You are on page 1of 7

A NUMERICAL FORMULATION APPLICABLE TO SURFACE PIERCING HYDROFOILS AND PROPELLERS by Cedric Savineau Department of Ocean Engineering Massachusetts Institute

of Technology Cambridge, MA 02139 Spyros Kinnas1 Department of Civil Engineering University of Texas at Austin Austin, TX 78712

ABSTRACT The flow field around a fully ventilated two-dimensional hydrofoil of arbitrary geometry is considered. The presented theory is a non-linear, timemarching, potential based boundary element method. The perturbation potential on the cavity is a function of submergence and time. The non-linear cavity geometry is determined iteratively within the solution at each timestep. A linearized free surface condition, at infinite Froude number, is enforced by using the negative image approach. The developed analysis is shown to converge well with iterations per timestep, even though the cavity geometry has been found to be sensitive to the panel arrangement at the foil leading edge. Pressure distributions compare very well with linearized analytical results at small angles of attack, but the non-linear effects are noticeable at larger angles of attack. NOMENCLATURE c Dij ds Fr g H n p q qn Foil chord length Influence coefficient due to a normal dipole Panel arc length Froude number Gravitational acceleration (9.81m/s2) Cavity thickness height Unit normal to the hydrofoil or cavity surface Pressure Total flow velocity Velocity normal to cavity surface

qs t Sij u U x y yFS

Velocity tangential to cavity surface Time Influence coefficient due to a source Horizontal perturbation velocity Foil entry velocity Horizontal coordinate Vertical coordinate y location of the free surface Density Total flow potential Perturbation potential

INTRODUCTION With the increased demand of high speed vessels, surface piercing propellers have come back as an efficient alternative mode of propulsion. The design of surface piercing propellers has traditionally been based on empirical and experimental methods. Analytical methods based on conformal mapping techniques were developed during the 1970s by B. Cox [1]2, B. Yim [4] and D.P.Wang [5] for the analysis of ventilating and cavitating 2D hydrofoils. This paper describes the latest developments by the authors, for the flow analysis around a surface piercing 2D foil of arbitrary geometry, using a non-linear, potential based, boundary element method [3], which is applied in a time-marching sense.
2. Numbers in square brackets designate References at the end of the paper. 1. When this work began, Department of Ocean Engineering, MIT, Cambridge MA

2.0 PROBLEM STATEMENT Formulation Consider a two-dimensional hydrofoil as shown in Fig. 1, entering an air-water interface, which is initially at rest, with a constant velocity, U, normal to the free surface. The flow is considered with respect to the foil. The speed of entry is sufficiently high for the foil to fully ventilate starting at the sharp leading edge and along the suction side. Ventilation can be treated as cavitation with the cavity pressure being equal to the atmospheric pressure at the free surface.
y

cient, the Froude number defined by equation (2) are in the 5-10 range. U F r = --------(2) gc where U is the entry velocity, g is the gravitational constant and c is the foil chord length, or alternatively, the propeller diameter. In this regime, gravity effects may be neglected [4], hence the infinite Froude number assumption made in this analysis. By implementing the infinite Froude number assumption in the linearized dynamic boundary condition, we obtain that the free surface condition reduces to zero horizontal perturbation velocity, u, on the linearized free surface: = u = 0 x (3)

p = p(atm) Uoo Uoo

Dynamic and Kinematic Foil Boundary Conditions The flow is required to be tangential to the wetted hydrofoil surface, as well as to the cavity surface. The kinematic boundary condition is given by: = U n (4) n where n is the unit normal to the hydrofoil or cavity surface, directed into the fluid domain. This condition is applied to the exact non-linear cavity surface and wetted foil surface. The perturbation potential on the cavity will be a function of submergence and time, and will be convected with the flow. In complete linear theory [4], the perturbation potential is a function of y and time only. To find the cavity shape at subsequent iterations, the cavity surface is convected with the local velocity. Henceforth, this condition is non-linear in nature. This is somewhat analogous to that of the shed vorticity in the wake of an oscillating foil or a foil in unsteady inflow. Finally, at the leading edge, continuity of perturbation potential must be satisfied: (5) wetted = cavity
LE LE

q Uoo+ v u

c y(FS)= tU oo x a

Uoo
Fig. 1: Sketch of Surface Piercing Hydrofoil with coordinate system moving with the foil

Assuming the fluid is inviscid and incompressible, and that the resulting flow is irrotational, the total velocity flow field, q, can be expressed in terms of either the total potential , or the perturbation potential , as: q = = U + (1)

Free Surface Boundary Condition For applications where surface piercing hydrofoils or propellers are effi-

This is similar to the Kutta condition at the trailing edge of a hydrofoil. NUMERICAL IMPLEMENTATION Free Surface Boundary Condition The method of negative imaging [2], is used to enforce the free surface boundary condition (3). The imaged foil and cavity, shown in Fig. 2, is represented by sinks and dipoles with opposite normal.

right hand side, we obtain the following system of equations:


D 1, 1 K 1 DN DN KN
--2

N 1, --- + 1 2

S 1, N SN SN

1 N
--2

---, 1 2

SN N SN

---, --- + 1 - 2 2

---, N 2

--- + 1, 1 2

KN

--- + 1 2

N --- + 1, --- + 1 2 2

--- + 1, N 2

N --- + 1 = 2 n N n 1 n N n 0 LE
--2

(7)

D N, 1 K N

S
N N, --- + 1 2

S N, N

S 1, 1 S SN
DIPOLE with opposite normal

N 1, --2

N 1, --- + 1 2

D 1, N DN DN DN

---, 1 2

SN N
---, --- 2 2

DN N DN DN

---, --- + 1 - 2 2

---, N 2

SN SN

--- + 1, 1 2

SN SN

N --- + 1, --2 2

N --- + 1, --- + 1 2 2

--- + 1, N 2

SINK

--- + 2, 1 2

N --- + 2, --2 2

N --- + 2, --- + 1 2 2

--- + 2, N 2

F.S.
i=N i=N-1
SOURCE DIPOLE

S N, 1 S N
N, --2

D
N N, --- + 1 2

D N, N

N t = --- 1 2

LE t = 1

i=1 i=2

i=N/2+2 i=N/2+1

i=N/2-1 i=N/2

where Ki=Di,n/2+Di,n/2+1. This system of equations is solved at time t=N/2.T, and t=0 at the moment the foil enters the free surface. By combining (7) and (5) we made the following substitution for the cavity potentials (lower half of right hand side vector): (8) j = LE
time = T time = T i

Uoo
Fig. 2: Potential Distribution on actual foil and cavity and its image.

where j=[n/2+1,n} and i=[0,n/2-1] Non-Linear Cavity Shape At each timestep, as the foil enters through the free surface, the above system of equations is solved. For numerical purposes at the first four timesteps, the cavity is assumed to be on the upper surface. After that, the alignment of the cavity shape is found iteratively, such that (4) is satisfied. The correction to the cavity height, H, at each iteration, measured normal to the current iteration of the cavity surface, is calculated at each panel edge by: qn H i + 1 = H i ---- s (9) q-
s i

Governing System of Equations The discretized form of Greens theorem is applied to each panel on the actual foil and cavity surface: i =

Sij n Dij j j j

(6)

where Sij and Dij are the influence coefficients due to a source and normal dipole respectively, including the effects of the free surface images. Rearranging (6) for the unknown quantities, dipole strength on the wetted surface and source strength on the cavity in the left hand side and the known quantities, source strength on the wetted surface and dipole strength on the cavity in the

where qn and qs are the normal and tangential velocities to the cavity respectively, and s is the arc length of the

panel. At each timestep, equations (7) and (9) are solved until convergence of the cavity geometry is attained, or qn=0. Usually, no more than 3 or 4 iterations per timestep are needed for convergence. Unsteady Foil Pressures Consider the flow with respect to the moving foil as shown in Fig. 1. After the initial entry, the flow at infinity and the free-surface move with speed U. The unsteady Bernoulli equation will be: 2 (10) + -- q + gy = C ( t ) t 2 where p is the pressure in the fluid domain, , is the density. The constant C(t) can be found by applying equation (10) at a point on the free surface, far from the intersection with the foil and cavity. 2 C ( t ) = p atm + -- U + gy FS (11) 2 where yFS is the y location of the free surface. y FS = U t (12) p+ Introducing the following non-dimensional quantities: U y = --------- ; t = t ------ ; y = (13) c U c c we finally obtain
2 p atm t q y p - 2- ------ + + -- ------ + ----- = --------- + -- + ----2 2 2 2 2U t Fr U 2 Fr U

converges slowly with number of panels, even though the difference in cavity shapes for increasing number of panels are not as large. This is primarily due to the fact that the panels arrangement at the leading edge is dictated by the constant time step approach, as shown in Fig. 4. However this does not affect the pressure forces on the foil

Cavity Convergence with Number of Panels


0.10

0.08

0.06

x/c
0.04

60 panels 120 panels 240 panels


0.00 0.02

-0.02 -0.8 -0.6 -0.4 -0.2 0.0

y/c

Fig. 3: Convergence of Cavity Shape with increasing number of panels.

(14)

where we also substituted for the Froude number. Being consistent with our infinite Froude number assumption, the hydrostatic terms in equations (10) and (11) drop out. So finally, the pressure equation to be applied on the fully wetted part of the foil surface and cavity surface is: qB 2 p B p atm C p = --------------------- = 1 ------ 2 (15) U - t 2 -- U 2 RESULTS Convergence with number of panels Fig. 3 shows the cavity geometry convergence with number of panels. It appears that the overall cavity shape
Fig. 4: Sensitivity of Cavity Shape to Leading Edge Paneling.

Convergence with iterations/timestep Fig. 5 shows the correction to the

cavity surface from each iteration per timestep. When the cavity growth is less than 10-5, it can be considered that convergence of cavity geometry is achieved. In other words, equation (4) is satisfied, the flow is tangential along the cavity surface.

Comparison Linear Theory vs. Non-Linear Theory for Cavity Shape


0.040 0.035 0.030 0.025

Linear [Yim] Non-Linear

Cavity Thickness Change

Cavity Convergence with Iterations/Timestep

x/c
0.020 0.015

5E-4

s/c = 1
Flat Plate Geometry
4E-4

0.010

=1

0.005 0.000

3E-4

1st iteration 2nd iteration 3rd iteration

s/c = 2/3

-0.8

-0.6

-0.4

-0.2

0.0

y/c

2E-4

Fig.6: Linear vs. non-linear cavity shapes


s/c = 1/3

1E-4

0E0 -0.8 -0.6 -0.4 -0.2 0.0

4.0
y/c

Comparison Linear Theory vs. Non-Linear Theory for Pressure Distribution on Flat Plate

3.5 3.0

Fig.5: Convergence of cavity shape with increasing number of iterations per timestep.

Comparison with Linear Theory B. Yim [4] and D.P. Wang [5] used conformal mapping techniques to address the linearized problem of surface piercing hydrofoils. Under this assumption, the cavity shape, and resulting flow field is symmetric with respect to the y-axis of the foil. Fig. 6 shows there is a very large difference from the non-linear cavity shape. The linear analytical calculations of the pressure distribution on a surface piercing flat plate are compared to the current method in Fig. 7. The current non-linear method compares very well with the linear analytical results at small angles of attack. With increasing angles of attack however, the effect of non-linearity are more apparent, and there is a significant difference with the linearized theory.

Cp / 2.5
2.0 1.5 1.0 0.5 0.0 -0.8 -0.6

Yim

Non-linear (=1o) Non-linear ( = 3 ) Non-linear (=7o)


o

y/c

-0.4

-0.2

0.0

Fig.7: Linear vs. non-linear pressure distributions on a flat plate.

CONCLUSIONS OF CURRENT STUDY A non-linear, time-marching, potential based boundary element method was developed to analyze the flow around a surface piercing 2D hydrofoil of arbitrary geometry. The method converges rapidly with increased number of iterations

per timestep, as well as with increased number of panels. The final cavity shape is sensitive to the cavity shape near the foil leading edge. Pressure distributions compare very well with linear analytical results for a flat plate at small angles of attack. FUTURE WORK Regridding At Each Timestep The sensitivity of the converged cavity shape to the leading edge cavity geometry, suggests that a finer panel arrangement locally at the leading edge might further improve the convergence. A cosine spacing panel arrangement will be implemented. It has to be cautioned that the cavity potentials will need to be splined and extrapolated to their new panel locations at each new timestep, since the panel spacing is no longer equal. Fully Submerged Analysis The current method can be extended for the fully submerged phase, where the foil is entirely below the free surface, and the cavity extends beyond the foil trailing edge, as shown in Fig. 8.
y

wetted

TE

= cavity

(16)
TE

Here to, the perturbation potential on the cavity behind the foil trailing edge is convected with the resulting flow. The system of equations to be solved in this phase, are an extension of equation (7) and can be symbolically represented by equation (17):
a e AB C b = E F G f c g

(17)

The left hand side of equation (17) consists of three parts. A and C are respectively the matrix including the influence coefficients on the lower cavity part and upper cavity part due to the source distribution everywhere else. The vectors a and c are the corresponding unknown source strengths, on the lower and upper cavity respectively. B is the matrix including the influence coefficients on the wetted part of the foil due to the dipole distributions and b is the vector including the unknown dipole strengths on the wetted part of the foil. On the right hand side of the equation the same analogy follows. E and F are respectively the matrices consisting of the influence coefficients on the lower and upper cavity part due to the dipole distribution everywhere else. The vectors e and f are the known dipole strengths on the lower and upper cavity part respectively. The matrix F consists of the influence coefficients due to the source distribution, and the vector f, is the known source strength on the wetted part of the foil. The numerical solution of this new system of equations is an extension to the results shown in the previous sections. AKNOWLEDGEMENTS

U oo

Fig. 8: Sketch of Fully Submerged Surface Piercing Foil

This research was supported by an international consortium of the following companies and research centers: DTMB, Daewoo, El Pardo MB, HSVA, IHI, KaMeWa. Mercury, OMC, Rolla, Volvo-Penta, Hyundai, Sulzer-Escher Wyss, Ulstein and Wrtsil Propulsion. REFERENCES

The additional boundary condition to be incorporated in the governing system of equations, is:

[1] B. Cox. Hydrofoil Theory for Vertical Water Entry. Ph.D Thesis, Department of Naval Architecture,Massachusetts Institute of Technology. May 1971. [2] S.A. Kinnas, S. Mishima, C. Savineau Application of Optimization Techniques to the Design of Cavitating Hydrofoils and Wings. International Symposium on Cavitation. Deauville, France. May 2-5, 1995. [3] S.A. Kinnas, N.E. Fine. Non-Linear Analysis of the Flow Around Partially or Super-Cavitating Hydrofoils by a Potential Based Panel Method. In Boundary Integral Methods-Theory and Applications, Proceedings of the IADEM-90 Symposium. Rome, Italy. October 15-19, 1990. [4] B. Yim. Linear Theory on Water Entry and Exit Problems of a Ventilating Thin Wedge. Journal of Ship Research, Vol. 18, No. 1. March 1974. [5] D.P. Wang. Water Entry and Exit of a Fully Ventilated Foil. Journal of Ship Research, Vol. 21, No. 1. March 1977.

You might also like