by
Elwyn S. Baker
DEPARTMENT OF THE NAVY
NAVAL SHIP RESEARCH AND DEVELOPMENT
CENTER
BETHESDA, MD. 20034
ANALYTICAL PREDICTION OF WALL EFFECT ON FULLY CAVITATING LIFTING FOILS, USING
NONLINEAR THEORY
APPROVED FOR PUBLIC RELEASE: DISTRIBUTION UNLIMITED
TABLE OF CONTENTS LIST OF FIGURES 11 Figure Figure 1 2 -Flow-Field Boundary
Mapping Planes- -Choked Flow in an Arbitrarily
Shaped Tunnel, From Reference 3
Page
5
Figure 3 - Mapping Planes--Nonchoked Flow in a Straight-Sided
Tunnel, From Reference 3 16
Figure 4 - Lift and Drag Coefficients versus Cavitation Number
at Various Angles 34
Figure 5 - Percentage Reduction of the Predicted Lift and
Drag Coefficients Due to Tunnel Wall Proximity for the
CaseH/L=4 .
40Page
ABSTRACT. 1
ADMINISTRATIVE INFORMATION 1
INTRODUCTION 1
WALL EFFECT THEORY 6
CHOKED FLOWS 6
FINITE CAVITY FLOWS 15
Equation Set 15
Changes to Equations 21
EXPLANATION OF THE COMPUTER PROGRAM 31
DISCUSSION OF RESULTS 60 CONCLUS IONS 69 RECOMMENDAT IONS 70 AC KNOWLEDGMENTS .71 APPENDIX A - w INTEGRAL 73 0
APPENDIX B - INTERPOLATION FORMULA . 81
APPENDIX C - COMPARISON WITH 1955 THEORY OF WU 83
APPENDIX D - COMPARISON WITH LAROCK AND STREET THEORY 93
APPENDIX £ - PROGRAM NOTES AND LISTING 99
APPENDIX F - COMPUTED DATA POINTS 129
REFERENCES . 134
Page Figure 6 - Choked Cavitation Number versus Incidence Angle and
Tunnel Wall Spacing at Various Angles 42
Figure 7 - Free Streamline Location versus Tunnel Wall Spacing
Under Various Conditions 45
Figure 8 - Pressure Coefficient Distribution versus Tunnel Wall Spacing Incidence Angle = 10 Degrees, Cavitation
Number = 0.4 48
Figure 9 - Comparison of Pressure Coefficient Predictions from Wu's Theories and Larock and Street's Theory for the Flat Plate in an Infinite Stream at Incidence
Angle = 10 Degrees 49
Figure 10 - Comparison of Streamline Locations Flat Plate
Infinite Stream c'. = 10 Degrees Wu's 1969 Theory
Roshko (or Transition) Wake Model 50
Figure 11 - Comparison of Streamline Locations Flat Plate
Infinite Stream ci. = 10 Degrees Larock and Street
Theory Tulin Single-Spiral Vortex Wake Model 51
Figure 12 - Comparison of Streamline Locations Flat Plate
Infinite Stream ci. = 10 Degrees Wu's 1963 Theory
Wu 1963 Wake Model 52
Figure 13 - Lift Coefficient versus Cavitation Number for a Flat
Plate Hydrofoil at c = 15 Degrees, Infinite
Stream--Comparison of Various Theoretical Predictions 53
Figure 14 - Comparison of Predicted Streamline Locations with Experimental Data Circular Arc Hydrofoil,
2Y = 16 Degrees, c = 10 Degrees, H/L = 6,
a=0.26
54Figure 15 - Comparison of Predicted Coefficients with
Experimental Data 55
Figure 16 - Sample Cross Plot for Cavitation Number 59
Figure 17 - Comparison of Mapping Planes between Wu's 1955 and
1969 Theories 85
Figure 18 - Sketch of Common Parameter Plane Cut Values for Various Theories for Flat-Plate Hydrofoil in
Infinite Stream 94
Figure 19 - A Mapping Conversion from Common Parameter Plane of Larock and Street Theory to that of Wu 1969
Theory 96
Figure 20 - Comparison of Common Parameter Plane Cut Values between Wu 1969 Theory and Larock and Street Theory--Flat
Plate, = 10 Degrees, Infinite Stream 98
Thicknesses of upstream stream above and below dividing streamline, respectively
General line integral Imaginary part of
NOTAT ION
b
C
C
K Parameter in Larock and Street theory C
L Arc length of foil; segment of real axis
LE Leading edge
in Natural logarithm of real positive quantity
log Natural logarithm of complex quantity, principal
value
M Image point used in mapping
m Algebraic parameter on axis
Pc Cavity pressure, not normalized
p Pressure at upstream infinity, not normalized
A Real constant scale factor used in complex
potential mapping
Al Power series coefficients
A,B,B',C,C',D,D',E,E' Image points used in conformal mapping
a,b,b',c,c' Algebraic parameters on axis
CL, CD Lift and drag coefficients based upon chord length
and upstream velocity
Pressure coefficient (P.-P,) / p
ds Infinitesimal element of arc length
dx, dy Components of ds
d1, d2 Thickness of upper and lower downstream streams
e Base of natural logarithm system 2.71
f Complex potential f = + i
f, g General real functions
h Thickness of upstream stream
hD, hD, Heights of leading and trailing edges of foil in
unit tunnel
h0 or hL Height of chord centerline in unit tunnel
h1, h2
I Im
;
q Real magnitude of normalized velocity vector,
called "speed"
R General radius
S Total arc length of foil
SP Stagnation point
S(), S(')
Arc length as a function of parameter plane axis ,transformed axis '
s Arc length
TE Trailing edge
t Common parameter plane in Larock and Street theory
tb Parameter in Larock and Street theory
U Upstream velocity magnitude--normalized on the
free-streamline speed in the nonchoked case only
U Speed at upstream infinity
u,v Component magnitudes of complex velocity w
V Downstream velocity magnitude--same normalization
as u
w Complex velocity, u - iv, or qe_10
x Real axis of physical coordinate plane
y Imaginary axis of physical coordinate plane
z Complex physical plane variable z = x + iy
Inclination angle of foil chord line with respect
to the horizontal direction
-Asymptotic inclination angle of curved wall at downstream infinity and upstream infinity
ct(s) Local wall inclination angle as function of arc
length s along the wall
Local wall inclination angle as function of
parameter plane axis.
s(s) Local inclination angle of foil tangent as a
function of arc length along the foil, measured with respect to the chord line, positive clockwise; characterizes shape of foil
Same quantities as functions of parameter plane
axis , transformed axis '
One-half the included angle of a circular arc
Normalized pressure coefficient (C + a)/(l + c)
Small quantity
c. Complex parameter plane variable = + Imaginary axis of complex parameter plane
0 Real angular direction of normalized velocity
vector, positive counterclockwise, called direction
00, 01 Components of 0
Real axis of complex parameter plane
(x) Values of on axis as a function of transformed
variable X
Point on axis
IT 3.14
a Cavitation number (ç_P)/ p U2
aCH Chokd cavitation number or blockage constant
a Cavitation number based on minimum tunnel wall
wall
pressure
T Logarithm of inverse speed
T0* Nonsingular component of
T0, .t1 Components of T
General contour integral
Velocity potential; general complex function Variable of integration corresponding to
trans-formed values of F, > 1
Point on axis
Angular parameter, corresponding to positions
along the axis from c to c'
Values of x corresponding to the positions of
a,b,b'on
Two-dimensional, irrotational stream function Square root of complex quantity, branch to be
determined
Square root of positive real quantity, sign positive Superscript, limiting value on boundary from the upper half plane
n a wh i is ne s m coc rat by is wh the on me ve
th'
x
V \i I Aa "b' "b 11) ( )l/2 +e
ABSTRACT
A Fortran computer program, compatable with the CDC 6000 series, has been developed to evaluate the effect of a
closed-jet tunnel-test section on the force coefficients of
arbi-trarily cambered, sharp-nosed, fully cavitating, two-dimensional
hydrofoil sections. The numerical solution to this prcblem has been obtained using the nonlinear theory developed by
T.Y. Wu in 1969. The theory employs an open-wake model, commonly called the Roshko or transition-wake model, to
represent the cavity downstream of the foil section. Results
of the program include predicted force coefficients,
stream-line shapes and pressure coefficient distributions. Some
comparisons are made with the predictions of two other nonlinear theories available for the infinite-stream case only.
ADMINISTRATIVE INFORMATION
This project was authorized and funded under the General Hydro-mechanics Research Program and partially by the In-House Exploratory De-velopment program.
INTRODUCTION
The past decade has seen remarkable progress in the development of
naval hydrofoil craft. These ships have hulls of the planing boat form,
which in the cruise condition rise up out of the water. The ship weight
is then supported by fully wetted wings of the aerofoil type running
be-neath the water surface. These wings are characterized by a relatively
small thickness ratio when compared to aircraft wings, by low lift coefficients, and by high efficiencies as measured by the lift-to-drag
ratio. The speed with which such wings can move through water is limited by cavitation erosion of the material, the inception of which phenomenon
is characterized by low values of the dimensionless cavitation number,
P - Pc
1 2
p U0
where Pc, the cavity pressure, is generally taken as the vapor pressure of
the fluid. At high speeds and low values of a, small vapor bubbles appear on the suction side of the hydrofoil, which upon being swept into a higher
pressure region of the flow, can collapse and damage the material of the wing.- The design of fully wetted wings is therefore characterized by an attempt to make the thickness and camber ratios as small as possible consistent with structural requirements so that the lift coefficient, and hence the pressure difference across the foil, which encourages the growth
of cavitation bubbles, is small.
An alternative approach to the design of hydrofoil wings is to
encourage the growth of these bubbles to the point where the entire suction
face of the hydrofoil is covered with a large bubble of fluid vapor. The
resulting design is referred to as being supercavitating or fully
cavi-tating. The leading and trailing edges of such foils are generally
sharpened to encourage the separation of the fluid at both points. The
most notable example of such.a foil is the surface-piercing forward foil
on the Canadian Ship BRAS D'OR (renamed FHE-400). These foils are
characterized by somewhat higher lift coefficients and lower lift-to-drag ratios than their fully wetted counterparts; however, they do not suffer cavitation damage at high speeds.
The section shapes-for supercavitating foils are generally tested in
-cavitation tunnels. In contrast to the case of fully wetted foils, for which the tunnel correction method is well established, there is no simple method of calculating the effect of the tunnel walls on measured force data
of supercavitating sections. Indeed, the calculation of two-dimensional
wall effect on fully-cavitating lifting hydrofoils using nonlinear theory has hitherto only been attempted for flat-plate hydrofoil sections in choked (minimum cavitation number) flows.1
The computational difficulties involved in considering the cambered
foil case invite the use of linearized theory. However, in several cases
the linearized theories have been shown to be in error;2 in particular the
D.K. and Z.L. Harrison, "Wall Effects in Cavity Flow," California Institute of Technology Hydrodynamics Laboratory Report 111.3 (Apr 1965).
2Wu, T.Y., "A Note on the Linear and Nonlinear Theories for Fully
Cavitated Hydrofoils," California Institute of Technology Hydrodynamics Laboratory Report 21-22 (Aug 1956).
1: ti p tJ n( b di U] ac i
1
cc iT dilinearizing assumption of small perturbation velocities is not valid near the leading edge of the foil, where, because of high loads, an accurate prediction of the free-streamline shape, which controls the thickness of
the foil, is necessary for good design. The report by Wu et al.3 offers a
nonlinear theoretical model of the problem; the purpose of this work has been to convert the model into a computer program to numerically evaluate
the wall effect on arbitrary two-dimensional hydrofoil shapes.
While a three-dimensional wall-effect theory would most certainly be preferable to the two-dimensional theory considered here, the mathematical difficulties involved in analyzing three-dimensional cavity flows (partic-ularly the difficulty of applying exact boundary conditions at an unknown cavity boundary location) have so far limited developments in this
direction. It is to be hoped that the two-dimensional theory will give
satisfactory results for flows over high-aspect-ratio hydrofoils. In
addition to two-dimensionality, the present theory assumes that the flow is irrotational and inviscid and that separation occurs from the sharp
leading and trailing edges of the foil. Therefore, the lift-and-drag
coefficients derived from it are hydrodynamic coefficients only and do not-include viscous effects on the flow, such as skin friction and the
diffusion of vorticity.
In real cavity flows the steady upstream flow degenerates into turbulence at some distance behind the obstacle; therefore, representation of the flow downstream by means of steady-state, discrete streamlines can
be only a poor approximation to the true wake flow. Fortunately, the
hydrodynamic forces on the obstacle appear to depend primarily on the up-stream flow, where the potential flow representation is much more reason-able.
Several wake models have been developed which approximate the
vis-cous wake by a perfect fluid
flow.
These can be divided into two classes:(1) closed-wake models, in which the free streamlines join at some down-stream point to form a closed cavity, and (2) open-wake models, in which
3Wu, T.Y. et al., "Wall Effects in Cavity Flows," California Instutute of Technology Report E-lllA.5, Chapters 7 and 8 (Apr 1969).
3
n
the free streamlines remain separated until downstream infinity. Since
most potential flow theorems take their simplest forms when applied to simply connected regions, which requires that the flow be "enclosed"
within a single continuous contour forming the boundary of the flow region, it is natural in considering the wall effect problem to first attempt an
open-wake cavity representation. Then the walls and the hydrofoil-free
streamline boundary can both be considered as a part of a single boundary. The up- and down-stream streams can also be considered as part of this boundary since the direction of the flow in them is known and constant, as
it is along a flat wall.
3
In the open-wake model used by T.Y. Wu, he assumes that the wake is divided into two regions: a near wake immediately behind the obstacle
and a far wake further downstream. In the near wake the pressure remains
constant at its minimum value anywhere in the flow field; in the far wake
it gradually regains part (but not all) of the upstream pressure. The
point at which this transition to the far wake begins is that at which the free streamlines have become parallel to the walls and have the free
streamline (maximum) speed. These bOunding streamlines continue downstream
from this point parallel to the walls but lose part of their speed to
account for the increase in wake pressure. If the velocity of a fluid
element is expressed as a magnitude and a direction, it can now be seen that one component of the velocity is known everywhere along the boundary
shown in Figure 1. The velocity magnitude is known (constant). on the free
streamlines, and the direction of the flow is known on the foil, at up and
downstream infinity, and on the walls. This is a sufficient specification
of boundary values for a potential flow solution in a simply connected
region.
In prograimning this problem, reliance has been almost entirely on
the theory developed by Dr. Wu, and only those changes have been made which were necessary for the numerical calculation of the results and to apply
q=U p= p00 q= 1 LE (LEADING EDGE)
8= ir-fl(s)
SP (STAGNATION POINT) B = . (S) TE (TRAILING EDGE) q= 1 q-V\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\
d1 q=V 8 ci FAR WAKE P >P NEAR WAKE p=Pcd - (c-a) (c-b) (c-b')
using the Schwartz-Christoffel transformation, where A is a real constant
(scale factor). To map the velocity field onto the same plane, the velocity is expressed as the logarithmic hodograph variable
= log--=
T +
i.Owhere
t =
ln V/q, normalized on the free streamline velocity V, whichequals the downstream velocity for infinite cavity choked flows (the value
of V is not normalized). This variable is decomposed into two parts, w0
and c
w=w0+W1,
W0T0+ie0,
()1T1+iO1
4Holt, M., "Basic Developments in Fluid Dynamics," Vol. 2, Academic Press, New York (1968).
WALL EFFECT THEORY
CHOKED FLOWS
The theory of wall effect for arbitrarily shaped two-dimensional bodies in choked (minimum cavitation number) flows was given by Wu in
Reference 3, Chapter 7. (See also References 1 and 4 for more
infor-mation on this problem.) In this presentation the shape of the tunnel
walls may also be arbitrary. Since ideal choked flows are characterized
by infinitely long cavities, the choice of a wake model does not enter this problem; both the free streamlines extend to downstream infinity, and the speed along them remains constant at the upstream value.
A summary of this theory follows, using the Wu's notation shown in
Figure 2. The complex potential plane f = + i is mapped onto the upper
half of the parameter plane = + i by
Figure 2 - Mapping Planes--Choked Flow in an Arbitrarily Shaped Tunnel, From Reference 3
7 ity f-PLANE A
L1'
IJh.=Vd1 B' B E U B A U' B' ' =-Uh=-Vd B'Following the boundary of the flow region counterclockwise, one
sees that at every point one component of the velocity is known.
On the free streamline the velocity magnitude is given as V; on the foil surface
its direction is specified by s(s); on the wall surface, by ce(s).
Across
the upstream and downstream flows the direction of the velocity is uniform
at angle c, and its magnitude is U and V, respectively.
To correspond to this decomposition, the boundary conditions
on
the velocity are divided as follows:
parameter plane +
<-1
11=0
+-l<<b,, b'<<1 11=0
+b<<b' n=O
>+l
ii=0
w0 velocity 00 = ir + = 0T00
01 0 =The superposition of and w1 gives a velocity field in terms of
con-sistent with the- conditions along the physical flow boundary. Once the
f-plane and the w-plane are expressed in terms of a common parameter plane
, they can be combined to give the physical
z-plane and the physical
parameters of the problem.
Wu gives the solution to this boundary value problem for as
= = log 2 1/2) 1/2 C + (C -1)
-(-
(2))l/2
1b' [cL() - 0()] d b (c-C) 3/(-b) (b'-)(2)
w1 velocity 01 = 0 = 0 = ct() -01 =.0 +co-
r
lI
(c-C)
(2) 5 Flow Eng i pro and wher (1) knowwhere w0(C) must be analytic (throughout the 'upper half plane
except at
°°) and satisfy the previously described boundary conditions.
More details
of the derivation of this expression will be found in Appendix A.
Similarly,
The derivation of this may be sketched by considering the
Rieinann-Hubert problem in an upper half-plane, as summarized by Larock (l966). Supposing the imaginary part of some complex function to be known everywhere along the real axis E, the analytic function Q will be given in the upper half t-plane by
'm
Q()
=S
d
+'
A.
-where = + ir, A. real. The statement of the problem w1 is
Where = + i01, and 00 is known from the solution of the w0
problem and depends only on the function () and not on the parameters b
and b'.
Suppose a function H() such as
H()
= i4-') (c-b) (c-b') (-(-l)) (4)= /(-b) (b'-)
(2)
where the argument of the square root changes sign at the points where the
known part of becomes real or imaginary. Let
wl
- H()
5 ,,
Larock, B.E. and R.L. Street, A Nonlinear Theory for Fully Cavitated
Flows Past an Inclined Flat Plate," Stanford University Department of Civil Engineering Technical Report 64 (Jun 1966).
9 j =0 (3) L) [y, 01 = 0
>+1
<-1
n=0'
T1 = 0-1<<b b'<<+l
=0
01 = o() -b<<b'
on the half-plane boundary to give
Thus, it is seen that the imaginary part of Q() is known everywhere along
the axis. Hence, from Equation (3), the extension of Q into the positive half-plane may be expressed as
Im Q
Q()
=f
d +2'A.
i j=0 1 b' ct() -d A. ?=J
b(-)
and the velocity field is given by
= H() Q()
=
- (-
(-b')(2-[c() - O0()J d
(-)
The power series term is eliminated by considering the behavior of w near
the stagnation point. Since Re w
= T =
log l/q, as q - 0 at the stag-nation point, T must -- c logarithmically there. This behavior is exhibitedby the term of the velocity field expression and must not be changed by
the addition of T1.
Therefore, the complete solution to this Riemann-Hilbert mixed-boundary-value problem in the upper half t-plane, continuous at the transition points, is given as
plal as a cc
bow
Corn undc The by Im Q() Im Q() = = 0-l>,
0-l<<b,
[cL()->+l
=0
b'<<+l
=0
Im Q(fl - -I
b<<b'
n=0
V(F-b) (b'-) (21)
)l 2= + i0() = w0() + q () = log (1 + 2 1/2
(21)1/2)l(
-')(1'
1/2 +(b')
(2l)1
jbt
=J
q:()
d L - __________ b(-)
The first term may be thought of as the velocity field due to a flat
plate in an infinite stream at the ch'oked-flow conditions; the second term,
as a correction due to cambering the plate at angle ; the third term, as
a correction due to the presence of a wall at angle c.
That this expression for the velocity does indeed satisfy the given boundary conditions can be seen from the limit formula for Cauchy integrals. Consider the wall effect term w.
2 1/2 1 1 l))
jbt
-d b(-)
under the conditions
= - 00(e) when b < < b'
when-l<<b
orb' <<+1
01 = 0 when
II
> 1The limit from the positive half-plane of a Cauchy-type integral is given
by Plemelj's formula as6'7
=7r i
-
I
\c)d
Jl )()p/i
[c() - 0()]
d 4.vpc() d
J
L (5)6Mushelishvili, N.I., "Singular Integral Equations," Noordhoff Groningen,
Holland, p. 237 (1946).
7Pogorzelski, W., "Integral Equations and Their Applications," Pergamon
where the term on the right is the Cauchy principal value (real valued, if
J. is the real axis).
Comparing this with the expression for w,, with L
defined as the segment of the real axis from b to b', we see that the
limit
of as
-F0, ii -' is
wl(o) = .
(b'-0)
(2)
Then taking the integrand function in Plemelj's formula as
- 0o
Va-b) (b'-)
(l_2)and comparing with the Cauchy integral gives
= -b' [ct()
- O()]
d +-- 0-b)(b'-0)(l-02)J
b(o
which has the required imaginary part.
Furthermore in the range -1 < <
b, b' <
< +
1 the termis purely imaginary; since the integral
VP 1b'
[c() -
oo()I
db ()
is real in this range, the expression for must be purely
imaginary,
satisfying the condition that T1 = 0. A similar argument shows that
= 0 for IEI > 1.
b'
Ic() - e0(fl
d(o
j(-b) (b'-)
(l2)
Now with expressions being given for both the velocity
field and the
velocity potential in terms of a common parameter plane (?-plane), the
physical z-plane is derived from
or wh e foL upF anc th Th an
fr
and axiit df = w, z = 0 at = -1 1 e° df
z()
=J
1and the arc length of the foil as a function along the parameter plane axis becomes (since IdzI = dS)
S()
=
-J
) d n = (7)where S(l) = S, the total arc length.
Using the same distance Equation (6) allows an expression to be found for hD, the tunnel position of the nose, by integrating along the
upper streamline. b / -Ia - -Ia0
i0() df
hD = d1 + Im te 0 (zB_zD))= d1 ++
5
Im Le ejd(8)
-1Similar integral equations are given for the positions of the upper and lower walls with respect to the foil.
The distance Equation (6) may also be used to find the thickness of
the up and downstream streams. Let vertical distances be given by
= liii
[e0 (zz0)]
= Im[e0+J
eW--d]
Then, supposing that the streams up and down stream asymptotically approach
an angle of ct0, at the upstream infinity we have - a, and by definition from the Riemann-Hilbert problem w = ln V/U + ia0, so
a+c d -1 A
[rO
ii4 5
(-a)(-b)(-b')]
Ah=Im
eCO
a-creplacing w by its value in the limit gives
13
n>
0 (6)a+c
[-ict0
(in-4j_+ icx0)j
d = - A e a-c (c-a) (c-b) (c-b')]
a+c = - A . 1im[iiml
1 d (a-b)(a-b')cO j
(c-a)]
tr a-c= - A/U 1 un [log(a+c_a) - log (a_c_a)]
(b-a) (b'-a)
1
liii [iogf-!-'l= log (-1)
L
\-i]
(b-a)(b'-a)Take the negative log branch to give the upstream thickness as
Air Air - A/U A d2 = + (b-a) (b'-a) h (b'-b)(b'-a) U(b-a)(b'-a) U(a-b)(b'-a)
The negative sign which would naturally appear in this expression were it not for the choice of logarithm sign expresses the sense of the boundary-a positive c-direction corresponds to boundary-a jump from the upper to the lower
wall. In a similar fashion, the thicknesses of the downstream streams are
r.
I v.\
b+c I -iclnv + lcL0)
. 1 1 1d --AImie
elimi
L
b-c(-a)(-b)(-b')
A it =(b-a) (b-b')
wherein the positive logarithm sign has been used.
Thus, when the functions ct() and () are given, where is the real
axis of the common parameter ?-plane, a sufficient number of equations is given to determine all the physical parameters of the flow, including the
shape of the resulting hydrofoil, which may be found by calculating s()
Fl0 Tec TN tat (J u and (S fin cal the FIN the flo cas Ref rep Fig edg t Un fun ang vel ye 1 and
e
ea 1
and eliminating the parameter between that function and () to give
s(s). However, if it is desired to specify s(s), then it is necessary to
find some function s() such that the substitution 8(s())
= () and the
calculation of s() results in a generated function which is identical to
the original s(). This may be done by integral iteration.8
FINITE CAVITY FLOWS
Equation Set
For the case of cavity flows with a finite cavity, we specialize.
the method used by Wu in Chapter 7 of Reference 3 in solving the choked
flow case by adopting his boundary conditions used in the solution of the
case of finite cavity flow over a flat plate with wall effect, Chapter 8,
9,10
Reference 3. This corresponds to the use of the Roshko wake model to
represent the finite cavity. The mapping planes then become those shown in Figure 3. The physical coordinate system is centeredon the foil leading
edge, with its x-axis along the chord. We choose the case of a straight
tunnel with parallel walls at angle a to the chord of the foil, so that the
function a() reduces to a constant a. The upstream velocity becomes U at
angle a; the downstream velocities, V at angle a. Having normalized the
velocity field on the free streamline velocity, taken as unity, the
velocities have the relation
U<V< 1
and the cavitation number is given by
1
a 1
U
8Wu, T.Y. and D.P. Wang, "An Approximate Scheme for the Theory of Cavity
Flows Past Obstacles of Arbitrary Profile," California Institute of Technology Report 111-1 (Jul 1963).
9Roshko, A., "A New Hodograph for Free Streamline Theory," NACA TN 3168 (1954).
T.Y., "A Free Streamline Theory for Two-Dimensional Fully Cavi-tated Hydrofoils," California Instutute of Technology HL Report 21-17
(Jul 1955).
z - plane
- plane
Figure 3 - Mapping Planes--Nonchoked Flow in a
Straight-Sided Tunnel, From Reference 3
f - plane
n -'p.-r
I4 3 o 9In the choked flow case the cavity extended downstream to infinity, and consequently the points C, C' at which the free streamlines became parallel to the walls coincided with the points B, B' at downstream
in-finity. For flows .with a finite cavity, the open-wake model represents
the real physical case of convex free streamlines meeting in a turbulent wake as streamlines extending to downstream infinity from the points at
which they become parallel to the walls (C, C'). The velocity along these
lines then varies in such a manner that the net pressure effect on the foil
ahead of them remains unchanged. Thus, for finite cavities the streamline parallel points C, C' are some finite distance behind the foil, marking
the transition from the near wake, where the free streamline speed is
unity, to the far wake, where the streamline velocity approaches V.
Conse-quently, as the boundary of the flow region is traced around the path
shown in Figure 1, we let = c at C and = c' at C'. The statement of
This may be written more concisely as
- _ - i e =
ii +
B()
T0
+C<<+c'
c'<<+l
r=O
17 +T10
H the Riemann-Hilbert boundary value problem corresponding to this case
be-comes
D
Location Segment Parameter Velocity + -4 Foil Forward of Stagnation Point ED
n= 0
-<<- 1
0 = t + Upper Streamline DC- 1<
< C ci ci Upper Wake CBc<<b
0=c
-4 I) Upper Wall BA b<<
a0=ch
Lower Wall AB' a < < b' 0= c
Lower Wake B'C' b' < < c'
0=ct
Lower Streamline C'D'
C' <
< 1Foil Rearward of
Stagnation Point D'E'
1<
which, remembering the principle of superposition, used in the last w
section, can be decomposed into i
C
<-
1 - 1 < < c, c' < < + 1C <
< C'
01=0
t0=0
T0 = 0 01 = a-01=0
By comparison with the choked-flow case we see that the statement of the Riemann-Hilbert problem would be the same if the parameters c and c'
replaced b and b', and the constant a replaced the function ct(E).
There-fore we replace b, b' with c, c' and adopt Wu's solution for the previous Riemann-Hilbert problem, giving
W
W + ()
= ln--+-i0 0 1 q 001 U w0 log [ 2 1/2] 1 2 1/2[f1
()
d -j()( -1)
1 =+( -1)
] jf ( -1) -2 1/2 (11) C r ) d C' a-00( 1/2 (12) 1Wl =
[(c-c)
(c-c')
(2l)]
1/2()
[-c)
(c'-)(i2)]
CThe condition at the stagnation point becomes
Jc,
oo() d
ct a - d C0,or a-
(13) c-I
d cV(-c)
(c'-E)(2)
A theoretical explanation of this condition, necessary because of the singu-lar behavior of w at infinity, may be found in Reference 7, pg. 654.
Now we apply the boundary conditions used in solving the flat-plate
t
case. The complex potential expression remains unchanged, as do the
algebraic stream thickness relations, see Reference 3, page 40. However,
1
a
- ..
we must now require that the downstream streams have equal velocities at
infinity, to correspond to a physical situation of a closed wake with a
constant under pressure, or
= T1(b')
= in (i-)The upstream velocity is calculated from
1
T1(a) = in
and m, which is the point in the c-plane corresponding to the minimum
velocity position on the wail, is calculated from the maximum point of the
curve.
T1(m) = maximum , c < m < c'
We choose our tunnel position parameter as hL, the height of the
midchord point of the foil, and require this to equal one-half in a tunnel
of unit height. Then hL = (hD + hD')/2, and hD and hD, may be determined by integrating the distance formula along the upper and lower streamlines.
There is now a sufficient number of equations to determine the
physical characteristics of the problem. The solution is expressed in
terms of five unknown algebraic parameters a, b,.b', c, c' and one unknown
function s()/S. For inverse problems these parameters are arbitrarily
chosen, and the foil shape is calculated from them; for direct problems it
is necessary to iterate for s()/S. The full equation set is
=+in
w=ifl-+iOW0+W1T.10
'S
= log 1/21 - _ + (;2l) iT2(
I-iJ)
()
-
1(-)
w) =
[(c
(c-c')(21)]
jC?
-
d C(-)
19 f 1) 3)(2)
TJ(b)= T1(b')
= in j T1(a) = in = in (1 + ci) df -A (c-a) (c-b) (c-b') fb'-a\ (a-b) = (U/V) Ib'_b) d2 = (U/V) (b'-b) A =- U (a-b)(b'-a) hD = hD,0() d
s() =-s=-
-hD = di - A hD, = d2 - A:1:
(.1
+ sin (0(c) -a) dJ1
(c-a) (c-b) (-b') sin (0(c) -a) d (c-a) (c-b) (c-b')T()
df d e d j e d + / d Arclength of foil df ds() = d(s()) = eT d-T()
P()=l-q()2=
p1 +ci
Normalized Pres sure Coefficient He ye no mo arSi
pa ne th Ch w0 si 0p
Lift = Drag = Lift Drag
CL=
p/2
U2 chord CD =p/2
U2 chordHere all distances are normalized on a tunnel height of unity; all velocities, on a free streamline velocity of 1.
Changes to Equations
The previously described equations as written provide an exact, nonlinear description of the wall-effect problem using the Roshko wake
model. However, for practical applications and computer applicability they
are unsatisfactory for two reasons: (1) the integral expressions for w are
singular; and
(2)
at the low-incidence angles of practical interest theparameters c, b, a, b', c' approach each other and +1.. Therefore, it is necessary to investigate various transformations of the equations, with
the object of eliminating the more serious of these drawbacks.
Changes to w0(). First consider the expression for = + i 00
= + in
= log
+(2i)
1/2]
(2i)
1/2
(l
-8(e)
d - J1I
Since all the physical quantities of interest in the problem are defined
on the boundaries of the flow field, which correspond to the real axis of the
parameter plane, Equation (11) is of primary interest when ii + 0
Further-more, it appears from the statement of the Riemann-Hilbert problem for
21 Ep
cos
()
cos
ds d ds - sin()
sn
p 2w0() that w0 is purely imaginary over the range +
-l<<+l
n=O
Therefore, for < 1, i = 0+ 1 / 2\ log [+.(2l)h/2]
= log +(2l)
) = log ( +On an Argand diagram it is seen that
and
i.e . . -1
log (e ) = 10 = 1 CO5
so. that for < 1, r =
log ( 2
l/2\
+(-l)
)=icos'
For the integral term, it is desirable from a numerical standpoint
to avoid the integration to infinity. Consider the change of variable11
= - sec (c'ir) = 0
lI >
1then = - 1 '=0 leading edge
= * ' = 1/2 stagnation point
= + 1 '
= 1 trailing edge
2 1/2
The square-root coefficient (t -1) was defined in such a way that
,, 1/2
1
0
-(-l)
+ for complex ; therefore, as 11 -'- O', we must take thenegative real branch, if < -1, and the positive real branch, if > +1.
If j < 1, n = 0, we take the positive root. The term within the
1Abrainowitz, M. and LA. Stegun, "Handbook of Mathematical Functions," National Bureau of Standards, p. 889 (1964).
integ chang see ate as fc This Simi Car roo Ri e +
e0
II
< 1 , realintegral is always real and takes the plus root also. With this
change of variable the integral expression can be written as
-1 +
J
')
d'
I<l
(14) 0 +see Appendix A. This integral is nonsingular but still difficult to
evalu-ate as -' ± 1. Therefore, we subtract out the near singularity at - + 1
as follows:
(sb') d' Cl
I)
+ (l) -(l)
I
-I
d'
1 + cos ('ir) 1 + cos (q"ir)
1 (c') - (1) = I
d'+(l)
I J0 1 + cos(4'ii)
J 1 + cos ('ir) Cl (ct') - (l) (l) = Id' +
1 + cos ('ii)This gives an expression for 00 as
1
(t)
-
(1)d'
+(i)
I<'
(15)oo()
= cos() +
1 + cos ('r)
Similarly, we subtract out the near singularity at -1 and get
1
-=
cos'()
+j
1 + cos (4'ir)d' +
(0) F1<1Care must be taken in numerically evaluating the integrand and the
square-root term because of the subtractions of nearly equal numbers. FrOm the
Riemann-Hilbert boundary conditions, the limits of 00 are
23
liiii
O()
= (1)++
1urn U = it .'- (0)
which are seen to be the correct limits of Equation (15).
When
+
such that > 1, r O, the conditions of theRiemann-1-Lilbert problem state that must have a real part T0 and an imaginary
part equal to () or it + fit + )\ =
T0()
+ or ) II>i-'-\
)/
Unfortunately the integral is now singular. We consider the loga.rithin term
first and as let
= -
sec ('ir)/
l/2\I
I
log i +
(2.i)
) -
log\-
sec ('ir) +sign('- 1/2)
(-I I l/2\\.
Re log + ç2_l)
))=
ln sec ('ir) + tan ('ir)11
= in
tan (T--') I
where we have remembered the sign convention that
2 1/2
( -1) - as - complex infinity
and used the trigonometric identity,
secO+tanO=tan(ir/4+U/2)
For the singular integral term we again make the change of variable
sec 'TT in the variable of integration and also in the limit value of
-3. Ti
II
> 1= -
sec (40trr)and then substitute into the integral form of the w0 expression shown in
Equation (14) sec
)2)
(16) us i ho ofi tan however,
C1
1/2 (- sec(oI)2_1)
01
cos (0tir)using the same square-root sign convention. This integral is singular when
'V = Subtracting out the singularity in the conventional manner leaves
1 (c1t) d
Jcos ('ir
[.1
[(t)(qot)] d4'
IJr
cos ('iT)lo
11-L
L c°s (0'ir) dOd'
cos (')
- cos 0 X=iir
/ 2 1,2 \ 1 I 1 '1 ((sec°o4
tan + 1 - sec 0021og
2[(sec
o_l)
J\(sec
001)
tan-i - 1 + sec 000
= ± cotan 0 [log(l) - log (-1)] = ± i cotan
Since for the real part T0 we are only interested in the real value
of the integral, we ignore this purely imaginary term and let
25 cos
1')
d'
1 + (- seccf'7T)
cos
+ (qo')d'
J
01
cos (qr) cos ("ir) 0 < 00 < 710<0<11
0 = (17) (18) =+ tan (07r) rmH)4)
16) e £ inConsequently the expression for is now
To.
)=
in tan (1r/4+ (Tr/2)o')ll + sin (q0tTr)where the square-root sign convention is reflected in the sign of the sine
term. This integral is nonsingular but again involves a difficult limit
as
' '
during the integration.
Changes to w1(). Again we are interested in evaluating the expression for
on the boundaries of the flow field, where the physical flow quantities
are of interest and which correspond to the real axis of the parameter. plane.
[(c-c) (c-c') (2l)]
1/2
Consider first the range
+
-3.E; n-'-O
C<<C'
and the change of variable'1
.2x
= c + (c'-.c) sin (-fwhich we make in both the square-root coefficient, using
x0
and in the2 1/2
variable of integration, using X. The term {(l-c)(-c')( -1)] as
-3.
;, n -'-
becomesC <
< C
whic 1/2 2 1/2 [1-c ] 26 [c- 8()] d
(-)
h-c)
(t)
(2)
1 [(q:')-(q0')] d4'cos ('ir) - cos ('ir) R C
'i', C
and c wher Re 1 1ü th I = tan-d'
(19) cos c'7r 1 cos
-)
nie.
With this change of variable
cl-c R
2
2 2 2
(c-c) (c'-) = R (1-cos x) (1 + cos x) = R (1 - cos. x)
2.2
= R sin X and consequently 2 1/2 [(-c)(c'-)(1- )] =where (x) = c + (c'-c) sin2 x/2. The integral expression transforms similarly
'ct-c' Re = T1(X0) 2 )
Sjfl
x0 P4-
2(x0) c' -c 2 (c-c) = R(1-cos x) sin X and (c'-) = R(l+cos x) fc'-c\ 2 ) sin x cl-c 2 x=o = . sinjXT
[_00(X))] d/dX
dx Ic'-c[(x)-(x0)]sin
2 (20)C'
-c = 2 [cos X0 - cos x]which results in an integral expression of the form iT
42(x)
$
a- ((x))] d 0 [cos x0 - cos x]1/j2(x)
27 (21) [i - cos X - (1-cos x0)]This' equation is singular when X = x0 and is of the same form as the
inte-gral Equation (19). Therefore, the singularity can be subtracted out directly leaving
T1(X0) = sin
'C
a-00((x))
ct-Oo((x6))cos - cos x
Finally we multiply through the square-root coefficient to give
E/1(xo))
a-00 ((x))]
El-x0fl
(l+(X)) ('-(x)) 0
cosx0-cosx
[a- 00 [a-00((X))] /((1+o)/(l-)
\
Tr8i(0)
= (0-c)(0-c')J
(l+(X))
0 ' (x)Similarly for the range > 1 we have
T(t)
=1
(0-c)(0-c')
$
[a-8((x))]V(
0 (x)sin2--sec (o') and
(x) = c + (c'-c)
=-2
Changes to a. When this change of variable is applied to the expression for a, Equation (13), it becomes
(22) (23) wh c Iig2 C
use
Chai hi where (24) To\(o-1)
)
to'(x)i
l-(X)
d (25)which is nonsingular but again involves a difficult limit as x+ x0 during
the integration.
For the range
+
, -1 < < C, Ct < < +1 fl+
0+ there is no
whe
singularity to subtract, and -'
Thu:
-where T = T
+ T1.
Thus
s(00') = Air
= Air
tan (ir/4+(ir/2)
esin (tTi) cos @'iT) dt
(i + a cos (t
i + b cos ('
i+ b' cos ('
which shows that
e((X)) dx
o
V l2(x)
-11 dX
SO
where
(x) = c + (c'-c)
sin2 4- .
This integral is nonsingular at both
limits and has no limit points.
The denominator of the expression has an
exact expression in terms of elliptic integrals, which, however, is not
used in the computer program.
Changes to s().
The expression for s in terms of
wass()
(J
+J)
eT(7)
29+ A d
I (c-a)
(c-b) (-b
')I
1 /2\
log
(+(2l)
)+ T*
(26)
O
Itan (/4+(/2) a')!
e
sec (ir) taii
'ir d'
oI(- sec (q'ir)-a) (- sec (4'ir)-b)
(- sec ('Tr) -
b') I
(27)
s()
++c211/2l)
e[TO*+Tl
4)
To convert this to the foil parameter
'let
sec
irT(E) = t((q')) =
to give
t(')
=5)
2)
tods(4')
-(i + a cos (q'ir)) (]. + b cos ('ir)) (i + b' cos (4'Tt))
These expressions are nonsingu].ar but again have a near singularity as a, b, b' at low incidence angles.
The normalized velocity magnitude on the foil is
_[T0*(t) +
q(') =
Icotan(ir/4+(iT/2)
4')I eand the normalized local pressure coefficient is
*AP@t)
= 1 - q2+
1 + c
Changes to Algebraic Equations. Since values of between c and c' are
represented by angles
x
such that= c
(c'-c)
sinit would seem logical to write the parameters, a, b, b' in the same manner.
Thus a Xa b Xb b' - Xbt where 2 Xa a = c
+ (c'-c)
sin --- etc.Then the algebraic equations transform into
2 1
XbXa\
fXa_Xb) /Xa4Xbt\A = - (a-b)(b'-a) =
(c'-c)
sin ( 2 ) Sfl I\ 2 sin 2 , Sin TI 0the B thprog
foun curv negli trip to Si vo1v trip one-i inte: the arefunc
of " 1 cad12 .
Greville, T.N.E., Theory and Applications of Spline Functions, Academic Press, Inc., New York (1969).
31 (XaXbT\ U (b'-a\ U
Sifl\
2 )S1flt
2 / - V b'-b) V fXbXb,\fX1X
2 1 2 (XalXb) sin(XX)
U(a_b)
2 d2 =- v
'Xb+xb) (xht-xb\Sifl(
2 21
Other material relating to calculation methods will be found in Appendices
B through D.
EXPLANATION OF THE COMPUTER PROGRAM
The problem described previously has been converted to a Fortran IV program for the CDC 6000 series digital computer system; the listing may be
found in Appendix B.
The use of discrete, digital representation for smooth mathematical curves always involves some loss of accuracy, which in general becomes more
negligible as the number of discrete points becomes larger. Because of the
triple integral in the expression for s@') in thi.s problem, it was decided
to store individual functions as tabular values, so that the functions
in-volved could be integrated successivel' without resorting to a cumbersome
triple integration. Therefore, a given function is stored internally as a one-dimensional array, and individual function values are developed by
interpolating from this array)2 The parameter integrations (integrals of
the form
C'
f() d
g(0)
=
.i:
c-;0for example)
are likewise accomplished by storing individual values of the integrand function f() as arrays of values and calculating a new integrand function
Th program is divided into main program sections and functions and requirc
subroutines. The main section of the program contains a master program In part
loop which computes the values 0:
s(4,').
The heart of the iterativerequir
process is to assume some starting function s(') and to use this to calcu-
previot
late a better value of s(i'). This process is repeated until the quantil
difference between successive s(4')-curves is less than some specified c
-amount. Then the foil shape, lift- and drag-coefficients, and cavitation proper number are calculated.
This problem involves six unknowns, namely, the algebraic values c,
b, a, b', c' and the normalized function s('). Similarly there are six
equations which relate them to the physical parameters of the problem: foil
shape (s/S); length of foil in a unit tunnel S, which is the inverse measure of the wall spacing; foil position in the tunnel h0; cavitation
number a which is directly related to the upstream velocity U; and V1
velocity of the downstream flows, which must be equal for the upper and
lower streams if the open-wake model is to b.e a valid approximation to the
real flow with a closed cavity. Examination of the Riemann-Hilbert mapping
shows that the parameters c + c' always occur in the order
plate
first
-1 < c < b < a < b'
< C' <
+1 of thc During the programing, we will arbitrarily "associate" each of theseparameters with a physical flow quantity
c a or U b' - S a - h0 b - [requirement that V = V ] upper lower c' - ci.
The function
s(4i') is
associated with the foil shape $(s/S). The directspecification of c, b, a, c'parameters and a function
sW)
will resultafter a single calculation in the output of a hydrofoil shape which has some combination of the previously described physical parameters.
)ng
es e
require more iteration from the program to search for them simultaneously.
In particular the specification of a required shape for the foil (s/S)
requires iteration for the function s(c')/S, during each cycle of which the prev&ously described algebraic quantities must be iterated, if physical
quantities are to be given as well. Besides the requirement that the
c -' c' parameters occur in the given order, it is possible to deduce other
properties of their behavior from the free streamline model. In particular
aOin°°stream
c'c
choked flow--a =
minimuinI
(00 cavity length) b-'.c
-b'--c'
S -'- maximum
ci.
+
0°00 -
stream
h0
- 1/2These relationships, as well as the curves of c, c', m for the flat-plate case, are used to select starting values of the parameters for the
first iteration. The order of the calculations performed during each cycle of the main program is:
a Keeping the starting value of c constant during the cycle, use
the starting s(') to generate (4,')
b Evaluate w0()
c Search for c' so that c equals the valUe required
d Evaluate w1(), based on these values of c
and c'
e Find a value of b consistent with the starting b' so that
't1(b) =
T1(b')
f Search for "a" such
that h0
= 1/2; calculate the arc length Sg Calculate a new s(')
h Adjust the new value of c, according to the value of in relation
to the desired value. If s(') has converged, generate foil
shape, lift
and
drag coefficients.33 c, b, a, b', c' -'- +1 b, a, b' -'-m a (1/2)(b + b') C, C :h e
CL 1.4
1.3
1.2 1.1 1.00.9
0.8
0.70.6
0.5
0.4 0.3 0.0 01Figure 4 - Lift and Drag Coefficients versus Cavitation Number
at Various Angles
02
03
0.405
a
Figure 4a - Lift Coefficient versus Cavitation Number, c'. = 20 Degrees
06
07
08
09
100.4
0.3 CD 0.2 0.1 0.0LIFT COEFFICIENT VS CAVITATION NUMBER
INCIDENCE ANGLE = 200
I,
27=160
CIRCULAR ARC/
4,
H/L
10,
V4.
FLAT
PLATE 6AA'
20 10H/L
-7,
ooU.
I.
4
a
a
a
U
a
U
U
U
U
U
U
U
LAT
ATE 0.2 0.1Figure 4b - Drag Coefficient versus Cavitation Number,
a = 20 Degrees 35 ;:i. DRAG COEFFICIENT I I VS I.
CAVITATION
I I NUMBER .1 INCIDENCE ANGLE = 20°1427
CIRCULAR ARCUI
aU
UUUUUUU
H/L
6 10 FLAT PLATE 20 U6-10.
-4 20
H/L
---
00U
0.4 0.3 CI,00
0.102
0.3 0.405
06
0.7 0.809
-100.9
0.8
0.7 0.4 0.3 0.200
0102
03
04
05
06
07
Figure 4c - Lift Coefficient versus Cavitation Number, c'. = 10 Degrees 36
08
09
0.05 0.04 0. LIFT COEFFICIENT VS ICAVITATION
I I I NUMBER I INCIDENCE ANGLE II
= 100ICIRCULAR
2'y16°
ARCH/L
/
20 10FLAT
PLATE 6'r'
20 (A,
'.3
0.6 CL 0.5 0.14 0.13 0.12 0.11 0.10 0.09 CD0.08
0.07 0.060.1 0.13 0.12 0.11 0.10 0.07 0.06 0.05 0.04
/
1
CIRCULAR ARC27=160
FLAT PLATE4jrH
4 /
H/L
6 /
20/
4P00
INCIDENCE ANGLE = 10°DRAG COEFFICIENT VS CAVITATION NUMBER
I
I I 1 I I00
01
02
0.3 0.4 0.5 0.607
0.8a
Figure 4d - Drag Coefficient versus Cavitation Number,
a = 10 Degrees 37
ARC
-o09
0.09 CD 0.080.30 0.28 0.26 0.24 0.22 CL 0.20 0.18 0.16 0.14 0.12
/
H/L
INCIDENCE ANGLE = 50LIFT COEFFICIENT VS CAVITATION NUMBER
00
01
a
Figure 4e - Lift Coefficient versus Cavitation Number,
c'. =
5 Degrees
3802
03
CL 0.28 0.26 0.24 0.22 0.20 0.18 0.16 0.14 0.12 0.10 0.0800
a
aa
a
a
a
a
a
a
a
a
a
a
a
ERaa
p3a
Figure 4f - Lift Coefficient versus Cavitation Number, a. = 3 Degrees H/L
3
6
1020
CoLIFT COEFFICIENT VS CAVITATION NUMBER
FLAT PLATE
INCIDENCE ANGLE 3039
0.28
0.26
0.24
0.22
0.20
0.18
CL0.16
0.14
0.12
0.10
0.08
00
0.102
03
U U -9% -8% -7% -6% -5% -3% -2% 0% +1% 0 01 40
rercentage Redaction of the Predicted Li ft and Drag Coefficients
Due to Tunnel Wall I roxiaiity for he Case lI/I. 1
02 03 04
Figare S - Percentage Reduction of the Predicted Lift an) Fr,
Due to Thenel lull Pronini to For the Cone Il/I.
-05 06 0.7 CAVITATION RUM Figure Sn - lift Co PLATE - PLATE o5' PLATE CIRCULAR 216 0_losARC
_J__
DS2 o = 4.2307 CL4 - C1 PERCENT LIFT I WALL EFFECT COEFFICIENT H/L4 I . ON I C1_, PLATE_._L_.1I
PLATE a=30 I I CIRCULAR 2n165 a205ARC I I I I . PLATE I'
/
0 OS2 0 0.2041061
UPPEFI FOIL SURFACES NOT
°6L'
TO SCALE
0.6 0.7 08 1)9 10 11 1.2 1.3
41 PLATE PLATE . a PLATE 10
---I I ---,-15 ---PLATE -C0 _-C0PERCENT WALL EFFECTON
-DRAGCOEFFCIENT H/L4 . -. - H--cD - ---. CIRCULAR ARC . I LOS 2 a4.2307' . . .-.1--. .
--..1
II
01 02 03 04 0.5 06 0.7 0.8 0.9 1.0 1.1 12 CAVITATION NUMBER Fig,r SI, - llrag Cofficicr,tFigure 6 - Choked Cavitation Number versus Incidence Angle
and Tunnel Wall Spacing at Various Angles
INCIDENCE ANGLE (DEGREES) Figure 6a - Incidence Angie
0.5 0.4
7
RCULAR 2 ARC ci 16° H/L = 8 PLATE////777
CHOKED CAVITATION INCIDENCE I NUMBER I I (BLOCKAGE ANGLE I CONSTANT) VS I I I 0 :0 FLAT CIRCULAR PLATE ARC 1969 DATAtEXrERIET
ROSHKO FROM WAKE PARKIN 1956_j MODEL WU I I I I I 10 15 20IIIIIIiIIiI
VITATION NUMBER (BLOCKAGE CONSTANT) VS TUNNEL SIZE I I I I I INCIDENCE ANGLE = 100CIRCULAR ARC 27 = 16°
E 2 4 6 8 10H/L
0.6 0.5 0.4 0.3 0.2
0
CHOKED CAVITATION NUMBER
VS TUNNEL SIZE (BLOCKAGE CONSTANT)
\
\
I I 1 I\
INCIDENCE ANGLE =20°
CIRCULAR ARC 27
16° FLAT PLATE 2 4 6 8 10 12 14 16 18 20H/L
1.0
'(IL
0.9
0.8
0.7 0.60.5
Figure 7 - Free Streamline Location versus Tunnel Wall Spacing Under Various Conditions
01
02
03
04
05
06
07
08
XIL
09
Figure 7a - Flat Plate, c = 90 Degrees
45 1.0 11
12
[1
FLAT UPPER PLATE FREE PERPENDICULAR 1 STREAMLINE - CHOKED TO TUNNEL STREAM FLOWI
THEORETICAL RAYLEIGHa=0 I I I WU19690=1.6 EXPERIMENTAL---TUNNEL TEST NSRDC 1964
10"-12"TEST SECTION - 2" FLAT PLATE L/H =
0.1794
..'
0
6"r4d
/
/
to
0.9
0.8
0.7.0.30 0.28 0.26 0.24 0.22 0.20 0.18 0.16 Y/L 0.14 0.12 0.10 0.08 0.06 0.04 0.02 0.0 - 0.1-I
STREAMLINE ABOVE A CIRCULAR ARC HYDROFOIL
I I I
.1
I I I F CAVITATION NUMBFDfl)
1°
INCIDENCE -ANGLE = 100H/L
_'Pp
TAil
411;
-A ..,,i,LIIUIIiWjwi 'I..AUJIh1/II111'4WI/I1IIhilhifl)h.
7I1II
UU4W45øZ9.
11'
DIRECTION OF FLOW 00 0.910
Y/L
06
0.7 0.8 0.0 0 1 0.2 0.3 0.4 0.5X/L
Figure 7b - Circular Ai'c Section, 2? = 16 Degrees, ci. = 10 Degrees, a= 0.2
I I I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
10
0.30 0.28 0.26 0.24 0.22 0.20 0.18 0.16YIL
0.14 0.12 0.10 0.08 0.06 0.04 0.02 0.0IOU
H1I!Iii1HIHhII
IUIIIIIIIIIIIIIHH
STREAMLINE. ABOVE A FLAT PLATE HYDROFOIL INCIDENCE ANGLE a = 100
CAVITATION NO. a = 0.2
H/L00
01
02
03
04
05
06
07
08
X/L
Figure 7c - Flat Plate, a = 10 Degrees, a = 0.2
47
k
H
CIRCULAR ARCH/L = 20U
2)' = 16°H/L=10
H/L =6
H/L = 4
H/L=20
H/L=10
FLAT PLATEUH/L=6
SUSUUK1
111111111
NORMALIZED PRESSURE COEFFICIENT VS CHORDLENGTH
-CAVITATION NO. a = 0.4
INCIDENCE ANGLE a = 10°-0.8 0.7 0.6 0.5
C +
a1 +0
0.4 0.3 0.2 0.1 0.0 0.0 0.1 0.2 0.3 0.4 0.5 0.60_i
08
09
10x/L
Figure 8 - Pressure
Coefficient
Distribution versusTunnel
Wall Spacing Incidence Angle = 10 Degrees, Cavitation Number = 0.4
1.0'
0.9
0.8
0.7
0.6
C +a
0.5
0.4
0.3
0.2
0.10
02
04
06
08
10
XILFigure 9 - Comparison of Pressure Coefficient Predictions from Wu's Theories and Larock and Street's
Theory for the Flat Plate in an Infinite Stream at Incidence Angle = 10 Degrees
CL CL
;IL
X/L
1969-FLAT ROSHKO PARALLEL.PLATE
NORMALIZED
PLATE
PRESSURE COEFFICIENT
a10° INFINITE STREAM
WU 1963 WAKE
LAROCK & STREET - TULIN
MODEL
SING LESPI RAL VORTEX
WU
LINE.
a = 0.1 = 0.4 = 0.27234 = 0.000305 = 0.48749 = 0.000495 CL = 0.27665 CL = 0.27241 XSIL = 0.000324 )çIL = 0.000316 CL = 0.56532 CL = 0.49350 XIL = 0.000932 X5/L = 0.000724\
N..
\
' ...,
S -.5.-
SS--5. -5 .5.-
-5-
.5-.5
-5-5-
1-a0
--.
a0.1
a0.4
-
-5-5.'\
0.12 0.10 0.08
Y/L
0.0.6 0.04 0.02
0
0.12
FLAT PLATE a = 100 INFINITE STREAM
/
UPPER FREE STREAMLINE
100
IN EVALUATiON
- RAYLEIGH a = 0 CHOKE
WU 1969 ROSHKO WAKE
MODEL a0
03
.
04
UPPER FREE STREAMLINE
05
06
X/L
ACCUMULATED NUMERICAL ERROR
Figure 10 - Compari.son of Streamline Locations
Flat Plate Infinite
Stream
c = 10 Degrees
Wu's 1969 Theory Roshko (or Transition)
Wake Model a' 0
01
02
0
07
08
.09
10
11FLATPLATE
UPPER FREEa10°
STREAMLINEINFINITESTREAM
ciO
o0.1
-J!I
= 0.4a = cii
a=02
a=0.4UU
o0
/
/
/
LAROCK VORTEX RAYLEIGH
WAKE a & STREET = 0 MODEL CHOKED TUL
aO
N SINGLE SPIRAL-
-A\/aroo
----0 0.102
03
04
05
06
07
08
09
10 11X/L
Figure 11 - Comparison of Streamline Locations
Flat Plate
Infinite
Stream
ct = 10 Degrees
Larock and Street Theory
Tulin
Single-Spital Vortex Wake Model
0.12 0.10 0.08
YIL
0.06 0.04 U, 0.02 0u0l
FLAT PLATE UPPER a FREE = 100 STREAMLINE INFINITE STREAM a 0 U = 0.2-a = 0.4 !hhh15 .-a = 0.4a0.2 p4
EXTRAPOLATED ci =oØ'
RAYLEIGH 1963 WAKE a = 0 MODEL CHOKEDWU
a0
oa-1O
---I.
01
02
03
04
05
06
07
0.8 0.9X/L
Figure 12 - Comparison of Streamline Locations
Flat Plate Infinite Stream a. = 10 Degrees Wu's 1963 Theory Wu 1963 Wake Model 1.0 1.1 0.12 0.10 0.08
Y/L
0.06 0.04 0.02 00.60 0.55 0.40 0.35 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50
CAVITATION NO. a
Figure 13 - Lift Coefficient versus Cavitation Number for a Flat
Plate Hydrofoil at c#. = 15 Degrees, Infinite
Stream--Comparison of Various Theoretical Predictions
53
LI FT COEFFICIENT VS. CAVITATION NO.
FLAT PLATE a=15° INFINITE STREAM //
1 /
/
,/
I/
1I,,,
/liv
/
-
a WU 1969 0.3670 0.3905 0.4188 0.4500 30 0.4831 0.5200 C.5564 0.5955 0.10 0.15 0.20 0.25 0.35 0.40 0.45 WU LAROCK 1963 & STREET 0.3685 0.3661 0.3964 0.3911 0.4288 0.4193 0.4652 0.4504 0.5053 0.4940 0.5489 0.5200 0.5957 0.5580 0.6452 0.5978 , /f
//
0 1/
1'
1"//
,' ,, /,/I
ri
,'I,
/
,/:/'
,,
/
,.,
/1
-,//
/
//1'
1969 THEORY ROSHKO (OR TRANSITION)
WAKE MODEL
WU1963WAKEMODEL
& STREET TULIN SINGLE-SPIRAL
VORTEX WAKE MODEL
LINEARIZEDWAKEMODEL 1966 SECOND-ORDER REF. 31
---WU1963THEORY
WU- - LAROCK
----SCHOT1971
- HSU
0.6 0.5 0.4 0.3
YIL.
0.2 0.1 00.1
I I STREAMLINES INCIDENCE ABOVE CAVITATION A 2y=16° ANGLE CIRCULAR = NO. 100 = 0.26 ARC H/L = HYDROFOIL 6 -PRESENT THEORY02
04
06
08
10
12
14
XIL
Figure 14 - Comparison of Predicted Streamline Locations with Experimental Data Circular Arc Hydrofoil, 2? = 16 Degrees,
a = 10 Degrees, H/L = 6, a = 0.26
16
18
Figure 15 - Comparison of Predicted Coefficients with Experimental Data 1.2 1.1 1.0 C-)
z 0.9
w
C) U-w0
C.) 0.8U--j
0. 0. 0. 0..T 0 1 0.2 0.3 0.4 0.5 0.6 0 7 0 8 0.9 1.CAVITATION NUMBER a
Figure iSa - Circular Arc Hydrofoil, 21 = 16 Degrees,
c = 10 and 20 Degrees
55
/ LIFT COEFFICIENT VS.
CAVITATION NO.
CIRCULAR ARC a =
100 & 20°2'yl6°
Iv-/ //
/ / // /
/
/1
/
a=20°
/
/
/
/
,.U
CAV CHOKED TATIONa=20°
uuumiiim
NO.U
/
/
r//
/y.
/
/.:
II./
/
'A
rA _.R
/
I, -THEORETICAL CURVES WU 1963 WAKE MODEL - IN WU 1969 ROSHKO WAKE MODEL WU 1969 ROSHKO WAKE MODELEXPERIMENTAL DATA 0 BASED ON MEASURED CAVITY PRESSURE Cl BASED ON FLUID VAPOR PRESSURE a1O° FINITE o e - PARKIN - HIL STREAM - INFINITE 6 a-20° -EXPERIMENT STREAM
-
-UUI
-/ CAVITATION
IiUUI
CHOKEDa=100
-'
NO.URU
1.1 1.0 0.9 0.8
-I
C) I-0.7 C.) L1 U. w0
'
0.6 I- U--J 0. 0. 0. 0.01
0.2 0.304
05
06
07
08
0.9 1.0 1.1 CAVITATION NUMBER aFigure 15b - Flat Plate Hydrofoil, c. = 10
and 20 Degrees
(*Data from TX. Wu)
56 1.2 1.1 1 .0 -J C)
I-zoc
Lii0
U-LI. Ui0
0
LI. -J 0.4 I I L I ILIFT
PLATECAVITATION
COEFFICIENT a 100 NO. VS. & 20°//jFLAT
0 = a 20° CHOKED a = 20° CAVITATION NO. 0UI
//
a=lo°7V
/
//
//
THEORETICAL CURVES L1969 ROSHKO WAKE MODEL - INFINITE STREAM
1969 ROSHKO WAKE MODEL - H/L 8
-EXPERIMENTAL DATA - PARKIN EXPERIMENT
a1O° a2O°
-a a -PRESSURE PRESSURE OBASEDONMEASURED CAVITY GBASEDONFLUID VAPOR WU WU/
CAVITATION I CHOKEDa=10°
I I NO. I-1.2 1.1 1.0
0
I.-0.90
U-LU0
C.) 0.8 U--J 0.7 0.6 0.5 0.4LIFTCOEFFICIENTVS.
CAVITATION NO.
FLAT PLATE a = 15°
-I
/0
I/
/
/
/
/0
O //
/
/
/
//
0 0/
/
/
/
0/
/
/
//
CAVITATION CHOKEDa=15°
NO./
/
/
THEORETICAL CURVE ROSHKO WAKE MODEL 1969 ROSHKO WAKE MODEL1971 LINEARIZED-INFINITE 1971 LINEARIZED -EXPERIMENTAL DATA a=15° PRESSURE
0
ON FLUID 0 PRESSURE WU1969 SCHOT GBASEOONMEASURED CAVITY aBASED VAPOR ---WU1963WAKEMODEL-INFINITESTREAM H/L - PARKIN - INFINITE - H/L = 8 (4 IN STREAM EXPERIMENT STREAM* 8* REF. TEXT) -WU/;"/
----SCHOT
,,/
,,
/
02
0.304
05
06
0.7 0.809
1.0 11 1.2 1.3 1.4 CAVITATION NUMBER aFigure l5c - Flat Plate Hydrofoil, a = 15 Degrees
(*Data from T.Y. Wu)