Re-Entrant Jet Modelling of Partial Cavity
Flow on Two-Dimensional Hydrofoils
Gert Kuiper MARIN and TU Deift
Jie Dang,
MARIN
Report 1151-P
Project Code; 963
7 - iO April 1998
Presented on the Third Symposium on Gavitatiön, Grenoble, France, Proceedings Volume i
Edited by f.M. Michel and H. Kato
TU Deift
Faculty of Mechanical Engineering and Marine TechnologyShip Hydromechanics Laboratory
Third International Symposium
on
Cavitation
APRIL 7-10, 1998
-
GRENOBLE, FRANCE
PROCEEDINGS
Volume i
Cavi
tation
April 7-10, 1998
GRENOBLE,
FRANCE
Volume 1
Edited by J.M. Michel
and H. Kato
98
SCIENTIFIC COMMITTEE
H. Kato, Chairman,
University of Tokyo, JapanJ.M. Michel, Chairman,
Centre National de la Recherche Scientfique, Grenoble, France
L. d'Agostino, University
of Pisa, italyF. Avellan, Ecole
Polytechnique Fédérale de Lausanne, SwitzerlandG. Bark, Chalmers
University, Got eborg, SwedenJ.R. Blake, University
of Birmingham, UKM.L. Billet, The
Pennsylvania State University, USAC. Brennen, Calfornia
Iñstitufe of Technology, USAJ. Field, Cavendish Laboratory, UK
D.H. Fruman, Ecole
Nationale Supérieure de Techniques Avancées, FranceB. Gindroz, Bassin
d'Essais des Carènes, FranceS. Kamiyama Tohoku
University, JapanA. Karimi, Ecole
Polytechnique Fédérale de Lausanne, SwitzerlandS.A. Kinnas,
Massachusetts Institute of Technology,USA
G. Kuiper, Marin,
The NetherlandsJ.T. Lee, Korea Research Institute of Ship & Ocean
Engineering,
Y. Matsumoto,
University of Tokyo, JapanS.S. Pan, China
Ship Scientific Research Center,R. Simoneau,
Hydro-Québec, CanadaB. Stoffel, Technische
HOchschule Darmstadt, GermanyLOCAL EXECUTIVE COMMITTEE
J.P. Franc, Chairman,
Centre National de laRecherche Scientifique, Grenoble, France
A. Archer, Electricité
de FranceBriançon-Marjollet, Bassin
d'Essais des Carènes, FranceP. Dupont, Ecole
Polytechnique Fédérale de Lausanne, SwitzerlandF. Jousselljn, University
of Grenoble, FranceT. Maître, University
of Grenoble, FranceMarchadier; University of
Grenoble1 FranceJ.M. Michel, Centre
National de la Recherche Scientifique, Grenoble, FranceJ. Pauchet, Centre National
d'Etudes Spatiales, FranceC. Pellone, Centre
National de la Recherche Scientfique, Grenoble, FranceM. Riondet, Centre
National de la Recherche Scientifique, Grenoble, FranceBubble dynamics in cavitation
Y. Matsumoto
University of Tokyo, Ja pan
CONTENT OF VOLUME i
INVITED LECTURES
3
Vorticity and pressure distributions in numerical simulations
of turbulent shear flows
9M. Lesieur
Institut National Polytechnique de Grenoble, France
The prediction of unsteady sheet cavitation
19S.A. Kinnas
University of Texas at Austin, U.S.A.
BUBBLE DYNAMICS
Details of asymmetric bubble collapse
39C.D. Ohi, O. Lindau, W. Lauterborn, A. Philipp
Technische Hochschule Darmstadt, Germany
Bubble formation by a corona discharge in dielectric liquids
45
F. Aitken, F. Jomni, A. Den at
Laboratoire d'Electrostatique et de Matériaux Diélectriques, Grenoble, France
Interaction of a cavitation bubble with a curved rigid boundary
51Y. Tornita
Hokkaido University of Education, Japan
J.R. Blake, P.B. Robinson
University of Birmingham, UK
Effects of bubble-bubble/bubble-wall interactions
on the nonlinear oscillations of bubbles
57H. Takahira, A. Yasuda
Osaka Prefecture University, Ja pan
Non-spherical motion of two cavitation bubbles
produced with/without time delay near a boundary
63K. Sato
Osaka University, Ja pan
Y. Tornita
Hokkaido University of Education, Japan
Spherical bubble dynamics : a model to predict red blood cell damage
by hydrodynamic cavitation
69S. Chambers, R. Bartlett, S. Ceccio
University of Michigan, USA
Numerical simulation of bubble dynamics in vortex core by the BUBMAC method
75KYan
China Ship Scientific Research Center
J.M. Michel
Laboratoire des Ecoulernents Géophysiques et Industriels, Grenoble, France
Prediction of Francis turbines efficiency alteration by travelling bubble cavitation
81C. Arne, F. Avellan, P. Dupont
Ecole Polytechnique Fédérale de Lausanne, Switzerland
Interfacial tension in water at solid surfaces
87N.A. Mortensen, A. Kühle, K.A. Morch
Technical University of Denmark, Lyngby
InteractiOn between cavitating btibbles and vortical structures
93G.L. Chahine, R. Duraiswami
Dynafiow Inc., Fulton Maryland, USA
Linear stability of Blasius boundary layers containing gas-vapor bubbles
99F. Burzagli
Cent rospazio, Pisa, Italy
d'Agostino
Università degli Studi di Pisa, Italy
NUCLEI AND INCEPTION
Effect of boundary layer tripping on the onset of cavitation in jets
109G. Shridhar, J. Katz, O. Knio
The Johns Hopkins University, Baltimore, USA
An attempt to predict cavitation level in turbulent jets by a
statistical approach
117Pauchet
GEC-ALSTHOM-ACB Grenoble, 'France
Influence of large-scale vortices on cavitation inception in steady, turbulent flows
.123R. Leucker
HOCHTIEF Consult, Essen, Germany
Cavitation by macroturbulent pressure fluctuations in hydraulic jump stilling basins
129RA; 'Lo pardo
Instituto Nacional del Agua y dei Ambiente, Argentina
On the cavitation of orifices in a pipe
(measurement of pressure reductions caused by vortices) ...
135Ogura
Liquid overheat upon
139V. Prisniakov
Dniepropetrovsk Sta te University, Ukraine
Nucleation and bubble dynamics in vortical flows
143REA. Arndt,
St. Anthony Falls Hydraulic Laboratory, Minneapolis, USA
B.H. Mames
Lockheed-Martin TAS, Fort Worth, USA
The influence of nuclei content on .the inception of bubble and vortex cavitation
149A.V. Chalov, V.P. Ilyin , Y.L. Levkovsky
Krylov Shipbuilding Research Institute, St.-Petersburg, Russia
Effects of dissolved gas on cavitation inception in free surface flows.
155T. Bau r, J. Köngeter,. R. Leucker
Aachen University çf Technology, Germany
On the cavitation occuring at the bottom of an accelerated circular cylinder
161H. Tanibayashi, K. Ogura, Y. Matsuura
Tamagawa LIn iversity, Tokyo, Ja pan
The experimental investigation of the cavitation inception
in the flow around NACA 16-018 hydrofoil
167L. Wilczynski
Institute of Fluid Flow Machinery, Gdansk, Poland
Tip vortex cavitation.: effect of the hydrofoil fractionation
173Viot, C. Jeanson, D.H. Fruman
Ecole Nationale Supérieure de Techniques Avancées, Palaisèau France
Method of similarity for cavitation
179L. Espanet, A. Tekatilan, D. Barbier
CEA Cadarache, France
H. Gouin
Laboratoire de Modehsation en Mecani que et Thermodynamique Marseille France
Numerical and experimental study of cavitating vortices in turbulent wake
185B. Belahadji, j.M. Michel
Laboratoire des Ecoulements Geophysiques et Industriels Grenoble France
HYDROFOILS AND PARTIAL CAVITIES
Three-dimensional characteristics of the cavities
formed on a twodimensional h.ydrofoil
191Kawanami, H. Kato, H. Yamaguchi
Uniziersity of Tokyo, Japan
Flow in the closure region of closed partial attached cavitation
. 197K.R. Laberteaux, S.L. Ceccio
University of Michigan, Ann Arbor, USA
Two-phase flow structure of cavitation:
experiment and modelling of unsteady effects
203J.L. Reboud, B. Stutz, O Coutier
Laboratoire des Ecoulements Geophysiques et Industriels Grenoble France
Influence of cavity thickness and pressure gradient
on the unsteady behaviour of partial cavities
209M Callenaere, J.P. Franc, J.M. Michel
Laboratoire des Ecoulenients Géophysiques et Industriels, Grenoble, France
Investigation of unstable cloud cavitation
215T.M. Pham; F. Larrarte, D.H. Fruman
Ecole Nationale Supérieure de Techniques Avancées, Palaiseau, France
Evaporation rate at sheet cavity interface
221Y. Kawatiami, H. Kato, H. Yarnaguchi, M. Maeda, F. Oshima, J. Doi
University of Tokyo, Ja pan
Cavitation inception and development on two dimensional hydrofoils
227P. Dorange, JA. Astolfi, J.Y. Billard, D.H. Fruman
Ecole Navale, Brest, France
I. Cid Tomas
University of Valladólid, Spain
Oscillating partial cavity on elastic hydrofoil
233Kovznskaya, E. Amromin
Krylov Shipbuilding Research Institute, St-Petersburg, Russia
Numerical simulation of 3D cavity behaviour
239Maître, C. Pellone, E. Collard
Laboratoire des Ecoulements Géophysiques el Industriels, Grenoble, France
Partial sheet cavities prediction on a twisted elliptical pianform hydrofoil
using a fully 3-D approach
245R. Hirschi, P. Dupont, F. Avellan
Ecole Polyt&inique Fédérale de Lausanne, Switzerland
PROPELLERS
Air injection as a means of reducing propeller cavitation induced ship vibration
253j.W. English
Maritime Technology, UK
Correlation investigations for higher order pressure fluctuations
and noise fòr ship propellers
259J. Friesch
Hamburg Ship Model Basin, Germany
Cloud cavitation - Preliminary classification of mechanisms
and observations on full scale ship propellers
267G. Bark
Propeller flow field analysis by means of LDV phase sampling techniques
273A. Stella, G. Guj
Università di Roma Tre
F. Di Felice
INSEAN, Roma, Italy
M. Elefante, F. Matera
CEIMM, Roma, Italy
A new algorithm fór numerical investigation of unsteady cavitating screw propeller
with use of variational approach
279A.S. Achkinadze, G.M. Fridman
St-Petersburg State Marine Technical University, Russia
A method to calculate characteristics of supercavitating propeller
making use of 2-D theory for cavitating wing
285Ando, S. Maita, K. Nakatake
Kyushu University, Ja pan
Prediction of screw cavitátion characteristics on the basis of model tests results
291'V.P. ilyin, A.A. Roussetsky, AV. Chalov
Kry!ov Shipbuilding Research Institute, St-Petersburg, Russia
Prediction of cavitation noise radiated from a marine screw propeller
295H. Kamiirisa
Akishima Laboratories, Mitsui Zosen Inc., Tokyo, Ja pan
New blade section design and its application
301J. Dang, D.H. Tang, J.D. Cheng,X.X. Peng, Q. Qin, S.S. Pan
China Ship Scientific Research Center
Design of supercavitating propellers for a racing boat
309T. Kudo, Y. Ukon
The Ship Research Institute, Tokyo, ¡a pan
FLUID MACHINERY
Turbine design for improved cavitation performance
317
Brekke
Norwegian University of Science and Technology, Trondheim, Norway
Study of hydrodynamic cavitation, in inducer centrifugal pumps
323V.V. Pilipenko, Y.A. Semenov, O.V. Pilipenko
Institute of Technical Mechanics of the NA'S of Ukraine
Experimental investigations concerning the influence of liquid property
and speed of rotation on the inception of blade cavitation in a centrifugal pump
329Awad, G. Ludwig, B. Stoffe!
3D cavity shape in an inducer:
experimental investigations
and numerical predictions
335F. Joussellin, T. Maître
Laboratoire des Ecoulements Géophysiques et industriels, Grenoble, France
P. Morel
Société Européenne de Propulsioñ, Vernon, France
Cavitation at low flow rates in
centrifugal pumps
341T. Okamura, Y. tie yama, T. Satoh
Tsuchiura Works, Hitachi Ltd. Ibasraki, Ja pan
Linear analyses of cavitation instabilities
347Watanabe, Y. Tsujimoto
Osaka University, Japan
J.P. Franc, J.M. Michel
Laboratoire des Ecoulements Géophysiques et Industriels, Grenoble, France
A simple prediction method
of pump cávitation performance
353Tsugawa
Kobe, Ja pan
Would turbine uprating be
allowable with respect to cavitation?
An example of vibroacoustic
diagnosis
359B. Bajic
Brodarski Institute, Zagreb, Croatia
Development of long-lived type inducer for multi-stage pump
363M. Maeda, M. Tagawa
International Symposium on Cavitation,
Grenoble, April 1998.
Re-Entrant Jet Modelling of Partial Cavity Flow on
Two-Dimensional Hydrofoils
Jie Dang
Maritime Research Institute,Netherlands P.O.Box28, 6700 AA Wagen ingen, The Netherlands
Telephone: +31-317-493223 Fax: +31-317-493245 Email: j.dang@marin.nI
Gert Kuiper
Maritime Research Institute, Netherlands and Delfì University of Technology P.O.Box28, 6700 AA Wageningen, The Netherlands
Telephone: +31-317-493273 Fax: +31-317-493245
Email: g.kuipermarin.ni
Abstract
A potential based panel method is developed to predict the partial cavity flow on two-dimensional
hydrofoil sections The Dirichiet type dynamic boundary condition on the cavity surface and the
Neumann type kinematic boundary condition on the wetted section surface are enforced A re entrant Jet cavity termination model is introduced A validation is accomplished by comparing the present calculations with cavitation experiments of a modified Joukowsky foil and a NACA 66(MOD) a=O 8
Re-Entrant Jet Modelling of Partial Cavity Flow
on
Two-Dimensional Hydrofoils
Jie Dang
Maritime Research Institute, Netherlands P.O.Box28, 6700 AA Wageningen, The Netherlands
Telephone: +31-317-493223 Fax: +31-317-493245 Email: j.dang@marin.nl
Abstract
A potential based panel method is developed to predict the
partial cavity flow on two-dimensional hydrofoil sections. The
Dirichiet type dynamic boundary condition on the cavity
surface and the Neumann type kinematic boundary condition on the wetted section surface are enforced. A re-entrant jet
cavity termination model is introduced. A validation is
accomplished by comparing the present calculations with
cavitation experiments
of
a modWed Joukowskyfoil and aNACA 66(MOD) a=O.8 section.
Introduction
Cavitation on ship propeller blades is a major source of
noise and vibrations. In order to prevent excessive noise and vibrations cavitation has to be eliminated entirely orhas to be controlled in its behaviour. Since for reasons of efficiency the
propeller always operates in the wake of the hull, which is
highly non-uniform, cavitation can generally not be avoided. The increasmg speed and power installed in ships nowadays make the problems worse. It is therefore important to control the dynamic behaviour of cavitation.
Cavitation on a ship propeller is three-dimensional and
unsteady. The inflow variations into the propeller have a low
frequency, however Although the bläde loading is highly
unsteady and has to be calculated as such, it seams acceptable as a first approximation to consider the cavity as quasi-steady.
Only few authors have investigated the three-dimensional
characteristics of cavity flow (eg in [1,2]). Most attention has been given to two-dimensional cavity flows. Excellent reviews of the research on cavity flows have been given by Wu131 and Uhlman141. The classic linear solutión of cavity flow around a hydrofoil by Tulin5, Wu161 and Geurst'71 was modified by
Kinnas18' to predict the leading edge partial cavity flow.
Different from traditional linear method, this method predicts the cavity length to decrease with an increase of foil thickness.
A systematic investigation from 2 D foil to 3 D propeller
cavitation has been performed recently at MIT121191°1. A so-called split panel technique is used to avoid re-panellmg of the cavity-foil surface At the end of the cavity, a simple algebraic
expression of the pressure recovery is empirically enforced
over a given range of the cavity length at the end of the cavity
in the method But this approach of artificial recovery may
influence the final results. Also the detailed flow structure at
the trailing edge of the cavity is ignored and no insight is
Gert Kuiper
Maritime Research Institute, Netherlànds and Deift University of Technology P.O.Box28, 6700 AA Wageningen, The Netherlands
Telephone: +31-317-493273 Fax: ±31-317-493245
Email: g.kuiperniarin.nl
noise by its implosion. The re-entrant jet also influences the volume of the cavity. In this paper a method is developed to
numerically predict the re-entrant jet at the end of a
two-dimensional cavity and to assess its effect on the cavity shape
and volume.
In our investigation, a potential based lower order panel
method is used. A re-entrant jet cross section.is introduced as a
boundary of the problem, on which surface a normal velocity
into the cavity is prescribed which equals the free-stream velocity on the cavity surface. A Dirichlet type dynamic
boundary condition is enforced on the cavity surface, while
Neumann boundary conditions are enforced both on the jet
boundary and on the wetted surface of the foil
sectionrespectively. An initial cavity surface is firstly assumed and then iterated. The kinematic boundary condition on the cavity surface is satisfied by iteratiñg the cavity length and:shape. Upon convergence, both the dynamic boundary condition and
the kinematic boundary condition on the cavity surface are
satisfied and a re-entrant jet with a certain thickness
isautomatically formed.
Mathematical Formulation
Expression of the problem
Consider a cavity flow around a two-dimensional foil
section in an unbounded fluid as show in Figure 1. The inflow V0 has an angle of incidence a to the nose-tail line of the foil section. A steady partial cavity is formed on the surface of the
foil when the ambient pressure is low enough so that the pressure on some part of the foil surface is lower than the
cavitation inception pressure, which can be taken as the vapor pressure. A re-entrant jet is formed at the end of the cavity and flows inward to the cavity with the same speed as the speed on the cavity surface. Let us suppose that the re-entrant jet flows through an intersection surface of the jet (S in figure 1) and disappears so. Then the flow around the cavity-foil system can be treated as an inviscid fluid flow field with a total potential '»which satisfies the Laplace equation,
V2(1)=0,. (I)
If we define a perturbation potential to the inflow potential
at infinity (D0 as q, then, where,
si
f Sr
Figure 1. Partially cavitating hydrofoil with a re-entrant jet The universal solution ofequation (4) can be obtained by
Green's identity on a
closed boundary.Since the two
dimensional flow field around the foil section is not a simply connected zone, a cut has to be introduced to connect the foil trailing edge to infinity. A boundary condition should also be enforced on this cut. Then we have,
2,r = fço -- In
r
ds -f_!in
r
ds (5)s
where,S=Sh+Sc±S;v+Sj,q is a point on the boundary S and p ¡s a point in the field
Instead ofdeveloping a singularity distribution code based
on the induced velocity, we developed a potential based panel code subject to a mixed boundary condition ofDirichlet and
Neumann. When equation (5) is applied on the Neumann boundary conditions, a Fredholm integraI equationof the
second kind is obtained. But when equation (5) isused onto
the Dirichlet boundary condition, we obtain a Fredholm
integral equationofthe first kind.
Boundary conditions
The kinematic boundary condition on the surface ofthe
cavity and the foil are as follows,
ñV=O
onSbandSC (6)Substituting equation
(2) and (3)
into (6), a boundarycondition for the perturbation potential is found,
ûn_.n. o
on Sb and S (7)where, ñ denotes the outward normal unit vector both on the foil surface and on the cavity surface.
A dynamic boundary condition is also needed on the cavity surface which states the pressure on the cavity surface to be
constant for the steady cavity flow and equal to the vapor
pressure Pc
(8)
While from the Bernoulli's equation, we have,
Po
I-.
P'o
-
pc+plÇ2
I
(9)
where, V denotes the constant velocity on the cavity surface
and po denotes the pressure at 'infinity. Define a cavitation
numberas,
_PoPC
°)/pv2
then, =
In order to prescribe easily this velocity on the cavity
surface, an alternative expression is used. Consider a curved coordinate s
along the cavity surface
starting from thedetachment point o of
the cavity near the leading edge
towards the end of the cavity. Equation (Il) can be rewritten
as;
(12)
where, Ï denotes the tangential unit vector on the cavity
surface alongs. Then,
ôs
=ï+V0V1+a
onS.
(13)Integrating from the detachment point to any points on the
cavity, we get the prescribed potential on the cavity surface
based on the potential at s0,
onS (14)
where is the perturbation potential at the detachment point.
On the jet cross section S, only the kinematic boundary
condition is needed, which is,
(15)
where, ñ is the normal unit vector on the jet boundary toward the fluid field For the perturbation potential, we have,
1jT47.n
onS3 (16)On the wake surface Sfl,, the velocity is considered to be continuous while the potential has a jump across the wake, but
is kept constant along the wake from the trailing edge to
infinity. It is:expressed in the perturbation potential as,
ûn - ûn
on S,, (17)L'=Ç: q
onS,, (18)where, A is the potential jump across the wake surface and
superscripts + and - denote the variables on the upper and
lower surfaceofthe wake, respectively.
A Kuna condition is enforced at the trailing edge to keep the velocity at the trailing edge finite.
At the detachment point
of the cavity, a so-called'Brillouin-Villat condition is always demanded in potential
flow, which states that the curvature of the cavity surface at the detachment point must be continuous with the curvatureof
the foil surface at the same position. In real flow, it has been found e. g. by Shen & Peterson'1 that the detachment point is always a little bit downstream ofthe negative pressure peakof
the wetted flow or starts from the separation point of the laminar boundary layer near the leading edge. lt is still
difficult, up 'to present dày, to treat the detachment point
precisely in numerical simulation. For simplicity, the
detachment positionofthe cavity is treated as an input for this
method.
Substitute equation (17)and (18) into equation (5) we get,
7rÇO =
J
ço-_lnrpqds
f
--lnr,,ds
o
Sb+S+SJ S+S,+Sj +fitçof-lnrpqds
(19) sThen equation (19) can be solved under the mixed
boundary conditiOnsof(7), (14) and (16) for different parts ofthe boundary together with a Kutta condition at the trailing
edge. The uiiknown potential or the normal derivative of the potential on the boundary is obtained.
Instead of solving the equations by giving a fixed cavity length and calculating the corresponding cavitation number, a
more direct way is used in the present method, which
prescribes a cavitation number but solves the equation by
iterating the cavity surface until it is converged.
Iteration scheme for the cavity surface
An estimated cavity length and shape, and a re-entrant jet cross section boundary are first assumed at 'the beginning of
the calculation. The dynamic Dirichlet boundary condition
(14) is imposed on the cavity surface, and the kinematic
(10)
on S. (Il')
and,
CL_ (28)
pVcC
D (29)
pV02C
are lifting coefficient and drag coefficient respectively. Here, C is the chord length of the foil.
The cavity volume coefficient is.defmed as,
c, = V/c2
(30)Numerical Implementation
Discrete expression of the problem
Instead of using the higher order panel method, we used the lower order panel method with constant source and dipole
distributions on every panel, because the solution always approaches the analytic one when the number of panels approaches infinity. For a two-dimensional problem, CPU
time consuming is not a considerable problem.
The foil, cavity and jet surfaces are represented by N
straight line panels. The control points are located at the center
of each panel. The constant source and dipole strength
distributed on the1th
panel are () and
,. respectively. Thus equation (1.9) can be expressed by the following discrete form
for every control. point i,
=A11qi1
B11()1 +C1Aço
1=1 j=l
¡=1,2,3...N (31)
where Nis the total number of panels on the boundary andA11,
B1 and C, are induction coefficients of thé dipoles and the sources on the surface and the induction coefficienti of the.
wakedipoles respectively, A. = i
lnr. .
ds 1,1 J rê
(32) (Sb +S +S).B,1 =
fIn r,1.ds (33) (Sb +S+Si)-and,C,=J-Slnrjds
(34)where (Sb+S*SJ)Jis the boundary of thejth panel on 5b,S and
Si.
The kinematic boundary conditions of equations (7) and (16) prescribe the source strengthon the wetted part of the foil
and on .the jet boundary respectively, while on the cavity surface the dynamic boundary condition of equation (14)
prescribes the dipole strength on its surface The unkÌowñs are
the source strength on the cavity surface and the dipole
strength on the foil surface.
There are different approaches. to the implementátion of the. Kutta condition 23I for steady and unsteady flows. In the
present work only the simplest numerical Kutta condition
given by Morinot'41 is chosen For two dimensional cases, this condition states that the wake dipole strength should be equal
to the difference of the dipoles. on the upper surface and on the lower surface at the trailing edge. It is written as,
Aç =
-
(35)th
jet. Only the kinematic boundary condition on the cavity surface is not satisfied at this moment. The potential on the
wetted part of the foil and on the cross section of the jet, and. the normal derivative of the potential on the cavity surface are calculated by solving equation (19). This calculated normal derivatives of the potential on the cavity surface are usually
not equal to the value prescribed
in equation (7.). Thedifference AV,,
AV_19
ôn
ôq
¿9ço (20) req. cal.where, subscriptreq. and caL denotes the value prescribed by equation (7).and the calcu ated value respectively.
Our aim is to re-align the cavity surface to make the flow on the cavity surface tangential to its surface.
d Ai7 AP', AV,
(21)
ds V v0-.Ji
where, ,j is the direction orthogonal to. the cavity surface s.
Then,
s
AV
A ì
-
J
j---
ds (22)where Ai = O when s = s.
The jet thickness of S is the distance between the end
point of the cavity and the foil surface When the end point of the cavity moves to a new location after an iteration, S is also moved to this location.
Cavity volume and hydrodynamic forces
The most important parameter of the cavity is the volume. of the cavity. The accelerations of the cavity volume are the source of pressure fluctuations on the ship's afterbody. If the cavity surface is described by y=77(x), and the foil surface is described by y=J(x), then the cavity volume is obtained by the following integration,
V
=J[ii(x)
- f(x)Ix
(23)Xd
where, x,, denotes the position of the cavity detachment point and x,, denotes the position of the maximum cavity length (see Figure :1). The volume occupied by the jet is not considered in the calculation since it is verythin.
The lift L and drag D on the foil section is calculated by integrating the pressure all over the surface. We have,
=
-
J pn, dl
(24a)F=_Jpndl
(24b)and,
L=.F cosa-1 sina
(25)D=Fsina+Fcosa
(26)where, n and ny are the two components of the. outward
normal unit vector on the foil surface and ¡ is the arc length
along the foil surface, respectively. The pressure, under the cavity but on the foil surface between the detachment point
and the .posit ion of the jet cross section, is set toPc
a
priori inthis force calculation.
The following non-dimensional coefficients are also
Øfl M
ybpJl
tr Qt45 pf,S3fÇ1 4tJ1panel just ahead of the detachment point of the cavity as
described in equation (14) by . Suppose the cavity starts at
panel N, and ends at panel Ne, the integration of equation (14)
is written numerically as a sum giving by the following
expression, = k Na Ne where, AS'k =(1k ''k_Ê) (37)
and 'k is the length of the kth panel.
Because of the mixed type boundary conditions, the linear
equations (31) are composed of both Fredholm equations of the second kind and the Fredhoim equations of the first kind. The matrix is no longer a diagonal dominated. Further more,
equation (36) makes the condition of the matrix even worse. A more accurate direct solver ¡s needed.
The kinematic boundary condition on the cavity surface is
used to iterate the cavity length and the cavity surface shape. The numerical form of the equation (13) is as follows,
AV,,
k=Nd...Ne (38)
where, ¿tSk is given by equation (37). The surface correction is made by changing the coordinates of the control points by z1i,
in the normal unit vector direction for every panels on the
cavity surface. The end points of panels are interpolated from
the coordinates
of the
control points, except for thedetachment point, which is actually given as input, and the iñtersection point of the cavity and jet which will be described
in the next paragraph. This numerical scheme can prevent the cavity surface from becoming saw-tooth shaped.
lt has been found in reference [I] and [15] that the re-entrant jet is hardly formed before the cavity is fully developed. lt means that the re-entrant jet will not form when the cavity is still growing. Our calculation shows the same phenomenon. At the beginning of each calculation, after the length of the cavity is assumed, a re-entrant jet surface with very small height is firstly erected vertically to the surface at this point and the cavity thkkness distribution is assumed to
increase from the detachment point to the jet linearly. If the cavity length is under estimated, the cavity will grow up very quickly over the jet boundary. 1f the end of the cavity surface goes into the foil surface, the program simply truncates that part. But if the end of the cavity is above the foil surface, the jet boundary is set there to connect this end point vertically to the foil surface. No more restrictions are enforced and the
re-entrant jet can evolve automatically. Surface panelling
The surface of the cavity and foil system is divided into
small panels Around the leading and trailing edge and around
the reentrant jet of the cavity, the cùrvature of the surface is quite large. Especially at the end of the cavity and at the thin
re-entrant jet, very fine grids are needed to obtain a converged result. The following panelling is adopted.
On the lower surface of the foil, a cosine distribution of the panel from the leading edge to trailing edge is used,
(il)
x. = 05 OScos
ir
¡=1,2,3...N1+1 (39) where, N, is the total panels number on the lower surface of the foil.the panels on the cavity surface are used for the last 5% of the arc length on the cavity surface.
.s; =0.95(0.5-05cos
2ir)
cae
s, = 0.95+ 0.05
N N /2
iN0J2...Nc+1(41)cae cae
where, is the total panel numbers on the cavity surface.
(36) On the foil surface downstream of the cavity, the surface is
also divided into two parts,,
NACA1 6-006
Re-entrant Jet ModeF o=0.8751 3 re-entrant jet cavity surface
t-. t
t t foil surface 0.56 0.57 0.58 ¡=1,2...Ncr,/2 (40) 0.59 0.60 0.61 x/CFigure 2. Surface paneling around the re-entrant jet surface
and the cavity end. (NACAI6-006 section at anangel of attack 40, and cavitation number 0.875 13)
Results and Discussion
Convergence test
It is certain that when the number of panels increases to
infinity, the solution converges to the correct solution
described by the analytical equation of the problem. In order
to make sure how many panels are the minimum requirements to reach a reasonable result,, the convergent test of the method was done. We took NACAI6-006 at an angle of attack 4° and
a cavitation number of 0.87513 as a test case. Figure 3 and
Figure 4 show the test results.
It is obvious that the solution is quickly converged with the increase of the panels. When both of the number of panels on
the cavity surface and on the foil surface downstream of the cavity is increased to 80, the cavity length and the re-entrant
jet thickness are converged. So, 100 panels on the cavity
surface and 100 panels on the foil surface downstream of the
cavity are recommended. A total of 260 panels all over the cavity-foil system are needed if the number of panels for the
jet %9Rsh9face, two sets of panels are used. Half of
X, = 0.1
NdOw,, / 2
i1
¡=1,2...Nd0,,.,/2 (42)
and,x. = 0.1 + 0.9(0.5 - 0.5 cos - NdO,,,, / 2
dow,, - down
where, Nd0,,,,, is the total panel numbers on the foil wetted part
downstream of the cavity Figure 2 shows an example of the
panel arrangement around the re-entrant jet under the above scheme of panelling. 0.07 y/C 0.06 005 0.04 0.03 002
pro&ffi4F o
A006 §ttbntIàÇ aii1eÇlNfl41!
4° for two different cavitation numbers respectiely. lt is found that the convergence is slow toward' the final result.
From our experience, although the steps needed for the
iteration depend on the initial assumption of the cavity loo steps are always needed to achieve a result with a maximum
error, between the calculated velocity on the cavity surface
and the prescribed freestream cavity velocity, less than 1%. The maximum errors 'always occur at the intersection of the
cavity and the re-entrant jet. But on 'most part of the cavity
surface, the tangential velocity is already exactly equal to the prescribed velocity.
The influence of the detachment point on the final cavity
volume and cavity length has been well investigated by
Uhlman in reference [4]. Since we have not found a good way to. treat the position of the detachment point, a detachment is always set at the leading edge in all of the calculations in the following paragraphs. 1.0 - -I/c 0.9 0.8 0.4 40 50 60 70 100 110 "down
Figure 3. The cavity length varies with thepanel ñurnbers
0 .0060 h/C 0.0055 0.0050 0.0045 80 90.
Figure 5. Changes of the cavity shape with the iteration steps.
(NACA 16OO6 a=4°, YO.87513, linigja/C=O.37550, converged l/C=O.59279) 0.20 0.25 0.1 0.0 -0.1 -0.2 -0.3
00 01
02 03 04 05
0607 08 09 10
'-'CFigure 7. The final converged cavity shape With. re-entrant jet. (NACAI'6-006, ct=4°, cr=O.875l3) 0.5 0.0 -0.5 -1.0 y/C ÑACA16-006 0.2 o0.87513
a4°
Re-entrant Jet Model
NACA1 6-006 Re-entrant Jet Model
c=0.87512 a4° 1.0
i
'00 01
02 03 04 05
0.607 08 09
10x/C
Figure' 8. Pressure distribution on the cavity-foil system
Predictionsfor'NACAI6 series sections
A calculation for the cavity flow with a re-entrant jet on the NACA 16-006 sectionatan angle ofattack of4° is carried out at a cavitationnumber of 0.87513. This cavitation nUmber is chosen, because the authors wanted 'to makea comparison
with the result of Uhlman's cavity termination wall model. The converged shape of the cavity and its re-entrant jet is
shown in Figure 7 and the fmal pressure distribution on the cavity and on the wetted part of the foil is drawn in Figure 8. Detailsabout the re-entrant jet are shown in Figure 2 with the
panel arrangement.
Figure 8 shows that the velocity on the cavity surface
equals to the prescribed freestream velocity. lt means that both the dynamic boundary condition and the kinematic boundary condition on the cavity surface are very well satisfiéd. Only at the control point of the last panel on the cavity, the velocity is a little bit higher than the freestream velocity, but the relative error is still less"than 1%. A stagnation point is clearly shown downstream the end of the cavity.
The flexibility of the present program makes it very easy to change the kinematic boundary condition on the jet section from equation (16) to a non-penetrative condition similar to equation (7) and find the solution with a cavity términation wall' model. Figures 9 and 10 show the cavity shape and'the pressure coefficient distribution on the cavity-foil surface with
this cavity model respectively. A stagnation' point is also
shown at the corner of the wall:. The difference of the
predicted cavity shapes by these two cavity models is shown
more clearly in Figure 11 and' Figure 12 for two cavitation 0.0040
40 50 60 70 80 90 100 110
Figure 4. The jet thickness varies with the.panel numbers
1.5
-cp
0.7
0.6
u.., y/C 0.2 y/C y/C 0.1 0.0 -0.1 -0.2 -0.3
00 01
02 03 04 05 06 07 08 09 10
ilCFigure 9. The finally converged cavity shape with termination
wall model (F ACAI6-006, czr=4°, c=0875 13)
1.5 -cp 1.0 0.5 0.0 -0.5 -1.0 0.3 0.2 0.1 0.0 -0.1 -0.2 -03 0.3 0.2 I 0.0 -0.3
NACA16-006 Terrnination'Wall Model a=0!87513
NA A16.006 -mina
c0.E
a4°
7512ion Wall Mdel
00 01
02 03 04 05 06
0.708 09 10
.rJC
Figure 10. Pressure distribution on the cavity-foil system
NACA1 6-006 i Wal Model Re-entrant J
leni
inatio t Modela4°
00 01
02 03 04 05 06 07 08 09 10
xJCFigure 11. Comparison of the cavity shapes for the different cavity models.
NACA1 6-006
Term nationWall Re-e
MOdl\>/
itrantJetMa=08
odel
7513
calculations for NACA 16 series sections with different
thickness to chord ratio to study the relationships between the cavity volume, the cavity length the cavitation number and the re-entrant jet thickness. All of the results are listed in Table I
to Table 3.
Table 1. Calculated Results for NACAI 6006'ata=4°
Table 2. Calculated Results for NACA16-009 at ct=4°
Table 3. Calculated Results for NACA 16-012 at ct=4°
The special feature of the present method is the re-entrant
jet calculatiOn. Figure 13 shows the re-entrántjet thickness
versus cavitation number for three sections. It can be found in
this figure that the re-entrant jet thickness decreases with increasing cavitation number. And also ound is that the
re-entrant jet is thicker on a thinner profile (tJC=6%) and thinner on athicker profile (t/C=12%).
But when we plotted the ratio of the re-entrant jet to the maximum cavity thickness against the cavity length, we got Figure 14. It shows that the re-entrant jet thickness is always a
certain percentage of the maximùm cavity thickness
m,
irrespective of the cavitation number and the profile thickness. This percentage is around 8% to 10%. It is also found that the maximum cavity thickness is also always located at 60% of the total cavity length (Figure 15 andX,max/Cin Table 1 to 3),
irrespective of the cavitation number and the profile thickness as well. These characteristics of the re-entrant jet could also be
found in Gilbarg's calculation of the re-entrant jet after an
obstacle61 0006 h/C 0.005 0.004 0.001 O N°CA16-0C6 -O-- N°CA16-009 N \CAI 6-012
a40
00 01
02 03 04 05 06 07 08 09
10 0.000 xJC 06 08 10 1.2 141618
Figure 12. Comparison of the cavity shapes for the different
cavity models. Figure 3. Cavity re-entrant je.t thickness WC versus cavitation
nune?&am
analysis isperformed by a
series ofi/C
i/C xJC
Cv h/C CL CD 0.83 0.7324 , 0.06272 0.4452 0.03496 0.00549 0.6252 0.0270 0.85 0.6624 0.05737 0.3897 0,02888 0.00448 0.5920 0.0241 0.87513 0.5928 0.05197 0.3464 0.02338 0.00424 0.5614 0.0220 0.90 0.5355 0.04767 0.3124 0.01936 0.00384 0.5383 0.0206 lOO 0.3917 0.03721 0.2319 0.01106 0.00300 0.4940 0.0179 1.10 0.3066 0.03082 0.1866 0.00717 0.00275 0.4752 0.0164 1.20 0.2460 0.02607 0.1487 0.00487 0.00234 0.4635 0.0153 1.30 0.2070 0.02318 0.1220 ' 0.00365 0.00225 0.4562 0.0142 1.40 0.1724 '0.0201i 0.1014 ' 0.00264 0.00207 0.4498 0.0133 iSO 0. 1454 0.01757 0.0853 ' 0.00195 0.00185 0.4445 0.0125 1.60 0.1241 0.01549 0.0727 0.00147 0.00165' 0.4412 0.0118 1.70 0.1013 0.01378 0.0628 0.00113 0.00154 ' 0.4384 0.0112 . VCtJC xJC
Cv WC C1 C0 0.81 ' 0.6998 . 0.04799 0.4212 0.02516 0.00339 0.5581 0.0160 0.90 ' 0.4748 0.03407 0.2841 0.01202 O.00244 0.4957 0.0117 LOO 0.3376 002593 . 0.2088 000650 0.00180 0.4622 0.0094 1.10 0.2629 0.02117 0.1613 0.00412 0.00173 0.4606 0.0087 120 0.2019 0.01683 0.1234 0.00251 0.00136 0.4522 0.0076 1.30 0.1579 0.01367 ' 0.0970 0.00159 0.00118 0,4368 0.0057 1.40 0.1271 0.01122 0.0778 ' 000105 0.00103 0.4361 0.0050 1.50 0.1029 ' 0.00915 0.0640 0.00069' 0.00088 0.4349 0.0044 VCiJC x/C
Cv WC C'L 'C0 0.80 06544 0.03373 0,3982 0.01606 0.00226 05037 0.0076 0.90 0.4261' 0.02348 0.2608 0.00719 000148 04714 0.0051 1.00 0.2843 0.01595 0.1823 0.00321 0.00100' 0.4575 0.0040 '1.10 0.1821 0.00994 0.1195 0.00125 0.00082 0.4260 0.0015 i20 0.1258 0.00652 0.0082 0.00054 0.00048 04260 0.0008 0.003 -0.1 0.002 -0.21.0
x,/I
0.8 0.6 0.4 0.2 0.0 O NA A 16-006 -n- NAiA16-009FigUre 14. Thç re-entrantjet thickness as apercentage of the cavity maximum thickness versus the cavity length
a4° -o- NACA16-oo9O NACA16-006
' NÄCAI6-012
00 02 04 06 08 I/C 10
Figure 15. The maximum cavity thickness position
Comparison With other linear and nonlinear results
In order to find out the difference of the prediction results between the present method and those of other linear methods
with thickness correction and the nonlinear methods with
termination wall model, comparisons are made for NACA 16 series sections at an angle of attack 4° for different cavitation
numbers.
Uhlmant41's method is
based on velocity. A cavity
termination wall model is employed in his method. The cavity length is prescribed, while a cavitation number is calculated.
When the shape of the cavity is converged, the cavitation number is obtained. By assuming the velocity on the jet
boundary to be zero in the present method, we can also obtain the solution with a termination wall model. A comparison is shown in Table 4. It can be said that the results are very close to each other even if some differences exist for some valùes. The cavity length predicted by present method is 2% longer than the length predicted by Uhlman.
Table 4. Comparison of different approaches for the.same cavity termination wall model
correction gives very good results comparing to the present
nonlinear method for NACAI6-006 but over-predicts the
cavity length for NACAI6-Ol2. The present method predicts a cavity length decrease with increasing of thickness while the linear method predicts a cavity length increase with increasing foil thickness. Since the tendency is different, we can not say
the linear method is accurate enough for a cavity flow on a
very thin section, especially when the cavity is long.
As' for the cavity volume, a comparison is shown in Figure 18 of the present method and the linear method. The tendency that the cavity volume decreases' with increasing foil thickness
is the same but the results are very different. The linear
method seams to
over-predict the cavity volume. ForNACAI6-O'12 section, the volume predicted by linear method
'is twice as large as that predicted by the present method.
09 I/c 08 0.7 0.5 0.4 03 Ò.2 0.1 0:0 0.6 -0.5 0.4 0.3 0.2 0.1 0.0 0.03
The calculated cavity lengths are now compared with the
nonlinear results of reference [4] and shown in Figuré 16. lt is 0.01
found that the results for NACAI6-009 are very close to each 0.00
other. While the cavity predicted by present method is longer
02 03 04 05 06 07 08
6% Ref [4] results (non-linear) g% Present Method 12%
-06%
o A 12% O.OÄ
0
o '2'
A' O Ref. (4] result (linear) Present Method Uc=12% t'C=O% o t/C=6ó/o D UC=9% -- t/C=6% N N.%
j
'1 Auiiii
Uhhnan's results Present Results Relative Differences Cav.Numbercr 0.87513 0.87513 0.00% Cavity Length VC 0.5000 0.51 00 +2.00% Cavity Volume C 0.0 1670 0.01794' +7.4% Lift Coeff. CL 1 0.53562 ' 0.51705 -3.5% 003 0.04 0.05 0.06 0.07 008 0.09 0.10 a/aFigure 16. Comparison between present method and the
non-linear method of ref. [4]
0.04 0.05 0.06 0.07 0.08 0.09 0.10
ciJa
Figure 17. Comparison between present method and the linear
method of ref. [4] 04 06 08 I/C 10 1.0 I/C09 0.8 0.7 0.07 C 0.06 0.05 0.04 0.03 0.02
of the re-entrant jet
formation by a high speed cinephotography method. From the framing speed of the film, the velocity of the re-entrant jet was measured. The measurement shows that the maximum speed
of the re-entrant jet is a little smaller than the freestream
velocity of the cavity. To investigate the influence of the
re-entrant jet velocity on the cavity volume and length, the predictions for NACA 16-006 at an angle of attack 4° are
calculated both for a prescribed re-entrant jet velocity and for a free jet. The result is shown in Figure 19.
The free jet here means that we did not use the jet
boundary and the jet is no longer cut off. So, With the iteration continuing, the jet evolves automatically. Figure 19 shows that when the free jet is developed long enough, these tWo methods give exactly the same result. It seams that the jet boundary is
not really needed, but the free jet approach takes more
iteration and the boundary condition on the jet surface could not be satisfied finally, even if these errors do not influence the final solution. However, in three-dimensional condition, it might be easier to use this free jet approach since it is not easy to erect a jet boundary on the surface of a three dimensiónal
body. 0.15 o .0.lo NACA 16-006
a1.O a4°
y
jet not cutc
jetcutoif
if
-0.15
00 01
all over the section surface. In the real flow the section is
actually de-cambered by the boundary
layer. Thisde-cambering could be treated accurately by a careful calculation of the boundary layer development on the section surface. A viscous/inviscid interaction calculation1201 for a 2-D section has shown this strong effect already. Any simple comparison based on equal angle of attack or equal lifting coefficient does
not work. In the equal angle of attack comparison, the
potential method always over-estimates the cavity length and volume because the negative pressure on the suction side is
always over-predicted. But in the equal lifting coefficient comparison, the potential theory always under-predicts the
negative pressure peak.
The authors of this paper are not going to carry out the
boundary layer calculation for this comparison because the deference of the pressure distribution may result not only from the boundary layer flow but also from the blockage of the test
section1211, even if the blockage was very small. We think that
ifthe cavity is small, the boundary layer on the suction side of
the section is not influenced too much by the cavity, and a
comparison could be done on an equivalent section that has
the same pressure distribution as measured at wetted condition.
A simple empirical method like Pinkerton's1221 is used. An
arbitrary function is used to de-camber the section,
A(f/C)=(f/C)TE .(l(x/C)2)
where, where (f/C)TE is the total de-camber amount at the trailing edge
For this NACA 66(MOD). a08 section,
(f/ÇT.E.=-OOO9 gives a good correlation between the
experiments and the calculations both for pressure distribution
and for the lift, as shown in Figure 20. The de-cambered
section geometry is plotted in Figure 21, which shows that the deformation is mainly around the trailing edge
Based on this de-cambered profile, the calculations are
done for three cavitation numbers (0.84, 0.91, and 1.00) at an angle ofattack of40 The pressure distribution is shown in
Figure 20 and
the calculated cavity length and thecorresponding lifting coefficients are listed in Table 5. The agreements are quite good.
In order to estimate, roughly, how strong the viscous effect is, a test for the same section at a conditionofa=4°and a=l.0
is calculated both for the original geometry and for the
de-cambered geometry. The results are plotted in Figure 22. As expected, the predicted cavity length on the original geometry
is 2.5 times as long as that predicted on the de-cambered
section.
But this result could not be generalized because it strongly depends on the pressure distribution on the section surface. For a NACA 66 section, which has a relative flat pressure (as shown in Figure 20) on a large part of the suction surface at wetted flow condition, the cavity length and volume are very sensitive to the pressure. On the other hand, for a section like the modified Joukowsky sectiont231, a de-cambering seam not
needed because the pressure is not so flat on the suction side.
Figure 23 shows the pressure distribution calculated by the present potential theory and the measured data in experiment
cietht js JtW1bWSk
MrtMhffitw ihluMww
Table 5. Comp.ofthe cavity length and lifting coefficients
Cavitation Number Experiments Calculations IIC CL I/C GL LOO 0.20 0.645 0.223 0.619 0.91 0.36 0.670 0.363 0.652 0.84 0.60 0.699 0.6096 0.678 02 03 04 0.5 x/Chord
Figure 19. Comp. of the cavity shapes with/without jet section Comparison with experimental measurements
In order to validate the present method, we compared the calculations with Shen's experimental measurements both for a modified Joukowsky profile1 and for a NACA 66 (MOD) a=0.8 section'71. Since the leading edge sheet cavity is very sensitive to the pressure distribution on the suction side of the profile, a careful investigationofthe influenceofthe viscosity,
the boundary layer separation and the wall effect in
experiment should be done before the comparison could be
done.
The viscosity has not only strong effects on the cavity flow, but also on wetted flow81. The experimental results show that the lifting coefflcient of NACA 66(MOD) a=0.8 section from experiment is 0.531 at an angle of attack of 3°
when the Reynolds number is 3x106, while the potential theory
predicts 0.626, which is
15% larger than
experiment.Correspondingly, the calculated negative pressure distribution
is also hiher than the measured ones. For the cavity flow,
AvellanU9J's recent experiment shows that the leading edge cavity on a NACA 009 symmetric profile at 2.5° incidence is 30% chord length when the water speed is 20 mIs, but 45%
chord length when the water speed is 35 m/s.
This may be resulted by the differences of the boundary development on the suction side and on the pressure side at
different Reynolds numbers. For the NACA 66 (MOD) a=0.8
section, the transition points of the boundary layer on both
sides are reported to be at 13% of the chord length on the
suction side and at 89% ofthe chord length on the pressure
0.05
0.00
-0.05
f10
without viscous correction. ihe calculated cavity lengths for
two different angle of attacks are, listed in Table 6 The
agreement is very satisfactory.
Since there is no available good experimental results for the cavity volume, no comparison is done at present.
Table 6. Comp. of cavity length calculations with experiments
u u cc Expernents Calculations 0.25 o 9 a) 15 resent method cT1 .0 a0.84 E1.0 a) C) 8
p.5
0.5 --0.2 3.8 4.3A°'
A wetted flowFigure 21 The.geometry of thede-cambered section.
E 03
0.249
experiment
NACA 66(MOD) t/C=9%
NACA a=0.8 fIC=2%
-1.0 i I i
--0 --0 0.2 0.4 0.6 0.8 1.0
i/Chord
Figure 20. Comparison of the pressure distributions at an angle
ofattack of40 2 0.4 o
o
>.
0.2 0.1 -0.0 0.1 -NACA66(MOD)t/C=9%NACA a0.8 fIC2%
2.6 1.0 0.5 0.0 Experimenl o Present Method 0.39 0.395
-Cp ModifiedJoukowsky Foil ai3.25°
00
02 04 06 08 10x/Chord
Figure 23. Comparison of pressure
disfributionon the
modified Joukowsky profilé.
Conclusions
A potential based surface panel method is developed to
predict the cavity flow around an arbitrary two-dimensional
foil section. Lots of numerical calculations show that this
method is quite stable for different geometry. Although the convergence is still not very fast, it always converged to the
same result for different initial cavity length.
The result obtained by using a cavity termination wall
model in present method shows a good agreement with
Uhhnan's non-linear termination wall result. Lots of calculations for NACA 16 series sections with three different ,
thickness to chord ratio (6%, 9% and: 12%) have been
performed. The predicted cavity lengths and volumes with re-entrant jet modeliñg have been compared with the Uhiman's nonlinear results and the linear results with thickness
corrections respectively The present calculations show the
same trend like other non linear methods that the cavity
Iengthdecreases with increäÑing foil thickness.
As for the cavity volUme, the predicted results for NACA 16-012 are only half of the cavity völumes predicted by the
linear méthod. This valué is also smaller than Uhlman's
nonlinearresults. The cavity lengthspredicted with the present re-entrant jet model are larger than those of the termination wall model for NACA 16-006, but shorter than the results for
NACA 16-012. The re-entrant jet thickness also decreases
with increasing foil thickness
It is very iniportant to have found that, at least in our calculation cases, the re-entrant jet thickness is always a
certain percentage
of the
cavity maximum thickness,irrespective of the cavitation number, the cavity length and the foil thickness The locations of the maximum cavity thickness
for different conditions are also fixed at 60% of the cavity
length from thedetachment point.
The present method is validated by comparing with Shen's
experimental data.. Since the experiments show a strong
viscous effects both for the cavity flow and the wetted flow around NACA 66 MOD) a=08section the comparisons were
done by de-cambering the original NACA section into a
00 0.6 0.8 1.0
IChord
..sults and the
,-nother comparison, is made for a modilied Joukowsky
profile at two different angles of atiack. Because the pressure
distribution on the suction side of the section is not very flat,
the calculated cavity length is not so sensitive to the pressure
difference between the calculation and the measurement. A
good agreement has been found between the present
calculation and the experimental data. Viscous correction
seams not needed for this case.
In summary, the re-entrant jet modéling with the potential
theory for 2-dimensional cavity flows is a quite stable and
convergent method. If the viscous effect could be included in
the method, it could provide a rather precise prediction. A benchmark test of the cavity volume for steady condition
should be carried out to verify the present theory. A further extension of the method to three-dimensional predictions and unsteady cavity flow around hydrofoils and propeller blades
seams feasible.
Acknowledgments
This work was supported by the Maritime Research Institute Netherlands (MARIN). The authors would like to
express their appreciation to the comments of H. C. J. van Wijngaarden and to the discussions with Prof. S. L.. Ceccio when he was visiting MARIN.
References
[1] D. F. de Lange, 'Observation and Modeling of Cloud Formation behind a Sheet Cavity', Ph. D. Thesis of
Twente University, April, 1996.
[2] S. A. Kinnas & N. E. Fine, 'A Numerical nonlinear
Analysis of the Flow around Two- and
Three-Dimensional Partially Cavitating Hydrofoils', Journal of
Fluid Mechanics, Vol. 254, 1993, pp. 151-.181..
T. Y. Wu, 'Cavity and Wake Flow', Annual Review of
Flùid Mechanics', Vol. 4, 1972.
[4] J.S. Uhlman ir, 'The Surface Singularity Method Applied
to Partially Cavitating Hydrofoils', Journal of Ship
Research, Vol. 31, No.2, June 1987, pp. 107-124.
M. P. Tulin & C. C. 1-Isu, 'New. Application of cavity
Flow Theory', Proceedings.of 13thSymposium on Naval Hydrodynamics, Tokyo, Japan, 1980.
[6] T. Y. Wu, 'A Note on the Linear and Nonlinear Theories for Fully Cavitated Hydrofoils', Technical Report on No. 2 l-22, CIT, California, August, 1956.
J. A. Geurst, 'Linearized Theory for Partially Cavitated Hydrofoils', International Shipbuilding Progress, Vol. 6,
No.60, Aug. 1959, pp. 369-384.
[8] S. A. Kinnas, 'Leading Edge Corrections to the Linear
Theory of Partially Cavitating Hydrofoils?, Journal of
Ship Research, Vol. 35, No. I, March 1991, pp. 15-27.
S. A. Kinnas & N. E. Fine, 'Nonlinear Analysis of the
Flow around Partially or Super-cavitating Hydrofoils by a
Potential Based Panel Method', Proceedings ofBoundary
Integral Methods - Theory and Applications, IABEM-90
Symposium of the International Association for Boundary Element Methods, Rome Italy, October, 1990.
[IO] S. A. Kinnas & N. E. Fine, 'A nonlinear Boundary Element Method for the Analysis of Unsteady Propeller Sheet Cavitation', Proceedings of 19th Symposium on
Navel Hydrodynamics, Seoul, Korea, August 1992.
[Il] Y. T. Shen & F. B. Peterson, 'Unsteady Cavitation on an Oscillating Hydrofoil', Proceedings of 12thSymposium on Naval Hydrodynamics, Washington, 1978.
[3]
[5]
[7]
[9]
Annual Review ofFluid Mechanics, 1985, 17:41 l-45. [l3]N. Bose, 'Explicit Kutta Condition for an Unsteady
Two-Dimensional Constant Potential Panel Method', Technical
Notes, AIAA Journal, Vol. 32,.No. 5, May, 1994.
[l4]L.
Morino &
C. C. Kuo, 'Subsonic PotentialAerodynamics for Complex Configurations: A General
Theory', AIAA Journal, Vol. 12, No. 2, 1974, pp.
191-I 97.
[IS] R. A. Furness & S. P. Hutton, 'Experimental and
Theoretical Studies of Two-Dimensional Fixed-Type
Cavities', Journal of Fluids Engineering, December 1975,
pp.5 15-522.
[16]D. Gilbarg & H. H. Rock, 'On Two Theories of Pläne
Potential Flows with Finite Cavities', NOLM report No.
8718, Washington'D. C., 1946.
[17] Y. T. Shen & P. E. Dimotakis, 'The Influence of Surface Cavitation on Hydrodynamic Forces', Proceedings of22uid
ATTC, St. Johns, 1989.
[18]Y. T. Shen & P. E. Dimotakis, 'Viscous añd Nuclei
Effects on Hydrodynamic Loadings and Cavitation of a
NACA 66 (MOD) Foil
Section', Journal of FluidsEngineering, Vol. III, Sept. 1989, pp306-3 16.
[19]F. Avellan, P.
Dupont & I.
Ryhming, 'GenerationMechanism and Dynamics of Cavitation Vortices
Downstream of A Fixed Leading Edge Cavitation',
Proceedings of l7 ONR, The Hague, 1989.
[20]S. A. Kinna, S. Mishima & W. H Brewer, 'Non-linear
Analysis of Viscous Flow Around Cavitating Hydrofoils', Proceedings of 2O" ONR, University of California, Santa Barbara CA, 1994.
[21JiM. Deshpande, J. Feng & C. L. Merkle, 'Cavity Flow Predictions Based on the Euler Equations', Journal of
Fluids Engineering Vol 116 March 1994 pp36 44 [22]R M. Pinkerton 'CalcUlated and Measured Pressùre
Distributions over the Midspan Section of the NACA
441.2 .irfoil', NAC report No. 563, 1936.
[23jY T. Shen & F B. Peterson, 'The Influence of Hydrofoil Oscillation on Boundary Layer Transition and Cavitation Noise', Proceedings of 13th Symposium on Naval
Hydrodynamics, Tokyo, 1980.