Anda di halaman 1dari 113

Assessment of slam induced loads on two

dimensional wedges and ship sections

Shan Wang

Thesis for obtaining the degree of Master in

Naval Architecture and Marine Engineering

Jury

President: Prof. Yordan Ivanov Garbatov

Supervisor: Prof. Carlos Antonio Pancada Guedes Soares

Member: Prof. Nuno Miguel Magalhaes Duque da Fonseca

December 2011
Acknowledgement

The work has been performed in the scope of the project EXTREME SEAS – Design for Ship

Safety in Extreme Seas, (www.mar.ist.utl.pt/extremeseas), which has been partially financed

by the EU under contract SCP8-GA-2009-234175.

Special thanks are due to Prof. Carlos Guedes Soares for providing the opportunity for me to

study and stay in Portugal, and all the help and guidance in research. I also want to express

my gratitude to Prof. Hanbing Luo, who has been sharing his knowledge and giving valuable

advises throughout the process.

The author is grateful to Prof Odd Faltinsen for having provided the detailed experimental

data used in this paper.

The author also wants to thank his family for constant understanding and support.

Lisbon,

Shan Wang

I
II
Abstract

The slam induced loads on two-dimensional bodies have been studied by applying an explicit

finite element code which is based on a multi-material arbitrary Lagrangian-Eulerian

formulation and penalty coupling method. This work focuses on the assessment of total

vertical slamming force, pressure distributions at different time instances and pressure

histories on the wetted surfaces of typical rigid bodies. Meanwhile, the simulation technique

involved in the two-dimensional slamming problem is discussed through related parameter

study.

This thesis first validates the fluid-structure interaction method of LS-DYNA by comparing the

slam induced loads on a two-dimensional wedge with deadrise angle of 30° with published

experimental results, and then discusses the relationship between the slamming loads and

the deadrise angle of the wedge through simulated calculations of wedges with different

deadrise angles which is varying from 10° to 80°, while analytical results are included as well

in terms of maximum pressure coefficient and pressure distribution. The slam induced loads

on a typical bow-flare section are evaluated, and also compared with available measured

values. Based on the water entry models of the wedge and the bow-flare section, parameters

study of the finite element modeling is carried out, to obtain acceptable predictions from

above numerical simulations. Importantly, the problem of fluid leakage through the structure

involving in the simulations of the water entry is discussed as well. Furthermore, the effects of

the roll angle on the slamming loads are investigated through the calculations for the bow-

flare section with different roll angles which include 0°, 4.8°, 9.8°, 14.7°, 20.3° and 28.3°.

Keywords: slam induced loads, multi-material arbitrary Lagrangian-Eulerian formulation,

penalty coupling, LS-DYNA, roll angle.

III
IV
Resumo

As cargas induzidas pelo impacto na água de corpos de duas dimensões foram estudadas

através da aplicação de um código de elementos finitos explícito que é baseado em uma

formulação Lagrangiana-Euleriana arbitrária multi-material e no método de acoplamento de

penalidades. Este trabalho tratada da avaliação da força vertical total, das distribuições de

pressão em pontos de tempo diferentes e das histórias de pressão sobre as superfícies

molhadas dos corpos rígidos típicos. Enquanto isso, a técnica de simulação envolvendo o

problema do impacto a duas dimensões é analisado por meio do estudo de parâmetros

relacionados.

Esta tese valida primeiro o método de interacção fluido-estrutura utilizando o LS-DYNA,

comparando as cargas induzidas pelo impacto de uma cunha bidimensional, com ângulo de

30° de inclinações publicadas com resultados experimentais, e, em seguida, discute a

relação entre as cargas de impacto e o ângulo de inclinação da cunha através de cálculos

simulados de cunhas com ângulos diferentes de inclinações que está varia de 10° a 80°,

enquanto os resultados analíticos se obtêm em termos de coeficiente de pressão máxima e

distribuição da pressão. As cargas de impacto induzidas numa secção de proa com

desenvolvimento típico são avaliadas, e comparadas com valores medidos que também

estão disponíveis. Com base nos modelos de entrada da cunha na água e da proa

desenvolvida, os parâmetros de estudo da modelação de elementos finitos é realizada, para

obter previsões aceitáveis de simulações numéricas. O problema de escoamento do fluido ao

longo da estrutura que envolve nas simulações da entrada de água também é discutido.

Além disso, os efeitos do ângulo de balanço nas cargas de impacto são investigados por

meio de cálculos para a secção de proa com diferentes ângulos de inclinação que incluem 0°,

4.8°, 9.8°, 14.7°, 20.3° e 28.3°.

Palavras-chave: cargas induzida, formulação Lagrangiana-Euleriana arbitrária multi-material,

acoplamento de penalidades, LS-DYNA, ângulo de balanço.

V
VI
Table of Contents

Acknowledgement .......................................................................................................................I

Abstract......................................................................................................................................III

Resumo ..................................................................................................................................... V

Table of Contents .................................................................................................................... VII

List of Figures ........................................................................................................................... XI

List of Tables ...........................................................................................................................XV

List of Nomenclatures ............................................................................................................XVII

Chapter 1 Introduction ...........................................................................................................1

1.1 Slamming problem ....................................................................................................1

1.2 Research background ...............................................................................................2

1.3 Motivation and objectives of this thesis.....................................................................4

Chapter 2 Theoretical backgrounds.......................................................................................7

2.1 Slamming theory of two-dimensional wedge ............................................................7

2.1.1 von Kármàn theory................................................................................................7

2.1.2 Wagner theory ......................................................................................................9

2.2 Analytical formulations ............................................................................................12

2.2.1 Maximum slamming pressure.............................................................................12

2.2.2 Vertical Slamming force ......................................................................................13

2.2.3 Time history of the vertical slamming force ........................................................15

Chapter 3 Explicit finite element method LS-DYNA ............................................................17

3.1 Introduction..............................................................................................................17

3.2 Numerical method for fluid-structure interaction .....................................................17

3.2.1 Lagrangian formulation .......................................................................................17

3.2.2 Eulerian formulation ............................................................................................17

3.2.3 Arbitrary Lagrangian-Eulerian (ALE) formulation................................................18

3.2.4 Governing equations...........................................................................................18

3.3 Numerical simulation of slam induced loads on two-dimensional structures..........20

VII
3.3.1 Lagrangian formulation for structures .................................................................21

3.3.2 Multi-material Eulerian formulation for fluids.......................................................21

3.3.3 Fluid-structure coupling algorithm.......................................................................21

3.3.4 Advection algorithm ............................................................................................22

3.3.5 Element type and material model .......................................................................22

3.3.6 Equation of state .................................................................................................23

3.3.7 Boundary conditions ...........................................................................................24

3.3.8 Analytical control.................................................................................................24

Chapter 4 Slamming load on the two-dimensional wedge during water entry ....................27

4.1 Description of test....................................................................................................27

4.2 Description of finite element model .........................................................................28

4.3 Verification of the explicit finite element method.....................................................30

4.4 Results and Discussion ...........................................................................................31

4.4.1 Vertical slamming force.......................................................................................31

4.4.2 Pressure distributions at different time instances ...............................................32

4.4.3 Pressure histories in different points...................................................................34

4.5 Parameter study ......................................................................................................37

4.5.1 Acceleration of gravity.........................................................................................37

4.5.2 Mesh density.......................................................................................................39

4.5.3 Penalty factor ......................................................................................................39

4.5.4 Time step ............................................................................................................40

4.6 Conclusions.............................................................................................................40

Chapter 5 Slamming load on the two-dimensional wedges with various deadrise angles
during water entry .....................................................................................................................43

5.1 Introduction..............................................................................................................43

5.2 Finite Element Model...............................................................................................43

5.3 Results and Discussion ...........................................................................................43

5.3.1 Maximum value of pressure coefficient ..............................................................43

5.3.2 Pressure distribution and free surface elevation ................................................44

5.3.3 Vertical slamming force.......................................................................................49

VIII
5.3.4 Results for different impact velocities .................................................................52

5.4 Conclusions.............................................................................................................54

Chapter 6 Slamming load on a two-dimensional bow-flare section during water entry.......57

6.1 Introduction..............................................................................................................57

6.2 Description of test....................................................................................................57

6.3 Description of finite element model .........................................................................58

6.4 Results and Discussion ...........................................................................................59

6.4.1 Gravitational acceleration of bow-flare section ...................................................59

6.4.2 Slamming force ...................................................................................................60

6.4.3 Pressure distribution at different time instances .................................................62

6.4.4 Pressure histories at different points ..................................................................63

6.4.5 Pressure variations during water entry ...............................................................65

6.4.6 Free surface elevation and pressure contours at different time instances .........67

6.5 Fluid leakage ...........................................................................................................68

6.6 Parameters study ....................................................................................................69

6.6.1 Mesh density.......................................................................................................69

6.6.2 Penalty factor ......................................................................................................69

6.6.3 The number of coupling points ...........................................................................70

6.6.4 Time step ............................................................................................................70

6.7 Conclusions.............................................................................................................71

Chapter 7 Slamming loads on a bow-flare section with various roll angles ........................73

7.1 Introduction..............................................................................................................73

7.2 Description of finite element model .........................................................................73

7.3 Results and Discussion ...........................................................................................74

7.3.1 Impact force ........................................................................................................74

7.3.2 Impact velocity ....................................................................................................77

7.3.3 Pressure history ..................................................................................................77

7.3.4 Pressure distribution ...........................................................................................81

7.3.5 Pressure variation ...............................................................................................82

7.3.6 Water surface elevation and pressure contour ...................................................83

IX
7.3.7 Discussions about the secondary impact ...........................................................84

7.3.8 Discussions about the ventilation and air pocket on the left side .......................85

7.4 Conclusions.............................................................................................................86

Chapter 8 Conclusions ........................................................................................................87

Bibliography ..............................................................................................................................91

X
List of Figures

Figure 1.1: Slamming problems of marine structures.................................................................1

Figure 2.1. Definition of coordinate system for 2-D symmetric wedge.......................................7

Figure 2.2. Sketch of Wagner’s theory .......................................................................................9

Figure 4.1. Geometry of the test section ..................................................................................27

Figure 4.2. 2-D water entry model for wedge with deadrise angle of 30°. ...............................28

Figure 4.3. Parts of the mesh for the model of wedge in LS-DYNA.........................................29

Figure 4.4. Predicted slamming forces on the wedge with a deadrise of 30° from LS-DYNA,
drop test and analytical solutions. ............................................................................................30

Figure 4.5. Pressure distributions on the symmetric wedge with a deadrise angle of 30º for
LS-DYNA at the time instance of t = 0.0158s . ..........................................................................31

Figure 4.6. Slamming forces on the symmetric wedge with a deadrise angle of 30° for
different mesh sizes..................................................................................................................32

Figure 4.7. Comparison of pressure distributions at different time instances for the models
with different mesh sizes. (a) t1 = 0.00435s ; (b) t 2 = 0.0158s ; (c) t 3 = 0.0202 s . .......................33

Figure 4.8. Pressure histories at different points for the models with different mesh sizes. (a)
Point P1, the depth from the keel to the position is 12.5mm; (b) Point P2, the depth from the
keel to the position is 37.5mm; (c) Point P3, the depth from the keel to the position is 62.5mm;
(d) Point P4, the depth from the keel to the position is 87.5mm; (e) Point P5, the depth from
the keel to the position is 131.8mm. .........................................................................................34

Figure 4.9. Free surface elevations and pressure contours of the model with mesh size of
1.25mm during water entry of the wedge. (a) t1 = 0.00435s ; (b) t 2 = 0.0158s ; (c) t 3 = 0.0202 s .
..................................................................................................................................................36

Figure 4.10. The predicted pressure distribution at different time instances on wedges with
deadrise angle 30ºduring the water entry by LS-DYNA. t = 0 corresponds to that the keel
touches the water surface. (a) Pressure distributions on the wedge before flow separation. (b)
Pressure distributions on the wedge after flow separation.......................................................37

Figure 4.11. Comparison of slamming force between the model considering gravity and that
neglecting gravity......................................................................................................................37

XI
Figure 4.12. Comparison of pressure distributions at three time instances between the model
considering gravity and that neglecting gravity. (a) t1 = 0.00435s ; (b) t 2 = 0.0158s ; (c)
t 3 = 0.0202 s . .............................................................................................................................38

Figure 4.13. Slamming forces for the models with different values of penalty factors. ............39

Figure 4.14. Pressure distributions for the models with different values of penalty factor. (a)
t1 = 0.00435s ; (b) t 2 = 0.0158s . .................................................................................................39

Figure 5.1. Comparison of slamming pressure coefficient for rigid wedge with different
deadrise angles. .......................................................................................................................44

Figure 5.2. Pressure distributions on symmetric wedges with constant vertical velocity during
water entry. ...............................................................................................................................46

Figure 5.3. Predicted pressure distributions on wedges with constant velocity when the
deadrise angle is 30°. (a) before flow separation; (b) after flow separation .............................47

Figure 5.4. Predicted pressure distribution on wedges with constant velocity when the
deadrise angle is 60°. (a) before flow separation; (b) after flow separation. ............................47

Figure 5.5. Predicted water jet flows and pressure contours in water by LS-DYNA for the
wedge with deadrise angle 30º. (a) t=0.00608, (b) t=0.0165, (c) t=0.0190. t=0 corresponds to
that the keel touches the water surface....................................................................................48

Figure 5.6. Predicted water jet flows and pressure contours in water by LS-DYNA for the
wedge with deadrise angle 60º. (a) t = 0.00608s , (b) t = 0.019067 s , (c) t = 0.025967 s . t=0
corresponds to that the keel touches the water surface...........................................................49

Figure 5.7. Maximum values of slamming force for symmetric wedges with different deadrise
angles. ......................................................................................................................................50

Figure 5.8. Comparison of the predictions of non-dimensional vertical force ..........................50

Figure 5.9. Comparison of the predictions of coefficient of vertical slamming force ................51

Figure 5.10. Predicted time histories of the vertical slamming force when the deadrise angle is
15°. ...........................................................................................................................................52

Figure 5.11. Predicted time histories of the vertical slamming force when the deadrise angle is
30°. ...........................................................................................................................................52

Figure 5.12. Predicted time histories of the vertical slamming force when the deadrise angle is
45°. ...........................................................................................................................................52

Figure 5.13. Maximum values of pressures for different impact velocity when the deadrise
angle is 15°. ..............................................................................................................................53

Figure 5.14. Maximum values of pressures for different impact velocity when the deadrise
angle is 30°. ..............................................................................................................................53

XII
Figure 5.15. Maximum values of pressures for different impact velocity when the deadrise
angle is 60°. ..............................................................................................................................53

Figure 6.1. Locations of gauges in bow flare section (Zhao et al. 1996)..................................57

Figure 6.2. Water entry model of bow flare section..................................................................58

Figure 6.3. Meshing of the water entry model of 2D bow-flare section. ...................................59

Figure 6.4. Comparison of vertical slamming force on a symmetric bow-flare section between
the model considering gravity and that neglecting gravity........................................................60

Figure 6.5. Comparison of experimental slamming forces and predicted results on bow-flare
section for different mesh sizes. ...............................................................................................61

Figure 6.6. The pressure distributions at three time instances during water entry of a bow-flare
section. (a) t = 0.06 s ; (b) t = 0.07 s ; (c) t = 0.08s ......................................................................62

Figure 6.7 Comparison of pressure distributions between theoretical methods and LS-DYNA.
(a(a) t = 0.06 s ; (b) t = 0.07 s ; (c) t = 0.08s ;(d) pressure distribution for the bow-flare section
with constant impact velocity. ...................................................................................................64

Figure 6.8. Pressure histories at different points on the surface of bow-flare section. (a) P1;
(b) P2; (c) P3; (d) P4.................................................................................................................64

Figure 6.9. Pressure distributions at different time instance for the bow-flare section mode
with mesh size of 2.5mm. .........................................................................................................66

Figure 6.10. Pressure distributions at different time instance for the bow-flare section model
with mesh size of 2mm. ............................................................................................................66

Figure 6.11. Free surface elevation and pressure contours at different time instances during
water entry by LS-DYNA. (a) t = 0.04 s ; (b) t = 0.042 s ; (c) t = 0.05s ; (d) t = 0.06 s , (e) t = 0.07 s ,
(f) t = 0.08s . t = 0 corresponds to that the keel touches the water surface. .............................67

Figure 6.12. Fluid leakage through the structure (a) Fluid leakage (b) leakage prevented .....69

Figure 6.13. Comparison of slamming force between the simulation with leakage and that
without leakage.........................................................................................................................69

Figure 6.14. Predicted slamming force for the models with different penalty factors...............70

Figure 6.15. Predicted pressure distributions for the models with different penalty factors. (a)
t = 0.06 s ; (b) t = 0.07 s ; .............................................................................................................70

Figure 7.1. Model setup of the asymmetric water entry of a two-dimensional bow-flare section
with constant roll angle θ. .........................................................................................................74

Figure 7.2. Part of the meshing of the water entry model for the bow-flare section with roll
angle 9.8°..................................................................................................................................74

Figure 7.3. Vertical impact force for various roll angles ...........................................................75

XIII
Figure 7.4. Vertical impact force for different impact velocity. (a) θ = 0 o ; (b) θ = 9.8 o ; (c)

θ = 28.3 o . ..................................................................................................................................76

Figure 7.5. Comparison of impact velocity for various roll angles. (a) θ = 9.8 o , initial velocity

v = −0.61m / s ; (b) θ = 20.3 o , initial velocity v = −0.75m / s ........................................................77

Figure 7.6. Pressure histories on the points P1-P4 on the right side of the bow-flare section
for different roll angles. (1 bar=105 N m-2) ................................................................................79

Figure 7.7. Pressure histories on the points P1-P4 on the left side of the bow-flare section for
different roll angles. (1 bar=105 N m-2) .....................................................................................80

Figure 7.8. Pressure distributions for different roll angles. (1 bar=105 N m-2) ..........................81
5
Figure 7.9. Pressure variations on a bow-flare section with different roll angles. (1 bar=10 N
m-2)............................................................................................................................................82

Figure 7.10. Free surface elevation and pressure contour for different roll angles..................84

Figure 7.11. Pressure contours at the time instance when the secondary impact at the left part

happens for different roll angles. (a) θ = 4.8 o , v = −0.57 m / s , (b) θ = 9.8 o , v = −0.61m / s ,(c)

θ = 14.7 o , v = −0.61m / s ; (d) θ = 20.3 o , v = −0.75m / s . ................................................................85

XIV
List of Tables

Table 3.1. Parameter for equation of state of water .................................................................23

Table 3.2. Parameter for equation of state of water ................................................................24

Table 4.1. Primary data of the tests for symmetric wedges .....................................................27

Table 4.2. The main data for three models of a wedge with different mesh densities. ............29

Table 6.1. The main data of the test sections ..........................................................................58

Table 6.2. The main data for three models of a wedge with different mesh densities. ............59

Table 6.3. Predicted peak pressures by using LS-DYNA, FLUENT and Aarsnes (1996)’s test
..................................................................................................................................................65

XV
XVI
List of Nomenclatures

β deadrise angle of wedge

B half-width of wedge

L(t ) wetted half-width

l wetted half-width in calm water

x horizontal coordinate of the bottom surface of the 2-D body

Y vertical coordinate of the bottom surface of the 2-D body

Yk vertical coordinate of the keel of the wedge

Yd under water penetration

V0 initial velocity of wedge

V impact velocity of wedge during water entry

M total mass of the wedge

Ma added mass of the wedge

F vertical slamming force

a acceleration of wedge

φ velocity potential in 2-D flow filed

ν drop velocity of body

n unit normal vector to the body surface

ρ density of fluid

σ half infinite fluid field

S boundaries of the entire fluid field

m added mass

P0 atmospheric pressure

η ( x, t ) variation of free surface

P pressure on the body surface

Pmax maximum pressure on the body surface

XVII
ηb variation of the bottom surface

βb shape curve of wedge

Cp pressure coefficient

CF vertical force coefficient

γ the splash-up coefficient which measure the rise of the water surface

C P max maximum value of pressure coefficient

k pressure coefficient for theorem of impulse

k1 parameter obtaining from a large amount of experiments

g ( x) shape function which depends on the geometry of the section

f (t ) parameter of time variable

d 1/10 of the design draft

Td time period

XVIII
Chapter 1 Introduction

1.1 Slamming problem


When a ship travels in rough seas, it will impact water because large vertical relative motions
between the ship and the wave surface. Impulse loads with high pressure peaks occur during
impact between a body and water. This is often called ‘slamming’ and occurs for instance
when a ship bottom hits the water with a high velocity. As shown in Figure 1.1, the probability
of slamming is highest on the fore part of a ship where the relative vertical velocity between
the ship and the waves is largest. During and immediately after slamming, there is a period of
localized responses when the hull plate will respond immediately, because of direct contact
with the slamming load. The keel, floors and nearby frames structures will react to the local
responses. Even it leads to global responses. Furthermore, slamming can happen on the
underside of the deck between the two hulls for a catamaran, and it occurs when breaking
waves hit the column of a platform as well. The slamming pressure has the complex nature of
impulse on time scale, moving rapidly on structural shell and unevenly distributed over the
impacted surface.

Figure 1.1: Slamming problems of marine structures

Different physical effects occur during slamming. An air cushion may be formed between the
body and the water when the local angle between the water surface and the body surface is

1
small at the poison of impact, which affects the interaction between air and water. The large
slam induced load can lead to great local dynamic hydroelastic effects. The relative
importance of hydroelasticity was studied by Faltinsen (1999), who presented that the
importance of hydroelasticity for the local slamming-induced maximum stresses increased
with decreasing deadrise angle and increasing impact velocity and natural period, and that
the deadrise angle has to be quite small for hydroelasticity to matter. The vibrations of the
body cause cavitations and ventilation.

In all, the impulsive pressure loads induced by slamming will lead to damage on the hull of
marine structure, and affect the structural responses. Many ships have reported local
structural damage due to the slamming loads, especially in heading waves with high forward
speed. For example, the tragedy of MV Estonia in the Baltic Sea on 28 September 1994, one
of the deadliest marine disasters of 20th century, was initialized by the break of the bow door
due to the severe slamming, which the Ro/Ro ferry experienced. The maximum slamming
pressure on a body in its lifetime is of great important for designing and improving the
structures. An underestimated pressure might lead to damage of the structures under harsh
condition, while an overestimated one might induce additional weight and low performance.
On the other hand, the impulse characteristic of a slam induced load is much more related to
the structural damage.

Therefore, assessment of slamming load is of great significances to study the safety and
movement responses for marine structures.

1.2 Research background


The assessment of the slam induced load on structures has been researched for decades by
analytical, experimental and numerical methods.

The pioneering work on impact between a body and water was carried out by von Kármàn
(1929) who developed an asymptotic theory on the idealized problem of a two-dimensional
wedge impacting with a calm water surface which neglects the local uprise of the water.
However, this idealized theory based on momentum conservation underestimates the impact
load for wedges with small deadrise angle. Based on his work, Wagner (1932) proposed an
asymptotic solution for water entry of two-dimensional bodies with small local deadrise angles,
accounting for piled-up water on the wedge. The flow was divided into two fluid domains. The
inner flow domain contains a jet flow at the intersection between the body and the free
surface, while the body boundary condition and the dynamic free surface condition were
transformed to a horizontal line in the outer flow domain. For his theory, the pressure is
singular on the edge of the expanding plate when the deadrise angle is small. Much work has
been done by other researchers to further develop this theory. Dobrovol’skaya (1969) derived
an analytical solution by transferring the potential flow problem for the constant water entry of
a wedge into a self similar flow problem in complex plane, which took advantage of the
simplicity of the body geometry and is valid for any deadrise angle. However, such similarity

2
solution cannot be obtained for arbitrary bodies. Armand and Cointe (1987) and Howison et al.
(1991) introduced the effect of nonlinear jet flow into the intersection region between the body
and free surface by using asymptotic matching.

Zhao et al. (1993) generalized the work of Wagner (1932) and presented a numerical method
for studying water entry of a two-dimensional body of arbitrary cross-section which is a
nonlinear element method with a jet flow approximation. As a further development work, a
fully nonlinear numerical simulation method which includes flow separation from knuckles of a
body and an approximation solution which dose not include flow separation were presented
by Zhao and Faltinsen (1996) to predict slamming loads on two-dimensional sections.

Motivated by the work of Zhao and Faltinsen (1996), Mei et al. (1999) raised an analytical
solution for the general impact problem by adopting the conformal mapping technique. For
numerical confirmation, they developed a fully nonlinear simulation by using a Cauchy-
integral method with a matching jet solution near the intersection between water and the body.
This solution is valid for a wedge with arbitrary deadrise angles, however, the velocity of the
wedge was assumed constant. Yettou (2007) presented an analytical solution to symmetrical
water impact problems of a two-dimensional wedge by taking into account the effect of
velocity reduction of the solid body upon impact.

In the field of experimental research, several studies were carried out for prediction of
slamming loads on bodies during water entry. Chuang (1966, 1970) carried out the
experimental research on the water entry of rigid and elastic bodies which include rigid flat
and rigid wedges with deadrise angles of 1°, 3°, 6°, 10° and 15°, and obtained the relationship
between maximum pressure and deadrise angle. Ochi and Motter (1973) obtained the
slamming loads in terms of slamming pressure, the pressure distribution and the time
variation of the total slamming load by analyzing lots of test results. Recently, the drop tests
for a wedge with deadrise angle 30° and a bow-flare section were carried out by Zhao et al.
(1996) in MARINTEK. They concluded that the maximum value of pressure occurred at the
moment before flow separation. Aarsnes (1996) performed free drop tests of two ship
sections, i.e., one wedge section and one bow-flare section which are the same ones in the
tests of Zhao et al. (1996), aiming at investigating the pressure distributions and the impact
forces for different roll angles. Compared to the experimental results of him, Sun and
Faltinsen (2009) studied the two-dimensional water entry of a bow-flare section with a
constant roll angle, or heel angle by using a boundary element method. They found that, for
the ship studied, the vertical slamming force did not change much with the roll angle when the
roll angle is small, whereas the horizontal force obviously increased with the increasing roll
angle, and the high localized pressure in the flare area was occurred for the larger roll angle.

Recently, with the development of computing technology and capability, codes based on
explicit finite element method (FEM) began to be applied to predict local slamming loads.
Bereznitzki (2001) analyzed hydroelastic problems using the MS-Dytran code. Stenius et al.
(2006) studied modeling techniques for rigid wedge impact problems using the LS-DYNA

3
code. Several parameters that influence the convergence of simulation, such as mesh density
and contact stiffness were discussed. Aquelet et al. (2006) discussed the influence of penalty
factor on the damping effect. Luo et al. (2010) used MS-Dytran to study the impact of one
stiffened panel, showing that an explicit code has the potential to predict water entry impact
problems for both rigid and elastic structures. Alexandru et al. (2007) carried out a
comparison of simulation of 2D slamming problems, using Boundary Element Methods
(BEMs), Computational Fluid Dynamics (FLUENT and FLOW-3D codes), Smooth Particle
Hydrodynamics (SPH), and Explicit FEM (LS-DYNA code). Fairly good agreements were
achieved. Some differences in time domain after the peak value for slamming pressure on
one wedge with dead rise angle 25° were observed. There were even much differences
between the pressures predicted with each other on an impacting rigid bow section.
Nevertheless, the results obtained are encouraging overall, but more validation work still
needs to be carried out on tuning parameters in order to get better numerical results.

1.3 Motivation and objectives of this thesis


As known, assessment of slam induced loads is a fluid-structure interaction problem, of which
the full governing equations are too complex to be solved analytically. As to the theoretical
and analytical methods, some simplifications on boundary conditions or shape of section are
applied to obtain general solutions, which result in some errors. Furthermore, one difficulty in
the fully nonlinear numerical approach is the modeling of the flow in the water jet which is a
very thin layer of fluid attached to the body surface and is developed during the impact.
Because the pressure in this domain is very small and almost constant, a solution for the jet is
to cut if off from the flow as mentioned by Zhao and Faltinsen (1993). This means that the
pressure of the thin jet is taken out of the calculation. Hence, this work aims on describing the
water entry problem by a numerical simulation method which is induced into the field of
slamming and includes the modeling of structures, fluids and the coupling between the
interacting parts.

In this work, a fluid-structure coupling algorithm is applied, which was developed and
implemented in LS-DYNA, an explicit finite element code for general fluid-structure interaction
problems. Unlike existing algorithms that couple two separates codes, a CFD and a structure
code, the fluid-structure interaction algorithm is fully coupled.

By applying a commercial explicit finite element code, the slam induced loads on the wetted
surfaces of typical two-dimensional bodies are evaluated in this thesis. Combining with
experimental results and analytical values, the simulated slamming loads are discussed, and
the numerical modeling technique is investigated though parameters study.

The current situation of slamming problem and developments achieved during past years are
presented firstly in Chapter 1. Chapter 2 describes the theoretical background on slamming
problem of two-dimensional bodies, such as the pioneering work of von Kármàn (1929) and
Wagner (1932), and presents related analytical formulations which will be used in this thesis.

4
In Chapter 3, explicit finite element method is introduced for the field of fluid-structure
interaction, followed by the description of the numerical simulation of slam induced loads on
two-dimensional structure.

The slam induced loads on the wetted surface of a rigid wedge with deadrise angle of 30°
subjected to water impact are evaluated in Chapter 4, and the simulated vertical slamming
force, pressure distribution and pressure histories are compared with the experimental results
from published drop tests which validate the explicit finite element method. Furthermore, a
parameter study about the simulation is carried out.

When the impact velocity is assumed constant, the relationship between slamming loads and
the deadrise angle of wedge is discussed through the simulations of wedges with varying
deadrise angles from 10°s to 80° in Chapter 5, in which theoretical and analytical values are
presented as well. The relationship between the slamming pressure and impact velocity is
presented as well in this part.

In Chapter 6, the slamming loads on a two-dimensional bow-flare section are calculated, and
then the simulated predictions are compared also with the results from drop tests. Similarly, a
parameter study about the numerical simulation is carried out in this part. Furthermore, fluid
leakage through structure is studied due to the complexity of the section shape. The roll angle
is taken into account in Chapter 7 later. The predicted total slamming force, impact velocity
and pressure histories on points on a bow-flare section with various roll angles are compared
with published results of test and numerical calculations. In addition, the simulated pressure
distributions on the two-dimensional body and free surface elevations are presented as well,
from which the secondary impact, ventilation and air pocket on the left side of the section are
analyzed.

Finally, Chapter 8 draws the conclusions from the work mentioned above.

5
6
Chapter 2 Theoretical backgrounds

2.1 Slamming theory of two-dimensional wedge


2.1.1 von Kármàn theory

von Kármàn (1929) studied initially on the slamming theory in order to analyze the slam
induced loads on the bottoms of seaplanes during landing, which was simplified as the
slamming problem of a wedge. He proposed the first theoretical solution for the wedge with a
deadrise angle around 20°s by applying the linear boundary conditions, and the local uprise of
the water was accounted for.

Figure 2.1. Definition of coordinate system for 2-D symmetric wedge.

As seen in Figure 2.1, the deadrise angle of the wedge is β which is defined as the angle

between horizontal line and the surface of the body, the wetted half-width in calm water is
L(t ) , the actual wetted half-width is l , and the underwater penetration is Yd . Assume the

initial velocity of wedge is V0 , and the gravity, buoyancy and other external forces are

neglected. According to conservation of momentum, he obtained,

MV0 = ( M + M a )V (2.1)

Therefore,

MV0
V = (2.2)
M + Ma

where, M a is the unit added mass of the wedge, and M is the total mass of the wedge.

Thus,

7

dV V0 M a
F = Ma = − M = (2.3)
dt M
(1 + a ) 2
M

The equation shows that the impact force is related to added mass and the time derivative of
that. Therefore, it can be concluded that the calculation of the hydrodynamic forces on the
bottom of wedge depends on the solution of the added mass which is a potential flow problem.
Furthermore, the pressure distribution on the wedge cannot be obtained from conservation of
momentum. Therefore, fluid mechanics was applied to the slamming problem.

Assume a two-dimensional body entering into water surface vertically, which means that the
flow field is symmetrical. The fluid is inviscid, incompressible, and the water surface is calm
initially. Therefore, the flow field satisfies the Laplace equation which is described as

∇ 2φ = 0 (2.4)

where, φ = φ ( x, y, t ) is the velocity potential. The associated boundary condition on free


surface is

Dφ 1  ∂ 2φ ∂ 2φ 
=  +  (2.5)
Dt 2  ∂x 2 ∂y 2 

The boundary condition on the wetted body surface is

(∇φ − v) ⋅ n = 0 (2.6)

where v is the downward vertically velocity of the body, and n is the unit normal vector to
the body surface and positive direction into the fluid domain.

The initial condition is

φ = φ ( x, y , t ) = 0 on t = 0 (2.7)

Once the velocity potential is obtained, the added mass can be given as

ρ 2 −ρ ∂φ
m=
V 2 ∫∫σ [gradφ ] dσ = V ∫ φ ∂n dS
2 S
(2.8)

where, ρ is the density of fluid, σ means the half infinite fluid field, and S is the boundaries

of the entire fluid field.

The pressure distribution on the body can be expressed as:

1  ∂φ   ∂φ  
2 2
p − p0 ∂φ
=− − gη −   +    (2.9)
ρ ∂t 2  ∂x   ∂y  
 

where, p 0 is atmospheric pressure, η ( x, t ) represents the variation of free surface.

8
Due to the nonlinear boundary condition of the free surface and the varying shape of that, it is
difficult to solve such an unsteady potential flow problem. von Kármàn (1929) considered that
the impact occurred in a very short time period, neglected the second-order differential part of
velocity, and considered the free surface at any time instance as the original calm water
surface.

Therefore,

1
Ma = ρπl 2 (2.10)
2

where, l is the horizontal size of the wedge immersed in water, so the total impact force is
given by:

V02 ρπl
F= 3
(2.11)
 ρπl 2 
tan β 1 +
 2M 

The maximum value of pressure is

ρ
Pmax = V0 2π cot β (2.12)
2

It is easy to show that, once the deadrise angle β is close to zero, the impact force will be

infinite. The formulation above is not applicable when the deadrise angle is small.

2.1.2 Wagner theory

Based on the work of von Kármàn, Wagner (1932) developed the theoretical method on the
water entry of wedge by taking the free surface elevation into consideration. Assume the
following:

a. The fluid is inviscid and incompressible.

b. The acceleration of fluid is much larger than that of gravity, so the gravity can be neglected.

Figure 2.2. Sketch of Wagner’s theory

9
c. The draft of wedge is much smaller than the wetted width.

Therefore, the water entry model can be simplified as Figure 2.2.

The flow field can be described as

∂ 2φ ∂ 2φ
2
+ = 0, y < 0 (2.13)
∂ x ∂2 y

The free surface boundary condition is

φ = 0, x > L(t ), y = 0 (2.14)

The boundary condition on the wetted body surface is

∂φ
= −V , x ≤ L(t ), y = 0 (2.15)
∂y

The velocity potential on the surface of the body can be given as

φ = −V ( L2 − x 2 )1 / 2 , x ≤ L(t ) (2.16)

The pressure distribution can be obtained from Bernoulli equation

p − p0 ∂φ 1 → →
=− − V ⋅ V − gy + C (2.17)
ρ ∂t 2

where, V is a constant which can be neglected.

2 2
→ →  ∂φ   ∂φ 
V ⋅ V =   +   (2.18)
 ∂x   ∂y 

This second-order part needs to be keep, in order to get the pressure distribution on the
surface of the body. The vertical distance between the keel of the wedge to the calm surface
t


is V (t )dτ , so
0

2
∂φ 1  ∂φ  1 2 x2
p − p0
ρ
=− −   =V 2
∂t 2  ∂x 
L dL
( L − x 2 )1 / 2 dt
+ L2
− x 2 1 / 2 dV
(−
dt 2
V 2
L − x2
) (2.19)

From the boundary conditions

∂φ Vx
= , when y = 0, x ≤ L(t ) (2.20)
∂y x − L2
2
( ) 1/ 2

The free surface elevation is given by

t
Vx
η (t ) = ∫ (x
0
2
− L2 )
1/ 2
dt (2.21)

10
Therefore, at one time instance when the fluid particle comes up to the surface of the wedge
in the position x , the vertical coordinate in the position x of free surface is equal to that of the
body surface, namely η (t ) = η b , assume that

dt
u ( L) = V (2.22)
dL

Then equation (2.21) becomes

u ( L) x
η (t ) = ηb ( x) = ∫ (x 2
− c 2 (t ) )
1/ 2
dL (2.23)

If the bottom of the body is given by a series as

ηb = a1 x + a2 x 2 + a3 x3 + a4 x 4 + a5 x5 + ⋅ ⋅ ⋅ (2.24)

Assume that

a1 a2 a a a
u = u (c) = + L + 3 L2 + 4 L3 + 5 L4 + ⋅ ⋅ ⋅ (2.25)
A0 A1 A2 A3 A4

Therefore, equation (2.24) becomes

2
u ( L) = a1 + a 2 L + a 2 L2 + a 3 L3 + a 4 L4 + ⋅ ⋅ ⋅ (2.26)
π

where,

2 2 ⋅ 4 ⋅ 6 ⋅ ⋅ ⋅ ⋅n n
 π ⋅ 1 ⋅ 3 ⋅ 5 ⋅ ⋅ ⋅ ⋅(n − 1) , 2 = int eger
a n = a n +1  (2.27)
1 ⋅ 3 ⋅ 5 ⋅ ⋅ ⋅ ⋅n n
 , ≠ int eger
 2 ⋅ 4 ⋅ 6 ⋅ ⋅ ⋅ ⋅( n − 1) 2

Therefore,

t
2 1 1 1

y = V (τ )dτ =
0
π
a1 L +
2
a 2 L2 + a 2 L3 + ⋅ ⋅ ⋅
3 n +1
a n Ln +1 (2.28)

For a wedge, with the deadrise angle β , the shape curve is

β b = x tan β (2.29)

So, the wetted width at t time instance is

t
π
L(t ) =
2 tan β ∫ V (τ )dτ
0
(2.30)

Then

dL π
= V (2.31)
dt 2 tan β

11
When the impact velocity is constant, the second term of equation (2.19) equals to zero. By
substituting equation (2.31) in equation (2.19), we can get

p − p0 π L x2
CP = = 2 2 1 / 2
− 2 (2.32)
1
ρV 2 tan β ( L − x ) L − x2
2

2.2 Analytical formulations


2.2.1 Maximum slamming pressure

The definition of the maximum value of the pressure coefficient is:

pmax − p 0
Cp = (2.33)
1
ρV 2
2

Assume that the wedge enters into water with a constant velocity, the peak value of pressure
coefficient is obtained by Wagner (1932):

π2
C p max = 1 + (2.34)
4 tan 2 β

The theorem of impulse assumes that the maximum impact pressure is proportional to the
quadratic impact velocity, as:

2
1 Dwrel ( x, t )
Pmax ( x) = ρk (2.35)
2 Dt

where, ρ is the density of fluid, and k represents the pressure coefficient

Dwrel ( x, t )
is the impact velocity at the time instance t , which can be obtained from
Dt
experiments, theoretical solution or empirical formulation.

The k value can be evaluated for any section by Ochi and Motter (1973), which is expressed
as following

k = exp(1.377 + 2.419a1 − 0.873a3 + 9.624a5 ) (2.36)

where, a1, a3 , a5 were calculated based on mapping of the section for any conventional hull

shape below 1/10 of the design draft waterline into a circle.

Stavovy and Chuang (1976) proposed an empirical formulation for pressure coefficient k as
following, which was obtained through a large amount of experiments.

k = 288k1 / cos 4 ξ ( x, t ) (2.37)

where cos ξ ( x, t ) is the effective impact angle, and k1 is expressed using a series of

polynomials that fit experimental results as:

12
 = 0.37ξ / 2.2 + 0.5,0 ≤ ξ < 2.2o...........................................
 2 3
 = 2.1820894 − 0.9451815ξ + 0.203754ξ − 0.0233896ξ
 + 0.0013578ξ 4 − 0.00003132ξ 5 ,2.2o ≤ ξ < 11o
k1 =  2 3 (2.38)
 = 4.748742 − 1.3450284ξ + 0.1576516ξ − 0.0092976ξ
 + 0.0002735ξ 4 − 0.00000319864ξ 5 ,11o ≤ ξ < 20o

= (1 + 2.4674 / tan 2 ξ )0.76856471 / 288,20o ≤ ξ ..................

2.2.2 Vertical Slamming force

By integrating equation (2.19) along the surface of the body, the vertical slamming force can
be given as

L L L
dL dx dV dL π dV
F= ∫
−L
pdx = ρVL
dt ∫
−L
2
L −x 2

dt ∫
−L
L2 − x 2 dx = ρπVL
dt
+ ρ L2
2 dt
(2.39)

For a wedge,

t
π
L(t ) =
2 tan β ∫ V (τ )dτ
0
(2.40)

The vertical force coefficient for a unit length wedge with constant velocity during water entry
is

F
CF = 3
tan 2 β (2.41)
ρV t

So, for Wagner’s model, the vertical force coefficient is equal to 3π / 4 .

Due to neglecting the free surface elevation, the vertical force coefficient of von Kármàn’s
model is

F
CF = tan 2 β = π (2.42)
ρV 3t

The first term in the right of equation (2.39) is the item of impact, and ρπL2 / 2 means the two-

dimensional added mass in heave, the equation (2.39) can also be expressed as:

d
F= (a33V ) = a33 dV + V da33 (2.43)
dt dt dt

where, Vda 33 / dt is the impact force obtained from the von Kármàn (1929) solution which is

based on a flat plate solution, similar as Wagner (1932) did. The other von Kármàn (1929)
solution is obtained by using the exact body boundary condition and the principle of
conservation of momentum. This is referred to as the von Kármàn-momentum solution. The
impact force is given by:

d
F= ( A33V ) = A33 dV + V dA33 = A33 dV + V 2 dA33 (2.44)
dt dt dt dt dz

13
where, A33 is the infinite frequency added mass in heave as a function of submergence. V is

the drop velocity.

Based on equation (2.41), Zhao and Faltinsen (1996) compared the maximum pressures on a
wedge with different deadrise angles by using the methods of von Kármàn (1929), Wagner
(1932), von Kármàn-moment, the simplified solution and the similarity solution.

After obtaining the maximum impact pressure from equation (2.35), the sectional impact force
can be determined by:

FSlam ( x, t ) = Pmax ( x) g ( x) f (t ) (2.45)

where, Pman (x) is the maximum value of slamming pressure, g (x) represents shape function

which depends on the geometry of the section and f (t ) is the parameter of time variable.

Ochi and Motter (1973) assumed that the slamming pressure had a linear distribution
vertically with the maximum value at the bottom and the zero value at one tenth of the design
draft. The total vertical slamming force was obtained by integrating along the section

π /2
y (θ ) A'
F = 2 Pmax ∫
0
d
cos α (θ )dθ = Pmax
d
(2.46)

Substituting the maximum pressure in equation (2.35), it becomes

A′
F= ρkv 2 (2.47)
2d

In the Stavovy and Chuang’s (1976) work, the pressure in a wedge is assumed constant and
so the vertical force is obtained by using:

b
F = pmaxb = ρkv 2 (2.48)
2

where, b is the width of the section.

In the Zhao and Faltinsen (1993)’s work, the pressure distribution has a sharp value in a z -
position, according to the asymptotic theory, equal to (0.5π −1)z′ . Based on the direct

integration of the pressure, they gave the values of the total non-dimensional vertical force:

F
FAD = (2.49)
ρv 2 z′

When the impact velocity is constant, z ′ is equal to vt .

Based on equation (2.49), Ramos and Guedes Soares (1998) presented the non-dimensional
vertical force using the solutions of Ochi and Motter (1973), Stravovy and Chuang (1976) and
derivative of added mass.

14
2.2.3 Time history of the vertical slamming force

For the time history of the vertical slamming force, Ochi and Motter (1973) proposed a simple
form to evaluate the time variation of the vertical slamming force. They assumed that the time
history had a triangular varying from 0 to the maximum value and to 0 again during a
period Td . According to experimental work and Froude scale law, the relationship between the

period and the ship length was given as

Td = 0.00794 L (2.50)

Based on their work, Kawakami et al. (1977) proposed an exponential function to describe the
time history of vertical slamming force

et −t T 0
f (t ) = e (2.51)
T0

where,

T0 = 0.00088 L (2.52)

Mei et al. (1999) raised an analytical solution for the general impact problem by adopting the
assumptions of Zhao and Faltinsen (1996) and the conformal mapping technique.

Based on the momentum theory and ignoring gravity, the velocity of the wedge during impact
is given by equation (2.2). But the added mass of the wedge is calculated by:

M a = Ca ρY 2 , Ca = (1 − β / 2π )2 π / 2 (2.53)

ρV02Y
F (t ) = 2Caγ cot β (2.54)
(1 + M a / M )3

where γ depends on the deadrise angle β , and Y is given by:

ρCa γV02Y
Y3 +Y = (2.55)
3M (1 + M a / M )

15
16
Chapter 3 Explicit finite element method LS-DYNA

3.1 Introduction
LS-DYNA is a general finite element commercial code for analyzing a large range of practical
and complex problems, especially for the problems of large deformation and dynamic
responses of structures including impact, explosion, metal forming and structures coupled to
fluids as well. LS-DYNA has been widely used in related industrial fields since 1990s.

A large amount of element types are available, including four node tetrahedron and eight
node solid elements, two node beam elements, three and four node shell element, eight node
solid shell elements, truss elements, membrane elements, discrete elements and rigid bodies.
A variety of element formulations are available for each element type. Currently, LS-DYNA
can provide a wide range of material which corresponds to approximately one-hundred
constitutive models and ten state equations. Adaptive remeshing is available for shell
elements and is widely used in sheet metal stamping application.

The main solution methodology is based on explicit time integration which can be applied to
solve different kinds of nonlinear problems. Especially, it is suitable for analyzing impact,
explosion and fluid-structure interaction which are dynamic nonlinear problem. There are
three algorithms for explicit dynamic analysis in LS-DYNA: Lagrange algorithm, Euler
algorithm and ALE algorithm.

3.2 Numerical method for fluid-structure interaction


3.2.1 Lagrangian formulation

The Lagrangian algorithm is widely used for analyzing stress and stain variation of solid
structure, which is based on physical coordinate system. The volume of fluid changes in
shape, while the total mass remains constant. The deformation of the finite element mesh and
the corresponding structure described by the Lagrangian algorithm are exactly the same. In
other words, the mesh of the computational domain moves with the particle fluid velocity,
therefore, this formulation is preferred for moderate large deformation problems.

Although the Lagrangian formulation allows easy tracking of free surfaces and interfaces
between different materials, its weakness is its inability to follow large distortions of the
computational domain.

3.2.2 Eulerian formulation

In the Eulerian formulation which is based on space coordinate system, the finite element
mesh is dependent on the structure to be analyzed, and the mesh does not move in space

17
during the analysis, while the materials move through the fixed mesh in space. Therefore, this
formulation is preferred for problems with large distortion, such as fluid-structure interaction.

Obviously, the size and the space coordinate of the mesh are not changing which means that
the numerical precision is the same during every iterative process. It is difficult to capture the
material boundary, which leads to costly computation. Since the computational domain is
fixed, the Eulerian description has the advantage of preserving the mesh regularity.

3.2.3 Arbitrary Lagrangian-Eulerian (ALE) formulation

Arbitrary Lagrangian-Eulerian (ALE) formulation was initially applied in finite difference


method for numerical simulation of hydrodynamics. It has both the advantages of Lagrangian
and Eulerian formulations. The structural movement on boundary is calculated by Lagrangian
formulation which can effectively capture the moving boundaries, while the finite element
mesh algorithm is based on an Eulerian mesh. Actually, although the interior mesh is
independent with solid material, it is not exactly the same to Eulerian mesh. The mesh in ALE
formulation is capable of updating the position according to related parameters to prevent
large distortions. Compared to pure Eulerian methods, it is also better suited for moving
boundaries and large volume changes of the computational domain.

The ALE algorithm consists of a classical Lagrangian step in which the mesh moves along
with the modeled material, a rezone step in which the mesh is modified to preserve good
quality through the computation, and a remapping step in which the solution is conservatively
transferred from the old mesh to the new, rezoned one.

3.2.4 Governing equations

The governing equation for incompressible and unsteady Navier-Stokes fluid is described as

∂u
+ u∇u − 2u F ∇ε (u ) + ∇p = b (3.1)
∂t

∇⋅u = 0 (3.2)

where u is the flow velocity, p is the pressure of fluid, b means body force acting on the fluid

and ε (u ) represents the deviatoric stress tensor.

The boundary condition and initial condition are

σ = − pl + 2v F ε (u ) (3.3)

ε (u ) =
1
2
(∇u + (∇u )T ) (3.4)

In ALE (Arbitrary Lagrangian-Eulerian) formulation, a reference coordinate which is not the


Lagrangian coordinate and Eulerian coordinate is induced. The differential quotient for
material with respect to the reference coordinate is described as following equation.

18
∂f ( X i , t ) ∂f (xi , t ) ∂f (xi , t )
= + wi (3.5)
∂t ∂t ∂xi

where, X i is Lagrangian coordinate, xi is Eulerian coordinate, and wi is relative velocity.

Therefore, the Arbitrary Lagrangian-Eulerian formulation can be derived from the relation
between the time derivative of material and that of reference geometry configuration.

Assume that v represents the velocity of the material, and u means the velocity of the mesh.
In order to simplify the above equation, relative velocity w is induced, which is given
by w = v − u . Therefore, Arbitrary Lagrangian-Eulerian formulation can be obtained from
following conservation equations:

1) The mass conservation equation:

∂ρ ∂v ∂ρ
= − ρ i − wi (3.6)
∂t ∂xi ∂xi

2) The momentum conservation equation

The governing equation of fluid is Navier-Stokes equation which is described by ALE method:

∂v i ∂v
υ = σ ij , j + ρbi − ρwi i (3.7)
∂t ∂x j

The stress tensor is expressed by:

σ ij = − pδ ij + µ (vi , j + v j ,i ) (3.8)

The initial and boundary conditions are:

vi = U i0 on Γ1 domain (3.9)

σ ij n j = 0 on Γ2 domain (3.10)

while

Γ1 U Γ2 = Γ Γ1 I Γ2 = 0 (3.11)

where, Γ represents the whole boundary of computed field, while Γ1 and Γ2 means the parts

of Γ . ni represents the unit vector of boundary in outward normal direction, δ ij is

Kronecker δ function. Assume that the velocity field of the whole computed field at time t = 0 is
know,

v i ( x i ,0 ) = 0 (3.12)

3) The energy conservation equation

∂E ∂E
ρ = σ ij vi , j + ρbi vi − ρw j (3.13)
∂t ∂x j

19
The Euler equation is derived based on the assumptions that the velocity of reference
configuration is zero, and the relative velocity between the material and the reference
configuration is the velocity of the material. The terms of velocity in the equation (3.7) and the
equation (3.9) are known as convective terms which are used to calculate the transportation
volume that the material flows through the mesh. The additional items are the reason that the
numerical solution of the ALE equation is much more difficult than that of a Lagrange equation
in which the relative velocity is zero.

There are two approaches to solve the ALE equation, which are similar to the methods
applied to Euler’s viewpoint in hydrodynamics. The first method is solving fully coupled
equations using computational fluid mechanics, which can only govern singular material in
singular element. The second one was called detached operator method, of which the
calculation in each time step is separated into two parts. First, the Lagrange approach is
executed, when the mesh moves with material. During this process, the equilibrium equations
are:

∂vi
ρ = σ ij , j + ρbi (3.14)
∂t

∂E
ρ = σ ij vi , j + ρbi vi (3.15)
∂t

In the Lagrange process, there is no material flowing through element boundary, so the
calculation satisfies the mass conservation. Then the transportation volume, internal energy
and kinetic energy of materials that flow through the boundaries of element are calculated in
the second stage. It can be considered as remapping the meshes back to their initial or
arbitrary positions.

As to each node, the velocity and displacement are undated according to following equation:

(
u n +1 / 2 = u n −1 / 2 + ∆tM −1 Fext
n n
+ Fint ) (3.16)

x n +1 = x n −1 + ∆tu n +1 / 2 (3.17)

where, Fintn is vector of internal force, and Fext


n
is vector of external force. They are in relation
with body force and boundary conditions. M is diagonal matrix of mass.

3.3 Numerical simulation of slam induced loads on two-dimensional


structures
The commercial code LS-DYNA provides appropriate material models and state equations for
that problem, furthermore, advection method and coupling algorithm are available in an ALE
method.

In this section, the related algorithms and parameters involving in the simulation of slam
induced loads on two-dimensional structures are presented.

20
3.3.1 Lagrangian formulation for structures

Because the structures involved in this work have no large deformation, it is better to use
Lagrangian formulation which can easily track the surface of the mesh.

3.3.2 Multi-material Eulerian formulation for fluids

The ALE solver involves a Lagrangian step, where the mesh is allowed to move and a second
step that moves the element state variables back onto reference mesh. The multi-material
Eulerian formulation is a specific ALE case where the reference, or background, mesh
velocity is zero. For this formulation, the material flows through a mesh, fixed in space and
each element is allowed to contain a mixture of different materials.

For slamming problem in this work, the fluid is subjected to large deformations, Lagrangian or
ALE meshes are strongly distorted which jeopardizes the simulation because distorted
elements have low accuracy and their stable time step sizes are small for explicit time
integration algorithm. For single material or multi-material Eulerian formulation, the mesh is
fixed in space and materials flow through the mesh using an advection scheme to update fluid
velocity and history variables.

Therefore, the Euler Lagrange coupling using Eulerian multi-material formulation for the fluid
is more suitable for solving slamming problems and more generally, fluid-structure interaction
problems.

3.3.3 Fluid-structure coupling algorithm

Euler-Lagrange coupling allows us to treat the impact problems involving fluids because this
coupling treats the interaction between a Lagrangian formulation modeling the structure and
an Eulerian formulation modeling the fluid. In addition, an algorithm is needed to make the
structure and fluid interacted.

The penalty coupling algorithm was implemented in the 950 version of LS-DYNA. It is defined
to preserve the total energy of the system as well as possible. The constrained based method
consumes some kinetic energy, which is a problem in many applications. The basic idea of
the penalty formulation is to track the relative displacements between the coupled Lagrangian
nodes and the fluid. The coupling forces are defined to be proportional to these
displacements.

In an explicit time integration problem, the main part of the procedure in the time step is the
calculation of the nodal forces. After computation of fluid and structure nodal forces, we
compute the forces due to the coupling, which will only affect nodes that are on the fluid
structure interface. For each structure node, a depth penetration is incrementally updated at
each time step, using the relative velocity at the structure node, which is considered as a
slave node, and the master node within the Eulerian element (Aquelet N. et al. 2006).

The penalty coupling algorithm behaves like a spring system and penalty forces are
calculated proportionally to the penetration depth and spring stiffness. The head of the spring

21
is attached to the structure or the slave node and the tail of the spring is attached to the
master node within a fluid element that is intercepted by the structure. Similarly to penalty
contact algorithm, the coupling force is described by

F = kd (3.18)

where k represents the spring stiffness, and d means the penetration. The force in equation
(3.18) is applied to both master and slave nodes in opposite directions to satisfy force
equilibrium at the interface coupling, and thus the coupling is consistent with the fluid-
structure interface conditions namely the action-reaction principle.

3.3.4 Advection algorithm

ALE codes allow material to flow through the mesh and therefore the remap step in the ALE
algorithm needs an advection algorithm to update fluid velocity and history variables. LS-
DYNA incorporates both first and second order accurate advection algorithms. First order
accurate solutions can be thought of as generally smoothing a solution and reducing peak
values. A comparison of first order (donor cell) and second order (Van Leer MUSCL) accurate
advection algorithms resulted in almost identical results (Benjamin and Anthony 2004).

3.3.5 Element type and material model

In the explicit finite element modeling of fluid-structure interaction in water entry, the fluid
which is subjected to large deformations at the free surface, is modeled as multi-material
Eulerian, while the structure is modeled as Lagrangian. They can be connected through a
coupling algorithm. The moving Lagrangian structure imposes displacement and velocity
boundaries conditions on the Eulerian fluid, which in turn imposes traction boundaries on the
structure. For the Lagrangian solid part, the element type can be shell element, solid element
and beam element, while only solid element can be applied for Eulerian mesh.

The fluid, water and air, are modeled with Solid164 element which is an 8-nodes brick
element, and the wedge is modeled with Shell163 element which is a 4-nodes element and
can only be used in explicit dynamic analysis.

In this work, the two-dimensional structures are assumed as rigid in order to compare the
results with experimental ones. Therefore, the Lagrangian structure is defined as rigid
material. As for the Eulerian fluids including the water and the air, the material model of
*MAT_NULL which has no yield strength and behaves in a fluid-like manner is chosen. This
null material must be used with an equation of state. Furthermore, this material has no shear
stiffness and hourglass control must be used with great care. In some applications, the default
hourglass coefficient 0.1 might lead to significant energy losses. In this work, the coefficient is
defined as 1E-6.

22
3.3.6 Equation of state

In the water domain, the Gruneisen equation of state is applied with a water density of
1000kg/m3, while the air domain is described as Polynomial equation of state. If the effect of
air is neglected, the air can be modeled as void.

The Gruneisen equation of state with cubic shock velocity-particle velocity defines pressure
for compressed materials as:

  γ0  a 
ρ 0 C 2 µ 1 + 1 − µ − µ 2 
  2  2 
P= 2
+ (λ0 + aµ )E (3.23)
 µ2 µ2 
1 − (S1 − 1)µ − S 2 − S3 
 µ +1 (µ + 1)2 
and for expanded materials as

P = ρ 0 C 2 µ + (γ 0 + aµ )E (3.24)

where C is the intercept of the µ s − µ P curve; S1 , S 2 and S 3 are the coefficients of the slope

of the µ s − µ P curve; γ 0 is the Gruneisen gamma; a is the first order volume correction to γ 0 ;

and µ = ρ / ρ 0 − 1 .

The values of the related parameters for the water are listed in Table 3.1.

Table 3.1. Parameter for equation of state of water


Item ρ C S1 S2
water 1000kg/m3 1480m/s 1.92 -0.096

The linear polynomial equation of state is linear in internal energy. The pressure is given by:

P = C 0 + C1 µ + C 2 µ 2 + C 3 µ 3 + (C 4 + C 5 µ + C 6 µ 2 ) E (3.25)

where term C 2 µ 2 and C 6 µ 2 are set to zero if µ < 0 , µ = ρ / ρ 0 − 1 , and ρ / ρ 0 is the ratio of

current density to reference density. ρ 0 is a nominal or reference density defined in the null

material card.

The linear polynomial equation of state may be used to model gas with gamma law equation
of state. This may be achieved by setting:

C 0 = C1 = C 2 = C 3 = C 6 = 0 (3.26)

C 4 = C5 = γ − 1 (3.27)

where, γ is the ratio of specific heats.

The values of the related parameters for the air are listed in Table 3.2.

23
Table 3.2. Parameter for equation of state of water
Item C4 C5
air 0.4 0.4

In this work, penalty coupling method is applied for the interaction of Eulerian fluids and the
Lagrangian solid, which is different from the penalty based contact coupling method used by
Stenius et al. (2006). The penalty factor used here is defined to simulate the coupling effect
between the fluids and the structure. The coupling effect is limited to the normal direction of
the solid surface. So the sliding is allowed in the tangent direction. Usually, compression
coupling direction option is selected for rigid body impact.

3.3.7 Boundary conditions

The simulation of fluid-structure interaction is time-consuming. In order to reduce the


computational time, the model of the water entry should be limited to some extend. Due to the
symmetry, the symmetric boundary condition can be applied on the symmetric plane. In
addition, the fluid exterior boundaries are defined as non-reflecting boundaries to prevent
artificial stress wave reflections generated at the model boundaries.

3.3.8 Analytical control

Double-precision solver of LS-DYNA 971 version is needed for this simulation. Since the
double-precision run times will be approximately 30% longer than single-precision run times,
the simulation for water entry of the sections is more time-consuming than other explicit
analysis. Much effort is needed to decrease the computational time, meanwhile, to obtain
accurate predictions.

The selection of related parameters involving in coupling formulation is important. The


parameter study about the following items is to be carried out for the simulations of the
slamming problems in this work.

3.3.8.1 Penalty factor

Penalty factor (PFAC) is a scale factor for scaling the estimated stiffness of the interacting
(coupling) system. It is used to compute the coupling forces to be distributed on the slave and
master nodes which were mentioned in section 3.3.3. To avoid numerical instabilities, the
factor is introduced as in the contact method.

3.3.8.2 The number of coupling point

The number of coupling points (NQUAD) is distributed over the surface of each Lagrangian
segment. Generally, two or three coupling points per each Eulerian element width are
adequate. Low coupling points may lead to fluid leakage through structure, while the
computational time increases greatly when that is too much.

24
3.3.8.3 Time step

For the explicit analysis using LS-DYNA, the time step is conditional stable, which is related to
the shortest duration for an acoustic wave to travel across any element of the model. The
critical time step size is the minimum time value that the sound travels through all elements
(solid and shell mesh). The time step used in the simulation can be set by one scale factor
based on the critical one. The default value of the factor is 0.9 which can be reduced to
ensure the stability of the simulation. However, low time step lead to more computational time.

25
26
Chapter 4 Slamming load on the two-dimensional
wedge during water entry

4.1 Description of test

The drop tests of one rigid wedge section with deadrise angle of 30° were carried out in
MARINTEK by Zhao et al. (1996) as shown in Figure 4.1. Table 4.1 presents the main data
for the test section of symmetric wedge. Results of measured pressure and slamming force
will be adopted to compare with predicted numerical results.

Figure 4.1. Geometry of the test section

It must be noted that the total weight of the drop rig include the measuring section, the free-
falling rig used in the drop tests and the vertical force transducers. In the explicit finite element
analysis of LS-DYNA, the mass of the model should be consistent with the extension of the
model in z -direction.

Table 4.1. Primary data of the tests for symmetric wedges


Breadth of section 0.50m
Length of measuring section 0.20m
Length of each dummy section 0.40m
Total length 1.00m
Dead rise angle 30°

The vertical slamming force was obtained by using two force transducers connected to the
ends of the measuring section (see in Figure 4.1), while, the pressures in point 1 to point 5
were measured by pressure gauges.

27
The sections were dropped against calm water, which means that the velocity of the wedge
was changing with time. Furthermore, at least two drop tests were performed for each
condition, which limited the test error.

4.2 Description of finite element model


The commercial explicit FEM code LS-DYNA (version 971, 2007) is used with double
precision for the numerical simulation. The FEA is based on a multi-material Arbitrary
Lagrangian-Eulerian (ALE) formulation and penalty coupling method. The penalty coupling
method is applied for the interaction of Eulerian fluids and the Lagrangian solid, which is
different from the penalty based contact coupling method used by Stenius et al. (2006).

The fluids, water and air, are modeled with solid164 element and defined as void materials
(*MAT_NULL), while the wedge is modeled with shell163 element and rigid body material.

Figure 4.2. 2-D water entry model for wedge with deadrise angle of 30°.

Figure 4.2 shows the wedge model setup in LS-DYNA. Two-dimensionality is applied by fixing
all nodes in the z -direction and assuring that the model has an only one element extension in
the z -direction. The boundaries of fluids are defined as non-reflecting in order to decrease
the effect of the reflection of acoustic wave. Due to the symmetric geometry of the wedge,
only half of the wedge, water and air are needed to be modeled, therefore, symmetry
boundary condition is applied on the surfaces where x = 0 . In this way, the simulation CPU
time will be decreased dramatically.

The size of air and water domain in x and y direction is selected to try to eliminate the effects
of limited boundaries as much as possible. The size of water domain is 1250mm*700mm in
plane x − y which is about five times of the wedge’s size, and that of air domain is

1250mm*200mm. Initial impact velocity above calm water surface is -6.15m/s in vertical y -

direction. Gravity is not considered.

28
From considerations of computational efficiency, it is not a good choice to mesh uniformly the
total fluids (air and water) domains. Only the domain near the wedge and also where the
wedge will pass through are meshed uniformly with finer meshes, while the domain far away
the wedge is moderately expanding toward the boundaries as shown in Figure 4.3.

Figure 4.3. Parts of the mesh for the model of wedge in LS-DYNA.

Much work about slamming loads on wetted surface of wedge subjected to water entry has
been done using LS-DYNA by other researchers, for example, Stenius et al. (2006) and
Alexandru et al. (2007) who proposed that mesh size is of great significance to the predicted
results. In the uniformly meshed area, different mesh densities varying from 25mm to 1.25mm
have been studied. It shows that satisfactory result is achieved when the mesh size is smaller
than 5mm. Therefore, the mesh size of the air and water in this domain is defined as 5mm,
2.5mm, 1.25mm respectively, and that of the wedge is set as the same as that of fluids.
Table 4.2 lists the main parameters for the three models. The predicted accelerations are
proportional to the slamming force on the measuring section in the test. The time instant when
the vertex of 2D wedge touches the element on the free surface is set as 0.0s. Figure 4.3
shows part of meshes in air and water domains near the vertex of the wedge from model 2.

Table 4.2. The main data for three models of a wedge with different mesh densities.
Parameters Model 1 Model 2 Model 3
Mesh size 5mm 2.5mm 1.25mm
Number of elements(Fluids + Structures) 5725+118 29500+215 127000+430
Number of elements on impacting wedge surface 58 115 230
CPU time 56 m 9 h 49 m 96h 54m

Apart from the mesh size, other related parameters involving in fluid-structure coupling
simulation, such as penalty factor, time step, and coupling point which have been mentioned
on section 3.2, are discussed through a parameter study.

29
The predicted results can be obtained by post-processing of LS-DYNA, and are compared
with experimental values from Zhao et al. (1996). The acceleration and velocity of the wedge
can be extracted directly from the result files in LS-DYNA by defining corresponding ASCII
keywords, however, pressure sensors are needed to be located on the center of each shell
elements at the coupling surface to obtain their pressure values.

4.3 Verification of the explicit finite element method


In order to validate the predicted results of the finite element method, the predicted total
slamming force and pressure distribution at one time instance on the wedge are compared
with the experimental results published by Zhao and Faltinsen (1996), the analytical results
from Wagner (1932) and Mei et al. (1999) are included as well.

In this section, the mesh size of the fluid and wedge are both 2.5mm (model 2). The instant
when the vertex of 2D wedge touches the element on the water surface is set as 0.0s. The
gravity is neglected.

As seen in Figure 4.4, the predicted slamming forces have very good agreement with the
experimental ones in the initial stage, while the predictions are about 5% larger than the
measured values in the middle stage during the water entry, which might be due to three
dimensional effects. For the results of Wagner (1932) and Mei et al. (1999), the flow
separation was not taken into account, so the slamming forces in the later stage are not
included in the figure. In the initial stage, the predicted slamming forces have great similarities
with the ones of Mei et al. (1999), while Wagner’s solution significantly overestimates the
impact force.

7000

6000
Slamming load(N)

5000

4000

3000
LS-DYNA
2000 Expt.by Zhao
Mei´s solution
1000
Wagner
0
0 0.005 0.01 0.015 0.02 0.025
Time(s)

Figure 4.4. Predicted slamming forces on the wedge with a deadrise of 30° from LS-DYNA, drop test
and analytical solutions.

Predicted non-dimensional pressure distributions on the wetted surface of the wedge are
presented in Figure 4.5, and the pressure values on P1-P5 of Zhao’s test are also included.

30

Y is the vertical coordinate on the wedge surface, Yd = V (t ) dt is the draft of the wedge, and

V (t ) is drop velocity of the wedge.

10
LS-DYNA
8 Expt.by Zhao

Cp 6

0
-1 -0.5 0 0.5 1
Y/Yd
Figure 4.5. Pressure distributions on the symmetric wedge with a deadrise angle of 30º for LS-DYNA at
the time instance of t = 0.0158s .

The predicted pressure distribution agrees well with experimental results. Generally speaking,
the measured values are smaller than the predictions, especially in the middle stage of the
impact, which can be observed also from the comparison between the results of a fully
nonlinear numerical method (Zhao et al. 1996) and of the test However, the pressure
variations trend on the body surface is exactly the same, and the locations that the pressure
coefficient comes up to the maximum value are almost the same.

4.4 Results and Discussion


In this section, the numerical predicted results of slamming loads from LS-DYNA, which are
divided into the vertical slamming force, pressure distribution along the wetted surface of
wedge at different time instance and pressure histories at different positions of the wedge, are
compared with published experimental values from Zhao at al. (1996). Meanwhile, through
comparisons between the results of the three different models, the effects of the mesh density
is discussed as well.

4.4.1 Vertical slamming force

Figure 4.6 compare the predicted vertical slamming forces with experimental results. The
predicted slamming forces have very good agreement with the experimental ones from the
beginning of slamming, to the flow separation on the knuckle at about time 0.0158s, while the
predictions are about 5% larger than the measured values in the middle stage during the
water entry, which might be due to three dimensional effects in the drop tests. However, this
difference is not as large as up to 20%, which was predicted by BEMs of Zhao et al (1996).

As seen in Figure 4.6, when the mesh size is 5mm (Model 1), there are some high frequent
oscillations on the curve because of numerical noise which are especially obvious in the initial

31
stage of water entry. The difference between predicted and experimental slamming force after
flow separation is relatively large.

When the mesh size is 2.5mm (Model 2), the result is better except the small oscillations at
the beginning. It can describe the flow separation very well at the knuckle. The slamming
force drops fast after flow separation, and then decays slowly. Furthermore, the CPU time
listed in Table 4.2 is also acceptable.

6000

5000
Slamming force(N)

4000

3000

2000 LS-DYNA 5mm


LS-DYNA 2.5mm
1000 LS-DYNA 1.25mm
Expt.by Zhao
0
0 0.005 0.01 0.015 0.02 0.025
Time(s)
Figure 4.6. Slamming forces on the symmetric wedge with a deadrise angle of 30° for different mesh
sizes.

Finally, when the mesh size is 1.25mm (Model 3), no obvious oscillations are observed in the
curve which shows best consistency, but it will take about 4 days to finish only one simulation
in a normal personal computer.

4.4.2 Pressure distributions at different time instances

Pressure distributions on the wetted surface of the wedge are presented at three time
instances, and are compared with experimental results in Figure 4.7 which are plotted in
terms of the non-dimensional pressure coefficient C P and the relative location on the wetted

part of the wedge surface. C P is the pressure coefficient, Y is the vertical coordinate on the

wedge surface, and Yk is vertical coordinate of the keel.

The predicted results agree well with the measured values. Generally speaking, the
measured pressures are smaller than the predicted ones, especially after the initial stage as
shown in Figure 4.7 (b) and (c). It is mainly because of three-dimensional effects in the drop
tests.

When a much finer mesh is applied in the simulator model, the predicted results are much
smoother. Obviously, the pressures from Model 3 show the best consistency with
experimental results. Only the value of pressure P2 in Figure 4.7 (a) from the experiment is
larger than the predicted results from LS-DYNA. The reason for this difference is not clear,

32
maybe mesh density at the initial stage here is not finer enough, or maybe due to
experimental errors, because this value of pressure P2 from the experiment is also larger
than the predicted results by BEMs in Zhao et al. (1996).

8
LS-DYNA 5mm
7 P2
LS-DYNA 2.5mm
6 LS-DYNA 1.25mm
Expt.by Zhao
5
P1
Cp

0
-1 -0.5 0 0.5 1
Y/Yd
(a)
8
LS-DYNA 5mm
LS-DYNA 2.5mm
6 LS-DYNA 1.25mm
Expt.by Zhao. P5
Cp

P1 P4
2 P2
P3

0
-1 -0.5 0 0.5 1
Y/Yd
(b)
7
LS-DYNA 5mm
6
LS-DYNA 2.5mm
5 LS-DYNA 1.25mm
Expt.by Zhao.
4
Cp

2 P1
P2 P5
P3 P4
1

0
0 0.2 0.4 0.6 0.8 1
(Y-Yk)/Yd
(c)
Figure 4.7. Comparison of pressure distributions at different time instances for the models with different
mesh sizes. (a) t1 = 0.00435s ; (b) t 2 = 0.0158s ; (c) t 3 = 0.0202 s .

In the initial stage, Figure 4.7 (a) shows that results from both Model 1 and 2 are not good,
while in Figure 4.7 (b) and (c), results from both Model 1 and 2 are better. It may be explained
as that the mesh density near the vertex of wedge in model 1 and 2 is not finer enough to
describe the slamming pressure in this case.

33
Comparing with slamming force results in Figure 4.6, it may be concluded that the mesh
density in the water and air domains near the vertex of wedge is most important. In other
words, if the mesh density in water and air domain of Model 2 near the vertex of wedge is set
much finer, for example, 1.25mm, then the predicted results at the initial stage in Figure 4.6
and Figure 4.7 (a) will become better and acceptable comparing with those from Model 3. Not
so much CPU time is needed as Model 3. It may be a good solution from the view point of
computational efficiency.

4.4.3 Pressure histories in different points

Figure 4.8 shows the pressure histories at different points on the wetted surface of the wedge
from P1 to P5, the locations of which can be seen in Figure 4.1.The variable of pressure is
plotted in terms of the actual value of that rather than the non-dimensional coefficient, due to
the varying drop velocity during water entry process, The predicted pressure results from
model 2 and 3 are shown, but those from model 1 are not included because the results are
not good due to too coarse meshes.

4
4 x 10
x 10 15
12
P1 LS-DYNA 2.5mm P2 LS-DYNA 2.5mm
10
LS-DYNA 1.25mm LS-DYNA 1.25mm
Pressure(Pa)
Pressure(Pa)

8 Expt.by Zhao. 10 Expt.by Zhao.


6

4 5
2

0 0
0 0.005 0.01 0.015 0.02 0.025 0 0.005 0.01 0.015 0.02 0.025
Time(s)
(a) Time(s) (b)
4 4
x 10 x 10
15 15
P3 LS-DYNA 2.5mm P4 LS-DYNA 2.5mm
LS-DYNA 1.25mm LS-DYNA 1.25mm
Pressure(Pa)
Pressure(Pa)

10 Expt.by Zhao. 10 Expt.by Zhao.

5 5

0 0
0 0.005 0.01 0.015 0.02 0.025 0 0.005 0.01 0.015 0.02 0.025
Time(s) (c) Time(s) (d)
4
x 10
15
P5 LS-DYNA 2.5mm
LS-DYNA 1.25mm
Pressure(Pa)

10 Expt.by Zhao.

0
0 0.005 0.01 0.015 0.02 0.025
Time(s) (e)
Figure 4.8. Pressure histories at different points for the models with different mesh sizes. (a) Point P1,
the depth from the keel to the position is 12.5mm; (b) Point P2, the depth from the keel to the position is

34
37.5mm; (c) Point P3, the depth from the keel to the position is 62.5mm; (d) Point P4, the depth from the
keel to the position is 87.5mm; (e) Point P5, the depth from the keel to the position is 131.8mm.

The predicted results show good agreement with the experimental ones in general. Generally
speaking, they are a little larger than the experimental ones, except the value at point P2 in
Figure 4.8 (b). The reason of the abnormal difference in point P2 was explained in 4.2.2
before.

The peak values of pressure histories from model 3 with finer mesh 1.25mm are larger than
those from model 2, while the pressure distributions after peak are almost the same. There is
one obvious impulse in pressure curves before the peak value from model 2 in Figure 4.8 (b),
(c) and (d). Maybe these impulses are due to numerical noise, because when finer mesh
1.25mm is used, the phenomena disappear.

The peak value of point P1 is smaller than that of P2, P3, and P4, and the shape of pressure
curve is not as same as others. It is because the mesh density near the vertex of wedge
including the location of P1 is still not fine enough in model 3.

In order to capture the pressure peak phenomena correctly and also to eliminate the
numerical noise in the curves, much finer mesh is needed in this case, for example 1.25mm.
For the case of slamming force, mesh size 2.5mm maybe is enough as shown in Figure 4.6.

4.4.3.1 Pressure distributions and free surface elevation during water entry

Figure 4.9 presents both water jet and pressure contour phenomena from the time instances
at 0.00435s, 0.0158s, to 0.0202s, corresponding to pressure distributions described in Figure
4.7. Half of rigid wedge and part of the water domain near the wedge is shown. The vertex of
the rigid wedge of model 3 will touch the calm water at time 0.0036s in the LS-DYNA
simulation, which corresponds to the time 0.0s in the drop test. Three coupling points option is
chosen for each coupled Lagrangian element, and no fluid leakage is observed on the
coupling wedge surface.

Corresponding to these pressure contours, the pressure distributions along the wetted
surface of wedge at different time instances are presented in Figure 4.10 in which x is
horizontal coordinate on the wedge surface and B is the half-width of the wedge. Figure 4.10
(a) shows the pressure distributions before flow separation, while Figure 4.10 (b) presents
those after flow separation.

Before flow separation, as shown in Figure 4.9 (a) and (b), the maximum pressure is located
in the inner domain, or the water up-rise which was described in Wagner (1932) theory. The
slamming pressure will decrease in the outer domain in the water, or in the jet flow along the
wedge surface. The pressure in the jet flow is so small that it can be ignored. That’s the
reason why the Wagner theory can explain the water impact almost correctly, and still be
widely used today. The maximum pressure appears near the root of water jet, and decreases
during the water entry process, as seen in Figure 4.10 (a), however, it does not decrease

35
much before flow separation so slamming force will increase gradually when the wetted
surface is increased as shown in Figure 4.6.

After flow separation, the maximum pressure on the wetted surface of wedge moves
downwards from the vertex to the keel during the water entry process. Figure 4.9 (c) shows
the water jet and pressure contour in the final stage after flow separation on the knuckle, and
Figure 4.10 (b) shows the pressure distributions after flow separation. The pressure is almost
the same along the coupling wetted surface along wedge at the time instance of 0.0202s, but
it is much smaller comparing with that before flow separation as shown in Figure 4.9 (a) and
(b). It is also observed from Figure 4.10. That is the reason why the slamming force will
reduce very much comparing with that before flow separation. It is consistent with what is
presented in slamming force curve in Figure 4.6.

(a)

(b)

(c)
Figure 4.9. Free surface elevations and pressure contours of the model with mesh size of 1.25mm
during water entry of the wedge. (a) t1 = 0.00435s ; (b) t 2 = 0.0158s ; (c) t 3 = 0.0202 s .

36
4 4
x 10 x 10
15 t=0.00658s t=0.0158s
t=0.0115s 10 t=0.0175s
t=0.0158s t=0.0185s

Pressure (Pa)
Pressure (Pa)
10 t=0.0202s

5
5

0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
x/B (a) x/B (b)
Figure 4.10. The predicted pressure distribution at different time instances on wedges with deadrise
angle 30ºduring the water entry by LS-DYNA. t = 0 corresponds to that the keel touches the water
surface. (a) Pressure distributions on the wedge before flow separation. (b) Pressure distributions on the
wedge after flow separation.

4.5 Parameter study


4.5.1 Acceleration of gravity

Predicted slamming force which related to acceleration of gravity and which neglects it are
shown as Figure 4.11 in which experimental results are also included. LS-DYNA data 1
represents the results of the model with considering acceleration of gravity, while LS-DYNA
data 2 represents the results of the model without considering that. Comparisons of the
predicted results with experimental values show a great deal of similarity. And it shows that
the acceleration of gravity of structure has limited influence on the results of vertical slamming
force. Taking the acceleration of wedge into consideration, the predicted maximum slamming
force is about 10% larger than the experimental one which is closer to the model neglecting
gravity. However, the time instances when the slamming loads come up to peak value are
almost the same in both cases.

7000

6000
Slamming load(N)

5000

4000

3000

2000 LS-DYNA data1


1000 LS-DYNA data2
Expt.by Zhao.
0
0 0.005 0.01 0.015 0.02 0.025
Time(s)
Figure 4.11. Comparison of slamming force between the model considering gravity and that neglecting
gravity.

37
8
LS-DYNA data1
LS-DYNA data2
6 Expt.by Zhao.

Cp
4

0
-1 -0.5 0 0.5 1
Y/Yd
(a)
8
LS-DYNA data1
6 LS-DYNA data2
Expt.by Zhao.
Cp

0
-1 -0.5 0 0.5 1
Y/Yd (b)
6
LS-DYNA data1
5
LS-DYNA data2
4 Expt.
Cp

0
0 0.2 0.4 0.6 0.8 1
(Y-Yk)/Yd
(c)
Figure 4.12. Comparison of pressure distributions at three time instances between the model
considering gravity and that neglecting gravity. (a) t1 = 0.00435s ; (b) t 2 = 0.0158s ; (c) t 3 = 0.0202 s .

Predicted pressure distributions on the wetted surface of the wedge at three time instances
are shown in Figure 4.12 which includes the measured values of five points. The comparison
shows small differences between the two models which perhaps due to the hydrodynamics.

Compare to the high velocity of rigid wedge, the acceleration of gravity is negligible. However,
Sun (2007) indicates that, in the case of moderate velocity, the gravity not only affects the
free-surface profile around the hull, but also influences the hydrodynamic force on the hull
surface.

38
4.5.2 Mesh density

The mesh density of the wedge is as same as that of the uniformly mesh area in Figure 4.3.
Three models with different mesh size were applied to analyze the slamming loads on the
wetted surface of the rigid wedge in section 4.4. The results show that the mesh size is of
great significance on the simulation. The finer the mesh size, the better the results are,
however, for the point view of computational time, the finer the mesh size, the more
computational time is required. For the water entry problem of the symmetric rigid wedge in
this work, the slamming force, pressure distributions and pressure histories have good
agreement with experimental results when mesh size is 1.25mm, while almost 97 hours are
required. When mesh size is 2.5mm, only the predicted slamming force and pressure
distribution are acceptable.

4.5.3 Penalty factor

The penalty factor (PFAC) is one parameter for scaling the estimated stiffness of the
interacting coupling system. It is applied to compute the coupling force to be distributed on the
structure and fluids. Simulations are run for three cases when PFAC are set as 0.5, 0.1, and
0.01 for model 3. Figure 4.13 and Figure 4.14 show comparison of slamming force and
pressure distributions with different values of PFAC respectively. Little difference is observed.
In this work, the default value 0.1 is used.

7000

6000
Slamming force(N)

5000

4000

3000

2000 PFAC=0.01
1000 PFAC=0.1
PFAC=0.5
0
0 0.005 0.01 0.015 0.02 0.025
Time(s)

Figure 4.13. Slamming forces for the models with different values of penalty factors.

8 8
PFAC=0.01 PFAC=0.01
PFAC=0.1 PFAC=0.1
6 6
PFAC=0.5 PFAC=0.5
Cp

4
Cp

2 2

0
0 -1 -0.5 0 0.5 1
-1 -0.5 0 0.5 1
Y/Yd
Y/Yd (a) (b)
Figure 4.14. Pressure distributions for the models with different values of penalty factor. (a)
t1 = 0.00435s ; (b) t 2 = 0.0158s .

39
4.5.4 Time step

The critical time step size is the minimum time value that the sound travels through any
elements in the model. One scale factor TSSFAC can be adopted to compute the time step
used in simulations. The time step calculated should not be larger than the critical one,
otherwise negative volume errors will appear. But if the time step is set to one value that is
too small, then the simulation CPU time will increase correspondently. The critical time step
size can be approximated firstly before the simulation, in order to set one sale factor to obtain
one appropriate time step. In this work, the time step factor is set as 0.4 for the simulation of
wedge.

4.6 Conclusions
Numerical simulation of slamming loads on a two-dimensional rigid wedge with the deadrise
angle 30° is carried out by the explicit FEM code LS-DYNA. The Arbitrary Eulerian-
Lagrangian solver and penalty coupling algorithm is used.

The predictions of slamming loads from LS-DYNA agree well with available experimental
results and analytical values, which validate the explicit finite element method.

The predicted 2D slamming force correctly describes the impact process from the initial stage,
flow separation, to the final one, and the values are a little larger than the 3D experimental
ones. So are the pressure values, especially after the initial stage of impact. This is mainly
because of the three-dimensional effects during water entry.

Correspond to the drop test of 2D rigid wedge, the velocity of the model varies with time in
this chapter, and the maximum one appears when the wedges impact with water. According
to the theory of hull-water impact, the peak value of pressure on the wetted surface
decreases with the reducing velocity. This process is also presented by the predicted
pressure distributions at different time instances which are shown in Figure 4.10. Before flow
separation, the maximum pressure is located near the water up-rise, while it moves
downwards from the vertex to the keel after flow separation.

A convergence study is carried out. Different mesh sizes, 1.25mm, 2.5mm and 5mm, are
compared. The finer the mesh size is, the better results can be obtained. But the CPU time
will increase dramatically once the mesh density is increased. The compromise between
mesh density and CPU time should depend on what will be predicted.

In order to capture the peak of slamming pressure, much finer mesh is needed, for example,
1.25mm in this case. The value of the pressure peak predicted by the coarser mesh model is
smaller than that by the finer mesh. For example, the maximum non-dimensional pressure
coefficient from model 3 at the time 0.0158s before the flow separation is 7.32, while that from
model 2 is only 6.65, which is about 9% smaller. The peak value of pressure P1 predicted in
model 3 is smaller than that of P2, P3, and P4, and the shape of P1’s peak is not as good as
others. Perhaps it is because the mesh size 1.25mm near the vertex of wedge and also P1 is

40
still not finer enough. Both models with mesh size 1.25mm and 2.5mm can predict pressure
histories after the peak very well. However, the coarser mesh, such as 2.5mm for this case, is
enough to predict the slamming force.

It is found that the mesh density is the most important parameter involving in the predicted
accuracy. Small differences are observed when the penalty factor is modified. One
appropriate value for time step scale should be set to reduce CPU time and also to prevent
the negative volume errors during the simulation.

The water jet flow and pressure contours in the water domain are presented during different
stages of the impact. It simulates the phenomena correctly in general from model 3 with
1.25mm mesh size. The value of pressure predicted in the jet flow is so small that it may be
ignored.

The modelling techniques described here can be applied to model wedges of different types,
and to study other hydrodynamic impact problems.

41
42
Chapter 5 Slamming load on the two-dimensional
wedges with various deadrise angles during water
entry

5.1 Introduction
The explicit finite element code LS-DYNA was applied to predict the slamming loads on a
rigid wedge with deadrise angle of 30°. Compared to experimental and analytical results, the
method is valid to the water entry problem.

In this chapter, the slamming loads on the surfaces of symmetric wedges with different
deadrise angles are studied by using explicit commercial software LS-DYNA. Wagner (1932)
theory, Stavovy and Chuang’s (1976) and Ochi and Motter (1973) formulations are used to
predict and compare with the simulated results of wedges with various deadrise angles
varying from 10° to 81°. The maximum value of slam induced pressure, pressure distributions
on the wetted surface, vertical slamming force and free surface elevation are presented.
Furthermore, different impact velocities are applied for the wedges with various deadrise
angles, to discuss the effect of that on slam induced pressure on the rigid wedge.

5.2 Finite Element Model


The models of the rigid wedges with various deadrise angle are as same as the one of the
wedge with deadrise angle of 30° as shown in Figure 4.2 and Figure 4.3, except for the
dimensions of the wedges due to the varying deadrise angle. Another different is that the
velocity of the wedge is constant during the entire water entry process.

The element types, material models of fluids and structures, and the boundary conditions are
as same as before as well. The penalty factor is set as 0.1, the number of coupling points is
three, and the time step factor is 0.4. In order to reduce the computational time, 2.5mm is
applied as the mesh size of wedge and the uniformly meshed area. The impact velocities
include 1m/s, 2m/s, 3m/s, 4m/s, 5m/s and 6.15m/s, for a given deadrise angle. If not
otherwise stated, impact velocity is 6.15m/s and the gravity is neglected.

5.3 Results and Discussion


5.3.1 Maximum value of pressure coefficient

Pressure coefficient is a non-dimensional factor that depends on the section geometry.


Obviously, the maximum value of pressure coefficients is of much significance, and it can be
generally obtained from the equation (2.33). When the wedge enters into the water with a

43
constant velocity, the method for determination of the maximum value was proposed as
equation (2.34) (Wagner 1932), equation (2.36) (Ochi and Motter (1973)) and equation (2.37)
(Stavovy and Chuang 1976).

In this section, the maximum slamming pressure coefficients are computed for the wedges
with deadrise angle varying from 10° to 45° by using the solutions analytical formulations and
LS-DYNA. The results are shown as Figure 5.1.

100
Stavovy&Chuang(1976)
80 Ochi&Motter(1973)
Wagner(1932)
Zhao´s similarity(1996)
60
Cp.max

LS-DYNA

40

20

0
10 15 20 25 30 35 40 45
β(degree)
Figure 5.1. Comparison of slamming pressure coefficient for rigid wedge with different deadrise angles.

As plotted, when the deadrise angle is small, the predicted values are much smaller than the
results from Wagner’s (1932), Stavovy and Chuang’s (1976) and Zhao’s similarity (1996)
solution, while they are closer to the results from Ochi and Motter (1973). When the deadrise
is larger, the predictions agree well with the results from the three analytical methods,
however, the values are a little smaller than those from Chuang’s formulation. In general, the
predictions of maximum pressure from LS-DYNA agree best with the values of Ochi and
Motter (1973).

As shown in Figure 5.1, the smaller is the deadrise angle, the larger is the pressure coefficient.
This was noticed previously by several researchers such as Dobrovolskaya (1969), Zhao and
Faltinsen (1993), Mei et al. (1999) and others.

5.3.2 Pressure distribution and free surface elevation

5.3.2.1 Comparisons between analytical and numerical solution

The simulated pressure distributions on the wetted surfaces of wedges with different deadrise
angle varying from 10° to 81° are presented in Figure 5.2, in which the published calculations
from Dorbrovol’s skaya (1969), Zhao and Faltinsen (1993, 1996) and Mei et al. (1998) are
included as well, together with the analytical predictions from Wagner (1932) which are
calculated according to the equation (2.19). The results from Zhao’s asymptotic (1993) are
only available for the deadrise angle smaller than 30°. For the predictions of LS-DYNA, the

44
pressure distribution for a wedge is varying from one time instance to another. Corresponding
to the analytical results, the plotted ones are the pressure distributions at the time when the
pressures come up to peak values.

The impact velocity is assumed constant in analytical solution, so the velocity of wedge in the
FEM model is defined as -6.15 m/s in the vertical y -direction from the beginning to the end of

the water entry. The wedge mass is 0.30125kg. The following curves are plotted in terms of
the non-dimensional pressure coefficient C P and the relative location on the wetted part of the

wedge surface.

100 40
Dobrovol´s skaya(1969) Zhao´s BEM(1993) β=15º
β =10º
80 Zhao´similarity(1993) Zhao´similarity(1993)
30
Mei et al.(1999) Zhao´asymptotic(1999)
60 LSDYNA
LSDYNA
Cp

Cp
20
40

20 10

0 0
-1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1
Y/ Yd (a) Y/Yd (b)
25 15
Zhao´s BEM(1993) Zhao´s BEM(1993)
β=20º β=25º
20 Zhao´similarity(1993) Zhao´similarity(1993)
Zhao´asymptotic(1999) Zhao´asymptotic(1999)
10
15 LSDYNA LSDYNA
Cp

Cp

10
5
5

0 0
-1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1
Y/Yd (c) Y/Yd (d)
12 10
Dobrovol´s skaya(1969)
10 β=30º Wagner β=35º
Zhao´similarity(1993) 8
Mei et al.(1996) LS-DYNA
8
LSDYNA 6
Cp
Cp

6
4
4
2
2

0 0
-1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1
Y/Yd (e) Y/Yd (f)
6 6
Zhao´s BEM(1993) Dobrovol´s skaya(1969)
5 β=40º 5 β=45º
Zhao´similarity(1993) Zhao´similarity(1993)
4 LSDYNA 4 Mei et al.(1999)
LSDYNA
Cp

Cp

3 3

2 2

1 1

0 0
-1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1
Y/Yd (g) Y/Yd (h)

45
2 2
β=60º Dobrovol´s skaya(1969)
Wagner β=70º
Zhao´similarity(1993)
1.5 1.5 LS-DYNA
Zhao´s BEM(1999)
LSDYNA

Cp
Cp
1 1

0.5 0.5

0 0
-1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1
Y/Yd (i) Y/Yd (j)
1 Dobrovol´s skaya(1969)
β=81º
Zhao´similarity(1993)
0.8
Zhao´s BEM(1993)
0.6 LSDYNA
Cp

0.4

0.2

0
-1 -0.5 0 0.5 1
Y/Yd (k)
Figure 5.2. Pressure distributions on symmetric wedges with constant vertical velocity during water entry.

When the deadrise angle is small, the predicted results are in good agreement with the
numerical and analytical calculations in the initial stage of the water impact; however, the
simulated peak values of the pressure are much smaller in the later stage. As the deadrise
angle increases, the agreement in the later stage between LS-DYNA and other solutions are
more satisfactory, in particular for the cases of 30° and 40°. When the deadrise angle is 60°,
they have good agreement except that the simulated pressure on the position near the keel of
the wedge is smaller than other values. The same difference can be found when the deadrise
angle is 81°. For the results of Mei et al. (1998), the pressure drop fast after coming up to the
peak value, this means that the free surface elevation was not considered.

For larger deadrise angles, the agreement is less satisfactory between LS-DYNA and
Wagner’s solution. In the case of Wagner’s solution, the pressure drops fast to zero after the
pressure comes up to peak value. However, for the predictions of pressure from LS-DYNA,
the pressures drop to zero gradually along the surface of the wedge.

Though there are some differences between these solutions, the same conclusion can be
obtained from the calculations. When the deadrise is small, the maximum pressure is located
near the spray root of the water jet before separation occurs, and the pressure distribution is
approximately uniformly along the surface of the wedge when the deadrise angle is close to
45°. When the deadrise angle is larger than 60°, the maximum pressure is located at the keel
of the wedge, and it means that the peak pressure appears at the time instance when the
wedge touches with water.

46
5.3.2.2 Pressure distributions and free surface elevation during water entry

Different from the analytical solution, the FEM can be used to predict the variation of pressure
distributions on the surface of wedges at different time instances. In this section, the wedges
with deadrise angle 30° and 60° are calculated to study the differences on variation of
pressure distributions. The vertical velocities of wedges are constant.

The pressure distributions for different time instances are presented in Figure 5.3, where x is
the horizontal coordinate on the wedge surface and B is the half-width of the wedge. Only
half of the section is modeled.

10 10
t=0.0115 t=0.0165
8 t=0.0135 8 t=0.0182
t=0.0190
t3=0.0165
6 6 t=0.0210

Cp
Cp

4 4

2 2

0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
x/B (a) x/B (b)
Figure 5.3. Predicted pressure distributions on wedges with constant velocity when the deadrise angle is
30°. (a) before flow separation; (b) after flow separation

Figure 5.3 shows the pressure distributions on the wedge with deadrise 30° at different time
instances, which indicate that the maximum pressures are located near the spray root of the
water jet before flow separation occurs, after the flow separation, the maximum pressures
decrease and move to the keel of the wedge, which are not much larger than the values on
other positions. As shown in Figure 5.4, the maximum pressures are located in the keel of the
wedge during the entire impact when the deadrise angle is 60°, where Figure 5.4 (a) presents
the pressure distributions before flow separation and Figure 5.4 (b) presents those after flow
separation. Furthermore, the maximum pressure has small differences before flow separation,
after that, the peak value decreases gradually.

1.5 1.5

1 1
Cp

Cp

t=0.011967
0.5 0.5
t=0.013967 t=0.019067
t=0.015467 t=0.022967
t=0.019067 t=0.025967
0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
x/B
(a) x/B (b)
Figure 5.4. Predicted pressure distribution on wedges with constant velocity when the deadrise angle is
60°. (a) before flow separation; (b) after flow separation.

47
The same conclusions can also be obtained from Figure 5.5 and Figure 5.6 which show the
free surface elevations and the pressure contours at different time instances for wedges with
deadrise angle 30° and 60°.

Before flow separation, as shown in Figure 5.5 (a) and Figure 5.5 (b), the maximum pressures
are located in the inner domain, or the up-rise which was described in Wagner (1932) when
the deadrise angle is 30°. The pressure in the jet flow is much smaller than the one near the
spray root of the jet. That is the reason why the pressure coefficient drop fast after the peak
value as seen in Figure 5.2(e). Figure 5.5 (c) displays the pressure distribution along the
wedge surface after flow separation, and it shows that the pressure reduces quickly and the
peak value move to the keel of the wedge.

(a)

(b)

(c)
Figure 5.5. Predicted water jet flows and pressure contours in water by LS-DYNA for the wedge with
deadrise angle 30º. (a) t=0.00608, (b) t=0.0165, (c) t=0.0190. t=0 corresponds to that the keel touches
the water surface.

48
As seen in Figure 5.6, corresponding to Figure 5.4 (a) and Figure 5.4 (b), the maximum
pressures are located on the keel of the wedge during the entire water entry process when
the deadrise angle is 60°.

(a)

(b)

(c)
Figure 5.6. Predicted water jet flows and pressure contours in water by LS-DYNA for the wedge with
deadrise angle 60º. (a) t = 0.00608s , (b) t = 0.019067 s , (c) t = 0.025967 s . t=0 corresponds to that the
keel touches the water surface.

5.3.3 Vertical slamming force

5.3.3.1 Non-dimensional slamming force

Generally speaking, the total slamming force can be obtained by directly integrating pressure
distribution with respect to the wetted surface of the wedge, or by applying equation (2.3).

To analyze the slamming force of the wedges with various deadrise angles, wedges with
deadrise angle from 15° to 80° are calculated by LS-DYNA in this section. The impact velocity
is -6.15m/s in the y -direction. The total masses of wedges are 241kg/m, and the mesh sizes

are 2.5mm which means that the mass of the models are 0.30125kg. According to equation

49
(2.3), the vertical slamming forces can be obtained by multiply the accelerations of wedges
with the mass of measuring section (Zhao and Faltinsen, 1996).

Figure 5.7 shows the maximum values of slamming force for symmetric wedges with different
deadrise angles. As expected, the larger is the deadrise angle, the smaller is the peak value
of slamming force.

Ramos and Guedes Soares (1998) compared the predicted non-dimensional vertical force
using the equation (2.49) for a rigid wedge with different deadrise angles. Figure 5.8 includes
the simulated results by LS-DYNA which are much larger than those from Ochi and Motter
(1973) when the deadrise angle is large and much smaller than those from Stavovy and
Chuang (1976) when the deadrise angle is small. The predictions obtained by using boundary
element method have best agreement with the values from LS-DYNA.

4
x 10
Maximum value of slamming force (N)

5
LS-DYNA
4

0
0 10 20 30 40 50 60 70 80
Deadrise angle (degree)

Figure 5.7. Maximum values of slamming force for symmetric wedges with different deadrise angles.

4
10
BE meth
Non-dimensional vertical force

Stavovy&Chuang(1976)
3
10 Ochi&Motter(1973)
Derivative of added mass
LS-DYNA
2
10

1
10

0
10
0 10 20 30 40 50
Deadrise angle(degree)
Figure 5.8. Comparison of the predictions of non-dimensional vertical force

50
10

Coefficient of vertical slamming force


Wagner(1932)
Von Karman(1929)
8 Von Karman-moment
Zhao(1993)Simplified
Zhao(1996)Similarity
6
LS-DYNA

0
0 10 20 30 40 50 60
Deadrise angle(degree)

Figure 5.9. Comparison of the predictions of coefficient of vertical slamming force

Numerically simulated coefficient of vertical slamming force on a wedge with different


deadrise angles by using the equation (2.41), and the corresponding analytical values are
plotted in Figure 5.9. The Wagner (1932) solution clearly overestimates the slamming force
for large deadrise angles, while the two different von Kármàn solutions, one of which is based
on a flat plate solution and the other one is obtained by using the exact body boundary
condition and the principle of conservation of momentum, underestimate the slamming force
for small deadrise angles. There are good agreement between the simplified and the similarity
solution proposed by Zhao and Faltinsen (1996). The simulated values by LS-DYNA are
between the results of Zhao and Faltinsen and von Kármàn when the deadrise angle is
smaller than 60°. For the deadrise angle around 45°, the simulated predictions have good
agreement with those from von Kármàn. Furthermore, the vertical slamming force for the 15°
deadrise angle is larger than the one for the 10°. Maybe it is due to the numerical noise of the
commercial code.

5.3.3.2 Time history of the vertical slamming force

Assume that the length of the ship is L , the period of the time history of the vertical slamming
force can be obtained by equation (2.50) which was proposed by Ochi and Motter (1973), and
thus the maximum value occurs at the time instance of Td / 2 . Since the maximum slamming

force of the simple form from Ochi and Motter (1973) is normalized to 1, the one from LS-
DYNA can be calculated using the same method. For different deadrise angles of the wedge,
the time histories of the vertical slamming force are instinct.

The predicted time histories of the vertical slamming force for the wedges with deadrise angle
of 15°, 30° and 45°, are respectively shown in Figure 5.10, Figure 5.11 and Figure 5.12, which
show that the slamming force from LS-DYNA is almost linearly related to time during the first
period of Td / 2 as same as the simple form, however it dose not drop to zero during the

period.

51
1.4
Kawakami(1977)
1.2
Ochi&Motter(1973)
1 LS-DYNA

0.8

0.6

0.4

0.2

0
0 0.002 0.004 0.006 0.008 0.01
t/sqrt(L)
Figure 5.10. Predicted time histories of the vertical slamming force when the deadrise angle is 15°.

1.5
Kawakami(1977)
Ochi&Motter(1973)
1 LS-DYNA

0.5

0
0 0.002 0.004 0.006 0.008 0.01
t/sqrt(L)
Figure 5.11. Predicted time histories of the vertical slamming force when the deadrise angle is 30°.

1.4
Kawakami(1977)
1.2
Ochi&Motter(1973)
1 LS-DYNA

0.8

0.6

0.4

0.2

0
0 0.002 0.004 0.006 0.008 0.01
t/sqrt(L)
Figure 5.12. Predicted time histories of the vertical slamming force when the deadrise angle is 45°.

5.3.4 Results for different impact velocities

Different impact velocities which include 1m/s, 2m/s, 3m/s, 4m/s, 5m/s and 6.15m/s, are
applied for the rigid wedges with the deadrise angles of 15°, 30° and 60° respectively. The
predicted maximum pressure on the wetted surface of wedges with different impact velocities
are compared with the published analytical results in the following three figures, where Figure

52
5.13 plots the comparisons for the wedge with a deadrise angle of 15°, and Figure 5.14
presents those when the deadrise angle is 30°, while the results for the wedge with a
deadrise angle of 60° are shown in Figure 5.15.

5
x 10
15

Maximum value of pressure (Pa)


Stavovy&Chuang(1976)
Wagner(1932)
Ochi&Motter(1973)
10
LS-DYNA

0
1 2 3 4 5 6 7 8
Impact velocity (m/s)
Figure 5.13. Maximum values of pressures for different impact velocity when the deadrise angle is 15°.

5
x 10
4
Maximum value of pressure (Pa)

Stavovy&Chuang(1976)
Wagner(1932)
3 Ochi&Motter(1973)
LS-DYNA

0
1 2 3 4 5 6 7 8
Impact velocity (m/s)
Figure 5.14. Maximum values of pressures for different impact velocity when the deadrise angle is 30°.

5
x 10
Maximum value of pressure (Pa)

4
Stavovy&Chuang(1976)
Wagner(1932)
3
LS-DYNA

0
1 2 3 4 5 6
Impact velocity (m/s)
Figure 5.15. Maximum values of pressures for different impact velocity when the deadrise angle is 60°.

53
As seen in Figure 5.13, Figure 5.14 and Figure 5.15, the maximum pressures increase with
the increasing impact velocity, and also increase with the decreasing deadrise angle when the
impact velocity is the same. When the deadrise angle is 15°, the analytical values from
Stavovy and Chuang (1976) are much larger than the simulated ones, while the results of
Wagner (1932) are the smallest. As to the values from Ochi and Motter (1973), the
differences between them and the predicted ones from LS-DYNA become larger when the
impact velocity is larger, as plotted in Figure 5.13.

As seen in Figure 5.14, when the deadrise angle is 30°, the predictions from LS-DYNA are
the smallest compared to other analytical ones, while those of Stavovy and Chuang (1976)
are the largest and the other two methods almost have the same results, especially when the
impact velocity is small.

Figure 5.15 shows that the analytical method of Stavovy and Chuang (1976) significantly
overestimates the maximum pressure when the deadrise angle is 60°, and the results from
Wagner (1932) somehow larger than those from LS-DYNA when the impact velocity is larger
than 3m/s. Actually, it can be found in Figure 5.1 that, the maximum pressure coefficient
calculated by Stavovy and Chuang (1976) are much larger than that from other methods
when the deadrise angle is larger than 30°.

5.4 Conclusions
Numerical simulation of slamming loads on symmetric wedges with various deadrise angles
are carried out by the explicit finite element method LS-DYNA to analyze the relationship
between deadrise angle and slamming loads.

When the impact velocity is assumed constant, the maximum slamming pressure coefficients
are computed for the wedge with deadrise angle varying from 10 to 45° by using the solutions
of Wagner, Chuang, Ochi and LS-DYNA. The results show that the smaller is the deadrise
angle, the larger is the pressure coefficient. The pressure distributions for wedges with
deadrise angle varying from 10º to 80º are presented as well, from which can be concluded
that the maximum pressure occurs near the spray root for the small deadrise angle, while it
moves to the keel of the wedge for larger deadrise angles.

Taking the deadrise angle 30º and 60º for examples, the pressure distributions on wetted
surface of wedges at different time instances are presented, and the free surface elevations
are included. The predictions of LS-DYNA can correctly describe the variation of pressure
along the surfaces of wedges from the initial stage, flow separation, to the end. For a small
deadrise angle, the maximum pressures are located near the spray root of the jet before flow
separation occurs, after the flow separation, the maximum pressures move to the keel of the
wedge, while the maximum pressures are located on the keel of the wedge during the entire
impact when the deadrise angle is large.

54
Finally, the maximum slamming forces for symmetric wedges with different deadrise angles
show that the larger is the deadrise angle, the smaller is the peak value of slamming force.

55
56
Chapter 6 Slamming load on a two-dimensional
bow-flare section during water entry

6.1 Introduction
In this chapter, the FE simulation extends to a rigid bow-flare section. The bow-flare section
with vertical velocity is modeled based on the experiment by Zhao and Faltisen (1996).
Predicted vertical slamming force, pressure distributions at different time instances and
pressure histories are compared with the available experimental data. Problems related to
fluid leakage through the structure may occur in the simulation of the water entry problem of
the bow-flare section, which is also discussed. For general problems, one solution to solve
these problems is to reduce the time step (Aquelet N. et al. 2006), however, it is a very
difficult problem to solve. Furthermore, the sensitivity in the selection of several related
parameters, such as penalty factor (PFAC), coupling points (NQAUD) and mesh size, are
studied.

6.2 Description of test


In this work, the main data for the bow-flare section are based on the drop tests that were
carried out by Zhao and Faltinsen (1993) at MARINTEK.

Force transducer

mm
25

m
0m
10 m
0m
203mm

15

Figure 6.1. Locations of gauges in bow flare section (Zhao et al. 1996).

Table 6.1 displays the main data of the bow-flare section which are applied in the numerical
simulations. As seen in Table 6.1, though the total length of the test section is 1.0 m, the
measuring section is only 0.1 m. The impact velocity is determined by the drop height of the
section.

57
Table 6.1. The main data of the test sections
Data items Bow flare
Breadth of section 0.32 m
Length of measuring section 0.10 m
Length of each dummy section 0.45 m
Total length 1.00 m
Total weight 261 kg
Weight of measuring section 6.9 kg

6.3 Description of finite element model


The concept of modeling is similar to that of wedge. The water entry model of a bow-flare
section is schematic as Figure 6.2. Only half of the section is modeled. Correspondingly, the
mesh detail in the area around the bow-flare section is shown in Figure 6.3.

Domain A in Figure 6.2 represents the uniformly meshed area in Figure 6.3 which should
satisfy that the finer the mesh of it, the larger size of the area is required. Except for this
domain, the mesh of other parts is moderately expanded towards the boundaries.

Figure 6.2. Water entry model of bow flare section.

Two-dimensionality is applied by fixing all nodes in the z-direction and assuring that the model
has an only one element in the z -direction. The boundaries of fluids are defined as non-
reflecting in order to decrease the effect of the reflection of acoustic wave. Furthermore,
pressures are obtained by applying sensors on the center of each shell elements at the
coupling surface.

The initial velocity is -2.43m/s in the y -directions and the mass of the model is proportional to

the measured sections in the test of Zhao (1996). The predicted accelerations are
proportional to the slamming force on the measuring section in the test. The instant when the
bow-flare section touches the element on the water surface is set as 0.0s.

58
Figure 6.3. Meshing of the water entry model of 2D bow-flare section.

Although the simulations of slam induced loads on the wedge section and the bow-flare
section are similar, the latter is more complicate due to its irregular body shape. Furthermore,
the water entry process of the bow-flare section requires much more time because of the
height of the section. Therefore, the conclusions got from the analysis of the wedge may be
not correct in the case of bow-flare section. For example, the gravitational acceleration has
limited effect on the predicted slamming loads on the rigid wedge, however, the initial velocity
of bow-flare section is much less than the wedge, which may lead to different effect.

A proper selection of mesh size is required for proper modeling of the slamming problems
(Luo et al. 2011). Generally speaking, the finer the mesh, the better the results are. However,
the simulation of fluid-structure interaction is time-consuming. Therefore, different mesh
densities are chosen to simulate the slamming loads on the wetted surface of the bow-flare
section, to obtain a balance between the precision and efficiency of the calculation.

Similarly to the wedge, three models with different mesh sizes for the uniformly meshed area
are used as well. As shown in Table 6.2, the primary data are listed. Based on these models,
the effect of gravity will be discussed at first in the following section.

Table 6.2. The main data for three models of a wedge with different mesh densities.
Parameters Model 1 Model 2 Model 3
Mesh size 5mm 2.5mm 2mm
Number of elements (Fluids + Structures) 4200+55 13550+110 58560+136
Number of elements on impacting wedge surface 55 110 136
CPU time 1h 41m 16 h 30 m 89h 43m

6.4 Results and Discussion


6.4.1 Gravitational acceleration of bow-flare section

Taking the computational time into consideration, the size of the water domain is limited to
700 mm+700 mm, while that of air domain is 700mm+250mm. Assume that the mesh sizes of
bow-flare section and fluids are both 2.5mm, the model has an extension of 2.5mm in Z-

59
direction. Corresponding, the mass of the model is 0.326kg. In order to decrease the
computational time, the size of domain A is limited to 200mm*(200 (water) +100 (air)) mm.
The instant when the vertex of 2D wedge touches the element on the water surface is set as
0.0s.

According to Newton’s law, if the gravity is not considered, the predicted accelerations of the
bow-flare section are proportional to the slamming force on the measuring section in the test.
In contrast to that, the gravity will be added to the slamming force. Two different models are
analyzed. The gravitational acceleration of the section is taken into account in case 1, while it
is neglected in case 2. The comparison of the predicted slamming forces is shown in Figure
6.4, together with the experimental result.

As seen in Figure 6.4, the predicted peak slamming force of case 1 is about 8% larger than
the experimental result, while that of case 2 is about 30% smaller than measured value.
Furthermore, the time when slamming force comes up to maximum is close to that of
experiment which is about 0.06s after the bow flare section touching water for the case 1.
When comes to the case 2, the predictions delay the peak force. Therefore, the gravitational
acceleration of the bow-flare section is of great significance to the slamming loads. It means
that gravity must be taken into account in this analysis.

500
LS-DYNA case1
400 LS-DYNA case2
Slammming Force (N)

Expt.by Zhao.

300

200

100

0
0.02 0.03 0.04 0.05 0.06 0.07 0.08
Time (s)
Figure 6.4. Comparison of vertical slamming force on a symmetric bow-flare section between the model
considering gravity and that neglecting gravity.

6.4.2 Slamming force

Different mesh sizes are applied to analyze the slamming loads of the bow-flare section,
including vertical slamming force, pressure distributions at different time instances and
pressure histories on points P1-P4 in Figure 6.1. When the mesh size is 5mm, the model has
an extension of 5mm in Z-direction, and it means that the mass of the bow-flare section is
0.652kg. Similarly, when the mesh size is 2.5mm, the mass is 0.326kg, while it is 0.261kg for
the model of 2mm mesh size.

60
Figure 6.5 compares experimental slamming forces with predicted results with different mesh
sizes. In general, the predicted slamming forces have good agreement with the experimental
results. In the initial stage of the impact, the slamming force is small, and it increases fast at
the moment when the water jet impacts with the section. It comes up to the peak value near
the time instance of 0.06s, and then decreases gradually The predictions of the blow-flare
section are larger than measured values before flow separation, which are as same as that of
the wedge shown by Luo et al. (2011). Perhaps, it is due to three-dimensional effect on the
drop test.

600 LS-DYNA 5mm


LS-DYNA 2.5mm
Slammming Force (N)

500 LS-DYNA 2mm


Expt.by Zhao.
400

300

200

100

0
0.02 0.03 0.04 0.05 0.06 0.07 0.08
Time (s)
Figure 6.5. Comparison of experimental slamming forces and predicted results on bow-flare section for
different mesh sizes.

When the mesh size is 5mm, the predicted slamming forces from LS-DYNA are somehow
larger than the measured results in the initial stage of the hull-water impact, while they are not
in good agreement with measured results from 0.04s to 0.05s, especially near 0.046s when
there is an impulse. Some high frequent oscillations exist around 0.05s and obviously in the
middle of water entry. In the later stage of water entry, the predicted values are much larger
than measured ones.

When the mesh size is 2.5mm, the predicted results are in better agreement with
experimental values, though there are some frequent oscillations as well. In the initial stage of
the impact, the predictions from LS-DYNA are smaller than measured value. There is an
impulse near 0.04s as well, after which the predicted results are larger than measured values.
It is because the mesh size is not finer enough, because the impulse becomes smaller when
the mesh size is finer.

When the mesh size is 2mm, the predicted results from LS-DYNA have great similarity with
the test values. As same as other models, an impulse exists in the initial stage of the impact,
which is not very obvious. Maybe, finer mesh can solve this problem. In the later stage, some
high frequent noises happen, which may be due to the free water elevation in this model.

61
6.4.3 Pressure distribution at different time instances

6.4.3.1 Compared to experimental results

Figure 6.6 shows the non-dimensional pressure distributions on the wetted surface of the
bow-flare section at different time instances, measured values on points P1-P4 (see in Figure
6.1) are included as well. Y is vertical coordinate on body surface, Yk is vertical coordinate of

the keel and Yd is the draft of the body. t = 0 corresponds to the time instance when the keel
touches the water surface.

6
LS-DYNA 5mm
5 LS-DYNA 2.5mm
LS-DYNA 2mm
4 Expt.by Zhao.
Cp

0
0 0.2 0.4 0.6 0.8 1
(Y-Yk)/Yd (a)
6
LS-DYNA 5mm
5 LS-DYNA 2.5mm
LS-DYNA 2.5mm
4
Expt.by Zhao.
Cp

0
0 0.2 0.4 0.6 0.8 1
(Y-Yk)/Yd (b)
7
LS-DYNA 5mm
6
LS-DYNA 2.5mm
5 LS-DYNA 2mm
Expt.by Zhao.
4
Cp

0
0 0.2 0.4 0.6 0.8 1
(Y-Yk)/Yd (c)
Figure 6.6. The pressure distributions at three time instances during water entry of a bow-flare section.
(a) t = 0.06 s ; (b) t = 0.07 s ; (c) t = 0.08s

62
The predicted results are in good agreement with measure values, especially on the locations
of P2 and P3 which are in the middle part of the section.

When the mesh size is 5mm, the values of P1 are much larger than the measured values,
especially at the time of 0.07s and 0.08s. It can be explained as that the mesh size of the
vertex of the sections is not finer enough. In the later stage of the water entry, some high
numerical noises exist around the vertex of the bow-flare section.

When the mesh size is 2.5mm, the predicted values are somehow larger than the measured
ones except that of P1 and P4 at the first time instance (see in Figure 6.6 (a)).

The simulated pressure distributions of the model with mesh size of 2mm have great similarity
with those of the model in which 2.5 mm is used, except for the pressure values near the keel
of the section. It indicates that finer mesh size has some effects on the pressures of the
locations near the keel when the mesh size is fine enough to describe the pressure
distribution.

6.4.3.2 Compared to theoretical methods

A comparison is made between the fully nonlinear solution and the finite element method in
Figure 6.7 which also shows the pressure distributions at three time instances for the bow-
flare section. There are good agreements between the results, although LS-DYNA gives
lower values near the spray root of water jet at the first time instance.

Figure 6.7 (d) compares the simulated slamming pressure distribution with that obtained by
the simplified and the fully nonlinear solution developed by Zhao and Faltinsen (1996). The
impact velocity of the section is constant. The time instance is near the time when the spray
root of water jet reaches the knuckle. They have quite good agreement at the lower half part
of the bow-flare section, while the value obtained from LS-DYNA is much lower than these
two numerical ones on the position near the spray root of water jet.

6.4.4 Pressure histories at different points

The pressure histories for bow-flare section at four measured locations of points P1-P4 are
plotted in Figure 6.8, together with the measured values at the three time instances. Due to
the complex geometry of the bow-flare section, the curves below are not as smooth as that of
the wedge-shape section.

According to the predicted pressure distributions for different model, the agreement of the
pressure value at point P1 is less satisfactory when mesh size is 5mm. Those numerical
noises indicate that mesh size of 5mm is not finer enough to obtain correct pressures.
Therefore, the pressure histories of model 3 are not included in this figure.

63
8 5
fully nonlinear solution fully nonlinear solution
6 LS-DYNA 2.5mm 4 LS-DYNA 2.5mm

3
Cp 4

Cp
2
2
1

0
0 0.2 0.4 0.6 0.8 1 0
0 0.2 0.4 0.6 0.8 1
(Y-Yk)/Yd (Y-Yk)/Yd
(a) (b)
5 8
fully nonlinear solution fully nonlinear solution
4 LS-DYNA 2.5mm simplified solution
6
LS-DYNA 2.5mm
3
Cp

Cp
4
2

1 2

0 0
0 0.2 0.4 0.6 0.8 1 -1 -0.5 0 0.5
(Y-Yk)/Yd (c) (Y-Yk)/Yd (d)
Figure 6.7 Comparison of pressure distributions between theoretical methods and LS-DYNA. (a(a)
t = 0.06 s ; (b) t = 0.07 s ; (c) t = 0.08s ;(d) pressure distribution for the bow-flare section with constant
impact velocity.

4
x 10 4
2 x 10
2
LS-DYNA 2.5mm
LS-DYNA 2.5mm
LS-DYNA 2mm
1.5 LS-DYNA 2mm
Expt.by Zhao. 1.5 Expt.by Zhao.
Pressure (Pa)

Pressure (Pa)

1
1

0.5
0.5

0 0
0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08 0.1
Time (s) Time (s)
(a) (b)
4
4
x 10
x 10 2
2
LS-DYNA 2.5mm
LS-DYNA 2.5mm
LS-DYNA 2mm LS-DYNA 2mm
1.5
1.5 Expt.by Zhao.
Expt.by Zhao.
Pressure (Pa)
Pressure (Pa)

1 1

0.5 0.5

0 0
0 0.02 0.04 0.06 0.08 0.1
0 0.02 0.04 0.06 0.08 0.1
Time (s)
(c) Time (s) (d)
Figure 6.8. Pressure histories at different points on the surface of bow-flare section. (a) P1; (b) P2; (c)
P3; (d) P4.

64
Generally speaking, the predicted results agree well with measured values when mesh size is
2.5mm. Usually, they are a little larger than the experimental ones, except at point P1 and P4
at the time instance of 0.06s as shown in Figure 6.8 (a) and (d). There are obvious impulses
at the beginning of the pressure histories at point P1, P3 and P4 in Figure 6.8 (a), (c) and (d).
It is due narrow shape of the lower part of the section.

When the mesh size is 2mm, the predicted pressure values are smaller than experimental
ones at point P1 as seen in Figure 6.8 (a). It also explains the differences between predicted
pressure distributions and experimental ones at P1 that shown in Figure 6.6. The pressure
histories on location of P2-P4 are in good agreement with the ones from model 2, except that
impulse dose not appear in the beginning of the curves of P3 and P4.

In general, there are lots of numerical noises in both of the model, which means that finer
mesh size is needed to smooth the curves neglecting the effect of computational time. If the
computational time is considered, the mesh size of 2.5mm is finer enough to predict the
general trend of the pressure history.

The corresponding predicted peak values, at the four points shown in Figure 6.8, are
presented in Table 6.3 which includes the predictions from FLUENT by Alexandru et al. (2007)
and the measured ones by the tests of Aarsnes (1996). The initial impact velocity of the
section is 2.43m/s.

Table 6.3. Predicted peak pressures by using LS-DYNA, FLUENT and Aarsnes (1996)’s test
Pressure from LS- Pressure from LS- Pressure from Measured
Point
DYNA(2.5mm) DYNA(2mm) FLUENT pressure
P1(kpa) 25.3 10.1 1302.0 33
P2(kpa) 13.5 15.9 7.0 12.5
P3(kpa) 17.8 15.7 10.6 14.0
P4(kpa) 17.7 14.3 14.7 17.0

Compared to the predictions from FLUENT, the results from LS-DYNA are much closer to the
measured one, especially the peak pressure on position P1. Generally speaking, the
predictions from LS-DYNA underestimate the peak pressures at position P1, while
overestimate those at positions P2 and P3. The peak pressure on position P4 depends on the
mesh size of the model, which means the mesh size has some effects on the predictions.

6.4.5 Pressure variations during water entry

The pressure distributions for different time instances when mesh size is 2.5mm are
presented in Figure 6.9, where the variable x is horizontal coordinate on the surface of the
bow-flare section, B is the half-width of the section. Figure 6.9 (a) shows the pressure
histories before flow separation, while Figure 6.9 plots those after flow separation. Therefore,
the pressure variation during the water entry can be observed in these figures.

65
As shown in Figure 6.9 (a), the maximum pressure is small and located in the keel of the bow-
flare section in the initial stage of the water entry, while it moves to the region near the spray
root of the water jet in the middle stage. Before flow separation, the pressures on the wetted
surface of the section increase with the variations of time instances.

As shown in Figure 6.9 (b), the pressure near the spray root of the water jet drops fast after
flow separation, while that near the keel of the bow-flare section increases, and the maximum
value of pressure coefficient moves to the keel of the section in the later stage of the water
entry.

4 4
x 10 x 10
2 2
0.04s 0.06s
0.045s 0.07s
1.5 0.05s 1.5 0.08s

Pressure (Pa)
0.06s
Pressure(Pa)

1 1

0.5 0.5

0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
x/B (a) x/B (b)
Figure 6.9. Pressure distributions at different time instance for the bow-flare section mode with mesh
size of 2.5mm.

4 4
x 10 x 10
2 2
0.04s 0.06s
0.045s 0.07s
1.5 1.5
0.05s 0.08s
Pressure (Pa)
Pressure (Pa)

0.06s

1 1

0.5 0.5

0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
x/B x/B
(a) (b)
Figure 6.10. Pressure distributions at different time instance for the bow-flare section model with mesh
size of 2mm.

The pressure distributions at different time instances for the model with mesh size of 2mm are
plotted in Figure 6.10, of which figure (a) means the predictions before flow separation, while
figure (b) represents those after flow separation.

The comparison of the pressure distribution at time instance of 0.04s between Figure 6.9 and
Figure 6.10 shows that the water jet touch the middle of the bow-flare section earlier when
finer mesh size is used. However, the variations of the pressure distributions for the two
models are similar, except for the pressure near the keel of the section, which means the

66
mesh size of the keel has some effects on the simulated slamming pressure. Furthermore,
there are more numerical noises in the curves from the model 3.

6.4.6 Free surface elevation and pressure contours at different time instances

Free surface elevation and pressure contours at different time instances for the bow-flare
section with mesh size of 2.5mm are shown in Figure 6.11.

Corresponding to the pressure distributions in Figure 6.9, the peak value is located near the
keel of the section before 0.04s after the impact. Then the free surface elevation reaches the
middle of the section and the peak pressure moves to the spray root of the water jet near the
time instance of 0.042s.

(a) (b)

(c) (d)

(e) (f)
Figure 6.11. Free surface elevation and pressure contours at different time instances during water entry
by LS-DYNA. (a) t = 0.04 s ; (b) t = 0.042 s ; (c) t = 0.05s ; (d) t = 0.06 s , (e) t = 0.07 s , (f) t = 0.08s .
t = 0 corresponds to that the keel touches the water surface.

67
As seen in Figure 6.5, the total slamming force is small in the initial stage, and then increases
gradually from 0.04s. It comes up to the peak value around 0.06s, and then decreases. When
the total slamming force is maximum value, the peak value of pressure is located near the
spray root of the water jet which can be seen in Figure 6.9 (a) and Figure 6.11 (d). After flow
separation, the pressure near the spray drops fast, and the peak value moves downwards
along the surface of the bow-flare section (see in Figure 6.11 (e)), meanwhile, the pressure of
the keel increase gradually. As seen in Figure 6.11 (f), the peak value of pressure moves to
the keel of the section in the later stage of the water entry.

6.5 Fluid leakage


Problems related to fluid leakage through the structure may occur for high velocity impact
problems. For general problems, one solution to solve these problems is to reduce the time
step. However, it is a complex problem and difficult to solve. In this work, the fluid leakage is
discussed by comparing several models with different related parameters.

Fluid leakage is the phenomenon that the fluid penetrates into structure during interacting
which can be seen in Figure 6.12(a). In such a case the fluid particle penetrates through the
structure that the coupling force in equation (3-18) is not large enough to return it to the
coupling interface. LS-DYNA Keyword User’s Manual (2007) indicates that penalty factor, the
number of coupling points, minimum volume fraction of a fluid to activate coupling and
coupling leakage control factor can be adjusted to prevent leakage. However, these
parameters do not work effectively for the slamming problem in this work.

Much work has been done on preventing the fluid leakage. It is found that advection method
for the remap step is in relationship with that. In general, it is better to begin an ALE analysis
with a Van Leer advection technique (ANSYS LS-DYNA user’s guide, 2009). In this work, the
fluid leakage is prevented by using the donor cell advection algorithm (METH=1) (see in
Figure 6.12(b)). However, fluid leakage through the structure is a very difficult and complicate
problem. It depends on the specific situation.

Figure 6.13 compares the predicted vertical slamming force from the model without fluid
leakage and that from the model with fluid leakage, and experimental results are included as
well. Though the fluid leakage begins at 0.04s after the impact, the results have limited
differences before 0.06s. Maybe, the fluid leakage is not too much in the initial stage.

There are much high frequent after 0.06s in the curve, however, the predicted total slamming
forces in the case of fluid leakage are similar to that from the model without leakage in the
middle stage. In the later stage, the fluid leakage extends to a large amount, so the slamming
force increases due to the large penetration through structure.

Therefore, the fluid leakage has small effect on the simulated results if it is limited to some
extend. Since it may lead to instability and high frequent oscillation in the simulation, it is
better to prevent the problem.

68
(a) (b)
Figure 6.12. Fluid leakage through the structure (a) Fluid leakage (b) leakage prevented

600
LS-DYNA 2.5mm
500 Fluid leakage
Slammming Force (N)

Expt.by Zhao.
400

300

200

100

0
0.02 0.03 0.04 0.05 0.06 0.07 0.08
Time (s)
Figure 6.13. Comparison of slamming force between the simulation with leakage and that without
leakage.

6.6 Parameters study


6.6.1 Mesh density

According to the results of slamming force, pressure distribution and pressure histories of the
bow-flare section for the three models with different mesh size. It can be concluded that the
finer the mesh size the better the results. Considering the computational time, the mesh size
of 2.5mm is suitable for the water entry model of the bow-flare section, though lots of
numerical noises exist in the curves. When the mesh size of 2mm is applied, the pressure
distribution has not much difference with that from the model with mesh size of 2.5mm, while
more than 4 times computational time is needed.

6.6.2 Penalty factor

The effect of the penalty factor is investigated by comparing the predicted slamming forces
and pressure distributions from the models with different penalty factors, which include 0.01,
0.1 and 0.5. As seen in Figure 6.14, where the simulated vertical slamming force is compared
for a penalty factor of 0.01, 0.1 and 0.5, the penalty factor has little effect in the initial stage of
the impact, and in the later stage, the effect is also small. When the value is 0.5, there are a
large amount of numerical noises on the curve in the last stage.

69
Figure 6.15 compares the simulated pressure distribution at three time instance. As shown,
the pressure near the middle of the section when penalty is 0.5 is large than that of the
models with the value of 0.01 and 0.1.

600
PFAC=0.01
500 PFAC=0.1
PFAC=0.5

Slamming load(N)
400

300

200

100

0
0.02 0.03 0.04 0.05 0.06 0.07 0.08
Time(s)

Figure 6.14. Predicted slamming force for the models with different penalty factors

6 6
PFAC=0.01 PFAC=0.01
5 PFAC=0.1 5 PFAC=0.1
PFAC=0.5 PFAC=0.5
4 4
Cp

Cp

3 3

2 2

1 1

0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
(Y-Yk)/Yd (a) (b) (Y-Yk)/Yd

Figure 6.15. Predicted pressure distributions for the models with different penalty factors. (a) t = 0.06 s ;
(b) t = 0.07 s ;

6.6.3 The number of coupling points

The value of coupling points was set as 3 for all the simulations above. The simulation with
one coupling points is studied and compared. The results show that, in this work, the number
of coupling points has little effect on the predictions of slamming loads on the wetted bow-
flare section, even on the problem of fluid leakage.

6.6.4 Time step

The smaller the time step factor, the more computational time is needed, however, the
smaller one can solve the problem of negative volume in this kind of simulation. Therefore, for
each model, different values of time step factor which is used to compute the time step are
applied in the simulations. The time step factor is set as 0.2 when the mesh size is 5mm and
2.5mm, while the value is 0.1 for the model with mesh size of 2mm.

70
6.7 Conclusions
The slam induced loads on a typical two-dimensional bow-flare section are evaluated by
using the explicit finite element method. The total vertical slamming force, pressure
distributions at different time instances and pressure histories are compared with the
experimental results from Zhao et al (1996).

In this chapter, the effect of the gravitational acceleration of the section is analyzed first. It
shows that, unlike the wedge, the gravity of the bow-flare section can not be neglected,
because the water entry process is longer and the impact velocity is less than that of the
wedge.

Three models with different mesh size are applied to predict the slam induced loads on the
wetted surface of the bow-flare section. Though the predicted slam induced loads have good
agreement with the published experimental results when the mesh size is finer enough, due
to the complexity of the body shape, more computational time is required to finish the
numerical simulation. For example, almost 90 hours is needed when the model with mesh
size of 2mm is used, however, the results from that model still exist lots of numerical noise.

Simulated pressure distributions and pressure contours show that the maximum pressure is
small and located in the keel of the bow-flare section in the initial stage of the water entry,
while it moves to the region near the spray root of the water jet in the middle stage. After flow
separation, the peak value moves downwards to the keel of the section.

Fluid leakage through the structure is a complicated problem and is difficult to prevent. In the
simulation of water entry for the bow-flare section, the problem is prevented by using the
donor cell advection algorithm instead of penalty factor and the number of coupling points.

The parameter study shows that the mesh density is most important factor for the predictions
of slam induced loads on the wetted surface of the bow-flare section. When the mesh size of
domain A and the section is 5mm, the simulated results are not satisfactory. The predictions
from the models with mesh size of 2mm and 2.5mm are acceptable, though including lots of
numerical noises. Combining the consideration of computational time, the model, for which
the 2.5mm mesh size is used, is more practical.

As to penalty factor and the number of coupling points, they have little effect on the
predictions. The smaller the time step factor, the more computational time is needed, however,
the smaller one can solve the problem of negative volume in this kind of simulation. Therefore,
the value of time step factor should be set as the largest one which can prevent the problem
of negative volume.

71
72
Chapter 7 Slamming loads on a bow-flare section
with various roll angles

7.1 Introduction
In a severe sea state, the bow of a ship can be lifted out of the water and then drop to the
fluid with an inclined angle. This may lead to complicated loads on the bottom of the ship. In
order to study the two-dimensional water entry of a ship-like section with a roll angle, Aarsnes
(1996) carried out the drop tests, aiming at investigating the pressure distribution and the
impact force for a bow-flare section with different roll angles. The measured section is
illustrated in Figure 6.1.

In the drop tests, different roll angles, including 0°, 4.8°, 9.8°, 14.7°, 20.3° and 28.3°, are used
for this bow-flare section. The initial vertical impact velocity is determined by the drop height.
The main parameters for some cases were listed in Table 1 of Sun and Faltinsen (2008)
which compared the numerical calculations with the experimental results for different roll
angles.

The predicted impact force and pressure distributions by LS-DYNA are later compared with
the measured values and the BEM solution as well. Meanwhile, the effects of the roll angle
are discussed based on the predictions. Due to the complexity of the geometry, the flow
pattern on the left side of the bow-flare section is complicated. Furthermore, the velocity of the
flow around the bottom can be very high, which will be companied by very low pressure that
can result in ventilation, the secondary impact or air pocket. They will be discussed later
based on the predictions on the left side.

7.2 Description of finite element model


The finite element model setup of the asymmetric water entry of a two-dimensional bow-flare
section is shown in Figure 7.1. As mentioned before, the section drops vertically into the calm
water as a constant roll angle θ . The x -axis is located at the calm water surface, and the y -

axis is placed in the vertical line which includes the lowest point of the section. Obviously, z -
axis means the three-dimensional direction of the model. Since only two-dimensional water
impact is investigated, the size of the model in z-axis is set as one element length.

Correspondingly, the meshed model is shown in Figure 7.2. Only part of the mesh is
illustrated and the mesh of the air is not included, in order to illustrate the detail mesh around
the impact domain clearly. Taking the computational time into consideration, the size of the
water domain is limited to 700 mm+500 mm, while that of air domain is 700mm+300mm. As
seen in Figure 7.2, the impact domain is uniformly meshed, and except for this domain, the
mesh of other parts is moderately expanded towards the boundaries. The mesh sizes of the

73
uniformly meshed area and the bow-flare section are all 2.5mm, which means that the model
has an extension of 2.5mm in z-axis. The initial impact velocity of the body is defined as the
estimated value in the drop tests of Aarsnes (1996).

Figure 7.1. Model setup of the asymmetric water entry of a two-dimensional bow-flare section with
constant roll angle θ.

Figure 7.2. Part of the meshing of the water entry model for the bow-flare section with roll angle 9.8°.

7.3 Results and Discussion


7.3.1 Impact force

During the water impact, vertical impact on the bottom surface of the two-dimensional section
is given by:

ma = Fv − mg (7-1)

where, m is the measured mass of the section, a is the acceleration of the moving body, FV is

the vertical impact force, and g is the acceleration of gravity.

The simulated vertical impact forces for various roll angles, which are obtained by equation
(7-1), are compared with the experimental ones in Figure 7.3, together with the numerical

74
results by using BEM solution. The impact velocity and roll angle for each case are illustrated
in these figures.

As seen in these figures, the predictions from LS-DYNA are in good agreement with the
BEM’s calculations, while the difference between them and experimental results becomes
larger. In general, the vertical impact force is larger for a larger roll angle which means the
angle between the body line at the windward side and the calm water is smaller. However,
actually, the impact force does not increase so much until the roll angle is 20.3°. From the
results in Figure 7.3 (a), (c), (d) and (f) which have the same initial impact velocity and instinct
roll angle, it can be found that the impact force comes up to the maximum value earlier for a
larger roll angle.

For the maximum vertical force, the simulated results of this method are in good consistence
with the calculations of BEM, except that when the roll angle is 28.3°. However, the drop test
gave smaller peak values than these two methods except for the case with 4.8° roll angle
plotted in Figure 7.3. Besides, the time instance when the simulated force comes up to the
peak value is just a little bit earlier than that from BEM’s solution, while is much earlier than
the experimental one. The differences between the tests and the numerical methods are
probably due to the experimental errors which are discussed by Sun and Faltinsen (2008),
and other effects caused by three-dimensionalities and hydroelasticity.

200 200
Slamming load(N)

Aarsnes(1996) Aarsnes(1996)
Slamming load(N)

150 150
BEM BEM
100 LS-DYNA 100 LS-DYNA

50 θ=0º,v=-0.61m/s 50 θ=4.8º,v=-0.57m/s

0 0
0 0.05 0.1 0.15 0 0.05 0.1 0.15
Time(s) (a) Time(s) (b)

200 200
Slamming load(N)
Slamming load(N)

Aarsnes(1996) Aarsnes(1996)
150 150
BEM BEM
100 LS-DYNA 100 LS-DYNA
50 θ=9.8º,v=-0.61m/s 50 θ=14.7º,v=-0.61m/s
0 0
0 0.05 0.1 0.15 0 0.05 0.1 0.15
Time(s) (c) Time(s) (d)

300 400
Aarsnes(1996)
Slamming load(N)

Aarsnes(1996)
Slamming load(N)

300
BEM BEM
200
LS-DYNA 200 LS-DYNA

100 θ=28.3º,v=-0.61m/s
θ=20.3º,v=-0.75m/s 100

0 0
0 0.05 0.1 0.15 0 0.05 0.1 0.15
Time(s) Time(s)
(e) (f)
Figure 7.3. Vertical impact force for various roll angles

75
As mentioned in equation (7-1), the vertical slamming force is related to the gravity which is a
constant during the entire water impact. It means that the effect of gravity depends on the
value of acceleration of the bow-flare section. Correspond to the vertical force; the
acceleration of the section is larger for a larger roll angle. Therefore, the gravity effects
become more apparent for a bow-flare section with a smaller roll angle which corresponds to
a larger angle between the windward flare surface and the calm water surface. This is
consistent with the fact proposed by Fairlie-Clarke and Tveitnes (2007) that the gravity effects
are greater for a wedge with a larger deadrise angle.

Furthermore, as seen in Figure 7.3, the impact forces keep around a value between 100N
and 200N for all cases in the later stage of the impact. It can be concluded that the gravity
plays an important role during this period.

Figure 7.4 shows the comparison of the impact force for different impact velocity. The bow-
flare section with roll angle of 0°, 9.8° and 28.3° are analyzed. For each case, three different
drop velocities of 0.5m/s, 1.5m/s and 2.43m/s, are chosen. As expected, the maximum value
of vertical impact force is larger for a higher velocity when the roll angle is constant, and it
happens earlier for a larger roll angle. As seen in Figure 7.4 (c), the peak value increases
faster compared to other two cases. It shows that the impact force is more sensitive to the
impact velocity when the roll angle is large.

500 500
0.61m/s 0.61m/s
400 400
Slamming load (N)

1.50m/s
Slamming load (N)

1.50m/s
2.43m/s
2.43m/s 300
300

200 200

100 100

0 0
0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2
Time (s) (a) Time (s) (b)
1000
0.61m/s
800
Slamming load (N)

1.50m/s
2.43m/s
600

400

200

0
0 0.05 0.1 0.15 0.2
Time (s) (c)
Figure 7.4. Vertical impact force for different impact velocity. (a) θ = 0 ; (b) θ = 9.8 o ; (c) θ = 28.3 o . o

76
On the other hand, as seen in Figure 7.4 (b), there are several impulses in the curves after
the first peak value. It seems that the secondary impact happens between the left part of the
section and the elevated water. When the roll angle is 9.8°, the impulse caused by the
secondary impact of the section is higher than the first peak value for the cases with an
impact velocity of 1.5m/s and 2.43m/s. We can find that that the effects of the secondary
impact on the vertical force become more apparent for a higher impact velocity. Furthermore,
the secondary impact occurs earlier for a higher impact velocity at a given roll angle. When
the roll angle is 28.3°, there is no obvious the secondary impulse in the curves, however it can
not easily shows that the secondary impact does not happen. Maybe it is just because the
effects of that are not important in this case. We cam discuss it later according to the profiles
of the elevated water.

7.3.2 Impact velocity

0 0
Aarsnes(1996) Aarsnes(1996)
Vertical velocity (m/s)

Vertical velocity (m/s) BEM


BEM -0.5
-0.5
CIP CIP
-1 LS-DYNA
LS-DYNA
-1

-1.5
-1.5
-2
-2
-0.05 0 0.05 0.1 0.15 0.2 -0.05 0 0.05 0.1 0.15 0.2
Time (s) Time (s)
(a) (b)

Figure 7.5. Comparison of impact velocity for various roll angles. (a) θ = 9.8 o , initial velocity
v = −0.61m / s ; (b) θ = 20.3 o , initial velocity v = −0.75m / s

The impact velocities of the bow-flare section with roll angle 9.8° and 20.3° are presented in
Figure 7.5. The predictions of this work are compared with the experimental and numerical
results, respectively published by Aarsnes (1996), Sun and Faltinsen (2008), and Zhu et al.
(2005). Though the curves of the impact force have lots of high frequencies, the ones of
impact velocity are smooth. At the first stage of the water impact, the four methods give very
similar result, but the experimental results become much larger than the values of other
methods in the later stage.

7.3.3 Pressure history

7.3.3.1 Points on the right side of the section

The predicted time histories of the measured points P1-P4 which are illustrated in Figure 6.1,
are plotted in Figure 7.6, together with the experimental and numerical results. For the
maximum value of the pressure, the predicted values by LS-DYNA agree well with the
measured ones and the numerical calculations on the locations of P2 and P3, but they are
usually smaller than the published ones on the locations of P1 and P4. Since P1 is at the

77
lowest part of the section, the under-estimated values on P1 may be due to the mesh size of
the model which was discussed in Shan et al. (2011). Correspond to the vertical force the
predicted peak pressure occurs earlier than the results of other two methods.

When the roll angle is 20.3° and 28.3°, the pressure on P1 become negative value in the later
stage as shown. The possible reasons for this, including the high velocity around the bottom
and the elastic ropes used in the test., were discussed by Sun and Faltinsen (2008). Since
that LS-DYNA gets rid of the negative pressure from the results in this work, here, we can not
determine if it has happened.

As known, negative pressure may lead to ventilation. Zhu et al. (2005) mentioned that

ventilation actually occurred in the case for θ = 28.3 o , but Sun and Faltinsen (2008) proposed
that it did not happen in the cases above, because the negative pressure on the left side have
not gone back to zero. In this work, the ventilation will be discussed later.

(a) θ = 0º , v = −0.61m / s
0.1 0.1
Aarsnes(1996) P1 P2
Aarsnes(1996)
Pressure (bar)
Pressure (bar)

BEM
BEM
0.05 LS-DYNA
0.05 LS-DYNA

0 0
0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2
Time (s)
Time (s)
0.1 0.1
P3 Aarsnes(1996) P4
Pressure (bar)

Aarsnes(1996)
Pressure (bar)

BEM BEM
0.05 LS-DYNA 0.05 LS-DYNA

0 0
0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2
Time (s) Time (s)

(b) θ = 9.8º , v = −0.61m / s


0.1
0.1
Aarsnes(1996) P1
Aarsnes(1996) P2
Pressure(bar)

BEM
Pressure(bar)

LS-DYNA BEM
0.05
0.05 LS-DYNA

0
0
0 0.05 0.1 0.15 0.2
0 0.05 0.1 0.15 0.2
Time(s)
Time(s)
0.1 0.1
Aarsnes(1996) P3 Aarsnes(1996) P4
Pressure(bar)
Pressure(bar)

BEM BEM
0.05 LS-DYNA 0.05 LS-DYNA

0 0
0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2
Time(s) Time(s)

78
(c) θ = 14.7 º , v = −0.61m / s
0.2 0.2
Aarsnes(1996) P1 Aarsnes(1996) P2

Pressure (bar)
0.15

Pressure (bar)
0.15
BEM BEM
0.1 LS-DYNA 0.1 LS-DYNA
0.05 0.05
0 0
0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2
Time (s) Time (s)
0.2 0.2
Aarsnes(1996) P3 Aarsnes(1996) P4

Pressure (bar)
0.15 0.15
Pressure (bar)

BEM BEM
0.1 LS-DYNA 0.1 LS-DYNA

0.05 0.05

0 0

0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2


Time (s) Time (s)

(d) θ = 20.3º , v = −0.75m / s


0.2 0.2
Aarsnes(1996) P1 Aarsnes(1996) P2
0.15
Pressure(bar)

0.15
Pressure(bar)

BEM BEM
0.1 0.1 LS-DYNA
LS-DYNA
0.05 0.05
0 0
0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2
Time(s) Time(s)
0.2 0.25
Aarsnes(1996) Aarsnes(1996) P4
0.15 P3 0.2
BEM BEM
Pressure(bar)
Pressure(bar)

LS-DYNA 0.15 LS-DYNA


0.1
0.1
0.05
0.05
0 0
0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2
Time(s) Time(s)

(e) θ = 28.3º , v = −0.61m / s


0.2
Aarsnes(1996) Aarsnes(1996) P2
0.1 0.15
Pressure(bar)

BEM
Pressure(bar)

BEM P1
LS-DYNA LS-DYNA
0.05 0.1

0 0.05

-0.05 0
0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2
Time(s) Time(s)
0.2
Aarsnes(1996) P3 Aarsnes(1996) P4
0.15 0.4
Pressure(bar)

BEM
Pressure(bar)

BEM
LS-DYNA 0.3 LS-DYNA
0.1
0.2
0.05
0.1
0
0
0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2
Time(s) Time(s)

Figure 7.6. Pressure histories on the points P1-P4 on the right side of the bow-flare section for different
5 -2
roll angles. (1 bar=10 N m )

79
7.3.3.2 Points on the left side of the section

Due to the symmetric geometry of the section, the points P1-P4 on the left side are defined as
the symmetric ones along the symmetric line. The predicted pressure histories are presented
in Figure 7.7, together with the numerical and experimental results, for θ = 9.8º and θ = 20.3º .

(a) θ = 9.8º , v = −0.61m / s


0.1 0.1
Aarsnes(1996) P1 Aarsnes(1996) P2

Pressure(bar)
Pressure(bar)

BEM BEM
0.05 LS-DYNA 0.05 LS-DYNA

0 0
0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2
Time(s) Time(s)

0.1 0.1
Aarsnes(1996) P3 Aarsnes(1996) P4
Pressure(bar)
Pressure(bar)

BEM BEM
LS-DYNA 0.05 LS-DYNA
0.05

0
0
0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2
Time(s) Time(s)

(b) θ = 20.3º , v = −0.75m / s


0.2 0.06
Aarsnes(1996) P1 Aarsnes(1996) P2
Pressure(bar)

Pressure(bar)

BEM 0.04 BEM


0.1 LS-DYNA LS-DYNA
0.02

0 0
0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2
Time(s) Time(s)

0.1 0.1
Aarsnes(1996) P3 Aarsnes(1996) P4
Pressure(bar)
Pressure(bar)

BEM BEM
0.05 LS-DYNA 0.05
LS-DYNA

0 0
0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2
Time(s) Time(s)

Figure 7.7. Pressure histories on the points P1-P4 on the left side of the bow-flare section for different
5 -2
roll angles. (1 bar=10 N m )

As shown, the pressures of P1 are similar to those on the right side of the section, however,
when the roll angle is 9.8º, an obvious increase happens in the late stage. The small impulse,
which can also be observed from the pressure histories of P2 and P3, is probably due to the

80
secondary impact on the left side. When the roll angle is 20.3º, the pressures on the left side
are smaller than those on the right side. In particular for P2, the predicted pressure by LS-
DYNA is nearly zero. It is because the ‘air pocket’ that created after the secondary impact on
the left side.

7.3.4 Pressure distribution

The predicted pressure distributions on the wetted surface of the bow-flare section with
different roll angles are plotted in Figure 7.8, together with the calculations from BEM. The
initial drop velocity and the roll angles are respectively illustrated in these figures. The time
corresponds to the instance when the simulated vertical force comes up to the maximum
value. In consistent with the variable of Fig.6 in Sun and Faltinsen (2009), the horizontal axis
is a curvilinear coordinate along the section surface, where s = 0 is at the keel of the section,
and the coordinates for s > 0 and s < 0 respectively correspond to the body surfaces on the
right and the left of the keel.

0.1 0.1
BEM θ=0º,v=-0.61m/s BEM θ=4.8º,v=-0.57m/s
0.08 0.08
LS-DYNA LS-DYNA
Pressure (bar)

Pressure (bar)

0.06 0.06

0.04 0.04

0.02 0.02

0 0
-0.3 -0.2 -0.1 0 0.1 0.2 0.3 -0.3 -0.2 -0.1 0 0.1 0.2 0.3
s (m) s (m)
0.15
BEM θ=9.8º,v=-0.61m/s BEM θ=14.7º,v=-0.61m/s
0.1
LS-DYNA LS-DYNA
Pressure (bar)

0.08
Pressure (bar)

0.1
0.06
0.04 0.05
0.02
0 0
-0.3 -0.2 -0.1 0 0.1 0.2 0.3 -0.3 -0.2 -0.1 0 0.1 0.2 0.3
s (m) s (m)

0.25
BEM θ=20.3º,v=-0.75m/s 0.6 BEM θ=28.3º,v=-0.61m/s
0.2 LS-DYNA
LS-DYNA
Pressure (bar)
Pressure (bar)

0.15 0.4

0.1
0.2
0.05

0 0
-0.3 -0.2 -0.1 0 0.1 0.2 0.3 -0.3 -0.2 -0.1 0 0.1 0.2 0.3
s (m) s (m)

Figure 7.8. Pressure distributions for different roll angles. (1 bar=105 N m-2)

81
When the roll angle is larger, the pressures on the right part are higher, while those on the left
part are lower. Compared to the BEM’s calculations, the predictions from LS-DYNA are lower,
especially for the maximum value at the right side for a large roll angle. As seen in Figure 7.6,
the simulated pressures of P1 are lower than the experimental and BEM’s results for all the
cases, which are consistence with the differences of the pressure at the keel ( s = 0 ) between
the results from LS-DYNA and BEM. On the other hand, the simulated pressures on the left
side are very low for the bow-flare section with a roll angle. It indicates that the free surface of
water does not reach the flare part of the section at this time instance for a roll angle larger
than 4.8° in this work.

7.3.5 Pressure variation

Figure 7.9 shows the predicted pressure distributions at different time instances for the cases
of a roll angle 14.7° and 20.3°. Only the right half part of the section is investigated here. The
variable in y -axis is given by (Y − Yk ) / Yd , where Y means the vertical coordinate on the

body surface, Yk is the vertical coordinate of the keel and Yd is the draft of the section when
then roll angle is 0°.In other words, this variable is related o the position on the section.
Obviously, 0 means the keel, 1 represents the highest point in the right part and -1 means the
highest part in the left part of the section.

(a) θ = 14.7 o , v = −0.61m / s


0.1 0.1
0.08s 0.12s
0.08 0.13s
0.09s
Pressure(bar)

0.14s
Pressure(bar)

0.11s 0.06
0.05
0.04

0.02

0
-1 -0.5 0 0.5 1 0
-1 -0.5 0 0.5 1
(Y-Yk)/Yd (Y-Yk)/Yd

(b) θ = 20.3 o , v = −0.75m / s


0.08
0.05s 0.12s
0.1
0.06 0.15s
Pressure(bar)

0.07s
Pressure(bar)

0.16s
0.08s
0.09s 0.04
0.05
0.02

0 0
-1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1
(Y-Yk)/Yd (Y-Yk)/Yd
5 -2
Figure 7.9. Pressure variations on a bow-flare section with different roll angles. (1 bar=10 N m )

82
For the evolution of pressure distribution on the right side, when the roll angle is 14.7°, the
maximum pressure on the wetted section is located at the keel of the body at the time
instance of 0.08s after the impact, as plotted in Figure 7.9 (a), and then it moves to the spray
root of the water jet. Before 0.11s, the pressure on each position of the section increases with
time. It seems that the flow separation occurs after the time instance of 0.11s, because after
that the pressure decreases fast, as seen the results of 0.12s. At the later stage of the impact,
the pressure distribution does not change too much with time and almost uniformly distributed
along the wetted bottom. When the roll angle is 20.3°, the pressure distributions are similar
with the results of the model with a roll angle of 14.7° before flow separation occurs. However,
at the later stage, the maximum pressure is located near the keel of the section, and it
increases with time. Maybe, it is due to the section impact at the left part of the body.

For the evolution of pressure distribution on the left side, the pressures on the left side are
very small for these two cases in the early stage of the water impact. When the roll angle is
14.7º, the pressures are increasing with time in the late stage, and the maximum value is
located in the middle part of the section. Even at t = 0.14 s the maximum pressure is larger
than the value on the right side. It is because that the flow separation occurs on the right side
but does not happen on the left side at that time instance. When the roll angle is 20.3º, the
maximum pressure on the left side is also located in the middle part of the section in the late
stage, and the pressures at t = 0.16 s are smaller than that at t = 0.15s , which means that the
flow separates from the section at t = 0.16 s . Furthermore, the pressures on the lower part of
the section are close to zero in the late stage. The possible reason is that the air pocket is
created during the water entry.

7.3.6 Water surface elevation and pressure contour

Correspond to the calculated pressure distributions at different time instances mentioned


above, the pressure contours at those time instances are plotted, together with the free
surface elevations, as seen in Figure 7.10. Since the simulated free surface elevation, to
some extend, depends on the mesh size, fluids domain and maybe some uncertain factors of
the model, it can not be predicted as precise as the real, but only be schematically estimated
for its evolutions during the water entry.

As plotted, the pressure contours at different time instances in Figure 7.10 are in good
consistent with the predicted pressure variations mentioned above. The secondary impact
that happens on the left part of the body is observed for both of these two cases. The
pressure variations on the left part can also be seen from these pressure contours, which
indicate that there is almost no pressure contribution on the left side in the initial stage of the
impact, but after the secondary impact, the pressure increases.

83
(a) θ = 14.7 o , v = −0.61m / s

(b) θ = 20.3 o , v = −0.75m / s

Figure 7.10. Free surface elevation and pressure contour for different roll angles.

7.3.7 Discussions about the secondary impact

The secondary impact, here, means the detached flow impact with the left side of the section.
Figure 7.11 shows the pressure contours and the free surface elevations at the instance when
the elevated water impact with the left side of the section for different roll angles. The results
for the cases with roll angles of 4.8°, 9.8°, 14.7° and 20.3° are presented, which indicate that
the secondary impact happens for any roll angle.

As mentioned before, the effects of the secondary impact are large when the roll angle is 9.8°,
while there is no obvious change of the vertical force for the 28.3° roll angle, as plotted in
Figure 7.4. The time instances included in these figures show that, as the roll angle increases,
the secondary impact happens later. When the roll angle is large, the secondary impact
happens in the later stage of the water impact, for which the relative velocity between the
body and the water is small; therefore, the force caused by the secondary impact is small as
well. When the roll angle is 4.8°, the effects of the secondary impact are also limit as seen in
Figure 7.3 (b). It is because that the force caused by the secondary impact is small compared
to the total vertical force on the body.

84
(a) (b)

(c) (d)
Figure 7.11. Pressure contours at the time instance when the secondary impact at the left part happens
for different roll angles. (a) θ = 4.8 o , v = −0.57 m / s , (b) θ = 9.8 o , v = −0.61m / s ,(c)
o o
θ = 14.7 , v = −0.61m / s ; (d) θ = 20.3 , v = −0.75m / s .

7.3.8 Discussions about the ventilation and air pocket on the left side

On the research of asymmetric impact of wedges, Xu et al. (1998) described two types of
impact. Type A flow occurred when there was small asymmetric and the water moved out
towards the chine on the both sides of the keel. Type B flow happened when there was large
symmetric and the flow separated from the hull at the keel on one side. As mentioned in
Judge et al. (2004), ‘Type B impact’ and ‘ventilation’ were used interchangeably, and both of
them implied a flow detachment at the keel of the wedge that may or may not eventually
produce reattachment to the body with an air pocket. From this point of view, ventilation does
not happen in present work, because the pressures of P1 in Figure 7.7 do not go to zero. As
seen from the pressure distribution in Figure 7.8, although the pressures on the keels of the
sections with a roll angle are very small, they are not zero.

On the other hand, although there is no ventilation, the secondary impact on the left side
happens as mentioned in section 7.3.7. As plotted in Figure 7.11, the air pocket is created in
the later stage of the impact or after the secondary impact occurs. It means that the flow
separates from the lower part of the left surface, but not the keel, and then touches the
section again. Furthermore, as the roll angle increases, the air pocket becomes larger and
moves towards the keel. Meanwhile, it is also the reason that the pressure value of P2 plotted
in Figure 7.7 (b) is close to zero. It shows that P2 is located in the air pocket when the roll
angle is 20.3º.

85
7.4 Conclusions
The slam induced loads on a bow-flare section with roll angle are evaluated by the explicit
finite element method, when it impacts on water with a vertical velocity. For the finite element
model of the bow-flare section with a roll angle, it is the same with the one in Chapter 6
except that full model has to be established due to its asymmetric shape.

The simulated vertical impact force, impact velocity and pressure histories on the positions of
P1-P4 are compared with experimental and numerical results. They are in very good
agreement, especially for the small roll angles.

The effects of the roll angle on the slamming load are investigated through the calculations for
different roll angles which include 0°, 4.8°, 9.8°, 14.7°, 20.3° and 28.3°. The results show that
the higher is the roll angle, the larger is the maximum vertical force is, and the gravity effects
become more apparent for a larger roll angle. For a constant roll angle, the maximum force is
more sensitive to the impact velocity when the roll angle is large.

The predicted pressure distributions on the wetted surface of the bow-flare section with
different roll angles are plotted at the time instance when the vertical impact force comes up
to the peak value. It‘s found that, when the roll angle is larger, the pressures on the right part
are higher, while those on the left part are lower.

Furthermore, the pressure variations and free surface elevations are presented as well. In the
initial of the water entry, the pressure distributions are similar for any case, but in the later
stage, the maximum pressure is located in the keel of the section and is increasing with time
when the roll angle is 20.3°.

From the schematically free surface elevations, we can see that the secondary impact
happens at the left side of the body for all the cases. When the roll angle is smaller, it occurs
earlier. Based on the predicted vertical force for different impact velocity, it can be concluded
that the effects of the secondary impact become more apparent for a higher velocity. In
present work, the effects on the vertical force are most obvious when the roll angle is 9.8°,
while they are limit for a small or a larger one. It means that the effects are related to the time
instance when the secondary impact occurs and the maximum value of the impact force.

At the left side of the sections, ventilation does not happen, but the flow separates from the
lower part and then reattaches the body. The process of detachment and reattachment
produce an air pocket between the body and the section surface in the left side. As the roll
angle increases, the air pocket becomes larger and moves towards the keel.

86
Chapter 8 Conclusions

Slam induced loads on two-dimensional wedges and ship sections are predicted using the commercial
code LS-DYNA. Multi-material Eulerian formulation, which is classed as part of the Arbitrary
Lagrangian-Eulerian (ALE) solver, and penalty coupling algorithm are used to describe the interaction
between the fluids and structures during water entry. A wedge with deadrise angle of 30° and a bow-
flare section, which are tested by Zhao et al. (1993) are simulated in this thesis which also discusses
the relationship between the slamming loads and the deadrise angle of the wedge.

The predictions of slamming loads on the wetted surface of the wedge with deadrise angle of 30°
agree well with available experimental results and analytical values, which validate the explicit finite
element method. The simulated vertical slamming force and pressure on the bottom are somehow
larger than the measured one due to the three-dimensional effects in the drop tests. Correspond to the
drop test of two-dimensional rigid wedge, the impact velocity is vary for this wedge, which means the
maximum pressure appears when the wedge touches the water according to the theory of hull-water
impact, and then decreases with the reducing velocity. This process is also presented by the predicted
pressure distributions at different time instances which are shown in Figure 4.10. Before flow
separation, the maximum pressure is located near the spray root of water jet, while it moves
downwards from the vertex to the keel after flow separation. Furthermore, the parameters study shows
that, the mesh density of the uniformly meshed area in the domain of fluids and the wedge are of great
significance on the predictions, while penalty factor and time step factor have limited effect on the
simulated results. When mesh size of 2.5mm is applied, the predictions of total slamming force and
pressure distributions are satisfactory, while the computational time is acceptable; however, finer
mesh size is required to describe the pressure histories on the bottom surface of the wedge.

In the simulations of slamming loads on symmetric wedges with various deadrise angles, pressure
histories are not included, so the mesh size of 2.5 mm is used considering the computational time.
The impact velocity is assumed constant in this part. The maximum pressure coefficients for the
wedges with deadrise angles ranging from 10° to 45° are computed by LS-DYNA, and the analytical
values from Wagner (1932), Chuang (1976), Ochi (1973) and Zhao’s similarity (1996) are covered as
well. The results show that the smaller is the deadrise angle, the larger is the maximum pressure
coefficient.

The pressure distributions for wedges with deadrise angle varying from 10º to 81º are calculated by
Wagner’s theory, similarity, simplified and boundary element methods, which are compared with the
predictions at one time instance from LS-DYNA. They have good agreement in the initial stage of
impact when the deadrise angle is small, but the agreement is less satisfactory in the region of the
peak value. When the deadrise angle is larger than 60°, the difference between analytical results and
simulated predictions becomes larger. In general, the maximum pressure occurs near the spray root

87
for the small deadrise angle, and it moves to the keel of the wedge for larger deadrise angles, while it
is uniformly distributed on the wetted surface when the deadrise angle is near 45°.

Taking the deadrise angle of 30º and 60º for examples, the pressure distributions on wetted surface of
wedges at different time instances are presented, and the free surface elevations are included. The
predictions of LS-DYNA can correctly describe the variation of pressure along the surfaces of wedges
from the initial stage, flow separation, to the end. For a small deadrise angle, the maximum pressure is
located near the spray root of the jet before flow separation occurs, after the flow separation, the
maximum pressure moves to the keel of the wedge, while the maximum pressure is located at the keel
of the wedge during the entire impact when the deadrise angle is large. The maximum slamming
forces for symmetric wedges with different deadrise angles show that, the larger is the deadrise angle,
the smaller is the peak value of slamming force.

The assessment of slam induce load extends to a two-dimensional rigid bow-flare section. Similarly,
the predicted vertical slamming force, pressure distribution and pressure histories on the wetted
surface of this section are compared with the experimental results. As same as the water entry model
of the wedge with deadrise angle of 30°, the impact velocity is varying with time. The simulated total
slamming force is a little large than the measured one due to the three-dimensional effects. Pressure
distributions and pressure contours show that the maximum pressure is small and located in the keel
of the bow-flare section in the initial stage of the water entry, while it moves to the region near the
spray root of the water jet in the middle stage. After flow separation, the peak value moves downwards
to the keel of the section. Parameters study shows that mesh density is the most important factor to
the simulated results of the bow-flare sections.

It must be noted that the gravity of wedge is neglected in the simulations of the water entry of wedge,
but for that of the bow-flare section, the gravity is taken into account, which is due to the longer
duration and less velocity of the bow-flare section. On the other hand, fluid leakage though structure
occurs in the latter case due to the complexity of the section shape, and it is prevented by donor cell
advection algorithm in this work, however, it may be related to other parameters.

Though it is concluded that the finer is the mesh size, the better is the predictions, from the
parameters study of the both sections, obviously, the simulation of the bow-flare section is more time-
consuming with the same mesh size of the wedge, while the predictions from the former are still not as
good as that from the latter. Moreover, the mesh size of the sections near the keels has great effects
on the simulated pressure of the locations near the keels.

With the smaller time step factor, the simulation of the water entry is more stable, while more
computational time is required. However, the problem of negative volume occurs when the time step
factor is large. Therefore, the proper value should be the largest one that can prevent the negative
volume in the modeling, which depends on the specific model.

Based on the parameter study for the finite element modeling a bow-flare section, the slam induced
loads on the same bow-flare section with different roll angles are predicted by LS-DYNA as well. The
water entry has only the vertical impact velocity in this work. The simulated vertical impact force,

88
impact velocity and pressure histories on the positions of P1-P4 are in very good agreement with the
numerical calculations of Sun and Faltinsen (2008) and Zhu et al.’s (2005), especially for the small roll
angles, while there are some differences between the numerical values and experimental results from
Aarsnes (1996), which were due to the experimental errors, three-dimensional and hydroelastic effects.

The effects of the roll angle on the slamming load are investigated. The predicted vertical force shows
that the higher is the roll angle, the larger is the maximum vertical force, and the gravity effects
become more apparent for a larger roll angle. In addition, the maximum force is more sensitive to the
impact velocity when the roll angle is large. It is also found that, when the roll angle is larger, the
pressures on the right part are higher, while those on the left part are lower. For different roll angle, the
evolution of the pressure distribution is distinct. In this work, the pressure distributions at different time
instances of two cases are presented. When the roll angle is 14.7º, the maximum pressure is located
at the keel of the right side of the section in the initial stage, and then it moves to the spray root of the
water jet in the middle stage, while it decreases and does not change too much in the late stage. On
the left side, the pressures are very small in the initial, however they become larger, and even larger
than those on the right side in the late stage. When the roll angle is 20.3º, the maximum pressure
moves to the keel of the right side of the section in the late stage, and the pressures on the left side
are also very small in the initial stage, while they become larger and decreases with time in the late
stage.

From the schematically free surface elevations, we can see that the secondary impact happens at the
left side of the body for all the cases. Its effects are related to the time instance when the secondary
impact occurs and the maximum value of the impact force. At the left side of the sections, ventilation
does not happen, but the flow separates from the lower part and then reattaches the body. The
process of detachment and reattachment produce an air pocket between the body and the section
surface on the left side. As the roll angle increases, the air pocket becomes larger and moves towards
the keel.

In all, when the roll angle is large, the slamming loads on the bow-flare section are much more
sensitive to other factors, and more difficult to predict precisely.

Through the assessment of the slam induced loads on both of a typical wedge and a bow-flare section,
it can be concluded that the modelling techniques described here can be applied to study the water
entry of two-dimensional structures, and to analyze other hydrodynamic impact problems.

Some of the work in this thesis has already been published in Luo et al (2011) and Wang et al (2011a),
while other results are ready and in the process of being submitted to Journals (Wang et al. 2011b,
Wang and Guedes Soares, 2011a,b)

89
90
Bibliography

Aarsnes J.V. (1996) Drop test with ship sections-effect of roll angle. Report 603834.00.01.
Norwegian Marine Technology Research Institute, Trondheim.

Alexandru I., Brizzolara S., Viviani M., Couty N., Donner R., Hermundstad O., Kukkanen T.,
Malenica S. and Termarel P. (2007). Comparison of experimental and numerical impact loads
on ship-like sections. Advancements in Marine Structures, Guedes Soares, C, and Das, P.K.,
(Eds), Taylor and Francis, UK, pp. 339-349.

Aquelet N., Souli M. and Olovsson L. (2006). Euler–Lagrange coupling with damping effects:
Application to slamming problems. Computer Methods in Applied Mechanics and Engineering.
Vol. 195, pp. 110–132.

Armand J.L. and Cointe R. (1987). Hydrodynamic impact analysis of a cylinder Proc. Fifth
inter. Offshore Mech. and Arctic Engng. Symp, Tokyo, Japan, Vol. 1, pp. 609-634.

Benjamin A. Tutt. and Anthony P. Taylor. (2004). The Use of LS-DYNA to Simulate the Water
Landing Characteristics of Space Vehicles. 8th International LS-DYNA Users Conference.

Bereznitski A. (2001). Slamming: the Role of Hydroelasticity. International Shipbuilding


Progress. Vol. 48, pp. 333-351.

Dobrovol’skaya Z.N. (1969). On some problems of similarity flow of fluids with a free surface.
Journal of Fluid Mechanics. Vol. 36, pp. 805-829.

E1-M. Yettou, Desrochers A. and Y. Champoux. (2007). A new analytical model for pressure
estimation of symmetrical water impact of a rigid wedge at variable velocities. Journal of Fluid
and Structures. Vol. 23, pp. 501-522.

Engle A. and Lewis R. (2003). A comparison of hydrodynamic impacts prediction methods


with two dimensional drop test data. Marine Structures. 16, Vol. 2, pp. 175–182.

Faltinsen O.M. (1999). Water entry of a wedge by hydroelastic orthotropic plate theory.
Journal of Ship Research, Sept., 43, Vol. 3, pp. 180-193.

Faltinsen O.M. and Chen, Z. M. (2005). A generalized Wagner method for three-dimensional
slamming, Journal of Ship Research. Vol. 49, 4, pp. 279-287.

Faltinsen, O.M., Landrini, M. and Greco, M. (2004). Slamming in marine application. Journal
of Engineering Mathematics. Vol. 48, pp. 187-217.

Judge C., Troesch A. and Perlin M. (2004). Initial water impact of a wedge at vertical and
oblique angles. Journal of Engineering Mathematics. Vol. 48, pp. 279-303.

91
Guedes Soares C., (1990). Effect of the heavy maneuvering on the wave induced vertical
bending moments in ship responses, Journal of ship Research. Vol. 34, 1, pp. 60-68.

Howison S.D., Ochendon J.R. and Wilson SK. (1991). Incompressible water-entry problems
at small deadrise angles. Journal of Fluid Mechanics. Vol. 222, pp. 215-230.

Korobkin A.A. and Scolan Y.-M. (2006). Three-dimensional theory of water impact. Part 2.
Linearized Wagner problem. Journal of Fluid Mechanics. Vol. 549, pp. 343-373.

Luo H.B., Hu J.J. and Guedes Soares C. (2010). Numerical simulation of hydroelastic
response of flat stiffened panels under slamming loads. Proceedings of the 29th International
Conference on Ocean, Offshore and Arctic Engineering (OMAE´10), 6-11 June, 2010,
Shanghai, China, ASME, Paper OMAE2010-20027.

Luo H.B., Wang S. and Guedes Soares C. 2011. Numerical prediction of slamming loads on
rigid wedge for water entry problem by an explicit finite element method. Advances in Marine
Structures, Guedes Soares, C, and Fricke, W., (Eds), Taylor & Francis, UK, pp: 41-47.

LS-DYNA, Keyword User’s Manual, Livermore Software Technology Corporation, Version 971,
May 2007.

Mei X.M., Liu Y.M. and Dick K.P. (1999). On the water impact of general two-dimensional
sections. Applied Ocean Research. Vol. 21, pp. 1-15.

Ochi M.K. and Motter L.E. (1973). Prediction of slamming characteristics and hull response
for ship design, Transactions SNAME. Vol. 81, pp. 144-190.

Peseux B., Gornet L. and Donguy B. (2005). Hydrodynamic impact: Numerical and
experimental investigations. Journal of Fluids and Structures. Vol. 21, pp. 277-303.

Ramos J. and Guedes Soares C. (1998). Vibratory response of ship hulls to wave impact
loads. International Shipbuilding Progress. Vol. 45(441), pp. 71-87.

Stavovy A.B. and Chuang S.L. (1976). Analytical determination of slamming pressures for
high speed vessels in waves, Journal of Ship Research. Vol. 20, pp. 190-198.

Sun H. and Faltinsen O.M. (2009). Water entry of a bow flare section with a roll angle. Journal
of Marine Science and Technology. Vol. 14, pp. 69-79.

Stenius I., Rosn A. and Kuttenkeuler J. (2006). Explicit FE-modeling of fluid-structure


interaction in hull-water impacts. International Shipbuilding Progress. Vol. 53, pp. 1031-121.

Sun H. and Faltinsen O.M. (2009). Water entry of a bow flare section with a roll angle. Journal
of Marine Science and Technology. Vol. 14, pp. 69-79.

Sun H. and Faltinsen O.M. (2007). The influence of gravity on the performance of planning
vessels in calm water, Journal of Engineering Math. Vol. 58, pp. 91-107.

von Kármàn T. (1929). The impact on seaplane floats during landing. National Advisory
Committee for Aeronautics. Technical note No. 321, pp. 309-313.

92
Wagner H. (1932). Uber Stossund Gleitvergange an der Oberflache von Flussigkeiten.
Zeitschrift fuer Angewandte Mathematik und Mechanik. Vol. 12, pp. 193–215.

Wang S., Luo H.B. and Guedes Soares C. (2011a). Explicit FE simulation of slamming load
on rigid wedge with various deadrise angles during water entry. 1st International Conference
on Maritime Technology and Engineering, in press

Wang S. Luo H.B. and Guedes Soares C. (2011b). Numerical prediction of slamming loads
on a bow-flare section during water entry. To be submitted

Wang S. and Guedes Soares C. (2011a). Water impact of symmetric wedges. To be


submitted

Wang S. and Guedes Soares C. (2011b). Slam-induced loads on a bow-flare section with
various roll angles. To be submitted

Wu G. X., Sun H. and He Y.S. (2004). Numerical simulation and experimental study of water
entry of a wedge in free fall motion. Journal of Fluids and Structures, 19, 3, 277-289.

Xu G.D. and Duan W.Y. (2009). Review of prediction techniques on hydrodynamic impact of
ships. Journal of Marine Science and Applications. Vol. 8, pp. 204-210

Xu. L, Troesch A.W. and Vorus W.S. (1998). Asymmetric vessel impact and planing
hydrodynamics. Journal of Ship Research. Vol. 42, pp. 187–198.

Zhao R. and Faltinsen O.M. (1993). Water Entry of Two-Dimensional Bodies. Journal of Fluid
Mechanics. Vol. 246, pp. 593-612.

Zhao R., Faltinsen O.M. and Aarsnes J.V. (1996). Water entry of arbitrary two-dimensional
sections with and without flow separation. Proc. 21st Symposium on Naval Hydrodynamics.
pp. 408-423.

Zhu X.Y., Faltinsen O.M. and Hu C.H. (2005). Water entry loads on heeled ship sections.
Proceedings of the 16th international conference on hydrodynamics in ship design, Gdansk.
pp. 407-416.

93

Anda mungkin juga menyukai