Supercavitation PDF
Supercavitation PDF
Supercavitation PDF
.
Igor Nesteruk
Editor
Supercavitation
Advances and Perspectives
of correspondent-member of
the National Academy of sciences of Ukraine
Yuriy Savchenko
.
Address About Yuriy Savchenko
vii
viii Address About Yuriy Savchenko
of hydrodynamic tunnels and test rigs. According to his projects the Impulse
hydrodynamic tunnels were built for Northrop Grumman Corporation (USA) and
Shipbuilding Centre in Wuxi (China).
The Yu. Savchenko’s contribution in development of new technologies was
awarded to the medal “For labour merit” in 1978, he is a laureate of the prize of
the Academy of technologic sciences of Ukraine in 1995 in the field of development
of new technologies and a laureate of State prize of Ukraine in the field of science
and engineering in 2002. The Yu. Savchenko’s scientific investigations are
represented in about 100 articles, 40 inventions, monographs and more than 200
scientific reports. Yuriy Savchenko has wide domestic and international recogni-
tion, he is an initiator of contacts with scientists and institutions of different
countries, a member of organizing committees of international conferences and a
scientific leader of researches on supercavitation according to contracts with
organizations in USA, Germany, China and Singapore.
The activity and life principles of Yu. Savchenko essentially promote the climate
of creative search, mutual respect and exactingness formed in the collective. He is a
considerable authority in the collective.
Collaborators of the Institute, his colleagues in the research work sincerely
congratulate him and wish him happiness and new creative achievements.
ix
x Contents
Abstract
Unsteady quasi-one-dimensional and two-dimensional bubbly cavitating nozzle
flows are considered using a homogeneous bubbly flow model. For quasi-one-
dimensional nozzle flows, the system of model equations is reduced to two
evolution equations for the flow speed and bubble radius and the initial and
boundary value problems for the evolution equations are formulated. Results
obtained for quasi-one-dimensional nozzle flows capture the measured pressure
losses due to cavitation, but they turn out to be insufficient in describing the
two-dimensional structures. For this reason, model equations for unsteady two-
dimensional bubbly cavitating nozzle flows are considered and, by suitable
decoupling, they are reduced to evolution equations for the bubble radius and
for the velocity field, the latter being determined by an integro-partial differen-
tial system for the unsteady acceleration. This integro-partial differential system
constitutes the fundamental equations for the evolution of the dilation and
vorticity in two-dimensional cavitating nozzle flows. The initial and boundary
value problem of the evolution equations are then discussed and a method to
integrate the equations is introduced.
1 Introduction
Tangren et al. [1] using a barotropic model. The problem has been reconsidered
by Ishii et al. [2] by taking into account unsteady effects, but still neglecting
bubble dynamics. A one-dimensional continuum bubbly flow model that couples
spherical bubble dynamics to the flow equations was proposed by van Wijngaarden
[3, 4] and was later employed in investigating shock wave structure [5]. Quasi-
one-dimensional steady-state solutions of bubbly cavitating flows through
converging–diverging nozzles are investigated using the continuum bubbly mixture
model [6, 7] by assuming that the gas pressure inside the bubble obeys the
polytropic law and by lumping all damping mechanisms by a single damping
coefficient in the form of viscous dissipation. These investigations have
demonstrated that steady-state solutions are possible only for some range of the
cavitation number, with the rest of the parameters kept fixed. Moreover, a recent
investigation [8] shows that the temporal stability of these quasi-one-dimensional
steady-state solutions suffer from being very sensitive to slight unsteady
perturbations. A numerical investigation of unsteady quasi-one-dimensional bubbly
cavitating flows have also been carried out [9] showing the possibility of
propagating bubbly shock waves in the diverging section of the nozzle.
The aim of this investigation is devoted to a detailed study of unsteady quasi-
one-dimensional and two-dimensional bubbly cavitating nozzle flows. For this
reason we first discuss the homogeneous bubbly mixture model previously
introduced for quasi-one-dimensional steady-state and unsteady cavitating nozzle
flows [5–9]. For quasi-one-dimensional cavitating nozzle flows, by a detailed
analysis the system of model equations is reduced to two evolution equations for
the flow speed and bubble radius and the initial and boundary value problems for
the evolution equations are formulated. For this case a numerical algorithm is
constructed for the solution of the initial and boundary value problems of evolution
equations. Results obtained for quasi-one-dimensional nozzle flows capture the
measured pressure losses due to cavitation, but they turn out to be insufficient in
describing the two-dimensional structures such as the formation and development
of the attached cavity, the formation of the re-entrant jet and bubble cloud shedding
and collapse. For this reason model equations for unsteady two-dimensional bubbly
cavitating nozzle flows are considered and, by suitable decoupling, they are reduced
to evolution equations for the bubble radius and for the velocity field, the latter
being determined by an integro-partial differential system for the unsteady acceler-
ation. More importantly, this integro-algebraic partial differential system seems to
form the fundamental equations for the evolution of the dilation and vorticity. In
particular, the evolution equation of vorticity is shown to be precisely Fridman’s
equation of vorticity [10], containing terms arising from non-barotropic flow. The
initial and boundary value problem of the evolution equations are then discussed
and a method to integrate the equations is introduced.
Mathematical Theory and Numerical Simulation of Bubbly Cavitating Nozzle Flows 3
2 Model Equations
r0 ¼ r‘ 0 ð1 bÞ (1)
@r0 @
A0 þ ðr0 u0 A0 Þ ¼ 0 (2)
@t0 @x0
du0 @p0
r0 0
¼ 0 (3)
dt @x
R0 3 ð1 bÞ 3
¼ ¼ constant: (4)
b 4p0 0
Dr 0
L¼ (6)
R0
4 C.F. Delale et al.
with Dr0 denoting the radius of influence of interacting bubbles from the center of
any fixed bubble [7, 19]. In Eq. 5 a polytropic law for the expansion and compres-
sion of the gas inside the gas/vapor bubble is used and all damping mechanisms, in
an ad hoc manner, [20–23] are assumed in the form of viscous dissipation,
characterized by a single viscosity coefficient m0 eff. Using the normalization
r p0 p0
r¼ 0
¼ 1 b; p ¼ 0 ; pv ¼ 0 v ;
r‘ p i0 p i0
0 0
pg u R0
pg ¼ 0 ; u ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ; R ¼ 0 ;
p i0 p0 i0 =r0 ‘ R i0
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 0
x0 A0 t0 p0 i0 =r0 ‘ t
x¼ 0 ;A¼ 0 ;t¼ 0¼ ; (7)
Hi Ai Y Hi 0
r ¼ 1 b; (8)
@r @
A þ ðruAÞ ¼ 0; (9)
@t @x
du @p
r ¼ ; (10)
dt @x
1b 1 bi0
R3 ¼ ¼ k3i (11)
b bi0
and
h i
3
pv p 1 þ ð3L 2
1ÞðR=k i Þ =2 d2 R
2
¼ h i R 2
L 1 þ ðR=ki Þ3 dt
h i
3 6
3 1 þ 2ð2L 2
1ÞðR=k i Þ þ L 2
ðR=k i Þ dR 2
þ h i2
2 dt
1 þ ðR=ki Þ3
S0 4 dR pgi
þ þ 2 2 3k (12)
L R L ðReÞR dt L R
2
R0 i0
L¼ ; (13)
H0 i
Mathematical Theory and Numerical Simulation of Bubbly Cavitating Nozzle Flows 5
1 bi0
k3i ¼ ; (14)
bi0
2S0
S0 ¼ ; (15)
p0 i0 R0 i0
and Re is a typical Reynolds number, based on the overall damping coefficient m0 eff,
and is defined by
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
r0 ‘ H 0 i p0 i0 =r0 ‘
Re ¼ 0
: (16)
m eff
Furthermore, by eliminating the void fraction b, the mixture density r and the
mixture pressure p between Eqs. 8–12, we arrive at the evolution equations for the
bubble radius R(x,t) and for the flow speed u(x,t) as
@R @R 1 1 dA @u
¼ u þ 2 ðR3 þ k3i Þ uþ (17)
@t @x 3R A dx @x
and
@u
¼ aðx; tÞ (18)
@t
where the unsteady acceleration satisfies the linear partial differential equation
@2a @R @a @R @R @u @ 2 u @ 3 u
þ gðR; ; xÞ þ hðR; ; xÞa ¼ sðR; u; ; ; ; ; xÞ (19)
@x 2 @x @x @x @x @x @x2 @x3
@R F1 ðRÞ @R 1 dA
gðR; ; xÞ ¼ þ ; (20)
@x F2 ðRÞ @x A dx
@R F1 ðRÞ 1 dA @R F3 ðRÞ d 1 dA
hðR; ; xÞ ¼ þ þ (21)
@x F2 ðRÞ A dx @x F2 ðRÞ dx A dx
6 C.F. Delale et al.
and
3
@R @u @ 2 u @ 3 u @ u
sðR; u; ; ; ; ; xÞ ¼ u 3
@x @x @x2 @x3 @x
F1 ðRÞ @R F4 ðRÞ @u F4 ðRÞ 1 dA F5 ðRÞ @ 2 u
þ u þ þ u þ
F2 ðRÞ @x F2 ðRÞ @x F2 ðRÞ A dx F2 ðRÞ @x2
2
F6 ðRÞ @R @u
þ
F2 ðRÞ @x @x
F7 ðRÞ 1 dA F3 ðRÞF5 ðRÞ @R @u
þ 3
F2 ðRÞ A dx RF2 ðRÞ @x @x
2
F4 ðRÞ 1 dA @u F8 ðRÞ d 1 dA
þ þ u
F2 ðRÞ A dx @x F2 ðRÞ dx A dx
2
F9 ðRÞ 1 dA F5 ðRÞ 1 dA F3 ðRÞ @u
þ u þ þ u
F2 ðRÞ A dx F2 ðRÞ A dx F2 ðRÞ @x
F6 ðRÞ 2 1 dA 2 F3 ðRÞF5 ðRÞ 1 dA
þ u 3 u
F2 ðRÞ A dx RF2 ðRÞ A dx
F1 ðRÞ 2 d 1 dA F10 ðRÞ @R
þ u þ
F2 ðRÞ dx A dx F2 ðRÞ @x
F9 ðRÞ 2 1 dA d 1 dA d 2 1 dA
þ u þ u2 2
F2 ðRÞ A dx dx A dx dx A dx
The functions Fj (R); j ¼ 1,2,. . .,10, entering Eqs. 20–22 are given in
Appendix A. The solution for the mixture pressure, the void fraction and the
density then follow by
S0 pgi L2 k6i
p ¼ pv þ 3k ð6L2 1ÞðR=ki Þ6
R R 18R4
4k3i h i
þ ð6L2 2ÞðR=ki Þ3 1 c2 1 þ ðR=ki Þ 3
c
3ðReÞR3
L2 k3i i dc
½2þð3L2 1ÞðR=ki Þ3 (23)
6R dt
and
R3
b¼1r¼ (24)
R3 þ k3i
Mathematical Theory and Numerical Simulation of Bubbly Cavitating Nozzle Flows 7
For the analysis of the 2D (or 3D) structures of partial cavitation and
supercavitation observed in experiments, the quasi-one-dimensional model
equations discussed above are insufficient. Therefore, the model equations should
be extended to multi-dimensional flows. In this section, for simplicity, we introduce
the model equations for two-dimensional unsteady bubbly cavitating flows to be
able to calculate, at least, some of the 2-D flow structures observed. Using the
homogeneous two-phase dispersed flow model and the classical Euler equations,
the continuity and momentum equations in two-dimensions take the form
@r0 @ @
þ ðr0 u0 Þ þ 0 ðr0 v0 Þ ¼ 0; (25)
@t0 @x0 @y
0 0 0
@u 0 @u 0 @u @p0
r0 þ u þ v ¼ (26)
@t0 @x0 @y0 @x0
and
0 0 0
@v 0 @v 0 @v @p0
r0 0
þ u 0
þ v 0
¼ 0 (27)
@t @x @y @y
where the mixture density r0 is given by Eq. 1 and the void fraction b is related to
the radius of mono-dispersed spherical bubbles by Eq. 4, assuming there is no
bubble creation and coagulation. Equations 25–27 together with Eqs. 1, 4 and the
modified Rayleigh–Plesset equation (5) constitute the model equations for unsteady
2-D bubbly cavitating nozzle flows. With the normalization given by Eq. 7 together
with y ¼ y0 /H0 i, the two-dimensional normalized model equations take the form
r ¼ ð1 bÞ; (28)
@r @ @
þ ðruÞ þ ðrvÞ ¼ 0; (29)
@t @x @y
8 C.F. Delale et al.
@u @u @u @p
r þu þv ¼ ; (30)
@t @x @y @x
@v @v @v @p
r þu þv ¼ ; (31)
@t @x @y @y
and
1 b 1 bi0
R3 ¼ ¼ k3i : (32)
b bi0
@u
¼a (34)
@t
and
@v
¼b (35)
@t
where the unsteady acceleration field (a, b) satisfies the linear system of integro-
partial differential equations
h i
3
@a @b R 2 þ ð3L 1ÞðR1 =ki Þ
2
@a @b
þ h i þ
@x @y R1 2 þ ð3L2 1ÞðR=ki Þ3 @x @y y¼0
ðy
6R b
h i h i dy1 ¼ Sa (36)
3
L2 k3i 2 þ ð3L 1ÞðR=ki Þ
2
0
1 þ ð R = ki Þ 3
@b @a 3R2 @R @R
3 b a ¼ Sb (37)
@x @y R þ k3i @x @y
Mathematical Theory and Numerical Simulation of Bubbly Cavitating Nozzle Flows 9
ðy
R
þh i sa dy1 (38)
2 þ ð3L 1ÞðR=ki Þ3
2
0
and
@o @o
Sb ¼ oc u v
@x @y
3R 2
@v @v @R @u @u @R
þ 3 u þ v u þ v (39)
ðR þ k3i Þ @x @y @x @x @y @y
(40)
In Eqs. 38–40, c and o, respectively, denote the dilation (in this case the
divergence of the velocity field) and the vorticity and are given by
@u @v
c¼ þ (41)
@x @y
and
@v @u
o¼ : (42)
@x @y
10 C.F. Delale et al.
Equations 36 and 37 for the unsteady acceleration field (a, b) constitute the
fundamental equations for the transport of the dilation c and of the vorticity o in
2D bubbly cavitating flows. In particular, Eq. 37 is precisely the non-barotropic
vorticity transport equation, called the Fridman equation [10], given by
@v 1
þ ðu:rÞv ¼ cv þ ðv:rÞu þ 2 rr rp (43)
@t r
where the term ðv:rÞu vanishes in 2D. Thus it forms the basis for the generation of
vorticity in non-barotropic flows and is responsible for the re-entrant jet in partial
cavitation and for all closure models of cavitation. In the absence of cavitation
where the source terms Sa and Sb vanish, Eqs. 36 and 37 reduce to the classical
Cauchy-Riemann equations (existence of the complex velocity potential). The
equations for the normalized pressure, normalized density and void fraction then
follow from Eqs. 23 and 24 with the dilation now defined by Eq. 41.
Fig. 1 Geometric configuration of the nozzle employed by Preston et al. [9] and the boundary
conditions used for the numerical simulation of quasi-one-dimensional bubbly cavitating nozzle
flows
Mathematical Theory and Numerical Simulation of Bubbly Cavitating Nozzle Flows 11
The initial flow field can be taken as the slightly perturbed steady-state quasi-
one-dimensional flow field (for the range of parameters where quasi-one-
dimensional steady-state solutions are not possible [6, 7], one may start with the
incompressible solution supplemented by an everywhere constant bubble radius
distribution). To be able to specify the boundary conditions at the nozzle inlet
(x ¼ xi) and at the nozzle exit (x ¼ xe), we have to discuss the nature of the
evolution equations (17)–(19). In particular, Eq. 17 for the bubble radius evolution
is hyperbolic for given flow speed so that we need only to specify the bubble radius
at the inlet so that
with R0(xi) ¼ Ri(0) to avoid a discontinuity in the bubble radius at the nozzle inlet.
On the other hand, Eqs. 18 and 19 can be combined into a single evolution equation,
coupled to the flow speed and bubble radius, as
@u
¼ aðx; tÞ ¼ K1 ðtÞA1 ðx; tÞ þ K2 ðtÞA2 ðx; tÞ
@t
ðx
sðx; tÞA1 ðx; tÞ þ uð@ 2 u=@x2 Þð@A1 =@xÞ
þ A2 ðx; tÞ dx
xi Wðx; tÞ
ðx
sðx; tÞA2 ðx; tÞ þ uð@ 2 u=@x2 Þð@A2 =@xÞ
A1 ðx; tÞ dx (46)
xi Wðx; tÞ
where W represents the Wronskian of the two linearly independent solutions A1 and
A2 of the linear homogeneous equation corresponding to Eq. 19 for the unsteady
acceleration a and is given by
@A2 @A1
Wðx; tÞ ¼ A1 A2 ; (47)
@x @x
and where K1(t) and K2(t) are time dependent functions to be determined from the
nozzle inlet and exit boundary conditions and s is given by
@R @u @ 2 u
sðx; tÞ ¼ sðR; u; ; ; ; xÞ
@x @x @x2
@R @u @ 2 u @ 3 u
¼ sðR; u; ; ; ; ; xÞ
@x @x @x2 @x3
@ 3 u @ 2 u @u @R
þ u 3 þ 2 ½ þ ugðR; ; xÞ: (48)
@x @x @x @x
In order to evaluate the time dependent functions K1(t) and K2(t) in Eq. 46, we
consider the appropriate boundary conditions at the inlet and outlet of the nozzle.
12 C.F. Delale et al.
For real cavitating flows, either of the following two sets of boundary conditions
can be specified:
(a) The inlet flow speed and exit pressure are specified, i.e.
together with Ui(0) ¼ u0(xi) and Pe(0) ¼ p(xe,0) to ensure continuity of the
solutions.
(b) The inlet and exit pressures are specified, i.e.
together with Pi(0) ¼ p(xi,0) and Pe(0) ¼ p(xe,0) to ensure continuity of the
solutions.
The evaluation of the time dependent functions K1(t) and K2(t) in Eq. 46
corresponding to the boundary conditions in each case are given in Appendix B.
It should be mentioned that the boundary conditions of case (b) require enormous
amount of computation time. Therefore, for simplicity, we adopt the boundary
conditions of case (a).
For the numerical method, we first evaluate the unsteady acceleration field by
Eq. 46 at every instant t using the flow speed distribution u(x,t) and the radius
distribution R(x,t) at that instant, starting with the initial distributions u0(x) and
R0(x). The homogeneous solutions A1 and A2 of Eq. 19 for the unsteady acceleration
are obtained by power series methods of second order linear ordinary differential
equations with variable coefficients. The time dependent functions K1(t) and K2(t)
are evaluated using non-reflecting boundary conditions. Using the unsteady accel-
eration field, the evolution Eq. 18 is integrated using a multi-stage Runge–Kutta
method in time to arrive at the flow speed distribution at the next time step. Using
the flow speed thus obtained, the first order hyperbolic equation (17) for the bubble
radius R is integrated by the classical method of characteristics. Thus the solutions
for the flow speed and radius distributions of the evolution equations are obtained
for the next time step. The procedure is repeated in a similar manner for all
subsequent time steps.
Fig. 2 Nozzle geometry and boundary conditions of the unsteady acceleration field for two-
dimensional bubbly cavitating nozzle flows
in such a way that it is also uniform at the nozzle inlet and exit. Taking into account
the hyperbolicity of Eq. 33 for the bubble radius for given velocity field, we need
only to specify the bubble radius at the inlet. Assuming that the inlet bubble radius
distribution is uniform in y at all times, we have
with Ri(0) being equal to the corresponding initial inlet bubble radius to avoid a
discontinuity in the bubble radius at the nozzle inlet. Similar to the procedure of
quasi-1D flows, we can specify two sets of boundary conditions:
(a) The inlet flow speed and exit pressure, both uniform, are specified, i.e.
for t 0 together with Ui(0) and Pe(0) matching the corresponding initial inlet and
exit values to ensure continuity of the solutions.
(b) The uniform inlet and exit pressures are specified, i.e.
for t 0 together with Pi(0) and Pe(0) matching the corresponding initial inlet and
exit values to ensure continuity of the solutions.
The above boundary conditions, similar to the procedure in quasi-one-
dimensional flows, should be converted to the boundary conditions for the unsteady
acceleration field for the integro-partial differential system, given by Eqs. 36 and
37. For this reason, assuming that the inlet velocity field is uniform and unidirec-
tional and that the bubbles are in mechanical equilibrium at the inlet and exit of the
nozzle and using Eq. 23 for the pressure field, we can arrive at the following
boundary conditions for the system of Eqs. 36 and 37 in each case:
Case (a) The inlet flow speed and exit pressure, both uniform, are specified.
a ¼ 0 and b ¼ 0 at x ¼ xi ; ax ¼ 0 and b ¼ 0 at x ¼ xe ;
ay ¼ 0 and b ¼ 0 at y ¼ 0; b ¼ a tan y at y ¼ hðxÞ (56)
where y ¼ h(x) denotes the shape of the upper wall of the nozzle and tany ¼ dh/dx.
Such a configuration of the boundary conditions are given in Fig. 2.
Case (b) The uniform inlet and exit pressures are specified.
ax ¼ 0 and b ¼ 0 at x ¼ xi ; ax ¼ 0 and b ¼ 0 at x ¼ xe ;
ay ¼ 0 and b ¼ 0 at y ¼ 0; b ¼ a tan y at y ¼ hðxÞ: (57)
For the numerical method, similar to the procedure for quasi-1D flows, we first
consider the integro-partial differential system of equations, given by Eqs. 36 and
37, subject to boundary conditions given by either Eq. 56 or Eq. 57. The system is
solved in two iterative steps. In the first step the integral on the left-hand side of
Eq. 36 is set equal to zero and the remaining elliptic system of first order partial
differential equations is first discretized by a central finite difference scheme. The
resulting linear system of algebraic equations, subject to the boundary conditions
given in Eq. 56 or in Eq. 57, are solved by Gauss-Seidel Over Relaxation Method.
In the second step, the skipped integral on the left hand side of Eq. 36 is evaluated
and treated as a source term. The first step is then repeated to obtain the unsteady
acceleration field at that instant. Using a multi-stage Runge–Kutta method in time
and the solution for the unsteady acceleration field, the evolution Eqs. 34 and 35 are
integrated to yield the velocity field in the next step. Using this velocity field, the
hyperbolic evolution equation (33) is integrated by the method of characteristics or
by using flux splitting methods to arrive at the bubble radius in the next time step.
The numerical scheme is then to be repeated for all subsequent time steps.
Fig. 3 Geometry and boundary conditions of the nozzle used in the numerical simulation of
quasi-one-dimensional flows for comparing the measured wall pressure distributions
respectively. The nozzle employed by Preston et al. [9], whose geometric configu-
ration is shown in Fig. 1, is considered in order to validate our numerical simulation
results against their results obtained by some other numerical means. The nozzle
whose geometric configuration is shown in Figs. 2 and 3 is employed to compare
the pressure distribution obtained for quasi-one-dimensional bubbly cavitating
nozzle flows against the measured pressure values at the wall of the nozzle under
the same inlet and exit conditions. In both cases we use nozzle inlet velocity and
nozzle exit pressure as boundary conditions. We follow the numerical method
described in Sect. 3.1 in each case. For the results to be shown we define the
cavitation number and the pressure coefficient as
p0 i0 p0 v
s¼ (58)
ð1=2Þr0 ‘ u0 2 i
and
p0 p0 i0
Cp ¼ : (59)
ð1=2Þr0 ‘ u0 2 i
Fig. 4 Comparison of the results for the pressure coefficient obtained by the present numerical
simulations against those of Preston et al. [9] for bubbly cavitating flow through the nozzle
employed by Preston et al. [9] for two different back pressures corresponding to s ¼ 1.2 and
s ¼ 0.932 with inlet void fraction bi0 ¼ 103
Fig. 5 Comparison of the results for the normalized bubble radius obtained by the present
numerical simulations against those of Preston et al. [9] for bubbly cavitating flow through the
nozzle employed by Preston et al. [9] for two different back pressures corresponding to s ¼ 1.2
and s ¼ 0.932 with inlet void fraction bi0 ¼ 103
Mathematical Theory and Numerical Simulation of Bubbly Cavitating Nozzle Flows 17
Fig. 6 The unsteady quasi-1D distributions of the pressure coefficient at three instants of time and
the time-averaged measured experimental values for the cavitating nozzle flow of water with air
bubbles with inlet void fraction bi0 ¼ 106, inlet bubble radius R0 i0 ¼ 50 mm, inlet flow speed
u0 i ¼ 8.2 m/s and exit pressure p0 e ¼ 0.388 bar for the nozzle shown in Fig. 3
Under the stated conditions, the steady-state solution shows that the bubbles grow
slightly reaching their maximum size and then they return to their initial size. In this
case the large growth and violent collapse of the bubbles do not occur and the bubbles
seem to be in local mechanical equilibrium [24]. To reach unsteady cavitating flow
conditions, we lower the exit pressure until the specified exit pressure under the
unsteady cavitating flow conditions is reached. The pressure coefficient, normalized
flow speed, normalized bubble radius and normalized unsteady acceleration
distributions along the nozzle axis obtained by the bubbly flow model are shown in
Figs. 6, 7, 8, 9 at three instants of time at the start of unsteady cavitation. In these
figures the transient distributions are ignored and the time t0 ¼ 0 is artificially set at
the begining of unsteady cavitation. It is seen in Fig. 6 that reasonable agreement is
achieved between the quasi-one-dimensional unsteady pressure distributions and the
measured values from the experiments performed at the Mechanical Engineering
Department at Istanbul Technical University under the same conditions. On the other
hand, a close examination of the flow speed and radius distributions, shown in Figs. 7
and 8, show that they seem to deviate only slightly fom the initially specified slightly
perturbed steady-state distributions, since the cavitation sheets attached to the nozzle
walls, in this case, have small thicknesses compared to the nozzle height, thus
influencing these distributions only slightly. However, the presence of unsteady
cavitation leads to pressure losses which are accommodated by relatively large values
of the unsteady acceleration, as shown in Fig. 9. These large values of the unsteady
acceleration are balanced by the pressure gradients. They do not contribute to the
flow speed significantly because of the very small characteristic times involved.
18 C.F. Delale et al.
Fig. 7 The unsteady quasi-1D distributions of the normalized flow speed at three instants of time
for the cavitating nozzle flow of water with air bubbles with inlet void fraction bi0 ¼ 106, inlet
bubble radius R0 i0 ¼ 50 mm, inlet flow speed u0 i ¼ 8.2 m/s and exit pressure p0 e ¼ 0.388 bar for
the nozzle shown in Fig. 3
Fig. 8 The unsteady quasi-1D distributions of the normalized bubble radius at three instants of
time for the cavitating nozzle flow of water with air bubbles with inlet void fraction bi0 ¼ 106,
inlet bubble radius R0 i0 ¼ 50 mm, inlet flow speed u0 i ¼ 8.2 m/s and exit pressure p0 e ¼ 0.388 bar
for the nozzle shown in Fig. 3
Mathematical Theory and Numerical Simulation of Bubbly Cavitating Nozzle Flows 19
Fig. 9 The unsteady quasi-1D distributions of the normalized unsteady acceleration at three
instants of time for the cavitating nozzle flow of water with air bubbles with inlet void fraction
bi0 ¼ 106, inlet bubble radius R0 i0 ¼ 50 mm, inlet flow speed u0 i ¼ 8.2 m/s and exit pressure
p0 e ¼ 0.388 bar for the nozzle shown in Fig. 3
Conclusions
Model equations for quasi-one-dimensional and two-dimensional bubbly
cavitating nozzle flows are presented and the evolution equations for the bubble
radius and velocity field in each case are obtained. In particular, in two-dimensional
flows the integro-partial differential system of equations for the unsteady accelera-
tion field, which enters the evolution equations for the velocity field, is shown to
constitute the fundamental equations of 2D cavitating flows, exhibiting the evolu-
tion of the dilation and of the vorticity. The initial/boundary value problems are
then formulated for both unsteady quasi-one-dimensional and two-dimensional
bubbly cavitating nozzle flows. Results obtained for the unsteady quasi-one-
dimensional case show that it is possible to determine the pressure loss due to
cavitation in this case.
However, two-dimensional structures of cavitation cannot be determined.
Thus the need for a two-dimensional numerical simulation of the model
equations is essential. Moreover, the model equations can then be modified to
include the boundary layer effect of the flow by using the Navier–Stokes
equations for the bubbly mixture and to include bubble nucleation, compress-
ibility and thermal damping effects left out in describing bubble formation and
bubble dynamics. These will be the subjects of future investigations.
Acknowledgment This paper is dedicated to Professor Yu.N. Savchenko on the occasion of his
70th birthday.
20 Appendix A
Appendix A
L2 k3i
F1 ðRÞ ¼ h i
3R2 1 þ ðR=ki Þ3 (A1)
h i
3L2 1 ðR=ki Þ6 þ 3L2 2 ðR=ki Þ3 1 ;
L2 k3i h i
F2 ðRÞ ¼ 2 þ 3L2 1 ðR=ki Þ3 ; (A2)
6R
1
F3 ðRÞ ¼ h i; (A3)
1 þ ðR=ki Þ3
L2 k6i h 6 3
i
F4 ðRÞ ¼ 21L 2
5 ð R=ki Þ þ 12L2
þ 2 ðR=ki Þ 2 ; (A4)
18R4
4k3i h 3
i
F5 ðRÞ ¼ 1 þ ð R=k i Þ ; (A5)
3ðReÞR3
L2 k6i h
F6 ðRÞ ¼ h i 12L2 2 ðR=ki Þ9
18R5 1 þ ðR=ki Þ3 (A6)
i
þ 6L2 ðR=ki Þ6 6 L2 1 ðR=ki Þ3 þ 4 ;
L2 k6i h
F7 ðRÞ ¼ h i 21L2 5 ðR=ki Þ9
9R5 1 þ ðR=ki Þ3 (A7)
i
þ 15L2 6 ðR=ki Þ6 þ 6L2 þ 3 ðR=ki Þ3 þ 4 ;
L2 k6i h
F8 ðRÞ ¼ 4
39L2 11 ðR=ki Þ6
18R i (A8)
þ 12L2 þ 14 ðR=ki Þ3 2 ;
L2 k6i h
F9 ðRÞ ¼ 4
12L2 2 ðR=ki Þ6
18R i (A9)
þ 12L2 4 ðR=ki Þ3 2 ;
S0 3kPgi
F10 ðRÞ ¼ : (A10)
R2 R3kþ1
Mathematical Theory and Numerical Simulation of Bubbly Cavitating Nozzle Flows 21
Appendix B
The boundary conditions presented for the cases (a) and (b) in quasi-one-dimensional
flows should be converted to the boundary conditions for the unsteady acceleration
field in order to pose a two-point boundary value problem for the unsteady accelera-
tion field given by Eq. 19. For this reason, using Eq. 23 for the pressure distribution in
quasi-one-dimensional flows, we can arrive at the boundary conditions for the
unsteady acceleration field corresponding to the inlet and exit pressure boundary
conditions as
@a 1 dA
þ ai ¼ Qi ðtÞ (B1)
@x i A dx i
and
@a 1 dA @A1 1 dA
þ ae ¼ Qe ðtÞ þ ðA1 Þe
@x e A dx e @x e A dx e
ð xe
sðx; tÞA2 ðx; tÞ þ uð@ 2 u=@x2 Þð@A2 =@xÞ
dx
xi Wðx; tÞ
(B2)
1 dA @u d 1 dA
Qi ðtÞ ¼ Ui þ Ui
A dx @x dx A dx i
h i i
i
k3i ð6L 1ÞðRi =ki Þ6 þ ð6L2 2ÞðRi =ki Þ3 1
2
i
3R3i ½2þð3L2 1ÞðRi =ki Þ3
2
1 dA @u
Ui þ
A dx i @x i
6Ri S pgi
þ i ðpv Þi 0 þ 3k Pi
L2 k3i ½2þð3L2 1ÞðRi =ki Þ3 Ri Ri
h i
8 1 þ ðRi =ki Þ3
1 dA @u
i Ui þ (B3)
L2 ðReÞR2 ½2þð3L 1ÞðRi =ki Þ
2 3 A dx i @x i
i
22 Appendix B
1 dA @u d 1 dA
Qe ðtÞ ¼ Ue þ Ue
A dx e @x e dx A dx e
h i
k3i ð6L2 1ÞðRe =ki Þ6 þ ð6L2 2ÞðRe =ki Þ3 1
i
3R3e ½2þð3L2 1ÞðRe =ki Þ3
2
1 dA @u
Ue þ
A dx e @x e
6Re S0 pgi
þ i ðpv Þe þ 3k Pe
L2 k3i ½2þð3L2 1ÞðRe =ki Þ3 Re Re
h i
8 1 þ ðRe =ki Þ3
1 dA @u
i Ue þ
L ðReÞRe ½2þð3L 1ÞðRe =ki Þ
2 2 2 3 A dx e @x e
@A1 1 dA
þ þ ðA1 Þe
@x e A dx e
ð xe
sðx; tÞA2 ðx; tÞ þ uð@ 2 u=@x2 Þð@A2 =@xÞ
dx
x Wðx; tÞ
i
@A2 1 dA
þ ðA2 Þe
@x e A dx e
ð xe
sðx; tÞA1 ðx; tÞ þ uð@ 2 u=@x2 Þð@A1 =@xÞ
dx (B4)
xi Wðx; tÞ
The time dependent functions K1(t) and K2(t) for case (a) and case (b) boundary
conditions in quasi-one-dimensional flows then follow as:
Case (a): The inlet flow speed and exit pressure are specified.
In this case, the functions K1(t) and K2(t) satisfy the following equations:
dUi
K1 ðtÞðA1 Þi þ K2 ðtÞðA2 Þi ¼ ai ðtÞ ¼ (B5)
dt
@A1 1 dA @A2 1 dA
K1 ðtÞ þ ðA1 Þe þ K2 ðtÞ þ ðA2 Þe ¼ Qe ðtÞ
@x e A dx e @x e A dx e
(B6)
and
@A1 1 dA dUi
þ ðA1 Þe ðA1 Þi Qe
@x A dx e dt
K2 ðtÞ ¼ e
(B8)
Da ðtÞ
and
@A1 1 dA
K1 ðtÞ þ ðA1 Þe
@x e A dx e
@A2 1 dA
þ K2 ðtÞ þ ðA2 Þe ¼ Qe ðtÞ (B11)
@x e A dx e
@A2 1 dA
þ ðA2 Þi Qe ½Db ðtÞ1 (B12)
@x i A dx i
and
@A1 1 dA
K2 ðtÞ ¼ þ ðA1 Þe Qi
@x A dx e
e
@A1 1 dA
þ ðA1 Þi Qe ½Db ðtÞ1 (B13)
@x i A dx i
24 Appendix B
References
1. Tangren RF, Dodge CH, Seifert HS. Compressibility effects in two-phase flow. J Appl Phys.
1949;20:637–45.
2. Ishii R, Umeda Y, Murata S, Shishido N. Bubbly flows through a converging-diverging nozzle.
Phys Fluids A. 1993;5:1630–43.
3. van Wijngaarden L. On the equations of motion for mixtures of liquid and gas bubbles. J Fluid
Mech. 1968;33:465–74.
4. van Wijngaarden L. One-dimensional flow of liquids containing small gas bubbles. Ann Rev
Fluid Mech. 1972;4:369–96.
5. Noordzij L, van Wijngaarden L. Relaxation effects, caused by the relative motion, on shock
waves in gas-bubble/liquid mixtures. J Fluid Mech. 1974;66:115–43.
6. Wang YC, Brennen CE. One dimensional bubbly cavitating flows through a converging-
diverging nozzle. ASME J Fluids Eng. 1998;120:166–70.
7. Delale CF, Schnerr GH, Sauer J. Quasi-one-dimensional steady-state cavitating nozzle flows.
J Fluid Mech. 2001;427:167–204.
8. Pasinlioğlu Ş, Delale CF, Schnerr GH. On the temporal stability of quasi-one-dimensional
steady-state bubbly cavitating nozzle flow solutions. IMA J Appl Math. 2009;74:230–49.
9. Preston AT, Colonius T, Brennen CE. A numerical investigation of unsteady bubbly cavitating
nozzle flows. Phys Fluids. 2002;14:300–11.
10. Saffman PG. Vortex dynamics. Cambridge: Cambridge University Press; 1992.
11. Brennen CE. Cavitation and bubble dynamics. Oxford: Oxford University Press; 1995.
12. Wang YC, Chen E. Effect of phase relative motion on critical bubbly flows through a
converging-diverging nozzle. Phys Fluids. 2002;14:3215–23.
13. Mørch KA. Cavitation nuclei and bubble formation: a dynamic liquid-solid interface problem.
ASME J Fluids Eng. 2000;122:494–8.
14. Delale CF, Hruby J, Marsik F. Homogeneous bubble nucleation in liquids: the classical theory
revisited. J Chem Phys. 2003;118:792–806.
15. Delale CF, Okita K, Matsumoto Y. Steady state cavitating nozzle flows with nucleation.
ASME J Fluids Eng. 2005;127:770–7.
16. Brennen CE. Fission of collapsing cavitation bubbles. J Fluid Mech. 2002;472:153–66.
17. Delale CF, Tunç M. A bubble fission model for collapsing cavitation bubbles. Phys Fluids.
2004;16:4200–3.
18. Blake JR, Gibson DC. Cavitation bubbles near boundaries. Ann Rev Fluid Mech.
1987;19:99–123.
19. Kubota A, Kato H, Yamaguchi H. A numerical study of unsteady cavitation on a hydraulic
section. J Fluid Mech. 1992;240:59–96.
20. Nigmatulin RI, Khabeev NS, Nagiev FB. Dynamics, heat and mass transfer of vapor-gas
bubbles in a liquid. Int J Heat Mass Tran. 1981;24:1033–44.
21. Prosperetti A, Crum LA, Commander KW. Nonlinear bubble dynamics. J Acoust Soc Am.
1988;83:502–14.
Mathematical Theory and Numerical Simulation of Bubbly Cavitating Nozzle Flows 25
22. Prosperetti A. The thermal behavior of oscillating gas bubbles. J Fluid Mech. 1991;222:
587–616.
23. Delale CF. Thermal damping in cavitating nozzle flows. ASME J Fluids Eng.
2002;124:969–76.
24. Franc JP, Michel JM. Fundamentals of cavitation. Dordrecht: Kluwer; 2004.
Experimental Study of the Inertial Motion
of Supercavitating Models
Abstract
The paper gives a brief overview of various types of available facilities for the
experimental study of the high-speed inertial motion of supercavitating bodies in
water. The paper reports the procedure of the experimental studies of high-speed
supercavitating models which have been conducted at the Hydrodynamics
Laboratory of the Institute of Hydromechanics of the National Academy of
Sciences of Ukraine under the direction of Yu.N. Savchenko since 1990. The
design philosophy of the electrochemical-catapult model firing system and the
motion parameter recording system is described. The paper gives examples of
model firing and reports the values of the initial parameters, video-recording
data on the motion of a supercavitating model, and motion parameter values for
models moving with a system of shock waves.
1 Introduction
Models can be put in high-speed motion through water in a number of ways [1, 2].
The types of existing facilities differ in the method of production of the energy
delivered to the model to speed it up.
Thus, a controlled-pressure ballistic chamber was built at the Naval Ordnance
Test Station, Pasadena, the USA, in 1951. It serves to study the water entry, water
exit, and underwater motion of engineless projectiles. A pneumatic piston catapult
system fires models of diameter 50.8 mm and mass up to 530 g from a tube into the
chamber with water entry and exit speeds up to 36 and 24 m/s, respectively.
The 0.9 m square chamber of length 2.4 m has glass windows on three sides and
can be set at an angle of 5–90 to the horizontal. The gas pressure in the chamber
over the water surface can be varied from the vapor pressure to 1.5 atm.
Observations are made using stroboscopic photography. To measure the variation
of the projectile angular velocity in water entry, use is made of a special camera
with optical scanning to compensate for the motion of the image relative to the film.
The more recent ballistic chamber at the California Institute of Technology has
an electromagnetic catapult system with controlled atmosphere and allows one to
study water entry and exit at different angles with waves on the free surface. Model
projectiles of diameter 25.4 mm can be fired (at the center of the chamber)
transverse to the water–gas interface up and down. Stainless steel models of
diameter 25.4 mm are fired underwater at a speed of about 27 m/s, the speed-up
distance being 50 mm. Increasing the energy to 54,000 W·s increases the speed to
130 m/s. The speed-up time can be varied by varying the circuit parameters, and an
oscillatory motion can be imparted to the model.
The largest controlled-pressure ballistic chamber is installed at the Naval Ord-
nance Laboratory in White Oak, Maryland, the USA, and it serves to test engineless
models. One powder gun fires into the water models of diameter 76.2 mm and mass
5.05 kg at 900 m/s. The other gun has a barrel of caliber 102 mm. Models of
diameter up to 76 mm are loaded in a strong titanium cartridge. The tray is stopped
by an aluminum braking nozzle of diameter 80 mm at the end of the barrel while the
model continues to fly. All operations: model firing, speed calculation, and photo-
graphy with the use of flash tubes – are performed automatically by a preset program.
The chamber length and width are 30 and 10.5 m, and the water depth is 19.5 m.
The Hydrodynamics Laboratory at the California Institute of Technology has a
centrifugal catapult system mounted inside a sealed reservoir with water and a
gaseous atmosphere over it. Models are fired in a vertical plane at any desired angle
with any angle of attack in the range 10 at any speed up to 75 m/s. The water
surface has area 3.6 9.16 m, and the water depth is 3.05 m. This special-purpose
facility makes possible a variety of experiments both with self-propelled projectiles
and with projectiles moving on inertia.
Experimental facilities to test high-speed inertial models have been built and are
currently being built in a number of European and Asian countries too. Some results
of foreign experimental studies of supercavitating bodies moving at high speeds are
presented in [3–5]. At the Institute of Hydromechanics of the National Academy of
Sciences of Ukraine, a firing bench has been in service since 1990.
the model is stopped using a metal shield or an obstacle filled with a soft material
such as sand, wood, etc. so that the model may not be damaged in stopping.
Models are fired using an electrochemical catapult, which uses ecologically
clean components: water, compressed air, hydrogen, and electric current and
provides high firing energy at transonic speeds (the sound speed in water at T ¼ 6 C
is 1,440 m/s). The action of the electrochemical (gas–vapor) catapult is described
and estimates of firing efficiency are given in [6]. The firing bench comprises a
hydraulic, a pneumatic, an electric, and a measuring system.
The hydraulic system (Fig. 1) serves to fill the water tunnel channel 10 with water
from a basin 7 through a pressure pipe 6 using a pump 2 and to empty the channel 10
through a drain pipe 5 after the experiment. Valves 3 and 4 control the pressure and
the drain pipe, respectively.
The electric circuit of the bench is shown in Fig. 3 where: 1 – personal computer;
2 – control panel; 3 – video camera; 4 – power unit; 5 – DC generator; 6 – fuse wire;
7 – electrolyzer; 8 – catapult; 9 – window; and 10 – lighting.
The electric system serves to accumulate the firing energy by water electrolysis,
fire the gas mixture, start the catapult, and record the test data.
The current and the voltage are checked on the amperemeter and voltmeter of the
control panel 2. The electric circuit also provides for the synchronous operation of
the personal computer 1, the control panel 2, the video camera 3, and the lighting 10
when recording the test data.
The energy is accumulated using the electrochemical process of water decom-
position into oxygen and hydrogen by the familiar chemical reaction [7]:
In the process, oxygen is liberated at the anode, and hydrogen is liberated at the
cathode. According to Faraday’s law, the mass of the oxygen and hydrogen
produced at the electrodes will be
M ¼ Z I t;
where DH289 ¼ 241.83 kJ/mole is the water formation heat [kJ/mole] at 289 К
(25 C) [8]; MH2O is the water (water vapor) mass in grams; V0 [m3] is the
combustion chamber volume; and P0 [Pa] is the combustion chamber pressure
prior to firing.
The consumed energy will be
ES ¼ I U t;
where I is the circuit current, U is the circuit voltage, and t is the chamber charging
(electrolysis) time.
The measuring system allows one to check the catapult charging parameters. They
are the electrolysis current, voltage, and time and the chamber pressure prior to and
after electrolysis. The measuring system also records the model motion in the
channel using a system of sensors and high-speed photography. Initially, SKS-
1 M and Pusk-16 high-speed 16-mm movie cameras with a frame frequency up to
5,000 frames/s were used for this purpose. Now we use an X-Sheam XS4 video
camera (Integrated Design Tools, Inc.) with a frame frequency of 1,000–20,000
frames/s.
32 N.S. Fedorenko et al.
The rather large illuminated area (0.8 0.8 m) requires a high lighting power of
about 10 kW. Figure 4 shows a schematic of the experimental setup with video
recording where: 1 – gas generator; 2 – catapult; 3 – moving model; 4 – video
camera; 5 – personal computer; 6 – control panel; 7, 8 – lighting; 9 – water wind
channel; 10 – window; and 11 – sensors. The instantaneous speed of models is
measured by two methods:
– From the recorded video frames by the technique described in [9]
– By measuring the time it takes for the model to travel the distance between two
measuring planes
3 Model Design
Fig. 5 High-speed
supercavitating model
V2
where m is the model mass (kg); a ¼ 0 is the acceleration; Lc is the barrel length;
2L
and Dm is the model aft diameter.
The test results listed in Table 1 show that in acceleration the model bottom
develops stresses of 235–785 MPa. Such high stresses call for special steels with
ultimate stresses of the order of 500–800 MPa.
The hydrodynamic drag force Fn and stresses sn acting on the cavitator of a
model can be estimated as
4 Test Results
Systematic tests on the IHM of NASU’s firing bench have been conducted since
1990. Over this period, the following has been investigated:
– The unsteady processes of high-speed water entry and supercavity inception
– The mechanisms of interaction of high-speed supercavitating models with vari-
ous obstacles
– The features of interaction between high-speed supercavitating models in group
motion
Starting in 1993, the obtained results have been published in Refs. [12–20].
Below are some of the test results obtained on the IHM of NASU’s upgraded firing
bench (see Table. 1). Table 1 gives the catapult charging parameters and the model
speeds calculated from the recorded video data for a series of tests. Among the
firings shown in the table, of especial interest is firing No 6 because in this case the
water sound speed a ¼ 1,422 m/s at water temperature T ¼ 4 C was exceeded.
Figure 6 shows video frames of the motion of the supercavitating model,
wherein the supercavity shape and shock waves can be seen.
34
The model speed measured from the frames was V ¼ 1,550 m/s, and the attained
Mach number was:
V 1550
M¼ ¼ ¼ 1:09:
a 1422
From the shock wave shape in Fig. 7, the Mach number was estimated as [19]
1
M¼ ¼ 1:082:
sin 67:5
References
1. Knapp R, Daily J, Hammitt F. Cavitation (in Russian). Moscow: Mir Publishers; 1974.
2. Gorshkov AS, Rusetsky AA. Cavitation tunnels (in Russian). Leningrad: Sudostroenie; 1972.
3. Kirscner IN. Results of selected experiments involving supercavitating flows. VKI/RTO
Special Course on Supercavitation. Brussels: Von Karman Institute for Fluid Dynamics; 2001.
4. Hrubes JD. High-speed imaging of supercavitating underwater projectiles. Exp Fluids.
2001;30(1):57–64.
5. Schaffar M, Ray C, Boeglen G. Behaviour of supercavitating projectiles fired horizontally in a
water tank: theory and experiments. 35th AIAA Fluid Dynamic Conference and Exhibit; 6–9
June 2005. Toronto; 2005.
6. Deinekin YuP. Firing of bodies using a gas–vapor catapult (in Russian). Gidromekhanika.
1993;66:40–4.
7. Goronovsky IT, Nazarenko YuP, Nekryach EF. Chemistry handbook (in Russian). Kiev:
Naukova Dumka; 1974.
8. Yavorsky BM, Detlaf AA. Physics handbook (in Russian). Moscow: Nauka; 1974.
9. Konovalov NA, Lakhno NI, Putryk ND, Skorik AD. Still and motion picture photography
methods in technical mechanics (in Russian). Kiev: Naukova Dumka; 1990.
10. Savchenko YuN, Semenenko VN, Putilin SI, Naumova EI. Software system to simulate the
motion of supercavitating bodies in water (in Russian). Matematicheskie Mashiny i Sistemy.
1999;2:48–57.
11. Semenenko VN. Software for designing the supercavitating vehicles. Proceedings of the 10th
International Scientific School “High Speed Hydrodynamics (HSH-2008)”; 10–14 September
2008, Cheboksary; 2008. p. 241–52.
12. Savchenko YuN, Semenenko VN, Serebryakov VV. Experimental study of developed cavity
flows at subsonic flow velocities (in Russian). Doklady AN Ukrainy. 1993;2:64–9.
13. Savchenko YuN, Semenenko VN, Serebryakov VV. Experimental verification of asymptotic
formulas for axisymmetric cavities at s ! 0 (in Russian). Problems in high speed hydrody-
namics. Cheboksary: Chuvash University; 1993. p. 225–30.
14. Savchenko YuN, Vlasenko YuD, Semenenko VN. Experimental investigations of high-speed
cavity flows (in Russian). Gidromekhanika. 1998;72:103–11.
15. Vlasenko YuD. Experimental investigations of high-speed unsteady supercavitating flows.
Proceedingsof the Third International Symposium on Cavitation. Vol. 2. Grenoble; 1998.
p. 39–44.
16. Savchenko YuN, Morozov AA, Savchenko VT, Semenenko VN. Mathematical models of the
motion of supercavitating bodies in water at transonic speeds and systems for its implementa-
tion (in Russian). Matematicheskie Mashiny i Sistemy. 1999;1:3–15.
Experimental Study of the Inertial Motion of Supercavitating Models 37
17. Vlasenko YuD. Experimental investigations of supercavitation flows at subsonic and transonic
velocities. Proceedings of the International Summer Scientific School “High Speed Hydrody-
namics”; June 2002. Cheboksary; 2002. p. 197–204.
18. Savchenko YuN, Semenenko VN, Putilin SI, et al. Designing the high-speed supercavitating
vehicles. Proceedings of the 8th International Conference on Fast Sea Transportation
(FAST’2005); 27–30 June 2005, St. Petersburg; 2005. ISBN 5-88303-045-9.
19. Savchenko YuN, Semenenko VN, Putilin SI, et al. Some problems of the supercavitating
motion management. Sixth International Symposium on Cavitation CAV2006; September
2006. Wageningen; 2006.
20. Savchenko YuN, Zverkhovsky AN. Procedure of experimental study of the high-speed motion
of supercavitating inertial models in water (in Russian). Prikladnaya Gidromekhanika.
2009;11(4):69–75.
Two Dimensional Theory of Cavitation
Flows Under Ship Bottoms
Michael V. Makasyeyev
Abstract
The two dimensional theory of cavity flows under ship bottoms is based on the
linear theory of wave motions of ideal incompressible fluid. The cases of high
speed and displacement ships are considered. In the case of high speed ship the
problem of planing hull with step and cavity at free fixing of trim angle and draft,
unknown shape and length of cavity and wetted borders of hull is solved. The
possibilities of modeling of ship hydrodynamic characteristics changing with the
help of cavity pressure control are shown. A reduction of the wave resistance can
be a result of such changes. In the case of displacement ship the cavitation flow
model behind wedge under solid wall is considered. It is shown that the gravity
waves with decreasing amplitude on cavity boundary are generated if the cavity
on the horizontal wall is closed. In theoretical model, the existence of countable
number of cavity lengths is possible. The characteristics of cavity shapes at
negative cavitation numbers are determined.
1 Introduction
The cavity flows under ship bottom are created in the purpose to reduce the drag and
to control the hydrodynamic characteristics (see examples [1–3]). The special steps
for a cavity creation on high speed hulls are designed. The cavities occur in the
areas behind steps as a result of high speed motion or pumping of air. On the
bottoms of displacement type of ships the artificial ventilated cavities are created
essentially. It is possible to control pressure distribution on bottom and make the
additional pressure or underpressure. In addition, the hull constructions with steps,
air cavities and controllable angles of installation allow obtaining the new
hydrodynamic effects associated with wave generation. Such effects can cause the
reduction of wave resistance and implementation of academician G. E. Pavlenko’s
idea [4] about wave energy regeneration in system of planing surfaces when some
surfaces use the energy which was generated by other surfaces at their right relative
position.
The research of real artificial cavity flow under ship bottom represents suffi-
ciently complicated theoretical and experimental problem. Different aspects of
vapor and artificial cavity flows near solid walls were researched in works by
R. Knapp [5], G.V. Logvinovich [6], Y.N. Savchenko [7, 8]. The works of A.A.
Butuzov [9–12] are dedicated to modeling of cavities under ship bottom. The
numerical model in these works is based on the two dimensional linear theory of
cavity flow behind wedge under solid horizontal flat wall. The calculation results
are compared with experiment which was obtained in a hydrodynamic channel for a
cavity under a flat plate with side discs. The description of Butuzov’s method can be
found in books [13, 14]. Later this method was also used by K. Matveev, see, for
example, [15].
Butuzov has spread his idea of solution method for the problem with a cavity
under wall to the modeling cavity on planing surface behind step [11, 12]. His
approach is based on simplified model of planing with the use of Ryabushinsky’s
cavitation scheme for coupling solid and free boundaries. The free boundaries of
fluid in this model are represented as linear solid wall and lengths of wetted
segments are given. Cavity length, unknown in physical problems, is given in
Butuzov’s method and cavitation number is defined from the problem solution. It
explains that the cavity length in mathematical problem defines the free boundary
and makes the problem nonlinear and in this time the cavitaton number enters
linearly into the problem.
In reality the cavitation phenomenon on moving ship bottom differ from
simplified models. The main fact is that the moving ship position on the water
surface cannot be arbitrarily defined. This position is defined by motion speed,
displacement amount, mass distribution and shape of hull. If the cavity exists under
the bottom, its shape and length are unknown. The sizes of cavity will depend
substantially on the pressure in the cavity, i.e., the cavitation number, and other
parameters – Froude number, bottom geometry. From physical conditions of
cavitation, it follows that the cavitation number is given. It is defined by the
saturated vapor pressure for natural vapor cavitation and by the cavity pressure
that is artificially created in the case of artificial gas ventilation. Thus, the cavitation
number can be positive or negative. Accordingly, the physical effects caused by
cavitation will be different. Particularly, at a negative cavitation number additional
backup is created which decreases the draft. Consequently, it is possible to control
the hydrodynamic characteristics of the high speed and the displacement type ships
with the help of steps with cavities behind it and artificial ventilation of hull bottom
segments.
The computational methods are need for comprehensive research of cavities
influence on hydrodynamic characteristics of hulls and these methods must corre-
spond with physical phenomena. The calculation method for planing hydrofoil that
Two Dimensional Theory of Cavitation Flows Under Ship Bottoms 41
gives possibility to define the wetted length, pressure distribution and trim angle on
given displacement, geometry and center of mass position is shown in [16]. This
method is suitable for calculation of system of planing surfaces that have segments
with cavities which are the boundaries with a known pressure. The general
approach to the study of system of planing surfaces is presented in work [17].
The present work consists of two logical parts united by common mathematical
theory. The theory is based on reduction of original problem for velocity potential
with linear boundary conditions to the singular integral equations. In the first part
the two dimensional theory of the planing hull with a cavity behind a step under the
bottom is presented. The examples of numerical results that show the model
possibilities are given. These results and physical effects are discussed. In second
part, the Butuzov’s approach to the modeling of the flow under the ship bottom with
a cavity behind a wedge under infinite horizontal wall is developed. This approach
can be used in the case of the displacement air cavity ship when the cavity area is
small in comparison with the hull bottom size. Improvement of this approach that
can define the cavity length at giving the cavitation number and the Froude number
is presented.
The part of this work – the results related to the problem of step planing hull with
cavity was presented on International Symposium on Cavitation CAV2009 in
Michigan university (Ann Arbor, USA) and published in proceedings of Sympo-
sium [18].
2 Statements of Problems
Fig. 1 Scheme of a planing boat with a gas cavity under the bottom after the step
42 M.V. Makasyeyev
The weight (volume displacement) of the boat is D and its center of mass is
situated by distance b from the trailing edge of the second planing surface. The
values D and b are given.
The motion of the boat is modeled by the motion of a system of two flat plates
rigidly joined into an integral structure, the x-projections of wetted sections of
plates are segments ½A1 ; B1 and ½A2 ; B2 . It is assumed that the angles of the plates
with the move direction a1 and a2 are small and assumptions of the linearized
theory of liquid wave motion are true.
The mathematical model of the physical problem is a boundary-value problem
for the perturbed velocity potential, and the boundary conditions are transferred to
the axis y ¼ 0. On the segments ½Ai ; Bi , i ¼ 1; 2, the unknown pressure difference –
the functions gi ðxÞ ¼ ðpðx; 0Þ p0 Þ=rV02 , x 2 ðAi ; Bi Þ, i ¼ 1; 2, are defined where
p0 is the pressure on the free boundary, pðx; yÞ is the pressure in the liquid, and r is
the liquid density. The pressure pc in the cavity aft of the step is specified by the
cavitation number s ¼ 2ðp0 pc Þ=rV02 . There is no pressure difference on the free
surface at x < A1and x > B2 , and the free surface shape is unknown. The Froude
pffiffiffiffiffiffiffiffiffiffiffi
pffiffiffiffiffi
number Fr ¼ V0 ga is defined by the characteristic length a ¼ 3 D=rg where g
is the gravity acceleration.
It is assumed that the free surface boundary and the plate boundaries represent
one streamline. The functions which describe the forms of flow on segments are
designated as fi ðxÞ ¼ hi þ ki x, i ¼ 1; 2, hi are drafts, ki ¼ tan ai . The amounts Da ¼
a2 a1 and Dh ¼ h2 h1 (height of step) are given as constructive parameters in
conditions of linear approximation.
’x ; ’y ! 0; y ! 1; (4)
8
>
> g ðxÞ; x 2 ½Ai ; Bi ;
< i
gðxÞ ¼ 0; x < A1 ; x > B2 ;
>
: s;
>
x 2 ½B1 ; A2 ;
2
is the function that defines the dimensionless pressure difference along the whole
liquid surface. Equation 2 is the kinematic condition for smooth flow past the
boundary, (3) is the dynamic condition for the pressure on the boundary, which is
the Bernoulli equation, (4) is the condition for disturbance attenuation at a great
depth, and the condition (5) means that the flow potential is specified at infinity in
front of the boat – undisturbed flow or steady-state independent waves.
The problem of the cavitation flow under the bottom of a displacement ship can be
considered as a particular case of the previous problem at zero angles a1 and a2 . The
segment ½A1 ; B1 will be presented as polygonal line in view of horizontal semi
infinite segment and short segment, which is cheek of edge, under angle of slope.
In this case the axis of abscissas must pass on the level that corresponds with
immersion depth of bottom. However, this level can be considered as zero and the
pressure and cavitation number needs to be corrected by corresponding addition.
The graphic illustration of the physical problem is shown in Fig. 2.
The characteristic length a can also be defined with the use of the ship displace-
ment and some characteristic value, for example, the wedge length c. It is assumed
that jd=cj < < 1, where d is the wedge height.
At high Froude numbers or at n ! 0 this problem is equal to the problem of
symmetrical cavitation flow of a weightless fluid over an edge. The Froude numbers
can be sufficiently small in the case of the displacement ship.
The effects of wave generation on cavity boundary would be expressed strongly
with reduction of Froude number. The sufficient long cavities can be generated at
intensive air ventilation or gas pumping. It is assumed that there can be long cavities
of wave shape whose boundary can intersect the wall level line. It is possible
Fig. 2 Scheme of physical statement of the problem of the cavitation flow behind a wedge under
the bottom of a displacement ship
44 M.V. Makasyeyev
physically if a hollow exists on the wall (ship bottom) in area in front of cavity
closure point on segment ½B1 ; A2 , in which the cavity goes into freely without
contact of walls.
The mathematical problem of cavity flow under bottom can be written formally in
the form (1)–(5). In this case the exceptions for conditions (2) and (3) will be the
point of flow return on beginning of wedge cheek and the junction point of the
cavity and the solid wall. The potential ’0 ðx; yÞ is identically equal to zero. It will
correspond to the absence of independent steady waves ahead.
Note that boundary conditions on segment ½B1 ; A2 correspond to the conditions
on the boundary of the cavity where the pressure is given and it is constant and the
boundary of the cavity is a part of the stream line. The presence of the wall on this
segment or its shape does not matter formally.
The problem (1)–(5) is solved using the Fourier method for the construction of
fundamental solutions [19]. After a conversion to the generalized functions in
(1)–(5) and a construction of the fundamental solution of the Laplace generalized
equation it is possible to obtain the relationship [20] with the use of the boundary
conditions (2)–(3):
where H ðlÞ ¼ F½ðxÞðlÞ and GðlÞ ¼ F½gðxÞðlÞ are generalized Fourier trans-
forms of functions ðxÞ and gðxÞ respectively. The functional relations (6) connect
the generalized Fourier transforms of pressure function and boundary form func-
tion. The velocity potential as an auxiliary function is excluded.
The inverse Fourier transformation gives
ð
1
1 ðsÞ
ds þ nðxÞ ¼ gðxÞ; 1 < x < 1; (7)
p ðx sÞ2
1
where the singular integral exists in the meaning of Hadamard. The relationship (7)
is true in generalized functions. The classical functions ðxÞ, gðxÞ and these
Two Dimensional Theory of Cavitation Flows Under Ship Bottoms 45
derivatives are sectionally continuous functions and if the boundary form ðxÞ or
this derivative are known on some section then pressure function gðxÞ is unknown
on this section. The opposite is also true. If the pressure is known in some section
then the form function is unknown there. Equation 7 gives the integral equation for
problems where the pressure is known on the part of the boundary and the shape
function is finite. The integral will be replaced by an integral with finite limits in
this case.
For other problems it is necessary to have expression for function ðxÞ per gðxÞ,
i.e., it is necessary to resolve the Eq. 7. The resolution of (7) can be found with the
use of the fundamental solution construction by Fourier method. Thus, the formal
solution of Eq. 6 in generalized functions should be found. The generalized
function
1
H ðlÞ ¼ GðlÞ reg þ ½AGðnÞ þ A0 dðl nÞ þ ½BGðnÞ þ B0 dðl þ nÞ;
jlj n
(8)
ð
1 ð
1 ð
1
A inðxsÞ B
ðxÞ ¼ gðsÞQðn; x sÞds þ gðsÞe ds þ gðsÞeinðxsÞ ds
2p 2p
1 1 1 (9)
A0 inx B0 inx
þ e þ e
2p 2p
where
1 1h p i
Qðn; xÞ ¼ F1 reg ¼ cos nxCinjxj þ sin njxj þ Sinjxj ;
jlj n p 2
Hence,
ð
1 ð
1 ð
1
A inðxsÞ B
ðxÞjx!1 ¼ gðsÞsin njx sjds þ gðsÞe ds þ gðsÞeinðxsÞ ds
2p 2p
1 1 1
1
ð 1
ð
A i B i
¼ gðsÞeinðxsÞ ds þ gðsÞeinðxsÞ ds:
2p 2 2p 2
1 1
(11)
It would be no waves on the left in infinity if one supposes that A ¼ pi, B ¼ pi.
Therefore, (9) can be written as follows:
ð
1
Here a0 and b0 are real constants which define the amplitude of independent
waves.
The differentiation of (12) gives the equation that corresponds to the boundary
condition (2):
ð
1
1 1
gðsÞ þ nRðn; x sÞ þ np cos nðx sÞ ds
p xs
1
¼ 0 ðxÞ nða0 cos nx b0 sin nxÞ; 1 < x < 1; (13)
where Rðn; xÞ ¼ p1 p2 sgnðxÞ þ SinðxÞ cos nx Cinj xj sin nx .
The Eqs. 7, 12 and 13 are the base for resolution of formulated problems.
In the problem of the planing hull, the Eq. 13 is used for determination of the
pressure function and (12) is used for determination of the free surface shape. The
integrals in (12) and (13) are replaced with the integrals between the finite limits A1
and B2 because the pressure is zero outside of ½A1 ; B2 .
The forces and the moments balance conditions [16] must be added to the Eq. 13:
ð2
B
gðxÞdx ¼ n; (14)
A1
Two Dimensional Theory of Cavitation Flows Under Ship Bottoms 47
ð2
B
The three Eqs. 13–15 resolve completely the main problem of determination of
pressure distribution, wetted length and trim angle in case of one single planing
hull.
If we have a cavity under bottom there are two rigidly bound planing surfaces
with unknown lengths. Therefore, the Eq. 13 will be written as a system of two
equations on each of surfaces. As result the follow integral equations system will be
obtained on the base of (13)–(15):
ð2
A ð1
B ð2
B
1 s 1
g1 ðsÞK ðn; x sÞds K ðn; x sÞds þ g2 ðsÞK ðn; x sÞds
p 2p p
A1 A2 B1
ð2
A ð2
B ð2
B
1 s 1
g1 ðsÞK ðn; x sÞds K ðn; x sÞds þ g2 ðsÞK ðn; x sÞds
p 2p p
A1 B1 B1
ð2
A ð2
B
s
g1 ðsÞds ðL l2 Þ þ g2 ðsÞds ¼ n: (18)
2
A1 B1
ð2 ð2
sh i
A B
ð2
A
g1 ðsÞ Q n;L þ l1 l2 s þ sin n L þ l1 l2 s Q n;s þ sin n s ds
A1
ð1
B
s
Q n;L þ l1 l2 s þ sin n L þ l1 l2 s Q n;s þ sin n s ds
2
A2
ð2
B
þ g2 ðsÞ½ Q n;L þ l1 l2 s þ sin n L þ l1 l2 s
B1
Q n;s þ sin n s ds ¼
¼ L l2 tg a1 þ Da l1 tga1 þ Dh;
(20)
The relation (7) gives the integral equation for the problem of cavity behind wedge
under solid wall:
ð
1
1 ðsÞ s
ds þ nðxÞ ¼ ; B1 < x < A2 : (21)
p ðx sÞ 2 2
1
ðl ðx ð0
1 qðsÞ s 1 q0 ðsÞ
ds þ n qðsÞds ¼ nð0Þ ds; 0 < x < l; (22)
p sx 2 p sx
0 0 c
where l is cavity length, q0 ðxÞ is known derivative of wedge form function. This
equation must be satisfied together with closure cavity condition ðlÞ ¼ 0, or
ðl
qðsÞds ¼ ð0Þ: (23)
0
Two Dimensional Theory of Cavitation Flows Under Ship Bottoms 49
4 Numerical Method
The systems of integral equations (16)–(20) and (22)–(23) may be solved by any of
the familiar methods of solution of singular integral equations. The key feature is
that they are parametrically nonlinear in the unknowns l and l1 , l2 . The use of any
numerical method gives a system of algebraic equations of the form
AX ¼ B; (24)
where the vector X is made up by the unknown values of the functions g1 ðxÞ,
g2 ðxÞ, and a1 , which enter into the system linearly, while the elements of the matrix
A ¼ Aðl1 ; l2 Þ depend on the unknowns l and l1 , l2 , which create nonlinearity. The
vector B is made up by elements from the known values of the right-hand sides of
Eqs. 22–23.
For the solution of systems of this type, it turns out to be efficient to use the
familiar method [22] of reduction of the problem (24) to the minimum search of
quadratic functional. Particularly in the case of planing the search problem has such
view:
In this work, the singular integral equations are solved using the discrete
singularity method [23, 24], and the problem (25) is solved using the Nelder–Mead
flexible polyhedron method (downhill simplex method) [25].
50 M.V. Makasyeyev
5 Calculation Results
It makes sense to relate the scales of values of the variables that are specified in the
problem and define the geometry of the planing boat to the scale of the generated
waves. Let us estimate the free surface shape from (11) which is generated by delta-
function pressure
Ð1 pulse gðxÞ ¼ cdðxÞ of strength c moving at velocity V0 . In this
case c ¼ 1 gðxÞdx ¼ n and free surface shape will be (without taking into account
independent waves, at a0 ¼ 0, b0 ¼ 0)
We can obtain
(
0; x < 0;
lim ðxÞ ¼
x!1 2v sin nx; x > 0:
This expression allows the scale of the planing-induced waves for the linearized
theory to be estimated. For example, at Froude number Fr ¼ 2 the wave amplitude
will be 0.5 and the wave length will be 25.13, and at Fr ¼ 1:5 the amplitude and the
wave length will be 0.88 and 14.14, respectively.
With this in mind, the results of calculations at L ¼ 15 (space between the
trailing edge and the step), Dh ¼ 0:7(step height), Da ¼ 0 (the planing surfaces
fore and aft of the step are parallel), and with the center of mass situated distance
b ¼ 10 from the trailing edge are presented below.
Figure 3a–h show the cavity and free boundary shape for the cavitation number
ranging from s ¼ 0:2 to s ¼ 0:0224 at Fr ¼ 2. Bold lines show here and further
the free surface, thin line segments show the wetted boundaries of planing hull. The
free surface boundary consists of three areas. The left area begins in minus infinity
and ends in a zero point, where a y-axis passes – in the contact point of free surface
with beginning of the wetted area of planing hull. The middle area is the free
boundary of cavity after step and right-hand area is the wake border.
The cavitation number s ¼ 0:0224 for parameters mentioned above is close to
the value when second wetted length goes to zero. That means that subsequent
increase of pressure in cavity can lead to tearing of stream from back edge and
subsequent undesirable unsteady effects.
Figure 4a–f show the cavity and free boundary shape for the Fr ¼ 1:7 and the
same center of mass on the distance b ¼ 10 from the trailing edge. In this case the
critical cavitation number is close to s ¼ 0:035.
Note that not all combinations of design parameters and sizes of planing boat
and step allow constructing a physically feasible flow or making the residual of
the system (16)–(20) or the value of the goal function in (25) smaller than a preset
small positive number. However, for all the results presented in the paper the goal
Two Dimensional Theory of Cavitation Flows Under Ship Bottoms 51
Fig. 3 (continued)
52 M.V. Makasyeyev
Fig. 3 Shape of the free surface, the cavity aft of the step, and the wetted boundaries of the
planing boat at Fr ¼ 2. (a) s ¼ 0:2 (b) s ¼ 0:1 (c) s ¼ 0:05 (d) s ¼ 0:0 (e) s ¼ 0:01 (f) s ¼
0:017 (g) s ¼ 0:02 (h) s ¼ 0:0224
function did not exceed 107 . For the rather small Froude numbers for parameters
mentioned above, the flows were constructed for only negative cavitation num-
bers. The samples of such flow for Fr ¼ 1:5 and Fr ¼ 1:2 are shown on the
Figure 5a–c.
The pressure distributions along the solid boundaries are shown in Fig. 6a–d.
Figure 7 shows the cavity length versus cavitation number at the parameters
indicated above. The calculations show that the shape of the cavity behind the step
is defined by two factors, namely, by the Froude number and the cavitation number.
The Froude number defines the cavity curvature, which correlates with the curva-
ture of the generated waves, and the cavitation number defines the cavity length. At
large Froude numbers the cavity curvature is small, and it is increased with the
decrease of Froude number. The waves of the same length as the waves in the wake
of the planing boat are generated on long cavities at sufficiently small cavitation
numbers.
Two Dimensional Theory of Cavitation Flows Under Ship Bottoms 53
The pattern of contact of the cavity boundary with the second planing surface
depends on the cavity curvature in the vicinity of the contact point. The curvature is
defined by the ratio between the cavity length and the wave length at a given Froude
number. Different contact patterns are illustrated in Figs. 3–5.
At negative cavitation numbers the pressure in the cavity is higher than that on
the free surface, and thus an additional lift is developed under the bottom. It can be
seen from the plots of the cavity and free boundary shape that at negative cavitation
numbers the planing boat draft decreases. The wave amplitude in the wake
decreases too. If the second planing surface behind the step is on the trailing
wave front, the wake amplitude increases. If the surface is on the leading wave
front, the wake amplitude decreases.
Fig. 4 (continued)
54 M.V. Makasyeyev
Fig. 4 Shape of the free surface, the cavity behind the step, and the wetted boundaries of the
planing boat at Fr ¼ 1:7. (a) s ¼ 0:1 (b) s ¼ 0:05 (c) s ¼ 0:0 (d) s ¼ 0:02 (e) s ¼ 0:03
(f) s ¼ 0:035
The analysis of the calculated data shows that one can select an optimum
combination of design parameters and factors such that the wave amplitude in the
wake is a minimum. The calculated data allow supposing that the consumption of
energy to form the wake decreases due to the fact that the second planing surface
behind the step uses the energy of the wave generated by the first surface. Multistep
planing surfaces with controllable angles of setting can enhance this effect many-
fold.
Fig. 5 Shape of the free surface, the cavity behind the step, and the wetted boundaries of the
planing boat at Fr ¼ 1:5 and Fr ¼ 1:2. (a) Fr ¼ 1:5, s ¼ 0:04 (b) Fr ¼ 1:5, s ¼ 0:05
(c) Fr ¼ 1:2, s ¼ 0:06
The cavity length decreases when the cavitation number increases in positive
side. The wave length will decrease correspondingly when the cavitation number
increases at fixed and finite Froude numbers. As the shape of cavity boundary is
wave like at finite Froude numbers, one can suppose that the resolution of problem
of cavity length determination at given cavitation number and Froude number will
be nonunique. The cavity length will correspond approximately to cross points of
wave and solid wall level line behind wedge, i.e. it will be multiple approximately
with the half of wave length.
The calculations results confirm these assumptions. The calculations results
show that the cavity boundary has wave shape at any Froude number. The cavity
length is determined by cavitation number and Froude number from resolution of
search problem of minimum (25). The search interval is preset for this problem and
corresponds to the expected cavity length. The cavities of minimal length are
56 M.V. Makasyeyev
convex, they do not have bends, and actually, they are half of wave. They are
elliptical in the closure area.
The cavity shapes of minimal length behind wedge at Fr ¼ 7 and different
cavity numbers are presented on Fig. 8. The cavity ordinates and cavity numbers
are divided on nondimensional wedge height. It is the case of relatively big Froude
number and big length of gravity waves. The dependences of lengths of minimal
cavities from Froude number at two values of cavitation number are shown on
Fig. 9. The horizontal lines show the limiting values of cavity length at Froude
number which tends to infinity. These values correspond to known resolutions for
symmetrical cavity wedge in weightless fluid [16].
Fig. 6 (continued)
Two Dimensional Theory of Cavitation Flows Under Ship Bottoms 57
Fig. 6 Pressure distributions along the solid boundaries. (a) Fr ¼ 2, s ¼ 0:2 (b) Fr ¼ 1:7,
s ¼ 0:0 (c) Fr ¼ 1:5, s ¼ 0:05 (d) Fr ¼ 1:2, s ¼ 0:06
The calculations of long cavities which length is more than one wave show that
the waves amplitudes on cavity boundary are not constants along the length. They
are maximal on first wave that comes down from cavitator and they decrease
asymptotically to zero on following waves. The cavity shapes at zero cavitaton
number s=jdj ¼ 0 and Froude numbers Fr ¼ 3, Fr ¼ 1:2 and Fr ¼ 0:8 are
presented on Fig. 10a–c respectively.
The obtained results show that the question about cavity length appears in model
which takes into accounts the fluid weightiness. For any value couple of Froude
58 M.V. Makasyeyev
Fig. 9 Dependence of minimal lengths of cavities from Froude number at constants cavitation
numbers
Two Dimensional Theory of Cavitation Flows Under Ship Bottoms 59
Fig. 10 Dependence of cavity shape from Froude number at zero cavitation number
number and cavitation number the theoretical model gives possibility to build the
stationary flows with cavities which are close on horizontal wall and contain a few
semi-waves. The cavity shapes which contain N ¼ 1; 3 and 5 semi-waves at
Froude number 0.6 and zero cavitation number are shown on Fig. 11a. The pressure
distribution on wall and wedge cheek for cavities, shown on Fig. 11a, is presented
on Fig. 11b.
The pressure distribution on wedge cheek and on the wall in front of wedge is the
same at any number of semi-waves. The pressure will be different only on the wall
behind cavity. Therefore, the cavity length will be defined by flow character in area
of cavity closure on the wall.
In practice, the cavity length can be controlled by position of beginning of solid
wall segment behind cavity on the basis of calculation dates. Theoretically, it is
possible to define the maximal cavity length at given Froude number for each
cavitation number. The cavity length with odd number of semi-waves, when the
wave amplitude on last segment is less that preassigned positive number can be
60 M.V. Makasyeyev
Fig. 11 Cavity shapes (a) which consist of different numbers Nsemi-wave and the pressure
distribution (b)for these cases at Fr ¼ 0:6 and s=jdj ¼ 0
such characteristic. Note that in previous researches from cavitation behind wedge
under solid wall [9, 10], there is no information about decrease of wave amplitude
on cavity boundary.
The calculations on the Butuzov’s equations and methodic [10] were made for
the purpose to make this fact more exact. Let us show the basic results of making
calculations. The use of Ryabushinsky’s scheme for base and imagine wedges
makes the possibility to write the equations system considered in [10]:
ð1 ðx
1 qðsÞ s b x1
ds n qðsÞds þ þ ln
p xs 2 p x 1 b1
0 0
a x c
¼ na
c ln ; 0 < x < 1; (26)
p x
ð1
qðsÞds bb1 ¼ a
c; (27)
0
Two Dimensional Theory of Cavitation Flows Under Ship Bottoms 61
Fig. 13 Cavity shapes at positive, negative and zero cavitation numbers, Fr ¼ 0:8
where b1 ¼ b1 =l, c ¼ c=l, n ¼ nl, b1 is the length of imagine wedge, b is the slope
angle of imagine wedge, ais the slope angle of base wedge. Where the function
qðxÞ does not have singularities and solution of (26) is found in class of functions
which are finite on ends of integration interval. The unknowns are cavitation
number sand the slope angle of imagine wedge b in addition to function qðxÞ. The
cavity length l, the length of base wedge c and the length of imagine wedge b1 are
defined.
The cavity shapes which was obtained from solution of (26), (27) at given
lengths l ¼ 5, 10, 20 and Froude number Fr ¼ 0:8 are presented in Fig. 12. As
we can see it is also the decrease of wave amplitude in solution on scheme
Ryabushinski. The effect of decrease of wave amplitude is more brightly expressed
at bigger cavity lengths.
It is possible to estimate the difference in cavity shapes at positive and negative
cavitation numbers on data on Fig. 13. The cavities at Froude number 0.8 and
cavitation numbers 0, 0.5 and 0.5 are presented there. The cavity lengths corre-
spond to three semi-waves. The cavity shape with positive cavitation number 0.5 is
plotted with the help of marker “+”, negative with help of “”, zero cavitation
number with help of solid line. The characters of cavity curvature and cavity length
are changed at negative cavitation numbers.
The curvature can change on opposite with the increase of cavitation number in
negative side, i.e. with the increase of pressure in cavity. In this time it is not
62 M.V. Makasyeyev
succeed to satisfy the condition of cavity closure in calculations. For more correct
describing of flows at negative cavitaton numbers, it is useful to add the accounting
of interaction of fluid and gas flow in cavity into the theoretical model.
In practical problems, there can be requirements of accounting the wall or the
depth of the deep under physical conditions. With help of proposed method, it is
possible to model the necessary flow conditions by way of matching of appropriated
positive or negative cavitation number and geometry of boundaries in area of cavity
closure.
Conclusions
The presented method of modeling cavity flows on bottom of planing and
displacement hulls gives possibility to resolve the problem in real physical
statement and determinate the cavity shape and length at given cavitation
number and Froude number.
The wetted lengths of planing surfaces, trim angle and draft are determined in
the case of planing hull.
The obtained new results have shown the efficiency of approach and allowed
determining the qualitative feature of planing with cavity for real conditions
with given displacement and free trim angle.
It is shown that the proposed theory gives the possibility to estimate the
ability of natural and artificial cavitation for control to hydrodynamic
characteristics of ships.
An example of such estimation is the ability to receive the minimal wave
wake and to reduce the wave resistance. This ability can be achieved by way of
creation of necessary technical conditions with the help of gas pumping up and
appropriated steps constructions.
The research of cavitaton behind wedge under solid horizontal wall with
taking into account of gravity forces has allowed determining the wave shape of
cavity boundary with decreasing amplitude.
The existence and physical meaning of countable number of cavity lengths is
shown. The definition of maximal and minimal cavity length is given.
The tendency of cavity length increasing and bending change at negative
cavitation numbers is shown.
The obtained qualitative and numerical results can be used for design and
research of air cavity ships for the purpose to reduce the drag and wave
resistance.
References
1. Voytkunski YI, editor. The manoeuvrability of displacement ships. Hydrodynamics of ships
with dynamic principles of support. Handbook on ship theory. In three volumes. Vol. 3.
Leningrad: Sudostroenie; 1985. 544p. (In Russian).
2. Pashin VM, Ivanov AN, Kaliuzhny VG, Lyakhovitsky AG, Pavlov GA. Hydrodynamic
design of artificially-ventilated ships. International Symposium on Ship Propulsion dedicated
Two Dimensional Theory of Cavitation Flows Under Ship Bottoms 63
Vladislav P. Makhrov
Abstract
The paper presents the some results of theoretical and experimental research of
axisymmetric supercavity flow formed by a ring type wing. These flows are
known as Lighthill-Shushpanov ones. It has been simulated by distribution of the
vortex singularities on combination the “body-ring wing-cavity” surface.
Numerical solutions of a set of integral-differential equations were obtained
using a spline function for the cavity shape with positive and negative cavitation
numbers. The results of the cavitation experimental testing have been cited as an
example of the new methods of the cavity formation.
1 Introduction
s ¼ 2ðp1 ps Þ=rV1
2
; (3)
V1, and p1 are the free-stream velocity and pressure, ps – pressure in the cavity,
r is the density.
The basic formulas had been proposed by Reichardt for the various test
conditions. An expanded concept about the current status of the supercavitation
research is presented, for example, by V. Serebryakov [2] and E. Paryshev [3] as
well. The practical use of these researches was realized first in the high-speed
Russian underwater rocket “Shkval” [4, 5]. However, until the present time there is
no evidence that the cavity may be controllable yet. The present paper is an
overview of investigations of the controlled cavitation flow and its boundary
formed by the hydrodynamic singularities.
In 1940s M.J. Lighthill [6] has proposed to use the hydrodynamic singularities
for the cavity boundary formation with a negative cavitation number. Later several
plane problems about a cavity under the vortex effect have been solved in our
country recently. For example, V. Migachev [7] solved such a problem for the pair
of vortices. At the Moscow State University (MSU) V. Prokofiev [8] solved a
problem for a cavity past a flat plane with the use of the Lighthill’s method. In
Moscow Aviation Institute (MAI) E. Maraqulin [9] solved the analogous problems
using a scheme by the Efros. To confirm theoretical Lighthill’s idea in [10] the
problem about the horizontal cavitation flow with positive and negative cavitation
numbers past a body of revolution formed by an axisymmetric ring vortex was
solved.
Systematic physical experiments on cavitation flows formed according to
Lighthill method were performed at MSU by Professor Vladimir F. Shushpanov
and his colleagues in MAI [11]. Shushpanov showed first that a cavity is formed by
the hydrodynamic singularities – the ring wing (annular airfoil) and others hydro-
dynamic singularities and depends on the geometry of the ring wing, the cavitator
and their combinations.
The similar flows were theoretically and experimentally obtained by using the
ring type wing [12–14]. We named such cavitation flows as Lighthill-Shushpanov
flows. Figure 1 shows the first Lighthill’s real cavity. It is obtained by using the
practical ring of hydrodynamic singularities – a ring water scoop.
Controlled Supercavitation Formed by a Ring Type Wing 67
2 Problem Formulation
Imagine the horizontal cavitation flow behind the body of revolution formed in the
horizontal potential stream of incompressible and imponderable ideal liquid when the
ring wing is under effect of the cavity formation in the unified combination “body –
ring wing – cavity”. Here, the cavitation number s may be positive and negative.
Figure 2 illustrates the cavitation flow pattern formed by the ring wing in the
cylindrical system of coordinates (x,r,’). Point Q belongs to the body-cavitator,
cavity and ring wing; point P belongs to outward flow accordingly.
Basing on a principle of superposition, the characteristics of this combination are
represented by a sum of non-disturbance and disturbance stream functions. It may
be written as [15]:
c ¼ c1 þ cb þ cw þ cs ; (4)
where indexes b, w and s relate to the body-cavitator, ring wing and cavity,
respectively.
For the mathematical model composition and problem solution one used a contin-
uous surface of ring singularities – a vortex layer with unknown intensity g – to form
the body, ring wing and cavity as a unitary body. It is known that mathematics of a
68 V.P. Makhrov
ring type wing uses combination of the spatial arrangement and the vortex layer
method allows the ring wing description with any camber of the hydrofoil to be used
as well. However, the hydrofoil contour is limited to be smooth everywhere by
Lapunov, excepting the hydrofoil trailing edge. For this gw ¼ 0.
The stream function with radius Z of the vortex layer for each component in
formula (3) is determined during numeric investigation of this problem:
ð ð
r gð’Þ cos ’ r gðx; Þ cos ’d’dl
cv ¼ dS ¼ (5)
4p R 4p R
S S
In (5) g(’) ¼ g(x,) is the unknown density function of the vortex layer
singularities; ds ¼ d’ dl; R is the radius of the vortex layer.
The first step of using the vortex layer method is the velocity ratio estimation for the
surface of this combination in the non-continuous flow. The main value of the
stream function for this case has the form [15]:
2 3
ðð ðð
r 2
r 6 cos ’ cos ’ 7
cðx; r Þ ¼ V1 4 g dS þ g dS5; (6)
2 4p Rb Rw
Sb Sw
where Rb (P,Qb), Rw (P,Qw) are the distances from point P(x,r) in the flow zone to
points at the surfaces of the body Qb and ring wing Qw.
Equation 4 in terms of Eq. 5 can be solved for the combination of bodies with the
use of the boundary conditions:
– The surfaces of body, wing and cavity are impenetrable:
un =S ¼ ð1=r Þðdc=dtÞS ¼ 0;
c=S ¼ 0;
ut =S ¼ ð1=r Þðdc=dnÞ=S ;
where n and t – are the unit vectors to the meridional bodies contour. For
combination of the two axisymmetric bodies, the limit values of the normal
derivative of the stream function will be a difference on the outside and inside
the body and ring wing surfaces.
Controlled Supercavitation Formed by a Ring Type Wing 69
Sb
ðð
@ 0
þ gw cos ’
1 Q dS c x b0 ðQ0 Þ ;
@nb RðQ0b ; Qw Þ 0w Lb
b0
Sw
ðð
@c @ 1
Q dS
i ¼ gb ðQ0 Þ b0 b0 ½ gb cos ’
@n0 2 4p @nb RðQ0b ; Qb Þ 0b
Sb
ðð
@ 0
þ gw cos ’
1 Q dS x b0 ðQ0 Þ cjLb ; (7)
@nb RðQ0b ; Qw Þ w0
b0
Sw
where R(Q0b, Qw) is a distance between point Q0b 2 Sb and point Qw 2 Sw, indexes
e and i are exterior and interior, respectively; index o – is a point on the surface in
which the velocity and stream function are determined.
Considering Rw(P,Qb) ¼ Rw(Qw0,Qb), Rw(P,Qw) ¼ Rw(Qw0,Qw) for the surface
ring wing, the equations for them will be identical
ðð
@c w0 gw ðQ0 Þ w0 @ 1
Q dS
e ¼ ½ gw cos ’
@n0 2 4p @nw RðQ0w ; Qw Þ 0w
Sw
ðð
@ 0
þ gb cos ’
1 Q dS x w ðQ0 Þ cjLw ;
@nw RðQ0w ; Qb Þ 0w
w0
Sb
ðð
@c w0 gw ðQ0 Þ w0 @ 1
Q dS
i ¼ ½ g cos ’
@n0 2 4p w
@nw RðQw0 ; Qw Þ w0
Sw
ðð
@ 0
þ gb cos ’
1 Q dS x w ðQ0 Þ cjLw ; (8)
@nw RðQw0 ; Qb Þ w0
w0
Sb
In the system of coordinates the distance between Q0 and Q is: Ro ¼ R(Qo, ;Q) ¼
[(x xo)2 + 2 + o2 – 2ocos’]1/2, where o ¼ (xo); Qo (xo,o) 2 Sa; Q
(x,) 2 S.
ðeÞ 1 @cðQ0 Þ ðiÞ 1 @cðQ0 Þ
If one considers: Vt0 ðQ0 Þ ¼ je ; Vt0 ðQ0 Þ ¼ ji ; and
0 @n0 0 @n0
excepts o, then (the internal flow is absent):
VtðeÞ
0
ðQ0 Þ ¼ gðQ0 Þ:
At the condition that vortex intensity is equal to the sum of the tangential velocity
in points on the surfaces of the body and ring wing, and there is jump on the vortex
70 V.P. Makhrov
layer, the surfaces may be presented by equations for the distribution of velocity g(Q)
on the surface for the body and the ring wing in the following form, [16]:
ðð
1 @ 1
gb ðQ0 Þ ¼ ½ gb cos ’ Q dS
2p @nb RðQb0 ; Qb Þ b0
Sb
ðð
@ 1
þ gw cos ’ Q dS þ V1 x0 ðQb0 Þ; (9a)
@nb RðQb0 ; Qb Þ b0
Sw
ðð
1 @ 1
gw ðQ0 Þ ¼ ½ gb cos ’ Q
2p @nw RðQb0 ; Qb Þ w
Sw
ðð
@ 1
þ gw cos ’ Q dS
@nw RðQw ; Qw0 Þ w0
Sw
ðð
1 x0 w ðQ0 ÞdS
þ ½ gb cos ’
2p w0 RðQb ; Qb0 Þ
Sb
ðð
x0 w ðQ0 ÞdS
gw cos ’ þ 2V1 x0 w ðQ0 Þ; (9b)
w0 RðQw ; Qw0 Þ
Sw
where gb and gw are identical by the velocity of flows at the surfaces of the body and
ring wing, respectively.
Here the integral of Sw in Eq. 9a allows the influence of the ring wing at the flow-
around of the central body to be estimated, and analogously integral of Sb in Eq. 9b
allows the influence of the central body at the ring wing to be determined. These
equations make it possible to calculate the velocity and pressure on the surfaces of
this combination. So, this result is applied for further consideration as a base for
obtaining the more necessary dependencies in the following calculations of
supercavitation flows formed by the ring wing. Hereafter, application of the method
of vortex layer allows the ratio of velocities on combination of the body of
revolution (cavitator) and the ring type wing to be estimated.
It is assumed that the cavitation flow formed by the ring wing may correspond to the
cavitation figures at s > 0 and at s < 0; that cavity boundaries closed at the flow
axis or on the body surface may be taken in the additional condition. When using
Eq. 1 for cavitation velocity and V1 1, interdependencies derivable from (9), we
can compose the set of integral-differential equations for the solution of the
problem (4). It may be written in the following form:
Controlled Supercavitation Formed by a Ring Type Wing 71
ðð
1 @ 1
0
gb ðQb0 Þ ¼ x b ðQb0 Þ ½ gb ðQÞ cos ’ Q dS
2p @nb RðQb ; Qb0 Þ b0
Sb
ðð
@ 1
gs cos ’ Q dS
@ns RðQs ; Qb0 Þ b0
Ss
ðð
@ 1
gw ðQÞ cos ’ Q dS; (10a)
@nw RðQw ; Qb0 Þ b0
Sw
ðð
1 x0 w ðQ0 Þ @ 1
gw ðQw0 Þ ¼ f gb ðQÞ cos ’½ Q ds
2p w0 RðQw0 ; Qb Þ @nkw RðQb0 ; Qb Þ w0
Sb
ðð
x0 w ðQw0 Þ @ 1
þ gs cos ’½ Q dS
w0 RðQkw ; Qs Þ @ns RðQw0 ; Qs Þ b0
Ss
ðð
x0 kw ðQw0 Þ
þ gw ðQÞ cos ’½
w0 RðQw0 ; Qs Þ
Sw
@ 1
dSg 2x0 w ðQw0 Þ; ð10bÞ
@nw RðQw0 ; Qw Þ
ðð ðð
1 1 1
rs ðQs0 Þ ¼ ½ gb ðQÞ cos ’ dS þ gs cos ’ dS
2p RðQs0 ; Qb Þ RðQs0 ; Qs Þ
Sb Ss
ðð
1
þ gw ðQÞ cos ’ dS; (10c)
RðQw ; Qs0 Þ
Sw
Figure 3 illustrates the calculation picture of the cavitation flow behind the body of
revolution in the falling vertical gravity flow. It is formed by a ring type wing also.
The set of equations for the vertical flow should be designed with taking into
consideration the gravity influence. It may be applied for the dynamical boundary
condition for quasi-steady flow and when ps(t) ¼ const. Then the velocity of the
flow along the axis may be determined by Bernoulli’s formula as follows:
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
x
gbs ¼ 1 þ s0 2 6¼ const:; (11)
Fr
xw xc V2
where: x ¼ ; Fr 2 ¼ 1 ; xc, c are the coordinates of the point of the cavity
c gc
beginning, and the cavitation number so is for these coordinates.
Fig. 4 Estimated cavity boundaries with positive cavitation number: 1 – behind the disk,
2 – behind the cone
The set of equations for this case will be analogous to the set of Eqs. 10a, 10b,
and 10c under the suitable conditions of the cavity closing.
3 Numerical Simulation
The existing mathematical model describing the cavitation flow as the set of
Eqs. 10a, 10b, and 10c allows all its characteristics to be determined. For the
continuous flow it was found that the present method gives good predictions for
practical interest. However, numerical difficulties can occur if too much
singularities are used to simulate the body and ring wing.
The solution of the set of nonlinear integral-differential equations of the
Fredholm’s type (10a, 10b, 10c) is obtained by a numerical method. For this
purpose, the surface integrals in this set may be converted by a series of integrals
identically to the finishing forms has been arrived. Here one can use different
approaches, for example, as in [17]. It should be a set of linear algebraic equations,
and it is solved by the squaring formula. It is necessary to use a spline fit function
also in order to determine the accuracy of the present method for the surface
velocity distributions on the body, ring wing and cavity boundaries with exact
analytic solutions [18].
Examples of numerical solutions for different schemes of cavities formed by the
ring wing using positive and negative cavitation numbers are given here. Details of
the method of solving the set of Eqs. 10a, 10b, and 10c are given in [12]. For
example, Figs. 4 and 5 illustrate the results of numerical calculations of approxi-
mation of real cavity shapes formed past a disk and a cone. Figure 4 shows a cavity
with the positive cavitation number (Г is the relative circulation of the ring wing,
Г ¼ 0.3). Figure 5 shows a cavity with the negative cavitation number. It illustrates
dependencies of the cavity shapes for geometric reciprocal relation body-cavitator
and a ring wing.
74 V.P. Makhrov
Fig. 5 Estimated cavity boundaries for the cavitation number s ¼ 0.045: a – behind the cone:
1 s ¼ 0.02, 2 s ¼ 0.035; b – behind the disk, s ¼ 0.045
Fig. 6 Tail cavity with a negative cavitation number formed by the ring wing
4 Experimental Data
The theory has been applied to many experimental tests in the hydrodynamic test
tunnel, towing and vertical test rig [19]. Experimental procedures were performed
for the testing models of bodies of various aspect ratio (l ¼ 5 – 20) and for various
model’s head and the ring wings at all the hydrodynamic test rigs. The cavitation
number is simulated by supported gas (air). Measurements of the drag were realized
for the qualitative analysis or for the comparison with analogs. The expression for
the gas injection rate has the form: GRT/po D2 V1. Here, G, R, T are the weight of
Controlled Supercavitation Formed by a Ring Type Wing 75
Fig. 7 Examples of a cavity formed by the ring wing: a: s ¼ 0.06, b: s ¼ 0.12, c: s ¼ 0.06
gas injected per 1 s, the gas constant, temperature of gas, respectively, and po is the
static pressure in the cavitator zone, D is the disk diameter.
Figures 6 and 7 represent several examples of such cavities observed during the
laboratory experiments [20]. Experimental data from the water tunnel tests are
compared for the well-known data and used in evaluating the present method.
Figure 8 shows a comparison of the cavity shapes past the disk without the ring
wing (A) with Cx ¼ 0.89, and formed by the ring wing around the disk (B) with
Cx ¼ 0.2. The cavitation number and gas rate Cg were constant here.
76 V.P. Makhrov
Conclusions
This paper is the first publication about the new method of forming the con-
trolled supercavitation flow applied earlier by Vladimir Shushpanov.
The most important characteristic of the supercavitation flow formed by the
ring wing is the possibility to make the closing flow with the minimal cavitation
drag and minimal gas loss from the cavity.
We hope the new method of the supercavitation flow formation stimulates
further theoretical and experimental investigations.
References
1. Reichardt H. The laws of cavitation bubbles at axially symmetrical bodies in a flow. Rep. and
Translations, N 766. Moscow, Ministry of Aircraft Production; 1946. p. 82.
2. Serebryakov VV. Some problems of hydrodynamics for high speed motion in water with
supercavitation. International Conference “Super FAST 2008”. St/Petersburg; 2008.
3. Paryshev EV. The dynamics theory of supercavitation. Proceeding of Scientific School «High
Speed Hydrodynamics». Cheboksary; 2002. p. 55–70.
4. Sokoliansky VP. Research of high-speed hydrodynamics in hydrodynamics department of
CAHI. Proceedings of Conference: The Problems of Body Motion in Liquid with High
Velocity. Moscow: CAHI-REGION; 2002. p. 19–30. (In Russian).
5. Shahidzhanov ES. Velocity and high-velocity underwater vehicles. Proceedings of Confer-
ence: The Problems of Body Motion in Liquid with High Velocity. Moscow: CAHI-REGION;
2002. p. 3–18 (in Russian).
6. Lighthill MJ. A note on cusped cavities. Aeronautical Research Council. Rep. & Mem. No.
2328.
Controlled Supercavitation Formed by a Ring Type Wing 77
7. Migachev VI. Axially flow-around wedge in the presence of vortexes. Proceedings of LIEWT,
N 113. Leningrad; 1967. p. 34–46. (In Russian).
8. Prokofjev VV. The task about ventilated gas bubble past a plate. Report IM MSU N1550.
Moscow; 1974. (In Russian).
9. Marakulin EM. Influence of vortex singularities at the geometric cavity with a recurrent
stream. In Set of Papers: Hydromechanics and Energy of Underwater Vehicles. Moscow:
MAI; 1991. p. 28–32. (In Russian).
10. Marakulin EM, Makhrov VP, Uzbashev AV. About thin axisymmetric cavity with a isolate
ring singularities. In Set of Paper: Hydromechanics and Power engineering of Underwater
Vehicles. Moscow: MAI; 1991. p. 23–7 (in Russian).
11. Shushpanov VF, Makhrov VP, Marakulin EM, Kerin NV. The results of experiences for large-
scale models with stern-placed cavity formed by a ring type wing in tower tank of A.N. Krylov
CNII. Lecture of Conference VMF MO USSR and SUDPROM. Sevastopol; 1984.
12. Makhrov VP. Theoretically investigations of cavity flows with negative numbers. Aerosp MAI J.
2001;8(2):30–9 (In Russian).
13. Makhrov VP, Kerin NV, Pushkarev AA. The using of supercavitation with negative cavitation
numbers for high-speed motion under water. Proceedings of Conference: The Problems of
Body Motion in Liquid with High Velocity. Moscow: CAHI-REGION; 2002. p. 140–50.
(In Russian).
14. Makhrov VP. Hydrodynamic of control flows has shaped free boundaries using of external
hydrodynamic singularity – the ring wing. Aerosp MAI J. 2009;16(5):264–73. Moscow: MAI-
Print (In Russian).
15. Lumb G. Hydrodynamics. Moscow: Gostechizdat; 1947.
16. Sobolev SL. Equations of Mathematical Physics. Moscow: NAUKA; 1966 (In Russian).
17. Guzevsky LG. Numerical analysis of cavitation flows. Preprint N 40-79 of CO AS USSR.
Novosibirsk: Heat-Physics Institute; 1979. (In Russian).
18. Inove M, Kuroumaru M, Jamacuchi S. A solution of Fredholm integral equation by means of
the Splaine fit approximation. J Comput Fluids. 1979;7(G.B.):33–46.
19. Kerin NV, Makhrov VP, Pushkarev AA. An experimental research cavity flow with negative
cavity number. Proceeding of Conference: The Problems of Body Motion in Liquid with High
Velocity. Moscow: CAHI-REGION; 2002. p. 146–150. (In Russian).
20. Grumondz VT, Korzhov DN, Makhrov VP. Some model dynamics problems of high-speed
underwater motion of a vehicle with stern-placed wing. International Conference “Super
FAST-2008”; July 2–4, 2008. St. Petersburg; 2008.
Drag Effectiveness of Supercavitating
Underwater Hulls
Igor Nesteruk
Abstract
The important problem of the drag reduction of underwater hulls was
investigated analytically and numerically. The axisymmetric flows of the ideal
and the viscous fluid were used. Different effectiveness criteria, such as: the
volumetric drag coefficient, the drag coefficients, based on the maximum body
cross-section area and the squared hull length, and the ranges of the inertial
motion were applied.
The use of known analytic dependences for the slender axisymmetric cavity
shapes after the slender or the non-slender cavitators, it was shown that the value
of the volumetric drag coefficient and the similar coefficients, based on the
squared values of the length and the caliber, can sufficiently be reduced at
small cavitation numbers. The smallest values of these drag coefficients corre-
spond to the largest aspect ratios and the slender cavitators. Comparison of the
drags of the supercavitating and unseparated flow patterns showed the existence
of the critical values of the volume and dimensions. The supercavitating flow
pattern is preferable for the values of these parameters smaller than critical ones.
The need of the buoyancy force compensation sufficiently diminishes the critical
values of the vehicle volume or its dimensions, which achieve maximum at a
certain value of the motion velocity. In the case of the base cavity existence, the
estimations of the supercavitating hull pressure drag and the comparison with the
unseparated flow pattern are presented. The critical values of the body volume
have a maximum at a certain value of the movement velocity and drastically
increase with the aspect ratio increasing.
Maximum range problems are considered for the supercavitating motion of
the axisymmetric body on inertia under an arbitrary angle to horizon. Different
isoperimetric problems were formulated and solved with the fixed values of the
I. Nesteruk (*)
Institute of Hydromechanics of National Academy of Sciences of Ukraine, Kyiv, Ukraine
e-mail: inesteruk@yahoo.com
body mass, kinetic energy, aspect ratio and caliber. Analytic and numeric
solutions for the maximal range and the optimal body shapes are obtained. It
was shown that infinite small exceeding some critical value of the initial depth
can cause a jump of the range and coming to the water surface. The
corresponding values of the critical initial depth are calculated.
1 Introduction
The drag reduction of the high-speed underwater hulls is the important and difficult
problem due to the very high water density. Really, for two vehicles with the same
shape, volume V and speed U moving in water and air respectively the drag can be
expressed as follows:
The volumetric drag coefficient CV can be the same for these two vehicles, if the
Mach and Reynolds numbers are similar. Then, due to the huge difference in
densities:
rwater
800;
rair
the drag in water can be estimated to be 800 times greater. Therefore, the drag of an
underwater vehicle can be reduced by decreasing the area wetted by water, i.e., by
changing the unseparated flow pattern (a) by supercavitating one (b) (see Fig. 1).
In the case of supercavitation the main part of the hull is located inside the cavity
(see Fig. 1b), therefore the skin-friction drag can be reduced sufficiently. This idea
was developed in many theoretical, numerical and experimental investigations in a
lot of countries. The Ukrainian scientific school, leaded by G. V. Logvinovych and
Yu. N. Savchenko, contributed sufficiently both in the experimental research of
physical principles of supercavitation, theoretical and numerical simulation of this
phenomenon, and in the practical applications (see, for example, [1–24]). Important
contributions to the experimental investigations of supercavitation have been done
with the use of High-Speed Multi-Purpose Water Tunnel available in the Institute of
Hydromechanics of National Academy of Sciences of Ukraine. The experiments
with supersonic underwater projectiles are a significant achievement of the
scientific school, conducted by Yu. N. Savchenko [8]. In Kyiv Institute of Hydro-
mechanics the computer code was developed to calculate the non-steady supercavity
flows, vehicle dynamics and stability [9, 10], which was successfully applied in
many countries.
The supercavitating flow pattern shown in Fig. 1b yields a large pressure drag,
because of the high pressure acting on the cavitator (a part of the hull, wetted by
water). In the case of unseparated flow (see Fig. 1a) the pressure drag is near to zero
Drag Effectiveness of Supercavitating Underwater Hulls 81
due to the d’Alembert paradox. It was necessary to compare the pressure and skin-
friction parts of the total drag and to conclude when the supercavitation is prefera-
ble. Such attempts are presented in [25–27]. In this paper the results of these
investigations are surveyed and some important conclusions are drowning out.
The flow patterns, shown in Fig. 1a,f correspond to the flow without boundary
layer separation and low pressure drag. The supercavitating flow patterns, shown in
Fig. 1b–e, ensure low skin-friction drag due to the small surface of the cavitator
wetted by water, but the pressure drag can be rather high. To create a cavity, the
slender (Fig. 1c,d) and the non-slender (Fig. 1b,e) cavitators can be used. The non-
standard flow pattern with a cavity which closes without any artificial closing body
82 I. Nesteruk
Fig. 2 Non-standard
cavitator and cavity which
needs no closing body
or re-entrant jet (shown in Fig.2, see also [28]) could provide minimal pressure drag
(due to the d’Alembert paradox). The skin-friction drag is reduced in comparison
with the unseparated flow pattern shown in Fig. 1a (due to the smaller area wetted
by water).
To compare the effectiveness of the different flow patterns, different criteria can
be used. If the vehicle velocity U1 and the hull volume Vb are fixed the simplest
and effective criterion is the volumetric drag coefficient:
2X
CV (1)
2 ðV Þ2=3
rU1 b
When the hull caliber Db or its length Lb are fixed, the coefficients CD or CL can
be used:
8X 2X
CD ; CL (2)
rU1
2 pD2
b rU12 L2
b
0:062
CVU 10=21 1= 7
lb ReV ;
U1 Vb 1=3 Lb
ReV ¼ ; lb ¼ (3)
n Db
The CV estimations for the supercavitating hull which use the total cavity
volume (Fig. 1b) can be found in [26] both for slender and non-slender cavitators.
In particular, in [27] the following formula was obtained
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
3 9ps4
CV ¼ (4)
16 ln s
Drag Effectiveness of Supercavitating Underwater Hulls 83
for conic cavitators with the angle 2y; y>250 . Equation 4 follows from the well
known semi-empirical formulas of Garabedian [29]
xð1 xÞ Rn s
R2 ¼ ; ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ;
l 2 L 2 Cx ln s
rffiffiffiffiffiffiffiffiffiffiffiffi rffiffiffiffiffi
L ln s D Cx
l¼ ¼ ; ¼2 (5)
D s Rn s
Here s is the cavitation number; RðxÞ is the cavity radius; Rn is the cavitator
radius; l is the cavity aspect ratio; D is the maximal cavity diameter; L is the cavity
length; Cx is the cavitation drag coefficient related to the base section area of the
cavitator pR2n . It must be noted that the value CV does not depend on y for these non-
slender cavitators and tends to zero with diminishing of the cavitation number s.
The relationship (5) is represented in Fig. 3 by the dashed line. The results of non-
linear numerical calculations for slender cones with the use of the method from
paper [30] are presented by dots. The linear calculations with the use of the
following formulas (see [15, 17])
R2 sx2 x
¼ þ 2b þ 1; (6)
Rn 2Rn ln b
2 2 Rn
(b is the derivative of the radius at the point of cavity origin) are shown in Fig. 3 by
solid lines.
Unfortunately, for the hull, which uses the total cavity volume (see Fig. 1b), the
cavitation number cannot be diminished to zero, since the appropriate cavity aspect
ratio l tends to infinity (see, for example, (5)) when s ! 0. The same value of lb
84 I. Nesteruk
has also the hull located in the cavity. The constructive considerations restrict the
body aspect ratio. For example, if lb is limited by the value lm ¼ 20, the possible
cavitation numbers cannot be less than 0.01 for both the slender and the non-slender
cavitators, and CV 1:5 103 (see Fig. 3).
Formula (3) shows that CVU <1:5 103 for ReV >107 and lb ¼ 20. Thus, the
standard supercavitating flow pattern (Fig. 1b) is preferable for smaller values of
the volumetric Reynolds number ReV <107 only. The cavitation number has to be
close to the minimal possible value s 0:01. The obtained critical value of the
Reynolds number ReV ¼ 107 means that supercavitation is preferable for very
small vehicles. For example, the critical volume of a hull must be limited by values
V 103 106 m3 for the velocity range 100<U1 <1000 m/s. To have a greater
effective supercavitating vehicle, we need to diminish the cavitation number. There
are two possible ways to realize such decreasing: (1) to use the flow pattern 1b and
to increase the body aspect ratio lb ; (2) to use the initial part of the cavity only (the
flow patterns shown in Fig. 1c,d) with the limited values of the hull aspect ratio.
In the first case the critical value of the volumetric Reynolds number can be
increased for the hulls with the greater aspect ratio and corresponding less values of
the cavitation number. For example, if lb ¼ 100, the corresponding values of the
cavitation number and CV (according to the formulae (4), (5)) can be estimated as
follows: s 0:00072, CV 4 105 . It means that the drag is 37 times smaller in
comparison with the case lb ¼ 20. The supercavitating flow pattern is preferable
for such slender hulls (in comparison with the unseparated one shown in Fig. 1a) at
all the values of the subsonic velocities and the vehicle dimensions of practical
interest.
The strength of the hull limits the aspect ratio increasing. Therefore, the
supercavitation drag diminishing is possible with the use the initial part of the
cavity only (as shown in Fig. 1c,d). It looks confusing. Really, if the hull uses only a
part of the cavity, then the volumetric drag coefficient increases (see formula (1)).
But this fact enables us to use smaller cavitation numbers and larger cavities. As a
result, the volumetric drag coefficient can be smaller for the flow patterns 1c and 1d
in comparison with the 1b one. The detailed proof can be found in [31]. The main
results for different values of the maximum hull aspect ratio lm are presented in
Sect. 2. The problem of the buoyancy compensation for the horizontal movement of
supercavitating vehicle is discussed in Sect. 3. The case of the flow patterns with the
base cavity (see Fig. 1e,f) is presented (see also [31]). The principal results are
reported in Sect. 4.
The maximum range problems are considered in [32] for the supercavitating
motion of the axisymmetric body on inertia under an arbitrary angle to horizon in
the case of very high velocities and non-slender cavitators. The vehicle can use the
initial part of the cavity only. The main results are presented in Sect. 5 for different
values of the maximum hull aspect ratio lm .
Drag Effectiveness of Supercavitating Underwater Hulls 85
If the hull is located in the initial part of the cavity only (such as shown in Fig. 1c,d),
the appropriate volumetric drag coefficients can be easily defined with the use of
(1) and (5) for non-slender cavitator (or (6) in the case of the slender one). The
analytical formulas can be found in [31], the calculation examples are presented in
Fig. 4 for different values of the maximum hull aspect ratio lm . The lines corre-
spond to the non-slender cavitators (the results do not depend on y); the dots show
the case of the slender cavitator with b ¼ 0:1.
It can be seen from Figs. 3 and 4 that CV can be sufficiently reduced for s<0:01.
The smallest values of CV correspond to the largest values of the hull aspect ratio. In
the case of the non-slender cavitators the function CV ðsÞ has a minimum. May be
this fact is connected with the limited accuracy of the Garabedian formulae (5) for
very small cavitation number in the region close to the cavitator. Usually, the
slender cavitators yield smaller values of CV . May be it is due to the limited
accuracy of the Eq. 6. In any case this interesting fact needs additional investi-
gations with the use of the second approximation equation [18] or a nonlinear
approach.
An example of the optimal shape with lm ¼ 10, b ¼ 0:1, the velocity 700 m/s
(the corresponding value of the cavitation number at small depth without ventila-
tion is 0.0004), Lb = Rn ¼ 62:62, CV ¼ 0:00099 is shown schematically in Fig. 1c.
The optimal hull shape must be as close as possible to the initial part of the cavity. If
the hull caliber or its length is fixed, the optimal hull must be only inscribed into the
initial part of the cavity, but its caliber must coincide with the diameter of the cavity
at the body end (see Fig. 1d). The dependences for coefficients CD and CL , which
can be obtained with the use of formulae (2), (5) and (6), are shown in Figs. 5 and 6
Fig. 4 Volumetric drag coefficients for different values of the hull aspect ratio
86 I. Nesteruk
Fig. 5 The drag coefficients CD for different values of the hull aspect ratio
Fig. 6 The drag coefficients CL for different values of the hull aspect ratio
(see details in [31]). The lines correspond to the non-slender cavitators (the results
do not depend on y); the dots show the case of the slender cavitator with b ¼ 0:1.
The following equation
CV ¼ CVU (7)
can be used to calculate the critical value of the volumetric Reynolds number
ReV which corresponds to the equal efficiency of the unseparated (Fig. 1a) and
Drag Effectiveness of Supercavitating Underwater Hulls 87
3:5 109
ReV (7a)
lm 10=3 C7V
sU12
pa
h¼ ; (7b)
2g rg
Fig. 7 (a) The critical volume of the effective supercavitating hull for different values of the
velocity. (b) The critical velocity of the effective supercavitating hull for different values of the
volume
The supercavitating hull moves in the gas (see Fig. 1b,c,d) with a very small value
of the buoyancy force in comparison with the case of the vehicle wetted by water
shown in Fig. 1a. Therefore, for supercavitating vehicles, which move horizontally,
the problem of their weight compensation must be solved (as in the case of
airplanes). For this purpose the hull planing on the cavity surface or underwater
wings are used. This situation causes an additional drag DCV , which can be
estimated with the use of the aerodynamic effectiveness k ¼ Cy =Cx . To calculate
the critical Reynolds number, a new equation
V1/3(m)
*
CV =0.0002 CV = 0.0002
8 CV =0.0004 CV = 0.0004
6
CV = 0.0006
4
2 CV =0.0008
0
10 500 1000 U(m/s)
Fig. 8 Dependencies of the critical volume at different values of the volumetric drag coefficient
90 I. Nesteruk
velocity range increases when the value of CV decreases, in particular U2 can
exceed the velocity of sound in water.
Figure 8 shows that the critical volume decreases drastically with the increase of
CV . It means that supercavitation is effective at very small cavitation numbers and
for not very great depth (if the vapor cavitation is used). Figures 8 and 9 show also
that the necessity of the buoyancy force compensation diminishes the critical
volume (especially at small values of k and large values of CV ). The same
estimations of the critical hull calibre and its length can be done in the cases
when these parameters are fixed (see details in [31]).
For the flows with the base cavity there are two options: (1) the hull is covered by
another cavity (the two-cavity flow pattern shown in Fig. 1e); (2) the hull is wetted
by the water flow without the boundary layer separation as shown in Fig. 1f. The
comparison of efficiency of these two patterns was done in [31] with the use of
formulae (6) for the pattern 1e and the parabolic unseparated shape
rffiffiffiffiffiffiffi
R bx2 1 x
¼ 2 þ 1; 0
Rn Rn b Rn
for the flow pattern 1f. Equation 8 was used to calculate the critical volume. The
numerical examples are presented in Fig. 10. The values n ¼ 1:3 106 m2 /s,
b ¼ 0:1, k ¼ 10 (solid lines) and k ¼ 1 (dots) were used for calculations. It can
be seen from Fig. 10 that some curves have a maximum (similar as ones shown in
Drag Effectiveness of Supercavitating Underwater Hulls 91
Fig. 10 Dependencies of the critical volume at different values of the hull aspect ratio for the base
cavity flow pattern
Fig. 8). The corresponding velocity increases with the hull aspect ratio increasing
and may approach to the sonic velocity in water. The critical volume increases
drastically with the increasing of the hull aspect ratio. Figure 10 shows also that the
buoyancy force compensation diminishes the critical volume (especially at small
values of k and large values of lm )
The results obtained in the previous sections stimulated the investigation of the
effectiveness of the supercavitating flow pattern for the inertial motion with very
small cavitation numbers. The horizontal supercavitating motion on inertia and the
problem of range maximization were considered by Putilin, Gieseke, Serebriakov,
Kirschner, Schnerr [14, 33–35] and other authors. The case of the non-horizontal
inertial motion with different isoperimetric conditions was investigated in [26,
36–39], but it was taken into account only the case of complete using the cavity
volume (Fig. 1b). The partial cavity use (Fig. 1c,d) is typical for very small
cavitation numbers and was investigated in [32] for non-slender cavitators. Here
the results of the papers [32, 36–39] will be shortly reported.
Let a model start a rectilinear movement in water on inertia with the velocity U0
under an arbitrary angle g to horizon. The distance S, passed by the supercavitating
body, should be maximal (see Fig. 11). It was shown (see, e.g., [37, 38]) that in
many cases the flow may be supposed to be quasi-stationary and the gravity effect
on the cavity and body motion may be neglected. If the cavitator is non-slender, the
semi-empirical relations (5) by Garabedian may be used with the current cavitation
number s at the cavitator immersion depth. If we neglect changes of the cavitation
92 I. Nesteruk
number s<<1, then Cx may be considered to be constant and the distance S passed
by the body is defined by the following formula:
2m U0
S¼ ln ; (9)
rCx pR2n U
For example, in the case of the fixed starting velocity the most interesting tasks may
be listed as follows:
1. The body mass and its caliber Db are fixed
2. The body mass and its length Lb are fixed
3. The body mass and its volume Vb or the average body density rb ¼ m=Vb and its
volume are fixed
4. The average body density and its caliber are fixed
5. The average body density and its length are fixed
Taking into account that a body practically stops after washing off by water, it
could be shown that the optimal body shape must coincide with the cavity shape in
the moment of washing off for problems 3–5 (see [36] and also [14, 34]). For the
isoperimetric conditions 1 and 2, the optimal body shape must be inscribed in the
cavity corresponding to the moment of washing off; its caliber must coincide with
the cavity caliber for the task 1 and its length must coincide with the cavity length
for the task 2. For these two cases the examples of the optimal body shapes are
presented in [36]. These five isoperimetric conditions can be applied for the
problems with the fixed final depth of supercavitating motion [36, 37] and for the
tasks with the fixed initial body depth too [36–38].
Another group of problems arises when the initial (or final) kinetic energy of a
body or its initial (or final) momentum are fixed. In these cases the initial velocity
Drag Effectiveness of Supercavitating Underwater Hulls 93
and body mass are not fixed. The solution of the problem must answer the question:
what are the best values of the initial velocity and body mass to achieve the
maximal range. For example, is it better to use a small body with a large velocity
or a large body with a small velocity? For the case of horizontal motion the
optimum body mass was calculated by Gieseke [33]. For every problem with the
non-horizontal body motion, the condition of the fixed initial or final depth can
be used. For this group of problems the optimal body shape coincides with the shape
of the cavity at the smallest possible cavity number. Usually we use the value
s ¼ 0:01, because it is practically impossible to use shapes with the aspect ratio
greater than 20. It is difficult to describe all 18 isoperimetric problems in details.
Let us consider their main features and present some examples.
5.1.1 Problems with the Fixed Initial Velocity and the Fixed Final
Depth
Such isoperimetric conditions were considered in [36]. In this case all the results
(for example, the maximum ranges) do not depend on the angle of motion g. This
conclusion is in good agreement [36] with the calculations, performed with the use
of the computer program SCAV [9, 10].
For example, the solution of the fifth problem yields the optimum value of the
final velocity (see [36–38])
The maximum range and the optimal value of the final cavitation number are
given by the following formulae:
Fr 2
r 3eh2 10 þ hf
S ¼ b 0 ; s ¼ ; h2 ;
3eh2 Fr02 L
pffiffiffiffiffiffi
where r b ¼ m=Vr; Fr0 ¼ U0 = gL is the initial Froude number. Knowing the
optimal value of the final cavitation number and the fixed body length (which has
to be equal to the cavity length of the moment of washing off, because in this case
the optimal body shape coincide with the shape of the cavity at s ¼ s ), the optimal
body shape and the optimal cavitator radius can be calculated with the use of
formulas (5), see [36–38]. For the isoperimetric problems 1–4, the corresponding
values of the optimal dimensionless final velocity U and the maximum range were
obtained in [36].
2gðh1 S sin gÞ
s¼
U 2 U
2
0
94 I. Nesteruk
where the initial depth h0 (the initial position of the center of the cavity separation
cross-section) and h1 ¼ 10 þ h0 are measured in meters. Therefore, the maximum
range and the optimal value of the final cavitation number depend on the motion
angle, but the optimum values of the final velocity are the same or very close to
those, obtained for the case with the fixed final depth (see [38]). In particular, for the
problems 1 and 5 this value coincide with (10), for problems 2–4 it was obtained
U ¼ e0:25 0:78, U ¼ e0:375 0:68 and U ¼ e1=3 0:72 respectively. The
solution of the fifth problem can be found in [37].
The values of the maximum range have to be calculated with the use of nonlinear
equations. For example, for the problems 1 and 5 this equation can be written as
follows:
2
S ðh1 S sin gÞ ¼ (11)
ep
The dimensionless values are connected with the physical characteristics by the
formulas
where the formulas for s depend on the isoperimetric conditions. For example, for
the first problem this relationship can be written as follows:
rffiffiffiffiffiffi
s DU0 r
¼
s
2 gm
The relationship (12) is shown in Fig. 12 by dotted lines. Knowing the optimal
value of the final cavitation number, the optimal body shape and the optimal
cavitator radius can be calculated with the use of formulas (5).
For g>0 the relationships S ðh1 Þ may achieve the infinite slope at some critical
values of the initial depth h1 (see Fig. 12). It means that for smaller values of the
ðcrÞ
initial depth the body can reach the free water surface without a loss of the
supercavitating flow pattern. The critical values h1 have been found in [38] for
ðcrÞ
Drag Effectiveness of Supercavitating Underwater Hulls 95
S*, σ*
r=0
1.5
r = –90º
r = –30º r = –30º
1
r = –90º
r=0
0
–0.5 0 0.5 1
h*1
Fig. 12 Dependences of the dimensionless maximum range and the optimal final cavitation
number at different angles of motion for the problem 1 or 5
Fig. 13 Dependences of the maximum range and optimal cavitation number for different values
of the hull aspect ratio
different isoperimetric conditions. For example, for the problems 1 and 5 the
following formula was proposed:
rffiffiffiffiffiffiffiffiffiffiffiffi
ðcrÞ 8 sin g
h1 ¼ (13)
ep
2=3 1=3
6as1=6 rb I0
U ¼ e3 0:0498; S ¼ ; a¼ 1=6 :
eCV r 2gðhf þ 10Þ
For all four isoperimetric problems the optimal value of the optimal final
cavitation number s must be as small as possible and the optimal hull shape
must coincide with the cavity shape at this cavitation number. For example, when
s ¼ 0.01, then S 654a. The optimal values of the body mass can be obtained
for the cases with the fixed starting and final momentum respectively:
0:00352I0 0:071If
m ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ; m ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi :
gðhf þ 10Þ gðhf þ 10Þ
Then the optimal body volume, length, caliber and cavitator radius can be
calculated with the use of Eq. 5. To determine the maximum range in the problems
with the fixed starting depth, it is necessary to solve the non-linear equations,
similar to (11).
2=3 1=3
3bs1=3 r T
U ¼ e1:5 0:223; S ¼ ; b ¼ b 0 1=3 (14)
eCV r gðhf þ 10Þ
For all the four isoperimetric problems the optimal value of the optimal final
cavitation number s must be as small as possible and the optimal hull shape must
coincide with the cavity shape at this cavitation number. For example, when
s ¼ 0.01, then S 152b. The optimal values of the body mass can be obtained
for the cases with the fixed starting and final kinetic energy respectively:
0:00025 T0 0:005 Tf
m ; m
gðhf þ 10Þ gðhf þ 10Þ
Then the optimal body volume, length, caliber and cavitator radius can be
calculated with the use of Eq. 5. To determine the maximum range in the problems
with the fixed starting depth, it is necessary to solve the non-linear equations,
similar to (11).
Thus, the presented simple analysis makes it possible to obtain the analytical
relations for the optimal supercavitating model parameters and the optimal starting
parameters. The obtained relations are in good agreement with the more accurate
computer calculations.
Drag Effectiveness of Supercavitating Underwater Hulls 97
Table 1 The value of D2 s=ð4Cx R2n Þ for the cones with different half-angles y
s 0.1 0.05 0.025 0.01
y ¼ 15
1.115 1.094 1.072 1.031
y ¼ 10
1.116 1.094 1.078 1.064
y ¼ 5
1.128 1.107 1.089 1.066
y ¼ 3
1.113 1.108 1.097 1.079
98 I. Nesteruk
aspect ratio of the optimal unseparated shape has to be minimal for the
isoperimetric conditions 1 and 5. For other problems the optimal aspect ratio
must be maximal (see [38]).
5.1.7 The Range Increase with the Use of a Cavitator with Changeable
Shape and Diameter
The initial stages of the supercavitating inertial movement correspond to higher
velocities and greater cavities in comparison with the final stages. Therefore, there
is no need to use a large cavitator with a high drag for the initial stages of
movement. The range can be increased with the use of cavitators with the change-
able diameter or shape. The corresponding estimations have been done in [37]. For
the horizontal movement with p theffiffiffiffi final cavitation number s ¼ 0.01, the optimal
value of the parameter c ¼ a o ¼ 1.438 was calculated. The values of the
cavitator radius and its drag coefficient for the first stage are connected with the
final characteristics as follows: Rn ¼ Rn1 a; a 1, Cx ¼ Cx1 o; o 1. The
increase of the range is approximately 64%.
2QDm
DS ¼ (15)
prUCx R2n
8QDm
DS ¼ : (16)
prUD2 s
Drag Effectiveness of Supercavitating Underwater Hulls 99
Therefore, the maximum range can be achieved with the use of minimal possible
value of the cavitation number (for example, s ¼ 0.01). According to (16), the
optimal range doesn’t depend on the shape of the cavitator, since there is no
dependence on its drag coefficient (similar to the case of inertial motion). The
same conclusions can be done for all six isoperimetric problems (see [38]).
For the first group of isoperimetric conditions and the fixed values of the body
caliber or its length, the optimal hull shape must be inscribed in the cavity
corresponding to s ¼ 0.01; its caliber must coincide with the cavity caliber or its
length must coincide with the cavity length accordingly. For other four
isoperimetric conditions the optimal hull must coincide with the cavity shape at
s ¼ 0.01.
It must be noted, that in the case of the unseparated hulls, their optimal aspect
ratio can be minimal. Such situation occurs for the following two problems: (1) the
mass of the fuel and the body caliber are fixed; (2) the density of the fuel rf , its part
in the total hull volume Kf and the body length are fixed. It is impossible to achieve
an unseparated flow pattern with the use of very thick hulls (with a near to zero
aspect ratio), but a special shape can be used to ensure the flow without separation
and cavitation (see [25, 26]). Other four problems need the maximal value of the
unseparated hulls aspect ratio.
If the body mass, its caliber and aspect ratio are fixed, there is no need to investigate
the case of the fixed body length. The volume of the hull located in the initial part of
the cavity after a non-slender cavitator can be estimated as a cone volume (see
Fig. 1c,d); therefore there is no need to use the isoperimetric condition with the
fixed volume. Thus, it is necessary to investigate the first problem only from the list,
presented in Sect. 5.1.
Both the natural and the ventilation cavitation will be taken into account with the
given value of the cavity pressure pc at the final moment of the hull washing off
(when the vehicle stops). The cavitation number can be rewritten as follows:
2gh2 U
s¼ ; U ¼ : (17)
U2 U
2 U0
0
velocity. The same trivial solution will be in the case of complete cavity volume
using (Fig. 1b) and for all five problems listed in Sect. 5.1.
If instead of mass, the initial body kinetic energy T0 is fixed, then the optimal
values of final velocity and body mass can be calculated (see details in [32])
2T0
U ¼ e0:5 0:607; m ¼ :
eU 2
The case of the fixed initial velocity is more difficult. But if the cavity volume is
used completely (Fig. 1b), the same relationship (see Eq. 10) was obtained in [36].
For other four listed isoperimetric conditions, other relationships for the optimal
velocities ratio were obtained in [36] with the use of the 1b pattern. The case of the
partial using of the cavity volume (Fig. 1c,d) needs solving the non-linear equations
and depends on the dimensionless parameter H ¼ gh2 =U02 .
The results for the dimensionless maximum range S
S gh2 rD2b
S ¼
mU02
(solid lines) and optimal final cavitation number s (dashed lines) are presented in
Fig. 13 for different values of the hull aspect ratio. The range increases with the
increasing of the aspect ratio, but the differences are sufficient for the very small
values of the parameter H only. For H>0:001
and lm >15 the obtained solution is
practically independent of H and coincides with the results for the flow pattern 1b
reported in [36].
The conclusion that the optimal hull caliber must coincide with the maximum
final cavity diameter (see [36]) is no more valid for the case of very high velocities
To illustrate this fact, an example of optimal shape is shown in
(small values of H).
Fig. 14. The parameters of this supercavitating hull are lm ¼ 3, H ¼ 0:001. The
optimal range S 0:083 exceeds the ranges of any other hull with the same values
of H and lm . For example, if the hull caliber coincides with the maximum final
cavity diameter, then S 0:063only.
solutions of the problem will depend on the angle g. The nonlinear equations make
the search for the optimal solution more complicated.
If in addition to the body mass, caliber, aspect ratio, initial depth, its final
velocity is also fixed, the maximum range corresponds to the maximal starting
velocity. The same trivial solution will exist in the case of complete cavity volume
using (Fig. 1b) and for all five problems listed in Sect. 5.1.
If instead of final velocity, the initial one is fixed, the solution is not trivial and
depends on the dimensionless parameters N, h1
rffiffiffiffiffiffi
mg h1 D b rg
N ¼ ; h1 ¼ :
rU02 D2b U0 m
rffiffiffiffiffiffi
S Db rg
S ¼
U0 m
(solid lines) and optimal final cavitation number s (dashed lines) are presented in
Figs. 15–17 for different values of the hull aspect ratio. The range increases with the
increasing of the aspect ratio, but the differences are sufficient for the very small
values of the parameter N only. For N>3 107 and lm >15 the obtained solution is
independent of N and coincide with the results for the flow pattern 1b obtained in
[38].
The numerical analysis showed that for g>0 the solution exist only for the values
ðcrÞ
of h1 which are greater than the critical one h1 . The situation is similar to the flow
pattern 1b investigated in [38]. It means that for smaller values of the initial depth
the body can reach the free water surface without a loss of the supercavitating flow
pattern. The critical values h1 can be seen in Fig. 17. Increasing the hull aspect
ðcrÞ
Fig. 15 Dependencies of the maximum range and optimal final cavitation number for different
values of the hull aspect ratio at N ¼ 107 , g ¼ 90
102 I. Nesteruk
Fig. 16 Dependencies of the maximum range and optimal final cavitation number for different
values of the hull aspect ratio at h1 ¼ 0, g ¼ 90
Fig. 17 Dependencies of the maximum range and optimal final cavitation number for different
values of the hull aspect ratio at N ¼ 107 , g ¼ 90
ratio increases h1 , which tends to the value (13) obtained in [38] for the flow
ðcrÞ
pattern 1b.
If a part of a vehicle is covered by the cavity, which closes itself (without any
fictitious closing rigid body or re-entrant jet, see Fig. 2), the hull’s pressure drag has
to be near to zero (due to d’Alembert paradox). Only the friction in the boundary-
layer determines the body drag. The skin-friction drag on such vehicle can be
reduced, since the large part of its surface has no contact with the water (an
Drag Effectiveness of Supercavitating Underwater Hulls 103
advantage of the supercavitating flow pattern). On the other hand, such vehicle has
no high pressure drag (a typical disadvantage of the supercavitation flow pattern).
Due to the smaller area of contact with the water, the skin-friction drag coefficients
cab be estimated as follows for the laminar and the turbulent boundary-layer
respectively, [41]:
rffiffiffiffiffi
4:708 Vb
CV ¼ pffiffiffiffiffiffiffiffi ; (18)
ReV V
0:073 Vb 6=7
CV ¼ ; (19)
ReV 1=7 V 13=21
Fig. 18 Axisymmetric
cavitator (x < 0) and cavity,
which closes itself (x > 0).
Shape and pressure
distribution
104 I. Nesteruk
measured at ReV ¼ 8:5 106 (see [42]). These small values of CV let the untypical
supercavitating hull shown in Fig. 18 be effective even in the cases of horizontal
motion (when the buoyancy force must be compensated). From Fig. 8 it can be
seen that the critical volume of the effective supercavitating hull can be rather
large for CV 5 104 and high enough values of the aerodynamic effectiveness
k ¼ Cy =Cx .
The diminishing Vb =V leads to the drag reduction. Nevertheless, the short
cavitators have more deep pressure minimum on their surface. This fact can
cause separation (and cavitation) upstream to the point x ¼ 0 and another flow
pattern with a large pressure drag. The separation behavior is very important for
such flow pattern and has to be investigated in a water tunnel. Some results of wind
tunnel tests with unseparated shapes are presented in [25, 26].
Conclusions
The value of the volumetric drag coefficient and the similar coefficients, based
on the squared values of the length and the caliber, can sufficiently be reduced at
cavitation number less than 0.01. The smallest values of these drag coefficients
correspond to the largest aspect ratios and the slender cavitators. Comparison of
the supercavitating and unseparated flow patterns showed the existence of the
critical values of the volume and dimensions. The supercavitating flow pattern is
preferable for the values of these parameters smaller than critical ones. For the
horizontal supercavitation motion, the necessity of the buoyancy force compen-
sation sufficiently diminishes the critical values of the vehicle volume or its
dimensions, which achieve maximum at a certain value of the motion velocity.
In the case of the base cavity, the comparison the supercavitating and the
unseparated flow patterns is presented. The critical value of the body volume
has a maximum at a certain value of the movement velocity and drastically
increases with the aspect ratio increasing.
Maximum range problems are considered for the supercavitating motion of
the axisymmetric body on inertia under an arbitrary angle to horizon. Different
isoperimetric problems were formulated and solved with the fixed values of the
body mass, kinetic energy, aspect ratio and caliber. Two dimensionless
parameters are proposed which influence the solution. At small values of these
parameters the optimal body shapes may use the nose part of the cavity only.
Analytic and numeric solutions for the maximal range and the optimal body
shapes are obtained. It was shown that infinite small exceeding of some critical
value of the initial depth can cause a jump of the range (body comes to the water
surface without loosing the supercavitating flow pattern).
The presented analysis showed that supercavitating hulls can be successfully
used to diminish the drag especially at very small cavitation numbers.
Supercavitation may be preferable for not very large hulls which use whole
cavity space or its nose part only. Non-standard cavitators and cavities shown in
Fig. 2 can be successfully used to reduce the total drag.
Drag Effectiveness of Supercavitating Underwater Hulls 105
References
1. Logvinovich GV. Hydrodynamics of Flows with Free Boundaries. Kiev: Naukova Dumka;
1969. (In Russian). English translation: Halsted; 1973.
2. Buyvol VN. Slender cavities in flows with perturbations. Кiev: Naukova Dumka; 1980
(In Russian).
3. Logvinovich GV, Buyvol VN, Dudko AS, et al. Free boundary flows. Кiev: Naukova Dumka;
1985 (In Russian).
4. Savchenko YuN. Supercavitating object propulsion. RTO-AVT/VKI Special Course on
Supercavitating Flows; February 12–16, 2001. VKI, Brussels; 2001. (Belgium).
5. Savchenko YuN. On motion in water in supercavitation flow regimes. Gidromehanika.
1996;70:105–15 (In Russian).
6. Savchenko YuN, Vlasenko YuD, Semenenko VN. Experimental study of high-speed cavitated
flows. Int J Fluid Mech Res. 1999;26(3):365–74.
7. Savchenko YuN, Semenenko VN, Putilin SI. Unsteady supercavitated motion of bodies. Int J
Fluid Mech Res. 2000;27(1):109–37.
8. Savchenko YuN. Perspectives of the supercavitation flow applications. Proceedings of the
International Conference on Superfast Marine Vehicles Moving Above, Under and in Water
Surface (SuperFAST’2008), 2–4 July 2008. St. Petersburg; 2008. ISBN 5-88303-393-8.
9. Savchenko YuN, Semenenko VN, Putilin SI, et al. Designing the high-speed supercavitating
vehicles. International Conference on Fast Sea Transportation (FAST’2005), June 2005.
St. Peterburg; 2005.
10. Semenenko VN. Some problems of supercavitating vehicle designing. Proceedings of the
International Conference on Superfast Marine Vehicles Moving Above, Under and in Water
Surface (SuperFAST’2008); 2–4 July 2008. St. Petersburg; 2008. ISBN 5-88303-393-8.
11. Logvinovich GV, Serebryakov VV. On the methods of calculating a shape of the slender
axisymmetric cavities. Gidromehanika. 1975;32:47–54 (In Russian).
12. Vlasenko YuD. Experimental investigations of supercavitating regime of flow around self-
propelled models. Int J Fluid Mech Res. 2001;28(5):717–33.
13. Savchenko YuN, Semenenko VN. Special features of supercavitating flow around polygonal
contours. Int J Fluid Mech Res. 2001;28(5):660–72.
14. Putilin SI. Some features of a supercavitating model dynamics. Int J Fluid Mech Res. 2001;
28(5):631–43.
15. Nesteruk I. Investigation of slender axisymmetric cavity form in fluid with gravity. Izv AN
SSSR MFG. 1979;6:133–6 (In Russian).
16. Nesteruk I. Form of slender axisymmetric cavity. Izv AN SSSR MFG. 1980;4:38–47
(In Russian).
17. Nesteruk I. Some problems of axisymmetric cavity flows. Izv AN SSSR MFG. 1982;1:28–34
(In Russian).
18. Nesteruk I. The slender axisymmetric cavity form calculations based on the integral-differential
equation. Izv AN SSSR MFG. 1985;5:83–90 (In Russian).
19. Goman OG, Semenov YA. Oblique entry of a wedge into an ideal incompressible fluid. Fluid
Dynam. 2007;42(4):581–90.
20. Faltinsen OM, Semenov YA. The effect of gravity and cavitation on a hydrofoil near the free
surface. J Fluid Mech. 2008;597:371–94.
21. Savchenko YuN, Semenov YA. Hydrodynamic drag of a surface with the mixed boundary
conditions. Prykladna Gidromehanika. 2005;7(2):54–62 (In Russian).
22. Savchenko YuN, Savchenko GYu. Efficiency estimation of the supercavitation using on the
axisymmetric hulls. Prykladna Gidromehanika. 2004;6(4):78–83 (In Russian).
23. Savchenko YuN, Savchenko GYu. Cylinder planning on the supercavity surface. Prykladna
Gidromehanika. 2007;9(1):81–5 (In Russian).
24. Savchenko YuN, Savchenko GYu. Near-wall cavitation on a vertical wall. Prykladna
Gidromehanika. 2006;8(4):53–9 (In Russian).
106 I. Nesteruk
25. Nesteruk I. The problems of drag reduction in high speed hydrodynamics. The International
Summer Scientific School “High Speed Hydrodynamics”; June 16–23, 2002. Cheboksary;
2002. p. 351–9.
26. Nesteruk I. Drag reduction in high-speed hydrodynamics: supercavitation or unseparated
shapes. CAV2006; 2006. Netherlands.
27. Nesteruk I. Drag calculation of slender cones using of the second approximation for created by
them cavities. Prykladna Gidromekhanika (Appl Hydromech), Kyiv. 2003;5(1):42–6
(In Ukrainian).
28. Nesteruk I. Partial cavitation on long bodies. Prykladna Gidromekhanika (Appl Hydromech),
Kyiv. 2004;6(3):64–75 (In Ukrainian).
29. Garabedian PR. Calculation of axially symmetric cavities and jets. Pac J Math. 1956;
6(4):611–84.
30. Nesteruk I. Simulation of axisymmetric and plane free surfaces by means of sources and
doublets. Prykladna Gidromekhanika (Appl Hydromech), Kyiv. 2003;5(2):37–44
(In Ukrainian).
31. Nesteruk I. Drag diminishing of long axisymmetric high-speed bodies. Prykladna
Gidromekhanika (Appl Hydromech), Kyiv. 2009;11(2):55–67 (In Ukrainian).
32. Manova ZI, Nesteruk I, Shepetyuk BD. Optimization problems for high-speed supercavitation
motion on inertia with the non-slender cavitators. Prykladna Gidromekhanika (Appl
Hydromech), Kyiv. 2009;11(4):54–9 (In Ukrainian).
33. Gieseke TJ. Toward an optimal weapon system utilizing supercavitating projectiles. Interna-
tional Conference on Cavitation “Cav2001”, Pasadena; 2001, Session B3.002.
34. Serebryakov VV. The models of the supercavitation prediction for high speed motion in water.
International Summer Scientific School “High Speed Hydrodynamics”. Cheboksary;
2002:71–92.
35. Serebryakov VV, Kirshner IN, Scherr GH. Some problems of high speed motion in water with
supercavitation for sub-, trans- and supersonic mach numbers. Proceedings of the X Interna-
tional scientific school “High-speed hydrodynamics” and International conference «Hydrome-
chanics. Mechanics. Power-plants» (to the 145-th anniversary of academician A.N.Krylov).
Moscow/Cheboksary: Cheboksary department of Moscow State Open University; 2008.
p. 73–104. ISBN 978-5-902891-35-2.
36. Nesteruk I, Semenenko VN. Problems of optimization of range of the supercavitation inertial
motion at the fixed final depth. Prykladna Gidromekhanika (Appl Hydromech), Kyiv. 2006;
8(4):33–42 (In Ukrainian).
37. Nesteruk I, Savchenko YuN, Semenenko VN. Range optimization for supercavitating motion
on inertia. Rep Ukrainian Acad Sci. 2006;8:57–66 (In Ukrainian).
38. Nesteruk I. Range maximization for supercavitation inertial motion with the fixed initial depth.
Prykladna Gidromekhanika (Appl Hydromech), Kyiv. 2008;10(3):51–64 (In Ukrainian).
39. Nesteruk I. Hull optimization for high-speed vehicles: supercavitating and unseparated shapes.
International Conference SuperFAST2008, July 2–4, 2008. St. Petersburg; 2008.
40. Savchenko YuN. Investigations of supercavitation flows. Prykladna Gidromekhanika (Appl
Hydromech), Kyiv. 2007;9(2–3):150–8 (In Russian).
41. Buraga OA, Nesteruk I, Savchenko YuM. Comparison of the slender axisymmetric bodies
drag by unseparated and supercavitation flow patterns. Prykladna Gidromekhanika (Appl
Hydromech), Kyiv. 2002;4(2):3–8 (In Ukrainian).
42. Lorant M. Investigation into High-Speed of Underwater Craft. Naut Mag. 1968;200(5):273–6.
Hydrodynamic Characteristics of a Disc
with Central Duct in a Supercavitation Flow
G. Yu. Savchenko
Abstract
Supercavitation flow past a disc cavitator with a round hole at its center is
studied experimentally. The cavity drag of a family of discs with different values
of the inner-to-outer diameter ratio Dd is measured in a water tunnel. The results
are presented as the drag coefficient and the factor of added mass versus the
diameter ratio and compared with the solution of the plane problem for two
symmetrical plates and with experimental results by other authors.
1 Introduction
2 Experimental Procedure
View
D = 47 mm
h
Δ= d
d
Fig. 1 The disc cavitator
3 mm
with different diameters of
the center hole
During the tests, the fairing 3 with the model 7 was sunk into the water flow
using a hinged suspension 6. While traversing the free boundary, the model 7
entrained some air, and for some time it was in the flow in the regime of developed
cavitation (see Fig. 3b). 10–20 s later, when the air from the cavity had been
entrained by the flow, the regime of continuous flow set in (see Fig. 3a). The
hydrodynamic resistance of the model in the regimes of supercavity and continuous
flow was measured using a type DFC 60–11 electronic digital dynamometer and a
storage oscilloscope.
During the experiments, the flow velocity in the water tunnel was kept constant
V1 ¼ 8:9 m=s, which was provided by keeping a constant water level of 4.04 m in
the head tank.
3 Experimental Results
The test results for the series of models are plotted in Fig. 4 as the drag coefficient
Cx versus a dimensionless parameter – the ratio of the inner diameter d of the
cavitator to its outer diameter D. The drag coefficient is calculated as
2X
Cx ¼ ; (1)
rV1
2 S
where r is the water density (1,000 kg/m3 at 15 C); V1 ¼ 8:9 m=s is the inflow
velocity; S ¼ p4 ðD2 d 2 Þ is the cavitator surface area according to the dimensions
in Fig. 1; and X [N] is the hydrodynamic resistance of the cavitator model.
The plots show that the model drag coefficient in the regime of continuous flow
tends to Cx 2:0 as the inner diameter increases to d =D ! 1, which corresponds to
the drag coefficient of a flat plate [2].
Thus it may be thought that at d=D close to one the model drag coefficient is
close to two
d
Cx ¼ 2:0; >0:85: (2)
D
At d=D ! 0 in the regime of continuous flow the drag coefficient is close to that
of a disc
110 G.Yu. Savchenko
d
Cx ¼ 1:2; <0:45: (3)
D
calculation of the drag coefficient of a disc with a center hole based on the solution
of the plane problem of flow past two symmetrical plates [3].
Since the theoretical dependence is too complex and cumbersome, we propose
the interpolation formula based on the experimental results (Fig. 4), which is more
convenient for practical use
d d
Cx s; ¼ 0:8 þ 0:08 ð1 þ sÞ: (4)
D D
where Cxd ¼ 0:82ð1 þ sÞ is the drag coefficient of a disc [3]; Cxp ¼ 4þp 2p
ð1 þ sÞ is
0 PcÞ
the drag coefficient of a flat plate [5]; s ¼ 2ðPrV 2 is the cavitation number; and
1
Pc ; P0 are the pressure in the cavity and in the flow, respectively.
A comparison shows that the disc drag coefficient obtained in the experiment is
10% smaller than the theoretical calculated value for Cx0 at small values of s.
However, a comparison with results obtained in water tunnels shows that this
decrease in Cx0 is typical for water tunnels [1].
8
m ¼ rR3 ; (6)
3
m ¼ prA2 L; (7)
where 2A and L are the plate width and length, respectively, the length being greater
than the width:
112 G.Yu. Savchenko
2A ¼ h; L>2A; 2R ¼ D:
ring will be
p2 3 d 2 d
m ¼ rR 1 1þ (8)
4 D D
At d ¼ 0, Eq. (9) must give the value of Km for a disc of radius R ¼ D=2.
When Km ¼ p4 ¼ 2:5 obtained from (9) for a disc ðd ¼ 0Þ is compared with the
2
8
Km ¼ ¼ 2:67; (10)
3
Km= m*3
ρR
8
3
π2
4
2⎛ 2⎛
Km = π 1 – d ⎛ 1 + d ⎛
4 ⎝ D⎝ ⎝ D⎝
2.0
D
d
1.0
4
disc : m ¼ rR3 ;
3
Equation 9 can be used in determining the factor of added mass of a flow passage
disc in a supercavitation flow (curve D in Fig. 5).
Conclusions
Simple formulas are proposed
to calculate the drag coefficient Cx s; Dd (4) and
factor of added mass Km Dd (9) of a ducting disc in a supercavitation flow. These
formulas are convenient for use in computer programs and in engineering
calculations of cavitating water intakes and braking devices.
References
1. Knapp R, Daily J, Hammitt F. Cavitation (in Russian). Moscow: Mir Publishers; 1974.
2. Devnin SI. Aerohydromechanics of bluff structures (in Russian). Leningrad: Sudostroenie;
1983.
3. Tseitlin MYu. On the pressure on two parallel plates in a jet flow (in Russian). TsAGI
Transactions on Hydrodynamics. Moscow; 1959. p. 296–308.
4. Deinekin YuP. Cavity flow past flow passage bodies (in Russian). Gidromekhanika.
1994;68:74–8.
5. Shashin VM. Hydromechanics (in Russian). Moscow: Vysshaya Shkola; 1998.
6. Logvinovich GV. Initial motion of a body in a liquid with developed cavitation (in Russian).
TsAGI Transactions. 1959. p. 3–39.
7. Zhuravlev YuV. Entry of a disc into a liquid at an angle to the free surface (in Russian). TsAGI
Transactions; 1959. p. 227–32.
8. Gurevich MI. Impact of a plate in a separated-jet flow (in Russian). Moscow: Prikladnaya
Matematika i Mekhanika. 1952;XIV(1):116–8.
9. Savchenko YuN. Hydrodynamic forces acting on a disc executing sinusoidal oscillations
(in Russian). Izvestiya AN SSSR, Mekhanika Zhidkosti i Gaza. 1971;2:186–7.
Gas Flows in Ventilated Supercavities
Abstract
The paper is concerned with a gas flow inside a ventilated supercavity. It is found
out that there exists a ring vortex flow inside a supercavity, which forms a re-
entrant counterflow and free boundary disturbances. Attention is drawn to the
specific character in which a gas flow inside a supercavity disturbs its free
boundaries. Formulas are given to estimate the gas flow rate and the supercavity
development time and to calculate the supercavity – moving object gap.
1 Introduction
cavity. In the adopted theoretical scheme of gas entrainment, the boundary layer on
the cavity boundary was assumed to be similar to a layer formed on a rigid wall.
In reality, the water and the gas flows in a ventilated cavity are interrelated. The
gas flow via its relative velocity
V e ¼ Vc Vg ; (1)
where Vc is the velocity on the supercavity boundary; Vg is the gas flow velocity,
may affect the breaking of the free liquid cavity boundary, which in its turn affects
gas entrainment from the cavity.
This work is aimed at elucidating the gas flow pattern inside ventilated
supercavities and its effect on cavity gas entrainment.
Experiments show that the gas entrainment rate from a free supercavity is governed
by free boundary disturbances and closure conditions. Free boundary deformations
may result under the action of gravity forces, when the Froude number Fr is small
[1] and the trailing part of the supercavity starts to buoy, its cross-section deforming
and transforming into two hollow vortices [2, 3].
Periodic variations of the instantaneous cavitation number may give rise to
waves on the free supercavity boundary, thus resulting in the portion type of cavity
gas entrainment [11, 12]. Small-scale supercavity boundary disturbances may be
caused by turbulence in the inflow and cavitator vibrations.
Small-scale disturbances on the free supercavity boundary in the presence of a
gas flow at the liquid–gas interface are due to the fundamental instability of a liquid
surface described by Taylor and Helmholtz [13]. This instability manifests itself in
waves on a free water surface and droplet detachment when the relative speed of the
gas flow exceeds some critical value (12–13 m/s for water).
If the relative gas speed is far greater than the critical speed, the free boundary
breaks to form a splash layer, wherein the droplets are accelerated to the gas flow
velocity. In the cavity closure region, this splash flow forms a two-phase gas–water
mixture, which is entrained continuously or intermittently into the wake in the form
of individual portions and jets.
The literature on the effect of a wind flow on a water surface is quite voluminous
([13–16], etc.). As a result of the studies conducted, the water surface state has been
related to the wind speed and action time [14]. However, supercavities feature gas
flows with high speeds of 102–103 m/s and a characteristic action time of the order
of 101 s while atmospheric processes exhibit speeds up to 25 m/s and a far longer
action time of the order of 104–105 s. Nevertheless, the obtained results are of
interest in that they allow one to find the minimum (critical) gas speed that
corresponds to the onset of splashing.
Gas Flows in Ventilated Supercavities 117
Splashing as a kind of surface wave instability in the presence of a gas flow in the
vicinity of interfaces was described by Taylor [17]. In this work, a definition of the
instability wavelength lв is introduced: this wavelength is defined as the diameter
2rk of detached droplets
!
2pB r B2 3p g
lB ¼ f ¼ 2rk ¼ ; (2)
rg Vg
2 rg 2 Vg2 rg Vg2
where z is the liquid surface tension; Vg is the relative speed near the interface; r, rg
are the liquid and the gas density, respectively; and is the liquid dynamic
viscosity; for the water–air interaction
!
r B2
f 1:5:
rg 2 V 2g
Vg 10 m=s: (3)
rg Vg2 cr
Wecr ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 8:8: (4)
rgB
At the following parameter values: air density (T ¼ 15o) rg ¼ 1.29 kg/m3, water
density (T ¼ 15o) r ¼ 1,000 kg/m3, and water surface tension B ¼ 0.074 N/m, the
critical gas speed will be Vg cr ¼ 13.5 m/s. This value is close to theoretical
estimates by the Helmholtz and the Taylor method, where Vg cr ¼ 11 m/s.
dW
¼ Qin Qout ; (5)
dt
where W is the supercavity volume, and Qout is the gas outflow rate from the cavity.
In ventilated supercavitation, two characteristic regimes exist:
1. Formation of a ventilated supercavity, when the injection Qin makes up for the
increase of the supercavity volume dW dt >0 and the gas is not entrained yet
(Qout ¼ 0) because the supercavity is still unclosed. In this case, the gas balance
is governed by the equation
dW
¼ Qin ; (6)
dt
while the cavity pressure Pc, the inflow velocity V1, and the cavitation number
remain constant.
2. Steady flow regime, when the cavitation number s ¼ s0 is constant, the
supercavity
has reached its maximum size, and its volume remains constant
dW
dt ¼ 0 . In this case, the gas entrainment rate is equal to the gas injection rate,
and the gas balance equation has the form
A similar regime takes place at the initial stage of penetration of cavitating objects
through a free water surface or in passing through a solid underwater obstacle [4,
20]. In this regime, the velocity and pressure in a supercavity are constant, and the
gas injection rate required for supercavity formation reaches its maximum value.
The importance of this regime is due to the fact it minimizes the supercavity
development time and allows one to assess the gas injection rate required for this
purpose.
The injection process in the supercavity formation regime can be explained
using the schematic shown in Fig. 1 where the cavitator starts moving from a
Y
r
Qin V∞ X
Y
Rc
Fig. 1 Supercavity Qin V∞ X
formation scheme with gas 0.5 Lc
Lc
injection
Gas Flows in Ventilated Supercavities 119
solid impermeable wall at a constant velocity. The schematic shows the develop-
ment of a supercavity to its full length Lc when the gas is injected into the cavity in
the vicinity of the cavitator.
If the cavity contour is assumed to be near-elliptical, then its radius can be
calculated as
x ð Lc x Þ
Rx 2 ¼ ; (8)
l2
sffiffiffiffiffiffiffi
ln s1 Lc
l¼ ¼
s 2 Rc
where l is the cavity aspect ratio, and the cavity volume can be estimated as
2
Wc ¼ p Lc R2c : (9)
3
p x ð Lc x Þ
Sc ¼ p R2x ¼ ;
l2
rffiffiffiffiffiffiffiffiffiffiffiffiffi rffiffiffiffiffi
2 Rn 1 Cx
Lc ¼ Cx ln ; Rc ¼ Rn
s s s
where Lc ; Rc are the cavity length and midsection radius, Rn, Cx are the cavitator
radius and drag coefficient, and the rate of increase of the cavity volume for the
cavitator moving at a constant velocity V1 is
dW px
¼ Qn ¼ p R2x V1 ¼ 2 ðLc xÞ V1 (10)
dt l
In view of the expression for the supercavity length Lc, the law of variation of the
injection rate Qin can be written as
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi !
p x s V1 2 Rn 1
Qin ðxÞ ¼ Cx ln x : (11)
ln s1 s s
The maximum gas injection rate will correspond to the maximum cross-section
of the supercavity (x ¼ Lc/2, Cx ¼ Cx0(1 + s))
120 Yu.N. Savchenko and G.Yu. Savchenko
1þs
Qin max ¼ p R2c V1 ¼ p R2n V1 Cx0 : (12)
s
For a steady motion W_ ¼ 0, and the required gas injection rate will only be limited
by the cavity gas entrainment rate Qout.
A water tunnel experiment on free axisymmetric cavities downstream of a disc
has shown that the steady flow inside a cavity consists of two regions: a through
flow region FP and a circulation flow region FR (Fig. 2).
The circulation flow inside a ventilated supercavity downstream of a disc
cavitator was observed in water tunnel experiments conducted at the Institute of
Hydromechanics of the National Academy of Sciences of Ukraine. Visualization
was made using aluminium powder added to the gas being injected. In the
1 2 3 4 5
Qin
FR Qout
Fig. 2 Scheme of gas flow V∞
inside the cavity FP
Gas Flows in Ventilated Supercavities 121
Ve ¼ Vo þ Vc 2V1 :
Section S2 is the start of the circulation flow. It also gives the limit dimension of a
supercavity with section F2 where the injection rate Qin is still sufficient for the
supercavity cross-sections to expand. Cross-section F2 is determined by the
conditions (10) and (15):
x2 ðLC x2 Þ
Qin ¼ F2 V1 ¼ p R22 V1 ¼ s p V1 (16)
ln s1
Qin
h¼ ; (17)
p Dc V1
pðDc 2hÞ2
FR ¼ : (18)
4
The region S3 – S4 is the cavity closure region where a re-entrant flow is formed
in the central part of the circulation flow.
Section S4 is the end of the circulation flow.
122 Yu.N. Savchenko and G.Yu. Savchenko
Section S5 is the end of the supercavity and the start of its hydrodynamic wake
where the volume of the gas determines its entrainment rate Qout from the cavity.
In a steady flow where the velocity, depth, injection rate, and cavitation number
are constant, the gas entrainment rate is equal to the injection rate (7).
6 Gas Entrainment
The 0.001 s supercavity surface photos show the presence of sizeable disturbances
(Fig. 3). Radial disturbances increase toward the trailing part of the cavity, and their
size is about 1/100 of the full cavity length Lc. A similar pattern of destruction is
shown by free jets issuing from a nozzle into the air. As shown schematically in
Fig. 4, the gas fills voids in free boundary defects outside of the undisturbed cavity
contour, and an equivalent volume of liquid in the form of droplets finds itself
inside the supercavity contour.
It is quite natural to assume that on cavity closure the gas that has filled the
defects outside of the disturbed boundary will remain in the cavity wake and the
splashes will form foam, which will also be entrained into the wake, but by another
entrainment mechanism.
The assumptions made allow one to estimate the gas entrainment rate from the
gas content in the vicinity of the supercavity midsection
Fig. 3 Supercavity surface photos with different exposition time (1/60 s and 1/1000 s)
V∞ DC
Fig. 4 Scheme of the
LC
disturbed supercavity surface
Gas Flows in Ventilated Supercavities 123
d ¼ K1 Lc ðReL Þn ; (20)
where К1 and n are to be determined for the conditions on the free cavity boundary.
The gas entrainment rate Qout and the gas entrainment coefficient Qout will be
Ve Lc pffiffiffiffiffiffiffiffiffiffiffi Lc
ReL ¼ ¼ V1 1 þ s Vg ;
2n 2n
pffiffiffiffiffiffiffiffiffiffiffi
Ve ¼ Vc Vg ¼ V1 1 þ s V g :
Here, Cx0 ¼ 0.82 for a disc, ReL is the Reynolds number based on the cavity
length, and Ve is the relative velocity.
It is easy to see that the form of Eqs. 20–22 opens up new possibilities for
varying the gas entrainment rate by varying the relative velocity in the number ReL.
Assuming that the layer thickness in Eq. 20 depends solely on the cavity length
d ¼ d(Lc), the gas entrainment rate (22) will be a function of the cavitation number
alone
rffiffiffiffiffiffiffiffiffiffiffiffiffi
1þs 1 1
Qout ðsÞ ¼ Const ln : (23)
s s s
An equation of the form of (23) has also been derived using dimensional theory
alone [19], thus confirming its fundamentality for the gas entrainment process.
124 Yu.N. Savchenko and G.Yu. Savchenko
Lðb
dðW Wb Þ
¼ Qin Qout : (25)
dt
Qin
Vg ðxÞ ¼ ; (26)
p hð R x þ R b Þ
In the case of a ventilated wall supercavity, the gas will be injected thereto from
the solid wall using special nozzles (Fig. 5). In this case, the velocity gradient on the
free boundary may be both positive and negative, and the relative velocity Ve ¼
Vc Vg may reverse sign depending on the gap width h and the gas entrainment
rate Qout ¼ Qin.
pffiffiffiffiffiffiffiffiffiffiffi Qin
Ve ¼ Vc Vg ¼ V1 1 þ s : (28)
p hðRx Rb Þ
Writing the condition for the absence of disturbances on the supercavity bound-
ary (4) in the form
makes it possible to obtain the relationship for choosing optimum values of the gap
h and the injection rate Qin
Gas Flows in Ventilated Supercavities 125
pffiffiffiffiffiffiffiffiffiffiffi
pðRx þ Rb Þ V1 1 þ s Vgcr < Qin ; (30)
h
Conclusions
It is suggested that Helmholtz–Taylor-type mechanisms of instability of the free
boundary of a supercavity under the action of a gas flow be taken into account
when considering supercavity boundary disturbances and calculating the gas
entrainment rate.
Special gas injection regimes in the formation of a supercavity and for a
steady supercavity are pointed out. Formulas are given to estimate the required
gas injection rate.
A circulation flow and a through flow inside a supercavity are pointed out.
Photos of disturbed supercavity boundaries are presented.
It is shown that there can exist near-wall gas flows with both a positive and a
negative velocity gradient in the gap between the solid body boundary and the
supercavity boundary.
References
1. Egorov IG. Ventilated cavitation (in Russian). In: Egorov IG, Sadovnikov YuM, Isaev II.
Leningrad: Sudostroenie; 1971. 284pp.
2. Logvinovich GV. Free-boundary flow hydrodynamics (in Russian). Kiev: Naukova Dumka;
1969.
3. Epshtein LA. Methods of dimensional theory and scaling in vessel hydromechanics problems
(in Russian). Leningrad: Sudostroenie; 1970.
4. Savchenko YuN, Semenenko VN, Putilin SI. Nonstationary processes in the motion of
supercavitating bodies (in Russian). Prykladna Gidromekhanika. 1999;1(1):62–80.
5. Savchenko YuN, Semenenko VN, Putilin SI, Savchenko G, Naumova E. Designing the high-
speed supercavitating vehicles. Proceedings of the 8th International Conference on Fast Sea
Transportation (FAST’2005), St. Petersburg; June 2005. p. 1–7.
6. Cox PN, Claiden WA. Air entrainment of the rear of a steady cavity. Cavitation in Hydrody-
namics. Proceedings of the Symposium, London; 1955.
7. Krylov VV. Experimental data on air entrainment from a cavity formed by air injection (in
Russian). Moscow: TsAGI Transactions; 1961. Issue 824, 284pp.
8. Epshtein LA. On the mechanism of pulsation processes in the trailing part of attached cavities
(in Russian). Proceedings of the Symposium on the Physics of Acustic/Hydrodynamic Phe-
nomena. Moscow: Nauka; 1975. p. 133–8.
126 Yu.N. Savchenko and G.Yu. Savchenko
9. Spurk IH. A theory for the gas loss from ventilated cavities. Proceedings of the International
Science School “High Speed Hydromechanics”. Cheboksary; 2002. p. 191–5.
10. Spurk IH. On the gas loss of ventilated supercavities. Acta Mech. 2002;155(3–4):125–35.
11. Semenenko VN. Instability and oscillation of gas-filled supercavities. Proceedings of the Third
International Symposium on Cavitation, Grenoble; April 1998. p. 25–30.
12. Franc J-P, Michel J-M. Fundamentals of cavitation. Dordrecht/Boston/London: Kluwer; 2004.
13. Ariel’ NZ, Bortkovsky RS. Refined model of the energy and mass exchange of splashes over a
storming ocean surface (in Russian). In: Typhoon-75 (Expedition materials). Leningrad; 1978.
Vol. 2, p. 101–15.
14. Bortkovsky RS. Atmosphere–ocean heat and water exchange in a storm (in Russian).
Leningrad: Gidrometeoizdat; 1983.
15. Preobrazhensky LYu. Estimation of the droplet and splash content in the surface layer of the
atmosphere (in Russian). Trudy GGO; 1972. Issue 282, p. 194–9.
16. Savchenko GYu. Hydrodynamics of wall supercavity flows (in Russian). Kiev: PhD thesis;
2009. 138pp.
17. Lane WR, Green HL. The mechanics of drops and bubbles. In: Batchelor GK, Davies RM,
editors. Surveys in mechanics. London/New York: Cambridge University Press; 1956.
p. 162–215.
18. Birkhoff G. Hydrodynamics, methods, facts and similarity (in Russian). Moscow: Izdatelstvo
Inostrannoi Literatury; 1963.
19. Mamenko YuN. Determination of conditions for the onset of splashing on a liquid surface
acted upon by a plane gas jet (in Russian). Transactions of Kaliningrad Technological Institute
of Fish Industry, Kaliningrad, USSR; 1980. Issue 90, p. 109–12.
20. Savchenko YuN, Semenenko VN. Wave generation on the boundaries of supercavities formed
in the water entry of a disc and cones (in Russian). Problems in high-speed hydrodynamics
(Transactions). Cheboksary: Chuvash University; 1993. p. 231–9.
Generalized Self-Similar Problem of Impact
Between Liquid and Solid Wedges
Abstract
This paper presents a generalized solution of the self-similar problem of impact
between solid and liquid wedges, which includes as a special case the problem of
asymmetric wedge entry into a liquid. The solution method is based on the
construction of an analytical expression for the complex flow potential in a
parameter region. From the dynamic and the kinematic boundary condition,
integral equations are obtained for the determination of the velocity magnitude
and angle with the free surface, which appear in the expression for the complex
potential. The free boundary shape, the pressure distribution along the wedge,
and the hydrodynamic force coefficients are calculated over a wide range of
given data.
1 Introduction
Problems of body entry into a liquid are of great importance in such applications as
the study of ship roll and pitch and offshore platform behavior in a heavy sea and
the design of planing boats, semisubmerged propellers, and seaplanes.
The initial stage of body-liquid interaction is of a shock nature, and it is
characterized by a high load level and substantial interplay between nonlinear
and unsteady effects. This class of problems involves additional difficulties,
which are due to the presence of a free boundary and a three-phase contact line at
the body-liquid-air interface. The body-liquid entry problem is overviewed in
considerable detail in Korobkin and Pukhnachev [1]. By now, only the symmetric
entry of a wedge into a liquid with an unperturbed free surface at a constant velocity
has been studied in detail.
In actual practice, the water surface interacts with the atmosphere, as a result of
which its shape becomes wavy. In this connection, a more accurate evaluation of
loads arising from hydroplaning, seaplane landing, and offshore platform – heavy
sea interaction calls for taking into account the actual shape of the free surface. As
known from wave theory, the limiting wave crest form is a 120 wedge. The model
of solid body – liquid wedge interaction presented in this paper allows one to
evaluate the effect of the wedge shape of a liquid on the unsteady hydrodynamic
loads arising from impact on a solid body. In this model, a flat free surface is a
special case.
Karman [2] and Wagner [3] were the first who obtained an approximate solution
of the entry of a wedge into a flat-surface liquid at small deadrise angles. They
considered the water entry process as a sequence of pulses generated by the body-
liquid impact. The study was continued, in particular by Garabedian [4], Borg [5],
Logvinovich [6], and Moiseev [7]. Gonor [8] considered the entry of a thin wedge.
Mackie [9] obtained a complete linearized solution. Logvinovich [6] showed,
among other things, that as the deadrise angle approaches zero, the liquid speed
in the tip jets becomes twice that in the flow turn and tip jet formation region.
Dobrovol’skaya [10] obtained a complete solution to the nonlinear problem of
the vertical entry of a symmetric wedge by constructing the Wagner function. She
found the Wagner function by conformal mapping and formulated a boundary-
value problem for the mapping function in such a way as to find it using Schwartz’s
integral formula. Dobrovol’skaya derived a singular integral equation in the real
part of the mapping function on the free surface and presented numerical results for
wedge angles less than 30 . A numerical solution of the singular integral equation
for larger wedge angles was given by Zhao and Faltinsen [11], Keady and Fowkes
[12], and Freankel [13, 14].
Logvinovich [6] set forth a theory of the entry of finite-size bodies, for which
Wagner’s assumptions do not hold, and his theoretical estimates were verified by
experiment. Since the forces that arise from high-speed water entry may be far
greater than their steady values, liquid compressibility and body elasticity may play
a significant role. These issues were considered by Sagomonian [15], Troshin et al.
[16], Kubenko [17], and Korobkin [18].
Cavitation in axisymmetric body entry was simulated using Logvinovich’s
semiempirical theory by Zhuravlev [19] and then by Savchenko and Semenenko
[20, 21]. They found and explained the interesting effect of wave generation on
cavities, which is due to oscillations of the air entrained by the cavity from the
atmosphere.
The application of numerical methods (the finite-element and the boundary-
element method) to water entry problems presents difficulties because of the salient
points of the flow boundary at the three-phase contacts. At these points, the flow
potential has singularities, thus making it difficult to obtain reliable numerical
results. As a consequence, in [11, 22] the tip jet flow region was truncated, and in
[21] an asymptotic analytical solution was used for the tip jet flow region. The
numerical results for the entry of an asymmetric wedge reported in the literature are
limited to small angles between the wedge axis and the vertical, at which effects of
Generalized Self-Similar Problem of Impact Between Liquid and Solid Wedges 129
asymmetric interaction between the wedge sides and the liquid are rather
insignificant.
This paper presents a generalized solution of the problem of impact between a
solid and a liquid wedge. A liquid wedge angle of 180 corresponds to an unper-
turbed flat surface. This case for the vertical and the oblique entry of a wedge into
an unperturbed liquid has been considered in [23–26].
This paper also discusses the limit combination of the horizontal component of
the entry velocity and the wedge orientation at which flow separation occurs and
only one wedge side interacts with the liquid [27]. The solution method is based on
the construction of an analytical expression for the complex flow potential in a
parameter region. From the dynamic and the kinematic boundary condition, integral
equations are obtained for the determination of the velocity magnitude and angle
with the free surface, which appear in the expression for the complex potential. The
free boundary shape and hydrodynamic force coefficients are calculated.
As distinct from the extensive literature on the vertical entry of a symmetric wedge
into a liquid, few works are devoted to a more general problem – the vertical entry
of an asymmetric wedge. Garabedian [4] and Borg [5] were the first to attack this
problem. More recent publications on the subject are Chekin [28] and Korobkin
[29].
Chekin’s method is based on the use of Sokhotski’s integral formula to find a
function that maps the upper half-plane onto a stationary flow plane in self-similar
variables. Chekin’s solution for asymmetric wedge entry assumes flow separation at
the wedge vertex followed by flow detachment to a wedge side. However, the
experiments by Judge and others [30, 31] have shown that this assumption does not
hold. Only in the special case where the stagnation point is at the wedge vertex,
Chekin’s solution describes nonseparated flow.
In [32], an attempt was made to estimate the hydrodynamic flow characteristics
for the entry of an asymmetric wedge by constructing an approximate physical and
mathematical model, which includes an approximate expression for the complex
flow potential. However, the calculated results differ considerably from the experi-
mental ones [30, 31].
In this paper, a similar approach is used, and the free boundary of the liquid is
assumed to be in the form of a wedge. The velocity at infinity has a magnitude V1
and makes an angle g1 with the symmetry axis of the liquid wedge; the bisector of
the solid wedge makes an angle d with the velocity direction. The angle d is positive
if the wedge is rotated counterclockwise from the velocity vector. Let 2a be the
wedge angle; then bL ¼ p a d g1 and bR ¼ g1 a þ d are the angles of the
left and the right wedge sides with the horizontal. For a constant entry velocity,
the introduction of the self-similar variables x ¼ X=V0 t and y ¼ Y=V0 t where t is
the time allows one to transform the time- In [23], a solution for reversible flow is
130 Yu.N. Savchenko and Yu.A. Semenov
O x
2 μR
μ α
A →
C n
μ∞
D′ γ D
′
V∞
b B ih
D′ i
c a x
O C A B
where ’ and c are the velocity potential and the stream function in the stationary
plane z. The function wðzÞ is to be found.
Following Joukovski [33], the solution is sought for by constructing two
functions: the complex velocity, dw=dz, and the derivative of the complex potential,
dw=dB, in the region of a parametric variable, B ¼ x þ i, for which the first
quadrant is chosen. Conformal mapping allows us to fix three arbitrary points in
the parameter region. If such points are O, B, and D (Fig. 1), then the wetted surface
of the wedge and the free surface will correspond to the real and the imaginary axis,
respectively. The stagnation point A and the wedge vertex C correspond to the
points B ¼ a and B ¼ c in the parameter plane, which are to be found as part of the
solution.
Generalized Self-Similar Problem of Impact Between Liquid and Solid Wedges 131
If the functions dw=dz и dw=dB are found, the velocity field in the parameter
plane is known, and the parameter plane and the stationary plane z are related to
each other as
ðB
dw dw dw
vx ivy ¼ ðBÞ; zðBÞ ¼ zð0Þ þ dB; (2)
dz dB dz
0
The boundary-value problem for the complex velocity function can be formulated
as follows. For now, let us assume that the velocity modulus along the free surface,
i.e. along the imaginary axis of the first quadrant is known
dw
vðÞ ¼ ; 0<<1; x ¼ 0; (3)
dz
makes it possible to solve the boundary-value problem (3) and (4) in the first
quadrant of the complex plane B and find the complex velocity as
132 Yu.N. Savchenko and Yu.A. Semenov
dw Ba B þ c ð12a=pÞ
¼ eiðg1 þaþdÞ
dz Bþa Bc
2 3
ð
1
i dlnv i B
þ exp4 ln d5: (6)
p d i þ B
0
To analyze the behavior of the velocity potential along the free boundary, it is
convenient to introduce two unit vectors, n and t, which are normal and tangent to
the free boundary, respectively. The normal unit vector points outward from the
liquid, and the tangent vector is directed so that the arc coordinate s increases in the
direction for which the liquid is on the left (Fig. 1). With this notation
where vs and vn are the tangential and normal velocity components, respectively.
Let y be the angle between the velocity on the free boundary and the unit vector t:
y ¼ tan1 ðvn =vs Þ. The behavior of the function y along the liquid boundary is
shown in Fig. 2.
Equation 7 gives the argument of the derivative dw=dB of the complex potential
dw dw ds
#ðBÞ ¼ arg ¼ arg þ arg
dB ds dB
(8)
y; 0<x<1; ¼ 0;
¼
y p=2; x ¼ 0; 0<<1:
Consider the behavior of the function yðBÞ along the whole of the liquid
boundary, i.e. along the real and the imaginary axis of the parameter region.
When moving along the free boundary from point O to point D, the function yðBÞ
increases from the value mR at B ¼ 0 to the value m1 þ g, which corresponds to the
velocity direction at infinity (the point B ¼ i). To reach the left infinite point of the
free boundary where the function y takes the value m1 g with the velocity
direction unchanged and equal to its value at infinity, we have to move along a
closed contour of infinite radius. To this contour there corresponds a circular arc of
infinitesimal radius centered at the point B ¼ i. On going around this semicircle, the
function yðBÞ is incremented by DyD ¼ ðp þ 2m1 Þ. The continuous variation of
the function yðBÞ is shown in Fig. 2 as solid lines, and its step changes are shown as
dashed lines.
Generalized Self-Similar Problem of Impact Between Liquid and Solid Wedges 133
D′
m•–g O+
B- m•+g
mL mR
O–, A+ B+, C, A–
Then the function yðBÞ varies continuously when moving along the free bound-
ary from point D0 to point B. On the interval a<x<1, ¼ 0, which corresponds to
the left side of the wedge and the part of the right side between points C and A, the
function yðBÞ 0 since vn ¼ 0 and vs >0. On the interval 0<x<a, ¼ 0 the
function yðBÞ p since vn ¼ 0 and vs <0. Thus at the point B ¼ a the function
yðBÞ undergoes the step change DyA ¼ p. One more step change, DyO ¼ mR p,
takes place at point O on going in an infinitesimal neighborhood of the point B ¼ 0
from the wedge side, x>0; ¼ 0, to the free surface, x ¼ 0; >0. The function
yðBÞ can be expressed in terms of the continuous function lðBÞ defined as follows:
8
>
> lðBÞ; x ¼ 0; 0<<1;
<
lðBÞ þ DyD ; x ¼ 0; 1<<1;
yð B Þ ¼ (9)
>
> lðBÞ þ DyO ; 0<x<a; ¼ 0;
:
lðBÞ þ DyO þ DyA ; a<x<1; ¼ 0;
which gives a solution of the boundary-value problem (8) in the first quadrant of the
complex plane B. Here, K is a real constant, and w1 ¼ wðxÞjx¼1 .
134 Yu.N. Savchenko and Yu.A. Semenov
Substituting Eqs. 8 and 9 into the first integral in (10) for B varying along the real
axis of the parameter region and into the second integral for B varying along its
imaginary axis and evaluating the integrals over the step changes of the function
yðBÞ, we obtain the following expression for the derivative of the complex potential
in the parameter plane
dw B 2 a2
¼ KB2mR =p1
dB ðB2 þ 1Þ1þ2m1 =p
2 1 3
ð
1 dl
exp4 ln B þ d 5:
0 0
2 2
(11)
p d0
0
Integrating Eq. 7 in the parameter region gives the function that conformally
maps the first quadrant onto the domain of variation of the complex potential
ðB
ð B 2 a2 Þ
wðBÞ ¼ wð0Þ þ K B2mR =p1
ðB2 þ 1Þ1þ2m1 =p
0
2 3
ð
1
1 dl 2
exp4 ln B þ 0 d0 5dB:
2
(12)
p d0
0
The velocity potential at point O, wð0Þ, can be put equal to zero. From Eqs. 6 and
11 we obtain the derivative of the mapping function
2 12a
dz ð B þ aÞ Bc p
¼ KBð2m1 =p1Þ
dB ð1 þ B2 Þ1þ2m1 =p Bþc
2 1
ð
1 dl 0 2 2 0
exp4 ln þB d þ ig1
p d
0
3
ð
1 0
i d ln v i B
þ ln d0 þ id þ ia5; (13)
p d0 i0 þ B
0
whose integration in the parameter region gives the shape of the free surface.
The unknown parameters a; c; K can be found from the condition for the flow
velocity at infinity and the condition for the length of the wetted parts of the wedge
OC и CB.
Generalized Self-Similar Problem of Impact Between Liquid and Solid Wedges 135
The Cauchy–Lagrange integral written in the physical region for point O and an
arbitrary point in the flow has the form
@F V 2 P @F V 2 Pa
þ þ ¼ þ 0þ ; (14)
@t z 2 r @t z¼0 2 r
dv v s sin y dy
¼ : (15)
ds v þ s cos y ds
dv v s sin y dy
¼ ; (16)
d v þ s cos y d
where
ð ð 2mR =p1
dz 2 þ a2
sðÞ ¼ d ¼ K
dB B¼i vðÞ ð1 2 Þ1þ2m1 =p
0 0
2 1 3
ð
1 dy 0 2 2 0 5
exp4 ln d d: (17)
p d0
0
On the free surface, the pressure is constant, and thus the acceleration of a liquid
particle is normal to the free surface. This condition can be written as [35]
@U
Re dZ ¼ 0: (18)
@t
136 Yu.N. Savchenko and Yu.A. Semenov
ð
1
1 d ln v 1 d ln v 0
þ d0
2 tan y d p d0 0 2 2
0
a 2a c
¼ 2 þ 1 : (19)
a þ 2 p c2 þ 2
The system of Eqs. 16 and 19 is closed and allows one to find the functions vðÞ
and yðÞ:
On the wedge surface, the normal velocity component is zero, and hence yR ¼ p
at point O and yL ¼ 0 at point B. At the same points on the free surface, the function
yðÞ is determined from the solution of the total system of equations. Thus the
contact angles mR and mL of the free surface with the wedge sides are
mR ¼ yR p lim yðÞ ¼ yð0Þ; (20)
!0
mL ¼ p lim yðÞ yL ¼ p lim yðÞ: (21)
!1 !1
2ðf þ s vÞ þ ð1 vÞ2
cp1 ðsÞ ¼ ; 0 x a; (22)
v21
from the stagnation point on left wedge side to the wedge vertex and on the right
wedge side
2ð f s v Þ þ ð 1 þ v Þ 2
cp2 ðsÞ ¼ ; a x 1; (23)
v21
dw
fðxÞ ¼ < wðBÞjB¼x ; vðxÞ ¼ ;
dz B¼x
ðx
dz
sðxÞ ¼ dx0 :
dB B¼x0
0
ða ðc
1 ds 1 ds
CnR ¼1 2 cp1 ðsÞ dx þ 1 2 cp2 ðsÞ dx; (24)
2 rV1 H 2 rV1 H
dx dx
0 a
ð
1
1 ds
CnL ¼1 2 cp2 ðsÞ dx: (25)
2 rV 1H dx
c
The section of the wedge by a straight line perpendicular to the velocity vector is
chosen as the characteristic length
H ¼ V1 t½ða dÞ þ ða þ dÞ:
3 Calculated Results
The system of the integral equations (16) and (19) is solved by the method of
successive approximations, in which the function dlnv=d is found from the
integral equation (19) in the ðk þ 1Þ-th iteration by taking the Hilbert
transformation
ð
1
d ln v ðkþ1Þ 4 1 d ln v a
¼ þ 2
d p 2 tan y d0 a þ 0 2
0
k
2a c 0
þ 1 2 d0 : (26)
p c þ 0 2 0 2 2
From Eq. 16 we can find the ðk þ 1Þ-th approximation for the derivative dy=d,
138 Yu.N. Savchenko and Yu.A. Semenov
kþ1 kþ1
dy vkþ1 þ sk cos yk d ln v
¼ ; (27)
d sk sin yk d
whose integration allows us to find the ðk þ 1Þ-th approximation for the function
yðÞ. The functions dy=d and yðÞ have a singularity at point O as can be seen
from Eqs. 26 and 27. On the other hand, these functions are bounded by their
definition. To evaluate the integral in Eq. 26, the lower limit of integration is put
equal to some small e. As the initial approximation, we put vðÞ V0 , yðÞ 0.
In discrete form, the functions dy=d и yðÞ are approximated by piecewise
constant functions on the intervals 0<i <1, which correspond to the right portion
of the free boundary. The fixed points i , i ¼ 1; N=2 are distributed along the
imaginary axis of the parameter region as a geometric series with the first term
1 ¼ 104 . On the intervals 1<i <1, i ¼ N=2 þ 1; N, which correspond to the left
portion of the free boundary, the points i are specified as i ¼ 1=Ni . The length
of the segment s1 nearest to point Ois calculated analytically using Eq. 17
0 1
ð
1
2mR =p
2 dy 0 A 1
s1 ¼ Ka2 exp@ ln 0
d : (28)
p d0 2mR =p
0
2m =p
K N L
sN ¼ : (29)
vB 2mL =p
To verify the obtained theoretical results and to analyze the effect of the distribution
of the points i and their number N on the calculation accuracy, the vertical entry of
a wedge into a liquid was calculated. This case is considered in detail in [9–11, 13].
Figure 3 shows the calculated results for an impact between a solid and a liquid
wedge, both of vertex angle 45 , which are compared with those obtained in [36]
using the boundary-element method. In [38], the problem is formulated using self-
similar variables. As illustrated, the calculated results are in reasonable agreement.
Figure 4 shows the profile of the free boundary and streamlines for an impact
between a solid and a liquid wedge. The wetted part of the wedge side is also shown
as a solid line, and its extension is shown as a dashed line. The case in Fig. 4b
corresponds to a cumulative motion. In the tip jets, the liquid speed is several
times as high as the upstream speed, and the horizontal velocity component is of
opposite sign.
Generalized Self-Similar Problem of Impact Between Liquid and Solid Wedges 139
Fig. 3 Streamlines (a) and pressure distributions along the wedge sides (b) for solid and liquid
wedge half-angles a ¼ 45 and g ¼ 45 . The results obtained in [36, 37] using the boundary-
element method are shown as dashed lines
140 Yu.N. Savchenko and Yu.A. Semenov
Fig. 4 Streamlines for an impact between a solid and a liquid wedge with half-angles a ¼ 20 ,
g ¼ 140 (a) and a ¼ 140 , g ¼ 20 (b)
The pressure distribution for the cases in Fig. 4 is shown in Fig. 5. One can see a
local pressure increase in the tip jet root, which accelerates the liquid in the tip jet.
This is the so-called slamming effect for the case in Fig. 4a or the cumulative effect
for the case in Fig. 4b. In the latter case, as can be seen from Fig. 5, the pressure
coefficient is several times as high as the pressure at the stagnation point for steady
flows.
The wedge orientation with respect to the free surface and the horizontal velocity
component may be such that the liquid at infinity moves away from one of the
wedge sides. With such a combination of given parameters, flow separation may
occur at the wedge vertex, which changes the flow topology and calls for problem
re-formulation. When flow separation occurs at the wedge vertex, only one wedge
Generalized Self-Similar Problem of Impact Between Liquid and Solid Wedges 141
Fig. 5 Pressure distribution along the wedge sides for the cases in Figs. 4a (solid line) and Fig.4b
(dashed line)
side interacts with the liquid, and the resulting flow corresponds to the entry of
a plate.
The experimental and theoretical studies of flow separation and the formation of
a ventilated cavity described in [30] made it possible to determine the angle of the
wedge velocity at which the wedge flow changes to the plate flow. Based on the
solution of the problem of the vertical entry of an asymmetric wedge, a flow
separation criterion was proposed in [25]. The criterion depends on the determina-
tion of the normal force acting on a plate: according to it, flow separation occurs if
the total force acting on one of the wedge sides becomes zero. The criterion is
formulated without regard for the pressure on the free surface, which is justified for
high wedge entry speeds. However, entry speeds may be moderate in such
applications as slamming and the carrying capacity of high-speed planning boats.
If in the solution presented above the liquid wedge vertex angle is put equal to
p=2, the flow will correspond to the oblique entry of a wedge into an unperturbed
liquid whose flat surface is inclined at angle g to the horizontal. Let O be the angle
between the wedge symmetry axis and the perpendicular to the free surface and g
be the angle between the velocity and the free surface.
Figure 6 shows the flow configuration and streamlines corresponding to the limit
angle g for angles O ¼ 0 (symmetric entry with respect to the free surface) and
O ¼ 20 . It can be seen that the streamlines on the left side of the wedge become
nearly rectilinear and uniform.
142 Yu.N. Savchenko and Yu.A. Semenov
This means that the velocity in this portion of the free surface differs only
slightly from the upstream velocity. There can be seen some elevation of the free
surface, which is due to liquid inflow from the right.
The tip jet contact angles mR and mL versus the angle g are shown in Fig. 7. The
contact angle mR decreases and the contact angle mL increases as the angle g
increases to some limit value, which depends on the wedge orientation angle O.
To the minimum angle g at which the system of the integral equations (16) and (19)
can be solved corresponds contact angle mL =p ¼ 0:1 for all wedge orientation
angles in Fig. 8. This is in agreement with the results reported in [10–12] and
[14] for the symmetric entry of a wedge. As a ! 0, the contact angle m=p ! 0:1.
Since for deadrise angles b<p=2 contact angles mL =p>0:1 are nonexistent, it can be
assumed that the value of g at which mL =p ¼ 0:1is the limit value of the velocity
angle at which a solution of the problem is existent for a given wedge orientation.
Calculated entry angles g versus the wedge rotation angle O are shown in Fig. 8
and compared with the experimental values of the velocity angle corresponding to
the onset of flow separation at the wedge vertex reported in [30]. The figure also
shows the entry angles at which the force coefficient on the left wedge side is zero
(dashed line), which corresponds to the flow separation criterion proposed in [25].
Generalized Self-Similar Problem of Impact Between Liquid and Solid Wedges 143
Fig. 8 Boundary of the region of nonseparated flow in the plane of the wedge rotation angle O and
the wedge entry velocity angle g : calculated results (solid line), experimental data [30] (squares),
and the criterion CnL ðg ; OÞ ¼ 0 (dashed line). The region of nonseparated flow is below the lines
As illustrated, the criterion gives somewhat smaller wedge entry angles g than
those determined in the experiment.
Figure 9 shows the hydrodynamic force coefficients on the right (a) and the left
(b) wedge sides versus the entry velocity angle for different wedge orientations with
respect to the free surface.
144 Yu.N. Savchenko and Yu.A. Semenov
For small values of g, for which the velocity is nearly perpendicular to the
unperturbed free boundary, the normal force coefficients vary nearly linearly:
with increasing g the normal force coefficient increases on the right wedge side
(a) and decreases on the left one (b).
At some g, the coefficient CnL becomes negative. Negative values may occur if
the average pressure P on the wedge side satisfies the condition Pv <P<P
a as can be
seen from the expression cpR ¼ ðP Pa Þ=rVy wherePv is the vapor pressure and Vy
2
Conclusions
We have presented a generalized solution of the self-similar problem of impact
between solid and liquid wedges, which includes as a special case the problem of
the asymmetric and oblique entry of a wedge into a liquid. The solution method
is based on constructing an analytical expression for the complex flow potential
and finding a function that conformally maps the parameter region onto the flow
region in the physical plane.
Generalized Self-Similar Problem of Impact Between Liquid and Solid Wedges 145
The flow geometry and the pressure distribution along the wedge sides are
analyzed numerically over a wide range of given parameters. It is shown that at
solid wedge angles greater than 180 the solution describes the cumulative
effect, which consists in a many-fold increase in the liquid speed in the tip jets
in comparison with the upstream speed.
The effect of the horizontal velocity component for wedge entry into a liquid
with a flat free surface is studied, and the limit combinations of given parameters
corresponding to flow separation from the wedge vertex are identified. The
calculated and the experimental regions of nonseparated wedge entry in the
velocity angle – wedge rotation angle plane are in good quantitative agreement.
For the limit combination of given parameters, the contact angle on the
leeward side reaches its maximum value 18 , which is also typical for the
symmetric entry of a wedge with a vertex angle approaching zero.
References
1. Korobkin AA, Pukhnachov VV. Initial stage of water impact. Ann Rev Fluid Mech.
1988;20:443–7.
2. von Karman T. The impact of seaplane floats during landing. Technical Note 321. Washington,
DC: NACA; 1929.
3. Wagner H. Uber Stoss Gleitvorgange an der Oberache von Flussigkeiten. Z Angew Math
Mech. 1932;12:192–215.
4. Garabedian PR. Oblique water entry of a wedge. Comm Pure Appl Math. 1953;6:157–65.
5. Borg SF. Some contributions to the wedge-water entry problem. Proc Am Soc Civil Engrs J
Eng Mech Div. 1957;83(EM2):1214.
6. Logvinovich GV. Hydrodynamics of free-boundary flows (in Russian). Kiev: Naukova
Dumka; 1969.
7. Moiseev NN, Borisova EP, Koryavov PP. Plane and axisymmetric self-similar problems of
liquid entry (in Russian). Prikladnaya Matematika i Mekhanika. 1959;23:490–507.
8. Gonor AL. Entry of a thin wedge into a liquid (in Russian). Doklady Akademii Nauk SSSR.
1986;290:1068–72.
9. Mackie AG. The water entry problem. Q J Mech Appl Math. 1969;22:1–17.
10. Dobrovol’skaya ZN. Some problems of similarity flow of fluid with a free surface. J Fluid
Mech. 1969;36:805–29.
11. Zhao R, Faltinsen O. Water entry of two-dimensional bodies. J Fluid Mech. 1993;246:
593–612.
12. Keady G, Fowkes N. The vertical entry of a wedge into water: integral equations and
numerical results. Third Biennial Engineering Mathematics and Applications Conference
(EMAC’98). Adelaide; 1998. Vol. 1, p. 277–81.
13. Freankel E. Problems of similarity flow of fluid with a free surface. J Fluid Mech.
1969;36:805–29.
14. Fraenkel LE, Keady G. On the entry of a wedge into water: the thin wedge and an all-purpose
boundary-layer equation. J Eng Math. 2004;48(3):219–52.
15. Sagomonyan AYa. Penetration (in Russian). Moscow: Moscow State University; 1974.
16. Eroshin VA, Romanenkov NI, Serebryakov IV, Yakimov YuL. Hydrodynamic forces in blunt
body entry into a compressible liquid (in Russian). Izvestiya Akademii Nauk SSSR,
Mekhanika Zhidkostei i Gazov. 1980;6:44–51.
17. Kubenko VD. Entry of elastic shells into a compressible liquid (in Russian). Kiev: Naukova
Dumka; 1981.
146 Yu.N. Savchenko and Yu.A. Semenov
18. Korobkin AA. Blunt body entry into a weakly compressible liquid (in Russian). Zhurnal
Prikladnoi Mekhaniki i Tekhnicheskoi Fiziki. 1984;5:104–10.
19. Zhuravlev YuF. Perturbation theory methods in spatial jet flows (in Russian). Trudy TsAGI.
1973;1532:1–22.
20. Savchenko YuN, Semenenko VN. Wave generation on the boundaries of cavities formed in the
water entry of a disk and cones (in Russian). Problems in High-Speed Hydrodynamics.
Cheboksary: Chuvash University; 1993. p. 231–9.
21. Savchenko YuN, Semenenko VN, Putilin SI. Unsteady processes in supercavitating body
motion (in Russian). Prykladna Gidromekhanika. 1999;1(1):62–80.
22. Iafrati A. Hydrodynamics of asymmetric wedges impacting the free surface. European Con-
gress on Computational Methods in Applied Sciences and Engineering. ECCOMAS 2000.
Barselona; 11–14 Sept 2000.
23. Semenov YuA. Analytical solution to the self-similar problem of the water entry of an
asymmetric wedge (in Russian). Prikladnaya Gidromekhanika. 2003;5(4, 77):64–72.
24. Semenov YuA. Oblique water entry of a wedge (in Ukrainian). Dopovidi Natsionalnoyi
Akademiyi Nauk Ukrayiny. 2004;2:48–53.
25. Semenov YuA, Iafrati A. On the nonlinear water entry problem of asymmetric wedges. J Fluid
Mech. 2006;547:231–56.
26. Goman OG, Semenov YA. Oblique entry of a wedge into an ideal incompressible fluid. Fluid
Dynam. 2007;42(4):581–90.
27. Faltinsen OM, Semenov YA. Nonlinear problem of flat plate entry into an incompressible
liquid. J Fluid Mech. 2008;611:151–73.
28. Chekin BS. Wedge entry into an incompressible liquid (in Russian). Prikladnaya Matematika i
Mekhanika. 1989;53(3):396–404.
29. Korobkin AA. Inclined entry of a blunt profile into an ideal fluid. Fluid Dynam.
1988;23:443–7.
30. Judge C, Troesch AW, Perlin M. Initial water impact of a wedge at vertical and oblique angles.
J Eng Math. 2004;48:279–98.
31. Xu L, Troesch AW, Vorus WS. Asymmetric vessel impact and planing hydrodynamics. J Ship
Res. 1998;42:187–98.
32. de Divitiis N, de Socio LM. Impact of floats on water. J Fluid Mech. 2002;471:365–79.
33. Joukovski NE. Modification of Kirchhoff’s method for determination of fluid motion in two
directions at a fixed velocity given on the unknown streamline (in Russian). Math Coll. 1890;
XV: 121–278.
34. Semenov YuA. Complex potential of an unsteady free-boundary flow (in Russian). Vestnik
Khersonskogo Universiteta – Kherson. 2003;2:384–7.
35. Gurevich MI. Theory of Jets in ideal fluids (in Russian). Moscow: Nauka; 1979.
36. Wu GX. Two-dimensional liquid column and liquid droplet impact on a solid wedge. Q J Mech
Appl Math. 2007;60:497–511.
37. Wu GX, Sun H, He YS. Numerical simulation and experimental study of water entry of a
wedge in free fall motion. Roy Soc A, J Fluids Struct. 2004;19(3):277–89.
38. Semenov YA. Method for solving nonlinear problems on unsteady free-boundary flows.21th
International Congress of Theoretical and Applied Mechanics. Warsaw; 15–21 August 2004.
Study of the Supercavitating Body
Dynamics
Abstract
In this paper, the results of investigations of dynamics of supercavitating (SC)
bodies are presented, which were performed by authors in cooperation with Yu.N.
Savchenko. Computer simulation of the SC-body motion based on the G.V.
Logvinovich principle of independence of supercavity section expansion [1, 2] is
the main research method. A general problem of the three-dimensional (3D)
motion of the SC-body is formulated. Special cases of both the longitudinal and
the lateral motion of SC-bodies are considered. Problems of the motion stability
and optimization of SC-bodies moving on inertia on the arbitrary angle to the
horizon are investigated. It is shown that the SC-vehicle motion in the regime of
planing within a cavity is unstable on the depth. A comparative analysis of
stabilization and control of motion (maneuverability) of SC-vehicles by inclination
of the cavitator having two degrees of freedom and by the vectoring thrust is given.
In this paper, some materials from the works [3–9] were used, a part of the
results were represented on the International conferences [10–13].
1 Introduction
Figure 1 shows a scheme of the 3D SC-body motion. The body coordinate system
O1 x1 y1 z1 and the flow coordinate system O1 x0 y0 z0 are shown. An origin of both the
coordinate systems is placed at the vehicle mass center O1 . The axis O1 x1 of the
body coordinate system is directed along the longitudinal vehicle axis. The axis
O1 x0 of the flow coordinate system is directed along the velocity vector V ~ of the
vehicle mass center. The axes O1 y1 and O1 y0 are placed in the diametral plane of
the body.
Also, we will use the fixed coordinate system Oxyz and the semi-body coordinate
system O1 xg yg zg . The direction of the semi-body coordinate system axes coincides
with direction of axes of the fixed coordinate system Oxyz at each time instant.
150 V.N. Semenenko and Ye.I. Naumova
dox
Ix þ oy oz ðIz Iy Þ ¼ Mx1 ; (4)
dt
doy
Iy þ ox oz ðIx Iz Þ ¼ My1 ; (5)
dt
doz
Iz þ ox oy ðIy Ix Þ ¼ Mz1 ; (6)
dt
dy d’
ox ¼ þ sin c; (7)
dt dt
d’ dc
oy ¼ cos c cos y þ sin y; (8)
dt dt
dc d’
oz ¼ cos y cos c sin y; (9)
dt dt
Study of the Supercavitating Body Dynamics 151
where m is the body mass; V ~ ¼ fVx1 ; Vy1 ; Vz1 g is the velocity vector of the body
mass center; o ~ ¼ fox ; oy ; oz g is the angular velocity vector relatively to the body
mass center; y is the roll angle; ’ is the yaw angle; c is the pitch angle; Ix , Iy , Iz are
the moments of inertia relatively to the axes O1 x1 , O1 y1 , O1 z1 , respectively; Fx1 ,
Fy1 , Fz1 , Mx1 , My1 , Mz1 are the projections of the resultant force vector and the main
moment on the same axes.
The body mass center trajectory relatively to the fixed coordinate system Oxyz is
defined by equations:
dx
¼ V cosðc aÞ cosð’ bÞ; (10)
dt
dy
¼ V sinðc aÞ cosð’ bÞ; (11)
dt
dz
¼ V sinð’ bÞ; (12)
dt
~ ; a is the angle of attack; b is the sliding angle. The angles a and b
where V ¼ jVj
define a position of the body coordinate system relatively to the flow coordinate
system (see Fig. 1). In this case the following relations are valid [36]:
One accepts the following assumptions for formulation of the general problem of
the SC-body dynamics:
1. The SC-body is a slender body of revolution, in this case Iy ¼ Iz
2. A disk with diameter Dn is the cavitator shape
3. The mass m, the mass center position xc , and the vehicle moments of inertia Ix , Iz
do not vary during motion
Then the right parts of Eqs. 1–3 must include the projections of the gravity force mg,
the hydrodynamic force on the cavitator F ~n , the force of interaction between the
~
body and the cavity wall Fs , and also the propulsor thrust F ~pr and the control force
~
on the fins and rudders Fc , if they are present:
fins moving under action of the propulsor thrust in the regime of planing within the
cavity.
The set of the differential equations (1)–(12) is integrated numerically at the
initial conditions:
The equation of expansion of the axisymmetric cavity sections [17, 37] is used to
calculate the unsteady cavity shape. It is expression of the G.V. Logvinovich’s
principle of independence of the cavity section expansion [1]:
@ 2 Sc ðt; tÞ k 1 D pð t Þ
¼ ; sðtÞ Lc ðtÞ x sðtÞ ; (14)
@t2 r
where Sc is the cavity section area (see Fig. 3); s is the arc coordinate of the
cavitator center; k1 is the semi-empirical constant; t t is the time instant of the
cavitator passage through the section x; Lc is the cavity length; pc is the cavity
pressure; DpðtÞ ¼ p1 ðxÞ pc . According to the independence principle the cavity
axis shape is defined by the trajectory of the cavitator center ðxn ; yn ; zn Þ:
Study of the Supercavitating Body Dynamics 153
where ðx; y; zÞ are the current coordinates of the body mass center; xc is the distance
from the cavitator to the mass center.
The cavity axis may be additionally curved under action of the lateral force
arising on the inclined cavitator and under action of the gravity forces as well. For
the steady cavities these effects are approximately taken into account by the
approximate formulae [38]:
where Rn is the cavitator radius; cny0 and cnz0 are the coefficients of the forces Fny0
and Fnz0 ; x ¼ x=Lc . As it was established [38], the cavity axis curving is the main
perturbations of the cavity shape, and deformations of the circular shape of the
transversal cavity sections may be neglected if Fr is large. Thus, owing to the
independence principle at each time instant the coordinates of contours of the axial
cavity section in the fixed coordinate system are determined by relations
The cavitator is the most important element from the point of view of the SC-body
dynamics, since it serves not only to forming a cavity, but to control of the motion.
~n , the force of interaction between the body
In addition to the force on the cavitator F
~
and the cavity walls Fs is of importance. The gravity mg, the propulsor thrust F ~pr ,
~
and the force Fc created by fins and rudders (if they are present) are concerned to
other forces acting onto the SC-body.
Let the cavitator is oriented in arbitrary way in relation to the coordinate system
O1 x1 y1 z1 . One introduces the cavitator coordinate system On xn yn zn (see Fig. 4). The
origin of coordinates On is placed on the longitudinal body axis O1 x1 ; the axis On xn
is directed along the normal ~ n to the cavitator plane in direction to the liquid; the
axis On yn is placed in the plane O1 x1 y1 ; dy is the angle between the axis On x1 and
the projection On K of the axis On xn onto the plane On x1 z1 0 (where On z1 0 jjO1 z1 ); dz is
the angle between the axis On xn and the plane On x1 z1 0 .
In the case of the disk cavitator the force from *the flow side is always directed
oppositely to the normal ~ n to the cavitator plane: Fn ¼ Fn~ n. If the normal to the
cavitator is inclined to the flow on the angle m, then the absolute value of the force
acting to the cavitator is equal to [1]
Fn ¼ F
~n ¼ Fx0 cos m; (16)
ðnÞ
cos m ¼ c11 ¼ cos dy cos dz cos an cos bn
(17)
sin dz sin an cos bn sin dy cos dz sin bn ;
~n on
where an ¼ a þ oz xc =Vx1 , bn ¼ b þ oy xc =Vx1 . The projections of the vector F
the axes of the body coordinate system are equal to:
Fnx ¼ Fn cos dy cos dz ; Fny ¼ Fn sin dz ; Fnz ¼ Fn sin dy cos dz : (18)
The projections of the same vector on the axes of the flow coordinate system are
equal to [9]:
Fnx0 ¼ Fn c11 ; Fny0 ¼ Fn c12 ; Fnz0 ¼ Fn c13 ; (19)
where
Any position of the cavitator may be obtained by realizing the two consequential
turnings of the cavitator on the angle dz 0 around the axis On z1 0 and on the angle dx
around the axis O1 x1 . In this case a connection between the pairs of angles (dz 0 ,dx )
and (dy ,dz ) is given by the relations
sin dy cos dz
cos dz 0 ¼ cos dy cos dz ; sin dx ¼ : (20)
sin dz 0
The first method of setting the cavitator orientation is convenient when solving
the dynamic equations, the second method is the more convenient when practical
realization of the SC-vehicle control system.
For the axisymmetric SC-vehicle without fins, the turning the cavitator on the
angle dx about the axis O1 x1 is equivalent to rolling of the whole vehicle on
the angle y ¼ dx . Thus, for such a vehicle the d-control on course is equivalent to
the “bank-to-turn” control [25]. In this case recalculation of the angles can be
performed by formulae (20).
156 V.N. Semenenko and Ye.I. Naumova
The force Fsy of planing of an elongated body within a cavity may be calculated by
the Wagner’s method [1] with using the solution of the two-dimensional problem
on immersion of a circular arc in a curvilinear free surface [19, 39]. In the case of
motion of the SC-body in the vertical plane it can be calculated at each time instant
by the formula:
" #
þ hÞ
hð2 2h
Fsz ¼ rpR2s V0 V1 þ V2 ; (21)
2
ð1 þ hÞ 1 þ h
rV 2 lw V
Fsx ¼ Sw cf ðRew Þ; Re ¼ ; (22)
2 n
where cf is the friction drag coefficient [40]; lw and Sw are the length and the area of
the wetted body surface, respectively.
Formulae (21), (22) may be applied in the case of arbitrary 3D motion of the
body within the cavity. In this case one should consider that the force F ~s is applied
at each time instant in that point of the contact arc where the immersion h is
maximal, and it is directed so that the force vector passes through the body section
center.
If the mass center coincides with the body transversal section center, then the
gravity mg does not create any moment. If the mass center is placed below the
section center on the distance hM (so called the metacentric height), then when
rolling the vehicle on the angle y the restoring moment relatively to the axis O1 x1
arises:
Using the traditional operating controls piercing the cavity walls (fins and
rudders) for SC-vehicles is ineffective by a number of reasons [10]. In this paper,
~c ¼ 0. However, it is
it is considered that the fins and rudders are absent, i.e. F
Study of the Supercavitating Body Dynamics 157
obvious that the effect of control of the SC-vehicle motion by the tail rudders will
be qualitatively similar to the effect of control by deflecting the propulsor thrust
~pr considered below.
vector F
Since the range of the SC-body motion on inertia is proportional to the model mass
m, then the shape of the “small” SC-model at the given density rb must be inscribed
maximally tightly into the frontal cavity part. Besides, the model shape must ensure
the motion stability.
Examples of calculations for the uniform model are given below. Its shape and
dimensions are shown in Fig. 5 and it has the following parameters: m ¼ 107 g,
xc ¼ xc =L ¼ 0:67, Iz ¼ 13:27 105 kg m2 .
Figure 6 shows the shape and dimensions of the SC-vehicle model designed in such
way that on the marching part of the flying path the model moves in the regime of
planing within the cavity. The model mass is m ¼ 600 g, the moments of inertia are
Ix ¼ 8 kgm2 , Iy ¼ Iz ¼ 900 kg·m2 . It is supposed that the model moves
horizontally on the marching part of the flying path on the depth Hm ¼ 5m with the
constant velocity Vm ¼ 120 m=s. In this case the cavitation number and the Froude
number are equal to s ¼ 0.0201 and Fr ¼ 144.8, and the cavity length Lcm ¼
6.52 m. For the accepted parameters the calculation gives the value of the required
propulsor thrust equal to the total drag of the SC-vehicle Fm ¼ 23.18 КN.
dVx1 1
V cosðc aÞ ¼ oz Vy1 þ Fx1 ; (24)
dx m
dVy1 1
V cosðc aÞ ¼ oz Vx1 þ Fy1 ; (25)
dx m
Study of the Supercavitating Body Dynamics 159
doz 1
V cosðc aÞ ¼ Mz1 ; (26)
dx Ic
dc
V cosðc aÞ ¼ oz ; (27)
dx
dy
¼ tgðc aÞ; (28)
dx
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
where V ¼ Vx1 2 þ V 2 ; a ¼ arctg Vy1 . For any x the elapsed time can be calcu-
y1 Vx1
lated by the formula:
ðx
ds
tðxÞ ¼ :
V cosðc aÞ
0
In the case of the longitudinal motion, the right parts of Eqs. 24–26 and relations
(18), (19) take the following form:
Let the body starts on the depth H0 with velocity V0 on an arbitrary angle to the
horizon g (g<0 at downstream motion). In this case it is convenient to define the
fixed coordinate system Oxy by directing the axis Ox along the starting velocity
160 V.N. Semenenko and Ye.I. Naumova
vector. Then the hydrostatic water pressure in formula (14) will be changed
according to the formula:
where yðtÞ is the deflection of the cavity section center formed at the time instant t
from the axis Ox.
The computer simulation has confirmed that the self-stabilization of “small” SC-
models by ricocheting the model tail from the cavity walls is the basic mechanism
of the motion stability [3–5]. In this case the degree of the SC-model motion
stability considerably depends on the model shape and its mass center position.
Figure 7 shows graphs of dependencies of the motion range S on the depth
H calculated for three cavitator diameters when g ¼ 0; V0 ¼ 1000 m=s and
oz0 ¼ 2:0 rad:=s. As is obvious, for the low H independently on the value Dn the
SC-model passes a distance about 10 m and then loses the stability. With increasing
H the motion becomes stable, and the range spasmodically increases.
The zones of stable and unstable motions on the plane of parameters ðH; oz0 Þ for
various Dn are shown in Fig. 8. For each of three cavitator diameters Dn the curve
divides the plane on two zones. The stable motion zone is placed to the right from
the curve, The unstable motion zone is placed to the left.
A comparison of the graphs in Figs. 7 and 8 shows that the SC-model motion is
more stable for smaller Dn and higher H, i.e. with the tighter cavity. The analogous
conclusion was made earlier in paper [41].
During inertial motion of SC-bodies their velocity rapidly decreases, and the cavity
dimensions decrease too. As the experience shows, the motion continues until the
body is fully covered with the cavity and practically instantly ceases when the body
is wetted by water [3]. Many parameters influences on the motion range: the body
shape, the body length, the body mass, the cavitator diameter, the starting velocity,
the depth and the angle of launching etc. Naturally, the optimization problem arises:
how to choose the body shape and other parameters that the range of the SC-body
motion on inertia would be maximal. Putting aside the motion stability problem
here, one considers two typical optimization problems for SC-bodies moving
rectilinearly.
Task 1. To choose the cavitator diameter Dn for the body with the given shape
and mass so that the range S would be maximal. Figure 9 shows the calculation
results when H0 ¼ 20 m; V0 ¼ 500 m/s, for a number of values of the starting angle g.
The calculation shows that the optimal values of Dn (marked by circles) depend on
the angle g and on the motion depth and do not depend on the model mass.
Task 2. To choose the optimal ratio of the mass m and the starting velocity V0
for given both the cavitator diameter Dn and the model initial kinetic energy
E0 ¼ mV02 =2 so that the range S would be maximal. Figure 10 shows the calculation
results when Dn ¼ 2.2 mm; E0 ¼ 13.392 kJ; g¼ 0. As one can see, the optimal ratio
of the model mass and the starting velocity when E0 ¼ const also depends on the
motion depth.
In the case of low cavitaton numbers and high Froude numbers we developed the
approximate analytical method of optimization of parameters of the SC-bodies
moving on inertia [6, 7]. The obtained simple relations allow the ultimately possible
162 V.N. Semenenko and Ye.I. Naumova
In this case we define the fixed coordinate system Oxy directing the axis Ox
horizontally. One considers some problems arising at practical application of self-
propelled SC-vehicles and successfully solvable in the frames of model of the
longitudinal motion.
One of the main problems consists in that the vehicle must overcome the very high
friction drag during its acceleration up to formation of a full-developed supercavity.
On the acceleration part the vehicle moves sequentially in the non-separated
regime, in the partial cavition regime and, finally, in the regime of planning within
the cavity. The different mechanisms of forming the motion drag act on each of
these three parts.
Figure 11 shows an example of calculation of the cavity length increasing during
the vehicle acceleration under action of the fixed propulsor thrust Fm . The graphs of
increasing the vehicle velocity look analogously [13]. Curve 1 corresponds to the
thrust value Fpr ¼ Fm . As is obvious, the marching values of the cavity length and
the velocity (in this case Lcm ¼ 8.71 m, Vm ¼ 150 m=s) are not achieved, and the
attained balanced values Lc ¼ 4:93 m, V ¼ 112:5 m=s correspond to the partial
cavitation regime. When the thrust increases (curve 2) the vehicle reaches the
balanced values Lc and V exceeding considerably the marching values Lcm and Vm .
A dependence of the final balanced cavity length after acceleration on the
relative propulsor thrust Fpr =Fm is shown in Fig. 12. The graph of dependence of
the final velocity of the vehicle on Fpr =Fm looks analogously. As is obvious,
transition from the partial cavitation regime to the supercavitation with increasing
Fig. 11 Changing the cavity length at acceleration when the propulsor thrust is fixed
1 Fpr =Fm ¼ 1; 2 Fpr =Fm ¼ 1:324
164 V.N. Semenenko and Ye.I. Naumova
the thrust occurs stepwise. In this case the vehicle attains uneconomical flow
regimes when Lc > Lcm , V > Vm .
Thus, it is impossible to come to the marching motion regime of the SC-vehicle
using only the mid-flight engine with the fixed thrust Fm beginning from the low
starting velocity. The two methods of solving this problem exist:
1. Using the additional starting propulsor (accelerator) generating very high thrust
Fst >>Fm during short time
2. Using the intensive gas-supply into a partial cavity with the purpose of its
development acceleration. In this case the problems connected with pulsation
of the ventilated cavities can arise [43]
During the SC-vehicle marching motion in the planing regime the vehicle weight
must be compensated by the hydrodynamic forces Fny and Fsy created respectively
by inclining the cavitator on the angle dz <0 and as a result of planning of the
vehicle tail along the lower cavity wall, and also, possibly, by deflecting the
~pr on the angle z >0 (see Fig. 13). In the steady horizontal
propulsor thrust vector F
motion a sum of the vertical projections of these three forces and the total moment
of these forces must be equal to zero.
To determine the balanced values of the angles dz , z , and c one considers
projections of the forces on the cavitator Fny and Fsy in the body coordinate system
(29), and Fnx0 and Fny0 in the flow coordinate system (30). In the steady horizontal
motion of the SC-vehicle the following relations must be fulfilled:
Study of the Supercavitating Body Dynamics 165
where xs is distance from the cavitator to the point of application of the force Fsy .
The connection between the projections Fny and Fny0 follows from (29) and (30):
sin dz
Fny ¼ Fny0 : (34)
sinðdz þ cÞ
From Eqs. 32 and 33 with taking into consideration (34) one obtains the values
of the force projections which are required for balancing the SC-vehicle:
sin dz xc
Fsy ¼ Fny0 Fpr sin z :
sinðdz þ cÞ xs xc
If the angle z is fixed, the balanced values of the angles dz and c are
determined numerically with the help of an iteration process. For the SC-vehicle
model (Fig. 6) when xc ¼ 0:6 and z ¼ 0 the calculation gives dz ¼ 5:781 and
c ¼ 0:334 . If the angle c is fixed, the balanced values of the angles dz and z are
determined analogously.
the SC-vehicle is unstable [8]. The instability appears in spontaneous arising and
increasing the vehicle oscillations by the pitch angle c and in increasing the
deviation yc of the vehicle mass center from the horizontal path. The vehicle
behavior after the stability loss strongly depends on its mass center position xc .
The example of development of the SC-vehicle oscillation on the distance 500 m
when xc ¼ 0:6 is shown in Fig. 14. It is designated x ¼ x=L in the figure;
¼ oz L=V0 , the magnitudes of c are in radians.
o
Amplitude of the angular oscillation of the vehicle grows until the opposite
cavity wall begins to confine them. After this the vehicle continues to oscillate
within the cavity with the approximately fixed frequency f , which may be named as
the natural frequency of the SC-vehicle. In this case the deviation y and the mean
value of the angle c grow monotonically. A dependence of the reduced natural
frequency k ¼ 2p fL=V0 on the dimensionless mass center position xc is shown in
Fig. 15. The deviations of the vehicle mass center y in meters at the end of distance
300 m are plotted there as well.
As is obvious, the natural frequency essentially depends on the mass center
position. In this case the value xc ¼ 0:5 approximately divides the function
Fig. 15 Influence of xc on
the natural frequencies of the
uncontrolled SC-vehicle
Study of the Supercavitating Body Dynamics 167
variation domain onto two zones. When xc < 0:5, the vehicle performs complex
oscillation with the high frequency. When xc > 0:5, the angular oscillations of the
vehicle are close to the simple periodic ones, their frequency decreases, and the
mass center deviation increases. The calculations show that it is possible for any
values of xc to determine such distance x that the deviation of the SC-vehicle mass
center exceeds any value given beforehand. When z ¼ 0, the uncontrolled SC-
vehicle usually exits on the water surface. Thus, introduction of the artificial
stabilization is necessary to maintain the horizontal motion of the SC-vehicle.
The lateral motion problem arises in connection with investigation of the SC-
vehicle maneuverability on course. The lateral vehicle motion consists in revolution
around the axes O1 x1 and O1 y1 and in transition in direction to the axis O1 z1 . If the
undisturbed motion is longitudinal, and the deflections are small, then the lateral
motion of the SC-body may be approximately considered independently on the
Study of the Supercavitating Body Dynamics 169
longitudinal motion [36]. When formulating the problem on the SC-vehicle lateral
motion one accepts that the longitudinal motion is steady, i.e. y ¼ 0, a ¼ c ¼ c ,
V ¼ V0 .
Then six Eqs. 3–5, 7, 8, and 12 remain from the general set of dynamic equations
(1)–(12). One should assign oz ¼ 0, dc=dt ¼ 0 there. Passing to differentiating
with respect to the absolute coordinate x and taking into account the relations (13),
we obtain the calculation set of six differential equations of the SC-vehicle lateral
motion:
dVz1 1
cosð’ bÞ ¼ ox sin c cos b þ oy cos c cos b þ Fz1 ; (37)
dx mV0
dox 1
V0 cosð’ bÞ ¼ Mx1 ; (38)
dx Ix
doy 1
V0 cosð’ bÞ ¼ My1 ; (39)
dx Iy
dy sin c
V0 cosð’ bÞ ¼ ox oy ; (40)
dx cos c cos y
d’ 1
V0 cosð’ bÞ ¼ oy ; (41)
dx cos c cos y
dz
¼ tgð’ bÞ; (42)
dx
where b ¼ arcsin VVz10 . For any x the elapsed time can be calculated by the formula:
ðx
ds
tðxÞ ¼ :
V0 cosð’ bÞ
0
170 V.N. Semenenko and Ye.I. Naumova
In the case of the lateral motion, the right parts of Eqs. 37–39 and relations (18),
(19) take the following form [9]:
ðnÞ
Fnz ¼ Fx0 c11 sin dy cos dz ;
ðnÞ
Fnz0 ¼ Fx0 c11 c13 ;
ðnÞ
c11 ¼ cos dy cos dz cos c cos bn
sin dz sin c cos bn sin dy cos dz sin bn ;
At intersection of the body contour and the cavity contour in the body section by
the horizontal plane O1 xg zg , the lateral planing force is calculated by the formula
" #
þ hÞ
hð2 2h
Fsz ¼ rpR2s V0 V1 þ V2 cos y:
2
ð1 þ hÞ 1 þ h
10 Maneuverability of SC-Vehicles
Fig. 18 Deviation z and mean trajectory angle wm at the SC-vehicle course maneuvering
a – d-control; b – -control
Also, we investigated the influence of the starting roll angle y0 6¼ 0 onto the SC-
vehicle trajectory [9]. In this case, if the vehicle metacentric height hM >0, then the
vehicle performs undamped oscillation on the roll angle y with the amplitude y0 .
The hM increasing leads to some increasing the frequency of this oscillation.
The vehicle oscillation on the roll causes the cavity oscillation in the horizontal
plane, which amplitude increases with decreasing hM . If in this case the vehicle
does not contact with the lateral cavity walls, then variations of z and wm are
insignificant. However, for sufficiently low values of hM the amplitude of the cavity
oscillation increases so that the lateral cavity walls begin to interact with the
vehicle. In this case the SC-vehicle oscillation on the yaw angle ’ arises, and the
mean angle of the trajectory wm abruptly increases. In practice with the purpose of
the course stabilization of the SC-vehicle motion one should increase the value of
hM , if it is possible, and/or damp the vehicle roll oscillation with the specially
designed fins.
Study of the Supercavitating Body Dynamics 173
Fig. 19 Deviation y and mean trajectory angle #m at the SC-vehicle depth maneuvering
1 – d-control; 2 – -control
Conclusions
The particular cases of the longitudinal motion and the lateral motion have been
considered basing on the general mathematical model of 3D motion of the SC-
body. The problems of the motion stability and optimization for the SC-bodies
moving on inertia on an arbitrary angle to the horizon were investigated. The
comparative analysis of application of two methods of control of the SC-vehicle
motion by inclining and/or turning the cavitator having two degrees of free-
dom (d-control), and by deflection of the propulsor thrust vector (-control)
has been performed to stabilize the SC-vehicle motion and to maneuver on
course and depth. The performed investigations allow the following main
conclusions to be made.
1. The motion of the “small” SC-bodies on inertia can be unstable or stable “in the
whole” due to periodic interaction of the body with the cavity walls. We have
developed the numerical and analytical method of optimization of the SC-body
parameters that have the purpose of increasing the motion stability degree and
attaining the maximal motion range.
2. The longitudinal motion of the uncontrolled SC-vehicle in the planing within the
cavity regime is unstable “in the small”. In this case, the automatic regulation of
the cavitator inclination angle and/or deflection of the propulsor thrust vector
with the linear feedback law permits the problem of practical depth stabilization
of the SC-vehicle motion at the specified distance to be solved.
3. The control of the SC-vehicle motion by inclining and/or turning cavitator
(d-control) and the control by deflecting the propulsor thrust vector (-control)
for the motion stabilization and maneuvering are equally effective.
4. Applying the -control for the course maneuvering the SC-vehicle is ineffective
in comparison with the d-control. Obviously, the same may be said about the
control by the tail rudders piercing the cavity walls. One notes that
G.V. Logvinovich has pointed onto inefficiency of the “tail control” for the
SC-vehicles in the beginning of the 1970s.
References
1. Logvinovich GV. Hydrodynamics of flows with free boundaries. Kiev: Naukova dumka; 1969.
208p. (In Russian). English translation: Halsted Press, 1973.
2. Logvinovich GV. Problems of the theory of slender axisymmetric cavities. Trudy TsAGI.
1976;1797:3–17 (In Russian).
3. Savchenko YuN, Vlasenko YuD, Semenenko VN. Experimental investigations of high-speed
cavitation flows. Hidromehanika. 1998;72:103–11. (In Russian). English translation: Int
J Fluid Mech Res. 1999;26(3):365–74.
4. Savchenko YuN, Semenenko VN, Putilin SI. Unsteady processes in supercavitation motion of
bodies. Prykladna hidromehanika. 1999;1(1):62–80. (In Russian). English translation: Int
J Fluid Mech Res. 2000;27(1):109–37.
5. Semenenko VN. Computer simulation of dynamics of supercavitatating bodies. Prykladna
Hidromehanika. 2000;2(1):64–9 (In Russian).
6. Nesteruk IG, Semenenko VM. Optimization problems for supercavitation inertial motion of
axisymmetric bodies. Prykladna Hidromehanika. 2006;8(1):51–9 (In Ukrainian).
Study of the Supercavitating Body Dynamics 175
7. Nesteruk IG, Savchenko YuM, Sememenko VM. Optimization of the range for the
supercavitation motion on inertia. Dopovidi NAN Ukrainy. 2006;8:57–66 (In Ukrainian).
8. Semenenko VN. Modelling the longitudinal motion of the underwater supercavitating
vehicles. Prykladna Hidromehanika. 2010;12(4):81–8 (In Russian).
9. Savchenko YuN, Semenenko VN. On the course maneuvering the underwater supercavitating
vehicle. Prykladna Hidromehanika. 2011;13(1):43–50 (In Russian).
10. Savchenko YuN, Semenenko VN, Putilin SI, et al. Designing the high-speed supercavitating
vehicles. Proceedings of the 8th International Conference on Fast Sea Transportation
(FAST’2005). St. Petersburg; 2005.
11. Savchenko YuN, Semenenko VN, Putilin SI, et al. Some problems of the supercavitating
motion management. Sixth International Symposium on Cavitation CAV2006. Wageningen;
2006.
12. Nesteruk IG, Savchenko YuN, Semenenko VN. Achievement of maximal range of
supercavitating body inertial motion. Proceedings of the International conference on subsea
technologies (SubSeaTECH2007); June 25–28, 2007. St. Petersburg; 2007.
13. Semenenko VN. Some problems of supercavitating vehicle designing. Proceedings of the
International Conference on superfast marine vehicles moving above, under and in water
surface (SuperFAST’2008); 2–4 July 2008. St. Petersburg; 2008.
14. Savchenko YuN. Experimental investigation of supercavitating motion of bodies. VKI/RTO
Special Course on Supercavitation. Von Karman Institute for Fluid Dynamics. Brussels; 2001.
(Belgium).
15. Savchenko YuN. Control of supercavitation flow and stability of supercavitating motion of
bodies. VKI/RTO Special Course on Supercavitation. Von Karman Institute for Fluid Dynam-
ics. Brussels; 2001. (Belgium).
16. Savchenko YuN. Supercavitating object propulsion. VKI/RTO Special Course on
Supercavitation. Von Karman Institute for Fluid Dynamics. Brussels; 2001. (Belgium).
17. Semenenko VN. Artificial supercavitation. Physics and calculation. VKI/RTO Special Course
on Supercavitation. Von Karman Institute for Fluid Dynamics, Brussels; 2001. (Belgium).
18. Semenenko VN. Dynamic processes of supercavitation and computer simulation. VKI/RTO
Special Course on Supercavitation. Von Karman Institute for Fluid Dynamics. Brussels; 2001.
(Belgium).
19. Logvinovich GV. Some problems of planing. Trudy TsAGI. 1980;2052:3–12 (In Russian).
20. Savchenko YuN. Investigations of supercavitation flows. Prykladna Hidromehanika.
2007;9(2–3):150–8 (In Russian).
21. Kulkarni SS, Pratar R. Studies on the dynamics of a supercavitating projectiles. Appl Math
Model. 2000;24(2):113–29.
22. Abe A, Katayama M, Saito T, Takayama K. Numerical simulation on supercavitation and
jawing of a supersonic projectile traveling in water. Symposium on Interdisciplinary Shock
Wave Research. Sendai; March 22–24, 2004.
23. Lindau JW, Kunz RF, Mulherin JM, Dreyer JJ, Stinebring DR. Fully coupled, 6-DOF to
URANS, modelling of cavitating flows around a supercavitating vehicle. Fifth International
Symposium on Cavitation CAV2006. Osaka; 2003
24. Kirschner I, Rosenthal BJ, Uhlman JS. Simplified dynamical systems analysis of
supercavitating high-speed bodies. Fifth International Symposium on Cavitation CAV2006.
Osaka; 2003.
25. Kirschner IN, Kring DC, Stokes AW, Fine NE, Uhlman JS. Control strategies for
supercavitating vehicles. J Vib Control. 2002;8:219–42.
26. Dzielski J, Kurdila A. A benchmark control problem for supercavitating vehicles and an initial
investigation of solution. J Vib Control. 2003;19(7):791–804.
27. Ruzzene M, Kamada R, Botasso CL, Scorceletti F. Trajectory optimization strategies for
supercavitating undervater vehicles. J Vib Control. 2008;14(5):611–44.
28. Botasso CL, Scorceletti F. Trajectory optimization for DDE models of supercavitating
undervater vehicles. Online preprint; 2008. http://www.aero.polimi.it/~bottasso/.
176 V.N. Semenenko and Ye.I. Naumova
29. Balas GJ, Bokor J, Vanek B, Arndt REA. Control of high-speed undervater vehicles. In:
B.A. Francis et al. (eds.) Control of uncertain systems. Berlin/Heidelberg: Springer-Verlag;
2006. p. 25–44.
30. Vanek B, Bokor J, Balas GJ, Arndt REA. Longitudinal motion control of a high-speed
supercavitation vehicle. J Vib Control. 2007;13(2):159–84.
31. Lin G, Balachndran B, Abed EH. Nonlinear dynamics and bifurcations of a supercavitating
vehicle. IEEE J Ocean Eng. 2007;32(4):753–61.
32. Lin G, Balachandran B, Abed E. Dynamics and control of supercavitating vehicles. J Dynam
Syst Meas Control. 2008;130(2):1–11.
33. Li DJ, Zhang YW, Luo K, Dang JJ. Motion control of underwater supercavitating projectiles in
vertical plane. Mod Appl Sci. 2009;3(2):60–5.
34. Li DJ, Luo K, Zhang YW, Dang JJ. Studies on fixed-depth control of supercavitating vehicles.
Acta Automatica Sinica. 2010;36(3):421–6.
35. Polyahov NN, Zegzhda SA, Yushkov MP. Theoretical mechanics. Moscow: Vysshaya shkola;
2000 (In Russian).
36. Lukomskiy YuA, Chugunov VS. Systems of controls of marine vehicles. Leningrad:
Sudostroenie; 1988 (In Russian).
37. Logvinovich GV, Serebryakov VV. On the methods of calculating a shape of the slender
axisymmetric cavities. Hidromehanika. 1975;32:47–54 (In Russian).
38. Buyvol VN. Slender cavities in flows with perturbations. Kiev: Naukova Dumka; 1980
(In Russian).
39. Vasin AD, Paryshev EV. Immersion of a cylinder in liquid through a cylindrical free surface.
Izvestiya AN SSSR, Mehanika zhidkosti i gasa. 2001;2:3–12 (In Russian).
40. Schlichting H. Boundary layer theory. New York: McGraw-Hill; 1961.
41. Putilin SI. Some features of a supercavitating model dynamics. Prykladna hidromehanika.
2000;2(3):65–74 (In Russian). English translation: Int J Fluid Mech Res. 2001;28(5):631–43.
42. Nesteruk IG, Semenenko VM. Problems of optimization of a range of the supercavitation
motion on inertia with a fixed final depth. Prykladna Hidromehanika. 2006;8(4):33–42
(In Ukrainian).
43. Semenenko VN. Computer simulation of pulsations of ventialed supecavities. Hidromehanika.
1997;71:110–8. (In Russian). English translation: Int J Fluid Mech Res. 1996;23(3 & 4):
302–12.
44. Savchenko YuN. Perspectives of the supercavitation flow applications. Proceedings of the
International Conference on Superfast Marine Vehicles Moving Above, Under and in Water
Surface (SuperFAST’2008); 2–4 July 2008. St. Petersburg; 2008.
Water Entry of Thin Hydrofoils
A. G. Terentiev
Abstract
First order methods conformably to entry of thin foils in incompressible fluids
are considered. Two of them have been based on complex variables and the
speed function as well as the partial time-derivative are obtained, another two
methods are based on the time depended Green function and Fourier transfor-
mation, respectively. For examples, an entry of inclined wedges with strait sides
into weightless and also into gravity medium are presented.
1 Introduction
Consider the motion of a partially submerged slender body (Fig. 1a). Let the
abscissa axis be parallel to principal velocity U of the body so that the y-component
of speed, v, is small. The linear value problem is formulated on the half plane with
Fig. 1 Inclined entry in fluid: (a) – flow in the z-plane; (b) – the parametric z-plane
an inclined slit (the definitional domain in Fig. 1a is marked by the dashed line). A
linear kinematical condition on the wetted surface of the body is imposed as:
@w
Im ¼ gðx; tÞ; x 2 ðl; 0Þ; y ¼ 0; (1)
@z
where gðz; tÞ is described by the given motion of the body. The dynamic condition
on the free boundary (inclined dashed line) is described by
@w ipg
Re e ¼ 0 on Im zepgi ¼ 0: (2)
@z
Besides, a solution of the mixed value problem should possess at the leading
edge a singularity of the type
Instead of function wz ðz; tÞ, it is more convenient to solve another value problem
of the partial derivative @w=@z ¼ wz ðz; tÞ, which satisfies the following conditions:
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Multiplying by function, gðzÞ ¼ ð1 zÞ ða þ zÞ, the mixed value problem is
reduced to Schwarz problem with boundary condition,
(
f ðxðx; tÞxx ðxÞgðxÞ; x 2 ½a; 1; ¼ 0;
Im½wz gðzÞ ¼ (8)
0 x 2 ð1; aÞ [ ð1; 1Þ; ¼ 0 :
ð1
lag1 a þ t 2 gtdt
12g
wz ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi : (9)
p ð1 zÞða þ zÞ 1t tz
a
Complex velocity, wz ¼ wz ðz; tÞ=zz ðz; tÞ, is a function of variable z and time t
together with Eq. 4, they represent the value problem solution presented in
Eqs. 1–4. Since variable z is a function of z and t, it is quite complicated to obtain
complex potential wðz; tÞ and its partial time derivative wt ðz; tÞ ¼ @wðz; tÞ=@t. But
on the surface of the body, potential ’ðx; tÞ can be determined integrating equation
(9) over interval (a, x), and then pressure can be expressed in the form:
@’ðx; tÞ @xðx; tÞ @’ðx; tÞ
p ¼ r þ ; (10)
@t @t @x
where
ðx
@w
’ðx; tÞ ¼ Re dx
@z z¼x
a
ð a þ tð12gÞ=g
l
¼ f ðtÞIðt; xÞ t dt; (12)
pa1g 1t
Ðx jGðt;xÞj
where Iðt; xÞ ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
dx ffi
¼ plnffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ;
a ðaþxÞð1xÞðtxÞ ðaþtÞð1tÞ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ða þ xÞð1 tÞ ða þ tÞð1 xÞ
Gðt; xÞ ¼ lnpffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi:
ða þ xÞð1 tÞ þ ða þ tÞð1 xÞ
180 A.G. Terentiev
Integrating the pressure along the wetted part of the foil, one can calculated force
and torque. The transverse force is equal to
ð0 ð1
@xðx; tÞ
Y¼ ðp1 p2 Þdx ¼ ðp p0 Þ dx: (13)
@x
1 a
@xðx; tÞ @xðx; tÞ l_
Since ¼ xðx; tÞ, the latter integral may be rewritten in the form
@t @x l
ð1 ð1
r _ @xðx; tÞ
Y ¼ 1g ½2lgðtÞ _
þ lgðtÞ
pa @x
a a (14)
ln jGðt; xÞj
t dxdt;
ða þ tÞg ð1 tÞ1g
ð1
rl _lgðtÞ þ l gðtÞ LðtÞdt;
Y¼ (15)
2 2
a
2 3
a þ t12g
4t 6 1t sin pg 7
where LðtÞ ¼ 2ð1gÞ 6 7.
a cos pg 4 aþt ð12gÞ=2 5
1t
Function L(t) has uncertainty as the inclination angle g ! 1/2; its limit is as
4t 1 þ t 1
LðtÞ ¼ ln ; g¼ : (16)
p 1t 2
Longitudinal force consists of concentrated force at the leading edge Oðx ¼ lÞ
and of longitudinal projection of pressure on the real axis [6], X ¼ Xc þ Xp , where
þ
ir w2z
Xc ¼ dz
4 zz
z¼0
21 32
ð a þ tð12gÞ=2
rl 4 gðtÞ
¼ dt5 ; (17)
2pa2ð1gÞ 1t
a
Water Entry of Thin Hydrofoils 181
ð1 ð1
rl
Xp ¼ qðt; xÞRðt; xÞdt dx;
2
a a
_ (18)
_ þ lgðtÞbðxÞ;
qðt; xÞ ¼ ½lgðtÞ
2tx ln jGðt; xÞj
Rðt; xÞ ¼ :
pa2ð1gÞ ða þ tÞ ð1 tÞ1g ða þ xÞg ð1 xÞ1g
g
ð1
@xðx; tÞ
M¼ ðp p0 Þxðx; tÞ dx;
@x
a
ð1
rl2 _ l
M¼ ½lgðtÞ _
þ gðtÞKðtÞdt;
2 3
a
3t h
KðtÞ ¼ 3ð1gÞ ða þ tÞ23g ð1 tÞ3g1 sin 2pg
a cos 2pg
3ð1 2gÞ a þ tð12gÞ=2
tþ : (19)
2g 1t
Consider forward movement of the wedge with speed U ¼ l._ Let angles a1 and a2
stand for the upper and lower sides of inclination angle to the x-axis, respectively,
Then we have following expressions for the transverse force, longitudinal force,
leading edge force and the moment respectively:
182 A.G. Terentiev
rl lU_
Y¼ U ða1 I1 þ a2 I2 Þ;
2
2 2
2
rl a1 I1 þ a22 I2
X ¼ ðU2 lUÞ _ þ ða1 a2 Þ2 I3 ;
2 2
rU l
2
Xc ¼ ða1 I4 þ a2 I5 Þ2 ;
2
rl2 Ul_
M¼ U2 ða1 I6 þ a2 I7 Þ; (22)
2 3
Here
ð1 ð0 ð1
I1 ¼ LðtÞdt; I3 ¼ Rðt; xÞdtdx;
0 a 0
ð1 ð1
1 a þ tð12gÞ=2
I4 ¼ pffiffiffi 1g dt; I6 ¼ KðtÞdt:
pa 1t
0 0
All integrals for coefficients Ik can be calculated numerically. They can also be
expressed by hypergeometric functions Fða; b; g; x zÞ [8]:
8 ð12gÞ=2 9
>
> 2g >
< ½ð1 gÞF1 F2 þ >
=
4 1 þ 2g
I1 ¼ ;
> þ sin pg ½F ð1 gÞF >
ð1 gÞ2ð1gÞ cos pg >
: >
;
4 3
2g
pffiffiffi
2 g
I4 ¼ pffiffiffi F4 ;
pð1 þ 2gÞð1 gÞ1g
Water Entry of Thin Hydrofoils 183
3 F6 ð1 gÞF5
I6 ¼ 3ð1gÞ
sin 2pg
ð1 gÞ cos 2pg 3g
ð1 4gÞð1 gÞ 2F7
þ gð14gÞ=2 F1 þ F2 ;
1 þ 2g 1 þ 2g
where
The hyper geometric function may be calculated by the power series expansion
X
1 Y
k1
ða þ mÞðb þ mÞ
Fða; b; g; zÞ ¼ 1 þ zk : (24)
k¼1 m¼0
ðg þ mÞð1 þ mÞ
For example, consider an entry of a plate and a wedge with flat sides. Let the
plate’s slopes are a1 ¼ a2 ¼ a, and wedge’s a1 ¼ a2 ¼ a. The following
coefficients are introduced:
The leading edge force and torque are negative in these examples, thus all the
coefficients are positive now. Parameter s ¼ 1 CM =CY is the distance of
the pressure center from leading edge related to length l. Dependences of the
coefficients are plotted in Figs. 2 and 3.
Note: the scale for dependences of sðgÞ is shown on the right side on both figures.
One can find that at g ! 0 the flat plate and the wedge give the same results:
CY ! p=2, CX ! p=4, CM ! 3p=8, C0 ! p=4 and s ! 1=4. That is because only
one side of the wedge is contiguous to liquid. It should be noted that the lift
coefficient is a half of that for a planning plate of a constant length. In another
limited case for perpendicular entry, g ! 1=2, we have for the flat plate CY ! 8=p,
CM ! 2, C0 ! 4=p, s ! 1 p=4, and for the wedge CX ! 8 ln 2=p; all other
coefficients CY ; CM and C0 vanish, so distance s is not meaningful but not zero.
It should be noted that the nonlinear problem of perpendicular entry of a wedge with
flat sides was considered by using Wagner function in [9].
4 Partial Time-Derivative
ðx
cðx; tÞ ¼ gðx; tÞdx þ TðtÞ: (27)
sðtÞ
Hence,
ðx
_
ct ¼ sgðx; tÞ gðx; _
_ tÞdx þ TðtÞ; (28)
s
where T_ is to be found.
The dynamic condition on the free boundary is
’t ¼ 0: (29)
186 A.G. Terentiev
_ lÞ þ TðtÞ:
c ¼ laðx (30)
Hence, the boundary conditions for partial time derivative of complex potential
on the x -axis are as follows:
The same function on the x-axis determines partial time derivative of the stream
function outside (0, 1), so that at any moment the stream function itself can be
obtained by integration with respect to time over interval (0, t). Equating with
function for T(t), one obtains an integral equation of the form
Fig. 4 Entry of plate with ventilated cavity: (a) – the z-plane; (b) – the parametric z-plane
Water Entry of Thin Hydrofoils 187
ðt sffiffiffiffiffiffiffiffiffiffiffi
z1
Re ðT_ I_ aÞ dt ¼ l_ ta:
2 2
(33)
z
0
For constant speed V ¼ l_ ¼ const, the solution of Eq. 33 and all the hydrody-
namic characteristics can be expressed in analytic form. Passing to new variable
s ¼ t=t, one can write both last equations as
ð1 sffiffiffiffiffiffiffiffiffiffiffi
x1
Re T_ U2 a ds ¼ U 2 a; (35)
x
0
z w1
w
s ¼ ð1 zÞ 1 : (36)
a
Since the function zðt; tÞ depends only on variable s, and equality (35) should
be satisfied for any value of time, unknown function T_ U 2 a should be a constant.
Hence,
:
T U 2 a ¼ U 2 a I 1 ; (37)
where
ð1 sffiffiffiffiffiffiffiffiffiffiffi
x1
I ¼ Re ds
x
0
In view of equality (36) the integral in (37) can be submitted through hyper-
geometrical and beta-functions [8] as
ð
1
ðwþ1=2Þ ð2wÞ
I¼a 1w
Re d1=2 epð1wÞi þ d d þ aepwi dd
0
pi=2 3 1 3 5 1
¼ a Re e
1=2
B ; 1 F þ g; ; ;
2 2 2 2 g
2 3
1
gg 6 F 2 þ g; g 1; g; g cosðpgÞþ 7
6 7
¼ 1=2 6 7: (38)
ð 1 gÞ 4 1 1 1g 5
þ B g; g
2 2
188 A.G. Terentiev
Resulting force can be found by integrating pressure along the body’s surface:
ðs ð1 sffiffiffiffiffiffiffiffiffiffiffi
rU 2 a 1 x @xðx; tÞ
Y ¼ pdx ¼ dx ¼
I x @x
0 0
rU 2 sagg 1 3 1
¼ B gþ ; F g; g þ ; g þ 2; g : (39)
Ia1g 2 2 2
The torque relative to the origin of the fixed coordinate is calculated similarly. Its
coefficient is
2M
CM ¼
rU 2 s2 a
(41)
pF 2g 1; 2g þ 12 ; 2g þ 2; g
¼ :
2ð1 gÞ2ð1gÞ ð1 þ 2gÞB 2g; 12 I
Consider the vertical entry of a wedge of finite length l into a fluid as shown in
Fig. 6.
The wedge sides are equal and form angles a1 and a2 with the x-axis. The wedge
moves at constant velocity U. The boundary conditions are:
(
_ x 2 ðb; 0Þ;
V 2 a1 þ T;
ct ¼
_ x 2 ð0; bÞ;
V 2 a2 þ T; (42)
’t ¼ 0 outer interval ðb; bÞ:
Fig. 6 Vertical entry of a finite-length wedge: (a) the z-plane; (b) the z-plane
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
V2b z2 b2 þ ib
wt ¼ ln
p z
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi! (43)
z 2 b2
þ iV 2 a 1 ;
z
190 A.G. Terentiev
ðt !
x
cðx; 0; tÞ ¼ U a mðtÞ 1 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi dt;
2
(45)
x2 s2 ðtÞ
0
where m ¼ T_ V 2 a 1. Substituting x ¼ sðtÞ and equating unknown function T, we
receive the required integral equation:
ðt
sðtÞ
mðtÞ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi dt ¼ t: (46)
s ðtÞ s2 ðtÞ
2
0
This equation, as Abel’s integral equation, may be solved analytically. But if the
wedge moves with a constant velocity, then sðtÞ ¼ Ut; sðtÞ ¼ Ut, and after
Ð
p=2
substitution t ¼ t sin y, integral equation (46) takes the form mðt sin yÞdy
0
¼ 1, which has the only solution as mðtÞ ¼ 2=p.
Consider now a motion of a fully immersed wedge (s>l). The stream function
pffiffiffiffiffiffiffiffiffiffiffiffiffiffion
the x-axis outer wedge is determined from (43) by substituting z ¼ s2 z2 =s
directly in the x variable as
cðx; 0; tÞ ¼
0 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1
ðt
x2 ðsðtÞ lÞ2 A (47)
¼ U 2 a mðtÞ@1 dt þ c0 ðxÞ:
x 2 s 2 ð tÞ
t0
Equating function (47) for x ¼ Ut; sðtÞ ¼ Ut, one obtains the integral equation
in the form:
Water Entry of Thin Hydrofoils 191
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ðt
t2 ðt t0 Þ2 2 t0
mðtÞ dt ¼ t 1 arcsin : (48)
t2 t2 p t
t0
þ
ir @wðz; tÞ 2 rU 2 la2 s 2
X0 ¼ dz ¼ p m ðtÞb2 ; (50)
2U 2 @z 2 l
z¼s
ðs ðs
rUlb2
X1 ¼ a1 p1 dx þ a2 p2 dx ¼ aY R; (51)
2
sl sl
Fig. 7 Dependency of
hydrodynamic parameters of
wedge on submergence depth
When distance s between the wedge and the free water surface approaches
infinity as (s ! 1), all hydrodynamic parameters approach the limits
corresponding to the Kirchhoff’s model: i.e. lift L1 ¼ prV 2 l, drag
D1 ¼ rV 2 b2 l=p, and pressure center l1 ¼ 1=4. Figure 7 shows dependencies
of ratios L=L1 , D=D1 and l=l1 on length-to-immersion ratio l/s.
transformation has been used in [13]. Two of these approaches, slightly modified
and added with numerical calculation, are discussed below.
Let the boundary of a moving body is given by equations y ¼ f ðx sðtÞÞ,
Fig. 8, which satisfy all required conditions of smallness. Then the entry problem is
determined by initial-boundary value problem for the speed potential satisfies the
following conditions:
• On the y-axis, y ¼ 0, x 2 ð0; hÞ,
• On the bottom,
x ¼ h; y 2 ð1; 1Þ ’x ¼ 0; (54)
t ’x ¼ 0; ’t g ¼ 0; (55)
where g is the gravity acceleration, ðyÞ is the free height, yðxÞ is the Heaviside
function. Besides, the speed potential and its time-derivative should vanish at the
initial time and in infinity at any time,
’ðx; y; 0Þ ¼ ’t ðx; y; 0Þ ¼ 0;
’ðx; y; tÞ ! 0; y ! 1: (56)
Gtt gGx ¼ 0; y ¼ 0;
G ¼ Gt ¼ 0; t ¼ t;
y ¼ 0; (57)
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
and behaves like ln r for x ! x; y ! with r ¼ ðx xÞ2 þ ð yÞ2 . The partial
derivative of speed potential is found by Green’s formula as follows
0 1
ð ð
1 @
’t ðx; y; t; tÞ ¼ ’t Gn ds ’tn GdsA: (58)
2p
C C
Here boundary C coincides with axes y and x that correspond to the free
boundary and the body’s wetted boundary. Unknown Green function, G, can be
obtained as follows:
r
Gðx; ; tjx; y; tÞ ¼ ln þ Fðx; ; tjx; y; tÞ; (59)
R
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
where R ¼ ðx þ xÞ2 þ ð yÞ2 , so that the first term in (59) determines the
source potential under the free boundary.
Substituting the latter equation into Eq. 57, one may obtain the partial differen-
tial equation on y-axis as
2gx
Btt gBx ¼ ; x ¼ 0;
x2 þ ð yÞ2
and setting
ð
1
2 pffiffiffiffiffi
Tðt; kÞ ¼ f1 cos½ gkðt tÞg: (62)
k
Thus,
ð
1
r ekðxþxÞ 2 pffiffiffiffiffi t t
G ¼ ln 4 sin gk cos½kð yÞ dk (63)
R k 2
0
Water Entry of Thin Hydrofoils 195
where symbol [.] denotes a jump of the value by crossing the x-axis.
The boundary value of partial derivative ½’t can be obtained from (53) but
partial derivative ½’t is unknown for an arbitrary foil. For a symmetrical wedge,
however, since ’t and G are both symmetrical with respect to the x-axis, the first
integral of Eq. 64 vanishes, and the speed potential and all dynamic parameters
could be calculated completely. From the boundary condition (53), one obtains
sf 0 ðx sÞ þ 2s_2 f 00 ðx sÞ
½’t ¼ 2€
(65)
2s_2 f 0 ðx sÞdðx sÞ; x 2 ð0; sÞ; y ¼ 0 ;
where d is the Dirac delta function. The potential can be calculated by double
integrating with respect to x 2 ð0; xÞ for ¼ 0and to time t 2 ð0; tÞ.
Let a flat-sided wedge with open angle 2a moves at a constant speed V, so that
the submergence depth is s ¼ Vt. Due to condition (65), potential (64) can be
written in the form
ðt
a 2
’ ¼ V Gðt; 0; tj x; y; tÞdt: (66)
p
0
a
p ¼ r V 2 ðp0 þ p1 Þ; (67)
p
where
Vt þ x
p0 ¼ ln ;
Vt x
196 A.G. Terentiev
ð
1 pffiffiffiffiffi pffiffiffiffiffi !
ekx gektV þ kg sinð kg tÞ
p1 ¼ 2 pffiffiffiffiffi dk:
kðk þ gÞ g cosð kg tÞ
0
Results of computations using two last equations are plotted in Figs. 9 and 10.
They show pressure coefficient Cp ¼ 2p=ar s_2 at the wedge and a ratio of drags
D ¼ D=D0 for different Froude numbers, Fr ¼ s_2 =gs. The drag D is calculated by
integrating the pressure distribution along the wedge side; the drag D0 ¼ rss_2 a2 4
ln 2=p corresponds to weightless fluid. High accuracy of the integral in p1 can be
obtained even when the upper limit is equal to 50 instead of infinity. As seen in
Fig. 9, the pressure is not equal to zero at point x ¼ 0 due to rising water level at the
wedge. The water level can be calculated using Eq. 55; the water level at the wedge
5
Cp
4
Fr = 0.5
2
Fr = 1
1 Fr = 2
Fr = 4
Fr → ∞
Fig. 9 Pressure distribution 0
0 0.2 0.4 0.6 0.8 x 1
at a wedge entering into water
2
D
1.8
α α
1.6
1.4
1.2
is ð0; tÞ ¼ pð0; tÞ=g. Figure 10 shows that the dynamic drag converges fast to the
drag at infinite Froude number.
’ ¼ ’0 þ ’1 ; (68)
where first term is a speed potential of entry into weightless liquid satisfying
conditions (53) and (54) and condition ’0 ¼ 0 at the free boundary
ðs
_
sðtÞ
’0 ¼ f 0 ðx sÞGðx; y; xÞdx; (69)
2p
0
with
cosh py pu
2h þ cos 2h
G0 ðu; yÞ ¼ ln py :
cosh 2h cos pu
2h
The second term ’1 ðx; y; tÞ satisfies Laplace’s equation in the flow domain and
the boundary conditions as follows:
ð
1
2
’1 ¼ ½Aðk; tÞ cosh kx þ Bðk; tÞ sinh kx cos ky dk;
p
0
and satisfying above mentioned boundary conditions, one can obtain the factor,
B ¼ A tanh kh, the ordinary differential equation as
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Att þ o2 A ¼ FðtÞ; o¼ kg tanh kh (70)
198 A.G. Terentiev
ðt
1
Aðk; tÞ ¼ Fðk; tÞ sin oðt tÞ dt; (71)
o
0
where
ðs
gs_
Fðk; tÞ ¼ f 0 ðx sÞ cosh kðh xÞ dx:
cosh kh
0
Now, the second function in (68) as well as other hydrodynamic parameters can
be calculated. Namely, the drag coefficient due to gravity is represented as
ð
1
00 8g ðcosh½kðh sÞ coshðhkÞÞ cosðo sÞ
C D ¼ Fþ
ph coshðhkÞ
0
(72)
ð
1
8
þ ðsinh½kðh sÞ sinhðhkÞÞo sinðo sÞFdk ;
ph
0
where
Using another approach Kuznezov and Manevich [14] obtained results similar to
Eq. 72, which can be transformed asymptotically to Yim’s results.
The water entry without gravity force is determined by speed potential (69).
Coefficients of pressure and drag in this case are calculated by
ðs
1 2
C0p ¼ Gðx; 0; sÞ; C0D ¼ Gðx; 0; sÞ dx: (73)
p ps
0
Figure 11 shows the effect of immersion depth on drag of the wedge entering
weightless water with the finite depth. Coefficient C0D o ¼ C0D ðh ! 1Þ ¼ 8 ln 2=p
corresponds to infinite depth; coefficient C1D ¼ C0D ðs ¼ hÞ ¼ 2:970 corresponds to
the case when the wedge top touches the bottom.
An effect of Froude numbers on drag coefficient C00D is shown in Fig. 12. It should
be noted, that the full drag of the wedge consists of dynamic drag, which is
proportional to the angle square, a2 , and hydraulic pressure ph ¼ rgx, which is
proportional to gravity acceleration g and angle a. Thus, the dynamic pressure and
Water Entry of Thin Hydrofoils 199
1.4 α α
1.2
1
0 0.2 0.4 0.6 0.8 1
s/h
α α Fr = 0.5
g
2 0.75
1
1 Fr = 2
0
0 0.2 0.4 0.6 0.8 1
s/h
dynamic drag should be taken into account if Fr 1=a, i.e. Froude number should
be quite large. Otherwise it suffices to take into account the hydraulic pressure only.
References
1. Logvinovich GV. Hydrodynamics of streams with free boundaries. Kiev: Nauk. Dumka; 1969
(In Russian).
2. Sagomonyan AYa. Penetrating. Moscow University; 1974 (In Russian).
200 A.G. Terentiev
3. Korobkin AA, Pukhnachov VV. The initial stage of water impact. Ann Rev Fluid Mech.
1988;20:159–95.
4. Terentiev AG. Problems in the theory of high-speed hydrodynamics. Proceedings of the
International Summer Scientific School “High Speed Hydrodynamics”. Cheboksary; 2002.
p. 11–27.
5. Yim B. Investigation of gravity and ventilation effects during the water entry of thin foils.
Proceedings of the IUTAM Symposium “Unsteady Flow of Water at High Speed”. Leningrad;
1971. p. 6.
6. Terentiev AG. Inclined entry of a thin body into incompressible liquid. Izvestia AN SSSR
MZhG. 1977b;5:16–24 (In Russian).
7. Terentiev AG. Inclined entry in ideal non-gravity liquid of thin body with ventilated cavity.
Izvestia AN SSSR MZG. 1979;3:66–76 (In Russian).
8. Gradstain IS, Ryzhik IM. Tables of integrals, sums, series and products. M. GIFML; 1962.
9. Dobrovolskaya ZN. On some problems of similarity flow of fluid with a free surface. J Fluid
Mech. 1969;36(part 4). pp. 805–829.
10. Terentiev AG, Mikhailov VM. Non-stationary moving of the slender bodies in an ideal fluid.
Proceedings of the Non-stationary moving of bodies in a fluid. Cheboksary; 1979. p. 111–48.
(In Russian).
11. Moran JP. The vertical water-exit-and-entry of slender symmetric bodies. J Aerospace Sci.
1961;28:803–12.
12. Porfiriev NP. Vertical moving of a thin axisymmetric body to free surface of a gravity liquid.
Proceedings of the “High Speed Hydrodynamics”. Cheboksary: Chuvash State University;
1981. p. 100–9. (In Russian).
13. Galanin AV, Terentiev AG. Boundary value problems of linear hydrodynamics. Cheboksary:
Chuvash University Press; 1984 (In Russian).
14. Kuznetsov AV, Manevich ASh. Gravity effect in water entry of thin foil. Proc. Trudy NKI.
Nikolaev: NKI; 1979. p. 152. (In Russian).
Study of the Parameters of a Ventilated
Supercavity Closed on a Cylindrical Body
Abstract
The paper reports the procedure and results of an experimental study of
ventilated supercavities on axisymmetric bodies of various geometries in a
water tunnel. Hysteresis processes for variously shaped model bodies are con-
sidered. The results of recording of unsteady processes of cavity size variation
caused by gas injection shutoff are presented. The effect of the body geometry
on gas entrainment and unsteady cavity size variation is shown.
1 Introduction
2 Experimental Procedure
Experimental studies were carried out on the small water tunnel (SWT) at the
Hydrodynamic Laboratory, the Institute of Hydromechanics of NASU. A schematic
diagram of the test model installed in the SWT working area together with auxiliary
equipment is shown in Fig. 1. The central element of the body is a hollow
cylindrical rod 1 of diameter 20 mm and length 1.1 m with a cavitator 2 of constant
diameter Dn ¼ 20 mm installed in its fore part. Replaceable elements were used to
install on the rod 1 axisymmetric bodies of various sizes, which included a
cylindrical middle part 3 of variable diameter Db and conical end parts 4 with
cone angles of 6 8 . The geometry of the test bodies is shown in Table 1.
A photo of the model with a cavity in the SWT working area is shown in Fig. 2.
The flow rate of the air injected into the cavity was measured with a type RS-5
rotameter 5. As shown in Fig. 1, air injection into the cavity was controlled using a
valve 6 and electric valve 7, the latter serving to obtain unsteady flow regimes. The
electric valve 7 is connected via a control unit 8 to a type ChZ-54 electronic
frequency meter 9, which served to time the development of unsteady processes.
Study of the Parameters of a Ventilated Supercavity Closed on a Cylindrical Body 203
Fig. 1 Schematic of the model and auxiliary equipment in the SWT working area
Cavity flows were recorded using a SHARP-type video camera 10, model
VL-H420, at a shutter speed of 1/50 s. To determine the cavity dimensions from
the recorded video data, the glass walls of the SWT working area had a line scale 11
marked every 100 mm.
The experiments were conducted at a constant flow velocity in the SWT working
area V0 ¼ 8:9 m=s, a constant cavitator immersion depth equal to 180 mm from the
free surface, and a constant cavitator diameter Dn ¼ 20 mm. At this flow velocity,
the effect of gravity is noticeable, which manifests itself in so called cavity
surfacing, as a result of which the cavity symmetry axis is at an angle to the flow
axis. Because of this, for the measured data to be comparable the test bodies were
installed at the corresponding constant angle to the flow axis thus providing
identical axisymmetric cavity closure on the body surface in all flow regimes.
In the experiments, the following parameters were changed:
– The geometry of the body downstream of the cavitator according to Table 1;
– The rate of air injection into the cavity over the range Qg ¼ 0 0:94 L=s
– The air injection regime (stepwise injection regime and air injection shutoff).
Changing the air injection rate changed the cavity geometry and the
corresponding cavitation number
p 0 pc
s¼
r V02 =2
where V0 ¼ 8:9 m=s is flow velocity; r is liquid density in the flow; p0 is pressure in
the flow at a depth of 170 mm from the free surface, which corresponds to the mean
position of the center of the cavity volume; and pc is the pressure in the cavity. For
steady cavities at a fixed air injection rate the cavitation number s ranged from
about 0.038 to about 0.079. As to the air injection regime, the following two types
thereof were studied.
In this case, the air injection rate was changed manually in steps D Qg ¼ 0:08 L=s
using the valve 6 (see Fig. 1). In doing so, for each type of the body downstream of
the cavitator the cavity length Lc was measured in two runs. In the first measure-
ment run, the air injection rate was increased from minimum values
Qg ¼ 0:020 0:028 L=s, at which a steady cavity starts forming, to the maximum
value Qg ¼ 0:94 L=s. Then in the second measurement run the air injection rate
was decreased from its maximum value to the value at which the cavity collapses.
The steady cavity length Lc for each fixed value of Qg was measured directly using
the line scale 11 provided on the glass walls of the SWT working area. The steady
flow regimes were also recorded using the video camera 10. The measured cavity
length Lc was plotted versus the air injection rate Qg for each direction of change of
Qg .
Study of the Parameters of a Ventilated Supercavity Closed on a Cylindrical Body 205
In this case, for each body type downstream of the cavitator experiments were run
in two stages:
At the first stage, the air injection rate was controlled manually in such a way as to
form the largest cavity for the chosen fixed air injection rate. To doso, with the electric
valve 7 open (see Fig. 1) the air injection rate was increased to Qg max using the valve
6, after which the air injection rate was gradually decreased to Qg ¼ 0:605L=s and the
cavity stabilized at this constant air injection rate in the fixed position of the valve 6.
This procedure provided a starting steady cavity that had a size close to the maximum
one and at the same time was sensitive enough to changes in the air injection rate.
At the second stage, the electric valve 7 was actuated via the control unit 8 to
shut off the air injection into the cavity synchronously with turning on the digital
timer/frequency meter 9 to time the evolution of the cavity. The experiment was
recorded by the video camera 10 so that the rotameter 5, the digital display of the
timer/frequency meter 9 and two line scales 11 on the glass walls of the SWT
working area were in picture together with the unsteady cavity under study. This
allowed one to measure the cavity length Lc from the recorded video data at known
time intervals starting from the air injection shutoff time, which was also seen in
picture as an abrupt sinking of the float of the rotameter 5.
The length of the cavity in the process of its collapse was measured every
second. Because of perspective distortions in the picture, the value of Lc was
measured using two line scales on the walls of the working area, and their average
was computed. In view of fluctuations, which are characteristic of ventilated
cavities due to the peculiarities of the gas entrainment mechanism [2–4, 9–11]
and are all the more probable in the case of an unsteady cavity, each experiment was
recorded by the video camera two to three times. For each video, at least two
independent measurements of the cavity time evolution were made, and for each
type of process the average was computed.
The measured data were processed to give the rate VL of change of the length Lc
D Lc
of the cavity VL ¼ for fixed times t from the start of the process, or, what is the
Dt
same, for fixed values of the length Lc of the unsteady cavity, and the rate VL of
cavity collapse along the body in the flow was plotted versus Lc .
In the plots below, the measured linear quantities are normalized to the cavitator
Lc VL 1
diameter Dn , i.e. Lc ¼ ; VL ¼ (s ). The air injection rate is shown as the
Dn Dn
4Qg
dimensionless air injection coefficient CQ ¼ . For the unsteady regimes
Vo p D2n
studied, the air injection coefficient for the starting steady cavity was CQ ¼ 0:216.
The results of measurement of unsteady cavities were also used in estimating the
cavity gas entrainment rate in the characteristic portions of the body profile. The
difference of the values of the volume of an unsteady cavity measured one second
apart (minus the corresponding part of the body volume on the same interval)
206 Yu.D. Vlasenko and G.Yu. Savchenko
characterizes the rate Qg of gas entrainment from the cavity. In this case, the length of
the body portion under consideration is numerically equal to the cavity collapse rate.
3 Experimental Results
In its initial portion, the descending branch of the dependence Lc CQ is also
identical in all plots, and it shows that the response of a developed cavity to a
sizeable decrease in the air injection rate is insignificant. However, here, too, on
decreasing the air injection rate to some threshold value a cavity rapidly diminishes
to the point of its collapse. At this stage, the plotted dependences differ quantita-
tively, which is due to different body geometries in the cavity closure zone.
By and large the plots demonstrate the complexity and limited possibilities of
controlling the parameters of a ventilated cavity by varying the air injection rate.
Experimental plots of variation of the cavity collapse rate along the body in the flow
V L ¼ f ðLc Þ are depicted in Figs. 5 and 6, which also show, on comparable scales,
the corresponding body profiles. Despite considerable quantitative differences, the
plots show that the variation of VL along the body is identical in form. The process
of cavity collapse after air injection shutoff includes three characteristic stages: two
maxima of the cavity collapse rate in the zone of the fore and the aft end of the
body, respectively, and a minimum of VL in its cylindrical middle part. A compara-
tive analysis of the plots based on known data on the mechanism of cavity gas
entrainment makes it possible to interpret the above features of the process under
study as follows.
208 Yu.D. Vlasenko and G.Yu. Savchenko
The first maximum of the cavity collapse rate VL near the aft end of the body
seems to be due to two factors:
– The features of the body geometry, namely, the presence of a conical aft portion
(Fig. 6), as a result of which the movement of the wash zone along the cone at the
initial stage increases the perimeter around which the gas is entrained from the
cavity;
– The features of the process itself because this effect in a weaker form is observed
on cylindrical bodies too (Fig. 5). This is confirmed by the data presented below
on the variation of the cavity gas entrainment rate Qg along the body, where the
maximum of Qg is also observed near the aft end of the body (compare with
Figs. 9 and 10).
In view of the aforesaid, it seems to be natural to relate the minimum of the
cavity collapse rate on the cylindrical middle part of the body to the above factors
ceasing to act. As to the abrupt increase in VL near the fore end of the body, a similar
effect was found out in experiments on the collapse of ventilated cavities in free
closure conditions [11]. In this case the cavity size also decreases at a progressively
increasing rate, and thus it is reasonable to relate this effect mainly to a change of
Study of the Parameters of a Ventilated Supercavity Closed on a Cylindrical Body 209
the cavity gas entrainment mechanism with decreasing relative cavity size. This
stage is characterized by a sharp increase in the intensity of re-entrant jets in the
cavity closure zone and thus in the level of cavity surface disturbances, as a result of
which the gas entrainment rate increases. In this connection, it is pertinent to note
that it is technically difficult to determine stable mean values of VL because the
process is fast and the measured parameters undergo sizeable fluctuations due to
pulsating gas release from the cavity in the form of foam.
Some generalized idea of the quantitative effect of the geometry of the body in
the flow on the cavity closure process can be inferred from the plots in Fig. 7, where
the maximum and minimum process rates are plotted versus the body diameter
(the maximum values of VL in the plot relate to the aft zone of the body). It can be
seen from the plots that the cavity collapse rate increases with the diameter Db of
the circumference around which the gas is entrained from the cavity. Note that the
measured maximum values of the cavity collapse rate are more than an order of
magnitude lower than the mainstream velocity V1 , i.e. VL <<V1 . This limits the
possibility of cavity control by air injection.
At the same time, this order of magnitude of VL suggests that the gas entrainment
process and the time evolution of the cavity size are quasi-steady in character. This
allows one to determine the instantaneous volumes of an unsteady cavity as an
ellipsoid of revolution [15–17] using the familiar dependences of the geometry of a
steady cavity on the cavitation number.
The volume Wc of an ellipsoid of revolution with semi-axes Lc =2 and Dc =2 is
determined as
210 Yu.D. Vlasenko and G.Yu. Savchenko
p p L3c
Wc ¼ Lc D2c ¼ ; (1)
6 6 l2c
Lc
where Lc ; Dc are the cavity length and maximum diameter, and lc ¼ is the
Dc
cavity aspect ratio. One notes that the parameter lc of a steady cavity, all other
factors being the same, is unambiguously related to the cavitation number. Using
the familiar dependence for the diameter of an axisymmetric cavity downstream of
a disc
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Dc Cxo ð1 þ sÞ 1þs
¼ 0:84 (2)
Dn Ks s
Lc 1:92 3 s
¼ ; (3)
Dn s
Similar calculations with the use of Epshtein’s formula for the cavity length [2]
rffiffiffiffiffiffiffi
Lc s Cxo
¼ 1:67 ; (5)
Dn K
Here, Cxo ¼ 0:82 is the disc cavitator drag coefficient at s ¼ 0 ; and K 0:98 is
an empirical coefficient.
The dependences lc ðLc Þ obtained by Eqs. 4 and 6 and plotted in Fig. 8 are in
rather close agreement with each other and in satisfactory agreement with the data
of the experiments described above. The spread of experimental points in the region
of small values of Lc and lc is due to the above-mentioned cavity instability at the
final stage of collapse. To determine the instantaneous cavity volume by Eq. 1, the
experimental dependence lc ðLc Þ was used.
The obtained data on the time evolution of gas cavity entrainment are plotted in
Figs. 9 and 10 as dependences CQ ðLc Þ; for comparison, the profiles of the
corresponding models are shown in the figures too. Here, the plots CQ ðLc Þ are
also qualitatively identical in form, and they show a well-defined maximum in the
zone of the aft end of the models. The characteristic feature is that the maximum gas
entrainment rate is clearly seen to be located at much the same distance from the
cavitator at the section that corresponds to Lc 45. In a quantitative sense, CQ max
increases with the model diameter. In the zone of the cylindrical middle part of the
test bodies, the gas entrainment rate monotonically decreases with a greater or
lesser intensity as the collapse progresses.
Fig.
11 Plots of CQ max and
CQ min versus the model
diameter
lines. These plots demonstrate the opposite behavior of the dependence CQ Db :
the gas entrainment rate decreases not only as we approach the cavitator, but also
with increasing body diameter. The latter seems to be due both to more intensive
gas release early in the cavity collapse and to a relatively smaller initial gas volume
in a cavity on large-diameter bodies.
Conclusions
1. Cavity size control by varying the gas injection rate into the cavity is highly
problematic because hysteresis is present, the gas injection rate is not a single-
valued function of the cavity closure location, and the rate of change of the
cavity length governed by gas entrainment is limited.
2. Experiments have shown that the gas injection rate required for sustaining a
ventilated cavity closed on a circular cylinder varies monotonically with the
cavity length within the cylindrical portion of the model.
3. The evolution of an unsteady cavity on air injection shutoff is characterized by
intensive gas release early in the cavity collapse, the maximum gas entrainment
rate being a linear function of the body diameter. Later in the cavity collapse, the
gas entrainment rate decreases monotonically with a greater or smaller intensity.
References
1. Logvinovich GV. Free-boundary flow hydrodynamics (in Russian). Kiev: Naukova Dumka;
1969.
2. Epshtein LA. Methods of dimensional theory and scaling in vessel hydromechanics problems
(in Russian). Leningrad: Sudostroenie; 1970.
3. Epshtein LA. Characteristics of ventilated cavities and some scale effects (in Russian).
In: Unsteady high-speed water flows. Moscow: Nauka; 1973. p. 173–85.
214 Yu.D. Vlasenko and G.Yu. Savchenko
Abstract
The objective of this paper is to discuss a new type of a shock-free
supercavitating wing with the controllable forward flap at the leading edge and
the spoiler mounted on the trailing edge. The viscous-inviscid interactive
method is applied to determine the hydrodynamic performances. The controlla-
ble forward flap plays a vital role to create cavity covered the whole surface of
the upper part at any speed in the operating range. The inviscid flow problem is
based on non-linear potential flow theory. The Tulin-Terentiev cavity closure
scheme with single-spiral vortex termination is adopted and smooth detachment
is taken into account near the stagnation zone in the spoiler vicinity. The
parametric study is carried out in the framework of inviscid flow problem. The
hydrodynamic performances of the hydrofoil are finalized with RANSE solver
of CFD code. Such a type could be used as a section of submerged
supercavitating wing.
1 Introduction
2 Problem Formulation
Let us consider the nonlinear shock–free cavitating problem for the flat plate with
the spoiler [5], at arbitrary cavitation number, see Fig. 2, the influence of gravity
being neglected. Tulin-Terentiev cavity closure model with single-vortex termina-
tion was adopted. The origin of the Cartesian coordinate system is taken at the
B
0.1
D
–1 –0.5
Fig. 1 Sketch of the shock-free supercavitating hydrofoil with spoiler and controllable forward
flap at the sharp leading edge
plate’s trailing edge, x-axis being directed downstream and y upwards. There is an
incident stream with speed V1 coming from the left. The region occupied by the
fluid is bounded by the rigid boundaries OD, OA and DB and the free surfaces AG,
BG, EF, interval DE, FA being unknown length. Note that the interval OD ¼ l,
OA ¼ e, and DB ¼ lF. The incidence angle is a, the inclination angle of the spoiler
is b and the inclination angle of the forward flap DB is g. It is of importance that the
dividing streamline would meet the cavitating plate at the vertex of the wetted
surface BDOA, namely at point D, to provide the shock–free cavitating mode. In
this case the length of the leading flap lF is to be treated as an unknown parameter of
pffiffiffiffiffiffiffiffiffiffiffi number sb0, the velocity
the problem, the angle g being given. Since the cavitation
absolute value on the free surfaces is V0 ¼ V1 1 þ s. The cavity terminates in
spiral vortex placed at points G and G. The velocity V1 on the boundary of
stagnation zone EF is unknown. The zone detaches smoothly (tangentially) at an
unknown point E on the plate and smoothly reattaches at another unknown point F
on the spoiler. Thus, the problem arising is a two-dimensional (2D) nonlinear
problem of the theory of jets in an ideal fluid and its solution can be achieved
through the corresponding methods.
With the correspondence between the physical z ¼ x + iy plane and auxiliary
quadrant ς ¼ + ix shown in Fig. 2, the exact solution can be written out in the
form of two derivatives of the complex potential o ¼ ’ + ic with respect to the
physical and auxiliary variables:
g=p 2ibB
dw ia y1 ðB id1 Þ y1 ð B Þ y1 ðBÞ
¼ V0 e e p exp ic1 (1)
dz y1 ðB þ id1 Þ y1 ð B g1 Þ y1 ð B þ g1 Þ
dw y1 ðB g1 Þ y1 ðB þ g1 Þ y1 ðB id1 Þ y1 ðB þ id1 Þ
¼ N1 y1 ð2BÞ 2 (2)
dB y1 ðB B1 Þ y21 ðB þ B1 Þ y21 ðB B1 Þ y21 ðB þ B1 Þ
ðB ðB
dz dz dw
z ð BÞ ¼ dB þ ZE ¼ dB þ ZE (3)
dB dw dB
pt pt
2 2
p
zð0Þ ¼ zB ¼ leiðpaÞ ; z ¼ zA ¼ eeiðaþbÞ ;
þ 2
dw dz
ð u1 Þ ¼ V1 ; du ¼ 0;
dz du
u1
dw pt p pt
¼ V1 eia ; z þ ¼ zA :
dz 2 2 2
dm
/¼ ;
ds
where m ¼ arg(dw/dz) denotes the tangential angle to the contour and s is the arc
coordinate. It is seen that the Brillouin’s condition implies that the stagnation zone
detaches smoothly in point E. The flat plate has zero curvature at any point and that
is why at point E the second derivative of the function of stagnation zone shape
ordinates with respect to x coordinate is to be equal to zero. It is seen that this
condition can be readily reduced to
dm d dw pt
¼ arg ¼0 (4)
du zE du dz 2
The numerical investigation covered all cavitation number s less than 1 and the
angle of attack a less than 30 . Figure 3 illustrates the lift coefficients, drag
coefficients and angle g versus the cavitation number at any angle of attack a. It
is obvious that the hydrofoil enables to create cavity in entire range of cavitation
number from smaller to bigger, i.e. not only high velocity inflow but also low one
respectively.
Besides it can be seen that hydrofoil provides foil-borne lift for any cavitation
number by two means, varying angle g without changes in angle of attach a and
varying the angle of attack a and angle g simultaneously. The former is more
efficient than the latter because lift to drag ratio is considerably high. For instance,
such ratios are 9.94 and 1.32 at angles of attack a of 3 and 30 respectively for the
cavitation number, s ¼ 0.21. As a result, it is noted that the difference in drag
coefficients is 16.87 times while lift coefficients just differ in 2.24 times.
Therefore varying only leading edge angle g is able to achieve highest lift to drag
ratio in the whole range of the cavitation number s, angle of attack a being kept as
small as possible. Moreover the hydrofoil can also facilitate a small variant of lift
force at any cavitation number through simultaneous adjustment of the angle of
Hydrodynamic Performances of 2- Shock-Free Supercavitating Hydrofoils 219
Fig. 3 CL, CD and g versus cavitation number s for the SC hydrofoil with a parameter set of
length lF ¼ 0.1, e/l ¼ 0.02, b ¼ 90
attack a and leading edge angle g. Meanwhile it should be aware that the hydrofoil
thickness affects performance because it has relations with determination of the
minimum angle of attack a and the relative spoiler length ratio e/l.
The thinner the foil thickness is, the better the hydrodynamic fineness will be. On
the other hand, there are some restrictions to reduce the foil thickness since the
thickness requirement of hydrofoil play vital role in view of strength. Figure 4
illustrates flow patterns for various cavitation numbers s at a unique angle of attack.
Cavity is shown by grey lines and the boundary of the stagnation zone can be
found near spoiler. The problem was solved under the assumption that the
Brillouine’s condition is satisfied. In figures, the relative height to chord length is
about 0.12 between the trailing edge and upper free surface boundary of cavity. The
authors left to consider strength viewpoints. However the numerical results reveals
that the larger angle a is required for a thicker hydrofoil to keep contact with water
vapour only on the entire surface of upper part of foil but it ensures lift to drag ratio
will be lower. It should therefore be noted that the larger angle of attack is favoured
to obtain enough clearance between upper and lower boundary of cavity for thicker
foil thickness while the angle of attack a is required as small as possible so as to
220 Z. Win et al.
Fig. 4 Flow patterns: shock-free supercavitating hydrofoil with stagnation zone in the spoiler
vicinity, and free stream line flow (gray lines); relative length of forward flap, lF ¼ 0.1, spoiler
length to chord ratio, e/l ¼ 0.02 and inclination angle b ¼ 908. Stagnation point coincides with
the leading edge. The Brillouine’s condition is satisfied
gain optimum the lift to drag ratio. It means that the designer must consider to
achieve the best compromise between hydrodynamic and strength aspects.
0.003 mm. One can determine the minimum cell thickness by using the equation in
[7].
qffiffiffiffiffiffiffiffiffiffi
yP ¼ yþ n= u1 Cf =2 ; where yþ <5 or yþ >30 (5)
Cf =2 ¼ 0:037Re1=5 (6)
The cavitating flow problems are generally solved in two steps, first in single
phase flow and second in two phase flows. That is why the cavitating flow problems
should be solved after having obtained the reasonable pressure fields around the
foil. Single phase flow needs to be converged until the residual of momentum
equation is below 106. In the first step, the boundary conditions of the flow domain
are taken into account with velocity inlet and outflow scheme. After convergent of
single phase solution, boundary condition has to be changed with a pair of velocity
inlet and pressure outlet and the cavitation mode is turned on. As for second step,
the solutions will be reasonable when all of residual are below 103. Steady
segregated solver is used at single phase flow but unsteady solver is suitable for
multi-phase flow.
The results from CFD code reveal that the hydrodynamic characteristics in the
frame work of the non-linear inviscid flow seem to be the same as in a viscous flow.
The cavity shapes are somewhat different between inviscid and viscous flow due to
viscous effect, see Fig. 5.
Conclusions
The 2-D hydrofoil presented in this paper ensures to achieve supercavitating
flow at any speed, low or high. Unfortunately authors left to conduct 3-D flow
analysis for this type. Such a hydrofoil can be used as a submerged wing in
hydrofoil vessels. Its features are as follows:
(a) Capability of obtaining foil-borne lift even at low speed
(b) Reducing take-off speed
(c) High lift to drag ratio
One can find the 2-D optimum shape of such types of hydrofoil in the frame
work of non-linear inviscid flow on the basis of initial parameters such as
cavitation numbers in the whole operating range, reasonable relative length of
forward flap, possible range of angle g, minimum foil thickness, inclination
angle b, and the relative spoiler ratio. Such a method is not a time consuming
one like CFD method but the resulting hydrodynamic characteristics are not too
far from that of CFD code.
References
1. Gurevich MI. The theory of jets in an ideal fluid. Moscow: Nauka; 1979 (In Russian).
2. Chaplygin SA. To the problem of jets in a incompressible fluid. Trudy Otdeleniya Phys. Nauk.
X(1), Moscow; 1899. (In Russian).
3. Terentiev AG. Mathematical aspects of cavitation. Cheboksary: Chuvash State University; 1981
(In Russian).
4. Achkinadze AS, Fridman GM. Optimal sections for supercavitating propellers with spoiler and
preset leading edge angle. St. Petersburg: St. Petersburg State Marine Technical University;
2000.
5. Fridman GM, Uryadov AK. Cavitating flat plate with stagnation zone in the spoiler vicinity.
Proceedings of The Second International Summer Scientific School “High Speed Hydrodynam-
ics”; June 2004. Cheboksary; 2004. p. 83–90.
6. Zaw W, Fridman GM. The study of viscous effect on spoiler mounted on the trailing edge of the
supercavitating hydrofoil. SubSeaTech 2009; June 2009. St. Petersburg; 2009.
7. Nikushchenko DV. The study of viscous flow of incompressible fluid from the basis of
FLUENT, Lecture note (in Russian). St. Petersburg: St. Petersburg State Marine Technical
University; 2006.
Index
A B
Added mass, 111 Bench, 108
coefficient, 112 electric, 30
disc, 111 firing, 29, 33
factor, 112 measuring, 108
flat plate, 111 Body(ies)
ring, 112 aft end, 208
supercavitation flow, 113 axisymmetric, 68, 192, 202
Aerodynamic effectiveness, 89 biconvex flat, 192
Air injection conical, 202
coefficient, 205 conical aft portion, 208
rate, 204 conical ends, 202
regime, 204 cylindrical, 202
threshold value, 206, 207 cylindrical middle part, 202
Air ventilation, 43 of revolution, 66, 67, 202
Angle, 154–156, 161, 164, 165, 168 partially submerged slender, 177
amplitude, 172 thin, 192
of attack, 28, 151 Bottom, 41, 44, 47, 53, 62
d-control, 173 Boundary conditions, 11–14, 42, 71, 179, 220
horizon, 159 dynamic, 129
incidence, 217 dynamical, 72
inclination, 173, 183, 217 kinematic, 129
leading edge, 219 Boundary layer, 115
pitch, 151, 166 Brillouin’s condition, 218, 219
roll, 151, 172 Bubbles
sliding, 151 cloud, 2
speed, 28 coagulation, 3
thrust, 171, 173 dynamics, 2
trajectory, 171–173 interacting, 4
trim, 39 mono-dispersed, 7
yaw, 151, 171, 172 radius, 2, 17
Aspect ratio, 74 spherical, 3, 7
body, 84 violent collapse, 17
cavity, 83
hull, 84 C
maximum hull, 84 Catapult, 30, 32, 33
optimal, 99 centrifugal, 28
P R
Parameter, 144 Range
plane, 130, 131 increase, 98
region, 130 maximal, 93
Parametric variable, 130 maximization, 91
Photography, 28 RANSE solver, 215, 220
high-speed, 31 Re-entrant jets, 2, 10, 82, 102, 209
stroboscopic, 28 Regime
Plane non-separated, 163
complex, 131, 133 partial cavition, 163
parameter, 134 supercavitation, 163
physical, 130, 144, 217 Reynolds Averaged Navier–Strokes (RANS)
Planing, 40–42, 49, 50, 53, 62, 89 solvers, 216
boat, 50, 53 Reynolds numbers, 5, 80, 123, 221
within the cavity, 157, 163 critical value, 84
hull, 39, 47, 62 Richardson number, 123
Potential Ring vortex, axisymmetric, 66
complex, 127, 130, 134, 177, 217 Rotameter, 202
speed, 193, 197 Rudders, 151, 154, 156, 157, 174
velocity, 42, 132
Pressure, 15, 42, 43, 53
center, 184, 188, 192 S
coefficient, 15, 17, 196 SC-body motion, 148
distributions, 7, 15, 39, 41, 47, 52, 59 3D, 149, 158, 174
dynamic, 195 inertia, 162
function, 45, 46 Scheme, 221
gradients, 17 central finite difference, 14
Index 229