Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
Numerical and analytical simulations of in-shore ship collisions within the scope of A.D.N.
Regulations
Ye Pyae Sone Oo
Master Thesis
presented in partial fulfillment of the requirements for the double degree:
“Advanced Master in Naval Architecture” conferred by University of Liege "Master of Sciences in Applied Mechanics, specialization in Hydrodynamics,
Energetics and Propulsion” conferred by Ecole Centrale de Nantes
developed at L'Institut Catholique d'Arts et Métiers, Carquefou in the framework of the
“EMSHIP” Erasmus Mundus Master Course
in “Integrated Advanced Ship Design”
Ref. 159652-1-2009-1-BE-ERA MUNDUS-EMMC
Supervisor: Prof. Hervé Le Sourne, L’Institut Catholique d’Arts et Métiers, France.
Ing. Stéphane Paboeuf, Bureau Veritas, Nantes.
Reviewer: Prof. Philippe Rigo, University of Liège.
Nantes, February 2017
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
3
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
ABSTRACT
Nowadays, due to the continuous increase in the inland waterway navigation, there has been a
higher risk of collisions, groundings, and of other undesired events. Especially for inland ships
carrying dangerous goods, the consequences of ship collision may lead to serious economic
losses as well as environmental pollution. Therefore, such accidents must be prevented at all
cost, and even if they were to occur, the ship designer has to make sure that the risk of cargo
tank damage or the oil spillage is the minimum. For the inland vessel construction, the rules are
governed by A.D.N. Regulations that consist in determining the probability of cargo tank
rupture using Finite Element Analysis (FEA). However, such numerical approach is often time
consuming and very expensive, and thus, is usually prohibited in the preliminary design phases.
In this context, ICAM and Bureau Veritas have been involved in the development of a
simplified damage assessment tool called SHARP based on the super-element method. Some
validation tests have already been performed on ocean-going tanker and FPSO application
cases. Nevertheless, the validity of the tool still needs to be verified for in-shore ship
applications. Thus, the main purpose of this thesis is to compare and validate the results of
SHARP with Non-linear Finite Element Explicit Code, LS-DYNA, within the scope of A.D.N.
Regulations. Some of the early results, however, show that although SHARP can be applicable
in the place of LS-DYNA, the tool still requires some more developments. Therefore, another
purpose of this thesis is to investigate the discrepancies in more details and at the same time, to
make suggestions for the future development of the tool.
P 4 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
This page was left intentionally blank.
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
5
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
CONTENTS
ABSTRACT 3
LIST OF FIGURES 9
LIST OF TABLES 12
ABBREVIATIONS 15
1. INTRODUCTION 16
1.1 Background and Motivation 16
1.2 Objectives 17
1.3 Scope of the Thesis 18
2. A.D.N. REGULATIONS 19
2.1 General 19
2.2 Carriage of Dangerous Goods by Inland Waterways 19
2.3 Alternative Design Procedure and its Approaches 20
2.4 Determination of the Collision Energy Absorbing Capacity 24
2.4.1 General 24
2.4.2 Creating the Finite Element Models 24
2.4.3 Material Properties 25
2.4.4 Rupture Criteria 25
2.4.5 Friction Energy 26
3. LITERATURE REVIEW 27
3.1 Inland Navigation Accident Study 27
3.2 Existing Simplified Ship Collision Models and Associated Software 29
3.2.1 Internal Sub-models of Structural Mechanics 29
P 6 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
3.2.2 External Ship Dynamics Sub-models 31
3.2.3 Coupled Approach of Internal and External Sub-models 33
4. SHIP COLLISION THEORY 34
4.1 General 34
4.2 Finite Element Theory 34
4.2.1 General Equation (Le Sourne, 2015) 34
4.2.2 The Newmark Method (Le Sourne, 2015) 35
4.2.3 Explicit Scheme (Le Sourne, 2015) 36
4.2.4 Implicit Scheme (Le Sourne, 2015) 37
4.2.5 LS-DYNA and MCOL 37
4.2.6 Advantages and Disadvantages of using LS-DYNA/MCOL 39
4.3 Super Element Theory 40
4.3.1 General Equations 40
4.3.2 SHARP Tool 44
5. LS-DYNA AND SHARP SIMULATIONS 48
5.1 LS-DYNA/MCOL Simulation Procedures 48
5.1.1 General 48
5.1.2 Modelling and Meshing 49
5.1.3 Elements, Materials and Parts 51
5.1.4 Boundary Conditions 55
5.1.5 Collision Scenarios 57
5.1.6 Contact Type and Friction 61
5.2 SHARP/MCOL Simulation Procedures 62
5.2.1 General 62
5.2.2 Modelling and Meshing 62
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
7
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
5.2.3 Materials and Rupture Strain 67
5.2.4 Collision Scenarios 68
6. COMPARISON AND ANALYSIS 70
6.1 Simulations without Rupture Strain 70
6.1.1 Case 1 (V-shape bow : 90 deg : At web : Under deck of struck ship) 70
6.1.2 Case 2 (V-shape bow : 90 deg : Between webs : Mid-depth of struck ship) 74
6.1.3 Case 3 (Push barge bow : 55 deg : At web : Mid-depth of struck ship) 78
6.1.4 Case 4 (Push barge bow : 55 deg : At bulkhead : Above deck of struck ship) 83
6.1.5 Case 5 (V-shape bow : 90 deg : At web : Above deck of struck ship) 87
6.1.6 Overall Analysis 90
6.2 Simulations considering Rupture (A.D.N. Regulations) 91
6.2.1 Case 1 (V-shape bow : 90 deg : At web : Under deck of struck ship) 92
6.2.2 Case 2 (V-shape bow : 90 deg : Between webs : Mid-depth of struck ship) 94
6.2.3 Case 3 (Push barge bow : 55 deg : At web : Mid-depth of struck ship) 96
6.2.4 Case 4 (Push barge bow : 55 deg : At bulkhead : Above deck of struck ship) 98
6.2.5 Case 5 (V-shape bow : 90 deg : At web : Above deck of struck ship) 99
6.2.6 Overall Analysis 101
6.3 Additional Simulations with Modified Rupture Strains 102
6.3.1 Case 1 (V-shape bow : 90 deg : At web : Under deck of struck ship) 102
6.3.2 Case 2 (V-shape bow : 90 deg : Between webs : Mid-depth of struck ship) 106
6.3.3 Case 3 (Push barge bow : 55 deg : At web : Mid-depth of struck ship) 108
6.3.4 Case 4 (Push barge bow : 55 deg : At bulkhead : Above deck of struck ship) 111
6.3.5 Case 5 (V-shape bow : 90 deg : At web : Above deck of struck ship) 113
6.3.6 Overall Analysis 115
7. CONCLUSIONS AND RECOMMENDATIONS 116
P 8 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
ACKNOWLEDGEMENTS 118
REFERENCES 119
APPENDIX 123
A. CONVERGENCE TESTS FOR MCOL SUB-CYCLING 124
B. CASE SENSITIVITY OF RUPTURE STRAIN 127
B.1 Case 1 (V-shape bow : 90 deg : At web : Under deck of struck ship) 127
B.2 Case 2 (V-shape bow : 90 deg : Between webs : Mid-depth of struck ship) 128
B.3 Case 3 (Push barge : 55 deg : At web : Mid-depth of struck ship) 129
B.4 Case 4 (Push barge : 55 deg : At bulkhead : Above deck of struck ship) 130
B.5 Case 5 (V-shape bow : 90 deg : At web : Above deck of struck ship) 131
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
9
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
LIST OF FIGURES
Figure 1 Definition of vertical striking positions (A.D.N. Regulations, 2015) ........................ 21
Figure 2 Number of accidents in inland water transportation in 8 EU countries, 2013 (Eurostat,
2015) ......................................................................................................................................... 27
Figure 3 Inland navigation accidents for 8 EU member states (Eurostat, 2015) ..................... 28
Figure 4 Distribution of vessel types involved in ship collisions (Youssef, Kim and Paik, 2014)
.................................................................................................................................................. 28
Figure 5 Minorsky’s Correlation (Minorsky, 1959) ................................................................. 30
Figure 6 Full-scale collision experiment in Netherlands. Figure available from: (Zhang, 1999)
.................................................................................................................................................. 31
Figure 7 LS-DYNA/MCOL collision simulation system (Le Sourne et al., 2003) ................. 39
Figure 8 Comparison of LS-DYNA/MCOL simulation with reality (Le Sourne et al., 2003) 39
Figure 9 Illustrations of super-elements for perpendicular collisions (Le Sourne et al., 2012)42
Figure 10 Descriptions of super-elements for oblique collisions (Le Sourne et al., 2012) ...... 45
Figure 11 Graphical user interface of SHARP ......................................................................... 46
Figure 12 Workflow diagram of SHARP (Le Sourne et al., 2012) .......................................... 47
Figure 13 Flow-chart of LS-DYNA simulation process .......................................................... 48
Figure 14 Finite Element model of struck ship (Type C inland double hull tanker) ............... 49
Figure 15 Finite Element models of striking ship bows ........................................................... 50
Figure 16 Arbitrary stress-strain curve for elasto-plastic material ........................................... 52
Figure 17 Organization of keywords to define structural parts in LS-DYNA ......................... 54
Figure 18 Typical collision simulation in LS-DYNA (Case 1) ............................................... 58
Figure 19 Undesirable effect (Conceptual diagram) ................................................................ 59
Figure 20 Definitions of the vertical impact locations (A.D.N. Regulations, 2015) ............... 59
Figure 21 Scheldt estuary – Zeebrugge (available from: Google Map) ................................... 60
Figure 22 Simulation procedures for SHARP/MCOL ............................................................. 62
Figure 23 Complete SHARP model of struck ship .................................................................. 63
Figure 24 Side shell becoming two super-elements due to virtual deck (Body plan view) ..... 64
Figure 25 Comparison of side shell super-element with and without virtual deck (profile view)
.................................................................................................................................................. 64
Figure 26 Model of the striking ship bow in SHARP (Besnard, 2014) ................................... 65
Figure 27 Models of striking ships in SHARP ......................................................................... 66
Figure 28 Adjustments in the push barge bow position in SHARP (Top view) ...................... 66
P 10 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
Figure 29 Comparison of the push barge bow models between LS-DYNA and SHARP ....... 67
Figure 30 Stress-strain curves considered for LS-DYNA and SHARP simulations ................ 68
Figure 31 Impact locations on side shell in SHARP ................................................................ 69
Figure 32 Typical collision scenario modelled in SHARP (case 1) ......................................... 69
Figure 33 Extent of damage in LS-DYNA – Case 1 (without rupture strain) ......................... 71
Figure 34 Extent of damage in SHARP – Case 1 (without rupture strain) .............................. 72
Figure 35 Comparison of the results – Case 1 (without rupture strain) ................................... 72
Figure 36 Extent of damage in LS-DYNA – Case 2 (without rupture strain) ......................... 75
Figure 37 Extent of damage in SHARP – Case 2 (without rupture strain) .............................. 76
Figure 38 Comparison of the results – Case 2 (without rupture strain) ................................... 77
Figure 39 Extent of damage in LS-DYNA – Case 3 (without rupture strain) ......................... 79
Figure 40 Extent of damage in SHARP – Case 3 (without rupture strain) .............................. 80
Figure 41 Comparison of the results – Case 3 (without rupture strain) ................................... 81
Figure 42 Extent of damage in LS-DYNA – Case 4 (without rupture strain) ......................... 84
Figure 43 Extent of damage in SHARP – Case 4 (without rupture strain) .............................. 84
Figure 44 Comparison of the results – Case 4 (without rupture strain) ................................... 85
Figure 45 Extent of damage in LS-DYNA – Case 5 (without rupture strain) ......................... 88
Figure 46 Extent of damage in SHARP – Case 5 (without rupture strain) .............................. 88
Figure 47 Comparison of the results – Case 5 (without rupture strain) ................................... 89
Figure 48 Comparison of damage extent in LS-DYNA and SHARP – Case 1 (With rupture
strain) ........................................................................................................................................ 92
Figure 49 Comparison of the results – Case 1 (With rupture strain) ....................................... 93
Figure 50 Comparison of damage extent in LS-DYNA and SHARP – Case 2 (With rupture
strain) ........................................................................................................................................ 94
Figure 51 Comparison of the results – Case 2 (With rupture strain) ....................................... 95
Figure 52 Comparison of the crushing force in LS-DYNA and SHARP (Case 2) .................. 95
Figure 53 Comparison of damage extent in LS-DYNA and SHARP – Case 3 (With rupture
strain) ........................................................................................................................................ 96
Figure 54 Comparison of the results – Case 3 (With rupture strain) ....................................... 97
Figure 55 Comparison of damage extent in LS-DYNA and SHARP – Case 4 (With rupture
strain) ........................................................................................................................................ 98
Figure 56 Comparison of the results – Case 4 (With rupture strain) ....................................... 99
Figure 57 Comparison of damage extent in LS-DYNA and SHARP – Case 5 (With rupture
strain) ...................................................................................................................................... 100
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
11
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
Figure 58 Comparison of the results – Case 5 (With rupture strain) ..................................... 100
Figure 59 Comparison of the results– case 1 (with modified rupture strain) ......................... 103
Figure 60 View of the deformation of weather deck and bottom in LS-DYNA .................... 104
Figure 61 View of the activated elements in SHARP (weather deck has not been impacted at
all) ........................................................................................................................................... 104
Figure 62 Comparison of the results– case 2 (with modified rupture strain) ......................... 106
Figure 63 View of the bending of weather deck due to deformation of the side shell .......... 107
Figure 64 View of the activated super-elements in SHARP (The weather deck has not been
collided) .................................................................................................................................. 107
Figure 65 Comparison of the results– case 3 (with modified rupture strain) ......................... 109
Figure 66 View of the side shell which has ruptured being still there and resisting the collision
– case 3 ................................................................................................................................... 110
Figure 67 Comparison of the crushing resistance of the side shell between LS-DYNA and
SHARP ................................................................................................................................... 110
Figure 68 Comparison of the results – case 4 (with modified rupture strain) ........................ 112
Figure 69 Comparison of the results – case 5 (with modified rupture strain) ........................ 113
Figure A - 1 MCOL convergence test for deformation energy (case 2 - without rupture strain)
................................................................................................................................................ 124
Figure A - 2 MCOL convergence test for penetration (case 2 - without rupture strain) ........ 124
Figure A - 3 MCOL convergence test for deformation energy (case 3 - without rupture strain)
................................................................................................................................................ 125
Figure A - 4 MCOL convergence test for penetration (case 3 - without rupture strain) ........ 125
Figure A - 5 MCOL convergence test for deformation energy (case 4 - with rupture strain) 126
Figure A - 6 MCOL convergence test for deformation energy (case 4 - with rupture strain) 126
P 12 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
LIST OF TABLES
Table 1 Values for uniform strain and necking for shipbuilding steel (𝑅𝑒𝐻 ≤ 355 N/mm2) 26
Table 2 Main characteristics of the striking ships .................................................................... 50
Table 3 Material failure strain calculation ............................................................................... 53
Table 4 Convergence tests for MCOL sub-cycling steps ......................................................... 56
Table 5 Results of convergence tests for MCOL ..................................................................... 56
Table 6 Collision scenarios ...................................................................................................... 57
Table 7 Location of the impact points in LS-DYNA ............................................................... 57
Table 8 Draft combinations of struck and striking ships ......................................................... 59
Table 9 Definitions of the parameters used in the SHARP bow model (Besnard, 2014) ........ 65
Table 10 Results calculated by LS-DYNA – Case 1 (without rupture strain) ......................... 70
Table 11 Results calculated by SHARP – Case 1 (without rupture strain) .............................. 71
Table 12 Comparison of the results – Case 1 (without rupture strain) .................................... 72
Table 13 Comparison of energy absorption – Case 1 (without rupture strain) ........................ 73
Table 14 Results calculated by LS-DYNA – Case 2 (without rupture strain) ......................... 75
Table 15 Results calculated by SHARP – Case 2 (without rupture strain) .............................. 75
Table 16 Comparison of the results – Case 2 (without rupture strain) .................................... 76
Table 17 Comparison of energy absorption – Case 2 (without rupture strain) ........................ 77
Table 18 Results calculated by LS-DYNA – Case 3 (without rupture strain) ......................... 78
Table 19 Results calculated by SHARP – Case 3 (without rupture strain) .............................. 79
Table 20 Comparison of the results – Case 3 (without rupture strain) .................................... 80
Table 21 Comparison of energy absorption – Case 3 (without rupture strain) ........................ 82
Table 22 Results calculated by LS-DYNA – Case 4 (without rupture strain) ......................... 83
Table 23 Results calculated by SHARP – Case 4 (without rupture strain) .............................. 83
Table 24 Comparison of the results – Case 4 (without rupture strain) .................................... 84
Table 25 Comparison of energy absorption – Case 4 (without rupture strain) ........................ 86
Table 26 Results calculated by LS-DYNA – Case 5 (without rupture strain) ......................... 87
Table 27 Results calculated by SHARP – Case 5 (without rupture strain) .............................. 87
Table 28 Comparison of the results – Case 5 (without rupture strain) .................................... 88
Table 29 Comparison of energy absorption – Case 5 (without rupture strain) ........................ 89
Table 30 Summary of result discrepancy (cases without rupture strain) ................................. 91
Table 31 Comparison of the results – Case 1 (With rupture strain) ......................................... 92
Table 32 Comparison of the results – Case 2 (With rupture strain) ......................................... 94
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
13
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
Table 33 Comparison of the results – Case 3 (With rupture strain) ......................................... 97
Table 34 Comparison of the results – Case 4 (With rupture strain) ......................................... 98
Table 35 Comparison of the results – Case 5 (With rupture strain) ....................................... 100
Table 36 Summary of result discrepancy (cases with rupture strain) .................................... 102
Table 37 Different rupture strain values considered in SHARP ............................................ 102
Table 38 Comparison of the results – case 1 (with modified rupture strain) ......................... 103
Table 39 Comparison of the energy dissipation – case 1 (with modified rupture strain) ...... 105
Table 40 Comparison of the results – case 2 (with modified rupture strain) ......................... 106
Table 41 Comparison of the energy dissipation – case 2 (with modified rupture strain) ...... 108
Table 42 Comparison of the results – case 3 (with modified rupture strain) ......................... 108
Table 43 Comparison of the energy dissipation – case 3 (with modified rupture strain) ...... 111
Table 44 Comparison of the results – case 4 (with modified rupture strain) ......................... 111
Table 45 Comparison of the energy dissipation – case 4 (with rupture strain) ...................... 112
Table 46 Comparison of the results – case 5 (with modified rupture strain) ......................... 113
Table 47 Comparison of the energy dissipation – case 5 (with modified rupture strain) ...... 114
Table 48 Result discrepancy of the simulations (cases with modified rupture strain) ........... 115
Table B - 1 Evaluation of energy distribution using different rupture strains – Case 1 ......... 127
Table B - 2 Evaluation of energy distribution using different rupture strains – Case 2 ......... 128
Table B - 3 Evaluation of energy distribution using different rupture strains – Case 3 ......... 129
Table B - 4 Evaluation of energy distribution using different rupture strains – Case 4 ......... 130
Table B - 5 Evaluation of energy distribution using different rupture strains – Case 5 ......... 131
P 14 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
DECLARATION OF AUTHORSHIP
I declare that this thesis and the work presented in it are my own and have been generated by
me as the result of my own original research.
Where I have consulted the published work of others, this is always clearly attributed.
Where I have quoted from the work of others, the source is always given. With the exception
of such quotations, this thesis is entirely my own work.
I have acknowledged all main sources of help.
Where the thesis is based on work done by myself jointly with others, I have made clear exactly
what was done by others and what I have contributed myself.
This thesis contains no material that has been submitted previously, in whole or in part, for the
award of any other academic degree or diploma.
I cede copyright of the thesis in favour of l'Institut Catholique d'Arts et Métiers (ICAM).
Date: Signature
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
15
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
ABBREVIATIONS
3D : 3 Dimensions
A.D.N. : European Agreement concerning the International Carriage of
Dangerous Goods by Inland Waterways
BV : Bureau Veritas
CCNR : Central Commission for the Navigation of the Rhine
DOF : Degree of Freedom
DTU : Technical University of Denmark
EMSHIP : Erasmus Mundus Master Course in Advanced Ship Design
EU : European Unions
FE : Finite Element
FEA : Finite Element Analysis
FEM : Finite Element Method
FEMB : Finite Element Model Builder
GUI : Graphical User Interface
ICAM : l'Institut Catholique d'Arts et Métiers
IMO : International Maritime Organization
ITOPF : International Tanker Owner’s Pollution Federation
MARPOL : International Convention for the Prevention of Pollution from Ships
MCOL : Mitsubishi Collision
MIT : Massachusetts Institute of Technology
OPA : Oil Pollution Act
SE : Super-Element
SEM : Super-Element Method
SHARP : Ship Hazardous Aggression Research Program
SIMCOL : Simplified Collision Model
SNAME : Society of Naval Architects and Marine Engineers
SSC : Ship Structure Committee
UNECE : United Nations Economic Commission for Europe
P 16 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
1. INTRODUCTION
1.1 Background and Motivation
Recently, transportation of hazardous substances by sea has increased considerably, resulting
the risks of accidents such as collision and grounding to increase. Collision of ships, especially
the ones carrying dangerous goods, can not only pollute the environments but also lead to
serious economic losses as well as casualties. In the past, there had been a lot of catastrophes,
human life losses and tragic accidents due to ship collisions. Examples of such tragic accidents
include RMS Titanic (1912) which sank in North Atlantic Ocean owing to the collision with an
iceberg. This incident resulted in the deaths of more than 1500 people. In 1979, a Greek Oil
Tanker, SS Atlantic Express collided with another Oil Tanker, Aegean Captain and eventually
sank, having spilled 287,000 tons of crude oil into the Caribbean Sea. Moreover, the collision
accident of Philippine-registered passenger ferry, MV Doña Paz (1987), with the oil tanker, MT
Vector, was also known as one of the worst disasters in the history of maritime, amounting to
an estimated death of over 4300 people.
In addition, according to the oil spill statistics of International Tanker Owner’s Pollution
Federation (ITOPF, 2016), from 1970 to 2015, a large amount of oil spill (> 700 tons) is due to
collision and grounding, accounting for 30 % and 33 % respectively. Therefore, it is obvious
that collision can cause serious environmental pollution as well as human life losses. Reducing
the risks of such accidents as much as possible is one of the top priorities of any maritime
organizations, classification societies and ship designers.
International Maritime Organization (IMO) introduced a series of measures such as SOLAS
(Safety of Life at Sea), MARPOL (International Convention for the Prevention of Pollution from
Ships), OPA 90 (Oil Pollution Act, 1990), etc. to improve the safety at sea as well as to prevent
oil pollution due to tanker accidents. Additionally, A.D.N. Regulation which is the European
Agreement concerning the International Carriage of Dangerous Goods by Inland Waterways
also entered into force on 29th February 2008 for promoting the safety level for the carriage of
dangerous goods and for preventing any pollution resulting from any such accidents.
Many scientists and researchers have also conducted a series of experiments and published a
lot of research papers concerning with methods to assess ship impact damage. In this context,
one of the important questions is how to evaluate the crashworthiness of the vessels during the
pre-design stage. Although non-linear finite element methods can be applied, such approaches
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
17
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
are not very well suited for the early design stages as the modelling and simulation time are
unusually too high. Recently, a method known as Super-Element Method (SEM) has been
developed by (Lützen, Simonsen and Pedersen, 2000) to rapidly assess the collision damage of
a vessel. Such methods are very useful for optimization in the preliminary ship design phase as
the time and computational requirement is significantly lower than that of Finite Element
Analysis.
1.2 Objectives
There are two main objectives for this thesis.
The first objective is to validate the software SHARP by comparing the results with
non-linear finite element code, LS-DYNA for the application of inland vessels within
the scope of A.D.N. Regulations. According to the A.D.N. Regulations, the energy
absorption capacity of the colliding vessels needs to be calculated in order to determine
the probability of cargo tank rupture. Conventionally, Finite Element Analysis has been
applied for this step. However, it will require an immense amount of work and time in
order to make a complete collision analysis with FEM. And hence, the purpose of this
thesis is to see whether SHARP can replace conventional Finite Element (FE) software
since the simulation time of SHARP is significantly lower than that of an FE Software.
Unfortunately, some of the early comparison results show that the tool still needs some
additional improvement for inland vessel structures and thus, the second objective of
this thesis is to investigate the discrepancies presented between the two methodologies
and to make suggestions for the future improvement of the software.
To achieve both objectives mentioned above, various simulations of LS-DYNA and SHARP
will be made using the same collision scenarios and the obtained results will be compared. The
struck ship is a typical Type C inland tanker. Two striking ship bows, push barge bow and V-
shape bow, will be defined according to the geometries given by the A.D.N. Regulations.
Associated collision scenarios will also be defined in accordance with the A.D.N. Regulations.
P 18 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
1.3 Scope of the Thesis
This thesis will be comprised of the following main Sections.
Section 2 will present a brief review on A.D.N. Regulations. Important points within the
regulations will be highlighted and the alternative design approach will be described.
In Section 3, literature review on inland navigation accidents around European waterway will
be discussed along with the descriptions of the existing simplified models for ship collision
analysis.
The basic theories concerning with ship collision mechanisms will be studied in Section 4. The
Numerical approach (Finite Element) and the analytical approach (Super-element) which are
implemented in LS-DYNA and SHARP respectively will be elaborated in details.
In Section 5, applications of LS-DYNA and SHARP in ship collision analysis will be presented.
All simulation processes, modelling, assumptions, and different collision scenarios will be
defined in that section.
Section 6 will emphasize the validation of SHARP results with LS-DYNA/MCOL code. MCOL
is an external dynamic tool used to calculate the external dynamics of the colliding vessels
during impact, taking into account the hydrodynamic forces which apply to both ships. The
results will be compared in the form of deformation energy as well as the penetration damage.
Any discrepancies between the two methodologies will be investigated and analysed in details.
Moreover, possible development for improving SHARP tool will be suggested.
Section 7 will be about the conclusions and recommendations, exposing not only the advantages
but also the limitations of SHARP in comparison with LS-DYNA. Moreover, recommendations
regarding the future improvement of the SHARP tool will be particularised.
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
19
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
2. A.D.N. REGULATIONS
2.1 General
A.D.N. is the European Agreement concerning the International Carriage of Dangerous Goods
by Inland Waterways. It has been issued in Geneva on 26th May, 2000 during the occasion of a
Diplomatic Conference held under the joint auspices of the United Nations Economic
Commission for Europe (UNECE) and the Central Commission for the Navigation of Rhine
(CCNR). It contains all the requirements for the design and construction of inland vessels
involved during the transport of dangerous goods. It entered into force on 29th February, 2008.
Refer to (A.D.N. Regulations, 2015).
The objectives are:
To ensure a high level of safety for international carriage of dangerous goods by inland
waterways;
To effectively contribute to the protection of the environment by preventing any
pollution resulting from accidents or incidents during such carriage; and
To facilitate transport operations and promote international trade in dangerous goods.
2.2 Carriage of Dangerous Goods by Inland Waterways
Within the context of A.D.N. Regulations, the followings are the type of dangerous goods that
are allowed to be carried by inland navigation:
Gases compressed, liquefied or dissolved under pressure;
Flammable liquids;
Oxidizing substances;
Toxic substances;
Corrosive substances;
Miscellaneous dangerous substances and articles.
The vessels which carry these dangerous substances listed above have to be met with certain
criterion. The rules required for constructing such vessels are presented in Chapter 9.3 of
A.D.N. Regulations. It documents various design requirements such as materials, protections,
P 20 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
hold spaces, cargo tanks, ventilation, engine room, piping system, fire system, electrical
installations, ship stability, and so on.
The rules are also divided according to the type of vessels, namely, type G, C and N: (Refer to
A.D.N. Regulations 2015, Chapter 9.3)
Type G: tank vessel intended for the carriage of gases under pressure or under
refrigeration.
Type C: tank vessel intended for the carriage of liquids. (usually double hull with flush
deck)
Type N: tank vessel intended for the carriage of liquids. (usually with open or closed
cargo tanks)
2.3 Alternative Design Procedure and its Approaches
One of the interests of using alternative design approach is to check if it is possible to promote
structural crashworthiness while increasing the capacity of cargo tank. It is done by comparing
the risk of the cargo tank failure between conventional construction and alternative one. The
alternative design vessel will be fitted with the cargo tank whose capacity may exceed the
maximum allowable one (but not greater than 1000 m3). Also, the distance between the outer
side shell and the cargo tank may deviate from the minimum requirement except the fact that
the ship will be protected with a more crashworthy side structure. Within the scope of A.D.N.
Regulations, the risk of a more crashworthy construction (alternative design) should be equal
to or lower than the risk of a conventional construction. Only then, a higher safety may be
approved and the vessel will be permitted. Details about the alternative construction can be
found in A.D.N regulation 2015, Section 9.3.4.
Referring to A.D.N., the risk of cargo tank rupture due to ship collision can be described by
Equation (1) below:
𝑅 = 𝑃. 𝐶 ( 1 )
where
R : risk [m2];
P : probability of cargo tank rupture; and
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
21
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
C : consequence (measure of damage) of cargo tank rupture [m2].
The probability of cargo tank rupture P depends on the probability distribution of the available
collision energy which the structures of the struck ship can absorb without any damage to the
cargo tank. This probability can be reduced by improving the structural crashworthiness of the
struck ship. The consequence of cargo tank rupture C can be defined as an affected area around
the struck ship.
There are 13 steps to calculate the probability of cargo tank rupture and the associated collision
energy absorbing capacity. Only the summary of those steps will be described in this thesis.
Step 1 includes preparation of a reference design and the associated alternative design. The
reference design should have at least the same dimensions (length, width, depth, displacement)
as the alternative design. Both designs should comply with the minimum requirements of a
recognized classification society.
In step 2, the vertical and longitudinal collision locations are defined. The vertical locations are
determined by using minimum and maximum draughts of the colliding ships. Then, a number
of possible draught combinations can be represented by an enclosed rectangular area shown in
Figure 1. Each inclined line has the same draught difference. In this thesis, however, to assess
the possible maximum collision energy, only the highest points on each respective diagonal line
are selected for the analysis.
Figure 1 Definition of vertical striking positions (A.D.N. Regulations, 2015)
P 22 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
On the other hand, three locations are suggested to determine the longitudinal impact locations.
They are:
at bulkhead;
between two webs; and
at web.
Therefore, a total of 9 collision locations is required to be studied for Type C vessels.
In step 3, the weighting factors representing the relative probability of the typical impact
locations are defined. The total weighting factor for each collision location is the product of the
factor for vertical collision location by the factor for longitudinal collision location. The
assumptions for these weighting factors must be agreed by a recognized classification society.
In step 4, collision energy absorbing capacity is determined for each collision location defined
in the previous steps 2 and 3 with the use of Finite Element Analysis. This energy is the amount
of collision energy absorbed by the structure of the struck ship up to initial rupture of cargo
tank. The following two impact scenarios are considered:
Scenario I: Push barge bow with 55 degree collision angle
Scenario II: V-shape bow with 90 degree collision angle
In addition, the following assumptions are made for struck and striking vessels:
The struck vessel is considered at rest, while the striking ship has a constant speed of
10 m/s.
The bow of the striking ship is assumed to be rigid while the structure of the struck ship
is considered as a deformable one.
In total, 36 finite element computations have to be simulated corresponding to 9 impact
locations and 2 bow shapes, each case for reference design and alternative design.
In step 5, probability of exceedance is calculated using the collision energy absorbing capacity,
𝐸𝑙𝑜𝑐(𝑖). To calculate this probability, cumulative probability density functions (CPDF) are
provided at A.D.N. Section 9.3.4.3.1.5.6. It is given by Equation (2) as follows:
𝑃𝑥% = 𝐶1(𝐸𝑙𝑜𝑐(𝑖))3 + 𝐶2(𝐸𝑙𝑜𝑐(𝑖))2 + 𝐶3𝐸𝑙𝑜𝑐(𝑖) + 𝐶4 ( 2 )
where
Px% : probability of cargo tank rupture;
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
23
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
C1-4 : coefficients from table in A.D.N Section 9.3.4.3.1.5.6; and
Eloc(i): collision energy absorbing capacity (from FEA).
In step 6, weighted probabilities of the cargo tank rupture 𝑃𝑤𝑥% are calculated by multiplying
each cargo tank rupture probability 𝑃𝑥% by the weighting factors corresponding to the maximum
displacement of the vessel and the characteristic collision speed.
In step 7, the total probabilities of cargo tank rupture 𝑃loc(𝑖) are calculated by summing all
weighted probabilities of cargo tank rupture 𝑃𝑤𝑥% for each collision location considered.
In step 8, for both collision scenarios, the weighted total probabilities of cargo tank rupture
𝑃𝑤loc(𝑖) are calculated by the multiplication of the total probabilities of cargo tank rupture 𝑃𝑙𝑜𝑐(𝑖)
with the weighting factors 𝑤𝑓𝑙𝑜𝑐(𝑖).
The scenario specific total probabilities of cargo tank rupture 𝑃𝑠𝑐𝑒𝑛𝐼 and 𝑃𝑠𝑐𝑒𝑛𝐼𝐼 are calculated in
step 9. This is done according to the table given at A.D.N. Section 9.3.4.3.1.
Step 10 is the calculation of weighted value of the overall total probability of cargo tank rupture
𝑃𝑤 according to the following Equation (3):
𝑃𝑤 = 0.8 ∗ 𝑃𝑠𝑐𝑒𝑛𝐼 + 0.2 ∗ 𝑃𝑠𝑐𝑒𝑛𝐼𝐼 ( 3 )
At step 11, the overall probability of cargo tank rupture 𝑃𝑤 is denoted as Pn for the alternative
design and as Pr for the reference design.
At step 12, the ratio between the consequence Cn of a cargo tank rupture for the alternative
design and the consequence Cr of a cargo tank rupture for the reference design is determined
using the following Equation (4):
𝐶𝑛/𝐶𝑟 = 𝑉𝑛/𝑉𝑟 ( 4 )
where
Vn : maximum capacity of the largest cargo tank in the alternative design; and
Vr : maximum capacity of the largest cargo tank in the reference design.
Step 13 is the final step in which the ratio 𝑃𝑟⁄𝑃𝑛 is compared with the ratio 𝐶𝑛/𝐶𝑟. If 𝐶𝑛/𝐶𝑟 ≤
𝑃𝑟⁄𝑃𝑛, the alternative design is proved to be at least as safe as the reference design and the design
can be accepted.
P 24 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
2.4 Determination of the Collision Energy Absorbing Capacity
2.4.1 General
Finite Element Analysis (FEA) is used to determine the collision energy absorbing capacity.
The applied FE code should be capable of dealing with both geometrical and material non-
linear effects. Examples of the applicable codes include LS-DYNA, PAM-CRASH, ABAQUS
and so on. The calculations will be validated by a recognized classification society.
2.4.2 Creating the Finite Element Models
FE models should be developed for both reference design and alternative design. Each model
should be capable of capturing plastic deformations corresponding to the collision scenarios
considered. The section of the cargo area should be modelled under the supervision of a
recognized classification society.
All three translational degrees of freedom are to be restrained at both ends of the modelled
section. The global horizontal hull girder bending of the vessel is not considered in most
collision cases. Thus, for the evaluation of plastic deformation energy, only half beam of the
vessel needs to be considered with the constraint in the transverse displacements at the
centreline CL. After generating the FE model, it is imperative to perform a trial collision
calculation to make sure that there is no plastic deformation near the constraint boundaries or
else the model should be extended. The collided area of the structures should use a sufficiently
fine mesh, while a more coarse mesh is applied for the other parts of the model. The fineness
of the element mesh must be adequate to capture the realistic rupture of elements including
local folding deformations. The maximum element size used should not exceed 200 mm in the
collision areas. The ratio between the longer and the shorter shell element edge should be
smaller than the value of three. In addition, the ratio between element length and element
thickness must be greater than five. Other values shall be validated with the recognized
classification society.
Plate structures, such as shells, webs, stringers, etc. can be modelled as shell elements and
stiffeners as beam elements, also taking into account cut-outs and holes in the collision areas.
When making the FE simulations, the ‘node on segment penalty’ method shall be activated for
the contact option, for example, “contact_automatic_single_surface” in LS-DYNA, “self-
impacting” in PAM-CRASH and so on.
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
25
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
2.4.3 Material Properties
Due to the extreme behaviour of materials and structures during a collision, a true stress-strain
relations shall be used to define the materials as shown in the following Equation (5):
𝜎 = 𝐶. 휀𝑛 ( 5 )
where
𝑛 = ln(1 + 𝐴𝑔);
𝐶 = 𝑅𝑚. (𝑒
𝑛)𝑛;
Ag = the maximum uniform strain related to the ultimate tensile stress Rm; and
e = the natural logarithmic constant.
The values Ag and Rm can be obtained from tensile tests.
If only the ultimate tensile stress Rm is known, for shipbuilding steel with a yield stress ReH of
not greater than 355 N/mm2, the following formula shall be used to approximate the Ag value:
𝐴𝑔 =1
0.24 + 0.01395. 𝑅𝑚 ( 6 )
If both Ag and Rm values from the tensile test are difficult to obtain when starting the
calculations, then minimum values defined by the recognized classification society shall be
used. If shipbuilding steel has a yield stress exceeding 355 N/mm2 or materials other than
shipbuilding steel are used, then material properties should be defined in accordance with a
recognized classification society.
2.4.4 Rupture Criteria
In order to capture the initial rupture of an element in an FEA, a threshold failure strain value
needs to be defined. This predefined value of failure strain will be used as a rupture criteria to
compare with the strain calculated in the finite element. The calculated strain can be effective
plastic strain, principal strain, or for shell elements, the strain in the thickness direction of this
element. If this value exceeds the rupture criteria, the element shall be deleted from the FE
model and the deformation energy in this element will no longer change in the next calculation
steps. The following Equation (7), suggested by (Lehmann and Peschmann, 2002), shall be used
to determine the threshold rupture strain value:
P 26 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
휀𝑓(𝑙𝑒) = 휀𝑔 + 휀𝑒 .𝑡
𝑙𝑒 ( 7 )
where
휀𝑓 : rupture strain;
휀𝑔 : uniform strain;
휀𝑒 : necking;
𝑡 : plate thickness; and
𝑙𝑒 : individual element length.
The values of uniform strain and the necking for shipbuilding steel with a yield stress ReH of
not exceeding 355 N/mm2 shall be taken from the following table:
Table 1 Values for uniform strain and necking for shipbuilding steel (𝑅𝑒𝐻 ≤ 355 N/mm2)
stress states 1-D 2-D
휀𝑔 0.079 0.056
휀𝑒 0.76 0.54
element type beam shell plate
Other rupture criteria may be used with the validation from the recognized classification society
if adequate tests can be provided.
2.4.5 Friction Energy
The collision energy absorbing capacity is determined by the sum of internal energy and friction
energy. The friction coefficient μc is defined by the following Equation (8):
𝜇𝑐 = 𝐹𝐷 + (𝐹𝑆 − 𝐹𝐷). 𝑒−𝐷𝐶|𝑣𝑟𝑒𝑙| ( 8 )
where
𝜇𝑐 = Coulomb friction coefficient;
FD = Dynamic coefficient of friction = 0.1;
FS = Static coefficient of friction = 0.3;
DC = Exponential friction decay coefficient = 0.01; and
|𝑣𝑟𝑒𝑙| = relative friction velocity of contact surfaces.
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
27
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
3. LITERATURE REVIEW
3.1 Inland Navigation Accident Study
In the past, navigation using inland waterways was solely for the transportation of bulk cargo.
However, nowadays, inland waterways have been used for the transport of not only bulk cargo
but also containers, general and liquid cargo. Therefore, the increase in the inland waterways
navigation has resulted a considerable increase in the probability of collisions, grounding, and
of other undesired events. Regarding this aspect, (Vidan et al., 2012) has proposed a new
approach in order to increase the safety level of inland navigation.
According to Eurostat statistics, in 2013, there was a total of 144 inland water transportation
accidents in the 28 EU (European Unions) member states and 56 % of such accidents took place
in Romania while 17 % were in Austria (See Figure 2).
Figure 2 Number of accidents in inland water transportation in 8 EU countries, 2013 (Eurostat, 2015)
In addition, every year starting from 2005 to 2014, the highest number of accidents occurred in
Romania with the exception of 2010 during which there were 32 accidents in Romania and 38
in Hungary. As for Czech Republic, the number of accident cases decreased by 74 % from the
year 2005 to 2014 (from 23 in 2005 to 6 in 2014). (See Figure 3)
P 28 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
Figure 3 Inland navigation accidents for 8 EU member states (Eurostat, 2015)
Figure 4 Distribution of vessel types involved in ship collisions (Youssef, Kim and Paik, 2014)
Figure 4 shows different types of ships involved in collisions. As can be seen, most of the
vessels taking part in the collisions are tankers, bulk carriers, cargo vessels and containers.
Although there exists statistics of inland navigation accidents for some of the EU member
states, the available data still suffer from incompleteness. To recreate the accident scenarios,
surrounding accident conditions such as ship speed, loading condition, and environmental data
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
29
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
are necessary. (Youssef et al., 2014) also mentioned that major efforts are still needed to build
up a database of ship collision and grounding accidents.
3.2 Existing Simplified Ship Collision Models and Associated Software
Since 1950s, ship collision models have been developed as the ships were needed to transport
radioactive materials. According to (Ship Structure Committee (SSC), 2002), the collision
analysis can be classified into three sub-models:
Internal sub-models of structural mechanics;
External ship dynamic sub-models; and
Coupled approach of internal and external sub-models.
There are various existing models that use different sub-models and coupling approaches.
3.2.1 Internal Sub-models of Structural Mechanics
Experimental approach or the correlation of actual collision data has been applied in order to
evaluate the internal mechanics of ship collision. (Minorsky, 1959) was the first to attempt a
simplified formulation of the ship resistance to collision. His formula was based on the
investigation of 27 ship-ship collisions. From these collisions, he was able to derive a relation
between damaged volume of the steel structure and the absorbed energy: (Figure 5)
∆𝐾𝐸 = 47.2𝑅𝑇 + 32.7 ( 9 )
where
ΔKE : energy absorbed by the struck ship [MJ]; and
RT : Resistance factor or damaged volume of steel structure of struck ship [m3].
The advantage of Minorsky’s formulation is its simplicity. However, it was only valid for large
energy collisions. There was no influence of material properties, side structural arrangement
and deformation mode in the assessment of total absorbed energy.
P 30 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
Figure 5 Minorsky’s Correlation (Minorsky, 1959)
Many experiments have also been carried out to analyse ship collisions since the early 1960s.
For example, during the period of 1967 to 1976, Germany conducted 12 ship model tests. A 7-
year project on the prediction of tanker structural failure and oil spillage was carried out in
Japan from 1991 to 1997. The purpose was to analyse the dynamic process of structural damage
and the process of oil spill and water ingress through the damaged hull. A series of full-scale
collision experiments was carried out in Netherland in 1991 (Carlebur, 1995). See Figure 6.
So far the experimental results were proved to be quite accurate but their application was limited
by the necessity of expensive production of side structure and striking bow. Thus, a lot of
researchers and scientists tried to find various simplified ways to deal with the ship collision.
The usual approach is to decompose the struck ship into various substructures or components,
such as plates, stiffeners, web frames and panels, etc. In the paper of (Le Sourne et al., 2012),
it was mentioned that the individual structural members could be classified into three categories,
namely, the web girders, the side panels and the intersection elements. The theoretical models
of each of these components can be found in literature. Some examples of these documents
include (Wierzbicki, 1995), (Wang and Ohtsubo, 1997), (Amdahl, 1983), (Zhang, 1999),
(Zhang, 2002), (Simonsen, 1997) and so on.
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
31
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
Figure 6 Full-scale collision experiment in Netherlands. Figure available from: (Zhang, 1999)
Using the above mentioned simplified approaches, closed form analytical formulations of the
resistance of each unit may be derived and by combining all individual resistances, a global
evaluation of the ability of a ship to withstand an impact with another vessel was formulated.
3.2.2 External Ship Dynamics Sub-models
There are various approaches to deal with the ship external dynamic behaviour during collision.
The approaches may be classified into two categories:
One or two degree of freedom models; and
Three degree of freedom models.
3.2.2.1 One or two degree of freedom models
One degree of freedom model is the simplest approach to deal with external dynamic of ship
collision proposed by (Minorsky, 1959) in which the surge velocity of the striking ship and the
sway velocity of the struck ships are in the same direction. Additional hypothesis are that the
kinetic energy in the longitudinal direction of struck ship is small, the collision is assumed to
be fully inelastic and the rotations are neglected for both colliding ships. In this method, the
worst case scenario for the crashworthiness evaluation of the vessels has been considered while
keeping the conservation of momentum to derive the formula for the sway velocity of struck
P 32 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
ships. Using the velocities, the absorbed kinetic energy in the struck ship in transverse direction
can be expressed as follows:
∆𝐾𝐸 =𝑀𝐴 𝑀𝐵
2𝑀𝐴 + 1.43 𝑀𝐵 (𝑉𝐵 𝑠𝑖𝑛 ∅)2 ( 10 )
where
ΔKE : total energy absorbed in collision;
MA : mass of the struck ship;
MB : mass of the striking ship; and
VB : velocity of the striking ship at impact.
Another approach to deal with the dynamics of ship collision is the computer program named
DAMAGE which was developed at Massachusetts Institute of Technology (MIT). Additional
degree of freedom and yaw motion of struck ship has been considered. The software can be
used to predict structural damage for grounding or right angle ship-ship collisions with
deformable side and deformable bow. One main advantage of DAMAGE is its modern
graphical user interface (GUI) that allows for making the analysis of ship structural
crashworthiness with no particular background in that field. However, one of the drawbacks is
that DAMAGE cannot be applied for collision with oblique striking angles or for struck ships
with initial velocity.
3.2.2.2 Three degree of freedom models
Three degree of freedom model was developed by (Hutchison, 1986). He generalized the
Minorsky’s method by considering all horizontal degree of freedom; surge, sway and yaw. The
virtual mass matrices were developed for both struck and striking ships, including the added
mass terms. Using these matrices and the velocity vector, the kinetic energy and momentum of
the ships were determined. (Rawson et al., 1998) also used similar model. Recently, the
Simplified Collision Model (SIMCOL) has been developed with the support of the Society of
Naval Architects and Marine Engineers (SNAME) and Ship Structure Committee (SSC). It
provides a simultaneous time-stepping solution of external ship dynamics and internal
mechanics. (Pedersen and Zhang, 1998) also provides expressions for absorbed energy
uncoupled with internal mechanics. By analysing the motions and impulses around the impact
point, the absorbed kinetic energy for longitudinal and transverse directions relative to the
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
33
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
struck ship is derived with the assumption of small rotation and constant ratio of absorbed
plastic deformation energy. This approach has been used in the Technical University of
Denmark (DTU) collision simulation model.
3.2.3 Coupled Approach of Internal and External Sub-models
When internal mechanics and external dynamics couple, both effects are taken into account in
one simulation model. This approach provides more accurate result of ship collisions. It is,
however, very complicated to consider the simulation with large water domain which will
require a lot of time and challenges. So, a simple and computationally fast method that includes
all the significant phenomena is needed for the numerical simulation of external dynamics of
ship collision.
(Brown, 2002) proposed a two-dimensional coupled method and compared the calculated
deformation energy with that evaluated with the decoupled approach of (Pedersen and Zhang,
1998). (Pill and Tabri, 2011) presented a coupled approach for the simultaneous analysis of
inner mechanics and external dynamics with finite element (FE) code LS-DYNA. The obtained
simulated results were compared with experimental results and good agreement was achieved.
However, in their simulation method, the restoring forces, the gravity force and the
hydrodynamic damping were not included. The added masses also need to be considered for all
motion components to simulate a wider range of collision scenarios.
(Le Sourne et al., 2012) presented a user-friendly rapid damage assessment tool named SHARP
for ship collision. In this tool, the upper bound theorem was applied to calculate resistant forces
and internal energies of structural elements involved in the collision process. The crushing
forces are calculated based on super-element theory, allowing for the calculation of roll, yaw
and pitch moments at the centre of gravity of the colliding ships. The results are then transferred
to Mitsubishi Collision (MCOL). MCOL is an external dynamic program used to determine the
striking and struck ship dynamics by solving the hydrodynamic force equations and returns the
positions, velocities and accelerations of their centres of gravity to the internal mechanics
solver, also see (Le Sourne et al., 2001). With that approach based on 6 Degrees-of-Freedom
(DOF) movements, it is possible to take into account ships large rotations as well as gyroscopic
effects and hydrodynamic damping (wave and viscous damping).
P 34 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
4. SHIP COLLISION THEORY
4.1 General
When two ships collide, the kinetic energy possessed by their motions is partly transformed
into the deformation energy in the structural elements of striking and struck ships. The process
will continue until the speed of both vessels become equal. The main laws governing the
dynamics and kinematics during the process are the conservation of momentum and the
equilibrium of force and energy.
Generally, the mechanics of ship collision can be divided into two main parts:
Internal mechanics and
External dynamics.
The internal mechanics involves structural behaviour in the striking bow and the side structure
of the struck ship. The structural crashworthiness such as material yielding, crushing, folding,
fracture, and so on are evaluated. Deformations taking place are usually many times larger than
the structural thickness and the energy is mainly dissipated in relatively localized regions and
is usually in the form of an inelastic straining. On the other hand, the external dynamics deals
with the rigid body global motion of the colliding ships and their interaction with the
surrounding water. Refer to (Paik, 2007) for further reference.
As the purpose of this thesis is to compare the analysis results of SHARP with LS-DYNA, it is
very important to understand the basis of these two main approaches. SHARP is developed
using an analytical approach based on Super-Element Theory while LS-DYNA is developed
using a numerical approach based on Finite Element Theory. Thus, in this section, the basic
theories underlying Finite Element and Super-Element Methods will be explored.
4.2 Finite Element Theory
4.2.1 General Equation (Le Sourne, 2015)
A recent development in computing technology has made it possible to use finite element
analysis in ship collision problems which especially require to solve non-linear matrix equation
system. The dynamic equation system which involves movement of the structural nodes can be
represented by the following discrete Equation (11):
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
35
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
[𝑀]. {��(𝑡)} + [𝐶]. {��(𝑡)} + [𝐾]. {𝑢(𝑡)} = {𝐹(𝑡)} ( 11 )
where
{��(𝑡)}, {��(𝑡)} 𝑎𝑛𝑑 {𝑢(𝑡)} : nodal acceleration, velocity, and displacement of the nodal vector
matrix;
[𝑀] : structural mass matrix;
[𝐶] : damping matrix;
[𝐾] : stiffness matrix; and
{𝐹(𝑡)} : total force vector matrix.
4.2.2 The Newmark Method (Le Sourne, 2015)
To solve Equation (11), a direct integration method known as Newmark Method is applied. It
solves the solution at time step 𝑡𝑛+1 = 𝑡𝑛 + ∆𝑡 when the solution at time step 𝑡𝑛 is known.
The initial conditions are:
𝑢(0) = 𝑢0 and ��(0) = ��0.
To get the approximate displacement and velocity at the next time step 𝑡𝑛+1, Taylor’s series
developments are used as follows:
𝑀��𝑛+1 + 𝐶��𝑛+1 + 𝐾𝑢𝑛+1 = 𝐹𝑛+1 ( 12 )
𝑢𝑛+1 = 𝑢𝑛 + ∆𝑡��𝑛 +∆𝑡2
2[(1 − 2𝛽)��𝑛 + 2𝛽��𝑛+1] ( 13 )
��𝑛+1 = ��𝑛 + ∆𝑡[(1 − 𝛾)��𝑛 + 𝛾��𝑛+1] ( 14 )
Where 𝛽 and 𝛾 are Newmark’s constants that will determine solution system of Equation (12).
For example, when 𝛽 value is zero, the corresponding integration scheme is said to be explicit.
On the other hand, it is said to be implicit if 𝛽 is not equal to zero.
Displacement and velocity predictors are defined so that the solution depends only on the
known quantities calculated at time step 𝑡𝑛.
P 36 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
��𝑛+1 = 𝑢𝑛 + ∆𝑡��𝑛 +∆𝑡2
2(1 − 2𝛽)��𝑛 ( 15 )
��𝑛+1 = ��𝑛 + ∆𝑡(1 − 𝛾)��𝑛 ( 16 )
Then, by substituting those predictors into Equations (12), (13) and (14), the following systems
of equations are obtained:
(𝑀 + 𝛾∆𝑡𝐶 + 𝛽∆𝑡2𝐾)��𝑛+1 = 𝐹𝑛+1 − 𝐶��𝑛+1 − 𝐾��𝑛+1 ( 17 )
𝑢𝑛+1 = ��𝑛+1 + 𝛽∆𝑡2��𝑛+1 ( 18 )
��𝑛+1 = ��𝑛+1 + 𝛾∆𝑡��𝑛+1 ( 19 )
Thus, initial acceleration becomes: (𝑀 + 𝛾∆𝑡𝐶 + 𝛽∆𝑡2𝐾)��0 = 𝐹0 − 𝐶��0 − 𝐾𝑢0.
4.2.3 Explicit Scheme (Le Sourne, 2015)
The system is said to be explicit if the above Equation (17) is solved by using 𝛽 = 0.
Considering central difference scheme (𝛾 =1
2) with undamped structural system, the following
equations are derived:
𝑀��𝑛+1 = 𝐹𝑛+1𝑒𝑥𝑡 − 𝐹𝑛+1
𝑖𝑛𝑡 ( 20 )
un+1 = un+1 ( 21 )
un+1 = un +1
2∆tun+1 ( 22 )
where
𝐹𝑛+1𝑒𝑥𝑡 : the vector of the external loads applied to the structure at time 𝑡𝑛+1;
𝐹𝑛+1𝑖𝑛𝑡 : the vector of internal forces and can be expressed as:
𝐹𝑛+1𝑖𝑛𝑡 = 𝐾𝑢𝑛+1 = ∫ 𝐵𝑇(𝜎𝑛+1)𝑑𝑣
𝑉 ; where 휀 = 𝐵(𝑢).
In the case of an elastic problem, 𝜎𝑛+1 is calculated from the increment of the deformation (Δɛ)
between tn and tn+1. Then,
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
37
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
∆𝑢 = 𝑢𝑛+1 − 𝑢𝑛 = ∆𝑡��𝑛 +∆𝑡2
2��𝑛 ( 23 )
∆휀 = 𝐵(∆𝑢) ( 24 )
𝜎𝑛+1 = 𝜎𝑛 + 𝐻∆휀 ( 25 )
where H is the Hooke matrix.
If mass matrix M is diagonal, then the system is decoupled. This scheme is known as explicit
integration scheme and widely used in LS-DYNA, ABAQUS explicit, DYTRAN, etc.
4.2.4 Implicit Scheme (Le Sourne, 2015)
If 𝛽 is not equal to zero, then the equations can be generalized as follows:
(1
𝛽∆𝑡2𝑀 +
𝛾
𝛽∆𝑡𝐶 + 𝐾) 𝑢𝑛+1 = 𝐹𝑛+1 − 𝐶��𝑛+1 − 𝐾��𝑛+1 ( 26 )
��𝑛+1 =𝑢𝑛+1 − ��𝑛+1
𝛽∆𝑡2 ( 27 )
��𝑛+1 = ��𝑛+1 + 𝛾∆𝑡��𝑛+1 ( 28 )
In the above equations, the Newmark constants β = 0.25 and γ = 0.5 are usually applied and the
method is known as medium acceleration method. However, this scheme of solving the
equations needs the system matrix to be inverted and recalculated for each time step, so it can
sometimes require a huge computational effort.
In general, the implicit formulation is applied for slow speed dynamic problems while the
explicit scheme is more useful for high speed dynamic problems such as vehicles collisions.
However, the explicit solver is only conditionally stable and so a very small time step as well
as a large number of iterations are necessary in order to maintain the computational convergence
and stability.
4.2.5 LS-DYNA and MCOL
For the ship-ship collision analysis, LS-DYNA finite element explicit code can be utilized by
coupling with outer collision dynamics program MCOL. The main reason for using the explicit
P 38 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
methodology is that high level of non-linearity, contact, friction and rupture processes are
taking part in ship collision. Therefore, so as to reduce the required computational efforts and
to ease the convergence of the solutions, rather than implicit methods, explicit methodologies
are more suitable as mentioned by (Wu et al., 2004).
Figure 7 shows ship collision simulation using LS-DYNA/MCOL. In this type of simulation,
only the collision area of both ships are meshed and the remaining parts are modelled by using
rigid bodies described by inertia matrix and the centre of mass position. The hydrodynamic
matrices such as stiffness, inertia, and damping are prepared as inputs. The crushing forces are
first calculated in LS-DYNA using the methodologies explained previously. Then, the results
are transmitted to MCOL in which the rigid ship motion equations are solved as described by
Equation (29). The obtained new position 𝑥, velocity �� and acceleration �� of ship centres of
gravity are then transmitted from MCOL to LS-DYNA for the next integration time step. In this
way, it is possible to consider both external dynamics and internal mechanics for ship collision
problems. Referring to (Le Sourne et al., 2003)
[𝑀 + 𝑀∞]�� + 𝐺�� = 𝐹𝑊(𝑥) + 𝐹𝐻(𝑥) + 𝐹𝑉(𝑥) + 𝐹𝐶 ( 29 )
where
x : the earth-fixed position of the centre of mass of the ship;
𝑀 : the structural mass matrix;
𝑀∞ : the added mass matrix;
𝐺 : gyroscopic matrix;
𝐹𝑊 : wave damping force vector;
𝐹𝐻 : restoring force vector;
𝐹𝑉 : viscous force vector; and
𝐹𝐶 : contact force vector.
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
39
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
Figure 7 LS-DYNA/MCOL collision simulation system (Le Sourne et al., 2003)
4.2.6 Advantages and Disadvantages of using LS-DYNA/MCOL
The advantage of using LS-DYNA/MCOL is its relative accuracy and in some cases, it may
even replace the model experiments. For example, as can be seen in Figure 8, the LS-DYNA
simulation result is superposed on the photo of the real ship hull after collision and the accuracy
of the software can be clearly observed. On the other hand, one major drawback is that to obtain
reliable results, the mesh size should be particularly small. This makes the computation time to
be very long. As the structure of the ship is already a complicated construction, a large number
of elements are required for modelling and as a result, the computation may take several
hundreds of hours to complete.
Figure 8 Comparison of LS-DYNA/MCOL simulation with reality (Le Sourne et al., 2003)
P 40 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
4.3 Super Element Theory
4.3.1 General Equations
The basic idea of using super-element method consists of splitting the vessel into several
structural macro-components also known as super-elements. A closed-form expression giving
the resistant force as a function of the element indentation is used to characterize these elements.
During the collision process, the energy absorption, collision forces and the penetration of the
striking ship are evaluated along with the activation of the super-elements involved in the
process. This is done according to the upper-bound theorem which states that “if the work rate
of a system of applied loads during any kinematically admissible collapse of a structure is
equated to the corresponding internal energy dissipation rate, then that system of loads will
cause collapse of the structure.” (Jones, 1997).
Mathematically, the maximal force causing the collapse of a given super-element with volume
V can be expressed by equating the external energy and the internal energy rates as follows:
(Available from: Buldgen et al., 2012)
��𝑒𝑥𝑡 = 𝐹. �� ( 30 )
��𝑖𝑛𝑡 = ∭ 𝜎𝑖𝑗 . 𝜖��𝑗 . 𝑑𝑉
𝑉
( 31 )
From (30) and (31), upper-bound theorem writes :
𝐹. �� = ∭ 𝜎𝑖𝑗 . 𝜖��𝑗 . 𝑑𝑉
𝑉
( 32 )
where
��𝑒𝑥𝑡 : external energy rate;
��𝑖𝑛𝑡 : internal energy rate;
𝐹 : maximal force responsible for the collapse of a given super-element;
�� : penetrating speed of the striking ship;
𝜎𝑖𝑗 : stress tensor of the super-element; and
𝜖��𝑗 : strain rate tensor of the super-element.
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
41
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
To solve the Equation (32), the following hypotheses were made:
Materials of the elements are assumed to be perfectly rigid-plastic to avoid strain
hardening and strain rate effects.
The total internal energy rate is the sum of the contribution of bending and membrane
effects which are assumed to be completely uncoupled.
For a plate in a plane-stress, the bending effects are assumed to be confined in a certain
number m of plastic hinge lines.
For example, with the assumptions made above, the bending energy rate ��𝑏 and membrane
energy rate ��𝑚 of a plate with is impacted out of its plane can be determined with the help of
the following two formulae:
��𝑏 = 𝑀0 ∑ ��𝑘𝑙𝑘
𝑚
𝑘=1
��𝑚 =2𝜎0𝑡𝑝
√3 ∬ √𝜖11
2 + 𝜖222 + 𝜖12
2 + 𝜖11𝜖22 𝑑𝐴
𝐴
( 33 )
where,
𝑀0 : fully plastic bending moment;
𝐴 : area of the plate;
𝑡𝑝 : thickness of the plate;
��𝑘 : rotation of the hinge number k; and
𝑙𝑘 : length of the hinge number k.
Solving the above equations, however, still presents a number of challenges, one of which is
how to obtain the strain rate tensor. To calculate this, the displacement fields which are close
enough with those observed on impact trials or on numerical simulations are chosen. One
problem with the upper-bound method is that it may overestimate the resistance if the
displacement fields are not in good agreement with the reality.
To be able to apply the above method, the considered vessels need to be divided into simple
structures (the so-called super-elements) and displacements fields are chosen for each of them
according to the collision scenario to be simulated, namely, right angle collision and oblique
angle collision. As in the case of right angle collision simulation, the structure of the struck ship
can be represented by four types of super-elements as follows: (Lützen et al., 2000)
P 42 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
Super Element SE1: a rectangular plate simply supported on four edges and will be
subjected to an out-of-plane deformation during collision (See Figure 9a). For example,
inner and outer side plating and longitudinal bulkheads.
Super Element SE2: a rectangular plate simply supported on three edges with the last
one being free edge and will be subjected to an in-plane loading during collision. The
plate will deform like a concertina with successive folds until fracture along with the
tearing of supported edges (See Figure 9b). For example, decks, transverse bulkheads,
web girders, frames, bottom and inner-bottom.
Super Element SE3: a beam subjected to a perpendicular transverse force. It will
collapse in two different phases; firstly, the occurance of three plastic hinges and
secondly, behaviour like a plastic string (See Figure 9c). For example, small stiffeners
such as longitudinals.
Super Element SE4: X-T-L form intersections which will be crushed axially until they
are completely deformed along with their initial length during collision (See Figure 9d).
For example, the junctions of vertical and horizontal structural members.
(a) Plate subjected to out-of plane deformation (c) Beam impacted eccentrically
(b) Plate subjected to in-plane deformation
(d) X-T-L form intersections
Figure 9 Illustrations of super-elements for perpendicular collisions (Le Sourne et al., 2012)
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
43
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
For the oblique angle collision case, the structural elements are divided into six different super-
elements as follows (Buldgen et al., 2012):
Super Element SE1: a plate simply supported by four edges and submitted to an out-
of-plane impact with oblique angle (See Figure 10a). For example, side shell, inner side
shell, longitudinal bulkhead.
Super Element SE2: a vertical plate simply supported on three edges with the
remaining free edge. Collision is occurred on the free edge at an angle other than 90
degree (See Figure 10b). For example, transverse bulkhead.
Super Element SE3: this element is similar to SE2 except collision is occurred inside
the structure and the modes of deformation are different (See Figure 10c). For example,
transverse bulkhead, web girders, frames, etc.
Super Element SE4: beam element which is considered to be clamped at both ends
(See Figure 10d). For example, longitudinal stiffeners.
Super Element SE5: this element is absolutely similar to the X-T-L form intersections
already mentioned above. The only difference is that the collision angle is assumed to
be different from 90 degree (See Figure 10e). For example, junction of vertical and
horizontal structural members.
Super Element SE6: a horizontal plate, simply supported on three edges and free on
the last one. Structure is similar to vertical one considered in SE2 and SE3. Collision is
assumed to occur at the unsupported edge with a certain angle in the horizontal plane
(See Figure 10f). For example, weather deck.
With these super-elements mentioned above, the structural components of most typical ships
can be modelled and their individual behaviour can be sufficiently evaluated. Each of these
elements has a closed-form expression and it is possible to derive an analytical formulation for
the estimation of the collision resistance. More details of the formula derivations can be found
in (Buldgen et al., 2012).
Generally, throughout the derivation stages, the bow of the striking ship is assumed to be rigid.
However, as in the case of the struck ship side structure being more rigid, then the bow of the
striking ship can be assumed to be deformable with approval from a recognized classification
society. The method for the evaluation of bow crushing force was developed by (Simonsen and
P 44 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
Ocakli, 1999) which is based on a modification of (Amdahl, 1983). They considered that the
striking bow is composed of angles, T-sections, and cruciforms. With the use of theoretical
consideration of energy dissipated during plastic deformation of these elements, a formula for
the average crushing strength can be expressed as follows:
𝜎𝑐 = 2.42 [𝑛𝐴𝑇𝑡2
𝐴]
0.67
[0.87 + 1.27𝑛𝑐 + 0.31𝑛𝑇
𝑛𝐴𝑇(
𝐴
(𝑛𝑐 + 0.31𝑛𝑇)𝑡2)
0.25
]
0.67
( 34 )
where
𝜎𝑐 : the average crushing strength of the bow;
𝜎0 : the flow stress;
t : the average plate thickness of the cross-section under consideration;
A : the cross-sectional area of deformed material;
𝑛𝑐 : the number of cruciform;
𝑛𝑇 : the number of T-sections; and
𝑛𝐴𝑇 : the number of angle and T-sections.
The total crushing force Fc is then obtained by multiplying this strength by the associated cross-
sectional area A of the deformed material as 𝐹𝑐 = 𝜎𝑐𝐴.
4.3.2 SHARP Tool
SHARP is a powerful software which has been developed using C++ program so as to analyze
ship collision rapidly. In SHARP, the internal mechanics are coupled with an adaptive version
of MCOL described in (Le Sourne et al., 2012). The crushing resistances are calculated with
the use of Super Element Theory mentioned in the previous section 4.3.1. The input parameters
such as structural design of struck ship, data of the striking bow, and different collision
scenarios are defined using a graphical user interface as shown in Figure 11.
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
45
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
Figure 10 Descriptions of super-elements for oblique collisions (Le Sourne et al., 2012)
P 46 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
Figure 11 Graphical user interface of SHARP
The definition of the struck ship structure can be limited to the impacted area, consisting of the
surfaces and their associated scantlings. In the case of defining the striking ship bow geometry,
a simplified model of the fore part of the ship which also includes both the bow and the bulb is
used. However, for estimating the bow’s crushing resistance, only longitudinal structural
members are required to model as the method implemented in SHARP does not consider
transverse structural elements of the bow or bulb.
To consider the external dynamic effects during collision, hydrodynamic matrices of both
struck and striking ships need to be prepared using a seakeeping software such as the one
developed by Bureau Veritas and named HydroStar. As can be seen in Figure 12, the crushing
force is calculated using super-element method and roll, yaw, and pitch moments are
determined at the centre of gravity of both ships. These results are then transmitted to an
external dynamic program MCOL which returns new acceleration, velocity, and position of
each ship. The simulation terminates when the surge velocity of the striking ship is equal to the
sway velocity of the struck ship. The outputs can be post-treated as crushing force and absorbed
energy in terms of penetration, the hydrodynamic forces acting on the ship side, and graphical
animations of the collision process.
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
47
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
Figure 12 Workflow diagram of SHARP (Le Sourne et al., 2012)
Yes No
P 48 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
5. LS-DYNA AND SHARP SIMULATIONS
In this section, the numerical and analytical simulation procedures carried out by using both
LS-DYNA finite element code and SHARP analytical solver will be presented.
5.1 LS-DYNA/MCOL Simulation Procedures
5.1.1 General
In general, the numerical simulations in LS-DYNA/MCOL consist of three main steps:
Pre-processing using FEMB (Finite Element Model Builder);
Running LS-DYNA collision simulation coupled with MCOL; and
Post-processing by LS-DYNA post-processor LS-PRE/POST.
The required hydrodynamic properties can be determined by using additional hydrodynamic
software such as HydroStar and ARGOS. In this thesis, the same input data obtained from the
previous EMSHIP master thesis (Uzögüten, 2016) will be used.
The Finite Element simulation with LS-DYNA software has been carried out according to the
following flow chart (See Figure 13).
Figure 13 Flow-chart of LS-DYNA simulation process
3D Modelling and Meshing
Define Elements, Materials and
Parts
Define Boundary Conditions
Define Collision Scenarios
Define Contacts Computation
Controls
Solution Controls
Run the AnalysisResults and Post-
processing
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
49
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
5.1.2 Modelling and Meshing
As can be seen in Figure 13, the first step for the LS-DYNA simulation is to build finite element
3D models for both colliding vessels. In this thesis, the struck and striking ships were modelled
and meshed by using FEMB (Finite Element Model Builder).
5.1.2.1 Struck Ship: Type C Double Hull Tanker
The struck ship is a typical Type C Double Hull tanker which has the following characteristics:
Length overall: 125 m;
Breadth: 11.4 m;
Draught: 4.5 m;
Depth: 6 m;
Displacement: 5863.3 tonnes.
In Figure 14, finite element model of Type-C inland tanker is shown. Note that only three cargo
holds were modelled. The rest of the ship was taken into account by defining a rigid body on
the two end bulkheads that were characterized by the ship’s true mass, inertia and the centre of
gravity. The size of the maximum element, the ratio between element length and thickness, and
the ratio between longer and shorter shell element edge were defined in accordance with the
A.D.N. Regulations. However, only reference design has been modelled in this thesis since the
main purpose is to validate the SHARP tool by comparing the obtained results.
Figure 14 Finite Element model of struck ship (Type C inland double hull tanker)
P 50 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
5.1.2.2 Striking ships: V-shape bow and Push barge bow
Two striking ship bows were considered according to the A.D.N. Regulations, namely:
V-shape bow and
Push barge bow.
These two bow shapes were modelled according to the geometries given by A.D.N. Regulation
at section 9.3.4.4.8. Figure 15 shows the finite element models of the striking ships. Note that
only the fore part of the striking ship needs to be modelled. The rest of the striking ship was
represented by a rigid body with associated inertia, true mass and centre of gravity. In addition,
it was not needed to model the detailed structure of the striking bow as it will be considered as
a rigid body as recommended by A.D.N. Regulations at section 9.3.4.4.6.2.
To be able to define the required hydrodynamic properties, two real ships were chosen, namely,
Touax for the push barge bow and Odina for the V-shape bow. They have the following
characteristics:
Table 2 Main characteristics of the striking ships
V-shape bow Push barge bow Units
Length overall 85.95 88.5 m
Breadth 10.95 11.4 m
Draught 3.65 3.4 m
Depth 4.6 4.32 m
Displacement 3040.6 3228.7 tonnes
(a) Push-barge bow
(b) V-shape bow
Figure 15 Finite Element models of striking ship bows
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
51
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
5.1.3 Elements, Materials and Parts
5.1.3.1 Elements
According to A.D.N. Regulations, plate structures such as side shells, inner side shells, web
frames, etc. were modelled as shell elements and stiffeners such as bottom longitudinals, deck
longitudinals, etc. were modelled as beam elements.
Belytschoko-Tsay formulation was used for the shell elements as it is more computationally
efficient than Hughes-Liu shell elements. Five integration points were considered throughout
the thickness of each shell element so as to capture the realistic plastic deformations. As for the
beam elements, Hughes-Liu beam element formulation was applied as this type of beam
provides an out-of-plane bending which is not provided by truss beam elements. Refer to
(Hallquist, 2006) for further details.
5.1.3.2 Materials
The material of the struck ship is Grade A shipbuilding steel with the following properties:
Young Modulus: 210 000 MPa;
Yield Strength: 250 MPa;
Ultimate Strength: 512 MPa.
In LS-DYNA modelling, two materials were defined as follows:
piecewise-linear-plasticity material and
rigid material
Piecewise-linear-plasticity material can be defined with or without considering a threshold
value for the failure strain. The rigid material was defined in order to take into account the
remaining part of the struck ship which was not included in the modelling.
With the use of piecewise-linear plasticity material, an elasto-plastic behaviour can be
represented and the corresponding stress-strain relation can be obtained by using the following
formula suggested by A.D.N. Regulation:
P 52 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
𝜎 = 𝐶. 휀𝑛 ( 35 )
where
𝑛 = ln(1 + 𝐴𝑔);
𝐶 = 𝑅𝑚. (𝑒
𝑛)𝑛;
Ag = the maximum uniform strain related to the ultimate tensile stress Rm; and
e = the natural logarithmic constant.
Using Equation (35), an arbitrary stress-strain curve was plotted and shown in Figure 16. Note
that the elastic behaviour of the material is only represented by 250 MPa and any values lower
than this were not considered as the deformations taking place during collision were usually
very high.
Figure 16 Arbitrary stress-strain curve for elasto-plastic material
The required material failure strain for the material was determined according to the formula
suggested by A.D.N. Regulations (see also Lehmann and Peschmann, 2002):
휀𝑓(𝑙𝑒) = 휀𝑔 + 휀𝑒 .𝑡
𝑙𝑒 ( 36 )
where
휀𝑓 : rupture strain;
0.0E+00
1.0E+08
2.0E+08
3.0E+08
4.0E+08
5.0E+08
6.0E+08
0 0.05 0.1 0.15 0.2 0.25
Str
ess
(N/m
^2
)
Plastic Strain
Arbitrary Stress-Strain Curve
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
53
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
휀𝑔 : uniform strain;
휀𝑒 : necking;
𝑡 : plate thickness; and
𝑙𝑒 : individual element length.
To use Equation (36), the following two assumptions were made:
First, the structures that were considered to fail due to excessive tension include side
shell, sheer strake, bilge plate, inner side shell, longitudinal bulkhead and webs. Note
that web frames do not fail by tension, however, they can fail by excessive shearing
especially at the connections of webs with weather deck or bottom plating.
Secondly, the parts which were subjected to compression such as deck, bottom, double
bottom, etc. were considered to still possess enough strength to resist the collision
without tearing even after plastic deformation. Therefore, for those parts, failure strain
was not considered.
With these two assumptions, the failure strain can be calculated for the concerned parts (See
Table 3). The definitions of parts and sections will be described in the next section 5.1.3.3.
Table 3 Material failure strain calculation
Part no. Item 𝒕
[m]
𝒍𝒆
[m] 𝜺𝒇
2 Web frame (upper part) 0.0064 0.126 0.08
6 Web frame (lower part) 0.008 0.126 0.09
5 Inner side shell 0.0075 0.1325 0.09
7 Outer side shell 0.0095 0.1325 0.09
9 Bilge plate 0.0115 0.1325 0.10
10 Sheer strake 0.0235 0.1325 0.15
14 Side longitudinals 0.0100 0.1325 0.14
15 Side longitudinals 0.0100 0.1325 0.14
The values of 휀𝑔 and 휀𝑒 required for the calculations were taken from Table 1 described in
Section 2.4.4. As can be seen, the values of the failure strain determined from Equation (36)
have some underestimations when they were compared with the reality. According to (Le
Sourne, 2015), a “classical” rupture strain value of 20 % was adopted in this thesis for the
P 54 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
simulations of LS-DYNA. Regarding this matter, also refer to (Simonsen, 2000), (Lehmann
and Peschmann, 2002), (Naar, 2002) and (Kitamura, 2002) for more references.
The material of the striking ship is Grade-A shipbuilding steel with the following properties:
Young Modulus: 210 000 MPa;
Yield Strength: 250 MPa;
Ultimate Strength: 512 MPa.
Note that according to A.D.N. Regulations, a rigid material was used to model the striking ship
bows.
5.1.3.3 Parts
Parts in LS-DYNA are used to define material information, properties of the section used,
hourglass type, thermal properties, and a flag for part adaptivity. Each part is comprised of
elements that have the same properties. In this way, same structural components can be grouped
under the same part. For example, bottom for part 33, weather deck for part 28, side shell for
part 7 and so on have been applied in the current LS-DYNA models. The following diagram
shows the way the keywords are organized in order to model in LS-DYNA:
Figure 17 Organization of keywords to define structural parts in LS-DYNA
In addition to the PART command, the PART_INERTIA option allows the inertial properties
and initial conditions to be defined instead of calculating them from the finite element mesh.
This option, however, can be applied for the rigid material only. Thus, in this case, the two
constraint boundaries and the striking ships were defined by using that command.
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
55
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
5.1.4 Boundary Conditions
Boundary conditions are imposed on the two transverse bulkheads which are located at both
ends of the struck ship model. As mentioned in the previous section 5.1.3, PART_INERTIA
command was used to define the associated true mass, inertia and centre of gravity for the
remaining part of the struck ship. The striking ship was also treated as a rigid body. At each
time step, the crushing forces were calculated in LS-DYNA and then were transferred to an
outer dynamic program named MCOL which will then calculate the translational movement as
well as rotational movement of the ship.
One problem with MCOL is that it is not parallelized, i.e., the calculations cannot be made
simultaneously in different processors of the computers. Generally, LS-DYNA uses very small
time steps (usually microseconds) and thus, the simulation needs millions of calculation steps.
Therefore, it is impossible for MCOL to perform all calculation steps with only one processor
or else the simulation will take several hundreds of hours to finish. To solve this issue, MCOL
sub-cycling option can be activated as it permits the MCOL to perform the calculations only at
every N cycle of LS-DYNA calculations. Since the evolution of the hydrodynamic forces is a
lot slower than the evolution of the stresses and strains in the structure, using such option will
not affect the final simulation results. For example, the hydrodynamic forces will not change a
lot during one millisecond while the stresses in the structure may change a lot in one
millisecond.
In order to determine how many calculation steps are required for the MCOL program, a
convergence test needs to be performed. The following test scenarios were chosen to perform
the convergence test for MCOL:
Case 2 (V-shape bow : 90 deg : Between webs : Mid-depth of struck ship);
Case 3 (Push barge bow : 55 deg : At web : Mid-depth of struck ship); and
Case 4 (Push barge bow : 55 deg : At bulkhead : Above deck of struck ship).
The collision scenarios mentioned above were taken from the next Section 5.1.5.
Cases 2 and 3 in the above scenarios were considered without rupture strain and case 4 with the
rupture strain. Different MCOL sub-cycling steps (100, 200 and 400) were considered for each
case, making a total of 9 simulations as follows (See Table 4):
P 56 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
Table 4 Convergence tests for MCOL sub-cycling steps
Test Simulation MCOL calculation
Case 2 without rupture strain 100 200 400
Case 3 without rupture strain 100 200 400
Case 4 with rupture strain 100 200 400
As shown in Table 5, the results of MCOL sub-cycling steps are given for each test case. Note
that the striking ship was considered at a constant speed 10 m/s for the convergence tests. The
obtained deformation energies and penetrations are plotted against time (See Appendix A).
Table 5 Results of convergence tests for MCOL
MCOL Deformation energy Penetration
Steps [J] Error % [m] Error %
Case 2 (V-shape bow : 90 deg : Between webs : Mid-depth of struck ship)
100 1.01E+08 4.31
200 1.01E+08 0% 4.36 1%
400 9.79E+07 3% 4.14 5%
Case 3 (Push barge bow : 55 deg : At web : Mid-depth of struck ship)
100 8.72E+07 3.6
200 8.71E+07 0% 3.74 4%
400 8.67E+07 0% 3.77 1%
Case 4 (Push barge bow : 55 deg : At bulkhead : Above deck of struck ship)
100 5.13E+07 3.09
200 5.32E+07 4% 3.19 3%
400 5.36E+07 1% 3.19 0%
In the above table, the error percentage is determined by using the following formulation:
𝐸𝑟𝑟𝑜𝑟 % =𝑀𝐶𝑂𝐿2𝑁 − 𝑀𝐶𝑂𝐿N
𝑀𝐶𝑂𝐿N ( 37 )
where, N = MCOL calculation steps = 100 or 200.
According to the results of the convergence tests shown in Table 5, the following deductions
can be made:
Case 2 has the same results for MCOL100 and MCOL200 calculation steps (0 %
difference for the deformation energy and 1 % for the penetration).
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
57
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
Case 3 has the same results for MCOL200 and MCOL400 calculation steps (0 %
difference for the deformation energy and 1 % for the penetration).
Case 4 has the same results for MCOL200 and MCOL400 calculation steps (1 %
difference for the deformation energy and 0 % for the penetration).
Therefore, it was decided to use MCOL sub-cycling step of 200 as it can give the converged
results for all simulations. The detailed comparisons of the graphs for different MCOL sub-
cycling steps can be found in Appendix A.
5.1.5 Collision Scenarios
The following five scenarios have been defined in LS-DYNA in order to compare the results
with SHARP:
Table 6 Collision scenarios
Scenarios Bow
Type
Collision Angle
[deg] Longitudinal Position
Vertical
Position
Case 1 V-shape 90 At web Under deck
Case 2 V-shape 90 Between webs Mid-depth
Case 3 Push barge 55 At web Mid-depth
Case 4 Push barge 55 At bulkhead Above deck
Case 5 V-shape 90 At web Above deck
In Table 6 shown above, the longitudinal position and vertical position refer to the locations of
the impact point on the struck ship. These locations can be again expressed in terms of
coordinate system in 3D plane (i.e., in terms of X, Y and Z coordinates) as shown in Table 7.
Table 7 Location of the impact points in LS-DYNA
Scenarios Impact Locations [m] Striking
Bow Type
Collision Angle
[deg] X Y Z
Case 1 21.55 5.76 5.65 V-shape 90
Case 2 22.67 5.76 3.72 V-shape 90
Case 3 16.14 5.78 3.89 Push barge 55
Case 4 24.10 5.78 6.24 Push barge 55
Case 5 21.55 5.76 6.37 V-shape 90
P 58 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
Note that in LS-DYNA, the origin point of the global coordinate system was located at the
intersection of the Centreline, the Baseline and 200 mm horizontally from the first bulkhead of
the struck ship model. When defining Y impact location, a margin distance was taken into
account which was defined by the sum of half thickness of struck ship’s side shell, half thickness
of striking ship’s bow and some additional margin distance (about 10-20 mm). A typical
collision simulation setup of LS-DYNA (case 1) is shown in Figure 18.
Figure 18 Typical collision simulation in LS-DYNA (Case 1)
Among those five cases mentioned above, cases 1 to 3 were defined using exactly the same
scenarios used in previous EMSHIP Master thesis (Uzögüten, 2016) in which the scenarios
were defined by following A.D.N. Regulations. However, the vertical impact locations for cases
4 and 5 will be slightly modified in this thesis. Because, the positions defined by (Uzögüten,
2016) were a bit too high for the striking ships and as a result, there were significant vertical
reaction forces, leading the simulation to undesirable condition. (See Figure 19)
This is because when simulating in LS-DYNA, the gravity loading is usually ignored for both
struck and striking ships in order to save time and avoid complexity in the simulation process.
This simplification will not affect the results unless the vertical reaction forces become
important. Therefore, in order to prevent the vertical forces from becoming very large, the
position of the striking ship was lowered by about 800 mm while keeping the draft combination
points to remain inside of the rectangular area bounded by maximum and minimum draughts
of colliding vessels. (See Figure 20)
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
59
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
Figure 19 Undesirable effect (Conceptual diagram)
(a) Push barge bow (b) V-shape bow
Figure 20 Definitions of the vertical impact locations (A.D.N. Regulations, 2015)
Table 8 Draft combinations of struck and striking ships
Case Bow Type Struck Ship [m] Striking ship [m]
Case 1 V-shape 4.5 3.5
Case 2 V-shape 2.72 3.65
Case 3 Push barge 2.72 3.4
Case 4 Push barge 4.5 2.43
Case 5 V-shape 4.5 2.78
P 60 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
In Figure 20, minimum and maximum draughts of striking and struck ships are shown. It can
be observed that even though the vertical positions of case 4 and case 5 were modified, the
choice of the draft combination was ensured to predict the worst case collision scenario, i.e., to
select the possible maximum displacements, in accordance with the A.D.N. regulations.
Hereby, the LS-DYNA simulations can be divided into two categories:
Simulations without rupture strain using a constant striking ship’s speed of 3 m/s; and
Simulations with rupture strain using a constant striking ship’s speed of 10 m/s.
The first category of the aforementioned LS-DYNA simulations aims at checking the validity
of software SHARP without any rupture strain. When rupture strain is not considered, it would
be more realistic to use a lower speed, 3 m/s, instead of using 10 m/s as proposed by the A.D.N.
The aim of the second category is to observe if SHARP tool can be applied for the A.D.N.
Regulations as a replacement of the conventional FE software. Therefore, a total of 10 LS-
DYNA simulations will be performed in the framework of this thesis.
To determine the hydrodynamic properties required for the rigid body calculation, the coastal
area of Belgium (Scheldt estuary – Zeebrugge) was chosen as the investigated incident area
(Figure 21). According to the wave data (12875 measured waves) in this region, the most
probable wave period is approximately 5.25 seconds and the wave frequency is about 1.197
rad/sec. The values of wave damping matrices were taken for this wave period and “.mco” files
were generated for different collision scenarios. The “.mco” files applied in this master thesis
were taken from the previous master thesis (Uzögüten, 2016).
Figure 21 Scheldt estuary – Zeebrugge (available from: Google Map)
Zeebrugge
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
61
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
5.1.6 Contact Type and Friction
Using ‘node on segment penalty’ method, the following two contact options were defined in
LS-DYNA.
CONTACT_AUTOMATIC_GENERAL_ID and
CONTACT_AUTOMATIC_SURFACE_TO_SURFACE.
The first option is to create the self-contact between each part of the struck ship. For example,
the contact between side shell and web frame, local folding of weather deck and bottoms and
so on. The second option is to create the contact between striking and struck ships. Depending
on the scenarios, the contact definition can change especially when failure is considered. In
order to get a rough idea of the collided part and behaviour of the elements involved, it is
important to run some test simulations and make some modifications in the input file if needed.
The friction considered in LS-DYNA is calculated by using the Coulomb friction relation given
by the following Equation (38):
𝜇𝑐 = 𝐹𝐷 + (𝐹𝑆 − 𝐹𝐷). 𝑒−𝐷𝐶|𝑣𝑟𝑒𝑙| ( 38 )
where
𝜇𝑐 = Coulomb friction coefficient;
FD = Dynamic coefficient of friction = 0.1;
FS = Static coefficient of friction = 0.3;
DC = Exponential friction decay coefficient = 0.01; and
|𝑣𝑟𝑒𝑙| = relative friction velocity of contact surfaces.
According to the scenarios simulated in this thesis, the friction coefficients can be computed as
follows:
With 90 degree collision angle ==> |𝑣𝑟𝑒𝑙| = 0 m/s and 𝜇𝑐 = 0.3.
With 55 degree collision angle ==> |𝑣𝑟𝑒𝑙| = 5.74 m/s and 𝜇𝑐 = 0.29.
In SHARP, the Coulomb friction coefficient 𝜇𝑐 has already been set at a predefined value of
0.3. Although there is no option to modify this coefficient in the current version of SHARP,
there is no problem as this is the same value recommended by A.D.N. Regulations.
P 62 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
5.2 SHARP/MCOL Simulation Procedures
5.2.1 General
In general, simulation in SHARP program consists of the following six main steps:
Figure 22 Simulation procedures for SHARP/MCOL
5.2.2 Modelling and Meshing
As shown in Figure 22, the first step of simulating in SHARP involves developing collision
models for both struck and striking ship.
5.2.2.1 Modelling of the struck ship
This step consists of hull form generation and structural modelling. Hull form modelling was
done by defining a parametric model based on the definitions of three sections (aft, middle and
forward) and longitudinal lines. Each section can be modified by shifting 6 control points which
are connected between the sections through longitudinal lines.
After modelling of the hull form, the structures were modelled. This includes defining the ship
surfaces such as decks, bulkheads, hulls and so on and the corresponding scantlings. A fixed
frame spacing in meters was determined before the start of the modelling and only starboard
side needed to be modelled. The other side was automatically created by the software by
Modelling Define materials
Define collision scenarios
Generate hydrodynamic matrices for
MCOL
Run the analysisPost process the
results
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
63
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
generating symmetrical surfaces about the centreline. The complete SHARP model of struck
ship is shown in Figure 23.
Figure 23 Complete SHARP model of struck ship
However, it should be noticed that when modelling structures in SHARP, some details cannot
be modelled such as holes and corrugated bulkheads. To solve this issue, an equivalent plate
thickness has been applied for those specific structures. The equivalent plate thickness was
determined based on the fact that the material volume of the actual and equivalent plate is the
same. The following formula depicts the required equivalent thickness:
𝑡𝑒𝑞 =𝐴𝑎𝑐𝑡𝑢𝑎𝑙𝑡𝑎𝑐𝑡𝑢𝑎𝑙
𝐴𝑎𝑐𝑡𝑢𝑎𝑙 + 𝐴ℎ𝑜𝑙𝑒
( 39 )
where,
𝑡𝑒𝑞 : equivalent thickness of the plate;
𝑡𝑎𝑐𝑡𝑢𝑎𝑙 : actual thickness of the plate;
𝐴𝑎𝑐𝑡𝑢𝑎𝑙 : actual area of the plate; and
𝐴ℎ𝑜𝑙𝑒 : total area of the holes on the plate.
In this thesis, a virtual horizontal deck was required to use as a limit for other surfaces. The
thickness of that surface was assigned as zero so that it would not be included in the calculation.
However, it is found out that this could impose some problems in the meshing process of
SHARP whose influence will be analysed in details in the next section of this thesis.
As can be seen in Figure 24a, SHARP will automatically divide the shell plating into two parts
when virtual plating is used, creating two side shell super-elements instead of one. Usually the
side shell super-element is bounded by two bulkheads at the end, by weather deck at the top
P 64 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
and by the ship’s bottom (See Figure 25b). By considering the virtual deck, this assumption
will change completely and can somehow affect the results. This is illustrated in Figure 25
which highlights the difference in the basic assumption of SHARP. As can be seen, a side shell
should normally be bounded by four clamped ends. However, the presence of the virtual deck
caused an extra clamp end at the position of the deck which could later result in a more rigid
side shell. Therefore, it is imperative to check the effects of the virtual deck in more details.
(a) With virtual deck (b) Without virtual deck
Figure 24 Side shell becoming two super-elements due to virtual deck (Body plan view)
(a) Two side shell super-elements
each with four fixed ends
(b) One side shell super-element
with four fixed ends
Figure 25 Comparison of side shell super-element with and without virtual deck (profile view)
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
65
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
5.2.2.2 Modelling of striking ship
In SHARP, only the fore part of the striking ships was modelled. Although it was possible to
model the striking ship structural configurations in SHARP, only the geometry of the bow was
modelled in this study as A.D.N. demands a rigid striking bow. On the other hand, SHARP is
also able to perform simulations with deformable striking bows.
The geometry of the striking ship bow is a parametric model defined by 9 parameters shown in
Figure 26. The symbols used in the figure are explained in Table 9. As can be seen in the figure,
the deck and bottom are defined by using ellipses and the bulb is assumed to be an ellipsoid.
Figure 26 Model of the striking ship bow in SHARP (Besnard, 2014)
Table 9 Definitions of the parameters used in the SHARP bow model (Besnard, 2014)
Identification of the parameter Notation
Semi major axis of the deck R1d
Semi minor axis of the deck R2d
Semi major axis of the bottom R1b
Semi minor axis of the bottom R2b
Depth D
Length of the bulb RL
Vertical radius of the bulb RV
Horizontal radius of the bulb RH
Distance between the bulb tip and the foremost part of the bow RD
P 66 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
(a) Push barge bow (b) V-shape bow
Figure 27 Models of striking ships in SHARP
Figure 27 shows the models of the two striking ship bows. By comparing these pictures with
finite element models presented in Figure 15, it can be observed that the V-shape bow could be
modelled quite accurately while it was not possible to get an exactly similar bow shape for the
push barge given by A.D.N. So as to obtain a similar push barge bow, the longitudinal curvature
of the ellipse was greatly reduced so that the weather deck would be a flattened curve shape.
Moreover, due to the limitation of the modelling of the knuckles, only one side angle can be
defined. This angle, however, will not be the first point of impact. The first impact point will
be at some point near that angle since the bow has been tilted to 55 degree.
In Figure 28, the intended impact location is defined according to the direction of the striking
ship although the initial impact point will be at the knuckle of the bow. Without making an
adjustment, the original striking ship model of SHARP cannot have the same shape as LS-
DYNA around initial impact point. Therefore, an extra distance of 0.95 m has been added to
the striking ship longitudinal position so that the shape of the bow around knuckle (around
initial impact point) would be as close to the LS-DYNA model as possible.
Figure 28 Adjustments in the push barge bow position in SHARP (Top view)
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
67
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
Figure 29 highlights the difference in bow shapes between the two models which should also
be taken into account when evaluating the results.
(a) Difference in bow shape (Top view) (b) Difference in bow shape (Profile view)
Figure 29 Comparison of the push barge bow models between LS-DYNA and SHARP
5.2.3 Materials and Rupture Strain
In SHARP, the material properties are characterized by the following parameters:
Young modulus E [MPa];
Yield stress S0 [MPa];
Rupture toughness Rc [N/mm]; and
Rupture strain Ec.
As the material of the struck ship is shipbuilding steel – Grade A that has yield strength of 250
MPa and ultimate strength of 512 MPa, the lower limit 250 [MPa] was taken as the material
yield stress S0 (See Figure 30). The young modulus was taken as 210000 [MPa] while rupture
toughness was taken as 500 [N/mm].
Although SHARP is able to make simulations with or without considering a rupture criteria,
the rupture strain value Ec is still required to define. Therefore, in the case without rupture
strain, very high value of Ec such as 1000 has to be applied. The reason is to prevent any shell
rupture in SHARP and sometimes it is found out that Ec value of 1 or 2 is not enough to prevent
this. With Ec = 1000, the prevention of shell rupture was ensured. As in the case with rupture
strain, the rupture strain value of 0.2 calculated in Table 3 of section 5.1.3.2 was applied.
P 68 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
Figure 30 Stress-strain curves considered for LS-DYNA and SHARP simulations
Figure 30 shows the comparison of stress-strain relation considered in LS-DYNA and SHARP
simulations. As can be seen in the figure, a true stress-strain relation is not applied in SHARP
but rather a plastic flow stress which remains constant until the rupture of the shell. In this
context, an average yield stress of (𝜎0 =𝜎𝑦+𝜎𝑢
2= 317.5 𝑀𝑃𝑎) has been considered in the
master thesis of (Uzögüten, 2016). However, under the assumption of very mild steel for the
inland vessels, only the yield stress up to 250 MPa has been considered in the framework of
this thesis.
5.2.4 Collision Scenarios
The same scenarios used in LS-DYNA will be applied for SHARP as well. However, in order
to take into account the variation inherent to the Super-Element Method, additional 8 collision
locations need to be created around the real impact point. So, for each LS-DYNA simulation, 9
simulations of SHARP will be analysed, making it a total of 90 scenarios. The total energy
absorbing capacity will be taken as the average results of these 9 simulations. The obtained
results were compared numerically as well as graphically and presented in the next section.
Figure 31 shows how those additional impact points can be considered in SHARP.
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
69
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
Figure 31 Impact locations on side shell in SHARP
In the above figure, the values shown are calculated using the structural information of the
struck ship as follows:
Impact point for X distance change (longitudinal) : ± 𝑊𝑒𝑏 𝑓𝑟𝑎𝑚𝑒 𝑠𝑝𝑎𝑐𝑖𝑛𝑔
2
Impact point for Z distance change (vertical) : ± 𝑆𝑡𝑖𝑓𝑓𝑒𝑛𝑒𝑟 𝑠𝑝𝑎𝑐𝑖𝑛𝑔
2
where, web frame spacing is 1.59 m and the stiffener spacing is 0.5 m.
Note that in SHARP, the impact locations were defined by using X and Z (longitudinal and
vertical coordinate) of the global coordinate system whose origin is located at the aft extreme
of the struck ship and at the interception of the ship’s baseline and centreline. In addition, it was
not necessary to determine Y coordinate in SHARP for the collision position as it will
automatically be calculated, taking into account the margin between the two ships. Figure 32
shows typical diagram of the collision simulation (case 1) defined and modelled by SHARP.
Figure 32 Typical collision scenario modelled in SHARP (case 1)
P 70 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
6. COMPARISON AND ANALYSIS
SHARP tool has already been validated for ocean-going tankers and FPSO application cases,
see (Paboeuf et al., 2015). However, in order to validate the use of SHARP program for inland
vessel structure, the following results have been compared between LS-DYNA and SHARP
calculations:
Penetration into the struck ship, and
Struck ship deformation energy.
6.1 Simulations without Rupture Strain
The main focus on this study is to check if SHARP results correspond well with LS-DYNA
without considering the rupture. The striking ship was considered to have a constant speed of 3
m/s for both simulations. Thus, it should be noted that the simulations presented in this Section
do not exactly follow the A.D.N. Regulations. However, the validation tests regarding the
A.D.N. Regulations will be presented in the following Section 6.2.
6.1.1 Case 1 (V-shape bow : 90 deg : At web : Under deck of struck ship)
Table 10 shows the penetration and deformation energy calculated from LS-DYNA when the
tanker has been collided perpendicularly by V-shape bow with a speed of 3 m/s. The impact
point considered in this simulation is at web and just under deck of the struck ship. The results
were taken at 1.2 sec which is at the end of the simulations.
Table 10 Results calculated by LS-DYNA – Case 1 (without rupture strain)
Simulation Time [sec] Penetration [m] Deformation energy [MJ]
1.2 1.1 8.5
In Figure 33, the extent of damage in LS-DYNA after the collision is shown. It can be observed
that not only the side shell but also the deck is damaged due to the impact. The reason why the
weather deck deforms without being collided is that in LS-DYNA, the elements are
simultaneously activated, i.e., there is an interaction between different structural components,
for example, side shell and weather deck in this case. This is usually the case for inland vessels
when the two colliding ships have nearly the same height. Another reason can be that the
thickness of the weather deck (9.75 mm) is quite small compared to the sheer strake thickness
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
71
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
which is 23.5 mm. Thus, it is obvious that the deformation of the thicker plate (the sheer strake)
has caused the deformation in the thinner one, the weather deck.
Figure 33 Extent of damage in LS-DYNA – Case 1 (without rupture strain)
Table 11 below presents the results of penetration into the struck ship and the deformation
energy calculated by SHARP at 9 impact locations. The extent of damage in SHARP is shown
in Figure 34. The impacted elements are shown in yellow colour while the destroyed elements
are in red. As there is no red colour, it can be said that there is no shell rupture. Moreover, it is
found out that the deck super-element has not been activated at the end of the simulation. This
is because in SHARP, the super-elements are independently activated only upon contact.
Table 11 Results calculated by SHARP – Case 1 (without rupture strain)
Simulations No. Penetration [m] Deformation Energy [MJ]
Simulation 1 0.68 8.70
Simulation 2 0.68 8.70
Simulation 3 0.81 8.50
Simulation 4 0.81 8.50
Simulation 5 0.81 8.50
Simulation 6 0.69 8.70
Simulation 7 0.84 8.20
Simulation 8 0.83 8.10
Simulation 9 0.83 8.10
Average 0.78 8.44
Standard deviation 0.07 0.25
P 72 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
Figure 34 Extent of damage in SHARP – Case 1 (without rupture strain)
The results of LS-DYNA and SHARP are compared numerically as well as graphically and
shown in Table 12 and Figure 35 respectively.
Table 12 Comparison of the results – Case 1 (without rupture strain)
Penetration [m] Deformation Energy [MJ]
LS-DYNA 1.1 8.5
SHARP (average) 0.78 8.44
% Difference 27% 1%
Figure 35 Comparison of the results – Case 1 (without rupture strain)
0.00E+00
2.00E+06
4.00E+06
6.00E+06
8.00E+06
1.00E+07
0.00 0.30 0.60 0.90 1.20 1.50
Def
orm
atio
n E
ner
gy [
J]
Penetration [m]
Struck ship - Deformation Energy (Case 1)
LS-DYNA
Simulation 1
Simulation 2
Simulation 3
Simulation 4
Simulation 5
Simulation 6
Simulation 7
Simulation 8
Simulation 9
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
73
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
As can be seen in Figure 35, SHARP results do not agree very well with LS-DYNA and it is
found out that SHARP model seems to be more rigid. However, the final deformation energy
is more or less the same in both simulations, only showing a discrepancy of 1 %. The
discrepancy regarding the final penetration is about 27 %. This is because the coupling effect
between the side shell and the weather deck creates different boundary conditions for the side
shell considered in LS-DYNA and SHARP. Since the weather deck has been deformed in LS-
DYNA, this makes the side shell in LS-DYNA to behave like a plate with three clamped ends
and one moving end (at the weather deck). On contrary, considering the SE1 super-element in
SHARP solver, all the four edges of the side shell are supposed to be clamped, causing the side
shell to behave more rigidly than in LS-DYNA.
In Table 13, deformation energies for each structural component are calculated and expressed
as percentages of the total energy and compared between LS-DYNA and SHARP. The results
are also given for SHARP with one side shell super-element in order to see the influence of
virtual deck. The presence of virtual deck can split the side shell super-element into two parts
and this has already been explained in Section 5.2.2.1. Note that among the 9 SHARP
simulations, the results of SHARP shown in Table 13 are the results of simulation 1 in which
the real impact point was considered. See Section 5.2.4 to recall the impact scenarios in SHARP.
Table 13 Comparison of energy absorption – Case 1 (without rupture strain)
PARTS LS-DYNA
SHARP
(Two side shell super-
elements)
SHARP
(One side shell super-
element)
E (MJ) % E (MJ) % E (MJ) %
Total Energy 8.5 8.71 8.87
Side Shell 3 35% 7.22 83% 7.61 86%
Inner Shell 0.13 2% 0 0% 0 0%
Web Frame 2.6 31% 1.47 17% 1.25 14%
Double Bottom 0 0% 0 0% 0 0%
Bottom 0 0% 0 0% 0 0%
Weather Deck 1.8 21% 0 0% 0 0%
Stiffeners 0.5 6% 0.02 0% 0.01 0%
Others 0.47 5% 0 0% 0 0%
Penetration (m) 1.1 0.68 0.64
P 74 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
According to the results from Table 13, the following analyses can be made:
Firstly, it can be observed that the presence of virtual deck does not cause much
difference in the final results of SHARP because all the results of SHARP with two side
shell super-elements and one side shell super-element are approximately the same.
In addition, it can be observed that in LS-DYNA, most proportion of the total
deformation energy was dissipated in side shell, web frame and weather deck (35%,
31% and 21% respectively). On the other hand, in SHARP, almost all of the total energy
was absorbed by the side shell, amounting to more than 80 % of the total energy. This
can be explained by the different assumptions on the behaviour of the elements
considered in LS-DYNA and SHARP. As already explained, the finite elements are
simultaneously activated when collision occurs whereas super-elements in SHARP are
activated only upon contacts. Therefore, the side shell in SHARP, which does not break
due to the fact that the rupture is not considered in the simulation, will behave like a
barrier that prevents the striking ship from hitting other elements. On contrary, the
deformation of the side shell in LS-DYNA will cause the deformation in other structural
components due to the coupling effect.
In addition, it can be observed that the weather deck in LS-DYNA has absorbed 21 %
of the total energy while the one in SHARP does not absorb any energy at all. This
coupling effect between weather deck and side shell is usually found in inland ship
collisions. An analytical formulation to consider this effect has already been developed
by (Buldgen et al., 2013) but still not implemented in SHARP yet.
All in all, the results of case 1 are highlighting the significance of the coupling effect
when simulating a collision on an inland vessel occuring just under its weather deck.
6.1.2 Case 2 (V-shape bow : 90 deg : Between webs : Mid-depth of struck ship)
Table 14 shows the penetration and deformation energy calculated from LS-DYNA simulation
when the tanker has been collided by V-shape bow at an angle of 90 degree with a constant
speed of 3 m/s. The location of the impact was considered between the webs and at mid-depth
of the struck ship.
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
75
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
Table 14 Results calculated by LS-DYNA – Case 2 (without rupture strain)
Simulation Time [sec] Penetration [m] Deformation energy [MJ]
1.2 1.1 9.5
Figure 36 Extent of damage in LS-DYNA – Case 2 (without rupture strain)
In Figure 36, the extent of damage in LS-DYNA after impact is shown. It can be observed that
in addition to the bending of the side shell, the weather deck also bent slightly. Table 15 presents
the results calculated by SHARP at 9 impact locations. The damage extent in SHARP at the
end of the simulation is shown in Figure 37. The impacted elements are shown in yellow colour
while the destroyed elements are in red. Although some red parts can be seen at the inner shell,
it is later found out to be the graphic error. This is also shown in Table 17 when the energy
absorbed by the inner shell is only 0.02 MJ which is negligible in comparison with total energy.
Table 15 Results calculated by SHARP – Case 2 (without rupture strain)
Simulation No. Penetration [m] Deformation Energy [MJ]
Simulation 1 1.06 8.60
Simulation 2 1.13 8.60
Simulation 3 1.16 8.50
Simulation 4 1.16 8.40
Simulation 5 1.17 8.50
Simulation 6 1.13 8.60
Simulation 7 1.06 8.70
Simulation 8 1.06 8.60
Simulation 9 1.06 8.60
Average 1.11 8.57
Standard deviation 0.05 0.09
P 76 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
Figure 37 Extent of damage in SHARP – Case 2 (without rupture strain)
The end simulation results calculated by LS-DYNA and the average results obtained from all 9
simulations of SHARP are given in Table 16. It is observed that the penetration has only 1 %
discrepancy whereas the deformation energy has 10 % discrepancy.
Table 16 Comparison of the results – Case 2 (without rupture strain)
Penetration [m] Deformation Energy [MJ]
LS-DYNA 1.1 9.5
SHARP (average) 1.11 8.57
% Difference 1% 10%
In Figure 38, these results are again compared graphically. It can be observed that the trends of
both simulations are approximately the same. Nevertheless, after about 0.5 m penetration, the
deformation energy results of LS-DYNA became lower than those of SHARP. Then, it
suddenly increased again around 1 m penetration. It has been found out that this sudden increase
in the energy in LS-DYNA comes from the web frames. However, this is not possible for
SHARP because, as already explained, the side shell super-element in SHARP will absorb
almost all of the total energy while preventing the striking ship from hitting other super-
elements. This can be clearly observed in Table 17 in which the deformation energy of each
structural component is shown as a percentage of the total energy.
Note that there is elastic energy deformation in LS-DYNA, that is, the shell goes backward after
the collision. This is because the speed and mass of the striking ship is small, resulting in the
low initial kinetic energy. However, for the comparison, the final result of LS-DYNA which
takes into account both elastic energy and plastic energy was used.
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
77
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
Figure 38 Comparison of the results – Case 2 (without rupture strain)
Table 17 Comparison of energy absorption – Case 2 (without rupture strain)
PARTS LS-DYNA
SHARP
(With two side shell
super-elements)
SHARP
(With one side shell
super-element)
E (MJ) % E (MJ) % E (MJ) %
Total Energy 9.5 8.56 8.26
Side Shell 4.2 44% 8.14 95% 7.61 92%
Inner Shell 0.2 2% 0.02 0% 0.15 2%
Web Frame 3.5 37% 0.3 4% 0.31 4%
Double Bottom 0.1 1% 0 0% 0.02 0%
Bottom 0.06 1% 0 0% 0 0%
Weather Deck 0.3 3% 0 0% 0 0%
Stiffeners 0.76 8% 0.09 1% 0.18 2%
Others 0.38 4% 0.01 0% -0.01 0%
Penetration (m) 1.1 1.06 1.22
According to the results from Table 17, the followings could be deduced:
The energy contribution of the side shell in LS-DYNA becomes higher in case 2,
amounting to 44 % of the total. This is because in this case, the point of impact, being
0.00E+00
2.00E+06
4.00E+06
6.00E+06
8.00E+06
1.00E+07
1.20E+07
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40
Def
orm
atio
n E
ner
gy [
J]
Penetration [m]
Struck ship - Deformation Energy (Case 2)
LS-DYNA
Simulation 1
Simulation 2
Simulation 3
Simulation 4
Simulation 5
Simulation 6
Simulation 7
Simulation 8
Simulation 9
P 78 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
the mid-depth of the struck ship, is far from both weather deck and bottom. Therefore,
the energy dissipations of the weather deck and bottom are quite low (about 1-3 %) and
as a result, the side shell absorbs more energy in their places.
On the other hand, in SHARP, the side shell absorbs very large amount of energy (95%
of the total). This means that the striking ship will not collide many super-elements such
as inner shell and webs that are behind the side shell. That is why the webs in SHARP
dissipate very small energy (only about 0.3 MJ) as compared with LS-DYNA in which
the energy dissipated by the webs is about 3.5 MJ.
The total energy and the penetration values of both SHARP results are in good
agreement with LS-DYNA. The result of SHARP (with only one side shell super-
element) is slightly more conservative than that of SHARP (with two side shell super-
elements). This is because with the assumption of two side shell super-elements, the
shell becomes more rigid as there will be an additional clamped end at the connection
point. (Refer to Section 5.2.2.1)
All in all, SHARP results in case 2 are quite satisfactory. It can be said that the
assumptions made in LS-DYNA and SHARP are also the same in this case as the deck
and bottom do not deform a lot. Therefore, it can be concluded that SHARP is more
accurate when the coupling effect remains negligible. Of course, this coupling effect
depends on the location of the impact point and the initial kinetic energy of the striking
ship.
6.1.3 Case 3 (Push barge bow : 55 deg : At web : Mid-depth of struck ship)
In Table 18, the penetration and deformation energy at the end of the simulation calculated by
LS-DYNA is presented. In this case, the tanker has been impacted by the push barge striking
ship at an angle of 55 degree with a constant speed of 3 m/s. The impact point is located at web
and at mid-depth of the struck ship.
Table 18 Results calculated by LS-DYNA – Case 3 (without rupture strain)
Simulation Time [sec] Penetration [m] Deformation energy [MJ]
1.2 0.65 6.65
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
79
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
In Figure 39, the extent of damage after impact is shown. It can be observed that the damage
on the side shell is very small and extremely localized. As a consequence, the weather deck and
bottom are only very slightly deformed.
Figure 39 Extent of damage in LS-DYNA – Case 3 (without rupture strain)
Table 19 presents SHARP results of penetration into the struck ship and the deformation energy
at 9 impact locations. The damage extent after impact is shown in Figure 40. As there are only
yellow parts in the figure, it can be said that SHARP does not result any rupture in the structures.
Table 19 Results calculated by SHARP – Case 3 (without rupture strain)
Simulation No. Penetration [m] Deformation Energy [MJ]
Simulation 1 1.16 6.80
Simulation 2 1.11 6.50
Simulation 3 1.13 6.40
Simulation 4 1.18 6.70
Simulation 5 1.04 6.70
Simulation 6 1.03 6.80
Simulation 7 1.01 6.90
Simulation 8 1.13 6.90
Simulation 9 1.08 6.60
Average 1.10 6.70
Standard deviation 0.06 0.17
P 80 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
Figure 40 Extent of damage in SHARP – Case 3 (without rupture strain)
In Table 20 and Figure 41, the results of LS-DYNA and SHARP are compared numerically as
well as graphically. As can be seen, there is a large discrepancy (about 69 %) between the
penetration results of LS-DYNA and the average result of SHARP. The deformation energy,
however, matches very well, only showing 1 % discrepancy. When observing Figure 41, it
appears that the large discrepancy in the penetration is due to the presence of elastic energy
during collision. Note that the energy post-processed in LS-DYNA and used for the comparison
includes both elastic and plastic energies.
Table 20 Comparison of the results – Case 3 (without rupture strain)
Penetration [m] Deformation Energy [MJ]
LS-DYNA 0.65 6.65
SHARP (average) 1.10 6.70
% Difference 69% 1%
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
81
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
Figure 41 Comparison of the results – Case 3 (without rupture strain)
As can be seen in Figure 41, if the elastic energy part is ignored, the maximum penetration in
LS-DYNA is around 0.85 m. Therefore, the discrepancy value will become a lot smaller
(approximately 30 %). Moreover, as can be seen, all of the results of SHARP are always higher
in the penetration, i.e., at the same deformation energy, the damage penetration in SHARP is
always higher than the one calculated by LS-DYNA. However, the slopes of these curves are
approximately the same. The reason is probably due to the many simplifications made when
modelling the push barge. As a result, the damage extent obtained from LS-DYNA and SHARP
are not exactly the same and it can be seen that SHARP has more damage extent. (Refer to
Figure 39 and Figure 40)
In Table 21, the distributions of energy for each structural component are presented. The results
are also given for SHARP (with one side shell super-element). Note that the results of SHARP
in Table 21 are only those of simulation 1 in which real impact point was considered.
According to the results shown in Table 21, the following important points could be extracted:
In LS-DYNA, the side shell absorbs maximum amount of the total energy (60 %) while
the webs and stiffeners absorb about 18 % and 14 % respectively. However, in SHARP,
the largest proportion of the total energy was dissipated only in the side shell (more than
80 %) and the rest were only absorbed by the webs (about 10 %).
0.00E+00
1.00E+06
2.00E+06
3.00E+06
4.00E+06
5.00E+06
6.00E+06
7.00E+06
8.00E+06
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40
Def
orm
atio
n E
ner
gy [
J]
Penetration [m]
Struck ship - Deformation Energy (Case 3)
LS-DYNA
Simulation 1
Simulation 2
Simulation 3
Simulation 4
Simulation 5
Simulation 6
Simulation 7
Simulation 8
Simulation 9
P 82 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
Table 21 Comparison of energy absorption – Case 3 (without rupture strain)
PARTS LS-DYNA
SHARP
(With two side shell
super-elements)
SHARP
(With one side shell
super-element)
E (MJ) % E (MJ) % E (MJ) %
Total Energy 6.65 6.84 6.71
Side Shell 4 60% 5.88 86% 5.48 82%
Inner Shell 0 0% 0.09 1% 0.2 3%
Web Frame 1.2 18% 0.7 10% 0.83 12%
Double Bottom 0.09 1% 0 0% 0 0%
Bottom 0 0% 0 0% 0 0%
Weather Deck 0.12 2% 0 0% 0 0%
Stiffeners 0.9 14% 0.16 2% 0.2 3%
Others 0.34 5% 0.01 0% 0 0%
Penetration (m) 0.65 1.13 1.29
The stiffeners in SHARP only absorbs 3 % of the total energy which is very small as
compared with LS-DYNA. This highlights the difference in the damage extent obtained
from LS-DYNA and SHARP due to the bow shape difference. It seems that the bow
model in SHARP has more in the longitudinal extent of damage but less in the vertical
extent of damage as compared with LS-DYNA.
In this case, the virtual deck have caused some changes in the final results. It can be
seen that the results with one side shell super-element is slightly more conservative than
those with two side shell super-elements.
One important aspect for the case 3 would be to investigate whether the discrepancies
are due to the bow shape difference or due to the coupling effect and the corresponding
boundary conditions for the side shell. However, in this case, since the location of
impact is at the mid-depth of the struck ship and the initial kinetic energy of the striking
ship is quite low, the deck and bottom do not deform a lot in LS-DYNA. In other words,
the simulation has only weak coupling effect. Therefore, in this case, it can be concluded
that the discrepancy is mainly due to the difference in the bow shape models. Therefore,
it is suggested to improve the user interface of SHARP which could consider a more
precise description of the push barge model.
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
83
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
6.1.4 Case 4 (Push barge bow : 55 deg : At bulkhead : Above deck of struck ship)
Table 22 shows the penetration and deformation energy computed by LS-DYNA simulation
when the tanker has been collided by push barge bow at an angle of 55 degree with a speed of
3 m/s. The location of the impact point is at bulkhead and just above deck of the struck ship.
Table 22 Results calculated by LS-DYNA – Case 4 (without rupture strain)
Simulation Time [sec] Penetration [m] Deformation energy [MJ]
1.2 0.6 4.3
In Figure 42, the extent of damage after impact is shown. It is found out that not only the side
shell deforms but also the deck has been crushed by the striking ship.
Table 23 presents the results of penetration into the struck ship and the deformation energy
calculated by SHARP at 9 impact locations.
The damage extent for SHARP is illustrated in Figure 43. It can be seen that only the weather
deck and upper part of the side shell are taken part in the collision process, which is just as the
same as LS-DYNA. The impacted elements are shown in yellow colour while the destroyed
elements are in red. As there is no red colour, it can be said that there is no shell rupture.
Table 23 Results calculated by SHARP – Case 4 (without rupture strain)
Simulations No. Penetration [m] Deformation Energy [MJ]
Simulation 1 0.55 4.70
Simulation 2 0.55 4.70
Simulation 3 0.54 4.50
Simulation 4 0.55 4.60
Simulation 5 0.54 4.50
Simulation 6 0.55 4.70
Simulation 7 0.55 4.70
Simulation 8 0.55 4.70
Simulation 9 0.55 4.70
Average 0.55 4.64
Standard deviation 0.00 0.09
P 84 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
Figure 42 Extent of damage in LS-DYNA – Case 4 (without rupture strain)
Figure 43 Extent of damage in SHARP – Case 4 (without rupture strain)
The results of LS-DYNA and SHARP are compared numerically as well as graphically and
shown in Table 12 and Figure 44 respectively.
Table 24 Comparison of the results – Case 4 (without rupture strain)
Penetration [m] Deformation Energy [MJ]
LS-DYNA 0.6 4.3
SHARP (average) 0.55 4.64
% Difference 8% 8%
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
85
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
Figure 44 Comparison of the results – Case 4 (without rupture strain)
As can be seen in Table 12, both penetration and deformation energy of SHARP show 8 % of
discrepancy in comparison with LS-DYNA. However, the results of SHARP are slightly less
conservative. Note that the penetration result of LS-DYNA is the final value at the end of the
simulation, i.e., at 1.2 seconds.
As can be seen in Figure 44, the structures in SHARP are slightly more rigid. The reason is
probably due to the simplification in the push barge model. Moreover, the side shell super-
element in SHARP has a virtual deck which will divide the side shell and create an extra
clamped end at the connection point. As a consequence, the shell in SHARP is slightly stiffer
than the one in LS-DYNA.
Table 25 shows the comparison of the deformation energy dissipated in each structural
component of the struck ship between LS-DYNA and SHARP. The results of SHARP (with
only one side shell super-element) are also given in order to check if the virtual deck causes
much influence in the results or not.
According to the results from Table 25, the followings could be deduced:
By observing the LS-DYNA results, it can be seen that most of the energy is absorbed
by the weather deck (about 40 % of the total). Other large amount of energy is
distributed to side shell and web frames, amounting to 30 % and 23 % respectively.
0.00E+00
5.00E+05
1.00E+06
1.50E+06
2.00E+06
2.50E+06
3.00E+06
3.50E+06
4.00E+06
4.50E+06
5.00E+06
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70
Def
orm
atio
n E
ner
gy [
J]
Penetration [m]
Struck ship - Deformation Energy (Case 4)
LS-DYNA
Simulation 1
Simulation 2
Simulation 3
Simulation 4
Simulation 5
Simulation 6
Simulation 7
Simulation 8
Simulation 9
P 86 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
Table 25 Comparison of energy absorption – Case 4 (without rupture strain)
PARTS LS-DYNA
SHARP
(With two side shell
super-elements)
SHARP
(With one side shell
super-element)
E (MJ) % E (MJ) % E (MJ) %
Total Energy 4.3 4.66 4.61
Side Shell 1.3 30% 3.54 76% 1.42 31%
Inner Shell 0.05 1% 0 0% 0 0%
Web Frame 1 23% 0.26 6% 0.8 17%
Double Bottom 0 0% 0 0% 0 0%
Bottom 0 0% 0 0% 0 0%
Weather Deck 1.7 40% 0.86 18% 2.38 52%
Stiffeners 0.2 5% 0 0% 0.01 0%
Others 0.05 1% 0 0% 0 0%
Penetration (m) 0.6 0.56 1.04
When observing SHARP results, the total energy of both results are almost the same.
Nonetheless, the rest of the results are not the same and the reason is due to the fact that
the shell becomes less rigid in the case with “one side shell SE”. Since “two side shell
SE” will have an extra clamped end due to the presence of virtual deck during modelling
of the struck ship in SHARP, it is obviously more rigid, leading to less penetration
results. This effect has been explained in the previous cases as well. It can be seen that
the results of SHARP with “one side shell SE” corresponds better with LS-DYNA than
the results of SHARP with “two side shell SE”.
The penetration result obtained from SHARP with “one side shell SE” was almost twice
larger than LS-DYNA. However, this could be improved if the equivalence thickness
between 9.5 mm and 23.5 mm was considered. Currently, “one side shell SE” has been
modelled using 9.5 mm as the thickness.
However, using such option still imposes some difficulties as every time the user wants
to consider only “one side shell SE”, one has to skip the user interface and manually
configure the super-elements in the input file. To conclude, it is suggested to make some
improvements in the user interface of SHARP so that even with the presence of virtual
deck, the side shell super-element will not be divided into two.
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
87
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
6.1.5 Case 5 (V-shape bow : 90 deg : At web : Above deck of struck ship)
Table 26 shows the results of penetration and deformation energy calculated from LS-DYNA
when the tanker has been impacted by V-shape bow at an angle of 90 degree with 3 m/s. The
impact point was considered at web and just above deck of the struck ship. In Figure 45, the
view of damage extent after impact is shown. It can be observed that the side shell was deformed
and the weather deck was crushed due to the collision.
Table 26 Results calculated by LS-DYNA – Case 5 (without rupture strain)
Simulation Time [sec] Penetration [m] Deformation energy [MJ]
1.2 0.7 5.2
Table 27 presents the results of penetration into the struck ship and the deformation energy
calculated by SHARP at 9 impact locations. The damage extent after impact is illustrated in
Figure 46. In SHARP, the impacted elements are shown in yellow colour while the destroyed
elements are in red. As there is no red colour, it can be said that there is no shell rupture. The
web frames visible are only the results of graphic errors when making the animation.
Table 27 Results calculated by SHARP – Case 5 (without rupture strain)
Simulation No. Penetration [m] Deformation Energy [MJ]
Simulation 1 0.61 5.90
Simulation 2 0.60 5.90
Simulation 3 0.61 5.90
Simulation 4 0.60 5.90
Simulation 5 0.60 5.90
Simulation 6 0.60 5.90
Simulation 7 0.61 5.90
Simulation 8 0.61 5.90
Simulation 9 0.61 5.90
Average 0.61 5.90
Standard deviation 0.00 0.00
The results of LS-DYNA and SHARP are compared numerically as well as graphically and
shown in Table 28 and Figure 47 respectively. It can be seen that both penetration and
deformation energy have discrepancy of about 13 %.
P 88 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
Figure 45 Extent of damage in LS-DYNA – Case 5 (without rupture strain)
Figure 46 Extent of damage in SHARP – Case 5 (without rupture strain)
Table 28 Comparison of the results – Case 5 (without rupture strain)
Penetration [m] Deformation Energy [MJ]
LS-DYNA 0.7 5.2
SHARP (average) 0.61 5.90
% Difference 13% 13%
According to Figure 47, it is obvious that LS-DYNA structures are less stiff than the structures
in SHARP because larger penetration was obtained in LS-DYNA. The reason for such
discrepancy is due to the presence of virtual deck when modelling the struck ship structures in
SHARP. This effect is highlighted by the distribution of energy in each structural component
as shown in Table 29.
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
89
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
Figure 47 Comparison of the results – Case 5 (without rupture strain)
As can be seen in Table 29, the energy distributions for each structural component are given in
terms of total energy and compared with LS-DYNA. Note that the value of LS-DYNA
presented in Table 29 includes elastic deformation.
Table 29 Comparison of energy absorption – Case 5 (without rupture strain)
PARTS LS-DYNA
SHARP
(With two side shell
super-elements)
SHARP
(With one side shell
super-element)
E (MJ) % E (MJ) % E (MJ) %
Total Energy 5.2 5.9 5.84
Side Shell 0.8 15% 3.82 65% 1.34 23%
Inner Shell 0.1 2% 0 0% 0 0%
Web Frame 1.2 23% 0.92 16% 2.36 40%
Double Bottom 0 0% 0 0% 0 0%
Bottom 0 0% 0 0% 0 0%
Weather Deck 2.8 54% 1.16 20% 2.12 36%
Stiffeners 0.1 2% 0 0% 0.02 0%
Others 0.2 4% 0 0% 0 0%
Penetration (m) 0.7 0.61 0.91
0.00E+00
1.00E+06
2.00E+06
3.00E+06
4.00E+06
5.00E+06
6.00E+06
7.00E+06
0.00 0.20 0.40 0.60 0.80 1.00
Def
orm
atio
n E
ner
gy [
J]
Penetration [m]
Struck ship - Deformation Energy (Case 5)
LS-DYNA
Simulation 1
Simulation 2
Simulation 3
Simulation 4
Simulation 5
Simulation 6
Simulation 7
Simulation 8
Simulation 9
P 90 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
According to Table 29, the following conclusions can be made:
In LS-DYNA, the side shell, webs and the weather deck are components which absorb
most proportion of the total energy while among them, the maximum amount is
dissipated in the weather deck (about 54 % of the total).
When observing SHARP results, the two simulations lead to different results. The
SHARP simulation with “one side shell super-element” leads to more penetration and
less deformation energy. However, the reason is the same as for case 4 because in this
modelling, only the thickness of 9.5 mm is used instead of considering equivalent
thickness.
All in all, in terms of penetration as well as energy distribution, it is obvious that the
results of SHARP simulation with “one side shell super-element” are in better accord
with LS-DYNA than the ones which considered “two side shell super-elements”. This
improvement should be considered when updating the user-interface of SHARP, i.e., to
be able to use virtual deck without causing any extra effect to the side shell super-
element. Moreover, it is obvious that in case 5, the discrepancy is coming from the
virtual deck since the striking bow model used in this simulation is the V-shape bow
which has almost the same geometry with LS-DYNA.
6.1.6 Overall Analysis
According to the results from case 1 to case 5 (from Section 6.1.1 to 6.1.5), it can be observed
that the modelling of the virtual deck, i.e., the use of two side shell super-elements instead of
only one, can change the final results in some cases. Because, with the presence of virtual deck,
the side shell super-element becomes more rigid and as a consequence, in most of the cases
(except cases 2 and 3), the penetrations calculated by SHARP are less than those calculated by
LS-DYNA. Regarding this matter, the user interface of SHARP needs to be improved so that
the virtual deck will not cause any extra effect on the side shell super-element.
Another important point to consider is the coupling effect between the structural elements such
as deck, bottom and side shell because inland vessels usually have similar depths and thus, the
deformation of the side shell can cause the deformation in the weather deck and/or bottom even
though these components are not directly impacted. An analytical formulation to consider such
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
91
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
effect has been developed by (Buldgen et al., 2013), but has not been implemented in SHARP
yet.
In addition, the bow shape geometries of the striking ship in SHARP, especially the push barge,
do not match exactly with the push barge model in LS-DYNA which takes the bow shape
definition from A.D.N. Regulations. Such difference in the bow shape can lead to discrepancy
which can be very large, for example, case 3 (about 69 %).
Moreover, the presence of elastic energy should also be noticed as this could make the side
shell in LS-DYNA to bounce back a lot at the end of the simulation. This happens when the
initial kinetic energy of the striking ship is quite low. Nevertheless, since SHARP can only
consider the completely rigid-plastic material, it can be said to be a more conservative approach
in that case.
The discrepancies of all results for all cases are summarized in Table 30. Note that in this table,
all results were taken at the end of the simulation time, i.e., when there was no more
deformation, and the results of SHARP were the results with two side shell super-elements
(with the presence of virtual deck).
Table 30 Summary of result discrepancy (cases without rupture strain)
Cases Penetration Deformation energy
1 27% 1%
2 1% 10%
3 69% 1%
4 8% 8%
5 13% 13%
6.2 Simulations considering Rupture (A.D.N. Regulations)
According to the A.D.N. Regulations, the probability of cargo tank rupture is determined by
using the energy absorption capacity of the struck vessel until the initial rupture of its cargo
tank. Therefore, the main focus in this section is to analyse and compare the results of SHARP
with those of LS-DYNA by defining every parameter exactly as suggested by the A.D.N.
Regulations. The comparisons will be made at 1 m penetration (not at the end of the simulation
time), that is, only when the initial rupture of the cargo tank occurs.
P 92 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
6.2.1 Case 1 (V-shape bow : 90 deg : At web : Under deck of struck ship)
Figure 48 compares the damage extent in LS-DYNA and SHARP at 1 m penetration. As can
be seen, the deck is crushed in LS-DYNA whereas it has not been activated in SHARP. This
coupling effect usually occurs when both striking and struck vessels have the same height. In
addition, the red part in SHARP simulation is showing that the side shell super-element has
been destroyed although in LS-DYNA, the side shell still remains unperforated.
(a) LS-DYNA (b) SHARP
Figure 48 Comparison of damage extent in LS-DYNA and SHARP – Case 1 (With rupture strain)
In Table 31, the comparison of the results of total deformation energy between LS-DYNA and
SHARP is presented when the penetration is at 1 m. It can be observed that there is a
discrepancy of 9 % and SHARP results are less than LS-DYNA in this case. Note that the
SHARP results given in Table 31 are the average value of all the results calculated at 9 impact
locations.
Table 31 Comparison of the results – Case 1 (With rupture strain)
Deformation Energy [MJ]
LS-DYNA 6.22
SHARP (average) 5.69
% Difference 9%
The results are also compared graphically in Figure 49 and good agreement is found between
LS-DYNA and SHARP. Also, SHARP results are slightly more conservative in this case.
However, with the knowledge gained from the previous simulations (Refer Section 6.1), the
following additional analyses could be made:
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
93
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
Figure 49 Comparison of the results – Case 1 (With rupture strain)
Firstly, there is a virtual deck which splits the side shell super-element of SHARP into
two. This causes an extra clamped end in the middle of these two elements and as a
result, the shell becomes more rigid. (Refer to Section 5.2.2.1)
Secondly, the weather deck in LS-DYNA was crushed during the collision. As a
consequence, the side shell in SHARP is bounded by four clamped ends while it behaves
like a plate with only three clamped ends and one moving end in LS-DYNA due to the
fact that the weather deck deforms.
Such combined effects from the first and the second might have caused the side shell in
SHARP to be stiffer than what it actually should be (before its rupture), leading to better
results in comparison with LS-DYNA. However, one has to be aware that if the
developments suggested in the previous sections were considered, then the results
would not be the same and in this case, they may become much smaller than LS-DYNA.
In this context, the behaviour of the side shell super-element becomes significant. In
SHARP, it is assumed that when the failure strain of the side shell exceeds the
predefined criterion, the associated resistant force is directly imposed to zero in the next
calculation step. However, in LS-DYNA (in reality as well), the resistant force will
slowly decrease until it reaches zero.
0.00E+00
1.00E+06
2.00E+06
3.00E+06
4.00E+06
5.00E+06
6.00E+06
7.00E+06
8.00E+06
0.00 0.20 0.40 0.60 0.80 1.00
Def
orm
atio
n E
ner
gy
[J]
Penetration [m]
Struck ship - Deformation Energy (Case 1)
LS-DYNA
Simulation 1
Simulation 2
Simulation 3
Simulation 4
Simulation 5
Simulation 6
Simulation 7
Simulation 8
Simulation 9
P 94 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
Therefore, to conclude case 1, it is important to consider how to correctly represent the
decreasing slope of the side shell resistance so that the loose of stiffness in the side shell
super-element caused by the developments could be properly compensated.
6.2.2 Case 2 (V-shape bow : 90 deg : Between webs : Mid-depth of struck ship)
Figure 50 shows the damage extent in LS-DYNA and SHARP at 1 m penetration. As can be
seen, the weather deck in LS-DYNA is only slightly deformed whereas it has not been impacted
at all in SHARP. More importantly, it is found out that the side shell has been completely
destroyed in SHARP while it is still unperforated in LS-DYNA. So, it is obvious that the side
shell super-element fails too rapidly in SHARP and, as a consequence, the side shell resistance
drops to zero too quickly.
(a) LS-DYNA (b) SHARP
Figure 50 Comparison of damage extent in LS-DYNA and SHARP – Case 2 (With rupture strain)
In Table 32, the results of total deformation energy obtained from LS-DYNA and SHARP at 1
m penetration are compared. It can be observed that there is a very large discrepancy (about 82
%) in the total deformation energies computed by SHARP and LS-DYNA. Note that the
SHARP results given in Table 32 are the average result of all 9 simulations of SHARP.
Table 32 Comparison of the results – Case 2 (With rupture strain)
Deformation Energy [MJ]
LS-DYNA 5.29
SHARP (average) 0.95
% Difference 82 %
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
95
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
Figure 51 Comparison of the results – Case 2 (With rupture strain)
Figure 52 Comparison of the crushing force in LS-DYNA and SHARP (Case 2)
The results are also compared graphically and shown in Figure 51. It appears that SHARP
results are much less, approximately 5 times, than LS-DYNA results. Although it is
conservative, the result discrepancy is still very large. The reason is due to the premature failure
of the side shell super-element in SHARP. As shown in Figure 52, the evolution of crushing
0.00E+00
1.00E+06
2.00E+06
3.00E+06
4.00E+06
5.00E+06
6.00E+06
7.00E+06
0.00 0.20 0.40 0.60 0.80 1.00
Def
orm
atio
n E
ner
gy
[J]
Penetration [m]
Struck ship - Deformation Energy (Case 2)
LS-DYNA
Simulation 1
Simulation 2
Simulation 3
Simulation 4
Simulation 5
Simulation 6
Simulation 7
Simulation 8
Simulation 9
0.00E+00
1.00E+07
2.00E+07
3.00E+07
4.00E+07
5.00E+07
6.00E+07
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Cru
shin
g F
orc
e [N
]
Time [sec]
SHARP lsdyna
P 96 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
force in SHARP is a lot smaller (about 10 times) than LS-DYNA. In addition, the crushing
force in LS-DYNA slowly decreases until it reaches zero while in SHARP, the side shell does
not participate in the overall struck ship resistance at all after its failure. This is the development
that is mainly needed to consider for the side shell super-element in SHARP in order to correctly
represent the decreasing slope of its resistance when it begins to fail. Also, investigations should
be made why the side shell super-element fails so rapidly.
6.2.3 Case 3 (Push barge bow : 55 deg : At web : Mid-depth of struck ship)
Case 3 is the scenario in which the tanker has been impacted by push barge. The angle of impact
considered is 55 degree together with the constant striking speed of 10 m/s. The location of
impact point is at web and at mid-depth of the struck ship. Failure modelling of 20 % was used
for both LS-DYNA and SHARP simulations.
In Figure 53, the damage extent in LS-DYNA and SHARP at 1 m penetration are compared. It
can be observed how the difference in the bow shape geometry can lead to different results.
Even though both figures are taken at the same damage penetration of 1 m, the longitudinal
extent of damage is small in LS-DYDA (See Figure 53a). On contrary, the longitudinal extent
of damage is larger in SHARP (Figure 53b). Thus, more elements will participate in the
collision in SHARP while only a few elements will resist the collision in LS-DYNA. This
makes the assumptions between the two approaches to be completely different.
(a) LS-DYNA (b) SHARP
Figure 53 Comparison of damage extent in LS-DYNA and SHARP – Case 3 (With rupture strain)
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
97
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
Table 33 shows the comparison of the total deformation energy between LS-DYNA and
SHARP when the penetration is 1 m. It can be observed that there is a discrepancy of 27 % in
which SHARP results are higher than LS-DYNA results. Note that the SHARP results given in
Table 33 are the average value of the results calculated at 9 impact locations.
Table 33 Comparison of the results – Case 3 (With rupture strain)
Deformation Energy [MJ]
LS-DYNA 4.4
SHARP (average) 5.6
% Difference 27 %
Figure 54 Comparison of the results – Case 3 (With rupture strain)
The results are also compared graphically and shown in Figure 54. According to the figure, it
can be seen that all SHARP results correlate well with the LS-DYNA. However, the results
calculated by SHARP are higher than LS-DYNA at 1 m penetration. As already mentioned, due
to the difference in geometry of the striking ship bow, the longitudinal extent of damage is not
the same even though both results are taken at the same penetration value.
It can be predicted that the results of SHARP will become higher after 1 m penetration because
the webs and inner shell have already participated in the collision process while in LS-DYNA,
only the side shell and very few elements have taken part in the collision. For this case, if a
0.00E+00
1.00E+06
2.00E+06
3.00E+06
4.00E+06
5.00E+06
6.00E+06
7.00E+06
8.00E+06
0.00 0.20 0.40 0.60 0.80 1.00
Def
orm
atio
n E
ner
gy
[J]
Penetration [m]
Struck ship - Deformation Energy (Case 3)
LS-DYNA
Simulation 1
Simulation 2
Simulation 3
Simulation 4
Simulation 5
Simulation 6
Simulation 7
Simulation 8
Simulation 9
P 98 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
more precise bow shape could be considered, SHARP results would become a lot smaller due
to the fact that the side shell is very rapidly destroyed.
6.2.4 Case 4 (Push barge bow : 55 deg : At bulkhead : Above deck of struck ship)
Case 4 is the scenario in which the tanker has been collided by a push barge which has a constant
speed of 10 m/s at 55 degree collision angle. The point of impact is located at bulkhead and just
above deck of the struck ship. The failure strain used for both LS-DYNA and SHARP
simulations is 20 %. As can be seen in Figure 55, while the shell in LS-DYNA breaks only a
little, the whole side shell has already failed in SHARP.
(a) LS-DYNA (b) SHARP
Figure 55 Comparison of damage extent in LS-DYNA and SHARP – Case 4 (With rupture strain)
Table 34 shows the comparison of the total deformation energy between LS-DYNA and
SHARP when the penetration is 1 m. Note that the SHARP results presented are the average
value of the results calculated at 9 impact locations.
Table 34 Comparison of the results – Case 4 (With rupture strain)
Deformation Energy [MJ]
LS-DYNA 10.8
SHARP (average) 3.9
% Difference 64 %
The results are also compared graphically and shown in Figure 56. According to the figure, a
discrepancy of 64 % is observed while the value calculated by SHARP is less than the one
calculated by LS-DYNA. The main reason is the same as case 2 in which the contribution of
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
99
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
the side shell super-element in the overall struck ship resistance is many times smaller than that
of LS-DYNA. This should be investigated in the future.
Figure 56 Comparison of the results – Case 4 (With rupture strain)
6.2.5 Case 5 (V-shape bow : 90 deg : At web : Above deck of struck ship)
In this case, the tanker has been impacted perpendicularly by the V-shape bow with a constant
speed of 10 m/s. The location of impact is considered at web and just above deck of the struck
ship. The failure strain used for both LS-DYNA and SHARP simulations is 20 %. Figure 57
shows the damage penetration extracted with LS-DYNA post processer, LS-PRE/POST and by
SHARP graphical interface respectively. Only the upper part of the side shell super-element in
SHARP seems to have ruptured.
Table 35 shows the comparison of the results of total deformation energy between LS-DYNA
and SHARP when the penetration is 1 m. It can be seen that there is a discrepancy of only 1%
with SHARP result being slightly more conservative. Note that the value presented for SHARP
is only the average value of the results calculated at 9 impact locations.
0.00E+00
2.00E+06
4.00E+06
6.00E+06
8.00E+06
1.00E+07
1.20E+07
0.00 0.20 0.40 0.60 0.80 1.00
Def
orm
atio
n E
ner
gy [
J]
Penetration [m]
Struck ship - Deformation Energy (Case 4)
LS-DYNA
Simulation 1
Simulation 2
Simulation 3
Simulation 4
Simulation 5
Simulation 6
Simulation 7
Simulation 8
Simulation 9
P 100 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
(a) LS-DYNA (b) SHARP
Figure 57 Comparison of damage extent in LS-DYNA and SHARP – Case 5 (With rupture strain)
Table 35 Comparison of the results – Case 5 (With rupture strain)
Deformation Energy [MJ]
LS-DYNA 7.12
SHARP (average) 7.03
% Difference 1 %
Figure 58 Comparison of the results – Case 5 (With rupture strain)
0.00E+00
2.00E+06
4.00E+06
6.00E+06
8.00E+06
1.00E+07
1.20E+07
0.00 0.20 0.40 0.60 0.80 1.00
Def
orm
atio
n E
ner
gy
[J]
Penetration [m]
Struck ship - Deformation Energy (Case 5)
LS-DYNA
Simulation 1
Simulation 2
Simulation 3
Simulation 4
Simulation 5
Simulation 6
Simulation 7
Simulation 8
Simulation 9
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
101
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
The results are also compared graphically as shown in Figure 58. According to the figure, it can
be said that all SHARP results correlate well with LS-DYNA. The trends of both approaches
are almost similar. This is because in this case, the geometry of the bow shape is similar and
also there are some factors, as explained in case 1 (Section 6.2.1) which could make the side
shell in SHARP to be stiffer than what it actually should be. It would be interesting to see what
the consequences will be if all the developments suggested were to be considered.
6.2.6 Overall Analysis
Studying the simulations where the rupture has been considered (from Section 6.2.1 to 6.2.5),
the following overall analyses could be made:
The location of the impact point and the associated striking bow are the deciding factors
in the agreement of the results. For example, if V-shape bow is used instead of the push
barge, the results should be the same. However, it is found out that only cases 1 and 5
are found to be in good accords with LS-DYNA while there is a very large discrepancy
of about 82 % for case 2. The reason is due to the difference in the locations of the
impact point considered. As both cases 1 and 5 considered the impact point around the
vicinity of the weather deck, the results are quite good because the deck super-element
contributes more in the collision rather than the side shell super-element. As for case 2,
however, the results do not match well because the impact point considered is at the
mid-depth of the struck ship where the side shell super-element plays a more important
role. Since the side shell fails too prematurely, this causes an underestimation in the
overall struck ship resistance. This highlights the need to consider more stiffness as well
as the behaviour after side shell rupture in the solver of SHARP. In the paper of
(Kitamura, 2002), it was stated that a simplified analytical approach should consider the
post rupture-initiation/propagation behaviour of the side shell in order to correctly
represent the overall resistance and the energy absorption capability of the struck ship.
Regarding the bow shape model, the push barge bow cannot be modelled as exactly as
the one defined by A.D.N. Regulations due simply to the fact that SHARP still needs
more surface tools to allow for a more precise bow shape description. This leads to some
discrepancies as compared with LS-DYNA, for example, cases 3 and 4.
The summary of all the result discrepancies with the failure strain can be seen in Table 36. Note
that both LS-DYNA and SHARP results are taken and compared only at the penetration length
P 102 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
of 1 m (not at the end of the simulations), i.e., only when the initial rupture of the cargo tank
occurs, in accordance with the A.D.N Regulations.
Table 36 Summary of result discrepancy (cases with rupture strain)
Cases Deformation energy
1 9%
2 82%
3 27%
4 64%
5 1%
6.3 Additional Simulations with Modified Rupture Strains
In the previous sections, comparisons have been made between LS-DYNA numerical approach
and SHARP analytical approach only up to 1 m penetration. According to the results, it has
been found out that the side shell super-element of SHARP fails too rapidly and, as its resistance
drops suddenly to zero, does not resist sufficiently. Therefore, in this section, the rupture strain
value considered in SHARP will be tuned in order to correctly model the loose of stiffness in
the side shell and the simulations will be run again. The obtained results will be compared with
LS-DYNA. Note that 20 % failure strain was used in LS-DYNA simulations for every
comparison. A list of new rupture strain values considered in SHARP simulations is given in
Table 37.
Table 37 Different rupture strain values considered in SHARP
Cases Rupture strain Ec
Case 1 2
Case 2 5
Case 3 0.5
Case 4 3
Case 5 3
6.3.1 Case 1 (V-shape bow : 90 deg : At web : Under deck of struck ship)
Table 38 presents the comparison of the results from LS-DYNA and an average result computed
from all 9 impact locations of SHARP. Both results are taken at the end of the simulations, i.e.,
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
103
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
when there is no more deformation. The results are also compared graphically as shown in
Figure 59.
Table 38 Comparison of the results – case 1 (with modified rupture strain)
Penetration [m] Deformation Energy [MJ]
LS-DYNA 4.5 111
SHARP (average) 4.37 84.67
% Difference 3% 24%
Figure 59 Comparison of the results– case 1 (with modified rupture strain)
According to the results, it can be seen that SHARP results are very close to the LS-DYNA
ones. The discrepancy for the penetration is 3 % while the discrepancy for the deformation
energy is 24 %. In both results, SHARP is less because in LS-DYNA, when collision occurs,
the weather deck was crushed and thus, absorbed some amount of energy during the process.
On the other hand, the weather deck in SHARP has not been impacted at all. This is because in
SHARP, the elements are independently activated when collision occurs. (See Figure 60 and
Figure 61)
In addition, the crush of the weather deck in LS-DYNA has caused different boundary
conditions for the side shell considered in LS-DYNA and SHARP. In other words, the side shell
in LS-DYNA behaves like a plate with three clamped ends and one moving end (at the weather
0.00E+00
2.00E+07
4.00E+07
6.00E+07
8.00E+07
1.00E+08
1.20E+08
0.00 1.00 2.00 3.00 4.00 5.00
Def
orm
atio
n E
ner
gy
[J]
Penetration [m]
Struck ship - Deformation Energy (Case 1)
LS-DYNA
Simulation 1
Simulation 2
Simulation 3
Simulation 4
Simulation 5
Simulation 6
Simulation 7
Simulation 8
Simulation 9
P 104 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
deck). On contrary, the side shell in SHARP is considered as four clamped ends. Hence, before
the side shell fails, the deformation energy value of SHARP is higher than LS-DYNA, i.e.,
around 1 m damage penetration.
Figure 60 View of the deformation of weather deck and bottom in LS-DYNA
Figure 61 View of the activated elements in SHARP (weather deck has not been impacted at all)
The results are also compared as the percentages of the total energy shown in Table 39. Note
that the values of SHARP presented in this table are taken only from the simulation 1 among
the 9 simulations of SHARP. Also, the results of SHARP with the virtual deck (with two side
shell super-elements) have been provided in this table because the sensitivity is not very high
due to the very fast velocity of the striking ship.
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
105
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
Table 39 Comparison of the energy dissipation – case 1 (with modified rupture strain)
PARTS LS-DYNA SHARP
E (MJ) % E (MJ) %
Total Energy 111 86.2
Side Shell 29 26% 10.2 12%
Inner Shell 10.4 9% 4.22 5%
Web Frame 12.8 12% 10.6 12%
Double Bottom 7.8 7% 0.24 0%
Bottom 4.7 4% 0 0%
Weather Deck 18 16% 0 0%
Stiffeners 5.8 5% 60.9 71%
Others 22.5 21% 0.04 0%
Penetration [m] 4.5 4.16
According to the results from Table 39, the following analyses could be made:
In LS-DYNA, the energy is well distributed between different structural components.
In SHARP, however, most of the total energy (about 71 %) is dissipated in stiffeners
which seems to be unrealistic. The reason for this should be investigated and improved
in the near future.
It can be observed that the weather deck in LS-DYNA absorbs a good amount of energy
(about 16 %) whereas the weather deck in SHARP has not been impacted at all. This
coupling effect between deck, bottom and side shell can usually be observed in the
collision of inland vessels.
In addition, it should be noticed that even with the modified failure strain, the side shell
super-element in SHARP fails more rapidly than in LS-DYNA. The amount of energy
dissipation can be compared, accounting for 29 MJ in LS-DYNA but only 10.2 MJ in
SHARP. This highlights the need to improve the modelling of the failure of the side
shell in SHARP.
P 106 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
6.3.2 Case 2 (V-shape bow : 90 deg : Between webs : Mid-depth of struck ship)
The results of SHARP and LS-DYNA for case 2 are compared in Table 40. Note that both
results are taken at the end of the simulations and SHARP result is the average result of all 9
impact scenarios. The results are also compared graphically in Figure 62. As can be seen,
SHARP results correspond well with LS-DYNA. The discrepancy is 2 % for the penetration
and 18 % for the deformation energy.
Table 40 Comparison of the results – case 2 (with modified rupture strain)
Penetration [m] Deformation Energy [MJ]
LS-DYNA 4.4 103
SHARP (average) 4.31 84.12
% Difference 2% 18%
Figure 62 Comparison of the results– case 2 (with modified rupture strain)
In this case, however, LS-DYNA results are higher in both penetration and deformation energy.
The reason is the same with case 1 in which the weather deck also deformed in LS-DYNA even
though it was not impacted. Figure 63 presents LS-DYNA view of the damage extent when the
weather deck was deformed due to the impact. In Figure 64, the view of the damage extent in
SHARP is presented.
0.00E+00
2.00E+07
4.00E+07
6.00E+07
8.00E+07
1.00E+08
1.20E+08
0.00 1.00 2.00 3.00 4.00 5.00 6.00
Def
orm
atio
n E
ner
gy [
J]
Penetration [m]
Struck ship - Deformation Energy (Case 2)
LS-DYNA
Simulation 1
Simulation 2
Simulation 3
Simulation 4
Simulation 5
Simulation 6
Simulation 7
Simulation 8
Simulation 9
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
107
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
Figure 63 View of the bending of weather deck due to deformation of the side shell
Figure 64 View of the activated super-elements in SHARP (The weather deck has not been collided)
In Table 41, the energy dissipation of each structural element is shown as a percentage of the
total energy. It can be observed that energy is well distributed in LS-DYNA for each component
while in SHARP, stiffeners absorb more than half of the total energy (about 66 %). The reason
for this should be investigated as already mentioned in case 1. Another important fact found
when observing Table 41 is that the webs are found to have destroyed and can only absorb 1.44
MJ which is very small (by about 8 times smaller) as compared with LS-DYNA. This effect
was not seen in case 1 as the striking ship position is just under deck of struck ship. The
sensitivity of the collision position of the web should be investigated.
P 108 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
Table 41 Comparison of the energy dissipation – case 2 (with modified rupture strain)
PARTS LS-DYNA SHARP
E (MJ) % E (MJ) %
Total Energy 103 83.3
Side Shell 28.5 28% 21.4 26%
Inner Shell 9 9% 4.12 5%
Web Frame 12.3 12% 1.44 2%
Double Bottom 14 14% 0.02 0%
Bottom 9.9 10% 1.16 1%
Weather Deck 5.2 5% 0 0%
Stiffeners 6.3 6% 55.2 66%
Others 17.8 17% -0.04 0%
Penetration [m] 4.4 4.34
6.3.3 Case 3 (Push barge bow : 55 deg : At web : Mid-depth of struck ship)
In Table 42, the results calculated by SHARP and LS-DYNA are compared numerically. Note
that in this case, the results are not the end simulation results but instead, the ones taken at 5.8
m penetration damage. The reason is that the striking ship has already penetrated more than half
the breath of the struck ship and in this case, SHARP simulation is forced to stop. Nevertheless,
there is only 1 % discrepancy in the deformation energy calculated from LS-DYNA and
SHARP at 5.8 m penetration.
Table 42 Comparison of the results – case 3 (with modified rupture strain)
Penetration [m] Deformation Energy [MJ]
LS-DYNA 5.8 29
SHARP (average) 5.8 28.8
% Difference - 1%
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
109
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
Figure 65 Comparison of the results– case 3 (with modified rupture strain)
In Figure 65, it can be seen that the trends of LS-DYNA and SHARP are similar before 1 m
penetration and after 4 m penetration. It is observed that LS-DYNA curve goes up suddenly at
the penetration value of about 5.8 m. This is due to the striking ship reaching the centreline
bulkhead of the struck ship.
In addition, it is observed that LS-DYNA results show the decreasing positive slope because
the side shell which has ruptured will continue to resist the collision (See Figure 66). On the
other hand, in SHARP, after the side shell breaks, nothing is there to resist the collision anymore
and the crushing resistance is assumed to have dropped to zero. The side shell crushing
resistances obtained from LS-DYNA and SHARP are compared in Figure 67.
According to Figure 67, it is obvious that the side shell impact resistances calculated from LS-
DYNA and SHARP are very different in nature as well as in amplitudes. It is seen that the
resistance given by LS-DYNA is almost 6 times larger than the one calculated by SHARP. This
is due to the fact that the penetration is very localized which leads to a very rapid rupture of the
side shell in SHARP while this is not the case in LS-DYNA.
0.00E+00
5.00E+06
1.00E+07
1.50E+07
2.00E+07
2.50E+07
3.00E+07
3.50E+07
4.00E+07
0.00 1.00 2.00 3.00 4.00 5.00
Def
orm
atio
n E
ner
gy
[J]
Penetration [m]
Struck ship - Deformation Energy (Case 3)
LS-DYNA
Simulation 1
Simulation 2
Simulation 3
Simulation 4
Simulation 5
Simulation 6
Simulation 7
Simulation 8
Simulation 9
P 110 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
Figure 66 View of the side shell which has ruptured being still there and resisting the collision – case 3
Figure 67 Comparison of the crushing resistance of the side shell between LS-DYNA and SHARP
Table 43 highlights the amount of energy dissipation in each of the structural component in LS-
DYNA and SHARP. Note that the values presented in the table are taken at 5.8 m penetration
for both simulations. It can be seen that in this case too, the stiffeners in SHARP absorb very
high amount of energy, accounting for 79 % of the total, which seems to be unrealistic. The
web in LS-DYNA has 17 % energy dissipation while the web in SHARP has 10 %.
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
111
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
Table 43 Comparison of the energy dissipation – case 3 (with modified rupture strain)
PARTS LS-DYNA SHARP
E (MJ) % E (MJ) %
Total Energy 29 33.2
Side Shell 7.26 25% 1.54 5%
Inner Shell 3.57 12% 1.73 5%
Web Frame 5.04 17% 3.44 10%
Double Bottom 0.09 0% 0 0%
Bottom 0.05 0% 0.14 0%
Weather Deck 0.09 0% 0 0%
Stiffeners 3.4 12% 26.3 79%
Others 9.5 34% 0.05 1%
Penetration [m] 5.8 5.8
6.3.4 Case 4 (Push barge bow : 55 deg : At bulkhead : Above deck of struck ship)
Table 44 presents the comparison of the results calculated by LS-DYNA and SHARP. Note that
both results in this case were taken at the end of the simulation time and SHARP results are the
average one calculated from all 9 simulation results. It is found out that there is 4 % discrepancy
in the penetration and 21 % discrepancy in the deformation energy.
Table 44 Comparison of the results – case 4 (with modified rupture strain)
Penetration [m] Deformation Energy [MJ]
LS-DYNA 3.61 55.5
SHARP (average) 3.45 43.80
% Difference 4% 21%
In Figure 68, the results are again compared graphically. It can be observed that with the
modified rupture strain, all SHARP results are in good accord with LS-DYNA and slightly
more conservative too.
P 112 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
Figure 68 Comparison of the results – case 4 (with modified rupture strain)
Table 45 compares the amount of energy dissipation in each of the structural component in LS-
DYNA and SHARP in the form of percentages. Note that the values of SHARP shown in this
table are only the results of simulation 1 in which the real impact point was considered.
Surprisingly, in this case, the side shell in SHARP absorbs more than half of the total energy.
Table 45 Comparison of the energy dissipation – case 4 (with rupture strain)
PARTS LS-DYNA SHARP
E (MJ) % E (MJ) %
Total Energy 55.5 45.8
Side Shell 8 14% 23.7 52%
Inner Shell 3 5% 0.34 1%
Web Frame 5.5 10% 8.42 18%
Double Bottom 0.02 0% 0.15 0%
Bottom 0 0% 0 0%
Weather Deck 21.6 39% 6.6 14%
Stiffeners 1.7 3% 6.6 14%
Others 15.68 28% -0.01 0%
Penetration [m] 3.61 3.74
0.00E+00
1.00E+07
2.00E+07
3.00E+07
4.00E+07
5.00E+07
6.00E+07
0.00 1.00 2.00 3.00 4.00
Def
orm
atio
n E
ner
gy
[J]
Penetration [m]
Struck ship - Deformation Energy (Case 4)
LS-DYNA
Simulation 1
Simulation 2
Simulation 3
Simulation 4
Simulation 5
Simulation 6
Simulation 7
Simulation 8
Simulation 9
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
113
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
However, with the modified value of rupture strain, it is understandable that the behaviour of
side shell are different for different cases. In this case, the rupture strain value used is ‘3’ which
makes the side shell to become more resistant. This in turn has caused a decrease of energy
dissipation in the stiffeners. This case somehow highlights that if the side shell super-element
was correctly modelled, the stiffeners would correspond correctly. Nevertheless, modelling of
both structural failure, i.e., stiffeners and side shell, are to be investigated more in the future.
The energies dissipated by the web frames are found to be not too different from LS-DYNA,
showing 5.5 MJ in LS-DYNA and 8.4 MJ in SHARP.
6.3.5 Case 5 (V-shape bow : 90 deg : At web : Above deck of struck ship)
The results of case 5 are compared in Table 46. It can be seen that there is 10 % discrepancy
for penetration and 6 % discrepancy for deformation energy. Note that both simulation results
are taken at the end of the simulation time.
Table 46 Comparison of the results – case 5 (with modified rupture strain)
Penetration [m] Deformation Energy [MJ]
LS-DYNA 3 56.4
SHARP (average) 3.31 59.53
% Difference 10% 6%
Figure 69 Comparison of the results – case 5 (with modified rupture strain)
0.00E+00
1.00E+07
2.00E+07
3.00E+07
4.00E+07
5.00E+07
6.00E+07
7.00E+07
0.00 1.00 2.00 3.00 4.00
Def
orm
atio
n E
ner
gy
[J]
Penetration [m]
Struck ship - Deformation Energy (Case 5)
LS-DYNA
Simulation 1
Simulation 2
Simulation 3
Simulation 4
Simulation 5
Simulation 6
Simulation 7
Simulation 8
Simulation 9
P 114 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
In Figure 69, the results are again compared graphically and a good agreement is found between
LS-DYNA and SHARP results. However, between the penetration damage of 0.5 m and 2 m,
SHARP results are found to be a bit higher (almost twice) than LS-DYNA results. The reason
is probably due to the behaviour of the side shell super-element. At the start of the collision, it
will have more stiffness due to the high rupture strain value (Ec = 3) used in this SHARP
simulation. The stiffness will abruptly decrease when the calculated strain exceeds the criteria,
and in this case, it is obvious that the overall deformation energy will decrease too. However,
this makes the results of LS-DYNA and SHARP to be in better comparison.
In Table 47, the comparison of the energy dissipation is given in the form of percentages of the
total energy. It can be observed that the energy distributions of different structural components
are in quite good agreement with LS-DYNA except for stiffeners and weather deck.
The weather deck of SHARP absorbs 1.6 times less energy than the weather deck of LS-DYNA
does. As in the case for stiffeners, similar to the previous simulations, the energy dissipation of
stiffeners in SHARP is very high (about 45 % of the total energy). Therefore, it could be said
that the extra energy absorbed by other elements in LS-DYNA such as brackets, deck transverse
beams, etc. seemed to have been compensated by the extra energy absorbed by the stiffeners in
SHARP.
Table 47 Comparison of the energy dissipation – case 5 (with modified rupture strain)
PARTS LS-DYNA SHARP
E (MJ) % E (MJ) %
Total Energy 56.4 60.2
Side Shell 13.5 24% 12.5 21%
Inner Shell 5.5 10% 3.42 6%
Web Frame 6.9 12% 6.65 11%
Double Bottom 0.3 1% 0.4 1%
Bottom 0.15 0% 0 0%
Weather Deck 16.2 29% 9.94 17%
Stiffeners 3 5% 27.3 45%
Others 10.85 19% -0.01 0%
Penetration [m] 3 3.43
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
115
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
6.3.6 Overall Analysis
According to the results from case 1 to case 5 (from Section 6.3.1 to Section 6.3.5), it can be
observed that with the tuned rupture strain, the results of SHARP correspond better with LS-
DYNA except for case 3. However, it should be noted that cases 3 and 4 use the push barge
bow whose shape has been modelled in SHARP with a lot of geometrical simplifications.
In most of the cases except in cases 3 and 5, the dissipated energies calculated by LS-DYNA
are found to be higher (around 20%) than those as assessed by SHARP. This is due to the fact
that, for inland vessels, in which the height of struck and striking ships are similar, the weather
deck and the bottom are simultaneously deformed in LS-DYNA even if they are not directly
impacted. On contrary, the super-elements are independently activated upon contact in SHARP.
In other words, some coupling between decks, bottom and side shell needs to be taken into
account even though they are not being impacted. Theoretical developments have already been
performed and presented in (Buldgen et al., 2013) but not implemented in SHARP solver yet.
The summary of penetration and deformation energy at the end of the simulation, i.e., when
there is no more deformation, is given in Table 48 below. The values presented for SHARP are
obtained by averaging the 9 scenario results.
Table 48 Result discrepancy of the simulations (cases with modified rupture strain)
Cases Penetration Deformation energy
1 3% 24%
2 2% 18%
3* - 1%
4 4% 21%
5 10% 6%
*Note that case 3 is the only case in which the comparison is made at the 5.8 m penetration damage.
All things considered, one of the interesting facts about this study is how to adapt the failure
modelling for the super-elements in order to correctly simulate the ship collision. In Appendix
B, the failure strain sensitivity check has been performed by applying different values of rupture
strain Ec in SHARP simulations and the results are compared with LS-DYNA.
P 116 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
7. CONCLUSIONS AND RECOMMENDATIONS
In this master thesis, some validation tests for a simplified analytical tool SHARP have been
performed by comparing the results with non-linear finite element explicit code, LS-DYNA.
The modelling of the colliding ships involved and the associated scenarios have been defined
within the scope of A.D.N. Regulations.
Generally, the simulations can be divided into two categories:
Simulations without modelling rupture at constant striking ship’s speed of 3 m/s; and
Simulations accounting for rupture at constant striking ship’s speed of 10 m/s.
According to the results obtained, the following conclusions and recommendations could be
made:
The use of the virtual deck which is defined with zero thickness to serve as a geometrical
limit during modelling of the struck ship is found to have caused some effect in the side
shell super-element of SHARP. As a consequence, the side shell behaves more rigidly
than it should be and this leads to some inconsistent results especially in cases where
rupture is disregarded. Therefore, it is recommended to improve the user interface of
SHARP in order to correctly characterize the actual structural behaviour of the super-
elements.
The coupling effect that occurs when the weather deck and bottom are not heavily
constructed as compared with the side shell and/or when the depths of the colliding ships
are similar (usually for inland ships) has not been considered in SHARP since the
application of the tool is originally intended for the collisions of FPSO and ocean-going
tankers. Analytical formulations have been developed by (Buldgen et al., 2013) and in
the future version of SHARP, it will become possible to take into account such effect.
Also, the geometrical simplifications that have to be made when modelling the push
barge has led to some discrepancy in the results. Therefore, more vessel lines and/or
surface tools are suggested to implement in the SHARP modelling interface to allow for
a more precise description of the barge bow shape.
In addition, a premature failing of the side shell super-element is observed when the
same rupture strain as LS-DYNA is applied. Moreover, it is also important to take into
account the post rupture behaviour of the side shell super-element to avoid an
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
117
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
underestimation in the overall struck ship resistance. In this thesis, see Section 6.3, this
lack of resistance in the side shell super-element has been compensated by defining a
larger rupture criterion. Nevertheless, this matter regarding the adaptation of rupture
strain for the super-element still requires an intensive research as the individual element
length le mentioned in A.D.N. cannot be applied for the super-elements due to the
methodology itself.
Regarding the material properties, a true stress-strain relation following the power law
has been applied for the LS-DYNA along with the consideration of elasto-plastic
material. On the other hand, a rigid-plastic material associated to a constant plastic flow
stress is defined in SHARP. However, neglecting the elastic part of the deformation can
sometimes have significant effect when the collision event involves smaller vessels with
slow speed applications.
Finally, the behaviour of the stiffeners that have not failed even in the event of shell
plating rupture should be investigated.
Considering all these aforementioned facts, it is obvious that the SHARP program needs more
developments in the solver as well as in the user interface. Nevertheless, SHARP still promises
a potential considering the time required for the simulations (a few seconds) as compared with
the Finite Element simulations (a few days). Also, the user interface of SHARP is very easy to
use since a full structural model can be generated within a few days unlike the building of Finite
Element models that usually requires tremendous effort and an immense amount of time.
To conclude, in the near future after all the developments will have been considered, SHARP
might become an effective as well as a reliable tool that can substitute the conventional finite
element method in terms of rapidity and simplicity when it comes to ship collision analysis.
P 118 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
ACKNOWLEDGEMENTS
First of all, I would like to give my wholehearted thanks to all the teachers and mentors who
have thought me and guided me to be in the right direction and to be able to proudly stand in
this current position.
Secondly, I would like to express my profound gratitude towards Prof. Hervé Le Sourne, my
thesis supervisor at l'Institut Catholique d'Arts et Métiers (ICAM), for sharing his insights and
valuable expertise with me to perform a successful study.
Thirdly, I would like to convey my genuine thanks to Stéphane Paboeuf, another supervisor of
mine from Bureau Veritas Marine & Offshore Division in Nantes, who always provided me
with precious advices and guidance throughout my internship period.
I would also like to deeply thank ICAM for their support with necessary computation power
which has let me finish all my LS-DYNA simulations in time. In addition, I would like to show
my appreciation to Bureau Veritas teams that welcomed me with open arms, creating an
atmosphere of warmth and friendliness.
Moreover, my deepest gratitude goes to Prof. Philippe Rigo, a coordinator of the EMSHIP
program, for inviting me to Europe and giving me a chance to participate in this challenging
master program in the first place. He is again the reviewer of this master thesis that deserves
him another round of my sincere regards.
Also, I would like to say great thanks to my parents for giving me encouragement and
motivation every time I needed it.
Last but not least, I would like to place on record my sense of gratitude to all the people, who
have directly or indirectly lent a helping hand in the development of this thesis.
This thesis was developed in the frame of the European Master Course in “Integrated Advanced
Ship Design” named “EMSHIP” for “European Education in Advanced Ship Design”, Ref.:
159652-1-2009-1-BE-ERA MUNDUS-EMMC.
Ye Pyae Sone Oo
Wednesday, 11 January 2017.
Carquefou, France.
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
119
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
REFERENCES
A.D.N. Regulations, 2015. European Agreement concerning the International Carriage of
Dangerous Goods by Inland Waterways. New York and Geneva: United Nations (UN).
A.D.N. Regulations, 2015. United Nations Economic Commission for Europe (UNECE)
[online]. Available from: http://www.unece.org/trans/danger/publi/adn/adn_e.html
[July 21, 2016].
Amdahl, J., 1983. Energy absorption in ship-platform impacts. Ph.D. Thesis. Department of
Marine Technology, Norwegian University of Science and Technology.
Besnard, N., 2014. SHARP 2 V1.0 User Guide. La Ciotat: PRINCIPIA.
Brown, A.J., 2002. Collision scenarios and probabilistic collision damage. 2nd International
Conference on Collision and Grounding (259-272). Copenhagen, Denmark, July 1-3.
Buldgen, L., Le Sourne, H. and Rigo, P., 2013. A simplified analytical method for estimating
the crushing resistance of an inclined ship side. Marine Structures, 265-296.
Buldgen, L., Le Sourne, H., Besnard, N. and Rigo, P., 2012. Extension of the super-elements
method to the analysis of oblique collision between two ships. Marine Structures 29.
(pp. 22-57). ELSEVIER.
Carlebur, A.F.C., 1995. Full-scale collision tests. Safety Science 19. (pp. 171-178). ELSEVIER.
Eurostat, 2015. Transport Accident Statistics [online]. Eurostat Statistics. Available from:
http://ec.europa.eu/eurostat/statistics-explained/index.php/Transport_accident_
statistics [June 15, 2016].
Hallquist, J.O., 2006. LS-DYNA Theory Manual. California: Livermore Software Technology
Corporation.
Hutchison, B.L., 1986. Barge collisions, rammings and groundings: an engineering assessment
of the potential for damage to radioactive material transport casks. Report No. SAND-
85-7165 TTC-05212.
ITOPF, 2016. Oil Tanker Spill Statistics 2015 [online]. Statistics - ITOPF. Available from:
http://www.itopf.com/fileadmin/data/Documents/Company_Lit/Oil_Spill_Stats_2016.
pdf [July 08, 2016]
P 120 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
Jones, N., 1997. Structural Impact. Cambridge: Cambridge University Press.
Kitamura, O., 2002. FEM approach to the simulation of collision and grounding damage.
Marine Structures, 403-428.
Le Sourne, H., 2015. Contribution à la modélisation de quelques problèmes de dynamique des
structures et de couplages fluide structure. Mémoire d’Habilitation à diriger des
Recherche. Université de Nantes, Ecole doctorale SPIGA.
Le Sourne, H., Besnard, N., Cheylan, C. and Buannic, N., 2012. A ship collision analysis
program based on upper bound solutions and coupled with a large rotational ship
movement analysis tool. Journal of Applied Mathematics. doi:10.1155/2012/375686
Le Sourne, H., Couty, N., Besnier, F., Kammerer, C. and Legavre, H., 2003. LS-DYNA
applications in shipbuilding. 4th European LS-DYNA Users Conference. Ulm,
Germany.
Le Sourne, H., Donner, R., Besnier, F. and Ferry M., 2001. External dynamics of ship–
submarine collision. In: Lutzen, M., Simonsen, B.C., Pedersen, P.T. and Jessen, V.,
editors. Proceedings of 2nd International Conference on Collision and Grounding of
Ships; 2001 July 1–3; Copenhagen, Denmark: Technical University of Denmark. p.
137–144.
Lehmann, E. and Peschmann, J., 2002. Energy absorption by the steel structure of ships in the
event of collisions. Marine Structures, 15, 429-441.
Lützen, M., Simonsen, B.C. and Pedersen, P.T., 2000. Rapid prediction of damage to struck
and striking vessels in a collision event. In SSC/SNAME/ASNE Symposium.
Minorsky, U.V., 1959. An analysis of ship collisions with reference to nuclear power plants.
Journal of Ship Research, Vol. 3, Page 1-4.
Naar, H., Kujala, P., Simonsen, B.C. and Ludolphy, H., 2002. Comparison of the
crashworthiness of various bottom and side structures. Marine Structures 15, 443-460.
Paboeuf, S., Le Sourne, H., Brochard, K. and Besnard, N., 2015. A damage assessment tool in
ship collisions. RINA Conference, Damaged Ship III. London, UK.
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
121
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
Paik, J.K., 2007. Practical techniques for finite element modeling to simulate structural
crashworthiness in ship collisions and grounding (Part I: Theory). In Ships and Offshore
Structures (Vol. 2:1, pp. 69-80). doi:10.1533/saos.2006.0148.
Pedersen, P.T. and Zhang, S., 1998. On impact mechanics in ship collisions. Marine Structures,
Vol. 11, pp. 429-449.
Pill, I., and Tabri, K., 2011. Finite element simulations of ship collisions: a coupled approach
to external dynamics and inner mechanics. In Ships and Offshore Structures (Vols.
Vol.6:1-2, pp. pp. 59-66). doi:10.1080/17445302.2010.509585
Rawson, C., Crake, K. (students) and Brown, A.J., 1998. Assessing the environmental
performance of tankers in accidental grounding and collision. SNAME Transactions,
Vol. 106, pp. 41-58.
Ship Structure Committee (SSC), 2002. Modelling Structural Damage in Ship Collisions.
Report No. 422.
Simonsen, B.C. and Lauridsen, L.P., 2000. Energy absorption and ductile fracture in metal
sheets under lateral indentation by a sphere. International Journal of Impact
Engineering, Vol. 24, 1017-1039.
Simonsen, B.C., 1997. Ship grounding on rocks - I Theory. Marine Structures, 10 (7), 519-562.
Simonsen, B.C., and Ocakli, H., 1999. Experiments and theory on deck and girder crushing. (J.
Loughlan, Ed.) Thin-Walled Structures, Vol. 34, 195-216.
Uzögüten, H. Ö., 2016. Application of super-element theory to crash-worthiness evaluation
within the scope of the A.D.N Regulations. Master Thesis, West Pomeranian University
of Technology, Szczecin.
Vidan, P., Kasum, J. and Misevic, P., 2012. Proposal of measures for increasing the safety level
of inland navigation. Journal of Maritime Research, Vol IX (1), pp. 57-62.
Wang, G. and Ohtsubo, H., 1997. Deformation of ship plate subjected to very large load. 16th
International Conference on Offshore Mechanics and Arctic Engineering (OMAE), Vol.
119, pp. 173–180.
Wierzbicki, T., 1995. Concertina tearing of metal plates. International Journal of Solids and
Structures, Vol. 32 (19), pp. 2923–2943.
P 122 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
Wu, F., Spong, R. and Wang, G., 2004. Using numerical simulation to analyze ship collision.
3rd International Conference on Collision and Grounding of Ships (ICCGS) (pp. 217-
224). Izu, Japan: ICCGS 2004.
Youssef, S.A.M., Kim, Y.S., and Paik, J.K., 2014. Hazard identification and probabilistic
scenario selection for ship-ship collision accidents. The International Journal of
Maritime Engineering, Vol 156, Part A1. doi:10.3940/rina.ijme.2014.a1.277
Zhang, S.M., 1999. The mechanics of ship collisions. Ph.D. Thesis, Technical University of
Denmark, Department of Naval Architecture and Offshore Engineering.
Zhang, S.M., 2002. Plate tearing and bottom damage in ship grounding. Marine Structures,
101-117.
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
123
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
APPENDIX
P 124 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
A. CONVERGENCE TESTS FOR MCOL SUB-CYCLING
Figure A - 1 MCOL convergence test for deformation energy (case 2 - without rupture strain)
Figure A - 2 MCOL convergence test for penetration (case 2 - without rupture strain)
0.0E+00
2.0E+07
4.0E+07
6.0E+07
8.0E+07
1.0E+08
1.2E+08
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Def
orm
atio
n E
ner
gy
[J]
Time [sec]
Deformation energy Vs TimeCase 2 (V-shape bow : 90 deg : Between webs : Mid-depth of struck ship)
MCOL_100 MCOL_200 MCOL_400
0.0
1.0
2.0
3.0
4.0
5.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Pen
etra
tio
n [
m]
Time [sec]
Penetration Vs TimeCase 2 (V-shape bow : 90 deg : Between webs : Mid-depth of struck ship)
MCOL100 MCOL_200 MCOL_400
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
125
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
Figure A - 3 MCOL convergence test for deformation energy (case 3 - without rupture strain)
Figure A - 4 MCOL convergence test for penetration (case 3 - without rupture strain)
0.0E+00
2.0E+07
4.0E+07
6.0E+07
8.0E+07
1.0E+08
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Def
orm
atio
n E
ner
gy
[J]
Time [sec]
Deformation energy Vs TimeCase 3 (Push barge bow : 55 deg : At web : Mid-depth of struck ship)
MCOL_100 MCOL_200 MCOL_400
0.0
1.0
2.0
3.0
4.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Pen
etra
tio
n [
m]
Time [sec]
Penetration Vs TimeCase 3 (Push barge bow : 55 deg : At web : Mid-depth of struck ship)
MCOL100 MCOL_200 MCOL_400
P 126 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
Figure A - 5 MCOL convergence test for deformation energy (case 4 - with rupture strain)
Figure A - 6 MCOL convergence test for deformation energy (case 4 - with rupture strain)
0.0E+00
1.0E+07
2.0E+07
3.0E+07
4.0E+07
5.0E+07
6.0E+07
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Def
orm
atio
n E
ner
gy
[J]
Time [sec]
Deformation energy Vs TimeCase 4 (Push barge : 55 deg : At bulkhead : Above deck of struck ship)
MCOL_100 MCOL_200 MCOL_400
0.0
1.0
2.0
3.0
4.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Pen
etra
tio
n [
m]
Time [sec]
Penetration Vs TimeCase 4 (Push barge : 55 deg : At bulkhead : Above deck of strcuk ship)
MCOL100 MCOL_200 MCOL_400
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
127
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
B. CASE SENSITIVITY OF RUPTURE STRAIN
B.1 Case 1 (V-shape bow : 90 deg : At web : Under deck of struck ship)
Table B - 1 Evaluation of energy distribution using different rupture strains – Case 1
PARTS LS-DYNA
SHARP* SHARP
Ec = 0.2 Ec = 0.5
E (MJ) % E (MJ) % E (MJ) %
Total Energy 111 29.5 81.3
Side Shell 29 26% 1.26 4% 5.06 6%
Inner Shell 10.4 9% 1.26 4% 1.85 2%
Web Frame 12.8 12% 12.7 43% 11.8 15%
Double Bottom 7.8 7% 0.65 2% 0.39 0%
Bottom 4.7 4% 0 0% 0 0%
Weather Deck 18 16% 0 0% 0 0%
Stiffeners (Side) 5.8 5% 13.6 46% 62.2 77%
Others 22.5 20% 0.03 0% 0 0%
Penetration (m) 4.5 9.3 5.72
SHARP SHARP SHARP SHARP
Ec = 1 Ec = 2 Ec = 3 Ec = 5
E (MJ) % E (MJ) % E (MJ) % E (MJ) %
85.6 86.2 86.8 87.7
6.77 8% 10.2 12% 14.7 17% 21.9 25%
2.7 3% 4.22 5% 7.05 8% 9.44 11%
10.7 13% 10.6 12% 10.5 12% 10.4 12%
0.33 0% 0.24 0% 0.21 0% 0.13 0%
0 0% 0 0% 0 0% 0 0%
0 0% 0 0% 0 0% 0 0%
65.1 76% 60.9 71% 54.3 63% 45.8 52%
0 0% 0.04 0% 0.04 0% 0.03 0%
4.39 4.16 4 3.8
*Results taken at 1 sec of collision simulation time.
Ec : Rupture Strain
E : Deformation energy [MJ]
P 128 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
B.2 Case 2 (V-shape bow : 90 deg : Between webs : Mid-depth of struck
ship)
Table B - 2 Evaluation of energy distribution using different rupture strains – Case 2
PARTS LS-DYNA
SHARP* SHARP
Ec = 0.2 Ec = 0.5
E (MJ) % E (MJ) % E (MJ) %
Total Energy 103 20.1 77.7
Side Shell 28.5 28% 0.65 3% 1.79 2%
Inner Shell 9 9% 1.35 7% 1.75 2%
Web Frame 12.3 12% 2.3 11% 1.66 2%
Double Bottom 14 14% 0.05 0% 0.02 0%
Bottom 9.9 10% 2.75 14% 1.5 2%
Weather Deck 5.2 5% 0 0% 0 0%
Stiffeners 6.3 6% 13 65% 71 91%
Others 17.8 17% 0 0% -0.02 0%
Penetration (m) 4.4 8.9 5.13
SHARP SHARP SHARP SHARP
Ec = 1 Ec = 2 Ec = 3 Ec = 5
E (MJ) % E (MJ) % E (MJ) % E (MJ) %
80.9 81.8 82.1 83.3
3.48 4% 8.45 10% 11.7 14% 21.4 26%
4.22 5% 4.34 5% 4.03 5% 4.12 5%
1.52 2% 1.47 2% 1.43 2% 1.44 2%
0.02 0% 0.02 0% 0.02 0% 0.02 0%
1.26 2% 1.22 1% 1.2 1% 1.16 1%
0 0% 0 0% 0 0% 0 0%
70.4 87% 66.3 81% 63.8 78% 55.2 66%
0 0% 0 0% -0.08 0% -0.04 0%
4.76 4.65 4.59 4.34
*Results taken at 1 sec of collision simulation time.
Ec : Rupture Strain
E : Deformation energy [MJ]
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
129
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
B.3 Case 3 (Push barge : 55 deg : At web : Mid-depth of struck ship)
Table B - 3 Evaluation of energy distribution using different rupture strains – Case 3
PARTS LS-DYNA*
SHARP* SHARP*
Ec = 0.2 Ec = 0.5
E (MJ) % E (MJ) % E (MJ) %
Total Energy 52.7 17.8 37.4
Side Shell 12 23% 0.7 4% 1.54 4%
Inner Shell 7 13% 1.2 7% 2.11 6%
Web Frame 8.8 17% 3.94 22% 3.83 10%
Double Bottom 0.9 2% 0 0% 0 0%
Bottom 0.5 1% 0.13 1% 0.14 0%
Weather Deck 0.12 0% 0 0% 0 0%
Stiffeners 6 11% 11.8 66% 29.8 80%
Others 17.38 33% 0.03 0% -0.02 0%
Penetration (m) 7 7.7 6.7
SHARP* SHARP SHARP SHARP
Ec = 1 Ec = 2 Ec = 3 Ec = 5
E (MJ) % E (MJ) % E (MJ) % E (MJ) %
44 66.5 67.8 69.7
2.4 5% 6.15 9% 4.79 7% 6.28 9%
4.02 9% 7.51 11% 6.75 10% 8.73 13%
3.97 9% 4.49 7% 3.61 5% 3.28 5%
0.01 0% 0.02 0% 0.01 0% 0 0%
0.16 0% 0.2 0% 0.26 0% 0.33 0%
0 0% 0 0% 0 0% 0 0%
33.5 76% 48.1 72% 52.4 77% 51.1 73%
-0.06 0% 0.03 0% -0.02 0% -0.02 0%
7.3 6.8 5.69 5.1
*Results taken at 1 sec of collision simulation time.
Ec : Rupture Strain
E : Deformation energy [MJ]
P 130 Ye Pyae Sone Oo
Master Thesis developed at l'Institut Catholique d'Arts et Métiers, Carquefou
B.4 Case 4 (Push barge : 55 deg : At bulkhead : Above deck of struck ship)
Table B - 4 Evaluation of energy distribution using different rupture strains – Case 4
PARTS LS-DYNA
SHARP* SHARP*
Ec = 0.2 Ec = 0.5
E (MJ) % E (MJ) % E (MJ) %
Total Energy 55.5 32.5 38.5
Side Shell 8 14% 2.62 8% 6.19 16%
Inner Shell 3 5% 0.58 2% 1.24 3%
Web Frame 5.5 10% 11.3 35% 13.7 36%
Double Bottom 0.02 0% 3.64 11% 2.75 7%
Bottom 0 0% 0 0% 0 0%
Weather Deck 21.6 39% 10.7 33% 7.46 19%
Stiffeners 1.7 3% 3.58 11% 7.11 18%
Others 15.68 28% 0.08 0% 0.05 0%
Penetration (m) 3.61 5.68 5.39
SHARP SHARP SHARP SHARP
Ec = 1 Ec = 2 Ec = 3 Ec = 5
E (MJ) % E (MJ) % E (MJ) % E (MJ) %
43.3 44.3 45.8 47.4
14.2 33% 17.5 40% 23.7 52% 28.8 61%
0.9 2% 1.02 2% 0.34 1% 0.15 0%
10 23% 9.12 21% 8.42 18% 7.44 16%
0.84 2% 0.5 1% 0.15 0% 0 0%
0 0% 0 0% 0 0% 0 0%
7.11 16% 6.96 16% 6.6 14% 7.86 17%
10.2 24% 9.21 21% 6.6 14% 3.09 7%
0.05 0% -0.01 0% -0.01 0% 0.06 0%
4.44 4.16 3.74 3.22
*Results taken at 1 sec of collision simulation time.
Ec : Rupture Strain
E : Deformation energy [MJ]
Numerical and analytical simulations of in-shore ship collisions within the scope of
A.D.N. Regulations
131
“EMSHIP” Erasmus Mundus Master Course, period of study September 2015 – February 2017
B.5 Case 5 (V-shape bow : 90 deg : At web : Above deck of struck ship)
Table B - 5 Evaluation of energy distribution using different rupture strains – Case 5
PARTS LS-DYNA
SHARP* SHARP
Ec = 0.2 Ec = 0.5
E (MJ) % E (MJ) % E (MJ) %
Total Energy 56.4 49.5 60.1
Side Shell 13.5 24% 1.86 4% 3.11 5%
Inner Shell 5.5 10% 0.72 1% 1.05 2%
Web Frame 6.9 12% 9.04 18% 6.8 11%
Double Bottom 0.3 1% 1.38 3% 0.62 1%
Bottom 0.15 0% 0 0% 0 0%
Weather Deck 16.2 29% 25.9 52% 15.4 26%
Stiffeners 3 5% 10.7 22% 33 55%
Others 10.85 19% -0.1 0% 0.12 0%
Penetration (m) 3 7.05 4.44
SHARP SHARP SHARP SHARP
Ec = 1 Ec = 2 Ec = 3 Ec = 5
E (MJ) % E (MJ) % E (MJ) % E (MJ) %
59.2 59.7 60.2 60.9
4.8 8% 8.22 14% 12.5 21% 20.7 34%
1.04 2% 1.95 3% 3.42 6% 2.43 4%
6.74 11% 6.52 11% 6.65 11% 6.49 11%
0.51 1% 0.41 1% 0.4 1% 0.29 0%
0 0% 0 0% 0 0% 0 0%
11.5 19% 10.7 18% 9.94 17% 10.9 18%
34.6 58% 32 54% 27.3 45% 20.1 33%
0.01 0% -0.1 0% -0.01 0% -0.01 0%
3.75 3.6 3.43 3.14
*Results taken at 1 sec of collision simulation time.
Ec : Rupture Strain
E : Deformation energy [MJ]