts
(x,y) = Cartesian coordinate system potential or stream function
as shown in Fig. 1 = vorticity — = pitchwise mean value
a + = pressure surface
z = coordinate of a point in the
complex x—y plane Subscripts — = suction surface
Zo = location of bound vorticity bl = at the blade ' = differentiation wrt argument

Journal of Engineering for Gas Turbines and Power APRIL 1984, Vol. 106/347
Downloaded From: https://gasturbinespower.asmedigitalcollection.asme.org on 06/28/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use
 We may write the blade boundary condition (9) in the form 3.1 Clebsch Formulation of Steady Rotational Flow. In
df _ Vyl + P, the Clebsch formulation of steady rotational flow, the
(12) velocity vector, V is decomposed into a "potential" part as
etc 2 well as a rotational part. Thus V is written in the form of a
Now sum
v= V(j>+Wn (16)
= Re(vxbi-iv}lb,)(/-/') where <f>, X, n are scalars, and the term \ V > is that part of the
velocity vector which accounts for the rotationality of the
= -/«(! +i/')- (13) flow (see Lamb [15], p. 248). By taking the curl of equation
fife (16), we obtain the vorticity as
where/' = df/dx. Q= v X x Vfi (17)
Writing w = F(z),
The Clebsch formulation has been used extensively in
dw I studies of rotational flows, e.g., shear and skew flows, [10]
=F'(*+(/)
fife I *' wake flows, [6, 7, 8, 9], and in nonaxisymmetric flows, i.e.,
Hence (13) becomes inlet distortion studies [7]. The scalars X and ^ are related to
appropriate physical surfaces in the flow, e.g., Bernoulli
vw« V a = —Im—F(x+if) surfaces (constant stagnation pressure, etc.), sheets of vor-
dx ticity (wakes from blades) and drift surfaces. In this paper, we
so that (12) becomes shall relate the scalars X and ^ to the blade circulation and a
function of the blade shape. The Clebsch approach is
rf/ tana,+tana2 1 d developed here for the two-dimensional cascade and then, in a
— = Im j - —F(x+if)
dx 2 Vx dx second paper [1], it is applied in the derivation of the flow in
or, using (5), annular blade rows.

dx
4 The Clebsch Approach for Two-Dimensional
2ir
— \ -^ /«lcosh—(x-t)-cos — Cascades
4x Jo Vy dx \ s s \
[f(x)-f(t)]) dt The bound vorticity on the blades can be written in terms of
a periodic delta function, viz:
tan a, +tan a2
+ (14) 8 = (VK,XVa)s Yi Ha-ms) (18)
m= —oo
If/ = 0 at x = 0, the leading edge, (14), may be integrated
to give Hence, in the blade row the mean vorticity becomes
/= i r v?
8=(VK,XVa)—I s 7,b(a-ms)dy
cosh — (x-t) - c o s — [f(x) -f{t)] y
s Jo "
s (19)
i- [ ^ 4 — - ]* = VK,,X V a
where the integral is taken over the complete blade pitch
4irJo Vx C 2Tf 2ir >
including the blade. In equations (18) and (19), s is the blade
cosh cos—f(t)
spacing.
s s
A Clebsch formulation for the mean velocity is then 0 < x
tan «i +tan a2 < 1
+x (15)
V=V<A(x) + K t ,Va = e x K c +e J ,F y (20)
The procedure for solution is to assume that y(x) or Vy is We note that the curl of (20) gives (19) as it should.
known and substitute initial values o f / a n d df/dx in the r.h.s. We obtain the complete velocity field by adding a velocity
of (14) and (15) and iterate. periodic in y
It is convenient to use the mean streamlines to give the
initial values of/. It is found that up to quite large values of v2=V*2-S(a)VF, (21)
the pitch, s, convergence can be obtained fairly rapidly Now
particularly if a relaxation factor is introduced. Details of the
numerical procedure are given in the Appendix. Q= V X V + V XV2
= (VK,XV«)(1+S'(tt))
3 The Clebsch Approach for Two- and Three- On comparing this with (18), we find
Dimensional Blade Rows
S'(a)=s £j 5{a—ms) -I
Section 2.1 shows the use of the Biot-Savart Law for the m= —a>
determination of the camber line for two-dimensional
cascade. However, its use for three-dimensional annular blade Lighthill [16] shows that a periodic delta function may be
rows is complicated and a simpler method is to be preferred. expressed in a Fourier series, i.e.,
For this reason, the Clebsch transformation was adopted in oo oo / 1 \
the procedure for determining the blade shape required to s \j 5(<x—ms)= \j expum J
produce a specified mean tangential or swirl velocity schedule m=-oo m=— oo s

in an annular blade row. The two-dimensional cascade was

the starting point for the development of this procedure. so that S(a) is the periodic "saw-tooth" function given by

348/Vol. 106, APRIL 1984 Transactions of the ASME

Downloaded From: https://gasturbinespower.asmedigitalcollection.asme.org on 06/28/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use
 (22)
£ e'V (XZ- \„2Xn) = V • (S(a) V K,) (31)
S(
.>-+a.L.---IC-r)
On substituting y(x) = dVy/dx and using (22) for the saw-
We separate the flow regime into three zones (Fig. 1). tooth function series we obtain
(0 upstream of the cascade in which the flow is irrotational
with velocity 4 2 4 ^ r ' V W
i *-n; . 'V!^ ^*n (32)
dx fk,
V=Vxex+Vyiey+\l
The solution by the method of variation of parameters gives
where
Y, = V * , (23)
" J 2X„ dt \ i\„ )
represents a periodic velocity which vanishes far upstream of K x
the cascade. + C\e n + c2e V (33)
(if) A flow within the blade row given by The constants c, and c 2 in the last two terms depend on the
limits in the integral and may be absorbed in the
V=Vx£x+Vy£y+v2 (24) homogeneous part of <j>2 (29b). A convenient choice of limits
where v 2 is given by (21). gives
(Hi) Downstream of the cascade in which the flow is x g X„(x-0 tf
irrotational with velocity
V=Vx£x+Vy2£y+yi
X„ (*) : i; i\„2
—
dt
(y(t)e~^)dt

x e-\ix-t) d

where 2— -j; (y(t)e~'xnf)dt (34)

Jo iX„2 dt
v3 = V 0 3 (25)
On integrating by parts we obtain
represents the periodic velocity which vanishes far down-
stream of the cascade. XAx) =
At the leading and trailing edges the boundary conditions \ /v A "1 »v A
are
•x e\(x-t)
at* = 0
+ [* — yWe-'W^dt
Vl=V2 Ji <X„
and at x = 1 •x e-X„(x-/)
-7(0e-'V(')rff (35)
v2=v3 (26) + Jo zX„
The remaining boundary condition is at the blade surfaces At this point we observe that if the Kutta condition at the
which may be expressed as trailing edge is to be satisfied then
Vw-Va = 0 7(1) = 0 (36)
or (see section (2.3) in Part II [1]) and that y(x) is not likely to
become singular except at the leading edge, where at the most
^+vw«Va (27) it may have a logarithmic singularity since the total cir-
etc culation round the blade must remain finite. For the moment,
Applying the continuity condition to expressions (23), (24), we shall assume that in this design problem we shall specify
(25), and (26) we obtain zero incidence at the leading edge so that
V20i=O (28a) 7(0) = 0 (37)
V2</.2 = V . ( S ( a ) V K ) , ) (286)
and leave nonzero and singular values of 7(0) for later con-
and sideration. Hence (35) becomes
V 2 4> 3 =0 (28c)
Xn(x) = ~^- re-M*-'>e-;>v,AD y(t)dt
Expressions satisfying these equations are of the form X„ Jo

+ — \ 'gM-*-') e->\At) y(t)dt (38)

4>i = Yd ^ « « ' V e V (29a) X„ Jx
n=l

On differentiating
e
4>2 = £ ' V ( B »/ eV + B^e •- K"x+Xn(x)) (29b)
> ]x tl
Xn ' (x) =i\ y(t)e~ * - e-W'>dt

<t>i=Yi CneiXnye~^ (29c)

i2y(x)e-ih»fM (39)

where X„ = n 2ir/s (30) Hence

X'n(Q) = KXn(0) (40a)
and An, Bni, B„2, Cn are constants, and real parts are to be
taken. and
The expressions (29a) and (29c) satisfy the boundary X'n(\)=-\Xn(\) (406)
conditions at infinity that the flow be uniform. The
inhomogeneous part of (296) may be obtained by substituting Matching conditions at the leading edge x = 0 (equation
(296) in (286) giving (26)) yields

Journal of Engineering for Gas Turbines and

Downloaded From: https://gasturbinespower.asmedigitalcollection.asme.org Power Terms of Use: http://www.asme.org/about-asme/terms-of-use
on 06/28/2019 APRIL 1984, Vol. 106/349
 o
CO = a = 45 a = 0°, a = 45°
°1 °°' 2 ° l D

Y = 6(tana - t a n a )x(l-x) - Y = 6(tana - tana ) x ( l - x ) ,- ~ -^

LU
LU
a
cr
ec o" s = 0.0 0.5 ).0 2.0 —- s = 0
UJ
en en
—-s = 0.5
cro ~—s = 1.0
0 r j

0.25 0.50 0.75

-1
d o
do —s = 2.0
0.00
CQ 1 1 1 1
RXIRL DIRECTION .00 0.25 0.50 0.75
Fig. 2 Two-dimensional guide vane: profiles RXIRL DIRECTION
Fig. 3 Two-dimensional guide vane: blade surface angle
4>\y = <j>2y

hence 0 a = 60°, a = -60°

OJ l 2
A„=Bni+B„2+X„(0) (41a) - tana )x ( 1 - -x)
UJ" Y = 6(tana s =
and
-z. • 1.5
_J
vXl=vX2=d-^-y(fl)S(a) \ _ ~ 3-0
0
OCco
or LLJ^-;-
\\r-°- 5
IXJ \ \ \ ^ 0 . 0
\An = \„Bni - \„B„2 + X'n(0) (416) 21
CE
Equations (40a), (41a), (41b) yield
(_>
.00

(42)
At the trailing edge, x = 1 1 1 1 1
^D.OO 0.25 0.50 0.75
4>iy = <j>2y RXIRL DIRECTION
hence
Fig. 4 Two-dimensional impulse turbine cascade
Cne-*n=Bni^n+Xn(\) (43a)
and 00 1
\-l X 1
Lx" = -— = - 1
„=i l-x l-x
so that the series of exponential terms in (45) and (46) may be
-\nCne-^ = \nBnxe^+X'n{\) (436) summed to give
Equations (406), (43a), and (436) yield
£ e x p \-n— (\x-t\-i(y-f(t)))\
B„, =0 (44)
Hence
00

Z 2 "", A 2ir „ 2ir

expl — \x-t\ I - c o s — (y-f) +1 sin— (y-f)
\ s / s s
°° P'
= 2^ I sgn(x-t)y(t)exp(-\„\x-t\ 21 cosh— (x— t) — cos— (y-f))
V s s /
+ iXBCy-/(0))* (45) (47)
Now noting that
where for x > 0, sgn (x) = + 1 and for x < 0, sgn (x) =
-1,and /2ir, A 27r
expl — \x—1\ I = c o s h — (x—t)
\ ss / is
vX2= YiX'n(x)ei^-S(a)y(x) 2-w
+ sgn(x—£)sinh— (x—t)
s
From (22) and (39) the real parts of (45) and (46) become
00
P'
v
x2= E* n 7(0exp{-\Jx-fl+/A„(.y-/(0)}* (46) 2-w
smh— (x—t)
1 f1 s
These results may be used in (27) by writing y = f(x) to
obtain v w . "2 Jo™ dt , 2ir 2ir
-sgn(x-t)
>cosh— (x—t) - c o s — (y—f)
As in section 2.1, equation (27) may be solved by iterative s s
methods. Now (48a)

350/Vol. 106, APRIL 1984 Transactions of the ASME

Downloaded From: https://gasturbinespower.asmedigitalcollection.asme.org on 06/28/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use
 o Curve ( 1 ) : y = 6(tana2 - tana )x(l-x)
oo = a = 45
"l °°' 2 °'
o~
Y= (16/TT) ( t a n a 2 - t a n t t j ^ J t x d - x )
3 1/2
] ' y^
/ Curve ( 2 ) : y= (16/ir) ( t a n a 2 - t a n a ^ [x (1-x) 3 ]. 1 / 2
LU
-Z- a = 60°, a_ = -60
t~—1
s = 0.0 0.5 1.0 2.0 X ^ ^ ^

LU
m
21
d<=>
0f\i
J/JJ^^
^ ^ _ — - ^
u i i | i
0.00 0.25 0.50 0.75 1.
nXIHL DIRECTION
Fig. 5 Two-dimensional Inlet guide vane: profiles
0.00 0.25 0.50 0.75

o °1 " °°' a
2 = 45
°' nXIRL DIRECTION .
LUO
- 3.1/2 Fig. 7 Two-dimensional turbine cascade: effect of loading
_J0 y= (16/lt) (tana - t a n a ) [x (1-x) ] distribution
CD to
Curve ( 1 ) : 6(tana - tana )x(1-x)
3 1/2
.Curve ( 2 ) : Y= (16/ir) (tana - t a n a ) [x (1-x) ]

0.00 0.25 0.50 0.75 1.0

QD
AXIAL DIRECTION
Fig. 6 Two-dimensional guide vane: blade surface angle

2ir
sin— (y-f)
LU 0 . 0 0 0.25- 0.50 0.75 1.0
"x2
y(t)dt[ az
2 Jo , 2x 2TT Q_ flXIRL DIRECTION
. cosh — (x—t) -cos — (y—f) Fig. 8 Two-dimensional inlet guide vane: pressure distribution nor-
malized by far upstream dynamic head

We note that
given by this expression satisfy the Kutta condition at the
ybi'Vot=vy2bl-f'(x)vx2bl trailing edge and give zero incidence at the leading edge.
and that As already noted the objective was not to produce a two-
dimensional design method but to prepare the way for a three-
1 r1 dimensional method by checking first whether the proposed
y ^sgn(x-t)y(t)dt
iterative technique would be satisfactory, exploring the use of
the Clebsch method and providing some basic results for
= i-i(v,n-(vy)1x} comparison with three-dimensional solutions. For this reason,
equations (A10) and (All) were used in writing the programs
V -V
in order to separate mean and periodic velocities. Only a few
= Vy~
results are presented to show the limits reached in the
(50)
numerous calculations.
so that the result given in equation (14) may be recovered Figures 2 and 3 show the profile and blade angles for a two-
from (27), (48a), (48ft), (49), and (50). dimensional cascade with air inlet angle cxy = 0 deg and air
Most calculations were performed using the equations given outlet angle a2 = 45 deg, representing a row of inlet guide
in section 2, but some numerical results were obtained from vanes. The computation involves an integration by a Newton-
the series solutions given above to explore results obtained Cotes formula (Appendix) with the integral evaluated at 101
from the three-dimensional analysis. equally spaced points on the axial chord. However, the results
given in Table I were obtained using 301 points and 18
iterations giving a maximum difference in the value of/(the
5 Results ^-coordinate of the blade profile) between the 17th and 18th
Programs were written for the computation of the thin iteration of 10 ~5 axial widths at any point.
blade profiles in a two-dimensional cascade. As already It is of interest to note the reverse curvature of the blade
noted, the fluid was assumed to be inviscid and in- near the trailing edge and the large negative blade angles
compressible. The variation along the chord of the strength of required to obtain zero incidence at the leading edge.
the bound vortices representing the blades was assumed A more stringent test of the iteration procedure is given by
initially to be parabolic, i.e., proportional to x(l -x). The the design of an impulse blade with 120 deg deflection, Fig. 4.
zero values of the vorticity at the trailing and leading edges At pitch to the axial chord ratios above 1.5, convergence

Journal of Engineering for Gas Turbines and Power APRIL 1984, Vol. 106/351
Downloaded From: https://gasturbinespower.asmedigitalcollection.asme.org on 06/28/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use
 = 45 , 10 = 1.0 Curve (1): Y = 6(tana - tana )x(1-x)
^ 1 D

•y « 6 (tana - tana ) x ( l - x ) -gcurve (2)s y = (16/TT) <tana2 - tana^ [x (1-x) 3 ] 1 / 2

60°, a2 = -60 0.75

^.00 0.25 0.50 0.75

RXIRL DIRECTION CC i
Fig. 9(a) Two-dimensional compressor cascade: profiles
o_ 0.00 0.25 0.50 0.75 1.0
RXIRL DIRECTION
Fig. 10 Two-dimensional turbine cascade: pressure distribution
Q^ = 45 , 10 1.0 normalized by far upstream dynamic head

y = 6 (tana, - tana ) x ( l - x )
Table 1 Two-dimensional inlet guide vane, * = 1.0
x Mean streamline Bilade profile Blade angle
^-coordinate y -coordinate deg
0 0 0 -10.49768
0.05 0.00012 0.01174 -14.23379
0.10 0.00095 0.02438 -13.51160
0.15 0.00312 0.03535 -10.76986
0.20 0.00720 0.04312 -6.52426
0.25 0.01367 0.04658 -1.10216
0.30 0.02295 0.04491 5.14522
0.35 0.03537 0.03749 11.79599
0.40 0.05120 0.02399 18.40724
0.45 0.07062 0.00424 24.58522
0.50 0.09375 0.02170 30.08641
0.55 0.12062 0.05360 34.80161
UJ 0.00 0.25 0.50 0.75 1.0 0.60 0.15120 0.09109 38.73239
az 0.65 0.18537 0.13369 41.95433
o_ RXIRL DIRECTION 0.70 0.22295 0.18087 44.55325
0.75 0.26367 0.23204 46.61130
Fig. 9(b) Two-dimensional compressor cascade: pressure distribution
0.80 0.30720 0.28656 48.18797
normalized by far upstream dynamic head
0.85 0.35312 0.34374 49.30482
0.90 0.40095 0.40270 49.92021
becomes slow. Figures 5, 6, and 7 show the effect of a dif- 0.95 0.45012 0.46230 49.84225
ferent loading, i.e., 1.00 0.50000 0.52039 47.93311

design of three-dimensional annular cascades reported in Part

70c(sin/3—— sin2/3) (51a) II [1].
where The work has been extended to radial cascades, e.g., dif-
fuser vanes for centrifugal compressors, and needs to be
* = (l+cosi3)/2 (516) extended to compressible flow and blades with thickness.
Figures 5 and 6 show curves similar to these of Figs. 2 and 3
for inlet guide vanes, and Fig. 7 compares high deflection Acknowledgments
turbine cascades with different loadings.
This research was partially supported by the Air Force
The expression (51) brings the point of maximum loading, Office of Scientific Research under Contract No. F49620-82-
i.e., bound vortex strength, forward to one-quarter chord K-0002, Dr. J. D. Wilson, Program Manager. The authors are
from the leading edge. Other loadings have been tried suc- also grateful to the staff of the Gas Turbine and Plasma
cessfully, including one with a singularity at the leading edge Dynamics Laboratory, M.I.T., and the Whittle Laboratory,
as in the Glauert series for the bound vortex strength on an University of Cambridge, and E. R. Beggs for their
isolated aerofoil. assistance.
Some pressure distributions are shown in Figs. 8, 9, and 10
for two-dimensional cascades of inlet guide vanes, com- References
pressor cascades, and turbines with the different loadings.
1 Tan, C. S., Hawthorne, W. R., McCune, J. E., and Wang, C , "Theory
Many other calculations were performed and the effect of of Blade Design for Large Deflections: Part II—Annular Cascades," ASME
number of terms used in the Clebsch method was explored. It JOURNAL OF ENGINEERING FOR POWER, Vol. 106, April 1984, pp. 354-365.
was found that good accuracy was obtained with 15 or 20 2 McCune, J. E., " A Three-Dimensional Theory of Axial Compressor
terms. Blade Rows—Subsonic, Supersonic and Transonic," Ph.D. thesis, Cornell
Univ., 1958 (see also USAFOSR TN-58-72, / . Aerospace Sc., 544 and 616,
1958).
6 Conclusions 3 Okuroumu, O., and McCune, J. E., "Lifting Surface Theory of Axial
Compressor Blade Rows: Part I—Subsonic Compressor; Part II—Transonic
The convergence of the iterative procedure for determining Compressor, "AIAA Journal, Vol. 12,1974, pp. 1363 and 1372.
the blade profile using both the Clebsch and the Biot-Savart 4 Namba, M., "Lifting Surface Theory for a Rotating Subsonic or
approach is satisfactory for the range of pitch-chord ratios Transonic Blade Row," ARC R&M, #3740,1974.
5 Homicz, G. F., and Lordi, J. A., "Three-Dimensional Lifting-Surface
and deflections of practical importance. Hence the procedure Theory for an Annular Blade Row," AIAA Journal, Vol. 19, No. 4,1980, pp.
was adopted for use in the development of the theory of the 492-500.

352/Vol. 106, APRIL 1984 Transactions of the ASME

Downloaded From: https://gasturbinespower.asmedigitalcollection.asme.org on 06/28/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use
 6 Tan, C. S., "Three-Dimensional, Vorticity-Induced Flow Effects in Because 7(0) = 7(1) = 0, the last term on the right-hand side
Highly-Loaded Axial Compressors," Ph.D. thesis, Massachusetts Institute of of this equation vanishes at x = 0 and 1. The integrand Fx (x,t)
Technology, June 1978 (also GT&PDL Report Nos. 131 and 151).
7 Hawthorne, W. R., McCune, J. E., Mitchell, N. A., and Tan, C. S., is now no longer singular as Fx(x, t=x) = 0 and the integral
"Nonaxisymmetric Flow Through an Annular Actuator Disk; Inlet Distortion can be computed by any of the standard methods.
Problem," JOURNAL OF ENGINEERING FOR POWER, Vol. 100,1978, p. 604. When t — x the integral in equation (15) has a term
8 Tan, C. S., "Vorticity Modelling of Blade Wakes Behind Isolated An-
nular Blade-Rows: Induced Disturbances in Swirling Flows," ASME JOURNAL
OF ENGINEERING FOR POWER, Vol. 103, No. 2, Apr. 1981. ^/„[cosh(fe) -co.(£/W=7tf>)j
9 Tan, C. S., "Three-Dimensional Incompressible and Compressible
Beltrami Flow Through a Highly-Loaded Isolated Rotor," MIT GT&PDL which approaches
Report 147,1980.

88
10 Hawthorne, W. R., "On The Theory of Shear Flow," MIT GTL Report ^)/w[(|^)2(1+/-W)/2] ( A6)
11 Costello, G. R., Cummings, R. L., and Sinnette, J. T., "Detailed
Computational Procedure for Design of Cascade Blades With Prescribed Equation (15) may then be written
Velocity Distributions in Compressible Flows," NACA Report 1060,1952.
12 Stanitz, J. D., "Design of Two-Dimensional Channels With Prescribed f(x) =x(tan a, + tan a 3 )/2 + \ F2 (x,t)dt
Velocity Distributions Along the Channel Walls," NACA Report 1115,1953.
13 Wilkinson, D. H., " A Numerical Solution of the Analysis and Design

^Kf 2 a + /-w)/2)
Problems for the Flow Past One or More Aerofoils or Cascades," ARC R&M
3545,1968.
14 Novak, R. A., Haymann-Haber, G., " A Mixed-Flow Cascade Passage
+
Design Procedure Based on a Power Series Expansion," ASME Paper No. 82-
GT-121.
15 Lamb, H., Hydrodynamics, 6th ed., Dover, 1945. + \joln{(x-t)2)dt\i (A7)
16 Lighthill, M. J., " A n Introduction to Fourier Analysis and Generalized
Functions," Cambridge University Press, 1969. where Vx is assumed to be unity and

APPENDIX 2ir~ 2x~

cosh — x—t-cos—f{x) —f(t)
The determination of the blade profile for the two- s s
dimensional cascade by the Biot-Savart method requires the F2(x,t) = ——In
4w , 1-K 2ir
integration of the functions in equations (14) and (15), each of cosh — f-cos—f(t)
which has a singularity at / = x. As t — x s s
f(x)-f(t)~(x-t)f'(x) (Al)
For t — x, the terms in the integral in equation (14) may be 2
expanded to give approximately _ ^ 4 ( J ^ ) W > ( * ) ) / 2 } (A8)
— (x-0U+/'2(x)) The final expression for f(x) is obtained by taking the
y(x) s principal value of the last integral in (A7),
2 J_/2x
( y ( * - ' ) ) d+/'2(x)) f(x) =x(tan a! -Kan a 3 ) / 2 + I F2(x,t)dt
2

sy(x) r lie , 1 , (\+f'2(x)

(A2)
s 2ir i s 2 \ 2 J
Equation (14) then may be written
df f1 +x lnx + {l-x)ln(\-x)-l\ (A9)
—- =(tan «!+tan a 3 )/2+ F,(x,t)dt
dx Jo It was convenient to rewrite equations (14) and (15) in the
+ STT_0 f dt/(x_t) (A3) form
2 x Jo
d^_
where x
dx 'y
Fl (x,t) f
1 ' sinhfi+/' (x)sinA
-sga(x-t)}dt (A10)
_ y(t)
sinh — (*-/) + / ' (*)sin — (f(x) - / ( f ) )
s s
"2 !>>[ coshB — cosA

cosh— (x-t) -cos— (f(x)-f(t)) and

s s
Vydx+
sy(x)
1
(A4)
2TT (x-t) — y(t)l—ln(coshB-cosA)-\x-t\) dt (All)
and Vx is assumed to be unity. On taking the principal value
of the second integral we obtain
Where B = (2*/s) (x-t); A = (2r/s) (f(x) -f(t)). The
df singularities were removed from the integrals as before and
—— - ( t a n a, +tan a 3 ) / 2
dx the expression for Vy and its integral obtained analytically
x from the assumed value of the loading y(x). When 7(0) # 0,
VFAx>t)dt^j^lln (A5) a singularity occurs in the integral when t = 0 which may be
Jo 2ir \-x removed by similar techniques.

Journal of Engineering for Gas Turbines and Power APRIL 1984, Vol. 106/353
Downloaded From: https://gasturbinespower.asmedigitalcollection.asme.org on 06/28/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use