Level: CI-CD
Level: CI-CD
Level: CI-CD
Mark W. McBride
Technical Memorandum
File No. TM 79-33
February 13, 1979
Contract No. N00024-79-C-6043
Copy No.
CICD,
LUl
The Pennsylvania State University
APPLIED RESEARCH LABORATORY
Post Office Box 30
College, Pa. 16801D
L~A-...
C-2
~.
_State
E
D--U
Distribution Unlimited
NAVY DEPARTMENT
NAVAL SEA SYSTEMS COMMAND
. .
-,
1,
UNCLASSIFIED
SECURITY CLASSIFICATION OF T4IS PAGE (When Date Entered)
READ INSTRUCTIONS
BEFORE COMPLETING FORW
t,
REPORT NUMBER
C4.
j,
TM 79-33
Technical
e
IN AN
OF &RBOMACHINERY
AND VALYSIS
HEPESIGN
U(
ORESSIBLE,
_TEADY FLOW
USING THE STREAMLINE
URVATURE LTHOD P
1
0XamLand'a
1r
Mark W. McBride
9.
0024-79-C-6043
/'
10.
Office
Box 30a
State College, PA
1I.
16801
San Diego,
CA
TASK
12.
REPORT DATE
February 13,
13
'7]7
1979
NUMBER OF PAGES
161
92152
14. MONITORING AGENCY NAME & ADDRESS(Il dillfrent from Controlling Olfie)
1S.
UNCLASSIFIED
15.. DECLASSVl~CATi3N
SCHEDULE
IS.
~wwN5RAOi4G
I7
Distribution unlimited.
I()ARLPSU/IN..79..33
1I
SUPPLEMENTARY NOTES
IS
KEY WORDS (Continue on reverse side it necessary and identity by block number)
turbomachinery
streamline
"hydrodynamics
curvature
20
identifv by block
i.,bar)
required for the operation of such devices:, In recent years, the digital
computer has made possible the solution of equations of motion pertinent to
this field:'ie)hstudy described heretdeals vith the design and analysis of
turbomachinery in an incompressible, steady flow, such as hydraulic pumps
DD
jD
1473
1AN
' ai 'te
THS''' A<G'(thn
OF
SIIC TO
CL
RT
EC
EU
UNCLASSIFIED
UNCLASSIFIED
SECURITY CLASSIFICATION OF THIS PAGE(Whon Dot& Enter.d)
and turbines.
The basic numerical approach (*tkaedin the analysis of
these devices is the Streamline Curvature Method.
ificatioLo
DIstribut on/
iltbillty Codes
:Availand/or
speclal
UNCLASSIFIED
SECURITY CLASSIFICATION OF TWIS PAGE(Whon Pat*a Entredl
liii
ABSTRACT
In recent
years, the digital computer has made possible the solution of equations
of motion pertinent to this field.
steady
a way that allows the modeling of the blade row spanwise and chordwise
loading distributions on a blade row and determination of their effect on
the through-flow.
1'
the airfoil-shaped
An empirical method of
This method,
iv
distribution.
range of applicability of the Mean Streamline Method on a new correlation developed to aid in the design of airfoil section
Further improvement of turbomachine performance and the development
of a rational three-dimensional solution of a turbomachine through-flow
requires an analytical approach to blade section design.
A Streamline
inlet and exit are specified along with the loading distribution along
the blade chord that is desired.
A blade-to-blade analysis is
performed
In general,
the performance of
however,
It
is
flows require more sophisticated modeling than those used in this study.
TABLE OF CONTENTS
P'age
ABSTRACT .................
.........................
....................
NOMENCLATURE ............................
ACKNOWLEDGEMENTS .............
I.
II.
III.
IV.
V.
VI.
.....
.......
vii
......................
.....................
xii
.......
xv
ii
.....................
..............
...
14
17
4.2
4.3
4.4
.......
..............
......
.........
...
18
..
27
...
31
...
32
...
37
...
38
...
39
5.1
Introduction ............
5.2
5.3
5.4
5.5
52
5.6
55
5.7
Summary ....................
......................
............
...................
..........
...
.......
Pressure Distribution
...
...
43
51
57
...
58
vi
Page
6.2
6.3
6.4
6.5
6.6
Summary ............
....
66
................
....
66
..................
...
70
...
73
....
73
...................
.......................
VII.
VIII.
IX.
t
ip
..........
8.2
8.3
.....
...
85
....
85
...
105
. .
107
109
9.1
9.2
9.4
X.
79
8.1
9.3
...........
ii1
...
124
130
....
150
...
152
...
152
...
156
....
159
SUMMARY, CONCLUSIONS,
RESEARCH ...........
10.1
10.2
REFERENCES ............
............
...................................
vii
LIST OF FIGURES
Figure
...
19
...
21
..........
...............
....
...
29
33
34
36
41
10
11
12
13
14
15
..............
....
...
42
45
....
46
48
Loa(ing Parameter,
Offset,
Page
Title
0....................
o
max
..
..................
50
viii
Figure
16
Title
Parameters DefLning the Nondimensionalized Offset
Distribution ..............
....................
17
18
19
20
......
22
23
24
25
27
28
54
61
..
62
63
64
26
..
. .
21
...
...........
...
67
..
68
..........
69
....
65
. .
71
72
74
..
75
...
76
...............
.....................
29
30
80
81
31
32
33
34
Page
Design Boundary Conditions Input to the TwoDimensional Cascade Design Procedure .. .........
...
86
...
88
...
91
ix
Figure
35
37
...
93
96
99
38
39
103
106
108
40
41
42
43
44
I IiI
Page
36
Sthe
Title
...
100
110
112
...
113
45
115
46
....
116
47
117
48
vs a) .........
....
119
49
....
120
50
.....
122
51
125
52
128
i i
vs a) .......
x
Figure
Page
53
54
131
132
55
56
57
58
59
60
61
62
63
64
65
Title
.
.....
129
133
134
135
136
137
138
139
140
.......
141
142
xi
Eure
66
Title
Comparison of the Theoretical ard Experimental
Velocity Profiles in
Station Coincih'.n,
6/1
68
69
/0
Page
Lhe
...
....
143
144
145
.....
148
149
xii
NOMENCLATURE
Definition
Sytpbocl
A, B, C, D, E
polynomial coefficients
a, b,
polynomial coefficients
c,
d, e
CL
blade lift
C0
coefficient
max
CP
pressure coefficient
streamline curvature -
tAN
pressure
AP
radius
"Rk
projected length
(R
direction
xiii
Definition
Snyol
U
velocity
axial coordinate
radial coordinate
ai
incidence angle
13
B
a
or (y)
direction
fluid density
Subscript
I
S1
tlocal
L
condition
xiv
Definition
Symbol
m
meridional component,
blade thickness
partial differential W. R. T.
component in (y)
partial differential W. R. T.
ps
pressure surface
ss
suction surface
function of n
peripheral component
parallel to
direction
(x)
direction
(y)
I
I
xv
ACKNOWLEDGEMENTS
Robert E.
San Diego,
California.
The
I:L
,Ln.
CHAPTER I
steady flow.
an
fans,
and turbines.
a turbomachine
closely
The performance of
Geometri-
All
or total
one of specifying a geometric shape of duct and blading that will deliver
the desired flow parameters either at a single operating point or over a
continuous range of operating conditions.
considers a given blade row shape and determines the operating points of
the machine.
the turbomachine.
considerations
These
constraints due to
unsteady flow
effects,
and
two-
and three-dimensional
effects of viscosity,
R-4\1sI
FLOW
AXIS OF ROTATION
I
,F I/I
IfII
I l//I
AXIS OF ROTATION
3
manufacturing limitations.
The first is
is
subject to modification.
However,
Since the
to
solve for the inviscid flow field and superimpose the effects of real
fluid flow which are difficult to treat analytically.
----
--
"
-.
VELOC
ITY_
PER IPHERAL
MER IDIONAL
IVe
PLANE
//
MER IMD
IONAL
VELOICITY (Vm
M
INNER
BOUNDARY
AXIS OF ROTATION
Figure 2.
5
The solution must include physical effects peculiar to each type of
machine.
Generally,
If
Determination of
Again,
fixed and it
is desired
This should
give the same solution as the original analysis at the design point.
However.
in particular, measured
velocity and pressure profiles in the vicintty of the blade row may be
used to test the theoretical analysis and design.
A numerical simulation
6
of the turbomachine,
form, must be
affects the other and must be developed iteratively to produce a consistent model of the flow.
Once complete,
fails the
An additional
JJ
CHAPTER II
and,
in some
A basic understanding of
the flows in the complicated passages and blading was obtained through
experiment and,
problems.
in many cases,
Historically,
still
required.
though, at present,
such as hydroturbines
turbulent,
compressible flow
the approximation of various terms, particularly with respect to viscosity ard turbulence.
are caused in part by molecular forces associated with the fluid itself.
As such,
To reduce the
possible to
He
One,
stream surfaces.
flow on planes generally parallel to the hub surface of the machine and
perpendicular to the blading.
Wislicenus [2]
Most of
NASA
rr
f
~BLADE-TO- BLADE
PLANE
Figure 3.
10
Smith [4] rearranged the equations of motion in the meridional
piane to give a time and spatially averaged picture of the flow in a
blade row.
The
equations for the stream function at every grid point were solved by the
finite difference method, an approximation to the actual differential
equations.
but.
in general,
Additionally,
viscous and turbulence effects are much easier to incorporate into the
11
This analysis is
important because the effects of the chordwise and spanwise loading and
thicknegs distributions in blade rows are included, and specific loss
correlations for various turbomachine types are easily included.
The
that a free
An addi-
The design
Katsanis
[13] was one of the tirst to successfully compute the velocity and
pressure di.tribution on the blade-to-blade plane.
very similar to the Matrix 1hrough-Flow Method.
He used a method
12
adapted to the design problem; all were used fo:. analysis of existing
geometries.
Wislicenus [2] developed the Mean Streamline Method of blade
section design.
A relationship
into the nature of the blade boundary layers and the structure of the
wake shed from the trailing edge of a blade.
Lewis and Hill [19] formulated analys ,: which predict the effect of
these forces on the through-flow.
I''
iKatsanis
13
estimating the three-dimensional velocity field in blade rows.
Novak
It should be
stressed that the above techniques are for analysis of already designed
blade rows and do not apply to the actual determination of a blade shape,
i.e., the design problem.
using the Streamline Curvature Method to construct an averaged throughflow picture that satisfies the general design requirements.
Then two
methods, the Mean Streamline Method and the Streamline Curvature Method,
will be used to actually define blading that generates the flow field
prescribed by the through-flow analysis.
:1
CHAPTER III
and type of
it
becomes obvious that some portion of the design will not fit
is
started.
In
this step, the over-all design and dimensions of the blade row are determined.
or
made.
At this point,
performed.
Empirical
15
data place limits on the amount of work a single blade row can perform
without severely affecting the blade's efficiency by causing stall or
flow separation.
necessary to assure that the blade row will not be overloaded and perform
poorly.
The third phase of the design is
numerical model of the flow through the duct and blade row.
Here,
solving the
inviscid equations of motion for the flow and correcting the solution for
the flow losses due to viscosity and turbulence that are expected.
detailed velocity and pressure fields are calculated,
of these and the power requirements,
The
and by examination
the real fluid effects of viscosity and turbulence play a very important
role in the actual performance of the blade row.
16
The second is
Method, while specifying the blade shape, also calculates the flow field
around the blade sections.
With the blade geometry in hand, examination of the flow field may
reveal areas prone to separation or cavitation, indicating that adjustments must be made in the earlier phases of the design process.
The
Again, additional
This
Ii
CHAPTER IV
streamlines on the meridional plane as shown in Figure 3 and to subsequently determine the convective and streamwise normal fluid
accelerations based on their geometry.
compressors,
and turbines,
utilizing
systems.
analysis,
through-flow.
body forces.
The equations
are genera. enough that axial, mixed, and radial flows may be calculated.
I,
____ ____
____
18
4.1
steady,
incompressible flow
are
u Tav + v Tav
ay
ax
l aP
- p
pa y
and
(4.1)
3u
u -+
ax
D4
1 aP
ay
l =
dx
19
170
REFERENCE LINE
(PATH OF INTEGRATION)
t'---
AXIS OF ROTATION
Figure 4.
t1
20
"figures.
The pressure gradient between points a and b,
Figure 5, can be
represented by
b
=p as
as
aP
an
a
b
+---n
(4.2)
(42
an
-P = -p sin +2cos
as
ay
ax
(P
1P
= [u
PT
v + v 'v
ax
sin
ay j
2- + v
ax
cos
(4.4)
Note that
Vm sin 0
V cos
and
Ii
4.
m,
(4.5)
21
Rk
REFERENCE LINE
S.L.
dn
dr
S. L.,
ad
AXIS OF ROTATION
Figure 5.
22
ax
Vm Cos
x+
ax
-m
au~ ax
I
sin
ay
+ Cos 4V
ni
_vv
Vm
+m
Cos
in
x
y+
sin
Vm
(4.6)
iY
and
-V
au
3y
sinp
+ cos
Ym
The quantities determined in Equations (4.5) and (4.6) are substituted into Equation (4.4) and,
as
Vm cos
Vm
+ Vm sin
S~y
(4.7)
3Pm
p as
which is seen to be the differentia'
ms
(4.8)
is developed in a
;P-
3n
3 cos
ax
sin
(4.9)
23
and
oCo
+ v
sin
(4.10)
and (4.10),
(4.6),
+ vau
ing result:
+ V 2 sin
2 Cos
I aP =
3x ,1
+sy
pTn
(4.11)
which is equivalent to
.V
~~~an
IL
asx
(4.12)
sy
(4.12) is recognized to be
The quantity in the brackets in Equation
equivalent to a/Ds, the curvature (k)
or (Rk)-
of the streamline.
pn
.-.
(4.9),
2
k V
(4.13)
and (4.10),
to be
equilibrium equation for nonswirling flow
aP 3n
ap as
ap
n n
a + aI
Tn -s
and,
finally,
o aq
sin c + V k V
mas
m
cosa
(4.14)
24
1 DP
-p Tr
2
(4.15)
'
--
kV2 sna+1
kV
sin
_P
02
mV
sin(+u) + V
a m4as
m
mcosa
oT
(4.16)
j) Vm(n)
ni
(4.17)
25
where ni and n
In words,
is
point,
on the refer-
P V + PV
V8
!P-)
dn -
pVm
2P
.(4.18)
The first two terms on the left-hand side represent the total pressure
(static plus dynamic) available at the point . The term P
contains
all head loss and rotor energy changes between the upstream reference
condition and the station of interest.
terms is the static pressure at F.
static pressure difference between q . and the point , along the reference
line from Equation (4.16).
Equations (4.17) and (4.18) may be combined and integrated between
the two flow boundaries to give
{pV12
P1 -
1 ') d6
r, sin(,,) dn
!q
V
In
sin(i
(4.19)
26
Because P ft is
no
Si
Wdr
W
2
Poo
5P
no 0
30) d1
- yo
+ vc
qin
(4.20)
}W2dn
2 m
where
r
and,
sin (ac)
finally,
Pk + LzV
~2- 2
-vV0
m2
1 11
-2
Wd
W2 dn
Sni
o
(4.21)
ni
line is then
point along a reference
The static pressure at anj
rlqi
drP
(4.22)
27
r
V
+ IP(V
1/2
rll
then obtained
Equation (4.23) until the changes in the velocity profiles and streamline
locations are small between two successive passes.
4.2
the effect
the
changed in a m'anner
In a case
where the spanwise loading or work distribution is not uniform, the total
energy will be changed in a nonuniform way in the spanwise direction as
well.
The
28
(or,
in an inviscid fluid in
AP
21V
(4.24)
where V
The
0) plane,
The lift
force that
Equation (4.24).
which
As
perpen-
29
LEE
SSTRE.AM -
Figure 6.
1'
I4 ,
30
Occasionally,
The radial
equilibrium Equation (4.16) requires an additional term where a significant body force component perpendicular to the streamsurface is
present.
This term is a pressure gradient due to the normal component of the blade
force along the path of integration.
This
Additionally,
the
Once accomplished,
31
4.3
Because of
In
This
testing of a turbomachine.
In the case of a
comparison of the tested data and ideal flow characteristics can be made.
This comparison will reveal the magnitude and distribution of the loss in
total pressure that the flow experiences.
If a correlation is developed
and applied to the analysis, the result is a model of the real flow
field.
it
32
4.4
An actual
The first two examples are indicative of the type of data that must
be generated to begin the detailed design of a blade row.
passes
The velocity
and pressure field is calculated for each of the reference stations shown
in the figure.
leading and trailing edges are required to begin the detailed design of
blading to produce the calculated flow field.
The second example shown in Figure 8 is a simulation of the axisymmetric flow field in a Francis-type turbine.
In this example,
computing
stations internal to the blade row are required to model the blade
loading distribution accurately.
Again,
33
ROW
BLADE
BLADIGEGS
LEADING
EDGES
INTRA-BLADE
STATIONS
BLADE ROWEDGES
TRAILING
COMPUTING
STATIONS
AXIS OF SYMMETRY
Figure 7.
6*
34
VOLUTE CASING
MI
BY
COMPUTED
S. C.M.
--- WICKET GATE EXIT
INLET
IMPELLER
AXIS OF ROTATION
Figure 8.
35
The
'4
36
MEASURED DATA:
0 Ve N .o/
o AV N
/A
/
co
3a
S PREDICTED BY SCM
'-vV
co
-r 2 o
001
z
S1-
00
SVELOCITY
Figure 9.
0~
1 1\_.
0.2
0.4
I
0.6
I
0.8
I
1.0
1.2
CHAPTER V
solidity,
By accounting
lift
loaded blades and blades with higher than usual solidities (C/S).
computer analysis was made of experimental data for many available blade
shapes with differing loading distributions.
for the differences or deviations between the camber line and the mean
flow streamline as a function of the lift
ing distribution parameter,
coefficient,
solidity,
a load-
When a blade is
to be designed
38
5.1
Introduction
The Mean Streamline Method of blade section design due to
ing for blade thickness and boundary layer blockage and the effects of
nonuniform pressure and velocity profiles at the blade row inlet and
exit.
ture of the mean flow streamline from the actual airfoil camber line and
thus, account for the undertu-niag exhibited by blade rows due to viscous.
and potential flow effects.
The trailing
emphasis has been placed on extending the range of this design technique
by correlating a larger range of lift coefficient and data on trailing
edge loaded blades tested by Erwin [24]A computer program was developed to analyze available experimental
cascade data.
and
39
distribution.
the input data and provides the resultant offset distribution and niumerical results.
The resulting data relating the magnitude and shape of the offset
distribution to the cascade flow parameters were analyzed in several
ways.
[2].
offset was more dependent on the lift coefficient and less on the stagger
angle than previously reported.
cases, extrapolated sections designed using the original data would have
been somewhat over-designed, giving more than the prescribed turning.
This chapter will summarize the computer analysis of the cascade
data and the various correlations derived, and, finally, suggest a
method by which these new data may be utilized in a design Process.
5.2
Since
the inlet and exit flow angles are known from the test data, the total
turning is known.
I'
~I
40
are assumed constant and continuity is preserved, assuming an incompressible fluid.
AP x dx
Vy(x)
v1
t
1
51
APx dx
the projected
line may be computed by using the tangents of the "'low angle at each
axial station.
Once the mean streamline is determined and spline curve fitted, the
offset distr.bution is computed by geometrically comparing the mean
streamline to the blade camber line.
The computation of
41
avoi
IN3083d
cr..
0
q^
tnto
0
0w
U-0
coo
u0
Coo
L)
CD
-J
C)C)
LM.
I"00
0o
C
xela
'-N
42
avoi0
INfl3i3d
LPI
tA
(AJ
co00c
o
D
'
iz
#44
0Q
.0
LonJ
LALJCL
onI
CA:
*q
0
Li%
-CN
C-4I-
00C
xew
NH
43
velocities existing at a given percent of the chord and aesuming that the
total pressure relative to the blade is a constant.
5.3
Two sources
The cata
originally used by Wislicenus [2] were used to verify that the computerized procedure yielded the same results that he originally obtained.
These data agree when reduced by the computerized procedure.
series data documented in Reference
over an extended range of lift
blades.
NACA 65
Reference [25],
[26]
were used to verify trends and upper limits of camber line offset.
The resulting data relating the magnitude and shape of the camber
line offset distributions to the experimental cascade geometries were
analyzed in several ways.
The quantity
ma
This is
44
This plot,
Figure 13 results.
coef-
coefficients.
However,
there is
still
significant
dispersion of the data and only slight correlation with the stagger
angle.
None of the correlations tried yielded definitive results, indicating a much more complex interaction was occurring between four primary
cascade variables.
the solidity (a),
coefficient
(CL),
The loading parameter was defined as the percent of blade chord where
one-half of the total turning was accomplished.
A high loading
A curve fit
AN
C
a + bb
c
d
te
c * a .
8f
v
(5.2)
(
t!I
...
45
I
0.08IiIIII|
DATA SOURCE
0 REF 2
cc
0.06
v NACA O=1.0
"0.04
a NACA o=1.5
0REF23
23
X REF 24
o
0
0. 02
II
10
Figuce 12.
[4
20
30
40
II
50
pv (deg)
AN/C
60
CL,
vs Vane
70
46
0.10
II
o REF 2
v NACA ov = 30
0.08
SNACA
v = 4560
:-
0.6- 0 REF
NACA23 v = 60
x,/.
x REF 24
X0
o3
Ax
Xoo-.F
~0.04
00
VVV
0.02-
0.2
0.4
0.6
0.8
1.O
1.2
1.4
LIFT COEFFICIENT, CL
Figure 13.
o,
a
vs Vane
1.6
47
1 n
1
ERROR=
where n is
2}1/2
1
{-
cde
(5.3)
SLi
chosen, then each in turn was varied to find the minimum error with
respect to that coefficient.
occurred.
-N
max
AN1.508
-0.4779
0.0018 + 0.0345 CL
0.0086
24
0.1250
.
(5.4)
The magnitude of 2 in this equation varies from zero to one, while the
vane angle is measured in radians.
drawn from this equation is that the loading parameter and the vane angle
appear to be insignificant and may be readily eliminated from this
relation.
is
in general agreement
there is
nents of the lift coefficient and solidity are in general agreement with
observations made on the other correlations.
equation are plotted in Figure 14.
the fit and experimental values of the nondimensional offset are plotted
against the lift coefficient.
48
0.08-
0.06
X
" 0.04-
ZZ
0.021+
0
0.
0.5
+ EXPERIMENT
x CURVE FIT
1.
1.5
2.o
LIFT COEFFICIENT, CL
Figure 14.
2.5
49
loading distribution.
less than 0.005, less than 10 percent of the usual offset values used in
a design.
coefficients, which
AN
(5,5)
= 0.002 + 0.035 C
-a
closely approximates the derived curve fit and would lead to slightly
more than the desired turning angle in a dceign process.
The actual shale of the offset distribution varies considerably and
is highly dependent on the incidence angle of the cascade.
The NACA
however,
TVe data of
Here,
the offset distribution on the trailing edge side of the maximum point
was fairly consistent, while the shape and magnitude of the leading edge
varied considerably,
In a design exercise,
r21.
Figure
Although there
50
1.I
0.8-
o REF2
v NACA
N
c- 0.6-
0 REF 23
=v
<
00.2-
Ii
0.2
I
0.4
I 1
0.6
I
0.8
I
1.0
max
Figure 15.
max
XfX
i
-
4~
-
51
fit and indicates that the maximum offset is generally toward the
trailing edge and shifts toward the trailing edge as the loading shifts
rearward, again as would be expected.
There is no apparent correlation for the leading edge offset
distribution, leaving the designer relatively free to choose this shape,
either for cavitation resistance or zero incidence.
5.4
In more
precise terms,
X-t
AP(x)dx =
P(x)dx
(5.6)
50 peecent.
or if
This
0.5 + 0.525 2
C0
(5.7)
Max
where C
Then,
52
max
0.002 + 0.035 CL
(5.8)
the process.
An
smooth
5.5
A polynomial may be
Y/C(C
) =
max
max
0.0
Y/C(100% C)
Q
0.0
Y'/C(C 0
max
-I
sufficient
53
and
YLE
Y/C(O.0%)
offset distribution.
being determined.
Y/C
must be used,
(5.9)
The quan-
YLE
TAN(al)
AN
Ca
(TAN(i)-E 2
o max
3) + YLE(
Co
)/(lC
Co
0
max
max
max
AN
1 ) + YLE(
B (TAN(ai(
C
0 max
max
+ C
-(I
max
and
"-
1Z
.l9*
/(-o
1)
C
0
max
max
54
a,
max
YLE
Figure 16.
=________
ANICmax
__
55
= -
(+TAN(a i)+YLE+A+B)
5.6
the static pressure difference across the blade and the average throughflow velocity.
Additionally,
if any.
Pt
a Ps
{(Vm +AVmx)2
+ (V20 -Ux)
(5.10)
It
of relative velocity across the channel and at the same time assure that
the volume flow rate is conseived.
The static pressure difference between the suction surface (ss) and
the pressure surface (ps)
is known
(AP),
= 1AP Const
Ss
cos(y) dy
(5.11)
56
where (y)
ps
SJW(y)
cos(y) dy
(5.12)
ss
and,
PS
2 .2 ts-AP(y)) 1/2cos(y) dy
(5.13)
ss
where Pss is the suction surface static pressure and AP(y) is
the
pressure difference between the suction surface and some point in the
channel.
static pressure,
Pss
JPt
W(y)
- AP(y)} dy
(5.14)
Once the suction surface static pressure is known, the velocity and
static pressure gradients may be reevaluated in an iterative procedure
which will converge in about five steps.
velocity and pressure distributions,
pressure are used.
is checked for volume flow and corrected to match the proper valve.
This method produces a pressure distribution quite similar to the
estima~ton method used in Reference [2] but is slightly less
57
[2]
5.7
Summary
A significant range of cascade data was analyzed to provide refined
The maximum deviation or offqet of the camber line from the mean
on this deviation.
The shape
In all cases,
small compared to the chord length, and the error in specifying this
approach, and it
is
CHAPTER VI
This
Head rise
36 ft of water
Shaft speed
6000 rpm
Dt
t
0.312 ft
Dh
0.12 ft
Casing diameter
= 0.315 ft
Inlet head
H
sv
1000 gpm
150 ft
1:
59
The suction specified speed is
(6.1)
Hsv 3/4
In Reference [29],
it
defined as
Z=
- 0.121
(6.2)
In Reference [2],
it
is shown
selected.
From Reference [30],
An estimate of horsepower
550
. 11.36 HP
and
6.1
in an earlier chapter was used to obtain the flow field solution used in
iV
60
the design of the axial flow pump.
rows.
was estimated based on the Blaslus seventh-root law for turbulent flow in
smooth pipes and modified as recommended in Reference [31] to correct for
the higher Reynolds number effects incurred for this particular applica-
tion.
but there is no
based on the seventh-root law for boundary layers and also modified for
Reynolds number effect, are shown in Figure 18.
Also plotted is a distribution for a free vortex radial loading distribution, and it
shown
exit, as well 3s those at stator inlet and exit, are shown in Figure 21.
The integrated value of head based on a mass average equals 36 feet,
if
are used.
61
OUTER CASING
FLOW
=-
ROTOR
TRANSITION
1.75'
1.89'
SECTION
STATO
STATOR
S0.
72'
AXIS OF SYMMETRY
Figure 17.
62
,..,
I-CASING WALL
1.6-""-
'-
20
0.4-
LCYPF
AT STATION
0.2
SEVENTH-ROOT
LAW
QCORRECTED
FOR R
REYNOLDS NUMBE
0.4
0.6
0.8
1.0
1.2
MERIDIONAL VELOCITY
MEAN AXIAL VELOCITY
Figure 18.
J_
63
2.0
"!
1.6-
FREE
1.41.21.<
FORCED VORTEX
"o
,
'
VORTEX
LOADING
DISTRIBUTION
--
CASING WALL
1.0 -
0.8 -
(USED IN DESIGN)
LOADING
DISTRIBUTION
USED FOR DESIGN
HUB
0.2
0.4
0.6
0.8
1.0
1.2
64
OUTER CASING
S~~STATOR
AXIS OF SYMMETRY
Figure 20.
Design Example.
65
2.0
CASING WALL
1.8
ROTOR INLET
1
1z.6
"'
ROTOR EXIT
1.4
C-
1.2-
i,-,-
1.0-_
STATOR INLET
STATOR EXIT
7
.,_
0.8 0.60
0.2
0.4
0.6
0.8
1.0
1.2
MERIDIONAL VELOCITY
MEAN AXIAL VELOCITY
Figure 21.
66
6.2
Method was used as input to the Mean Streamline Method of blade section
design.
section, the velocity associated with the streamline 2 percent inside the
casing ia used to represent the tip flow condition.
The developed cylindricrl sections for the rotor and stator are
plotted in Figures 22 and 23,
respectively.
6.3
methods.
The fillet
A photograph of the
Tests
The axial flow pump was installed in the ARL/PSU six-inch wa.er
tunnel.
67
TIP
Sj
Figure 22.
ROTATION
68
LEO
n
TIP
gnO
Figure 23.
69
va
.?>--J
S4J
'a
Sr
0
-,4
I.~
'.-4
70
and stator (3.78 inches) to the six-inch diameter water tunnel test
section.
vided the necessary pressure drop to test the pump from 50 percent to 150
percent of its design capacity.
Cavitation performance was noted using the tunnel pressure system, but the
validity of the results is limited due to the &.ov response of the tunnel
pressure system and the uncontrolled high air content in the water.
6.4
Test Procedure
The axial flow pump was installed in the six-inch water tunnel as
A three-inch orifice
The data
taken over several tests indicated that over-all the data were repeatable
to within 3 percent.
The upstream and downstream probes were set at 0.7 of the blade span
and a setting of the water tunnel pump was determined that yielded the
correct total pressure rise for the corresponding RPM of the pump.
At
71
0 4w
CL
134
%-4
4-4
72
0H
431
00
Q;
0j
0. -44W
~0
0l
0
73
these settings, the upstream velocity profile and the downstream total
pressure profile were obtained as a function of radius.
Both profiles were integrated to give the mass-averaged velocity at
the pump inlet and the mass-averaged total pressure at the stage exit.
The upstream velocity probe and the downstream total pressure probe were
set at a radius that corresponded to the mass-averaged values.
It was
assumed that the shape of the velocity and total pressure curves did not
change appreciably with flow coefficient and the main tunnel pump was
used to systematically vary the flow coefficient.
average total pressure rise, and shaft torque were recorded for a range
of flow coefficients.
6.5
Test Results
The upstream velocity profile and probe location which corresponds
to the mass-averaged velocity for the rotor are presented in Figure 27.
The total pressure survey data are shown in Figure 28.
located as shown in Figurew 27 and 28,
Some care
must be used in interpreting the efficiency curves since very large (40
percent) bearing torque values were measured,
6.6
Summary
The axial flow pump was tested in the ARL/PSU six-inch water tunnel
Tests indicated
that the design placed 13 percent too much total pressure rise into the
fluid and, hence,
74
TIP______
MEASURED PROFILE
VELOCITY PROBE LOCATION
THEORETICAL PROFILE
RADIUS
(inches)
_\,"
HUB
0
0
0.5
1.0
75
..
TIP
0-m
THEORETICAL
,/
(KEIL PROBE
(inches)
LOCATION)
_.
SEXPERIMENTAL
HUB
1NOND IMNIA
NONDIMENSIONAL
,/
(P
-I
2p
76
0.3--
0.9
0.
ii
0.2-
DESIGN
POINT
0.7
"-0.6 >:
Lni
0.5
,
0.1 -
S\
-0.4
0
0.3
I
0.4
0.5
77
Analysis of the results of the axial flow pump design and test
reveals several possible explanations for the differences between the
design specifications and the test results.
profile in the test facility was different than that anticipated for the
design application.
At the tip of
the rotor, the blade incidence was considerably higher than anticipated
and could have contributed to the increased torque and pressure rise
noted.
Additionally,
(n = 0.80) was conservatively low by at least 5 pt'rcent for this particular application.
However.
total pressure rise is correct but that a uniform increment has been
added,
necessary to bring the pump t.,its design point and also indicated that
too much loading was specified in the preliminary design.
The design point efficiency as derived from powering data in Figure
29 is 0.78.
Howevec,
78
It
is not apparent that these test results support the use of the
While agree-
ment of the theoretical and experimental flow fields was not as good as
anticipated in this case, other design examples of a more complex nature
have shown excellent agreement as can be seen in Figure 9.
II
CHAPTER VII
This method,
Figure 30
New cnrrelations
The
coefficient
and cascade solidity had any important effect on the airfoil performance
or on the geometric shape of the theoretical camber line.
The direct theoretical determination of the required blade camber
line is
Additionally,
80
-,tm
<
<
CURVATURE DUE
~TC
POTENTIAL
FigurE 30.
DUE
FLOW 'UNDERTURNED'
TO COMBINED
,REAL
FLOW EFFECTS
TURBULENT WAKE
SHED INTO DOWNSTREAM
FLOW
81
CORRECTED
BLADE
CAMBER LINE
I
Figure 31.
82
explored the problem of designing arbitrary blade sections for compressors and turbines, Reference
[3],
Additionally,
However,
An additional
This
83
fairly simple
being able to do less turning of the fluid, hence, less work than
theoretically possible.
Since the problem at hand is
however,
In a turbulent
This
loss is distributed in some manner both through and across the blade--toblade channel.
a wake of low
84
The
problem
flow will affect the meridional plane flow, i.e., the effects of thickness and boundary layers on the blade cause blockage accelerations in
the meridional plane solution.
The following chapter presents a detailed solution of the equations
of motion to determine a camber line given the blade loading and thickness distribution and specified inlet and exit fluid angles.
An
ii
CHAPTER VIII
8.1
assumed steady,
The
blade shape which results from the analysis has prescribed inlet and
exit angles, loading and thickness distributions,
tional model for the blade surface boundary layer effects and downstream
wake structure.
the
Together,
86
STREAMSHEET
THICKNESS
RATIO
f2 -
BLADE
LOAD ING
DISTRIBUTION
VANE
THICKNESS
DISTRIBUTION
Figure 32.
AP(x)
!
v
87
Because the
I,
Therefore,
This
In essence,
flow streamline.
Thus,
in Figure 33.
88
INITIAL
ZERO-THICKNESS
BLADE APPROXIMATION
tJ
Figure 33.
F9
This
In essence,
equations on any pass represents a valid solution and, hence, only sm&all
changes in the solution are required on any particular pass.
The
AP(x) dx
where AVy is
inlet to exit.
This equation is
V (x)
where f[sst(x)J
- Vx1 f[sst(x)]
Normally,
90
The following
y(x) = Y(X)
Vx) dx
1Vx(x)
(8.2)
xi
tangents.
Taking a number of these streamlines and repeating them in the
defined spacing between adjacent blades in the cascade, as seen in
Figure 33, yields a geometric solution that can be utilized to calculate
the geometric derivatives necessary to start the iterative solution of
the momentum equation.
The equations
are solved along lines parallel to the cascade axis and that are
generally perpendicular to the direction of flow.
the angle of
91
xf
-- V
S
CALCULATION STATIONS
(PATHS OF INTEGRATION)
Figure 34.
ROM
92
degrees.
I 2E ,yVa--+
3V
Vy Oy
(8.3)
The form
V 2
- . k cos + Vm---sin
Pay
Rk
mas
.(8.4)
is obtained from a
finite difference approximation from the velocity field and the geometry
of the streamlinea.
The effect of this equation is to calculate the pressure gradient
throughout the flow field based on the local acceleration due to streamVmVm
line curvature ( -)
3- -).
The simultaneous application of the continuity equation and conservation of total pressure completes the essentidl portions of the
calculation.
this analysis:
93
{S
SSTREAMLINE
I.
Vy
Figure 35.
PATHS OF INTEGRATION
FROM ss TO ps
vxR
V.
ii
94
Const
P((XY)
D1 (x,a) dO
V (Xy)
dy .(8.5)
The term P(x) is the static pressure on the surface of one blade or
on the stagnation streamline and must be solved.
pressure determines the velocity field that satisfies both continuity and
the conservation of total pressure.
Once the static pressure field has been determined,
equation may be used to recalculate the velocity field.
Bernoulli's
The following
V2
my
P(x,y)
dy - P
(8.6)
,p(x)
s1
whee P,.x,y) is
at prescribed percentages
In practice,
it
is
Generally,
only a small
At this point,
the
Il
l l
95
The effect of these calculations in the blade space will show an
increase in velocity near the blade suction surface and a corresponding
decrease in velocity on the blade pressure side.
In general,
the integrated pressure force parallel to the cascade axis derived from
the blade pressure diagram will be less than that required to produce the
fluid deflection specified by the inlet and exit flow angles.
Additional
This indicates
This is
made to the shape of the stagnation streamline and the blade camber line
to cause an increment of momentum equal to the difference calculated
above to be added.
for each streamline.
96
r- x --
TE
FLOW-PATH MODIFICATION REQUIRED
TO SATISFY MOMENTUM EQUmTION
AND ONE-DIMENSIONAL FLOW
RESTRAINT SIMULTANEOUSLY
Figure 36.
97
VW(x) + AV(x)W
Ao(x)-
(x)
tan
(8.7)
This
However,
at convergence,
The
momentum,
t
I
98
whatever criteria are required, i.e.,
could use the method of Abbot and Von Dohenhoff [33] to add the thickness
distribution to the already defined thin airfoil camber line and subsequently correct the surface pressure distribution in the same manner.
However,
may be present that will cause additional perturbations to the flow field
and pressure distribution.
As each iteration is
In this way,
the pass-by-
the
stagnation streamline is separated into two separate streamlines corresponding to the pressure and suction surfaces of the blade.
An area of particular difficulty is encountered with the thickness
distribution at this point.
99
STAGNAT ION
Figure 37.
100
STAGNATION
POINT
Figure 38.
101
stagnation streamline is forced into a singular point.
In almost all
cases, polynomial functions that model the streamlines will fail or "blow
up" at this point.
into three segments:
In the
integration of the momentum equaticn, the radius of curvature was neglected at the leading and trailing edge points and the velocity set equal
to zero at the singular points.
necessary,
A boundary
This
102
method,
tionally.
included
in the momentum equation calculations and results in a general acceleration of the flow around the cascade trailing edge.
is approximated by allowing the displacement thickness to proceed downstream and gradually vanish at one blade chord downstream.
The equation for the conservation of total pressure must also be
modified to include the effect of viscosity.
approximately
At each
station in the blade row, the boundary layer calculation gives the
displacement thickness, and the integrated reduction of total pressure
can be calculated.
by "smearing" the loss across the duct, weighting the loss as the square
of the distance from each blade surface,
velocity.
In this way,
At present,
However,
such as proposed by
103
SUCTION SIDE
B.L. DISPLACEMENT
THICKNESS
STAGNATION
""CAPP ROXIMATE
BLADE WAKE
PRESSURE SIDE
B.L. DISPLACEMENT
THICKNESS
Figure 39.
104
two-dimensional Streamline Curvature solutions to produce a quasi-threedimensional analysis of compressor or turbine blade rows.
No previous attempt at coupling solutions for the indirect or design
problem is known.
However,
bility of a coupled,
However,
it
is
still
approximate in that a
The
In this way,
This is accomplished in
The
in
A boundary
105
layer and wake model are included to accommodate the viscous and turbulent flow effects found in turbomachine blace rows.
8.2
i.e.,
The
45 degrees,
the inlet
19.5 degrees,
Therefore,
the
Ct = 2 A
where AVy
Y2
- V
yl
(8.8)
- = 0.707
therefore,
makes it
An elliptical loading
106
Y
S
Vy2]
y2
V"
Figure 40.
107
ideal flow blade loadings employed in thin airfoil theory.
The thickness
distribution used is similar to that reported in Erwin [24] and has its
maximum thickness (10 percent of chord) at the 60 percent chord position.
8.3
of the streamline pattern through the cascade, the camber line and blade
surface coordinates, and a comparison with the initial approximation to
the streamllqes.
Figure 41.
required to
part,
This is due, in
108
BOUNDARY-LAYER
-- DISrPLACEMENT
S~
CALCULATED
S~
BLADE PROFILE
NOTES:
/
Figure 41.
CHAPTER IX
determination of
fabrication of the cascade of blades, the experimental
determination
the blade surface pressure distribution, the experimental
of these data with the
of the flow field, and the direct comparison
section design.
theoretical flow field determined during the blade
The
in Figure 42.
blade design procedure is shown graphically
This figure
in an
indicates the geometric shape and relative position of two blades
infinite cascade.
Also shown is
Jl
110
BOUNDARY LAYER
DISPLACEMENT THICKNESS
(P(SUCTION
J5
SURFACE)
9 11112
Fiur
4.
hortialSremlneDitibtin8n
Blde SapeDetemine bythe trealin
Cuvtr6
ld Dsg
ehd
1ll
9.1
the blade
span was 14.50 inches, and the blade-to-blade spacing was 4.65 inches.
The arrangement of these blades in the test facility is shown in the
photograph of Figure 43.
The
pressure taps were located so as to correspond to the computation stations indicated in Figure 42.
manner that assured that wakes shed from each hule would not affect
readings taken at downstream locations.
the approach
In practice,
these
thereby assuring
112
to 0
00
0
4-4 W.
0 C34
to0
0w
113
1.
AT CALCULATION
STATIONS
LE
Figure 44.
INSTRUMENTED CENTER
ISECTbON
I.I I*In
Ii
U~
It
IPIn
IF
114
pressure along the approach duct to the cascz.de,
A comparison of the flow field in the test cascade with the theorctJ-.al flow field requires obaining the velocities (Vy) and (V )
parallel and perpendicular to ':he cascade axis, respectively, and the
local static pressure (P a) at various locatmons in the flow field.
thre-hole
ment.
Is shown in
g?,ure 46.
An expanded
The basis for
probe tip located at the cente- of a 1.0 foot. diameter, low turbulence,
open jet.
This arrangement is
The jet
increments from a flow angle rlative to the probe of -20 degrees to +20
degrees.
recorded.
Pt, P)
115
TEST
THREE-HOLE
SETTLING SECTION
-S
PRESSURE IN
AVERAGE APPROACH
TUNNEL
SECTION
45.
Figure
PTest
E REEECELN
NOES P. ESRDNTNE
116
MEASURED
PRESSURES
_SOFT
V
Pt
-----
PS
.PT
O
0
__
SOLDER
FILLET
P0
4TUBE
Figure 46.
..
SIt
117
OPEN-JET TUNNEL
/SET
CLAMP TO
ANGLE
ADJUSTABLE-HEIGHT COLUMN
TO KEEP PROBE TIP ON
TUNNEL CENTERLINE
Figure 47.
l- I
118
To deter-
mine the flow angle, the difference in pressure between the two side
pressure tubes P+ and P
(CP)
is used.
is defined as
C
Pa
Pt -
(P+-P-)
/2(P++P)
(9.1)
The result of
CPt =-P
PT - Pt
,-.
/2(P++P)
(9.2)
119
0~
S-3
0.
I,-I
-20
'
Figure 48.
-1
20
FLOW ANGLE, a
Three-Hole Probe Calibration
120
'=-
0.3
00.2-
'
0.11
-0.1
-0. i-20
-10
10
20
FLOW ANGLE, a
Figure 49.
!t
vs a).
121
During
defined as
(P++P-,'2 - PS
CP
s
(9.3)
S
Pt -
./2(P +P-)
"I
to atmospheric pressure.
Calculation of the velocity components and static pressure at some
Spoint
determined.
).
Once known, an
interpolation on the curve fit of Figure 48 yields the local flow angle
(a).
Determination of the local total pressure is achieved by determining
the value of C
mined.
Figure 49.
rearranged to give
Equation (9.2)
is
122
C%
a-
a-26
1.26
Q,'r
FO
" 1.231.-2
-10
10
20
FLOW ANGLE, a
Figure 50.
t8
123
tz
"
itt
- 1/2(P++P-)
+ P
+
(9.4)
determined.
Figure 50.
Equation (9.3) is
rearranged to give
Ps
where P
st
(P++P-)/2 s
Ft
- l/2(P++P-)
(9.5)
The flow "ay have experienced a loss in total pressure between the
upstream reference conditions and the point at which measurements are
being made.
difference between the indicated local total pressure and the total
pressure measured at tne upstream reference conditions.
The loss of
total pressure is
PL = PT - P
(9.6)
This equation is
i1/2
V
(PT-PL-Ps)
VmfTL
"t---
(9.7)
S. J
~-
124
A reference velocity (V1 ) is defined as
VI
(PT-Ps)
(9.8)
/2
is easily deter-
is defined as
s i
P
Cp
s PS
(9.9)
ence velocity of 100 feet per second and a corresponding blade chord
Reynolds number of 240,000.
1f
125
0.4
0
0.2
-O
">% -0.0
X/
-0.
N1
-')
0.o.
-0.4
0'/
.
0.6
0.8
1.0
, 0.0
0.2
0.4
x/ii
Figure 51.
126
where VI is
(V2 /VI),
of
the
changes requires that the average blade force per blade, in the direction
parallel to the cascade axis, be equal to the momentum change of the
fluid in the blade passage in the same direction.
The equation
(9.10)
AP C cos(O) = pSVxxAVy
(AP)
is
the pressure acts on the projection of the blade chord (C) perpendicular
to the cascade axis (C.cos(8)).
the (x)
(8)
(pSV x),
is the density,
the change in
velocity parallel to the cascade axis, between the leading and trailing
edges.
If
(AP),
is nondimension-
Cp - AP/(l/2pV 2
1
(9.11)
.CCos a VIV 1
of 0.6067.
is formed.
The test
Integration of the
127
0.440.
value of Cp
In this case,
in Figure 53.
plotted
found to be 0.504,
3.7
degrees less than design and 2.7 degrees less than predicted by the
Douglass-Neumann method.
Inspection of the pressure distributions shown in Figures 51,
52,
and 53 reveals that good agreement is obtained among the various comparisons for the pressure side of the blade.
fair
Examination of the
128
0.6-
0.4
0.2
CL
0~
-0.4
-0.6 -0
"-0.2
0.6
0.4
0.8
x/#
Figure 52.
1.0
129
0.4
C.
"-0.2-
-0.I4-
-0.6'
0
0.2
0.4
0.6
0.8
x/I
Figure 53.
-.
t0
-- ow
1. 0
130
boundary layer calculations performed during the design indicated the
possibility of flow separation at 70 percent of the chord on the suction
surface, making the boundary layer and wake model inaccurate.
Addition-
9.3
indication of the extent and structure of the wake of low velocity fluid
shed from the blade trailing edges.
agreeing fairly
131
BLADE WAKE
I\
BLADE GEOMETRY
BLADE SUCTION
SURFACE
BLADE PRESSURE
SURFACE
VELOCITY VECTORS
v.
'00
0
/0
FFLOW
Figure 54.
V.
MEASURED BY THE
THREE-HOLE PROBE
SEPARA'(ED
REGION
132
5
THEORETICAL
EXPERIMENTAL
3-
-j
U
zC-,
0z
W- 2-
0
0
LUI
0
z
I-o
0-
-10.8
0.9
1.0
1.1
1.2
VmlV
Figure 55.
1i
133
SUCTION
SURFACE
4
3-
0F
U,
z
0
L.J
C-0)
.LL
00
M /V
LIv
PRESSURE
SURFACE
76
mVm
08
Figure 56.
ii
,a
0.19
1.U
1.1
.2
134
5I-
---
SUCTION
SURFACE
-4
U
CD
i-
2.
o0
00
0
PRESSURE
00
a]
0.8
Figure 57.
SURFACE
0.9
1.0
1.1
1.2
t-
--
135
SUCTION
SURFACE
4
U
C
C-j
L" 2.
0A~
z
0
00
I-
PRESSURE
SURFACE
-1
.8
0.9
1.0
1.1
1.2
IV
VmlI
Figure 58.
136
5.
SUCTION
SURFACE
S4
eU
_z 3Lai
0o
U- 2
o
S0
00
0j
z
00
0.8
Figure 59.
PRESSURE
SURFACE
0.9
i.0
1.1
1.2
137
SUCTION
SURFACE
4w0
r-J
U
_
zu.J
0.8
Figure 60.
I,
77.~...
~J
PRESSURE
SURFACE
0o
0.9
1.0
VmlV
VM
IVI
1.1
1.2
138
5-
4-
SUCTION
SURFACE
0
U4
-l
A.
,,..1
Z
00
-0
0-
-10.8
0.9
PRESSURE
SURFACE
1.0
1.1
1.2
V~mlV
VM IVI
Figure 61.
a,
-t-
-.---.--
-~
--
,-
139
5-
SUCTION
SURFACE
4-
U
0
3
-j
Z
EE
00
I-
S0
0.8
Figure 62.
J;
i!
PRESSURE
SURFACE
0 .9
1 .0iol
1.2
140
SUCTION
SURFACE
_
CU
0
0
3-j
L&J
U.J
0
I-)
01
PRESSURE
SURFACE
-11
0.8
0.9
1.0
1.1
1.2
VmlV
VIV
Figure 63.
141
SUCTION
0
SURFACE
3-
_00
L.=J
0
0
PRESSURE
00
SURFACE
-1
B7
0:8
u'.9
.1.1
VmlV
VM/VI
Figure 64.
It
Li
_____
142
POSSIBLE
SEPARATED FLOW
SUCTION0
4 - SURFACE
0
0
z 3
LU
0
00
Uj
00
I-1
PRESSURE
SURFACE
0.1
Figure 65.
I,
I'
0.8
om9
.1.0
1.1
143
5 ,
SUCTION
SURFACE
SEPARATED FLOW
-4
0
'C
L-
C_
z 3-o
-J
LAJ
L)
0
0
I-.
SPRESSURE
SURFACE
0.8
0.9
1.0
1.1
v IV,
Figure 66.
I.
V.
1.
144
5.
PREDICIED WAKE
0-
J--
00
3.
LU
2-
00
Uj
U-
00
So
o.-..-__--o
-
00
0
-1.
.6
0
0
0.7
0:8
0.9 0--
Vm/V
Figure 67.
145
5-
00
'N0
O
0
3-
o0
Ui
z
03
0-0
o
0
zF.s
00
-1
S~Figure
0.6
68.
0.10
.70l
0. 6
-,
0.7
O.g
0.9
'---
1.0
VmlV
Comparison of the Theoretical and
Experimental Velocity Profiles in
the Test Cascade Reference Station
50 Percent Blade Chord Downstream.
146
Figure 55 repre-
The
The
seen.
The
the
the flow.
In
147
general,
A tuft
approximately 2 feet long was attached to the trailing edge pressure side
of the central blade in the cascade.
the exit flow angle of the cascade relative to the inflow direction.
These measurements were accurate to 1/2 degree and result in the plot of
Figure 69.
flow regimes.
seen.
severe underturning is
nearly constant,
The most
148
16|A
&I
""13<
LLAMINARI I
OAS DESIGNED
7iRIP AT 60%,C SUCTION
0 FLOW-FIELD SURVEYS
SEPARA-
12 TION "
o1211
20250
360
350
149
REYNOLDS
NUMBER
0.4
01I0,000
a 240,000
0.2
0
01
O,,'-%j
'I
-0.2
-0.4
-0.6
0.6
0.2
0.8
xlt
Figure 70.
1.0
150
At the highest
facility was measured with a hot wire annemometer and found to be on the
order of 0.5 percent of the free stream velocity, which is quite low.
This could account for the unexpected laminar separation and for the
effect of the trip on the cascade performance.
Additionally,
the cor-
relations used in the development of the boundary layer model used on the
design process give no insight into the effect of the over-all turbulence
level on the shape or extent of the boundary layer.
9.4
The
In
151
are in
order.
chord lengths in
both directions.
Increasing
"his region sh
due to
calculation station,
stream-
CHAPTER X
SUM14ARY,
10.1
,rbomachine
A large
the development
tics of the blade rows which change the total energy of the fluid, and
the losses in energy due to viscous and turbulence effects.
The accurate
A computerized numeri-
cal analysis has been developed to solve the equations of motion and
tI
_It-
inviscid,
incompressible flow
153
methods that treat a blade row as a thick actuator disk with uniform
chordwise loading.
The effects of the blade thickness distribution and blade surface
boundary layers may be incorporated into the solution.
This point is
Determination of the
In some cases,
surfaces.
Several examples show the through-flow solution based on the
Streamline Curvature Method to be accurate in predicting the
6-.
.-
154
The design of
Due to the
an
155
motion, with the results corrected for the effects of viscosity and
turbulence.
Streamline Method and iteratively adjusts the camber line and thickness
distribution to satisfy the prescribed boundary conditions.
These
flow field is made to account for the effects of the blade surface
boundary layers and the blade wake.
cascade blade geometry and the associated flow field in the blade-toblade plane, which provides a specified performance.
An empirical cascade of blades was designed and tested in a cascade
wind tunnel to verify this analysis.
the potential of the design method to predict the flow field when the
effects of viscosity and turbulence are small.
However,
serious diffi-
culties were encountered in the boundary layer and wake regions of the
flow field and suggested that more sophisticated models of these regions
must be used.
The blade-to-blade flow solutions allow the computation of pressure
gradients normal to the streamsurfaces in a turbomachine.
These terms
Test
13
156
10.2
are employed to include real flow effects can be made more general for
axial flow units.
flow turbomachines are more difficult to design and their internal flow
processes, especially boundary layer flows,
The
Streamline Curvature Method developed in this study can be used for the
prediction of the flow in a radial flow pump or turbine.
Tests should be
designed to provide the analysis with the same type of correlation data
as used in the axial flow case.
The Mean Streamline Method of blade section design has been reliably
employed for axial flow turbomachinery and two-dimensional cascadeb.
major improvement,
Experimental verifi-
Sand
157
The Streamline Curvature Method developed in this study to design
cascade airfoil sections will require a great deal of additional effort
before reliable designs can be produced.
The empirical
data used to predict the boundary layer and wake effects will require a
test and analysis program, along with an advanced theoretical approach
to produce useful design correlations that properly account for Reynolds
number effects.
measurements of the boundary layer and wake profiles will help to produce
the required correlations.
The development of a three-dimensional design solution for turbomachinery is required if maximum possible performance is to be obtained.
An important part of this solution is the blade-to-blade flow solution.
The Streamline Curvature Method of blade section design produces this
flow field as a part of the design process.
be made, however,
Again,
experimental
Occasionally,
This causes redistribution of the through-flow and, consechanges the performance of the turbomachine.
A large rotating
158
In this solution,
If
at least twenty times the computer power and complexity currently required.
However,
Comparison
and design presented in this study can be used and improved and will lead
ultimately to the design of turbomachines that are more systematic and
reliable and with improved levels of performance.
159
REFERENCES
3.
4.
5.
Marsh, H., "A Digital Computer Program for the Through Flow Fluid
Mechanics in an Arbitrary Turbomachine Using a Matrix Method,"
Aeronautical Research Council, ARC R&M 3509, 1966.
6.
7.
8.
Bosman, C,, and H. Marsh, "An Improved Method for Calculating the
Flow in Turbomachines, Including a Consistent Loss Modefl," J. Mech.
Eng. Sci., 1974.
9.
10.
11.
Billet, M. L., "An Approximate Method for the Solution of the Direct
Problem of an Open Rotor," Applied Research Laboratory TM 78-161,
The Pennsylvania State University, May 1978.
12.
New York,
160
13.
14.
Wilkenson, D. H., "Calculation of Blade-to-Blade Flow in a Turbomachine by Streamline Curvature," Aeronautical Research Council,
ARC R&M 3704, 1972.
15.
16.
McBride,
M. W.,
J. Fluids Eng.,
17.
18.
Smith, L. H., and H. Yeh, "Sweep and Dihedral Effects in Axial Flow
Turbomachinery," Trans. ASME, J. Basic Eng., September 1963.
19.
20.
21.
22.
23.
24.
Erwin, J. R., et al, "Two Dimensional Low-Speed Cas:ade Investigation of NACA Comp-essor Blade Sections Having a Systematic Variation
in Mean-Line Loading," National Advisory Committee on Aeronautics,
NACA TN 3816, November 1956.
25.
26.
161
JI
S
4j
27.
28.
29.
30.
31.
32.
33.
34.
35.
Schlicting, H.,
Book Co., Inc.,
36.
37.
38.
4th ed.,
New York,
N. Y.,
Butterworths,
McGraw-Hill
London,
1958.
Commander
Naval Sea Systems Command
Department of the Navy
Washington, DC 20362
Attn: Library
Code NSEA-O9G32
(Copy Nos. 1 and 2)
Commander
Naval Ship Engineering Center
Department of the Navy
Washington, DC 20361
Attn: D. V. Burke
Code NSEC 6140B.1
(Copy No. 23)
Commander
Naval Surface Weapon Center
Silver Spring, MD 20910
Attn, Library
(Copy No. 24)
Commander
David W. Taylor Naval Ship R&D Center
Department of the 4avy
Bethesda, MD 20084
Attn: R. J. Boswell
Code 1544
(Copy No. 25)
6 - 17)
R. A. Cumming
Code 1544
(Copy No. 26)
Attn:
Cot.manding Officer
Naval Underwater Systems Cepter
Newport, RI 02840
Attn: J. F. Brady
(Copy No, 18)
(Copy No.
Commanding Officer
Naval Ocean System:n Center
Department of the Navy
San Diego, CA 92152
Attn: J. W. Hoyt
Code 2501
upy No. 20)
29)
Of f icr- in-Charge
David W. Taylor Naval Ship R&D Center
Department of the Navy
Annapolis Laboratory
A.-napoiis, MD 21402
Attn: J. G. Stricker
Code 2721
(Copy No. 30)
4aa
27)
a4a
Research Center
Bodapati
227-9
41)
Calspan Corporation
4455 Genessce Street
Buffalo, NY 14221
Attn: Head Librarian
(Copy No. 58)
MS-16 United Technologies
Research Center
Silver Lane
E. Hartford, CT 06118
Attn: E. M. Greitzer
(Copy No. 59)
Creare, Inc.
Box 71
Hanover, NH 03755
httn: W. Swift
(Copy No. 60)
Rocketdyne Division
North American Aviation
6633 Canoga Avenue
Canoga Park, CA 91303
Attn: E. Jackson
Hydrodynamics Group
(Copy No. 61)
Aerojet General
Post Office Box 15847
Sacramento, CA 15813
Attn: Library
(Copy No. 62)
Hydronauti-s, Inc.
Findell School RoLd
Laurel, MD 20810
Attn: J. Bohn
(Copy No. 63)
Allis-Chalmers Corporation
Hydro-Turbine Division
Box 712
York, PA 17405
ALtn: R. K. Fisher
(Copy No. 56)
Allis-Chalmers Corp.
Attn: W. Whippen
(Copy No. 57)
Professor J. H. Horlock
Vice-Chancellor
University of Salford
Salford, M5, 40T
England
(Copy No. 75)
Dr. John Foxwell
Admiralty Research Laboratory
Teddington, Middlesex
England
(Copy No. 76)
Whittle Turbomachinery Laboratory
Madingley Road
Cambridge
England
Attn: Library
(Copy No. 77)
Whittle Turbomachinery Laboratory
Attn: Sir William Hawthorne
(Copy No. 78)
Purdue University
Thermal "cience & Prop. Center
West Lafayette, IN 47907
Attn: Dr. S. N. B. Murthy
(Copy No. 71)
Dr. D. S. Thompson
Turbine Research Department
Rolls Royce Ltd.
P. 0. Box 31
Derby
England
(Copy No. 80)
University of Pennsylvania
Towne School of Civil & Mech. Eng.
Room 113 Towne, 220 S. 33rd St.
Philadelphia, PA 19104
Attn; Dr. Hsuan Ych, Director
(Copy No. 72)
Stevens Institute of Technology
711 Hudson Street
Castle Point Station
Hoboken, NJ 07030
Attn: S. Tsakonas
kCopy No. 73)
Stevens Institute of Technology
Attn: Dr. Bruce D. Cox
(Copy No. 74)
Professor
School of
Cranfield
Cranfield,
England
(Copy No.
R. E. Peacock
Mechanical Engineering
Institute of Technology
Bedford MK430AL
81)
dated
Professor J. P. Gostelow
School of Mechanical Engineering
NSW Institute of Technology
Broadway
Sidney
Austrailia
(Copy No. 86)
Dr. rer. nat. Horst Merbt
Forschungsbeauftragter fiir Hydroakustik
8012 Ottobrunn B Munchen
Waldparkstr. 41
Munich
Germany
(Copy No. 87)
Carl-Anders Johnsson
Statens Skeppsprovningsanstalt
Box 24001
S-400 22 Goetborg
Sweden
(Copy No. 88)
Dr. V. H. Arakeri
Department of Mechanical Engineering
Indian Institute of Science
Bangalore 560 012
India
(Copy No. 89)