FI

u
FINITE ELEMENT ANALYSIS OF SHIP-ICE COLLISION

USING LS-DYNA

By

© Rui Zo ng

A the sis s ubm itte d to th e

Sc hoo l of Gra dua te St udies

in parti al fulfi llme nt of the req uirem ent s for the degr ee o f

Maste r of Enginee ring

Fac ulty of Engineering and App lie d Sc ie nce

Me moria l Universi ty of New fou nd land

July 20 12

St. John ' s Newfo und land

 ABSTR A CT

The energy indu stry ' s increasin g interest in the Arctic reg ion dem and s more and stro nge r

polar ships. lACS has released a se t of docum ents titl ed Unifie d Requir em ents for Polar

Ships (U R[) to harmoni ze diffe rent icc classifi catio n specificatio ns. This thesis defin es a

procedu re fo r eva luating an " lACS Polar Class" ship und er ice imp act s usin g LS-DY NA.

an explicit finit e clem ent ana lysis tool. The final produ ct include s a numer ical model that

is ca pable of eva luating the globa l motion s of the ship and icc. the ship-icc cont act for ce.

and the local structural respon se o f the ship . A fe w icc mat eri al mod els who se pressure -

area rela tions hips comply with the URI arc propo sed as well. Restorin g force s arc

mod eled usin g user-d efin ed-cur ve-functions. Thi s innovat ive approach sig ni fi cantly

reduc es the comput at ion cos t by excluding the water dom ain from the ana lys is. The

Arbi trary Lagran gian- Eulerian meth od in LS- DYNA is d iscussed and empl oyed to

estimate necessary inputs for the user-defined-cu rve-fu nct ion s, Severa l ship-icc impac t

sce narios are model ed in LS- DY A and cont act forces arc co mpa red with the es timatio ns

by DDeP S. a simple ana lytica l so lutio n that is co nsis tent with the URI. [n the last part of

this thesis. the ship from the previou s analysis is icc-stren gth ened with interna l structures

in acco rda nce with the URI and the DNV speci fic ations. Local structural respon se of th is

ship und er ice imp act s is assesse d .

 ACKNOWL EDGEMENTS

To beg in with thi s thesis, I wo uld like to ex press my since res t appre ciatio n to my

supervisor Dr. C laude Dal ey for his ex ce llent suppo rt, dir ecti on , and guida nce. I a m a lso

than kful to STE PS2 and NSE RC C REATE program fo r the ir ge ne ro us financi al suppo rt.

T his thesis wo uld be imp ossible w ithout the he lp a nd suppo rt fro m a fe w grea t peopl e. I

wo uld a lso like to sta te my grea t gra titude to the foll owin g :

Dr. Wei Q iu, professo r at MUN, for brin ging me to thi s o utsta nd ing un ivers ity.

Dr. Bruce Co lbo urne, pro fessor at MUN, for all the ins ightful a nd inspi ring di scu ssion s

thro ug ho ut m y resea rch .

Mr. Bruce Qui nto n, a fell o w gra d uate stude nt at my o ffice , fo r yo ur wond erful help a nd

pati en ce.

Dr. .Junyon g Wa ng and Dr. Rob er Gag no n at N RC -lOT , fo r yo ur rema rka ble help .

Las tly, I wo uld like to th ank my fami ly and my g irl friend fo r yo ur uncondit ion al support.

T his thesi s is dedi cated to yo u.

iii
 Table of Contents

A BST RACT . .......... ii

AC KNO W LE DGE ME TS .... ........... iii

Ta ble o f Co nte nts ...... iv

List o f T ables ............................ ...................................................... xi

List of Figures ................... xiv

List o f Nomencla ture or A bbrevi atio ns xv iii

List o f A ppe nd ices ....... ..

C hapte r 1 Introdu cti on I

1. 1 Sco pe and O bjec tive s

1.2 T hes is Organizati on ..

1.3 LS- Dy NA .. ...5

1.3.1 Ge neral Inform ati on .. .....5

1.3.2 Ti me Ste p Contro l a nd To ta l T ime Cos t ...7

1.3.3 T he A rbitra ry Lagran gian- Eulerian (A LE) Me tho d..... .........9

1.3.4 Co ntac t Mod el .......... . 12

1.3.5 User Defin ed C urve Function . 15

1.3.6 Present ati on of N ume rica l Mod el s .. .. 16

C hapter 2 Litera ture Review .. .. 17

iv
 2.1. Unifi ed lACS Polar Rules .......................................... 17

2.1.1 Origin of the lACS Polar Rules ... ...... 17

2.1.2 Load Design Scenario ... ........20

2.2 Ship Icc Contact and Pressure-Area Curves .. ........23

2.2.1 The Ship-Icc Contact Mechanism.. . 23

2.2.2 Spatial Pressure Distribution 24

2.2.3 Process Distribution ..... .. .......26

2.2.4 Spatial vs. Process ... .... .......27

2.3 Studies using f inite Element Analysis Programs 29

2.3.1 Studies Using DYNA.... .............................................................29

2.3.2 Studies using other FEA Programs .......... .....31

2.4 Summary of Literature Review .........32

Chapter 3 Ice Material Model......... ........ 35

3.1 Simulation Setup ....... ...... 35

3.1.1 Geometric Model .. . 35

3.1.2 Material Models . .. 38

3.1.3 Element Choices .40

3.1.4 Boundary and Initial Conditions... . .40

3.1.5 Other Inputs .......4 1

 3. 1.6 Me sh Co nverge nce Study ... .. .41

3.1.7 Nom inal Co ntact Area .. . 42

3.2 Ice Material Models Based on the Crus hable Foam Material... .45

3.2. 1 Gag non's Crus ha ble Foa m Ice Model.. . .45

3.2.2 Ice Model A .. . 47

3.2.3 Ice Model B... . .49

3.2.4 Ice Model C ... . 51

3.3 Ice Mater ial Model s Based on the Elas tic-Plastic Materi al ... ..... .......53

3.3. 1 Ice Mod el D .. . 53

3.3.2 Icc Mo dcI E ..... . 54

3.4 Summa ry ... . 56

Chapter 4 A LE Meth od .. . 59

4.1 Simulations for Evaluating the Co mputation Cos t ..............................59

4.1.1 Geo me tric Model ........59

4. 1.2 Materi al Mod els ... . 62

4.1 .3 Element C ho ices .. . 63

4.1.4 Bound ary and Initial Co nditions ... ....64

4. 1.5 Other Input s .. . 66

4.1.6 Mes h Co nve rgence ... .. 66

vi
 4 .1 .7 Eva luatio n of the Co mputatio n Cost ., . . 69

4.2 Oscill ator y Ana lysis vs. Tran sien t Ana lysis . . 70

4.2 . 1 Geomet ric Model . . 71

4. 2.2 Material Mode ls ........ ........ .. 73

4 .2.3 Element C ho ices ............. 73

4 .2.4 Bound ar y and Initial Co nditions 73

4.2 .5 Loadi ng Co nd itio ns . . 75

4 .2 .6 Added Mass usin g the Os ci llato ry A na lys is 75

4 .2.7 Ad de d Mass usi ng the Tra nsie nt A na lysis 76

4 .2.8 Co mparison . . 79

4 .3 Estimation or Ad de d Mass and Damp ing Coefficient .... . 80

4 .3.1 Geometric Mode l . . 80

4.3 .2 Material Model s . . 82

4 .3.3 Element Cho ices . . 82

4 .3 .4 Bound a ry and Initial Co nd it ions . . 82

4 .3.5 Loadi ng Co nd itions 83

4 .3.6 Ship's Adde d Mass and Dampin g Ter ms ...... ........ ....... ............84

4 .3 .7 Ice Adde d Mass a nd Damp ing Te rms 88

4 .3 .8 Compariso n 90

vii
 4.4 Summa ry ... .. 91

Chapter 5 Ship lee Co llisio n Force ... .. 94

5. 1 Defin ed- Cu rve- Functi on s .. . 94

5.1. 1 Restorin g Forces .. .. 94

5. 1.21 mpleme ntation ... . .. 96

5. 1.3 Drag and Added Mass . .. 100

5.2 Mass Reducti on Coe ffic ient.. .. .. 101

5.2.1 Sim ulation Se tup ... . 102

5.2 .2 Result s ............................................. 104

5.3 Simulation Se tup ... . 107

5.3.1 Geo metric Mod el .. .. 107

5.3.2 Material Mod els ... .. 107

5.3 .3 Element Choices . . 107

5.3.4 Bound ary and initial co nditions ... .. 107

5.3.5 Loading Co nditio ns and Dampin g ... . 108

5.3.6 Mesh Co nve rge nce .. . 108

5.4 Ship-lee Co ntact Force .. . 111

5.4 .1 " Dry" Co llis ion Cases 111

5.4 .2 "W et" Co llision Cases .. . 113

viii
 5.4.3 "Dry't vs. ""WeC . ................ 115

5.5 Summary . . 119

Chapter 6 Ships Structural Response . . 121

6.1 Ship Structural Design. . . ............... 121

6.1.1 Main Frames .... .............. 122

6.1.2 Load Carrying Stringers . . 124

6.1.3 Deep Web Frames ..... 125

6.1.4 Bulkheads . .. ...................... 126

6.1.5 Summary . . 127

6.2 Simulation Setup . . 129

6.2.1 Geometric Model . ... 129

6.2.2 Material Model.... ......... 130

6.2.3 Element Choices .... 131

6.2.4 Boundary and initial conditions ... . 132

6.2.5 Loading Conditions and Damping ..................................... 132

6.3 Ship Structural Response 133

6.3.1 Contact Force and Pressure ... . 133

6.3.2 Von Mises Stress . .. ..... ................. 135

6.3.3 Pressure-Deflection Curve 138

ix
 6.4 Summary . ...... 140

Chapter 7 Conclusions and Recommendations . ..... 142

7.1 Conclusions . .................. 142

7.2 Recommendations... .......... 143

Chapter 8 Bibliography . .... 148

Appendices 156

Appendix A: STePS2 Cluster Specifications ............................ 157

Appendix B: DYNA's Keyword File of the Final Model 159

 List of Table s

Table 2- 1: Polar Classes in lACS Unifi ed Polar Rules.... .20

Tab le 2-2: Ice Strength Te rms in the URI

Tab le 3- 1: Geo metry of the Ship and Ice .. .. 38

Ta ble 3-2: Material Properties o f the Ship Model ..38

Table 3-3: Material Properties of the Ice Model. . ..39

Table 3-4: Element Cho ices for Ship and Ice . ....40

Ta ble 3-5: Surge Distan ce x and Nomin al Contact Area Anominal . ....44

Ta ble 3-6: Materi al Properties of Gagnon' s Icc Model 46

Ta ble 3-7: Stress -Strain Relationship in Gagnon ' s Ice Model .46

Table 3-8 : Material Properties of Ice Model A .48

Ta ble 3-9: Stress - Volumetric Strain Relationshi p in Icc Model A .48

Ta ble 3- 10: Stress - Vo lumetric Strain Relationship in Ice Model 13 .50

Tab le 3-1 I : Strcss - Vol umetric Stra in Relationship in Icc Mode l C .......5 1

Tab le 3- 12: Material Propert ies of Icc Mode l D . .......53

Ta ble 3- 13: Materia l Properties of Ice Model E . .55

Table 3- 14: Stress - Vo lumetric Stra in Relationship in Icc Model E 55

Ta ble 3- 15: Summary of Proposed Icc Mate rial Model s 56

Ta ble 4-1 : Element Choices for the AL E Simulations .........63

Tabl e 4-2: Summ ary of the Mesh Convergence Study 68

Table 4-3: Dimensions of the ALE Domain .. .. 72

Ta ble 4-4 : Influence o f the Magnitud e of the Force on the Hea ve Added Mass 77

xi
Table 4-5: Heave Added Mass Coefficients at Very-Low Frequencies and Very-High

Frequencies on the Unit Hemisphere . . 79

Table 4-6: Boundary Conditions on the Ship and Ice .. . 83

Table 4-7: Values of Applied Loads ..........84

Table 4-8: Added Mass and Damping Terms of the Ship 87

Table 4-9: Added Mass and Damping Terms of the Icc . 90

Table 4- 10: Added Mass Coefficients Calculated by DDePS (1967) and Present Work ..91

Table 5- 1: Load Definition for Restoring Forces.... . 99

Table 5-2: Radii of Gyration Estimated by DYNA and DDePS 105

Table 5-3: Mass Reduction Coefficients Co . . 106

Table 5-4: Load Definitions on the Ship and Icc . . 108

Table 5-5: Computation Time of Simulations using Various Element Sizes 109

Table 5-6: Maximum Contact Force (Finite Ice. Dry Collision) . . 112

Table 5-7: Maximum Contact Force (Infinit e Icc. Dry Collision) . . 112

Table 5-8: Maximum Contact Force (Finite Ice. Wet Collision) 114

Table 5-9: Maximum Contact Force (Infi nite Icc. Wet Collision) .. . 114

Table 5-10: Dry vs. Wet - Maximum Contact Force. Finite Icc 116

Table 5- 11: Dry vs. Wet - Maximum Contact Force. Infinite Icc.. . 116

Table 6-1: Hull Area Extents (lACS 2010) .......... 122

Table 6-2: Scantlings of Structural Members in the Bow Region ... .... 128

Table 6-3: Material Parameters for the Non-Rigid Part of the Ship 130

Table 6-4: Material Parameters of the Rigid Part of the Icc ..... . 130

xii
Table 6-5: Element Choices .. .. 132

Table 6-6: Load Definition on the Ship ... .. 133

xiii
 List of Figures

Figure I-I : Co mpar iso n of Lagrangian , Eulerian and A LE (LST c' 20 I0) 10

f igure 1-2: Co upling in the A LE meth od ... .... ...... .... 12

Figure 1-3: Co mpariso n o f SOFT Options .. . 14

Figure 1-4: Co ntact Mod el with SO FT = 2 ... . 15

Figure 2- 1: Ma p of the Ar ctic Ice-Covered Water Defin ed by IMO 19

Fig ure 2-2: Sketch of Ice Co ntact with a Str ucture (Daley 2004) 24

f igure 2-3 : Spat ial Pressur e-Area Relat ion ship (Da ley 2004). .. ...25

Figure 2-4: No minal, tru e and measured areas (Daley 200 4) 25

f igure 2-5: Process Pressure-Ar ea Relat ion ship (Da ley 2004) .. . 26

Figure 2-6: Link between Proc ess and Spat ial Distr ibuti on s (Daley 2004) 28

f igure 3- 1: Geo me tric Mod els o f the Ship and Ice in Rhin oceros® 36

Figure 3-2 : the Ice Block with Round ed Edges ... . 36

fig ure 3-3 : Ship Bow ... ....37

Figure 3-4 : Se pa ration betw een the Ship and Ice (To p View) 37

Fig ure 3-5 : Stress - Volum etric Strai n Curve o f the Icc Model in Co nve rgence Study ..39

Fig ure 3-6: Bound ary Co ndition on the Infinit e Ice .... . .41

Figure 3-7 : Mesh Co nve rge nce . . . . ........ .42

f igure 3-8: Intersect ion of the Ship and Ice .......... .......... .............43

f igure 3-9: Stress -Vo lume tric Stra in Relation sh ip in Gag no n's Ice Model .46

Figure 3- 10: Process Pressur e-A rea Curve of Gag no n's Crusa ble Foa m Ice Mod el .4 7

f igure 3- 11: Stress - Volum etric Strai n Curve in Ice Mod el A .48

xiv
Figure 3- 12: Pressure - Area Curve of Ice Model A ... . .49

Figure 3- 13: Stress - Volumetric Strain Curve in Ice Model B .. . 50

Figure 3- 14: Pressure - Area Curve of Ice Model B 51

Figure 3- 15: Stress - Volumetric Strain Curve of Ice Model C .. . 52

Figure 3- 16: Pressure - Area Curve of Ice Model C ... . 52

Figure 3- 17: Pressure - Area Curve of Ice Model D.. . . 54

Figure 3- 18: Pressure - Area Curve of Ice Model E... . 55

Figure 3- 19: Pressure - Area Curve of Ice Model C (All Data Included)... . 58

Figure 4- 1: Top View of the Geometric Model in Rhinoceros®... . 61

Figure 4-2: 3D Model in DYNA... . 62

Figure 4-3: the ALE Domain including Ambient Layers ... . 65

Figure 4-4: Convergence of the Ice Surge Motion .. . 67

Figure 4-5: Convergence of Sway Motion of Ice .. . 67

Figure 4-6: Convergence of the Icc Heave Motion . .....................68

Figure 4-7: 3D Model in DYNA... . 72

Figure 4-8: Prescribed Heave Motion of the Sphere in the Oscillatory Analysis 74

Figure 4-9: Heave Motion of the Semi-Submerged Sphere.... . . 76

Figure 4- 10: Time History of the Semi-Submerged Sphere' s Heave motion 78

Figure 4- 11: Influence of the Magnitude of the Force on the Heave Added Mass 78

Figure 4-12: 3D Model for Estimating the Added Mass Coefficients on the Ship... . 81

Figure 4-13: 3D Model for Estimating the Added Mass Coefficient on the Ice 81

Figure 4-14: Time History of the Ship Heave Motion ... . 85

Figure 4- 15: ri me History of the Ship Roll Motion .. ......86

Figure 4- 16: rime History of the Ship Pitch Motion . . ............................87

Figure 4- 17: Time History of the lee Heave Motion .. . ................88

Figure 4- 18: rime History of the Ice Roll Motion... . 89

Figure 4-19: rime History of the Ice Pitch Motion .. . 89

Figure 5- 1: Creating a Rigid Part . . 98

Figure 5-2: Creating the Local Coordinate System .. . 98

Figure 5-3: Normal Direction of the Contact Surface... . 103

Figure 5-4: Contact Location on the Ice .. . . 103

Figure 5-5: Mesh Convergence - Ship Glancing with Finite Ice 110

Figure 5-6: Mesh Convergence - Ship Glancing with Infinite Ice 110

Figure 5-7: Comparison of the Contact Force - Dry vs. Wet (Finite Ice) . 118

Figure 5-8: Comparison of the Contact Force - Dry vs Wet (Infinite lee)... . 118

Figure 6- 1: Hull Area Extents (lACS 2010) ... . . 123

Figure 6-2: Main Frames and the Hull... .... ................ 124

Figure 6-3: Load Carrying Stringers and the Hull 125

Figure 6-4: Deep Web Frames and the Hull .. .......... 126

Figure 6-5: Bulkheads including Stiffeners. and the Hull .. . 127

Figure 6-6: Bow Region with Internal Structural Members 129

Figure 6-7: Rigid and Non-Rigid Ship and Ice ... . 131

Figure 6-8: Comparison of the Contact Force: Rigid Ship vs Non-Rigid Ship 134

Figure 6-9: Time History of the Average Pressure ... . 134

xvi
Figure 6- 10: A Typical Von Mises Stress Distribution on the Main Frame ........ 136

Figure 6- 11: A Typical Von Mises Stress Distribution on the Deep Web Frame 136

Figure 6- 12: Time History of the Effective Stress . . 137

Figure 6- 13: Time History of the Effective Plastic Strain 137

Figure 6- 14 : Pressure - Deflection Curve of the Main Frame Member .. . 139

xvii
 List of Nom enclature or Abbrev iations

ID One Dimensional

2D Two Dimensional

3D Three Dimensional

ABS American Bureau of Shipping classification socie ty

ALE the Arbitrary- Lagrangian-Eulerian method

CAC Canadian arctic class

CA D Computer-aided design

Card A unit where users give inputs into DYNA

CFD Computational fluid dynamics

CG Center of grav ity

Computation cost/ time Length of time required for the solver to run the numeric al

model

Computer Cluster A collection of networked computers that function as a single

computer

DDcPS The so ftware Direct Design for Polar Ships

DOF Degree of' frccdom

DNV DetNo rskeVeritas class iticationsoceity

DYNA Refer s to either LS-DYNA or MPI'-DY NA interchangeably

FE Finite element

FEA Finite element analysis

GL Germanischer Lloyd classi fication soc iety

xviii
lACS International Association of Classification Societies

IMO International Maritime Organization

LSTC Livermore Software Technology Corporation

LR Lloyds Register class ifica tion society

Part A collection of elements in DYNA with similar properties

Rhinoceros® A CAD software developed by McNeel North America

Simulation time Length of time explicitly simulated within the numerical model

URI Unified Requirements for Polar Ships published by lACS

xix
 List of Appendices

Appendix A: STePS2 Cluster Spe ficicatio ns . ..... 157

Ap pendix B: DYNA ' s Keyword File of the Fina l Mo de l ......... 159

 Chapter I Introduction

The Arctic region is believed to hou se one o f the world's largest oil and gas resourc es. A

United States Geo logica l Survey estima tes that 530 billi on barrels of potential pet roleum

arc located be neat h th is area . The ice- infested sea wa ter and othe r harsh env ironme nta l

co nditions have been cha llenging the indu str y eve r since the first operat ion in the Arctic.

However, the increas ing dem and from the globa l econo my, is drivin g the oil and gas

industry to be more and more active in the A rctic region .

Ships opera ted in the Arc tic area can be d ivided into two main catego ries: icc-br eakin g

vesse ls and ice-stren gth ened ships. Ice-br eakin g vess els arc used to suppor t ot her

operat ing unit s and ac tivities . The ir stro ng hull structures ena ble them to take on heavy

tasks such as icc breakin g, maneu verin g in icc and icc mana gem ent. Icc-stren gth ened

ships, whose hull s are rela tive ly weaker than ice-b reakers, arc designed to w ithstand

possibl e ex pos ure to a cer tain level o f icc load . depend ing on their icc class. They have

lim ited ability in breakin g ice and man euver ing in icc cove red water. Co m mo n ice-

stre ngthe ned ships in the Arctic are vesse ls such as cargo ships, tank er s. and supply ships.

Historic all y. ice cla ssifi cation s gove rni ng polar ships arc reg ulated by va rious

classi ficatio n socie ties. In 2006 . the Intern ational Assoc iatio n of C lass iiicatio n Socie ties

(l ACS) released a se t of doc ume nts titled Uni fie d Requir em ent s for Polar Ships (U RI) to

harm on ize d iffer ent ice classification speci fica tions. More icc- strengthen ed ships

complying with the URI are ex pecte d in the near futu re.
Extensive st udies co nce rning icc-b reakin g vesse ls have been carr ied out to und erstand the

mechanism of the hull breakin g ice and the physics of the broken icc acting aga inst the

hul l. Research and ex perience on the icc-stren gth ened vesse ls arc relative ly lim ited . The

presen t thesi s is primarily co nce rned with ice-stren gthened ships under the new URI. It

prese nts a study using the state-o f-art finite eleme nt ana lysis (F EA) program LS-DYNA

to inve stigate the globa l mot ion and local structural res ponse of an icc-stren gthened ship

under icc impac t scenarios.

1.1 Scope and Obj ective s

This thesis deta ils a proced ure for analyzing ship-ice co llisio ns using the com mercia l

FEA progra m LS-DYNA. The fi na l product is a FEA model ing templ ate to eva luate the

globa l motion. and the globa l and local str uctura l respo nse of an icc strengthene d ship

under var ious ice imp act scenarios. This study is co mposed of four subto pics :

• Deve lop an ice materia l model whose pressu re-area relatio nship co mp lies

wit h the UR I.

• Estimate the added mass and da mping coe ffi cients of the ship and icc

using the Ar bitrary -Lagra ngian-Euler ian (AL E) method.

• Mode l var ious ship -icc co llisio n sce nar ios and co mpare the result s wit h

ca lculations using the Popov model that is co nsis tent with the URI.

• Combine resu lts from previou s subtop ics to genera te a so lution for

eva luating a ship's struct ura l respo nse under icc impacts fo r an ice-

strengthened ship .
1.2 The sis Organ ization

This thesis co nta ins six chap ters. This chapter presents the bac kgro und. objec tives and

outline of thi s thesis. and introdu ces reade rs to the co mme rcial finite ele me nt ana lysis

progra m LS-DYNA . Its detailed theory manu al (Hallquist. 2006) and user manual (LSTC.

2007a. 200 7b) are avai lable online. However . back ground knowled ge of co m putatio n

time cost. the ALE meth od . and impl em ent ation of the user-defin ed -cur ve- funct ion s are

briefl y present ed here to support discussion s in later chapters. The use of user-d e f ned-

curve - functions is an innovat ive approac h for thi s application developed in thi s thesis to

simulate the wa ter dom ain where the ship-ice co llision take s place .

Chapter 2 is the litera ture review. Ge neral inform ati on on previou s wo rk on the URI

incl uding a short intro ductio n to design sce narios is present ed first. foll owed by the

deve lop me nt of basic knowledge of the mechani sm s o f ship/str ucture- ice interac tion. the

ice pressure-area relati on shi ps. and a discu ssion of exis ting studies on ship-ice co llision

usin g FEA program s incl uding LS- DYNA. A sum mary of the literatu re review explains

the motivation and methodology for this thesis.

Eac h o f Chapter 3 to Chapter 6 addresses one of the subtopics listed in the previou s

sec tion. Chapter 3 fo cu ses on developing a prop er ice material mod el that fits the purpose

o f this thesis. Ice mater ial prop ertie s and its failure mechani cs are the mo st important

facto rs in determinin g the ship-ice contact forc e. The pressur e-area relati on ship is the

most direc t indi cati on of ice strength. The pressu re-area curve specifie d in the URI is
considered as the benchmark. Various ice material models are evaluated by simulating a

simple ship-ice glancing impact scenario. One ice material model is chosen based on

closest compliance with the URI.

Chapter 4 explores the possibility of implementing the ALE method. ALE simulations in

LS-DYNA have been successfully implemented to simulate the lluid domain in many

studies on ship-ice collision. So naturally it is selected as a tool for this thesis. However.

existing studies using ALE are all concerned with the global motion of floating bodies

and the global contact forces. This thesis aims at evaluating the ship-ice collision in both

the global and local contexts. A discussion in this chapter will show that the ALE method

is not an efficient approach due to the high computation cost. An alternative solution

featuring user-defin ed-curve-functions is then proposed and discussed in Chapter 5.

Rather than simulating the whole lluid domain. the ALE method is employed to estimate

added mass and damping coeffic ients which can be input into user-defined-curve-

functions. Simulations of transient and oscillatory analyses are conducted to estimate

those coefficie nts and results are compared with literature.

Chapter 5 explains modeling the global contact force of a ship-ice collision. The ship is

simplified as rigid and the ice is modeled using the material model developed in Chapter

3. Hydrodynamic forces are modeled using user-defined-curve-functions without actually

simulating water. Simulations of various ship-ice collision scenarios are performed and
res ults are co mpare d to ca lculat ions usin g the Pop ov model which is co ns istent with the

URI.

In Chap ter 6. the ship used in previou s sections is ice stre ngt he ned in accor dance with the

URI. Th is structured and defo rma ble ship is then put in the co llis ion models developed in

Chapter 5 in lieu o f the rigid one. Th e ship's globa l motion . and its globa l and local

structura l respon ses und er ice impacts are ana lyze d. The final FEA model ca n be used as

a temp late for ana lyzi ng other ship-ice co llision probl em s.

Cha pter 7 conc ludes the co mplete study and reco mme nds future wor k.

1.3 LS-DVNA

The co mmerc ial finit e eleme nt program LS-D YNA is the prim ary num erical tool for th is

research . This sec tion introdu ces readers to its genera l characte ristics, as we ll as so me of

its backgrou nd theori es that are releva nt to this thesis.

1.3.1 General Information

LS-DYNA is a genera l-purpose finit e eleme nt progra m develop ed by the Live rmo re

So ftware Tec hno logy Co rpora tion (LSTC) and wide ly used by the automo bile.

co nstr uct ion, m ilitary, aeros pace. manu factur ing. and bioe ngineering industr ies. Its core -

co mpe tency is highl y nonlin ear transient dynami c finit e eleme nt ana lys is using ex plicit

tim e integration. "Tra nsient dy na mic" impl ies the ana lysis of high- speed . short-d uratio n
eve nts where inerti al forces do mi nate. Ship-icc co llisio n co uld be a typica l tran sient

dynamic prob lem. " Ex plicit" means so lving equat ions that invo lve time and tim e-

dependent variab les (ve locity. acceleration. and inerti al. etc.) to accu rately ca pture the

dy nam ic effects. A " nonlinear" pro ble m is gene rally characte rized by at least one of the

fol low ing co mplications:

• Boundary nonlin ea rity --- Co ntact between part s o r objec ts chan ges ove r

time or restraint s on parts are time depend en t.

• Geo met rica l nonline arity --- Large deform at ion s occur. thu s requirin g new

equilibrium equations based on the deforming geometry.

• Material nonlin earity --- Materi als do not ex hibit ideall y elastic behavior

and th is leads to changes in the stress- strai n rela tio nship.

Obvio usly. a ship -icc co llisio n problem fits in all three cri teria of non linearity. This

makes LS-DY A the best avail able tool lor th is researc h. The detaile d theory manu al

(Hallquist. 2006) of LS-DYNA is ava ilable on LST C"s we bs ite. So me importan t theor ies

relate d to this thesis will be present ed in the chap ter.

T his thesis utilizes two versions o f LS-D YNA. The first one runs on one or more parallel

processors in a sing le comput er. This version is used mainly to run sma ll and simple

simulatio ns. Another vers ion is MPP-DYNA. which run s on a computer clu ster that

wo rks like a super comput er by co nnecting a gro up of indep end ent computers. The clu ster

used in th is thesis has 128 co res and is ve ry pow erful in so lving large model s that co ntain
elabora te geo me try. very relin ed mesh. co mplex materia l models. lon ger simulation time.

com plicated bo undary co nd itions or combinations thereo f. This efficie ncy is achieve d via

model decomp osition that dissects the who le mode l into parts. There are three

dceom posin g meth ods (LSTC 2007a): the automa tic Recur sive Coor dinate Bisect ion

(RC B) meth od, the simple heuri stic meth od (G REE DY). and thc manu al met hod. In

almos t all cases, the RCB is the superior meth od for its rob ustness .

M PP-DY NA is the too l for most of the simulatio ns presented in this thesis. Since LS-

DYNA and M PP-DYNA ess entially share the same theo ries and code s, they wi ll be bot h

refe rred as LS-DY NA from Chapter 4 onwa rds unless otherwise spec ified .

1.3.2 Time Step Control a nd Total Time Cost

The goa l of th is thesis is to produ ce a pract ical solution fo r rcal world ship-ice co llision

prob lems. As part o f this. co mp utation cos t mu st be taken into con siderati on. Du ring the

so lution. LS-DY NA loop s thro ugh all the possible e leme nts to update the stress and the

righ t hand side force vec tor. The new tim e step is determ ined by thc minimum va lue of

all the critical time steps ove r all clements. Ge nera lly spea king. the ship is ana lyze d usin g

shell element s wh ile icc. water. and air are mode led using so lid cleme nts.

For she ll elements. the cr itica l time step can be co mputed from:

Mc = ~ Eq ua tio n I >t

where Ls is the charac teristic length of a she ll e leme nt and c is the speed o f so und:
 Eq ua tio n 1-2

where E is the Young's modulu s, p is the material densit y and v is the Poisson ' s ratio.

The de fault equation fo r ca lculating Ls is:

Equation l-d

where L, is the length of side i of the clem ent , f3 eq ua ls I for triangle and 0 for

qu adrilateral elemen ts, and As is the sur face area of the eleme nt.

The crit ica l tim e step for so lid elements is co mp uted in a similar mann er :

Eq ua tio n I-u

where Le is charac teristic length .Q is a function of bulk viscos ity and c is the adia batic

speed of so und . Equatio ns for ca lculating Q and c arc very co mplicated and unn ecessar y

to be present ed here.

As show n in the equations above, ele me nt sizes and material prop erti es togeth er

de ter mine the critical time step. Note that in LS- DYNA. rigid cle men ts arc not co nside red

in the co mputation o f tim e step . Users sho uld defin e a proper tim e step va lue when the

model only co ntains rigid cleme nts.

Besides the critica l tim e step, the total co mputatio n cos t also de pends on the num ber of

cle ments. boun dary con ditions, and the ana lys is meth od . More DO F. more co mp licated
loadin g co nditions, and the ALE analysis genera lly requir e longer co mputation time. This

is a major co ns ide ration in thi s thesis and is further discussed in Chapter 4.

1.3.3 The Arbitrary Lagrangian-Eulerian (ALE) Method

The ALE meth od is cu rrentl y the only method for simulating wat er in DYNA. It has been

used in severa l studies . Its full detailed theory can be foun d in DYNA ' s theory manu al

(Ha llquist, 20 06). This sec tion only introdu ces readers to the basic knowled ge of the ALE

meth od. Impl em ent ati on is d iscu ssed in Chapter 4.

Figure I- I illu strates the differ ence of the Lagran gian . Eulerian. and the ALE meth od in

ana lyz ing a soli d piece o f mat erial (red) moving and deformin g. In the Lagran gian

simulatio n, the mesh de form s with the material. In the Eulerian so lution. the material

flow s throu gh the fixed mesh. The ALE meth od is a co mbination of these two . The mesh

is attac hed to the material (Lagra ng ian) and passes thr ough the fixed back ground

refer ence mesh (E ulerian). In other word s. the material deform s in a Lagra ngia n

formul ation at the first step . The seco nd step is the advec tio n. which mean s that clem ent

sta te va riables in the de forme d cleme nts (re d ones in Figure I-I ) are rema pped back onto

the Eulerian re fere nce mesh .

 Figurc I-I : Compa r ison of Lagrangian, Euler ia n and AL E (LST C, 2010)

Fluid -str uct ure int e racti on ana lys is usin g the ALE meth od requi res th ree addi tiona l

co m putations besid es the Lag ra ngia n ste p. The first one is the advec tio n men tion ed

ea rlier. It co ntro ls th e flowin g or flu xing of materi al s in the tot al ALE dom a in. T he

seco nd ca lculation is inter face recon st ruc tion which defin es multi-mat eri al co -exis te nce

in o ne e leme nt. T he last o ne is the co upling between Lagran gian c leme nts and ALE part s

(fluid-s truct ure in thi s thesis). Se tup o f advec tio n and inter fac e reco ns truct io n is very

sta ndar d and stra ightfor war d in the ALE s im ulations, wh ile co upling req uires user' s

defin ed input s. No te that the cl em ent size of Lag ra ng ian part s sho uld be sim ilar to th at of

the AL E parts for the ALE a lgo rithm to functi on acc ura te ly.

T he co upling ca lculatio n in the ALE meth od is pen alt y based and is dem on strat ed in

Figure 1-2. In the left part o f Fig ure 1-2, there is no coupling force since the she ll

str ucture (g ree n) is not in co ntact w ith the wa ter (E ulerian materi al in red ). O nce

pen et rati on occ urs, it is measur ed to co mpute the cou plin g force as a spring sys te m . T he

10
spring stiffness depends on the material properties of all bodies involved. The penalty

factor named PFAC. a scale factor of scaling the estimated stiffness is required for

calculating the coupling force. This PFAe. whose default value is 0.1. is recommended to

be redefined by the user. Its value can be either a constant or a function of penetrating

depth. Note that its value is different in each analysis. Even in the same analysis. if the

element size or the geometric model is modified. its value needs to be re-calibrated. Prior

to conducting a detailed ALE simulation. several experimental simulations are generally

needed to determine a proper value. In each ALE simulation presented in this thesis. the

PFAC is set to a value so that the floating body's neutral buoyancy in the simulation is

the same as that determined by a simple hydrostatic calculation based on its geometry.

However, the floating body still oscillates around the neutral position with very small

amplitude. It is almost impossible to determine the optimal PFAC value to completely

eliminate this small oscillation. Many hours were spent on calibrating the PFAC value

during this research to minimize the noise it may introduce to the solution.

11
 F pena'ty - k spring .dx" .
Moving Lagrangian "/ Penetration
shell structure Spring
stretch

ALE material interface

(fromIR) (note: MMG =AMMG)
Figurc 1-2: Coupling in thcALE met hod

1.3.4 Contact Mode l

T here arc two types o f co ntac t algorithms in LS-DY NA. The first one is "on e-w ay

co ntac t" . It only checks the user-specified slave nodes for pen etration o f the master

seg me nts. It the n tran sfers compr ession load s betw een the slave nod es and the ma ster

seg me nts. When co ntac t frict ion is acti ve. tang enti al load s are a lso tran smitt ed ifrelativ e

sliding happ en s. A Co ulomb fricti on formul ation is used with an ex ponential interpo lation

function to tran siti on from static to dynami c friction. Thi s tran siti on requir es a dec ay

coe fficie nt. It only work s wh en the static fricti on coeffi cient is larger than the dynami c

friction coe ffic ient. On e-way contacts may be appropriate when the master part is rigid . It

may also be used for def orm abl e bodies when the master part has a coarse mesh and the

12
slave par t has a relative ly fine mesh . Othe r co mmo n applicatio ns are co ntacts of beam- to-

surface or she ll-edge -to-s urface .

Th e othe r type is the " two-way con tac t" . It functions esse ntially in the sa me way as the

"o ne-way contac t". exce pt that the subro uti nes chec king the slaves nod es for pen etrat ion

are ca lled a seco nd time to chec k the master nodes fo r pe netra tion throu gh the slave

seg ments . In ot her wor ds. the treatm ent is symme tric and the defin ition of the slave

surface and master sur face is arbitrary . This meth od res ults in higher co mputa tion cos t

due to the extra subrout ine ca lls.

The auto matic -si ng le-s urface -co ntac t is a "two-wa y co ntact " and is reco mme nded as the

supe rior algorit hm by DY NA. The so ft co nstra int option (SOFT) ca n be ad ded into the

co ntact stiffness ca lculatio n by the user. When SOFT is set at I. the co ntac t algo rithm

adopts the soft co nstrai nt fo rmul ation . It is effec tive for co ntac ts invo lv ing d issim ilar

mesh sizes an d d issim ilar material properties. The pin ball seg me nt based co ntac t is

activa ted by sett ing SOFT at 2. It is the reco mmende d optio n for treatin g co ntac t at shar p

corners . Si mulatio ns of a ship impactin g an ice block with rounded edges we re carrie d out

to exa mi ne thei r di ffe rence. The ship and ice had dram at icall y diffe rent material

prop erties. In eac h simulation. a dif ferent SOFT option is chose n. Result ant co ntac t fo rces

are co mpare d in Fig ure 1-3. Ti me histories o f the co ntac t forces using differ ent SO FT

options are simi lar to eac h other and roughly have the same peak va lue. Meas uri ng the

di stance betwee n the shi p and ice indi cates that the co ntac t should starts at about 1.1

13
seco nds . In all simulations. DY NA detec ts a co ntact be for e the bodie s are actually in

co ntact. Thi s is marked as the ver tical line (purpl e) in the figur e. However. in the case

where SOFT = 2. the cont act occurs mu ch earlier than other cases. Figure 1-4 is a

sna pshot from the simulation where SO FT = 2. The ship is in red and the ice block is in

blue. It shows that the ice (blue) is already deform ed before the geo metries are in contact.

This phenom enon also exi sts in the case wh ere SOFT = 1. but is much less seve re. Th is

"ea rly contact" affect s the analys is o f the nomi na l contact area and hence the ice pre ssure -

area relatio nshi p. It is discu ssed in Chapter 3. The auto matic- sing le-surface -co ntac t is

used fo r all simulations in th is study. T he va lue of SOFT is set at 1 in almos t a ll

simulations.

E 1.4 I - - - - - -I -+ - -- - I - -+ - - - -H - - - - --

~ 1.2 I------I~-_,~-+__-_H,_---­
~ 1.0 I - - - - - - - - - ,... - -- I - - - -- + - -- \ -l - - - - - - - -

~ 0.8 I - - - --+-- J- ----., - - -- + - - -,;\ - - - - - -

Cl 0.6 I - - - , _f---f-- - I - - - - -- - t - - -t - t - - - - -
0.4 I--- - +- --t-I -- - ---+- - -\+-----

0.0 1....-~_~~ l----------'------~ _ ___'_::l2oo._ _ __ _ '

1.0
Tim c (s)

- No SOFT Option

f igllrc l-3 : Co mpa r isono fS OfTO plio ns

Notc: T he vert ical purpl e line mark s t he time instan t when the co ntac t sho uld initiate.

14
 Figu re 1-4: Co ntac t Model with SO FT = 2

1.3.5 User Defined C urv e Function

In DYNA. the *DEF INE _CU RVE _FUNCT ION car d defin es a cu rve wh er e the absc issa

is time. a nd th e ordi nate is ex presse d by a function of other curve defin ition . for ces.

kinem ati cal q uant ities. int rinsic functi on s. interp olatin g pol yn omi al s. and co m binatio ns

ther eo f For instan ce. the di spl acem ent c urve functio n repo rts the di spl acem ent (or dina te)

ove r the time (a bs cissa) . T he n an ex te rna l load ca n be defin ed as the di spla cem ent

multipl ied by a coe ffic ie nt. A fu ll list o f the *DE FINE _CU RVE_ FUNCTION is ava ilable

in DY NA ' s Keyw ord User's Manu a l (2007) . Func tions that give va lues o f z-trans latio na l

di spl acem ent (heave). y-ro tatio na l di sp lacem ent (pi tc h). and x-ro tationa l d ispl acem ent

(ro ll) arc used in thi s thesis. Detail ed im pleme ntation is addressed in C hapter 5.

15
1.3.6 Presentation of Numerical Model s

A numerical model is constr ucted in DYNA by enter ing input s in " cards". eac h of whic h

is for a speci fic purp ose. When a simulation is presented . only imp ortant inputs is se lected

and organized into the foll owin g categor ies :

• Geo me tric model : Thi s ca tegory gives the detailed informa tion on the

dim en sion s o f the geo me tric model. Geo metric models can be generated in

DYNA or other CA D program s. Rhin oceros® (McNeel No rth Amer ica) is

used in this thesis for produ cin g geo metr ic models .

• Material models: This sect ion discusses mater ial types and their

param eters.

• Element cho ices : This category present s choi ces o f eleme nt types (so lids.

she lls. ctc.) as well as eleme nt parame ters such as the she ll th ickn ess.

cleme nt formu lation s. ambie nt types. and integ ratio n algor ithms.

• Bound ary co nditions and initi al conditi ons : In DYN A. the bound ary

co nditions defin e the con fi neme nt on obje cts and their prescribed moti ons.

Th e initial co nditions includ e initi al veloc ities. init ial strains. the init ial

hydr ostatic pressure d istr ibution . and the initial vo lume fracti on . etc.

• Other settings: Th e section covers the load definiti on . the contac t mod el.

dampin g defin ition . user defin ed functi on s. etc .

• Mesh co nverge nce: The app rop riate element size is det ermin ed via the

mesh conv ergence study .

• Resu lts decl aration : This part present s and discusses the results.

16
 Cha pter 2 Liter ature Review

Studies on icebreaking vesse ls arc mostl y conc ern ed with the level icc failure mechani sm.

the globa l icc resistance on the ship. and the movem ent o f broken ice floes aro und the

hul l. These top ics are not cove red in this literature review. This literatur e review

exa m ines the top ics of ship impacting ice floes. bergy bits. and icebergs. The rev iew

focu ses on a few areas: origin and the orie s in the URI. mech ani sm s o f the ship-icc co ntact.

icc pressure- are a relation ship s. and the finite eleme nt ana lysis o f ship-icc interacti on .

Specia l atte ntion is de voted to studies using DY NA since it is the primar y too l for the

pre se nt thes is. Each topi c wi ll be discussed in a separate section follow ed by a brief

summa ry.

2. 1. Unified lACS Polar Rul es

This sec tion presen ts the origin o f the URI and a discu ssion of design sce na rio which is

imp ortant to the finite clem en t modelin g in later cha pters.

2. 1.1 O rigi n of the lACS Polar Rul es

The re arc seve ra l major po lar ship class ifica tio ns developed by various co untries to

protect their arctic waters and interests. They arc:

• Ca nad ian ASPPR/CAC ( 9 C lasses)

• Russian MR S/N SR (9 Classes. 4 Icebreaker)

• Finnish/S wedish (Baltic) (5 C lasses)

• A BS (USCG) (5 Po lar Classes . 5 Balti c Classes)

17
 • DNV (3 Icebr eaker. 3 Polar. 5 Balt ic C lasses)

• LR (5 Polar. 5 Baltic Classes)

A "class " refers to the ice class ass igned to a ship by a classificati on society. Eac h ice

class wi ll have its ow n requ irem ent s regard ing hull th ickn ess. structura l sca ntlings.

rudders. prop ellers. mechani cal sys tems. and heat ing sys tems.

In recent years. the increasin gly globa lized industry has dem and ed a harm oni zed se t o f

classificatio ns for ships operating in the Arc tic wa ters (see Figure 2- 1). In 20 06. lACS

released a se t of Unified Requir ement for Polar C lass Ships (U RI) to co mpleme nt the

Guide lines for Ships O pera ting in Arc tic Ice Cove red Waters publ ished by the IMO . The

IMO class ific ations pro vid e a frame wo rk for the design and operation o f po lar ships and

the lACS gives specific requir em ent s on structures and machin ery. Ta ble 2- 1 lists a

ge nera l descripti on o f lACS po lar classes. Background theor ies of the URI can be found

in Daley ( 1999. 2000. 2002) . Kend rick et al. (2000a. 2000b. 2009).

18
Figure 2-1: Map of the Ar ctic Icc-Cov er ed Water Defined by IMO

19
 Polar Class Ice Description (based on WMO Sea ice Nomenclature)

PC I Year-round operation in all Polar waters

PC 2 Year-round operation in moderate multi-year ice conditions

Year-round operation in second-year ice which may include multi-year

PC 3
ice inclusions

Year-round operation in thick first-year ice which may include old ice
PC 4
inclusions

Year-round operation in medium first-year ice which may include old

PC 5
ice inclusions

Summer/autumn operation in medium first-year ice which may include

PC 6
old ice inclusions

Summer/autumn operation in thin first-year ice which may include old

PC 7
ice inclusions

2.1.2 Load Design Scenario

The energy method (Popov et al.. 1967) solves the maximum ship-ice contact force by

equating the normal kinetic energy with the ice crushing energy. A further developed

version using the process pressure-area ice crushing model can be found in Daley (1999.

2000. 200 I. 2002). and Kendrick et. al, (2000b). and is adopted in the URI. The balance

of effective kinetic energy K Ee and the ice crushing energy I E is expressed as (Daley.

1999. and Kendrick et. al, 2000b):

20
 KEz = IE Eq ua t lon z-I

where

Equatiol12-2

Where Vn is the norm al ve loc ity and Me is the effec tive mass and is given as :

M
M=-
e Co Eq uation z-J

whe re Co is the ma ss redu ction coe fficie nt. Its detailed derivation ca n be found in Popov

et al. ( 1967). It wi ll be di scussed in Section 5.3.

This appro ae h rationa lly link s the ice load to the design sce nario o f an ang ular ice edge

(the edge of a floe or a channel) glanc ing the sho ulde r of the bow. The ship is ass umed to

surge forwar d at the design spee d. hit and penet rate the ice. and then rebound away. The

ice crushing force mu st be sma ller than the min imum bendin g fo rce cau sin g ice flexu ral

fai lure . Class depend ent factors such as ice th ickn ess. ice stre ngth, ship spee d. and the

bow shape are all includ ed in the deri vation . The norm al co ntac t for ce Fn at bow is given

_
¢
tan (Z )
] 1+eX} 3+~ex 1 Z 3Z+Zex
+Zex
Eq ua tion z-u
Fll -
{ Po [ sin (If') cos( fJ')Z [zM eVn (3 + 2ex]

where ¢. fJ' are the ice we dge ang le and norm al hull frame angle respectively. The Po and

ex are fro m the proc ess ice pressur e-area relat ion ship :

Equation z-S

21
where P is the total pressure. A is the nom inal co ntac t area. Po is the ice strength ter m

co rrespo nding to the press ure on 1m 2 nomin al loadin g area . ex is the ex po nen tia l ter m

which var ies over di fferent process pressur e-area relationships. In the URI. ex is spec ified

as -0 . 1 and Po is class dependent (see Ta ble 2-2). The ir va lues are ca re fu lly chose n to

ens ure that resul ting local loads are co mpatible with both the Western and Russian

approac hes . The pressur e-area relationship in the URI is given in Equation 2-6 . The

co ncep t o f press ure-area curve is ex plained in the next sec tio n.

Eq ua tion 2-6

T able 2-2 : Ice St rengt h T erm s in th e UR I

Ice loads on non- bow areas (bow -inter med iate. mid. stern. and bott om ) are co nve rted

from the load on the bow by mult iplying empirica l area factors. The design load is

considered as the average pressur e ove r a rectang ular load patch. It is statica lly applied to

the ship structure to determ ine the minimum sca ntlings . A co mplete derivat ion of the

design load and framing design is give n by Daley ( 1999. 2000). Daley et al. (2009a .

200%. 20 I 0). and Kend rick et al. (2000a. 2000 b. 2009).

22
2.2 Ship Icc Contact and Pressure-Area Curves

2.2.1 The Ship-Icc Contact Mechanism

In the ea rliest ice load model s, the total eontaet foree was the prim ary conce rn. It was

usually es timated with an ass umption of uniform pressure distr ibuti on within the cont act

region. After 1980. more field trials and measu rement s with evo lving techn olo gies

sugges ted that the pressures actu ally vary ove r man y orders of magnitud e within the

co ntac t regio n. This mechani sm is idealized in Figure 2-2 (Daley 2004). Extruded rubble,

spa lls, internal crack s, and ex trus ion can be observ ed in all ice- structu re contact sce narios.

Flexu ral crack s may not be present unless llexural failure takes place. Direct so lid cont act

will exe rt the highe st pressure on the structure and dama ge the ice. However. the

co nfine ment in the dire ct cont act region makes it capable of sustaining very high

pressures. Extruded rubbl e and cru shed ice will result in very low pressure at the edge of

the con tact region . Thi s e ffect can be represe nted using a pressure- area plot where the

area is the independ ent variable. Ice strength, thickn ess, and velocity ge nerally vary in a

much smaller range than contact area and have less influenc e on pressure. Nowadays . the

pressure-area relationship has becom e the mo st popul ar present ation of ice pressur e data.

It is a lso used to det ermine both global and local ice loads on structures and ships. Ther e

are two distin ct types of pressure-area relationship s (Frcderking 1998 . 1999 ; Daley 1985,

2004. 2007): the process pressure- area relation ship and the spatial pressure- area

relationship .

23
 extruded rubble
';~paIlS- -
lfr-- internal crack I flexural
~

~l comminution/
i crack
~· ~xt!!J_s i on

Fi/:lIre2-2 : Skctchoflcc Contactwitha Strllctllre([)alcy2004)

2.2.2 Spatial Pressure Distribution

Figure 2-3 ex plains the spatial pressure distributi on whi ch de scrib es the variation o r local

peak pressur e on local areas within a glob al co ntac t area. At any instant tim e t or an ice

contact eve nt. a very sma ll area Al and its co rres po nding peak local average pressu re PI

ca n be plott ed as the point (AI ' PI) ' A larger area A z will necessaril y result in a sma ller

average pre ssu re pz. So another point (A z, Pz ) can be located on the plot. Simi larly. the

average pressur e P, o f the wh ole cont act area At ca n be plott ed as the point (At , Pt ) . Th e

spatial pressur e-area curve is useful in dete rm ining the design load on local structures . It

can be ex presse d as:

Eqllation2-7

where C var ies fro m 0.5 to 5MPA and e varies from -0 .7 to -0 .25 in most cases . Note that

the area discu ssed here is the nomin al contact area . Th ere are two othe r area term s: tru e

area and measu red area. T heir differ enc e is demon strated in Figure 2-4.

24
Local distribution of pressure (ideal) all al li me - (

Spatial
Pressure Area Plot
1',

A : A: AT
Figure 2-3 : Spa tial Pressu re-Ar ea Rclati ouship ([)alc y2004 )

__ force __ . force

, _.
_ force
..tndenter f spalhng
.& extrusion
, .oaneuorces
:/ prossuro panels

. J • l .. ,:

A
r

nominal pressure lrue pressure measurecpressures

disrnhutinn

Figurc2-4: Nomina l, tr uca nd measur ed a reas (Daley 2004)

25
2.2.3 Proce ss Distribution

T he process press ure-area is o ften used to det ermin e the co ntac t for ce. It gives the

re lations hi p of the average pressur e and the total co ntac t area (sec Figure 2-5) . At the

instant time t l • the tota l co ntac t area A l • a nd its corres po nd ing average pressur e P, can be

plotted as the po int (A l , Pd . As the co ntac t eve nt progresses to the instant tim e t z. the

a ve rage pressure Pz over the tota l co ntac t area A z can be plott ed as the po int (A z• Pz ).

Sim ilarly. at the ins ta nt tim e t N' the po int (AN . PN) ca n be plotted . In th is thesi s. the

di scu ssion of the process pressu re-ar ea c urve is based on th e nomin a l co ntac t a rea .

at iime = l , at time = I: at time = t -;

'\
~i~ .-l : :1_,

I 1',
I I:
I I
[A'

Pressur
Pro ces se Are a Plot
P:
/1. l~_~
. --. --
•

A ,A · :I ..
Figllre2-S: Pr ocess Pre ssure-Area Relations hip (Da ley 2004)

26
2.2.4 Spa tial vs. Proce ss

Figure 2-6 shows the co nnec tio n between the proces s and the spatial press ure-area curves .

Basic ally . at any insta nt time of a contac t eve nt. there is a co mplete spatial pressure-area

curve but only one po int on the process-area curve . As the impact eve nt develops. there

wi ll be a set of spatial pressu re-area curves . Joinin g the ends of them w ill ge nera te a

co mp lete process pressu re-area curve of the contact eve nt. The co nnec tion of the two

ind icates that greater total co ntact area and total co ntac t forces tend to yield higher

pres sure s. The spatial curve inev itab ly has a trend of falling. while the process curve may

rise or fall as the total area increases (Da ley 2004. Frede rking 1998).

Both spat ial and process curve s are co ncep ts in the co ntext of a sing le ice co ntac t eve nt.

Mos t exis ting press ure-area ana lyses are based on an asse mblage o f data and

measurement s of mult iple eve nts therefor e cannot be simply catego rized as either spatial

or process relatio nships. Those relatio nships are ge nera lly present ed in the form :

Eq ua t ion z-S

where k is the pressu re over 1m 2 load ing area; A is the loaded area and n is a co nstant

less than I (Ma ste rso n et aI2 007). For exa mp le. the pressur e-area curve in CSA S47 1 and

AP I RP 2N is P = S.lA- u,s (de rived by Masterso n and Freder king 1993). A few other

relationship s in this form can be found in Mas terso n et al. (2007) . The pressure-area curve

specif ied by the URI is a process distribu tion. It is in the form of P = PoA- u, ! as

menti oned ea rlier (see Equation 2-6 and Ta ble 2-2)

27
 at time =I,
111'."~I, ur l'd
I'andprl''''ur\.',

P,

j 3,

I p ,=CI\'g ll/l .'"

at time = t.

1'.:
:-r
- ! A:

I P:=CI\'g ll/l A:
at time = I,
F=P, I,xA ,

A,
'\ j P" A,

/ I I'. .=CI\'g ll/l A,

Process
Pressure Area Plot ~ P/"
together wi th
corre spo nding
Spatial Pressure > -'; r : : p/a
Area Plots
A rea
Fig ure 2-6: Link betw een Process and Spatiall)ist ributi on s ([)al ey 2004 )

28
As state d earlie r. the goal o f th is thesis is to inves tigate icc-stre ngthe ned ships reg ulated

by the URI. There fore . deve lop ing an ice material model whose pressur e-area re lationship

co mp lies with the URI has the utm ost impor tance . It is the co rners tone o f th is study and is

addresse d in Chapter 3.

2.3 Studies using Finite Element Analy sis Programs

This sec tio n reviews stud ies of the ship-ice interac tio n probl em usin g finite e leme nt

ana lysis. A sub-sec tion is dedi cated to studies using LS-DY NA since it is the prim ary tool

for thi s resea rch. It includes studies using the ALE meth od . In add itio n. a few stud ies

usin g ot her FEA program s will be present ed as we ll.

2.3.1 Studies Using DYNA

Gag no n et al. (2004) publ ished a paper on a series of model tests of a tran sitin g tank er

passin g by floatin g icc floes. Gag non et al. (2006) report ed an ALE simulatio n of one of

the model tes ts. The num erical so lution showe d goo d agree me nt with the physical test in

te rms of sway moti on . In the sa me paper , Gag non proposed a crus hable foam material

model for simulating shi p co lliding with a bergy bit in DYNA. This innovative material

mod el was va lidate d agai nst data from actua l measur em ent s. No te that all simulatio ns in

thi s study only allowe d the ship to move fo rward and restrained it in all ot her DOf'.

Wa ng ct al. (20 1Oa) proposed a study of ice resistan ce on the Ca nad ian ice brea ke r Ter ry

Fox in level ice. The ice failure enve lop developed by Derradj i-Aou at (2003) was ado pted

29
and mod ified to model level ice . The fluid dom ain was mod eled usin g the A LE meth od .

The ship was fi xed in all DOr exce pt the surge moti on. Simulations includ ing water. and

not incl udi ng wate r. we re co mpared with fu ll-sca le measurem en ts. Water was prove n

impor tant in num erical ana lysis o f ship break ing level ice.

Wa ng et a l. (20 10b) further investigated model ing fluid structural interac tion using

DYNA. A wave mak er was simulated using the ALE meth od . The wave len gth and wave

heigh t from the num erical simulatio n were in reaso nab le ag ree me nt with the exper ime nta l

res ults . An ALE simulation o f a th in ice piece float ing in water was also perfor me d. and

showed goo d resul ts of the buoyan cy force on the ice and its vertica l displacem ent. Late r

in the paper . simu lations of the Terry Fox moving throu gh wa ter covered by ice pieces

we re co nducted and globa l ice for ces on the ship we re reco rded. In the study. the ship was

modeled as a rigi d body and free to move only in the surge diree tion. Ice pieces we re

treated as rigid bod ies with uni form shape and size . Mesh depend ency was not

invest igated .

Extra atte ntion was devoted to reviewin g literatur e on ship local str uctura l respo nse und er

ice loads using DYNA . Unfort unately, only a few studies were found. The first one was

the Master ' s thesis by Myh re (20 10) at thc Norwegian Univers ity of Scie nce and

Tec hno logy . In his analysis of an ice co llisio n with a sec tio n o f the mid- ship structure. the

part of the ice that co uld possibly be af fecte d by the co ntact was modeled usin g the ice

model developed by Liu et al. (2009). This is a material model based on the Tsa i-W u

30
failure cr iterio n. The rest of the hal f spherical ice was treated as rigid to save comput ati on

cos t. Thi s was a ve ry e fficient approach. Liu ' s ice model is discu ssed in the next sub-

section .

The other two studies usin g DY NA were very simi lar to eac h other. Lee et al. (2007)

ex plored the possib ility of globa l 2 0 modelin g of ship-ice interacti on using DYNA. The

ship-ice co ntac t forc e was det ermin ed via a globa l analysis. and then a sectio n o f LNG

side structure was analyzed in a local FEA. Kim et al. (20 II ) foll ow ed the sim ilar

approach. They first estima ted the load by globa l analysis. and then applied it to a sec tio n

o f a cargo ship to determ ine the local structura l strength.

2.3.2 Studies using other FEA Programs

Kwak et al, (2006) an alyzed a sec tio n of the bow structure o f an Arctic tank er und er ice

load s. Ice models with differ ent elastic modulu s, failure stresses, and yield stresses were

tested in simulatio ns o f co llision betw een the rigid bow and deform abl e ice . One ice

model gave the co ntact fo rce that compl ies with the URI. Th en thi s ice mod el was used to

co llide with the flexibl e bow to eva luate the ship struc tura l stre ngth. Water and

hydrod ynam ic effec ts were not includ ed in the analysis. The meth odol ogy o f adj usting ice

material properties in thi s study is useful to the present work.

Wang et al. (2008a) eva luated the struct ura l integ rity o f an LNG ship und er a ship-ice

co llision . T hey used a co mbination of globa l and local finite element analysis modelin g.

31
T he globa l simu latio n trea ted th e shi p as e lastic- plastic and ice as crus ha ble foa m wi th a

mate rial fai lure c rite rio n. It es tima ted the ship-ice co ntac t force, co ntac t a rea, materi al

deform ation , and mat erial failure . In th e loc al fin ite eleme nt ana lysis model, the ice load

was applied statica lly to a sec tio n of the m id-sh ip str ucture to dete rmin e the cr itica l load .

T his work defin ed a proced ure fo r eva lua ting hull struc ture in LNG shi ps und er ice load s.

Fo llowing thi s proce d ure. Wan g ct al. (200 8b) invest igated ano the r ca rgo ship's struct ura l

respon se und e r ice im pac t. Di ffer ent from their prev ious wo rk. they ado pted the U R I to

de te rmi ne the va lues o f ice load and loadin g area rathe r than a globa l simulation. T he ice

patch load s fro m six d iffer ent co llis ion sce na rios were then applied to a local mod el of the

mid- ship to assess its stre ngth.

Liu et a l. (2009) prop osed a n ice mate rial mod el based on the Tsa i-W u failure criter ion.

w hich assoc iates da mage with plasti c strai n. for ana lyz ing a co llisio n between a be rgy bit

and a sec tio n of mid -ship struc ture . T he pressu re- ar ea curve P = 7.4A- o.7 defi ned by ISO

(20 08) was the be nc hmar k for Liu's ice mod el. Co m pare d to the press ure -a rea

re lations hip speci fied in th e UR I. Liu 's so lutio n overestima ted pressur e whe n the co ntac t

a rea was sma ll, i.e .. a ship im pac ting a sma ll ice floe.

2.4 Summa ry of Literature Review

A few conclusio ns ca n be d rawn from the lite ratur e review. T here is a need fo r an ice

mat erial mod el that is show n to co m ply w ith the U R I for ship-ice co llis io n ana lyses usin g

32
DYNA. It is necessary to verify if the crus hab le foam mod el (Gag no n et al. 2006) co uld

be used in this study . Ifnot , developin g a suitable ice model will be a priorit y.

Exist ing finit e eleme nt so lutio ns for ship-ice co llision probl em s can be catego rized as

foll ows:

• Mode ling ice impac ting a sec tio n of the ship struct ure: Kwak et a l. (2006).

and My hre (20 10).

• Modelin g the local ship structure und er static ice loads rather than

simulating the impac t: Wa ng et al. (2008 b)

• Using a simplified globa l ship-ice co llision mod el to determin e the contac t

force and then applying that fo rce statica lly to the ship structure in a

se para te ana lysis of the local ship structure: Wa ng et al. (200 8a). Lee et a l.

(2007). Kim et al. (20 11).

• Ana lyzi ng ship-ice co ntac t using globa l mod el ing where hyd rod ynami cs is

includ ed but the ship structura l res ponse is not: Wang et a l. (20 10a) and

Wa ngetal.(20 10b).

Eac h of their meth od s has pros and co ns. The first type does not includ e globa l moti on or

hyd rodynam ic for ces. The seco nd one does not co nsider globa l mot ion. hyd rod ynam ic

forces. ice st rength. or the dynami c e ffec t of ice load . How ever. both of the m are ve ry

quick sol ut ions. The thi rd one is more co mpre hensive than the previou s two but the

procedu re is com plicated . Co nduct ing two sepa rate ana lyses co uld be time co nsuming.

33
Although the last category is the only one that models hydrodynamic effects, it is only

concerned with the global contact and motion. It may not be a cost-effec tive solution once

the ship structural response is involved. An ideal solution would combine hydrodynamic

forces, the global motions of the ship and ice, the contact force, ice failure. and the ship

structural response in one efticient analysis. This is the goal of the present thesis.

34
Figure 3-1: Geometric Model s of the Ship and Ice in Rhinoc ero s®

Figur e 3-2: the Ice Block with Round ed Edges

36
 Figllre3-3: S hip How

Figllr e3-4 : Sep ar ationhetwccnthc Shipalld Icc (Top View)

37
 Table3-I : C eomet ry ofthe Ship and lee

Sh ip Ice

Overall Leng th 66.0 m 8.494 m

Leng th at Wa te rline 6 1.8m 8.494m

Beam 12.0m 15.154m

Il e igh t 7.20m 4 .00m

Dra ft 4.80m 3.5 10m

Co rner Rad ius N/A 1.0m

Wa te rline A ng le c 30 0 N/A

Sheer Ang ley 60 0 N/A

Fra me Ang le ~ 45 0 N/A

Waterplane Coeffic ient 0. 75 1.0

Block Coefficient 0.79 1.0

3.1.2 Materhll Mod els

The ship is always treated as a rig id bod y for the wor k cove red in thi s cha pter . Its materi al

prop erties are listed in Ta ble 3-2 .

Table 3-2 : Materiall' rope rti esofth e Sh ipM odel

Card ID MAT _ R IG ID (MA' r _020)

Ma teria l Ty pe Den sit y I Youn g' s Mo dulus I Poi sson 's Ratio

Rigid 7850kg / m 31207G Pa 03

1 .

38
Each simulation investiga ted a dif ferent ice material model. Those that showed the best

result s are presented in this chapter. An ice model based on the crushable foam material is

used in the mesh co nve rgence study. Its parameters and the stress-stra in relationship are

show n in Ta ble 3-3 and Figur e 3-5. Those parameters have minor influ ence on the mesh

co nverge nce .

Tabl e 3-3 : Mat erial Properties of the Ice Mod el

Car d ID MAT_CR USHALBEJ OAM (MAT _063)

Material Ty pe Density I Poisson's Ratio I Tensile Stress Cutoff

IE
Elastic-Plastic 900kg / m3 19 GPa I 0.00 3 800MPa
I

Figure 3-5: St ress - Volumctrtc Strain C urve of the lee Model in Convergence Study

39
3.1.3 Element C hoices

Au tomatic mes hing is used to crea te the ship and ice mesh mod els. The rig id ship is

meshed usin g shel l eleme nts and the ice block is meshed w ith so lid eleme nts. Inform ation

on the ele me nt form ulatio n and eleme nt type are listed in Ta ble 3-4 . The fu lly integra ted

formul ation is a very fast algorit hm and it is chose n for the rigid she ll eleme nts. If she ll

eleme nts are use d to mes h a non-r igid bod y. the Belytschk o-Tsay form ulation wi ll be the

bes t choice . It is the reco mme nded optio n for most structura l analys is (Q ui nton. 2009) .

The defa ult so lid clemen t ( I point so lid) is chose n for ice for its super ior rob ustness .

Ot her fully- integrated so lids are less stable when the deformat ion is large because one of

the integra tion points may have a negative jaco bia n while the who le ele me nt maint ain s a

posi tive vo lume. The co nve rge nce study that de termi nes the proper ele me nt size is

presen ted in Sec tion 3. 1.6.

Tablc3-4: Eleme nt C hoices for S hip a nd Icc

Part Eleme nt Type Formulation Option Ambie nt Ty pe

Ship Shel l 16 (Fully Integrated ) N/A

Ice So lid I (Defa ult) 0

3.1.4 Bounda ry lind Initial Cond itions

In eac h simulation. two faces o f the ice block are fixe d (sec Figure 3-6). The ship is free

to move in the longitud inal dir ection . but confined in all other 5 Oa F. It starts movin g

forwa rd towards the icc at an initial speed of 3m/s. After movin g for about 4.02 m. the

ship bow begi ns impacting the ice block at the rounded co rne r. The ice is then gradua lly
40
crushed and deforming as the collision proceeds. At the same tim e. the ship slows down

until the end of the simulation.

Figllre3-6: Boundary Condi tion on the Infinite Ice

3.1.5 Other Inputs

T he reco mmended automatic-s ingle-surface -contact is used . As discussed in Sectio n 1.3.4 .

its SOFT option is set at 1 since the material propert ies of the ship and ice are

dramatically different. There is no grav ity or any other ex terna l load . No damp ing is

added to the system.

3.1.6 Mesh Convergence Study

A mesh conversion study is co nducted by co mparing the time histories of the co ntact

forces. Figure 3-7 shows that co nve rgence is reached when the cle ment size is smaller
41
than 0.35 m. For subse quent simulatio ns, 0.24m is then cons idere d as an appro pria te

cleme nt size for subse que nt simulations.

Z 6.0 f------------------~~ /
~ 5.0

~ 4.0
~ 3.0
U 2.0

Timc(s)

Elem ent Size:

Figurc3-7:Mcsh Collvcrgcllcc

3.1.7 Nominal Contact Area

A fter eac h simulation is completed. the tim e history of the cont act for ce is directly

obtai ned from the simulation's outputs. The time history o f the nom inal co ntac t area

co uld not be acc urately give n by DY NA due to the coarse mesh, so it is deri ved using the

CA D progra m Rhinoceros®. The procedur e can be illustrated in Figure 3-8. Afte r the

ship is moved forwar d fo r a distan ce x from its initial positi on A to the new location B, an

intersect ion of the ship and icc can be created as the ye llow cur ve. The surface area of the

ye llow curve is co nside red as the nomin al co ntact area co rres po nd ing to the surge

d istance x .
42
 Figur e 3-8: Inter secti on of the Ship and Icc

Va lues of the ship surge di stances and co rres pon di ng nomin al contac t are as a re listed in

Ta ble 3-5. The rel ati on ship between x a nd A nomi n al , as o bta ined by the line of best tit, is

show n in Eq uatio n 3- 1. It is a pplica ble fo r a ll si m ulatio ns present ed in thi s c hapte r. In

eac h sim ulat ion. the time hist ory of the ship surge di stan ce is prov ide d by DY NA . It is

then subst ituted into Equat ion 3- 1 to yie ld the time histor y of the nomin al con tac t area for

that s im ulation . T he process pressur e- ar ea c urve of the ice is then ge nera ted by a na lyz ing

the time histo ry of the co ntac t force and the tim e histo ry of the nom inal co ntac t area

43
 Ta ble 3-5: S urge Distan ce x a nd Nomi na l Co ntac t Area Allom/lIl1l

x( m) An ominal x(m) Anominal

4.530 0.000 4.999 1.045

4.549 0.042 5.000 1.048

4.569 0.089 5.500 2.200

4.609 0.182 6.000 3.54 1

4.649 0.273 6.500 5. 101

4.709 0.408 7.000 6.88 1

4.749 0.496 7.500 8.882

4.789 0.583 8.000 11.103

4.849 0.7 15 8.500 13.545

4.889 0.802 9.000 16.207

4.939 0.9 13 9.500 19.089

4.969 0.979 10.000 22. 193

0, X < 4.53

Anominal = { 2.2207x - 10.054, 4.53 < X < 5 Eq uat ion Jvl

0.00 16x 3 + OA0 14x 2 - 2.0801x + 1.217 1, x 2: 5

Recall the discussio n in Sect ion 1.3.4 and Figure 1-4. which show that DYNA detects a

co ntac t before the geome tries are actually in contact. This phenome non means that the

nom ina l contact area derived in Rhinoceros® is different from that in DYNA. Although

setting SOFT = 1 helps minimi ze this discrepancy. it still makes the ana lysis of the

44
pressure -a rea re latio ns hip less acc ura te, es pecia lly whe n the con tact area is sma ll.

T he refo re, ana lys is in thi s c hapter does not inc lude da ta fro m co ntac ts where the nom inal

co ntact area is less tha n OAm 2 .

3.2 Icc Material Model s Based on the Crushable Foam Material

T his section presen ts the res ults of mode ling ice using the crus hab le foam material model

available in DYNA. Different mo de ls are deve lope d by changing the param et ers in th e

crus hab le foam mo de l. Mo re than 30 model s we re eva luated an d several of them have

showed the desir ed resu lts . In addi tion, a previo us model (Gag no n et a l. 2006) is

introd uced in this sec tio n.

3.2.1 Gagnon' s Crushable Foam Icc Model

Gag non's ice mode l (Gagno n et a l. 2006) was initiall y deve lo ped to reprodu ce the spa tia l

pressu re-ar ea curve wit h a high cen tra l pea k load . It is necessary to de ter mi ne if it tits the

purposes of this study . Gag no n's mod el is base d o n the cr us ha ble foa m material model

where the deformation is mos tly unrecoverable. Its key para me ters a re listed in Table 3-6.

T he sma ll Poisson's rat io limit s the material ' s defo rmation in d irecti on s other than the

load ing di rectio n. T he re lations hip of stress and vo lume tric stra in is show n in Ta ble 3-6

an d Figu re 3-9. Note th at in the cr us hab le foam material model, the material ' s beh avior

follows the stress-stra in re latio ns hip rather tha n the Yo ung's modulu s.

45
 Ta ble 3-6: Mal eri al I' ro pe rliesofGag no n's Ice Mod el

Ca rd ID MAT_CR USHALB EJ OAM (MAT _063)

Density Young' s Mod ulus I Po isson's Ratio I Ten sile Stress Cuto ff

900kg j m 3 9GPa 0 003 8M Pa

1 .
1

Tab le 3-7: St ress-Slr a in Relali onship in Ga gnon' s Ice Mod el

Figure3-9 : Sl ress-Vo lumetricStrain Relalions hip inG ag non's Ice Mod el

46
A process press ure -area curve o f Gag non's ice mod el is show n in Figure 3- 10. This curve

does not fi t the form of P = Po A- O. 1 • In later sec tions the prop erti es are mod ified to

develop model s wi th the desired pressure-area relatio nship to serve the purposes o f th is

st udy .

~ 8 f - - - - - - - - - - - - - - -f - - - --

I' f - - - - - - - - - - - -- r - - - - - - -

Nomina l Contact Ar ea (m" 2)

Figur e 3-10: Pr ocess Pre ssur e-Area C urve of Ga gnon's Cr usa ble Foam Ice Model

3.2.2 Icc Model A

The ten sile stress cuto ff (TSC) va lue in Gag non's model is 8M Pa . Ice models with

sig nifica ntly diff erent TSC va lues we re tried and they all displayed unsuit abl e behavior.

The Yo ung's Mo dulus has a min or impac t on the pressu re-area curve as lon g as its va lue

is in the real istic range . Th e stress-volumet ric strai n relat ion sh ip is the do minant factor in

the form of the pressu re-area curve . Material density and Poisson' s ratio are not alte red.
47
Table 3-9 and Figure 3- 11 shows the redefin ed stress-strain relationship in the modifi ed

crushable foam ice model-A. Other parameters are listed in Tabl e 3-8.

Table 3-8: Material Prop erties of lee Mod el A

Card II) MAT_CR USII ALB EJ OAM (MAT _063)

Density You ng' s Modulu s I Poisson's Ratio I Te nsile Stress Cutoff

900kg/m 3 5GPa 0 003 18.00M Pa

1 .

Ta ble 3-9: St ress - Volumetric Stra in Relation ship in Icc Mod el A

Figure 3-11: St ress - Volumetric Strain C urve in Icc Model A

48
The corre sponding pres sure -area curve is shown in Figure 3- 12. It is very clo se to the

benchmark. ote that Po = 5.91MPa in this model. It is sim ilar to the value spec ified for

Pola r Class I in the URI (see Table 2-2 ).

P = 5.9 1MPa *k O l 19

~4
-

omilia I Co ntac t Area (m"'2)

F i~ u re 3- 12: Pressur e - Area C u rve of Ice Mo de l A

3.2.3 Icc Model B

Tab le 3- 10 and Figure 3- 13 show the redefined stress -strain relation ship in the modified

crushab le foam ice model -B. Other para meters are the same as in Ta ble 3-8 . The

cor respo ndi ng press ure-a rea curve is show n in Figure 3- 14. The pressu re-area curves

spec ified for Po lar C lass-4 and Polar Class-5 are plotted as we ll for a visua l co mpar iso n.

49
 Table 3- 10: St ress - Volumet ric Stra in Relati on ship in Icc Mod el B

I
I
I

0.0 l - - " - _ - - ' -_ ----'-----_ --'--_ L - -- " -_ - - ' - _----'-----_ --'---------'

0.0

Figure 3- 13: St ress - Volum etric St ra in C urve in Icc Mod el H

50
 1'( --I I' ~ -t -\11'"

Ice Model-B: P = 2.16MPa * A-0I

PC-5: P = 2MPa * A-0I

Nominal Contact Area (m "2)

- lce Model-B - Polar Class 5 - Po lar Class 4

Figure 3-14: Pres sure - Area C urve of Ice Model B

3.2.4 Icc Model C

The ice model C is developed by slightly altering the ice model B. Table 3- 11 and Figure

3-15 show the redefined stress-strain relationship in the ice model C. Other parameters

are the same as in Table 3-8. Figure 3- 16 compares its pressure-area curve with the one

defined for the Polar Class-3 in the URI.

Table 3-1 t: St ress - Volumetric St ra in Relation ship in Ice Model C

51
 J

0.0
0.0

Figure 3-15: Stress - Volumetric Strain Curve of lee Model C

~rC-3' P= JMP" ' A'"

Model C: P = 2.898MPa * A- 0! 04

Nominal Contact Area (m "2)

- lcc Modcl C - PC-3

Figure 3-16: Pressure - Are a Curve of Ice Model C

52
3.3 Icc Ma te ria l Models Based on the Elastic-Plast ic Mate r ial

T he c lastic-plas tic mater ial model avai lable in DY NA was also eva luated . Un for tunately,

no usab le res ults we re acco mp lished . Two of the cases that ca me close to the desired

pressu re-area cu rve arc introduc ed here.

3.3.1 Icc Model D

The ice mod el D is a simple clastic-plastic mat eri al model. This type of material model

undergoes an clastic pha se then a simple linear plastic pha se when under compression .

Inputs for the simulation are listed in Tab le 3- 12. Its pres sure-area curve is shown in

Figure 3- 17.

Tab le 3-12: Material Prop er ties of Icc Mo de l I>

Card ID MA' r _ PLAST IC_ KINE MAT IC(MA'r_003)

Den sity Youn g' s Modulu s I Poisson 's Rat io I Yield Stress I Ta nge nt Modulu s
900kg / m 3 5CPa 03 5MPa 50
1 . 1 .
1

53
 Nominal Contact Are a (m I\2)

Figllre3-17: Pr esslIre -AreaClIrveoflceModell)

3.3.2 Ice Model E

Mod els based on the nonl inear clastic -plastic material (card MAr_24 in DYNA) were

also eva luated. T he icc mod el E is one o f them. Its material prop erti es arc listed in Ta ble

3- 13. A ve ry simple relati on ship of stress and pla stic stra in is defin ed (se c Ta ble 3-13) .

Cases with much more co mplica ted stress- plastic stra in relation ship s were also tested .

However. their pressure-area curves arc nowhere ncar P = 1;,A- 0 1 • The icc model E is the

one that has the best resul t. Its pressur e-area curve is shown in Fig ure 3- 18. It is c lear that

its pressu re-area re lations hip cann ot be acc urate ly ex presse d in the form o f P = ~J A-O.l .

54
 Tab le 3- 13: Materia l Pro pert ies of Icc Mode l E

Card ID MAT_PIECEWISE_L1NEAR_PLASTI CITY (MXI'_024)

Density Young' s Modulus I Poisson's Ratio I Yield Stress

900kg/m 3 9GPa 0 03 10. l M Pa

1 .

Table 3- 14: Stress - Volumetric Strai n Rela tion sh ip in Icc Mode l E

P = 3MPa * A-1I 3.H

Nomina l Con tact Area (m A2)

Fil:ure3-18: I' ressu re - Area Curve of Icc Mode l E

55
3.4 Summary

More than 80 different iee models have been evaluated. Their pressure-area relationships

are compared with the one specified in the URI. The modified erushable foam ice model

Ba nd C are very close to the benchmark. The ice model C is chosen to be used for the

rest of this study. Their relationship with the URI is shown in Table 3- 15. It seems likely

that ice material models corresponding to all the PC classes could be developed by

modifying the material parameters of the ice model A, Ba nd C. This needs to be further

explored.

Ta ble 3- 15: Summa ry of Propo sed Icc Mat eri a l Modcl s

lee Material Pressure-Area Correspo nding lee Material Model

Model Relationship Specified in the URI

A P = 5.9 1M Pa x A- D.1l 9 Similar 10 PC- I: P = 6 M Pa x A- D. l

Lies in between of

B P = 2.16 MP a x A- D. l PC-5: P = 2M Pa x A- D. l
and PC-4: P = 2.45 M Pa x A- D. l

C P = 2.898 M Pa x A- D. l D4 Similar to PC-3: P = 3M Pa x A- D. l

PC-2: P = 4.2 M Pa x A- D. l

To be developed Not available PC-6: P = 1.4MP a x A- D. l

PC-7: P = 1.25M Pa x A- D. l

56
Ice mod el D and E are based on the clastic-plastic mater ial model s in DYNA. Th ey show

goo d ag ree me nt wi th genera l form of the desir ed pressure-area relati on ship . but arc not as

satis facto ry as the icc mod el A. B and C. However . the clastic -plastic materi al mode ls

give users a large co ntro l ove r the mater ial' s stress -stra in relation ship . whi ch mean they

have the potential to give superior results. Moreover. the cla stic-pl ast ic materi a l mod els

arc more robust than the crushable foam model, which allows users to simulate cases with

very sma ll co ntac t areas . and co ntacts invol ving icc block s with sharp edges with out

encou ntering the negative vo lume probl em s. Further development o f ice mod els using the

clast ic-plastic mater ial sho uld be carried out.

In eac h case. va lues of the contac t forces are di rectly given by DY NA. Va lues of

co rres po nding nomin al contact areas are derived in Rh inoceros®. whi ch arc larger than

the actua l va lues in DYN A. Thi s unavoid abl e discrepancy is due to the natur e of the

co ntac t algorithm in DY NA. As discussed in Sec tion 1.3.4 . co ntact in DY NA takes place

be for e the geo me tries arc actually in con tact. This discrepan cy is minimi zed by excluding

data of sma ll co ntac t areas from the ana lysis. Ta king the ice model C for exa mple. its

pressure-area curve (Fig ure 3- 16) only co ntains data o f contact areas larger than a.5m 2 . If

the analysis is ex tended to the co ntac t area as sma ll as a.1m 2 • the pressure area curve will

becom e the blu e one in Fig ure 3- 19. Its trend line is in the form of P = 3 M Pa x A- o. I SS

rather than P = 3 M Pa x A- O. I 0 4 as illustra ted in Figure 3- 16. The increase in the

ex po nential te rm is to acco mmo date very large pressu res over sma ll co ntact areas. Eve n

57
so, this mathematical approximation still significa ntly underestimates pressures over very

small contact areas. This will affect the analysis of the contact force in Chapter 5.

Another limit regarding the proposed ice models is that the analysis of each ice material

model does not the cover nominal contact areas larger than I Dm2 due to the dimension of

the geometric model. Simulations using bigger ice blocks should be carried out to confirm

that the pressure-area relationships of the proposed ice models will still comply with the

P = Po A- o.1 for larger nominal contact areas. Note that such large cases are unlikely for

this study but may happen in real life.

- Icc Modcl C

PC-3 :P = 3 MP a * A ·O.l

Nominal Contact Ar ea (M "2)

Figur e 3-19: Pr essure - Are a C urve of Ice Mod el C (All Ilat a Includ ed)

58
 Chapter 4 ALE Method

As discussed in the literature review . the ALE method is utilized in severa l studies to

simu late the fluid do ma in in the ship-ice co llisio n analysis . It is natura lly co nsidered as a

pote ntial approac h for this thes is. However. all releva nt stu dies included in the literatu re

rev iew on ly allowed the ship to surge and rest rai ned it in all the other 5 DOr . while thi s

study is more intereste d in modelin g bot h the ship and ice as free float ing bodies. In

add ition. the prese nt study includ es the ship str uct ura l respo nse in the fin al so lution. It

w ill necessarily require a very refined mesh where the eleme nt size is gove rned by the

dimensions of ship str uctura l memb ers. These two factors ra ise a co ncern of the

co mputatio n cost of the ALE method. A set of AL E simulations simi lar to the model tests

by Ga gn on et al. (200 4) are co nducted to exp lore this co ncern . This revealed that the ALE

is not a practica l ap proac h for this study . The ALE meth od was however used to estima te

the added mass and damp ing coeffic ients. These we re then used to he lp deve lop user-

defin ed-cu rve- functi on s to repl ace the AL E meth od .

4.1 Simulations for Evaluating the Computation Cost

4.1.1 Geom etri c Model

The experi me nts by Gag no n et al. (2004) used a I :4 1 sca le tan ker model that was 7.20 01

in length. with a bea m of 1.1601. a dept h of 0.44m, and a draft of 0.37m. Cy lindrical,

pyra mid, and spherica l ice masses of various di me ns ions we re used as ice mode ls. In a

typica l test , the ship began to move forwa rd (the +x-direction) while the ice floe was held

at its neut ral buoya ncy pos ition . There was a se para tion between the ice and ship in the

59
tran sverse directi on (y- di rectio n) so that no co llision wou ld occur. The se paratio n is

measured as the dis tance betw een the CG o f the ship to the CG of the ice mass in the

tran sverse dir ecti on . A fter the ship accele rated from zero to the de signed speed. the ice

was fully released . The s urge and sway motion s of the ice we re measured as the ship

passed by at a co nstant spee d. Th e sway motion of the ice was wel l recorded but data of

surge mot ion were not avai lab le for all runs. Gagnon et a l. (2006) cond ucted an AL E

simulatio n o f one tcst that invo lved a spherica l icc mass. It showed a goo d ag reement

wit h the exper iment in the sway motion .

To eva luate the computatio n cost o f the A LE met hod. severa l simulations simi lar to the

expe rime nts descr ibed above have been conduct ed . Each simulation is in full sca le rather

than mod el sca le. Figure 4- 1 is the plane view sho wing the dim en sion s o f the full sca le

simulation where all un its are in meters. The geo metric model of the ship (red) was

provided by Dr. .Iungyon g Wang (Wang. 20 11). It is the sa me one used in the

ex perime nts by Gag no n et al. (200 4) and the num erical simu lation by Gag no n et a l.

(2006) . The radius of ice (blue) is 24.6 m and the se paration betw een ship and ice is 59 m.

They are dir ect ly sca led from the experime nt. The distanc e from the tip of the bow to the

cent er o f the spherica l ice in the x-dire ction is about 27m . In the ve rtica l (z-) directio n. the

ship is placed at its design ed dra ft. The ice is locat ed at the position of ncutra l buoya ncy.

The water do mai n (cya n) is 4 10m in length , 164m in wid th. and 69 .7m in depth . The air

dom ain has the same length and widt h as the water dom ain but on ly 12.3m in height. The

whole ALE dom ain (water and air) is a 41 0111 X I 64 111 X 82111 box . No te that the numb er of

60
elements is dominated by the size of the ALE domain so it is built as small as possible but

with enough space for the floating bodies to move around. The 3D model in DYNA is

shown in Figure 4-2 where the air domain is hidden and the exterior yellow layer is the

ambient water domain for a reservoir boundary condition that will be explained later. The

element size in Figure 4-2 is 4.1m.

Figu re 4-1 : Top View of the Geo met r ie Model in Rhinoc ero s®

61
 Fig ure 4-2: 3D Model in DYNA

4.1.2 Material Models

In ALE simulations , material models for water and air (including ambient entities) are

standard and straightforward as the water domain develops a proper hydrostatic

distribution. Users should strictly follow the DYNA 's instructions (LSTC , 2010).

Detailed information is presented here. The elastic -plastic material is chosen for modeling

ice (see Table 3-12) . The ice model C from Chapter 3 is not used because simulations

presented here were conducted before the investigation of the ice models introduced in

Chapter 3. The elastic-plastic ice and the crushable foam ice share the same Young's

modulus and material density , the dominant factors in the fluid structure coupling in the

ALE method . Therefore the difference in ice material should not affect the evaluation of

62
the co mputatio n cost. The shi p is modeled using the rigid material (see Tab le 3-2 ). This

co mbi natio n is very co mmo n in ot her similar st udies. Anothe r reason fo r not mod eli ng

bot h the ship and ice as rigid is that rigid c leme nts do not participa te in the co mputatio n

of time step (see Sec tio n 1.3.2). If the purp ose of the simulatio n is to va lidate the

exper ime nts men tion ed earlier. both the ship and ice sho uld be trea ted as rigid to save

comp uta tio n time .

4.1.3 Element Choices

In all simu lations in thi s chapter. the ship is modeled using she ll e leme nts and ice is

ana lyzed usin g the default so lid cleme nts (see Ta ble 3-4). Choices for clements of wa ter

and air follow the standard instructio n of the AL E meth od . They are summarized in Ta ble

4- 1. The proper eleme nt size is deter mined via a mesh convergence study . It wi ll be

disc ussed in Sect ion 4.2.6.

Ta ble ~-I : Eleme nt C ho ices for th e A LE Sim ula t ions

Part Element Formulation Op tion Ambient Type

Wa ter So lid 11 0

Air So lid II 0

Am bient Water So lid 11 4

Amb ient Air So lid II 4

63
4.1.4 Boundary and Initial Conditions

The simulation time is 15 seco nds for eac h ana lysis . Durin g the first seco nd. both the ship

and ice are held at their initi al position s when the wa ter dom ain develops a prope r

hyd rostatic distr ibuti on. At I seco nd, the ice is fu lly release d in all 6 Oa F. At the sa me

time. the ship starts moving forward at the prescrib ed spee d but is still restrai ned in the

ot her 5 Oa F. T his is similar to the model test. The ship acce lera tes from 0 m/s to 5m/s

from I seco nd to 2 seco nds . Star ting from 3 seco nds. the ship moves forwar d at a constant

spee d of 5m/s until the end of the simulatio n. The ship's forwa rd spee d is not sca led from

the ex perimen t becau se the purp ose of the sim ulation is to eva luate the co mputation cost

rathe r tha n va lida ting the ex peri me nta l results.

There are two types of bound ary co ndi tions ava ilable fo r model ing water using DY NA.

They are refe rred to as the "sw imming pool" bound ary co nd ition and the rese rvo ir

boundary co ndi tion. A "swimming poo l" bound ary conditio n allows the waves ge nerated

by floating bodi es to boun ce back and forth between the wall-lik e boundar ies . This

phenome no n signi fica ntly disturb s the hydrostatic distributi on in the fluid dom ain and

thus co ntami nates the res ults. Unless a finit e sized "sw imming pool" is de sired . a

reservo ir bound ary co ndition should be applied by addi ng ambie nt layers to the regul ar

fluid domain. The hyd rostatic distr ibution in the amb ient layers mu st be defin ed using the

*AL E_AM BIENTJIYDROSTAT IC car d . Waves ge nerate d by float ing bodi es will flow

into the ambie nt layers and not boun ce back . In other words. the ex tra ambie nt layers

tra nsfe r a fi nite flu id dom ain to a pseud o infi nite one witho ut using more eleme nts or

64
increasing the size of the fluid domain. Figure 4-3 shows the ALE domain with ambient

layers. Part of the domain is cut away for a clear demonstration. The water domain (blue)

is surrounded by the ambient water layer (yellow) on all four sides and the bottom. The

air domain (red) is surrounded by the air ambient layer (green). There is no ambient layer

on top of the air domain simply because it is not necessary. The reservoir boundary

condition is used in all simulations presented in this chapter.

Initial conditions for the ALE domain are defined strictly following DYNA's instructions

for a realistic hydrostatic pressure distribution that takes about 0.5s to develop. Note that

ambient entities and regular ALE clements require separate inputs.

Figure 4-3 : the A LE Domain including Ambient La yer s

65
4.1.5 Oth er Input s

Gravi ty is included in the ana lysis. Hyd rostatic and hydrod yn ami c forces are includ ed

usin g the ALE meth od . When defin ing the fluid- stru ctu re co upling force . the Pf-AC va lue

is ca refu lly ca librated so that eac h floatin g bod y is in neut ral buoyan cy at the initi al

posit ion . In eac h simulation. dampin g is applied to the water and air domain s for the first

0.5 seco nds. It help s the A LE dom ain to fo rm the rea listic hydro stati c pressure

distribution fa ster. Th ere is no ship-ice contact in eac h simulatio n.

4.1.6 Mesh Convergence

The co nverge nce study is conduc ted by obse rvi ng the surge. sway . and heave moti on of

the ice in simulations using various clement sizes. Results are show n in Fig ure 4-4 . Fig ure

4- 5. and Figure 4- 6. It is obv ious that the con vergenc e in the sway moti on is exce llent.

Co nvergence in the surge moti on is acce ptable wh en the eleme nt size is sma ller than

4.1m. Co nve rge nce is not reac hed for the hea ve mot ion . Note that the simulation report ed

by Gagno n ct a l. (200 6) cont ain ed about 2 milli on eleme nts and had good ag reeme nt with

the model test in the sway motion . So overall, the mesh conver genc e is acce ptable. Th e

numb er of eleme nts and the total comput ation time for eac h case are summarized in Ta ble

4-2. It is reason abl e to conclud e that 1.33m is an appro priate eleme nt size . Further

refinin g the mesh may give bett er co nve rge nce in all three motion s. but it will also

significantly add to the total comput ati on cost. wh ich is the biggest conce rn regardin g the

use of the A LE method.

66
 Ti mc(s)

Elcmcnt Size: - 4. lOm - 2.67m - 2.05m

Fi ~lIrc 4-4: Convergence of thc Icc Surge Motio n

Ti mc(s)

Ele ment Size: - 4. IOm - 2.67m - 2.05m - 1.33m

Fi~lIrc 4-5: Convergence of Sway Motion of Icc

67
 Elcmcnt Size: - 4. 10m - 2.67m - 2.05m - 1.33m

Figu re4-6: Con \'e rgen ce of th el ceH ea \'eM oti on

Ta ble 4-2: Su mma ry of th c Mcsh C onvergence St udy

Case No. A B C [)

Mesh Size 4 . 10m 2. 67 m 2.05m 1.33m

No . o f Eleme nts 94 ,280 3 13.6 96 726 .354 2.40 1.832

Co m putatio n T ime l hr 16 m in 4 hr 17m in 12h r 34m in 50hr23 m in

No . of Co m pute No des 4 4 4 4

Note: All s im ula tions a re sol ved on STe l'S 2 C lus ter (see Appe nd ix A) whi ch has 8 cor es 111 eac h

comp ute nod e.

68
" .1. 7 Eva lua tion of the Co mputation Cos t

The total computation cos t for an ALE simulation depend s on four fact ors: element size.

the tota l number of elements. the co ntact model. and bound ary co nditions. In eac h case

presented in this chap ter. the ALE dom ain is built as sma ll as possible but with enough

space for the ship and ice to move aro und. One floatin g part (ice) is ass igned with 6 DO F

while the other one (ship) is only free to move in the surge d irection . Co ntact is not

includ ed. Case D. where the element size is 1.33m. co ntained about 2 mi llion elements

and took abo ut 50 hours to solve.

Ass uming a very small ship-icc model that is one third in size of thc model presented

above is used. the dim ension of the A LE domain will becomel00m x 50m x 20m . With

the same cleme nt size as Case D. the total eleme nt numb er will be about 43.000. This

number is abou t 1.8% of Case D. If a 5-seco nd simulatio n time ( 1/3 of Case D) is needed

for a ship-icc co llision analysis. then the total comp utation time ca n be roughly es timated

as 5 0111' x 1.8 % x ~ = 18 min = 0.3111'.However. this is for the simulation using solid

elements of 1.33m in length. In a ship-icc colli sion analysis involving local structura l

response. the eleme nt sizes of the ship. ice. and the ALE dom ain should be similar to each

other for acc urately modeli ng co ntact and co upling fo rces. The elem ent size will

necessarily be domi nated by the sma llest parts. which arc the structural mem bers of the

ship. The proper element size should be in the neigh borhood of 0. 15m. For the same AL E

simulation whose d imension is 100m x 50m x 20m. the total num ber of element s will

be about 100 x 5 0 x 20 -i- 0.15 3 "" 29.6 m illions . Recalling equations [3.1] and 13.4]

69
which show that computation time step is proportional to element length. the total

estimated computation is 0.3hr X 29.:;~~~ons X ~:~~: "" 1831 hour s « 76 Days .

Including the contact model and more DOF in the analysis will increase the computation

time even more. Although using a non-uniform mesh may cut the number of elements in

hal f, the total computation cost will still be about 38 days. This is not an efficie nt or

practical engineering solution.

A more cost effi cient approach. the user-defin ed-curve-function. can be used to model the

hydrodynamic effects and hydrostatic forces. The ALE domain can be completely

eliminated and this reduces the number of elements by about 80%. It also makes the

numerical model much simpler. Using user-defi ned-curve-functions to account for

hydrodynamic forces requires added mass and damp ing coeffici ents as inputs. These

values can be derived from model tests. analytical solutions. numerical simulations. or

emp irical estimations. In the present study. ALE simulations arc used to calcu late them.

The detailed discussion of implementing user defined function is addressed in Chapter 5.

4.2 Oscillatory Ana lysis vs. Transient Ana lys is

There arc two methods for estimating added mass terms: the oscillatory analysis and the

transient analysis. Taking the heave added mass for example. the oscillatory method

assigns an initial heave displacement to the floating body and then lets it osc illate in the

water. The heave added mass coefficient a 33 can be solved using Equation 4- 1:

70
 TZ k
G33 = 4rr Z ;j" - 1 Eq ua t ion 4-1

where T is the osci llato ry perio d, m is the mass of the float ing bod y, and k is the heave

stiffness.

T he transient ana lysis so lves the sa me pro blem by app lying a for ce in the heave d irec tio n

to the floating body . The heave ad ded mass coefficient G33 is give n as :

Eq ua tio n 4-2

w here F is the ex te rna l force applied. a is the acce lera tio n du e to the fo rce. a nd m is the

mass of the floati ng body. This sec tion w ill co m pa re the two approac hes using ALE

sim ulatio ns.

4.2.1 Geometric Model

Sim ulat ions usin g the tran sient and osci llatory meth od s have the sa me geo met ric model.

It is show n in f ig ure 4- 7 where the ai r dom ain is hidd en. T he blu e pa rt is the water

domai n and the ye llow part is the ambient wa ter layer . Di me nsio ns of the AL E do ma in

arc in Tab le 4- 3. The d ia me ter of the se mi -sub merged sphere (red) is 6 m. T he c leme nt

size in the fig ure is O.5m .

71
4.2.2 Material Models

The sphere is modeled as a rigid body using shell elements. Its density is set at 2960. 10

kg / 1113 so that it is in a state of neutral buoyancy when it is semi-submerged.

4.2.3 Element Choices

Element choices for the ALE Domain are the same as in Table 4- 1. The semi-submerged

sphere is meshed using shell elements rather than solid elements to reduce the total

number of elements. Information of the shell elements is available in Table 3-4. The

thickness of each shell is 0. 1694m. The element sizc is much smaller than that in Section

4. 1. therefore no mesh convergence study is conducted for this analysis.

4.2.4 Boundary and Initial Conditions

The reservoir boundary conditions are applied to the ALE domain in all simulations. The

hydrostatic pressure distribution properly develops in about 0.5 seconds.

All 6 DOF on the sphere are constrained for the first second. Starting from I second. there

arc two options for the boundary condition on the sphere. The first option is to set the

sphere to be completely free floating. A few simulations suggested that this type of

boundary condition makes it very difficult to analyze the heave motion for two reasons.

f-irst. since the perfect PFAC value is impossible to find, the sphere will always osc illate

around its neutral buoyancy position. Besides, motions in other DOF tend to couple with

the heave. This influence is especially significant when the heave is not the dominant

73
motion . i.e.. a sma ll initi al heave displacem ent or a sma ll extern al force is appli ed to the

sphe re. So me of the simulations report unreali stic all y negati ve valu es for the heave added

mass. Therefo re the other type of bound ary co ndition is chosen . Th e sphere is restrained

in all DO r exce pt the one that is bein g inves tiga ted; therefor e. the added mass in each

DO F is eva luated inde pende ntly. No te that wh en the applied externa l for ce is relati vely

large. these two bound ary co nd itions show similar res ults .

In the osc illatory anal ysis of the heave added mass. the sphere is ass igned a prescrib ed

heave mot ion fo r the first 3 seco nds (see Figure 4-8 ). At 3 seco nds . it is rele ased to be

free in the heave mot ion but fixed in all the other DO r . In simulations using the tra nsient

analysis. there is no presc ribed moti on for the sphere. The sphere is fixed in all 6 Do r

du ring the first seco nd. and then it is rele ased in the heave motion but still restra ined in all

othe rS DO F.

Time(s)

Figure 4-8: I'r eseribedHeaveMotionofthe SphereintheOscillatoryAn alysis

74
4.2.5 Loading Conditions

Grav ity is incl uded in all simulations. There is no other load in the simulation usin g the

osci llatory meth od . Mult iple simulations using the tran sient approac h are ca rried out. In

eac h sim ulatio n, a force is appli ed to the sphere in the -z -direction (dow nwa rds) to push

it into the water. The for ce ramps up from zero to the designed valu e in 0.00 9 seco nds

(from 2.00 1 to 2.0 1 seco nds). The value of the force va ries ove r simulations to investi gate

if the magnitud e o f the force af fects the hea ve added mass.

4.2.6 Added Mass using the Oscillatory Analysi s

The time history of the heave moti on of the sphere is show n in Figure 4- 9. The osci llatory

peri od is abo ut 3.65 seco nds . The mass of the sphere is 56,43 8.8 kg-. Its heave sti ffness is

276,39 4 N / m . Substituting those va lues into Equation 4-1. the heave added mass

coefficient is so lved as 0.654 . The tim e history sugges ts that dampin g is very small. A

calc ulatio n usin g the logar ithm ic dec reme nt meth od sugges ts that the dampin g rat io is

approximately 3%.

75
 Tim e( s)

Figllre4-9:HeaveMotionofthe Semi- SlIbmcr ged Sph erc

4.2.7 Added Ma ss usin g the T rans ient Ana lysis

Results o f the simulatio n where the externa l force is abo ut 22 times larger than the

sp here' s weig ht are show n in Figure 4- 10. After the force is applied, the sphere acq uired

a large acce leration in the - z-direetion. Buoya ncy increases as the sp here is pushed into

dee per water, which ca uses the acce leration to decrease . At abo ut 2. 1 secon ds, part of the

sphere is pushed out of the fluid domai n so the buoyancy begins to decrease. The

decrease of the acce leration is co nsequently slowed down . Even tually, the sp here is

comp lete ly pushed out of the water. This result s in zero buoyancy and hence a co nstant

acceleration due to gravity and the force applied. In the solution, both the force and the

76
acceleration a re negative va lues . Th e min imum val ue o f the accelerat ion is give n by

DY A ' s outputs. T he heave adde d mass coefficient is so lved us ing Eq uatio n 4-2. No te

that be fore the force is app lied, the sphere a lrea dy has a sma ll heave acce lera tion du e to

the unstable hydros tatic force . T his noi se is du e to the natu re of the AL E si mu latio n a nd

is tak en into co ns idera tio n d uring the a na lys is. A ll resu lts fro m a ll s imulatio ns arc

sum marized in Ta ble 4- 4 a nd Fig ure 4- 11. It is obvio us that the heave adde d mass in this

a na lysis is inde pende nt o f the force a pplied.

T abl c 4-4 : Innll CII CC oft hc Ma gnitude of t hc Fo rcc 0 11 th e Hea ve Adde d Ma ss

Magn itud e o f the Fo rce ( KN) 30 54 120 56 7 1.206 6, 03 0 12,060 18,090

Force/ Body -we ight Rati o (%) 5 10 22 103 2 18 1090 2 18 1 327 1

Heave Ad de d Mass Coe. (%) 8.9 6. 1 8.7 10.2 8.7 9. 1 9.0 9.4

77
50 I

O ~--="""T--'''''''''''''==:::::!I!9F=~~~-_''''''''=--_
15

Ti mc(s)

- Displacement (m) - Velocity (m/s) - Acceleration (m/s"'1)

Figu rc 4-1 0: Ti mc History ofthc Se mi-S ubme rged Sp hcrc's Heave motion

~
\ ~
.--------.

5.0%
0%
Load /Wcig htRatio

Figllrc4- 11: Infl llcn cc of th c Magn itlldc of t hc Force on the Heave Added Ma ss

78
4.2.8 Comparison

It is clear that the adde d mass est ima ted by the osc illatory meth od is la rger tha n th at g ive n

by the tra nsie nt a pproac h. In other wor ds, " water" in DY N A is mor e respon s ive and

se nsi tive to the low- frequ enc y mot ion of the tloat ing bo dy . A simi lar phen om en on is also

re port ed in man y tim e-d om ain studies o n the osci llating hem isph e re. T hey arc

sum ma rize d in Ta ble 4- 5 w here cases of ve ry-low freq ue nc ies share a si mi lar natu re with

the osc illatory ana lysis. a nd cases fea turing ve ry-h igh frequ en cies corres po nd to the

tran sient ana lys is.

Tahl e 4-5: Heav e Add ed Mass C oefficients at Ver y-Low Fr equencies and Ver y-High Freq uencies Oil
the Unit Hemi sphere

Very -Low Freq ue nc ies Ver y-H igh Freq uencies Refer en ce

0.8 0.4 Sie revoge l ( 1998). Prin s ( 1995)

0.8 0.5 Kor sm eye r et aI. ( 1989) . Lia pis ( 1986)

0.83 0.5 Hulm e ( 1982)

0.83 0.5 Stor ti et a I. (2 004)

It is a ppare nt that the heave added mass estima ted usin g the ALE meth od is s ignifica ntly

sma ller than th ose give n by the tim e-d omain stud ies . This d iscrep an cy is m ost likel y du e

to the natu re of the fl uid-structure co up ling in DYNA. As di scu ssed in Section 1.3.3. it

co m putes the co up ling force usin g a pen alt y method . i.e.. the fo rce is a lways a func tio n o f

the d isp lacem ent. Whil e in rea lity. th e adde d mass is in ph ase wi th acceleratio n o r

dece leration . Results usin g the time -do ma in ana lysis are mo re tru stwo rth y.

79
For the ship-ice co llision ana lysis. the contac t force is more like an impul se fo rce. i.e.. a

very-hig h frequ ency load . Thus. the tran sient approac h is more suitable for estimating the

adde d mass term s.

4.3 Estimation of Added Mass and Damping Coefficient

ALE simulatio ns follow ing the transient approac h are co nducted to estima te the ad ded

mass and da mping coe ffic ients of the ship and ice. They wi ll be input into user-d efin ed-

curve-fu nctions in the next cha pter to replace the ALE meth od .

4.3.1 Gcomctric Model

The ship and ice are ana lyze d in se pa rate simulatio ns. Geo met ric models and of the ship

and ice are the sa me as in Sec tion 3.1.1 exce pt that ice edges are not rounde d. The

dim en sion of the water dom ain is 80m x 24m x8m . The dim en sion o f the ai r dom ain is

80m x 24m x 4m . The 3D models are sho wn in Figu re 4-12 and Figure 4-1 3 where the air

dom ain is hidd en.

80
Figure 4-12: 3D Model for Estimating the Added Mass Coefficients on the Ship

Figure 4-13: 3D Model for Estimating the Added Mass Coefficient on the Ice

81
4.3.2 Materia l Models

Both the shi p a nd ice are m od eled usin g the rigid mat erial since the defo rm ati on is

irrele van t. Mater ia l prop erti es of the ship a re the sa me as in Tabl e 3-2 . Since the ice block

is mesh ed usin g she ll e leme nts. its m ater ia l den sit y is set at 6 12 1.69 kg / 11/3 so th at it has

the sa me mass and sa me neut ral bu oyanc y positi on as if it was a so lid block . Th e c hange

in the mom ent of iner tia du e to thi s adj ustme nt is taken into co ns ide ration in a ll the

ca lculations perform ed in thi s cha pter.

4.3.3 Eleme nt Choices

Eleme nt c ho ices o f the AL E doma in a rc the sa me as in Table 4- 1. Bot h the ship and ice

are mod e led usin g the sa me she ll clement s as tho se in Ta ble 3-4 .

4.3.4 Bou ndary and Initial Co nditions

Boundary conditions o n the A LE dom ain are the same as the A LE sim ulatio ns pre sent ed

ea rlier. Boundar y conditions on th e ship and ice are sum ma rized in Ta ble 4-6 .

82
 Table 4-6: Bound ar y C ond itions on the Ship and Ice

Case From 0 second to 1 second From 1 second and onwards

Surge All 6 DOF Restrained Free in Surge, Restrained in all other DOF

Sway All 6 DOF Restrained Free in Sway, Restrained in all other DOF

Heave All 6 DOF Restrained Free in Heave, Restrained in all other DOF

Roll All 6 DOF Restrained Free in Roll. Restrained in all other DOF

Pitch All 6 DOF Restrained Free in Pitch, Restrained in all other DOF

Yaw All 6 DOF Restrained Free in Yaw, Restrained in all other DOF

4.3.5 Loading Conditions

In each simulation. a force or moment is applied to the floating body (ship or ice) for 2

seconds (from 6 seconds to 8 seconds). When investigating the translational motion

(surge. sway and heave). the magnitude of the force is about 10% of the body weight to

generate an acceleration of 1m/ 5 2 if there is no "water". When studying the rotational

motion (roll. pitch and pitch). the moment applied to the body will generate a rotational

acceleration of 0.1 Ta d /5 2 if there is no "water" in the analysis. Values of the applied

loads are summarized in Table 4-7. Note that when analyzing the ship's heave motion. the

force is applied to push the ship downwards into the water, but when investigating the

ice's heave motion. the force is applied in the positive z-direction to lin the ice up. This is

because a large force downwards makes the ice completely submerged and hence difficult

to observe the heave motion.

83
 Tabl c4-7 : ValucsofApplicd Load s

Floating Heave Roll Pitch

Bod y Va lue ( K N) Va lue ( M Nm) Va lue (M Nm)

Ship -3,037 8,653,020 101,8 49,00 0

lee 453 1,290,580 564 ,483

4.3.6 Ship ' s Added Mas s and Damping Terms

The time history of the ship heave moti on is show n in Figure 4-1 4. Th e ship is released at

1 second . The heave acce leratio n begin s to vary dram atic all y for the next 2 seco nds du e

to the natur e o f the PFAC va lue. It then becom es more stable but still osci llates a bit. At 6

seco nds , a force is applied to the ship and generates a large acce lera tion in the - z-

direction . It pushes the ship down ward s into the water. The incre as ing buoyancy ca uses

the heave acceleration to decrease, The force is rem oved at 8 seco nds and buoyance

immed iately domin ates. It gives the ship a large positi ve acce leratio n. A fter that, the ship

starts osc illating in thc water. The magn itud e of the acce lera tion at 6 seco nds is 0.8 179

III I .1" • Using Equation 4-2 co mputes the heave added mass coeffic ient as 0.198. The

damp ing ratio is calcul ated using the logarithmi c decrem ent meth od . Its va lue is 0.722.

Rcsult s are sum marize d in Ta ble 4- 8.

84
 Time(s)

- Displaccmcnt (m) - Vclocity (m/s) - Acccicration (m/s"2)

Figure 4-14: Time History of the Ship Heave Motion

The lime history of the ro ll moti on is shown in Figure 4-1 5. A mom ent about the x-axis is

app lied to the ship at 6 seconds to initiate the roll motion . A lter it is rem oved at 8 seco nds .

thc restoring for ce ca uses the ship to osc illate. The accelera tio n due to the applied

mom ent is 0.08 98 ra d / S2 . The roll added mass coeffi cient is ca lculated as 0.114 usin g

Equatio n 4-2. The osc illation in Figure 4-1 5 co ntains two cres ts and two trou gh s. Th e first

cres t and trou gh have relative ly large amplitudes. The seco nd cres t and tro ugh arc mu ch

sma ller. Dam pin g barely ex ists if onl y the first crest and trou gh are ana lyzed . How ever .

the who le osc illation sugges ts that the sys tem is heavily damp ed co mpared to real-li fe roll

mot ion . Th e dampin g rat io is ca lculated as 0.7 8 usin g the logarithm dec rement meth od .

Results are summa rize d in Ta ble 4-8 .

85
 Tim e( s)

- Displaccmcnt (m) - Velocity (rn/s) - Acceleration (m/s"2)

Figure 4-15: Time Histor y ufthe Ship Roll Motion

Pilch mo tion is inves tiga ted in the sa me wa y as the roll moti on . Its time histo ry is show n

in f igure 4-1 6. A mom ent abo ut th e y-axis is applied to the ship at 6 seco nds to trigge r

the pitch mot ion . A fter it is re move d at 8 seco nds , the shi p begin s to osc illate abo ut th e y-

axis . T he acce lera tio n d ue to the applied mom ent is 0.08 71 r ad] 52 . A pply ing Eq uatio n

4-2 g ives the pit ch adde d mass coe ffic ient as 0./48. The logarithm decr em ent sugges ts

the damp ing ratio is 0.44 2. Result s are summa rized in Ta ble 4-8 as we ll.

86
 Timc( s)

- Displacement (m) - Vclocily (m/s) - Acceleraloll (m/sI\2)

Figur c"-16:TimeHisloryofth e Shipl'itchMotioll

Added mass coeffic ients in the surge. sway . and yaw motion s are estimate d in the sa me

mann er excep t that dampin g did not ex ist in those three moti on s sin ce there is no

osci llatio n. Result s are summar ized in Ta ble 4-8.

Tabl e "-8: Added Mass and Dampin g Term s of the Ship

Te rm Surge Sway Heave Roll Pitch Yaw

Added Mass Coe. 0.178 0.140 0.198 0.114 0.148 0.11 3

Dampin g Rati o N/A N/A 0.722 0.787 0.44 2 N/A

87
4.3.7 Icc Added Mass and Damping Terms

Added mass and dampin g term s of the ice are ana lyzed in the sa me way. T ime histori es

o f the heave, ro ll, and the pitch mot ion are show n in Figur e 4-17, Figu re 4-18. and Figure

4-19 respe cti vely. Each o f them dem on strates a similar pattern to the co rres ponding tim e

history o f the ship mot ion . No te that the osc illation in the heave mot ion is very unstabl e .

Va lues o f the first two trou ghs are then used to ca lculate the dampin g rati o. Added mass

coeffic ients in the surge, sway , and yaw moti on s are estima ted as well. All the result s are

summa rized in Tabl e 4-9.

Time(s)

- Displacement (111) - Velocity (111/5) - Acceleration (m/s"2)

Figure 4-17: Time Histor y of the Ice Hea ve Motion

88
 Timc(s)

- Displaccmcnt (m) - Velocity (m!s) - Accclcration (m!sA)

Figllrc 4-18: Ti mc History of th c Icc Roll Motio n

Timc (s)

- Displaccmcnl (m) - Vclocity (m!s) - Accclcration (m!sA2)

Figllrc 4-19:Tirnc History of th c Icc Pitch Mot ion

89
 Tablc4-9: Added Ma ss and Damping Terms of the lee

Ter m Surge Sway Heave Roll Pitch Yaw

Ad ded Mass Coe. 0.226 0. 165 0. 193 0.247 0. 147 0.149

Damp ing Ratio N/A N/A 0.51 1 0.354 0.072 N/A

4.3.8 C om pa r ison

Ta ble 4- 10 comp ares the added mass terms estima ted in the previou s sec tions w ith the

ones give n by the so ftwa re Direct Design for Polar ships (DDePS). a so lution that is

based on the Popov ' s deri vation (Popov et al. 1967) and further developed by Dale y et a l.

(2007 .200 8, 2009a) . A pplica tions o f DDe PS can also be found in Dale y et al. (2009b ,

20 10). and Kendri ck et al. (2009). Note that DD eP S and Popov' s estima tions are based on

ex peri me nta l and empirica l va lues rather than the transie nt ana lysis. There fore. the

discrep anc y in the results is ex pec ted. The added mass coeffic ients of the ship given by

DDcI)S are very reasonab le and agree well wit h the co mmo n knowledge of the added

mass o f ships. They are a lso in the same orde r of magnitu de wi th the va lues suggested by

DYNA. In ter ms of the ice block . estimations give n by these two so lutions are a lso in the

same ord er of magnitud e exce pt fo r the heave and pitch added mass coeff icie nts. It is

difficult to co nclude which so lution gives the more acc urate answe r since there are no

other studies to co mpare with for this parti cular case.

90
 Ta ble 4-10 : Add ed Mass Cocfficicnts Ca lculated by DDcl' S (196 7) and Pre sent Work

Ship Ice
Adde d Mass
Present Wo rk Present Work
Coe ffic ients DD eI' S DOd'S
Us ing DY NA Us ing DYNA

Surge 0.000 0. 178 0.827 0.226

Sway 0.800 0.140 0.464 0.165

Heave 0.675 0.198 1.443 0.193

Roll 0.250 0.114 0.250 0.247

Pitch 0.74 1 0.148 2. 164 0.14 7

Yaw 0.575 0.113 0.328 0.149

4.4 Summar y

The ALE meth od has been show n to be imp ract ical in sim ulating the fluid dom a in in th e

ship-ice co ll isio n a na lys is. It is repl aced wi th user-d efin ed -cu rve- fun cti on s in C ha pter 5.

w hich is a n approac h that req uires the adde d mass and damp ing coe ffic ients as inputs.

T hose coefficients are eva lua ted usin g the tran sient a na lys is in ALE si m ulatio ns.

As show n by studies using the time -do mai n ana lysis . the tra nsient ana lys is and osc illato ry

ana lysis g ive very diff er ent es tima tions for adde d mass coe ffic ients. Th is is co nfir me d by

si m ulations in DY NA . How ever. ther e is a di scr ep an cy in the res ults du e to the natu re o f

the ALE meth od . As menti on ed ea rlier. DY NA uses the pen alty based a lgori thm to

det erm ine the flu id-stru ctur e co upling forc e. i.e.. the forc e is a fun cti on of the pen etr ati on
91
depth between the bod y a nd wa ter. Since the adde d mass a nd dampin g are in phase w ith

the acce lera tio n a nd ve loci ty resp ecti vel y, D YNA is not the idea l tool to es tima te them.

T he adde d mass coefficient s give n by DD eI)S are ve ry di ffe rent from thos e es tima ted by

DYN A. This is ex pec ted s ince the DDcI )S so lution is based o n empi rica l va lues. whil e the

DYNA ' s res ults are deri ved from tran sient ana lyses . A ltho ug h the ship-ice co ntact is a

tran sient process, the adde d mass coefficient give n by DY NA is not ac c urate du e to the

natu re o f the prog ram . It is diffi cult to determ ine whe ther the a nswer g ive n by DY NA o r

DD eP S is closer to the true va lue since th ere a re no othe r studies to co mpare wi th for thi s

par ticular case . How eve r. in C ha pte r 5. adde d mass coe ffic ients a nd damping rati os

deri ved in th is sec tion usin g DYN A a re used to ca lculate input s fo r mod elin g wat e r usin g

defin ed- cu rve-fun cti on s. The re a re tw o reaso ns for choos ing va lues g ive n by DY NA.

Firstly. the prac tice performe d here is repl ac ing the ALE meth od w ith curve functi on s.

T he refor e. inp utting va lues give n by the ALE method to the c urve functi on ca n g ive the

res ults sim ilar to s im ulations using the ALE meth od . A no ther reaso n is that DD eP S is a

ve ry q uick so lution th at only tak es a few m inut es to finish one ana lys is . It is mu ch

si m pler to input var io us se ts of adde d mass coe ffic ie nts into DD eP S a nd co m pa re th e

res ults w ith the DY NA so lutio n than the other way ro und. If add ed mass coe ffi cie nts

s ugges ted by mo re reli abl e so urces are ava ilable, such as ex perime nts, a nd CF D

s im ulatio ns. they sho uld be ado pted into the use r-d efin ed- cu rve-fun cti on s.

92
Another factor that may contaminate the result is the PFAC value. It must be carefully

calibrated for each simulation. If the geometry model. the element size. or any other input

related to fluid-structure coupling is modified. the PFAC value must be recalibratcd. This

is a very time consuming process that may take hours or even days depending on the

mesh and complexity of the simulation.

93
 Chapter 5 Ship Icc Collision Force

Thi s chapt er discus ses model ing the globa l ship-ice cont act force . Restor ing force s are

includ ed in the solution by mode ling the wa ter domain as a spring syste m with user-

defin ed -c urve -functions. Several ship -ice glancing cases are ana lyzed . Results are

compared wit h calcu lation s using DDeP S.

5.1 Defined -Curve-Functions

This sec tio n d iscusses how to impleme nt the *DEFINE_CU RVEU :U NCT IO N card to

mod el the water instea d o f simulating the water dom ain usin g so lids elem en ts. Modelin g

the restori ng for ces is cove red in Sec tio n 5.1.1. Mod el ing the drag force and the added

mass effect is discu ssed in Sectio n 5.1.2.

5.1.1 Restoring Force s

As mentioned in Chapter I. the *DE FI E_CU RVE_ FUNC T IO card defi nes a curve

where the ab sc issa is and can only be time and the ordin ate is exp ressed by a func tion of

ot her curve defini tion. force s. kinematica l qua ntities. intr insic functio ns. inter polati ng

pol yno mia ls. or co mbinatio ns thereo f The ordinate in a user-de fined-cur ve- function is

updated at eac h tim e step as the simulation proceed s. The re fore the ordi nate is a func tio n

of tim e. Users ca n then defin e a force (or for ces) as a function o f the ord ina te va lue. For

exa mple. a force ca n be defin ed as a function o f the d ispla cem ent. This displ acem ent is

give n as the ord inate in a user-defin ed- cur ve-fun ct ion . and is updated over tim e by

94
DYNA. Three of ava ilable curve functio ns are used in thi s study to acco unt for restor ing

forces (LSTC 2007a) :

• DZ (node ): It reports the z-tra ns lationa l d isp lace me nt of node N in the

globa l coo rdinate sys tem.

• AX (node ): It reports the rotational d isp lace me nt o f the node N abo ut

the x-ax is in the globa l coor dinate sys tem. A local coor dinate sys tem mu st

be de fined at the node N.

• A Y (node N): It reports the rotat ional displacem ent o f the node N abo ut

the y-axis in the global coordinate sys tem. A local coordinate sys tem must

be defin ed at the node N.

Ass umi ng the node N is the center of grav ity (CG) o f a floatin g body and its local

coo rdi nate syste m shares the same orientat ion with the globa l coord inate sys tem. the three

functions (DZ. AX and A Y) will update the val ues of heave. roll. and pitch at eac h time

step . In a fi nite element ana lysis. ca lculation time step is ge nera lly very sma ll. This

implies none of those va lues will vary sig nifica nt ly at eac h time step . There fore. they ca n

be used to ca lculate restoring forces using the hydrostati c theor y if sma ll moti on s are

ass umed. The stiffness in heave. roll. and pitch are also requi red to defin e restorin g forces

in DY NA. Th ey ca n be eas ily calcul ated with the outputs from Rh inocero s® . In summa ry.

this approa ch calcul ates restorin g force s using the hydro static theory at each time step.

However . those for ces are time dependent and ap plied dynamic ally to the floatin g bod y in

the globa l co ntex t. The added mass coeffi cient s es tima ted in Sec tion 4.3 are ado pted to

95
ca lculate the da mping coeffi cient s. The three curve -functions have been tes ted and pro ved

to be res ponsive to the moti on of the body.

5.1.2 Implementation

Impl em ent ation of user-d eli ned- curve -functions w ill be ex plained usin g the ice block

from Sec tion 3.1 as an exa mple. T he bound ar y co nd ition (see Figure 3-6) on it is rem oved

so it becom es a free- floa ting body.

Step 1: Creating a rigid part on the non-rigid body. The rigid part is for reference purp ose

so that a local coordin ate sys tem can be built on the ice. The rigid part could be very big

or very sma ll as lon g as it does not affec t the ove rall be hav ior of the non-ri gid bod y. It

may co nta in as few as one eleme nt. In this case. cle men ts that are far away from the

poten tial co ntact regio n are redefi ned as rigid usin g the " move' co mma nd in DYNA. The

first step is shown in Fig ure 5- 1 where the ship is in red. the crushable ice is in blue. and

the redefined rig id ice is in gree n. This step is not necessary if the whole body is modeled

as rigid in the first place.

Step 2: Creating the loc al coo rd inate sys te m. It is show n in Figure 5-2 . The CG o f the ice

block can bc determined usin g the " meas ure inerti a" comm and in DYN A. Its globa l

coo rdinates are used to cre ate a nod e (Node 13800 3 in Figure 5-2 ) as the or igin of the

local coor dinate sys tem. Node 138004 and node 13800 5 are cre ated to define the x- and

y-axes o f the local coo rdinate sys tem. The z-axis is auto matica lly determined foll owin g

96
the right-h and rule. The local coo rdinate system is deli ned to share the same orientation

with the global one to make it convenient to defin e loads later. Note that a ll thre e newl y

created nodes are massless so they will not alTect the physics o f the model. The local

coor di nate sys tem is integrated into the ice block using the *CONST RA I ED_E XT RA_

NO DES_N O DE card which tics the three nodes to the rigid part of ice. Note that this card

only wo rks for rigid bodies and this is why a rigid part must be created first.

Step 3: Definin g restoring forces with user-de fi ned-cu rve-functions. Load defi nition s of

restorin g forces are listed in Tab le 5- 1. As mentioned earli er. functions DZ( 138003).

AX( 13800 3). and AY( 138003) are user-defined-curve -fun ctions reporti ng heav e. roll, and

pitch disp lacement s of the node 138003 (CG of the ice). They are multiplied with the

st iffness, which are computed using outputs from Rhinoce ros®. to defin e the restorin g

forces. Each restoring force ca n be applied either as a point load to the CG or distribut ed

eve nly amo ng a ll the nodes on the icc. These two approaches lead to very similar result s.

In this thes is. each restorin g forc e is applied as a point load for simplicity.

97
 FigureS-I: Creating a Rigid Part

Figure S-2: Creating the Local Coordinate System

98
 TahleS-I : Load Definition for Restoring Forces

Mo tio n Stiffne ss Loa d Definiti on s for Restorin g Forces

Hea ve 4.53 8.32 1 -4 .538. 32 1*DZ( 138 003)

Roll 23.63 1.4 12 -23 .63 1.412 * AX( 138003)

Pitch 6.66 1.882 -6.661.88 2*A Y( 138003)

Note: Node t38003 IS the CG of th e lee.

Ste p 4 : Defin ing damp ing. Dampin g for eac h indi vidu al bod y is defin ed using the

DAMI'IN G_I'ART_MA SS_ SET card in DYNA . Its a lgor ithm do cs not includ e mas s in

the cal cul ati on of dampin g co effici ents . For instance . the theor y of dynamics est imates

the cr itica l dampin g co effic ient as:

Equation S>!

w here k is the stiffness and 111 is the mass. DY NA uses a co nce pt of c ritica l damping

fac tor. w hich is ca lculate d as:

D
s
= 2w = 2 ~~= ! .:.
~ 111
Eqll at iollS -2

w he re w is osci llato ry freq ue ncy. No te that 111 is the sum of ac tual mass and the adde d

mass in th is thesis. Th e damping fac tor D for DY NA input is ca lculated as:

D = (D s EqllatiollS-3

wh ere ( is the damping rati o deri ved in Section 4.3. In thi s th esi s. the added ma ss

coe ffic ients es tima ted by DY NA a rc used to calculate the total ma ss and damping

coe ffic ients , T hei r va lues a re listed in Tabl e 4-10 .

99
5.1.3 Drag and Added Ma ss

Drag forces ca n be applied to the body usin g cu rve functions assoc iated with veloc ities

term s such as VX. VY. VZ. WX. WY. and WZ . They can be impleme nted in the sa me

wa y as presen ted in the prev ious sec tio n. The ir de tai led informat ion is avai lable in LSTC

(200 7a). These six curve-fu nctions arc tested as we ll and proved to be respo nsive . In thi s

thesis. it is assume d that the ship moves tow ard s the ice at a co nstant ve locity be for e the

impact takes plac e. The imp act lasts for a short peri od of tim e. and the ch ange o f veloc ity

is not significa nt. Includin g dra g will requir e thru st to be modeled as well. wh ich makes

the ana lys is unnece ssaril y complica ted . Therefore . drag is ignored .

DY NA also provid es users w ith six functio ns that report acc ele ratio ns in 6 DO F. They arc

ACCX . ACC Y. ACCZ . WDTX. WD TY . and WD TZ (LSTC 200 7a) . In ea ch OOF . the

added mass effec t ca n be theo ret ica lly modeled by de fining a force as:

F = -a . lim Eq ua t io n S-4

where a is the acce lera tio n of the CG of the body given by user-d efin ed -curve- function s

and lim is the added mass. and the minu s sign means the force is in the oppos ite directi on

o f the acce lera tio n.

Both Popo v ' s origin al mod el (Popov et al. 1967) and DDeP S suggest that the added ma ss

effect is ver y important in determining the ship-ice contact force . How ever , unfortunatel y.

the added ma ss effect cannot be includ ed in thi s thesis due to a bug asso ciated w ith the

acce lerat ion curve-functions in DY NA. When the acce leratio n curve - functions arc present

100
in the mod el. DYNA is not able to co nve rt externa l node numb ers to sequential interna l

nod e numb ers. Exte rna l nod es arc nod es added by users to de fine the CG of the bod y and

the local coo rdinate syste m (s uch as node 138003. nod e J 3800 4 and nod e J 38005 in

Sec tion 5. 1.2) . Intern al nodes arc the ones created by DY A whe n it ge nera tes the mesh .

Discu ssion s and effo rts have been made with other DYN A ' s users and DYN A' s tech nic al

suppo rt to so lve this issue (Kennedy. 20 12). A wo rkaround o f man uall y numberin g node

numbers has been sugges ted. but it still cannot report translation al accelerati on s and gives

d ifferent rotational acce lerations when so lving the same model using di ffere nt releases o f

DYN A. The author has been advised by DYNA's techn ical support that a new vers ion is

av ailab le and it ma y or may not have this issue resolv ed (Kennedy. perso na l

co mmunica tio n).

5.2 Mass Reduction Coefficient

The mass red uction coefficient need s to be addresse d before di scu ssin g the ship-icc

co ntac t force. It is one of the key param eters in determin ing the ship-icc co ntact forc e as

menti oned in Sec tion 2. 1.2. It was originally applied to the ship- icc co llision ana lysis by

Popov et al. ( 196 7). Its detailed derivation is availab le in Daley (20 00) and Kendrick et al.

(2000 b). This subsec tion investigate s ifDY NA and Pop ov' s original model have the

same estimation on the mass reductio n coefficie nt. Results from DYNA simulations arc

co mpared with calculations using the DDePS progr am.

101
5.2. 1 Simulatio n Set up

In the so lution using DYNA. the dim ensions of the ship and ice are the sa me as in Section

3. 1.1. The ship and ice are modeled as rigid since the deformation is not the conc ern. The

material densiti es are the same as those in Table 3-2 and Ta ble 3-3. Element choice s are

the same as in Ta ble 3-4. Both the ship and ice are modeled as free bod ies. The loading

co nditio ns in DDePS and in DYNA are the same to ensure compa rable result s. Loads are

applied to ship and ice as if there was a collision . A quick contact simulation deter mines

one node on the ship bow and anot her node on the ice as the co ntac t location . A local

coordin ate syste m is co nstructed at the poss ible co ntac t po int on the ship bow. Its x-y

plane lies in the ship bow plate. and its z-direction co incide s with the norm al directio n of

the bow plate (sec Figure 5-3). Thi s local coord inate system is the reference for definin g

the loadin g directi on . Then a force is applied on the ship bow in the - z-direction of the

local coordinat e system. Another force with the same magnitud e is applied on the ice

(node 4982 4 in Figure 5-4) in the z-di rection of the local coordin ate system in a separa te

analysis . The magnitud e of the for ce equated the estimation by the DDePS tor the same

co llision case. Restoring for ces and added mass terms are not incl uded in the analys is so

it is a "d ry" collision.

102
Figure 5-3: Normal Direction of the Contact Surface

Figure 5-4: Contact Location on the Ice

103
5.2.2 Result s

Acce lerat ions in the X-, y-, and z-direc tio ns o f the two bodi es are recorded in DYNA' s

outputs. They are map ped back to the normal direction defin ed . The mass red uctio n

coefficient is ca lculated as:

In
CO = - F-
E<luatioI\ S-S
a n or m al

where In is the mass o f the body (the ship or ice), a n orm al is the acce lera tion of the

co ntac t poi nt in the nor ma l di rectio n of the co ntac t sur face, and F is the contac t fo rce .

DDe rS and DYNA have di fferent estimations on radii o f gyra tion as show n in Ta ble 5-2,

hence diffe re nt mass mom ent s of inertia. For the radii o f gy ration of the shi p, the

discrepa ncy lies in the di fferent mass distributi on s of the geo metric mod el. In the DYNA

so lution , the ship is a she ll, while in the DDerS so lution, the ship is co nside red as a so lid

body. For the radii of gyration of the ice, the two so lutions give very similar estima tions

ofr, and ry . However, they indi cate very di fferent va lues for rz since DDeP S' s

esti ma tion is base d on empirica l va lues.

These two approac hes also pred ict di ffer ent added mass coefficients as discu ssed in

Sec tion 4.4. Var ious co mbinat ions o f added mass coeffic ients and radii of gyra tions are

input into DDeP S to ca lculate the mass reductio n coeffic ient. Result s are presen ted in

Table 5-3 along w ith DYNA ' s estima tions. Note that for the same co llision case, the

co ntac t for ce wi ll increase as the mass reduction coeffic ient decreases. DDcI' S CD and the

DYNA so lutio n sha re the same co nditio n, but DDcI' S CD gives higher estimations fo r the

104
CO. Co mpa ring DD eP S @ and DOerS CD indic ates that added mass decreases the mass

redu ct ion coefficient, which will con sequently increase the ef fective ma ss and co ntact

fo rce. A co mpariso n of DOerS CD and DDe!' S ® sugges ts that rad ii o f gyra tio n of the

bod ies play an imp ortant role in determin ing the Co. Radii o f gyra tio n given by DYNA

and DDcI' S are diffe rent. To minimi ze this differ ence' s infl uence, ca lculation of the

co ntact force in DDcI' S will adopt the radii of gyra tion give n by DY NA when comp arin g

these two approac hes. Co mpa ring DDe!' S ® and DDcI' S ® a lso shows that ad ded mass

terms sig nifica ntly redu ce the Co' and it impl ies that DOerS ® wi ll pred ict a mu ch

higher co ntac t for ce than DDeP S ® . The OOeP S @ gives sma ller va lues co mpared to

DDcI' S @ because the origina l DOerS model gives higher estima tions for the adde d

mass coefficients as discu ssed in Sec tion 4.4.

Table 5-2: Radii of Gyration Estima ted by I)VNA and [)l) eI' S

Ship Ice

DYNA DDeP S DY NA DOerS

rA m) 5.1 3.8 4.5 4.6

ry( m) 18.1 15.1 2.7 2.2

rz(m ) 18.4 16.5 5.0 2. 1

105
 Table 5-3: Mass Reduction Coefficients Co
So lution Descript ion Ship Ice

DY NA No added mass term s 2.73 2.27

DDeP SCD Us ing DYN A' s radii of gyration. but o added mass term s 3.5 1 3.22

DDeP S@ Using DY A's radii of gyration & add ed mass term s 3.08 2.69

DDeP S® Using original DDeP S' radii of gy ratio n. NO adde d ma ss terms 4. 38 4.86

DDePS ® Usi ng original DDePS ' radii of gyration & added mass term s 2.67 2.56

DDePS @ Using DYNA' s radii of gyration & DD ePS ' added mass term s 2. 16 1.84

It is apparent that DDeI>S and DYNA have diff erent es timations lor the added mass

red uction e ffec t. There are several reason s which may co ntribute to this discrep ancy. First

o f al l, the cross mom ent s or inertia (l xy . I y z and I zx ) are ignored in DDeP S but includ ed in

DY A. Besides. DDeI>S assumes that mom ent arms of the body do not ch ange du ring the

co ntac t. while this is not true in the DY NA so lutio n. Moreover. in DDeI>S and Popo v' s

model. it is ass umed that the co llision occ urs in an instant tim e period . and the bod y

rotates aro und its eG. In the DYNA model. althoug h the impulse load is applied. the

e ffect of force lasts slightly longer . and the rotational cent er o f the bod y ch ange s ove r

tim e. The difference in the mass reduct ion co effic ient should be taken into account when

comparing the cont act force s estima ted usin g DDeP S and DYNA .

106
5.3 Simulation Setup

This sectio n prese nts the DYNA model for eva luating the global ship-ice contact force .

Result s are discussed in Sectio n 5.4.

5.3.1 Geometr ic Model

The geo metric model is the same as the one in Section 3. 1.1.

5.3.2 Material Mod els

The material model for the ship is the same as the one in Section 3.1.2 (sec Table 3-2 ).

The mod ified cru shab le foa m ice model C from Section 3.2.4 is utili zed to model the ice

block . Its material properties are ava ilable in Ta ble 3-8 and Table 3- 11.

5.3.3 Element Choices

Element choice s are same as those in Section 3.1.3 (see Table 3-4).

5.3A Bound ar y and initial conditions

In all simu latio ns. the ship is modeled as a free body. For the colli sion wit h a fi nite ice

mass. the ice is treated as a free body as well. For the co llision invo lving an infini te ice.

the ice is fix ed on two sides as shown in Figure 3-6 to imitate an infi nite mass of the ice.

In each cas e. the ship starts moving fo rward at a dif ferent initia l speed. The ice is crushed

and deforming as the contact proceeds. The ship-ice co ntact ceas es when they bounce off

each other. The ship' s init ial speed is 3m/s in the mesh co nvergence study.

107
5.3.5 Loading Conditions and Damping

The model does not inc lude gravity or water. hence no net buoyancy. This co uld be

interpre ted as gravity eq ua ls buoya ncy. i.e.. each bod y is initially " floa ting" at the neut ral

buoyanc y position . Restorin g forces are app lied foll owing the instructi on in Sectio n 5. 1.2.

Si nce there are no heave. roll and pitch displaceme nts prior to the co llision. use r-de lined-

curve-func tions are applying zero forc es to the floatin g bodies. Once the co llis ion takes

place. the ship-ice conta ct for ce begin s to " push" the ship and ice to mo ve in all 6 DOF .

Heave, roll and pitch di splacem ent s then result in non -zero restorin g for ces. Drag and

forces assoc iated with the add ed mass are not included in the ana lysis as discu ssed

previou sly. Load defin ition s arc summa rize d in Ta ble 5-4. Th ere is no load definit ion on

the in finite ice since it will not have any globa l mot ion. Dampin g is applied . Th e added

mass term s are includ ed in the ca lculation of dampin g coe ffic ients.

TableS-.t: Load Definition s on the S hip and lee

Motio n Load Defin ition on the Ship Load Definition on the Ice

Heave -6.828.21 9*DZ(904 333) -4.538.321 *DZ(138 003)

Roll -35.247. 518*AX(90 4333) -23.63 1.4 12* AX( 138003)

Pitch -1.813.659.196 *A Y(904 333) -6.661.88 2*AY( 13800 3)

Note: Node 90.t33 IS the C G of the sh ip. Node 138003 IS the CG of the Ice.

5.3.6 Mesh Convergence

Mesh co nve rge nce studies arc condu cted for both the ship glanc ing w ith finite and

infinit e ice cases . Time histor ies o f the co ntact for ces from simulations usin g va rious

108
e1cmcnt sizes arc compared in Figure 5-5 and Figure 5-6. Note that thc contact starts at a

differe nt time instant as the mesh density changes. Measuring thc distance between thc

ship and ice suggests that the contact should initiates at about 1.5 seconds. Simulations

using elements no larger than 0.24m give similar results. Reducing the clement size from

0. 12m to 0.1m barely changes the results. However. analy ses using 0.1m e1cmcnt take a

significantly longer time to solve as shown in Table 5-5. Note that 0. 12m is also an

appropriate size lor meshing the ship structure assuming non-unifo rm mesh is used.

Therefore. 0.12m is co nsidere d as the optimal cleme nt size for modeling the ship-icc

contact force.

Ta ble 5-5: Co mputa tion Time of Simula tions using Vari ou s Element Sizes

Element Size 0.35m 0.24m 0.2m 0.15m 0.12m O. lm

Finite Icc Case 40sec 1.511lin 411lin 2111lin I Ah r 6hr

Infinite lee Case 37sec 2m i n 511lin 27m i n 1.8hr 7.8min

No. of Compute Nodes 4 4 4 4 4 4

Note. All simula tions are solved on STel) S2

'
Cluster (see Appendix A)

109
 0.0 L...-_~.......::;c~-------,---------~",--_ _--"

1.4 1.6
Time(s)

Element Size: - 0.35111 - 0.24111 - 0.2111 - 0. 15111

Fig ure 5-5: Me sh Co nve rg ence - Ship G la ncing w it h Finite Ice

z;;;:
~ 3 .0
tZ
~ 2.0 f - - - - - --F/.W-- - - - - - - - ---\JIl
\----

-5
1.0 f - - - - --// I - - - - - - -- - - - - ''''"- --

0.0 L...-""""""';&L- ------:SlIiliil......._

1.4
T ime(s)

Ele me nt Size : - 0.35111 - 0.24111

Fi~ lIre 5-6: Mes h Co nvergence - S hip G l a nci n~ with In fini te lee

lI D
5.4 Ship-Icc Contact Force

Ship-ice gla nci ng scenarios at vario us ship speeds are s imu lated using the nu meri ca l

model pre sente d in the previous sectio n. Result s are com pared with the calc ula tion us ing

the DDd)S program . For the finite ice case. the defor ma tion in the ice is sma ll a nd the

sha pe of the contact area is s im ila r to Sce nario 2C in D De PS. For the infi nite ice case. the

de fo rmation is mu ch lar ger and the shape of the co ntac t area is s im ilar to Sce na rio 2B in

DD eP S. As d iscu ssed in Sect ion 5.2.2. DY N A a nd DDd) S give d iffer ent va lues o f the

rad ii of gyra tion of a bod y. To minimi ze thi s dif fer enc e' s influence o n the co ntac t for ce.

a ll ca lcu latio ns o f the co ntact fo rce using DOerS ado pt the rad ii of gy ra tion give n by

DY NA.

5.4.1 " [)'1'''CollisionCascs

The term "Dr y" co llisio n mea ns an y effect related to water is excluded in the ana lys is.

Restoring force s modeled us ing user -defined-curve-functions and damping arc removed

from the DY A ana lysis. In the DDd)S ca lculation. a ll the added ma ss va lues arc set at

ze ro. As di scu ssed in the prev ious sectio n. the mass red uction coeff ic ient Co m ust be

ta ken into co nsi deratio n w he n eva luati ng the co ntact force . T he o rigi na l Co derived by

Popov ct a l. ( 1967) a nd the o nes esti mate d by DYNA a re inp ut into DDd)S to eva lua te

the con tac t fo rce. Co ntac t for ces of va rious "dr y" co llis io ns arc co m pa red in Ta ble 5-6

(fi nite ice) and Ta ble 5-7 (infi nite ice). No te that:

• DD eP S CD so lution d irectl y applies the Co g ive n by DYNA rath e r than

fo llow ing th e derivat ion by Pop ov et a l. ( 1967) .

111
 • DDeP S @ so lution ca lculates the Co follow ing Popov ' s or igina l derivation

wi th the radii of gyrat ion give n by DYNA.

• DDcP S CD and the DY NA so lutions arc based on the sa me mass reduction

coeffic ients. DDeP S @ has larger mass redu ct ion coeffici ent s.

Tabl e 5-6: Ma ximum Contact Forc e (Finite Ice, Dry Co llision)

Ship Speed (mjs) 2 3 4 5

DY NA (MN) 1.26 1.86 2.44 3.04

DDeP S CD( MN) 1.39 2.03 2.68 3.3 1

DDcP S @ (MN ) 1.19 1.74 2.28 2.82

Tab le 5-7: Maximum Conta ct Force ( Infin ite Ice, Dry Collision)

Ship Speed (mjs) 2 3 4 5

DY NA (MN ) 2.62 3.83 5.22 6.71

DDePS CD ( MN) 2.37 3.78 5.25 6.79

DDeP S @ (MN ) 2.09 3.32 4.62 5.98

It is clear that the mass rcduetion coe fficie nt has a signific ant influence on the co ntac t

force. DYN A and DDeP S CD have the same mass redu ction coe fficients and show very

goo d agree me nt. Espec ially in the case of the infinite ice (large deform ation in the icc),

the two so lutio ns give ident ical result s. In the case of the finit e icc (sma ll deformati on in
112
the ice). the differ ence in the results are slightly larger due to the ice material mod el. The

pressu re-area curve o f the ice materi al model in DDeI)S strictly foll ow s the P = A- o. 1

rela tions hip . In the DYNA solution. the pressu re-area rela tio nship of the ice mod el ag rees

ver y wel l with the curve P = A- o.t for large deform ation s. but does not perfe ctly lit it

when the de formation is sma ll as discusse d in Sec tio n 4.4 . Co mpared to its math em at ical

approx ima tion. the ice model actua lly exe rts higher pressu res over small co ntac t areas .

Overa ll. these two so lutions give simi lar estimations o f the ship-ice co ntac t for ce.

5.4.2 "We t" Co llision Cases

"W et" co llision mean s that restorin g for ces and dampin g are included in the DY NA

so lution. while added mass is co ns idered in the DOd'S so lution. There is no restorin g

force or damp ing in the DDeI)S so lution. Result s are listed in Ta ble 5-8 (fi nite ice) and

Table 5-9 (infin ite ice). Note that:

• Only the ship's Co matters here since the ice is modeled as infi nitly large.

• In all the DDeP S so lutions. the ca lculation o f Co foll ow s Pop ov' s

deriva tion (Popov et.a l 1967 ) with radii of gyra tio n give n by DYNA.

• The so lutio n DDeI' S CD uses Popov ' s origina l adde d mass terms. It has the
sma llest Co amo ng the thre e so lutions.

• The so lution DDd)S @ uses DY NA ' s added ma ss term s (see Table 4-10 ).

It has the largest Co' but its va lue is only slightly larger than the Co given

in the DYNA so lution.

113
 Tabl e 5-8: Maximum C onta ct Force (Finite Icc, Wet C ollision)

Ship Spee d (mjs ) 2 3 4 5

DYNA ( M N) 1.27 1.88 2.47 3.06

DDeP S CD ( M N ) 1.53 2.25 2.96 3.66

DDerS @ ( M N) 1.28 1.89 2.38 2.95

Table 5-9 : Max imum Contact Force (Infinit e Icc, Wet C ollision)

Ship Speed (mj s) 2 3 4 5

DYNA ( M N) 2.65 3.89 5.30 6.89

DDeI) S CD ( M N) 2.75 4.38 6.10 7.88

DDeI)S @ (MN ) 2.25 3.58 4.99 6.45

In the DYNA mod el, added mass term s are not includ ed , but in the DDcI)S so lutio n,

added mass term s are involved in co mputing the mass redu ct ion coe ffic ient and

co nseq uently have a di rect influ ence on the contact fo rce. These three so lutio ns sho uld

no t give simila r es tima tions fo r the co ntac t force becau se o f the different mass redu cti on

coeffic ients. The ag ree me nt o f DDeI)S CD and DY NA is du e to the appro xima tio n in the

ice material mod el as di scu ssed in the previou s sec tion. DDeP S @ has a slightly larger Co

than the DYNA so lutio n and thus predi cts a similar but smaller co ntac t fo rce in most

cases. DDerS CD has the sma lles t mass reduct ion coeffic ient and there fo re gives the

largest co ntact force . Th e DDcI) S @ so lutio n featu res the largest mass reductio n

114
coefficients an d suggests the sma lles t co ntact for ces. The adde d mass is ano ther factor

that co ntri butes to the di fferenc e in the result s.

504.3"Dry" vs. "'Vet"

In the DY A mod el. the ship and ice boun ce off eac h other imm ediately after the co ntac t

in the " dry" co llisio n, while restor ing forces and dampin g ca uses the two bodi es to stay in

co ntac t in the "wee co llision. Therefore. the contact for ce in a "we t co llision" is expe cted

to be larger and last lon ger than that in a "dr y" coll ision . It is verified by co mparing the

max imum cont act force s in Table 5-10 and Tabl e 5- 11. The difference in the infinit e ice

case is mu ch more obv ious than that of the finit e ice case. How ever . thi s increase in the

co ntac t force is not signifi ca nt co mpared to its maximum va lue. This mean s that restoring

forces are not imp ort ant in de termining the co ntac t force of the first imp act. This agrees

wit h Popov 's orig ina l ass umptio n.

In the DDeI)S so lutio n. the ship and ice are a lways ass umed to bo unce of f eac h othe r in

both the "dry" and " wet" co llis ions. The differenc e is that, in thc "wet" co llis ion, adde d

mass term s are dir ectl y applied to the ca lculation of the mass redu ction coe ffi cient and

sig nifica ntly red uces the mass reduct ion coe fficie nt. i.e.. increase the effec tive mass. Thi s

con sequ entl y incr eases the co ntact forc e. Th ere for e. the increa se in the contact force is

more signifi cant as the conditi on changes from "w et" to "dr y" in the DDeP S so lution.

115
The change of the co ntact forc e from the "dry" to "wet" collision is dem onstrated in

Table 5- 10 and Tab le 5- 11. In the DDerS solutio n. results are from the analyses where

radii of gyra tion. and added mass coefficients give n by DYNA are used to the ca lculate

the eo ntact force. This practice is done for minimizing the difference in the inputs of the

two solutions.

Ta ble 5-lll: Dr y vs, Wet - Max imum C onta ct Fo rce , Finite Icc

Ship Speed DYNA ( MN) DDePS (MN)

(m js) Dry Wet Increase Dry Wet Increase

2 1.26 1.27 0.8% 1.19 1.28 7.6%

3 l.8 6 1.88 1.1% 1.74 1.89 8.6%

4 2.44 2.4 7 1.2% 2.28 2.3 8 4.4 %

5 3.04 3.06 0.7% 2.82 2.95 4.6%

Ta b le 5- 11: Dry '·S. We t - Ma xim um C onta ct Force , In fin ite Icc

Ship Speed DYNA (MN) DDeP S (MN)

(m js) Dry Wet Increase Dry Wet Increase

2 2.62 2.65 1.2% 2.09 2.25 7.7%

3 3.83 3.89 1.6% 3.32 3.58 7.8%

4 5.22 5.30 1.5% 4.62 4.99 8.0%

5 6.7 1 6.89 2.7% 5.9 8 6.45 7.9%

116
In the DY A solution. the difference in the time histories of the contact forces is much

more dramatic. Taking thc case where the ship speed is Smfs for example, time-histories

oft hc contact forces are compared in Figure 5-7 (finite ice) and Table 5- 11 (infi nite ice).

Besides a higher maximum contact force. the "wet" collision also predicts a second

contact and even a third contact. Note that the DOer S program is only capable of

estimating the first impact.

In conclusion. restoring forces modeled by user-defined-curve-functions have a small

influence on the contact force of the first impact. but are very important in simulating thc

bodies' motions after the first impact as well as the second impact. This suggests that the

contact force is dictated by the contact speed. mass. and the material strength of the

contacting bodies. Another important aspect is that the solution using user-defined -curve-

functions only takes about one hour to solve. which is significa ntly more efficient than

the ALE method.

117
0.0 --- ~ __.~~___JI ___..........
0.8
Tim e( s)

- Wet · · · · Dry

Figure 5-7: Co mpa r ison of the Co ntact Force - Ilr y vs. Wet (F inite Ice)

O l.-.J:--------~-----~~~ ....... ...-.-.-..

0.8
Tim e( s)

- Wet · · · · Dry

Figure 5-8: Compariso n of th e Contac t Forc e - Il ry vs Wet ( Infinite Icc)

118
5.5 Summary

This chapter explains the impl em entati on o f user-de fined-cur ve-functi on s to model

restorin g fo rces. Theo retica lly. forc es related to added mass and drag ca n be mod eled in

the sa me way . How ever. forces related to added ma ss are not includ ed due to a probl em

in DYNA. Drag is ignored becau se it is not necessary in thi s ana lysis. T his practice

grea tly reduces computation cos t compared to the AL E meth od . The mass redu ction

coe ffic ients estimated by DYNA and DDeP S are co mpared and shown to be important in

det ermining the contaet forc e.

Wh en the same ma ss redu ction coeffici ent s are applied. DY NA and DDe/>S give identical

es tima tio ns fo r the cont act force. This agreement is more obvious in the infinit e ice case

than in the finit e ice case . For the finite ice cases. the present DYNA model gives higher

estima tions for the cont act for ces. This is becau se the ice model in the DY A so lution

does not strict ly follow the form P = A- O. l • As discu ssed in Section 4.4 . co mpared to the

mathem atical approxima tio n. the ice mod el actu all y exe rts higher pressur es ove r sma ll

co ntac t areas . Ove rall. these two approa ches have dem on strated a goo d agree me nt.

Restorin g forc es mod eled by user-defined-curve-functions have a sma ll influ ence on the

maximum contact force of the first impact. but are very important in simulating the

motion s of the ship and ice as well as the second imp act. Thi s suggests that the contact

force is dictat ed by the contact speed. ma ss. and the materi al strength of the cont actin g

bodi es. Anothe r imp ortant aspect is that the solution usin g user-defined-curve-function s

119
only takes about one hour to solve. whi ch is significa ntly more e ffic ient than the AL E

meth od .

120
 Chapter 6 Ship's Structural Response

The previous chapters have addre ssed modelin g the ice material. the wa ter dom ain. and

the ship-ice contac t model. This chapter proposes the final model by co mbining all the

previous wo rk togeth er and incorporatin g the ship' s local structural respon se. However.

thi s cha pte r does not intend to co nduct a co mprehe nsive or so phisticated structural

ana lysis. Its purp ose is to give a simple demon stration of how to use the final model to

perform structural ana lysis. It serves mor e as a ge nera l guide line rath er than valid atin g

any resu lts. Reader s can follow the guida nce give n in th is chapter to build more elabor ate

model s for more comprehen sive and detai led structura l analysis.

6. 1 Ship Str uct u ra l Design

Th e ship hull used in previou s chapters is ice-stre ngthened with intern al struc tures in

accor da nce with the URI and DNV specifi cations. Si nce the ship icc co ntac t will occur at

the ship bow . the structura l design is only co nducted lo r the bow area. The bow region .

inclu din g its structural mem bers. will be modeled using elastic-plastic material. wh ile the

rest o f the ship is mod eled as rigid . The ship struc tura l design present ed in this chapte r is

not for ship-building purp oses. Thi s practic e aims to give a simple structural model to

proceed with the analysis of the ship structura l response. If the ship' s structural model is

avail ab le. it can be direct ly inp ut into DYNA to generat e the geometric mode l and the

me sh.

121
6. 1.1 Main Frames

The design of the main web frames complies with Section 12 of thc URI (lA CS 20 10).

which utilizes the plastic strength of thc structural mcmbcrs. Derivation of the

formulations for the framing design in the URI can bc found in Daley (2002). and

Kendrick et al. (2000b). The URI divides the ship hull into several regions, A different

level of icc load is expected in cach region. The hull is divided into four regions in the

longitudinal direction: bow. bow intermediate. midbody. and stern. The bow intermediate,

midbody. and stern arc thcn vertically divided into sub-regions: icebclt, lower, and

bottom. The extent of each region is shown in Table 6- 1 and Figure 6-1.

T ab le 6- 1: Hull Area Exte nts (lA CS 20 10)

Region/Area Notation

Bow B

Bow Intermediate Icebclt B1i

Bow Intermediate Lower B11

Bow Intermediate Bottom BIb

Midbody Iccbclt Mi

Midbody Low er Ml

Midbody Bottom Mb

Stcrn Iccbclt Si

Stern Lower Sl

Stern Bottom Sb

122
 F", PC1 2 3&1 . - 1 5 m
rOlPCS 6 7 J ·1 J nl.
~ltr · l · rnea5.Jre<Jatare "lO 01 0Cwr~ tOn

/
/
/
/

vn Ar<J'. ' O""'J_ ~ ' '\

Vn Angl.'lJd"'l" '"
lJ-dI~ld' K.e 1rV'l 1 :h t AP 1) , \clamIUI
nat ereactnaiuw,

UIWL
LlWL

Figure 6-1: Hull Area Extents (lACS 2010)

The ship is chosen as a Polar Class-4 ship. A ll nece ssar y geo me tric information for the

ship's structura l design is listed in Tab le 3- 1. The main frames are tran sversely arrange d

on the ship hull. T he ir dime nsio ns are listed in Ta ble 6-2 in Sec tio n 6.1.5. A snapshot of

the 3 D mod e l (in Rhin oceros® ) o f the main fram es is illustrated in Figure 6-2 . Th e gaps

between the main web fram es are occ upied by deep web frames and bulkh ead s.

123
 Figure 6-2: Main Fr am es and th e Hull

6.1.2 Load Ca r rying Stringers

The UR I doe s not address the scan tling requ irem ent s on load carry ing stringe rs. The ir

sca ntlings are determin ed follow ing the DNV 's specifi cations tit led Ships for Nav iga tion

in Ice (20 I I ). This lim its the stre ngt h of load car ryi ng stri nge rs to the clastic range . Load-

ca rrying stringers are or iented lon gitudin all y on the ship hull. Detailed sca ntlings of the

load carryi ng stri nge rs are listed in Ta ble 6-2 in Section 6. 1.5. The 3D mod els (in

Rh inoceros® ) of the load carr yi ng stringer s are illustrate d in Figure 6-3 .

124
 Figu rc6-3: Lna d Ca r ry ing St r ingc rsa nd th e Hull

6.1.3 Deep We b Frames

The URI does not addre ss the scantling requi rements for deep web frames. Their

scantlings arc determined followi ng the D V' s speci fications titled Ships lo r avigation

in Icc (20 11) as wel l. This a lso limits the strength of deep web frames to the elastic range.

Deep web frames arc orie nted transversely on the ship hull. Detailed sca ntlings o f the

deep web frames arc listed in Ta ble 6-2 in Section 6.1.5. The 3D models (in Rhin oceros® )

of the deep we b frames arc illustrated in Figure 6-4.

125
 Fi ~ lIrc6-4 : Deep Web Fra mcsa nd th e Hull

6. 1.4 Bulkheads

The design o f the bulk head s doe s not follow any spec ification s. A bulkh ead is placed on

the hull every 10 main frame s. The thickness of the bulkhead is 30mm. Each bulkh ead is

stiffened with flat bars that arc 300 m m in heigh t and 30m m in thickn ess. The 3D model

(in Rhinoceros® ) of the bu lkhead s is illustrated in f igure 6-5.

126
 F i~ u re 6-S: Bulkhead s inc lu di ng St iffe ne rs, a nd th e Hull

6.1.5 S ummary

T he scantlings o f the structural memb ers on the bow are summarized in Ta ble 6-2. The

3D model (in DYNA) of the bow region with inte rna l structural memb ers is illustrated in

Figure 6-6. It mu st be emphasize d that the structural de sig n is no t for ship buil ding

purp oses. This practice only aims to genera te a simple geo me tric mod el for the structura l

analys is.

127
 T a ble 6-2: Sca nt lings of St r uct ura l Me m bers in th e How Region

Loa d Ca rrying Deep We b

Item Mai n Frame
Stri nger Frame

Or ientation Tra nsverse Longi tud ina l T ransverse

Spacin g. m 0.5 2 2

Spa n. m 2 2 2

Il ull Pla te T hickn ess.mm 30 30 30

We b He igh t, mm 300 600 700

Web T hick ness. mm 22 30 30

Flange W idth . mm 0 0 120

Flange T hick nes s. mm 0 0 30

128
 Figure 6-6: Bow Region with Internal Structural Me mbe rs

6.2 Simulatio n Set up

A simulation of a ship-ice glancing scenario from Chapter 5 is repeated with the ship

modeled using a combination of elastic-plastic and rigid materials to include its structural

response. A simple evaluation of the local structural response of the ship bow is given as

an example of how to use the proposed model.

6.2.1 Geometric Model

The ship hull and the ice block are the same as in previous chapters . Their detailed

information is available in Table 3-1. The ship hull is strengthened with structural

members as discussed in the previous section.

129
6.2.2 Mate r ial Mode l

The ship is mod eled using a comb inati on o f rigid and non -rigid mater ials to sa ve

co mputation cost. Th e bow region and its intern al structura l memb ers arc modeled usin g

the clastic-plastic materi al. Its prop ert ies are shown in Ta ble 6-3. Th e rest of the ship is

treated as rigid . Properti es of the rigid materi al are the same as in Section 3. 1.2 (sec Ta ble

3-2) .

Ta ble 6-3 : Mal eri all>a ram eler s for lh e NolI-l{igill Part of the Ship

Ca rd ID MA 'U> IEC EW ISE_L1N EA R_ PLAST IC ITY (MAT _0 24)

Densit y Youn g' s Modu lus I Poisso n's Ratio I Yield Stres s I Tangent Mo dulus
7850kg/m 3 200GPa 03 I 350MPa I tGPa
1 .

The ice block is mod eled using a combination of rigid and non-r igid material s as well.

The modifi ed crus ha ble foam ice model C from Section 3.2.4 is util ized to model the

deform abl e part . Its materi al propertie s are availabl e in Ta ble 3-8 and Table 3- 11.

Materi al paramet ers o f the rigid part are listed in Ta ble 6-4. The 3D mod el is shown in

Figure 6-7.

Ta ble 6-4: Mat e rial Par amet er s of t he Rigill Part of the lee

Ca rd ID MAT _R IGID (MAT _0 20)

Materia l Ty pe Dens ity I You ng ' s Mo dulus I Poisso n's Ratio

Rigid 900kg / m315GPa 0
1 .03

130
 Figure 6-7: Rigid and Non-Rigid Ship and Ice

As shown in Figure 6-7, the non-rigid portions of the ship and ice are not small compared

to the overall dimensions of the ship and ice. In a more elaborate analysis, users can

model larger portions of the geometric models as rigid to save more computation cost.

6.2.3 Element Choices

Element choices are shown in Table 6-5. The Belytschko- Tsay formulation is

recommended for structural analysis (Quinton 2009) . An average element size of 0.15m

is used for the mesh without conducting a mesh convergence study . This is because the

present chapter only intends to give a simple demonstration of structural analysis rather

than giving an accurate answer. In a more elaborate analysis, not only the element size

should be carefully chosen , but also a mesh with non-uniformly sized elements should be

used. Areas far away from the possible contact region can treated with very coarse mesh

and the contact region on the ship and ice should be meshed with local refinements.

131
 Table 6-5: Element C hoices

Part Element Type Formulation Option Ambient Type

Ship Shell 2 (Belytschko-Tsay ) N/A

Ice Solid I (Default) 0

6.2.4 Boundary and initial conditions

This is a simulation ofa glancing scenario between the ship and an infinite icc. Therefore

the ship is modeled as a free body. The ice is fixed on two sides as shown in Figure 3-6 to

mimic an infinite mass. The ship starts moving forward at an initial speed of4rnjs . The

icc is crushed and deform s as the contact proceeds. The ship's bow region displays an

clastic-plastic structural response as well. I'he ship-icc contact ceases when the two

bodies bounce off each other.

6.2.5 Loading Conditions and Damping

The model does not include gravity or water: hence there is no net buoyancy. This could

also be interpreted as gravity equals buoyancy, i.e., each body is initially " floating" at the

neutral buoyaney position. Restoring forces are applied following the method discussed in

Chapter 5. Extra mass is assigned uniforml y to all the nodes on the ship so that it weighs

the same as in previous chapters. However, the CG of the ship is different from that in the

previous chapters due to a different mass distr ibution. This also changes the mass

moments of inertia of the ship, and consequently changes the load definiti on. Ih e updated

load definitions are summarized in Table 6-6.

132
 Table 6-6: Load Definition 011 the Ship

Motion Load Definition

Heave -6.828 .2 19*DZ(90 4333)

Roll -35.247 .5 18*AX(904333)

Pitch - 1.8l 3.659.l 96 *A Y(904 333)

6.3 Ship Structural Respon se

Th e model is solved using the STe PS2 clu ster with 4 compute nodes (32 co res) . The

computation tim e is about 36 hour s. which is reason abl y low con sid ering the am ount o f

plot s generated in the output. This co mputation cost can be significa ntly reduced by

modelin g a larger port ion of the model as rigid . usin g a non-uni form mesh . and requ est ing

fewer plots in the output. This section gives a short and simple discussion of the ship' s

structural response in terms o f the co ntact load. Von Mises stress . and the pressu re-

deflecti on cur ve.

6.3.1 C ontact Force and Pre ssure

T he time histo ry o f the res ulta nt ship-ice contact force is illustrated in Figure 6-8 . The red

line is the co ntact for ce o f the sa me co llision simulat ion exce pt that the ship is mod eled as

rigid . The value o f the contact force signifi cantly decreases when model ing the ship as

deform abl e rathe r than rigid. It is because the de forma tion in the ship structure abso rbs a

large amo unt of energ y. The time history o f the estimated averag e co ntac t pressure is

shown in Figure 6-9 . It sugges ts that average pressur e is not high eno ugh to ca use yie ld .

133
a I",..-_ Q: -----"""~---

1.0
Timc(s)

- Non-Rigid Ship - Rigid Ship

Fi!:urc 6-8: Co mpa riso n of the Co ntact Force : Ri!:id Ship vs Non- Ri!:id Shi p

T imc(s)

Fi!:lIrc 6-9: T imc Histor y of t hc Ave rage Pressur e

134
6.3.2 Von Mises St ress

Vo n Mises stress is one of the most imp ort ant criteria in structura l engi neer ing analysis .

In th is simulatio n. the ship-ice contact force is a moving load as the ice moves a long the

ship hull. The press ure di stribut ion on the ship bow is characterized wit h high pressu re

zo nes w ithin lower pressur e zo nes. wh ich cau ses so me elem ent s develop much higher

e ffec tive stress than their neighb orin g elem ent s. Thi s phenom enon is shown in Figu re

6- 10 and Figure 6- 11. As menti on ed ea rlier. the average cont act pressure is not high

enou gh to cau se structura l mem bers to yield. Thi s is confirmed by the resu lt tha t most

memb ers show a pure clastic respo nse as shown in Figure 6- 11. However. the high

co nce ntra tion of pre ssure still cau ses a co up le of sec tions on the main frame s to yield and

move to the plastic range as shown in Figur e 6- 10. The ana lysis o f the structura l respon se

fo cu ses on the main fram e memb ers with elem ent s that displ ay an ela stic-p lastic behavior.

Figure 6- 12 illu strat es the time history of the Von Mises stress of the elem ent that is

und er inves tiga tio n. T he sa me elem ent' s effective plastic stra in is shown in Figure 6- 13.

It is appare nt that the plastic stra in starts to devel op as the ef fective stress exceeds the

spec ifie d yield stress 01'355 MPA . Th is shows that the plastic ca pacity o f the main fram e

memb ers have been utilized as designed .

135
 Figure 6-10: A Typical Von Mises Stress Distribution on the Main Frame

Figure 6-11: A Typical Von Mises Stress Distribution on the Deep Web Frame

136
0 ..........- -"'--'- - - - -'------- - -----'----- - - -'--- - ----'
1.0
Time(s)

Figurc6-12 : Time Histor y ofth c Effective St ress

Effective Plastic Strain

(
/
/
I
Tim c(s)

Figur c6-13: T imcHist or y ofthc Effective Pla stic St ra in

137
6.3.3 Pre ssure-Deflection C urve

The last task is analyzi ng the pressur e-del1eet ion re latio nship of the ma in frame memb ers.

This ana lysis is fai rly easy to conduc t for the static FEA whe re the structure is fixe d on

the boundary. and the ship-icc con tac t is simply replaced by a po int load or a patch load

as the ice load. This setup implies that the del1ection of the structure is esse ntia lly the

disp lace me nt of the nod e (or nodes) on the mesh . whic h is ava ilable in DYNA output.

The va lue o f the load/pressur e that causes the del1ecti on ca n be acc urate ly determin ed

since it is dir ectl y defin ed by the user.

In thi s simulation. the ship-icc co ntact for ce is a movin g load . and the ship is free to mo ve

in 6 DO F. This makes it imp ossible to acc urately analyze the pressu re-defe cti on curve of

the structure. As ment ion ed earlier. the average co ntac t pressu re is not high enough to

cau se yield. there fo re the analysis of the pres sure-de llec tio n re lationship focuses on the

high pre ssure zo ne where cle me nts show an elastic-p lastic behavior. Va lues of the load

given by DY A inclu de the load on the memb er under investigation. and the loads on

other me mbe rs. A careful exam ination o f the data suggests that clem ent s that eve ntua lly

yield arc subj ect to the moving ice load roughly from 1.2 seconds to 1.6 seco nds. Va lues

o f the pressur e on the co ntac t interla ce are exa mined. Va lues of the co rrespo nding

loading areas arc manu all y measur ed in the DY NA output. The dell ection is rou ghl y

ca lculated by subtrac ting the displa cem ent due to the global motion from the total

displ acemen t give n by DY NA. which is an approxima tion to the actual del1ection.

138
Co m bining a ll the informa tion above g ives a roug h es tima tio n of the pressure-d ell ectio n

c urve . wh ich is sho w n in Figure 6- 14 .

,, 5
/
/
./"
----- /
/'
~

~
~
/
t
4

3
I /
I /
II /
o
I /
0.000
Deflc ct ion tm ]

- l'rcsslIrc-Dcllcction ClIrvc of thc Main Fralllcs

- Capacity lindcr Syllllllctrical Load
- Capacity lindcr Asyllllllctrical Load

Figure 6-14: Pre ssure - Dcflection Curve of Ihe Main Fram e Member

In the si m ulation. the load ing co nd ition o n the main frame cha nges fro m asy m me trica l to

sy m me tr ica l. a nd the n back to asy m me trica l as the ice mo ves a long the ship bow . T he

ca pac ities und e r asy m me trica l load and symmetrica l load are plott ed as wel l. T hey a re

ca lcul at ed usin g the eq uations pro posed by Ke ndri ck et a l. (2000) a nd Da ley (200 2) .

T he ir va lues are highe r tha n wh at is sugges ted by the pressure-d e llection c urve. It is

beca use the informa tion for ge ne rati ng thi s pressur e-d e llect ion plot is o bta ine d via
139
es timatio n and manuall y measu red from the DY NA output. w hich inevi ta bly intro d uces

e rrors to the plot. Howe ver. it is very clear th at the struc tura l resp on se featur es a n

exte nde d plastic reg ion . T his is becau se the ma in frames are des ign ed as memb ers o n a

PC- 4 shi p. while the ice mat erial model co m plies wi th th e o ne speci fied for PC- 3. It is

a lso appare nt that the plastic strengt h of the struc tura l memb ers is utili zed to resist the ice

load as des igned . T he struc tura l memb er s have de velop ed plastic stra in. residu al

de flec tio n. a nd post-yield residu al stress .

6.4 Summary

T his chapt er prop oses the final mod el by co m bining a ll the pre viou s work togeth er and

incorp or atin g the ship's local struc tura l resp on se . How ever . thi s cha pte r doe s not intend

to give a fu ll treatm ent to the struc tura l a na lys is. It is pu rel y for giv ing a ge ne ra l exa m ple

o f how to use the proposed mod el to ca rry o ut struc tura l a na lysis of the shi p und e r ice

im pact s. Use rs ca n fo llow the di rectio n give n in thi s chap ter to bu ild mor e e labora te

mode ls a nd ob ta in mor e co mpre he nsive and so phist ica ted result s.

As show n in this cha pte r. th e mod el develop ed in thi s thes is is ca pa ble of g ivi ng th e

glo ba l moti o ns o f the ship a nd ice. the globa l ship ice cont act load . ice failure. and ship's

structura l resp on se in one pack age. Th e co m putation time of the prop osed model is

reason ab ly low a nd ca n be furth er reduced by mod e ling a la rge r porti on of the mode l as

rigid . a nd usin g a non-uniform mesh . A mesh co nve rge nce study is not co nd ucted for thi s

140
sim ulation. A mar c refi ned mesh on the contac t region and local structural is necessary

fora ma rc acc urate structural analysis.

141
 Chapter 7 Conclu sions and Recommendations

The goa l of thi s study is to defin e a procedur e of ana lyz ing ship-icc co llis ion using the

co mmercia l FEA so ftwa re DY NA. The fina l produ ct is an efficie nt FEA so lutio n that is

ca pab le of eva luating the hydrod yn amic forces . the globa l motions o f the ship and icc. the

co ntac t fo rce. ice failu re. and the ship structural respon se in one e fficient ana lysis. In

order to ac hieve th is goa l. the word was co mpleted as fou r subto pics as stated in Sec tion

1.2. The co ncl usion fo r each subtopic has been discussed in the end o f the correspo nding

chapter. The wor d is summa rized and present ed in this chapter with ge nera l co nclusions.

7.1 Conclusions

First of all. three ice materia l models are propo sed . They arc the foundat ion o f the present

study . The ir pressure-area curves have dem on strat ed goo d ag ree me nt with those specifie d

in the URI. These icc materia l mod els have practi cal applications. They can also be

furt her modi fied to serve differ ent purp oses.

Modeling wa ter using the A LE meth od has been discussed . This study has sho wn that the

ALE meth od is not practic al when the mode l co nta ins a ve ry relin ed mesh co nstructed by

a large numb er o f clement s. Moreover. the A LE method is not an idea l approac h if the

hydro dyn am ic e ffe cts arc significant. du e to the natur e of the A LE meth od . However . the

AL E meth od is useful if the ana lys is foc uses on the low frequ ency globa l moti on s o f the

floating bodies as shown in severa l existing studies .

142
As an alte rna tive to the ALE meth od. this research prop oses to model the fluid effect

user-defined- cur ve-fun ction s in DYNA. Th e globa l ship-ice cont act forc es in various

ship-ic e glancing scenarios have been evaluated. Th e present solution agree s well with

the DDeP S so lution when the sa me mass reduction coe ffi cient is applied. Thi s research

simulates restorin g forces using displacem ent springs rath er than act ua lly model ing water

to imp rove efficie ncy . The restoring fo rces are show n to have sma ll influen ce on the first

imp act force . but are imp ort ant in es tima ting the seco nd imp act. This co nclus ion agrees

with DDeP S. Drag forc e is not necessary to be includ ed in this type o f simulations. On

the oth er hand. the added mass effect should be con sider ed . However. it is not pres ent in

the so lution du e to so ftware probl em s. The ship-ice cont act model prop osed by this thesis

is a very efficie nt so lutio n for assessi ng the globa l ship- ice con tact force,

In the last part of this thesis. the so lution develop ed previou sly is ex tended to inc lude the

ship-s tructura l respon se. The elastic-plastic beh avior of the ship structural members is

obse rved . The final so lution gives the global motion s of the ship and ice. the cont act force ,

ice fa ilu re. and the ship struct ural local res ponse in one package. The co mputation cos t is

fairly low and ca n be furth er redu ced by modelin g large r port ion s o f the bodies as rigid.

Overa ll, the initial goa l o f thi s thesis has bee n ac hieve d.

7.2 Recommendations

Severa l qu esti on s have arisen durin g this research that prompt furth er research . The most

impo rtant improve me nt that can be made is to include the add ed mass in the so lution.

143
Mod elin g the add ed ma ss usin g user-defined -cur ve-fun ction s is likely the most cos t-

effective method. It is necessary to determin e wh y the curve-functions of ac celera tions

beh ave incon siste ntly when so lving the sa me model using di ffer ent releases of DYNA.

Personnel from LSTC have advise d that the latest release of DYNA may have thi s issue

reso lved. [fn ot. a wo rkaro und can be made by ass igning extra mass to the ship and ice as

the adde d ma ss. Users ca n customize the va lue o f the added mass in each of the 6 DOF

using the *PA RT_ [NE RT IA card. [fthe new release can mod el the added mass as

discu ssed in Section 5.1 . the valu es o f the added mass coeffi cient s should be

appropriately cho sen . As di scu ssed in this the sis. for a given Iloatin g bod y. the added

mass effe ct under the instant impact load is different from that in the rotati on al and

tran slati on al moti on s. DYNA is not the ideal tool for estim atin g the added mass due to its

natur e o f ca lcu lating the fluid- stru ctu ral interacti on force. It is better to use CF D

program s or ex per imen ts to perf orm this task. The latest versio n o f DY A has

incorporated an incompre ssibl e flow so lver. It does not requir e the usage o f cur ve-

functions and m igh t be the most acc ura te so lution for sim ulating the water dom ain where

the ship-ice co ntac t occ urs.

Once the added mass issue is resolved. it is recomm end ed to furth er assess its influen ce

on the mass redu ction coefficient via FEA using DYNA. T he added mass can be ea sil y

incorporated into the model discu ssed in Section 5. 1 usin g user-defined-curve-function s.

DDeP S sugges ts that the added mass significantly redu ces the mass reduction coe ffic ient.

and co nsequently incr eases the effective mass and the co ntac t for ce. It is desired to

[44
investi gate th is ph en om en on in DY NA. A nothe r future wo rk dep endin g o n the adde d

mass is to si m ulate mor e shi p-ice co ntac t sce na rios usin g the co ntact mod el prop osed in

Cha pter 5. DDeP S pro vides ve ry qu ick so lutio ns for more than :W diffe rent shi p-ice

co ntac t sce na rios. A ll of them can be sim ulated by sim ply modi fyin g the geo met rica l

model in th e co ntac t model present ed in Chapter 5. It is o ptima l to comp are the result s

give n by DY N A again st the es tima tions by DD eP S.

A no the r futur e work relat ed to hydrod ynamics is to apply dra g force on the finite ice

ma ss. Drag force can slow the ice from moving away from the ship and give a better

es timatio n of the cont act force. In othe r ana lyses. it mi ght be nece ssar y to apply dra g

forc e o n the ship as well dep endin g o n the as sumption related to the ship vel oc ity.

A no ther reco m me nde d future work is to bett er measur e the nomin al co ntac t a rea w he n its

va lue is sma ll. Wh en developin g ice mat eri a l mod el s. the nom inal co ntact a rea derived

fro m Rhin oce ros® is not perfe ctl y co nsis te nt with the ac tua l nomin al co ntac t area in

DY NA . A ltho ug h th is di scr ep anc y has been redu ced. it still hinde rs the ac cura te

interpr etati on of the pressur e-ar ea cur ve o f the ice mat eri al mod el. It is optima l to so lve

thi s probl em to impro ve the pre sent so lution for an alyzing the imp act between a ship and

a finit e ice .

An oth er possibl e improvem ent co nce rns the ice materi al mod el. When developin g ice

mat eri al mod els. a pplying the the or y o f de sign of ex per ime nts will help with calibrating

145
mater ial param eters and their co mbinations to yie ld desired result s. In addition. modelin g

the ice using material mod els other than the cru shabl e fo am model sho uld be further

exp lored . It is recom mended to start wit h the nonl inear elastic- plastic mater ial model with

the MAT] IEC EW ISE_ LINEA R_ PLAST IC ITY car d in DYNA. This materi al model

gives users a large co ntro l ove r the materi al' s stress -stra in relat ion ship .

Durin g thi s research. num erical instabilit y occurred when modelin g ice block s with sharp

edges. Whe n the clement size is sma ll. the cont act involvin g sharp edges tend s to

introdu ce the negati ve volum e problem in DYNA. This issue co uld be miti gated by

prop erl y introducing material eros ion. or choo sing a more robu st material. such as the

nonlin ear elastic -plastic material.

This study does not investigate the size of the co ntac t area wh en eva luating the ship-ice

co ntac t force given by DY A ' s simulations. Th is sho uld be included in future studies .

The math em at ical approx imation of the ice materi al is not very acc ura te lor very sma ll

co ntact areas . In fo rmation of the exact co ntac t area can help users to ga in co nfide nce in

the result s. and to bette r und erstand the possibl e dif ference in the DY NA so lutio n and the

DDeP S so lution.

The final finite clement model prop osed by thi s thesis can also be improve d. First of all. a

mesh co nve rgence sho uld be conducted. It is possible that the ship structura l membe rs

should be meshed usin g sma ller clem ent s lo r a mor e accurate structura l analysis. Thi s

sho uld be a local refin em ent on the mesh to avoid signifi cantly increasi ng the total

146
num ber of cle me nts. Seco ndly. it is necessary to develop a method to more acc urately

measu re the pres sure and deflection for ana lyzing the pressure-defl ecti on curve . which is

very d ifficult to do when both the ship and ice are moving. It is much simpler to per form

the pressure-de flec tion analys is on a local model of the ship struc ture. The sec tio n o f the

ship can be fixed on the boun daries. and the ship-ice imp act ca n be repl aced by a movin g

load on the structure . Another future wo rk is to further reduc e the co mputation cost.

Using non-unifo rml y sized elements to generate the mesh can significa ntly reduc e the

total numb er of elem ent s. Use rs can also try modelin g lar ger partition s of the ship and ice

as rigid to reduce the computation time . since the effect of the impact is high ly local ized .

Fina lly. DY NA is a ve ry power ful and evo lving FEA program with so many functions to

be fur ther explored. The author has ex per ience wi th other co mme rcial FEA so ftwa re. and

stro ngly bel ieve s that DY NA is the best avai lable tool for add ress ing the ship-ice

collis ion problem . A ltho ugh th is thesis does not provide an swer s to all the iss ues . it serves

as the platform whic h further deve lopme nt ca n be bu ilt up on.

147
 Chapter 8 Bibliography
th
Daley. C.G .. 1999. Energy Based Ice Co llision Forces. Proceedings of the 15

International Co nfere nce on Port and Ocea n Engineer ing under Arctic Co nditions.

POAC' 99. Vo l.Z, pp. 687-696. Helsink i. Finland. August 23 -27. 1999.

Daley. C.G.. 2000. lACS Unified Requ iremen ts fo r Polar Ships - Backgroun d Notes -

Desig n Ice Loads. Prepared for lACS Ad-hoc Gro up on Polar Class Ships Tra nsport

Ca nada. 2000 .

Daley. CiG.. Oblique Ice Co llision Loads on Ships Based on Energy Met hods. Ocea n Eng

Int 200 1; 5(2) : 67-72.

Daley. e.G .. 2002, Deriva tio n of Plastic Framing Requi rements for Polar Ships. Marine

Structures IS (2002) 543 -559 .

Daley. e.G .. 2004. A Study of the Process-Spatial Link in Ice Pressu re- Area

Relat ionships.

Daley, e.G ., 2007 , Reanalysis of Ice Press ure-Area Relations hips. Marine Tec hnology .

Vo l. 44. No.4, October 2007, pp. 234-244.

148
Daley. e.G .. Kendr ick . A.. 200 8. Direct Design of Large Ice C lass Ships with emphasis

on the Midbody Ice Bell. Proceed ings of the 27t h Intern ational Co nference on O ffsh ore

Mec hanics and Arc tic Engi neering OMAE2008 Jul y 15-20. 2008. EstoriL Portu gal. pa per

200 8-57 846 .

Daley. C; Kendri ck . A.. Yu.H.. No h. B-L 2007. Structura l Design of High Icc Class

LNG Ta nkers. RINA Co nference Design & Co nstruction O f Vessel s Operating In Low

Te mpera ture Environments 30 - 3 1 May 2007. RINA I-IQ. London. UK.

Daley. e.G .. Liu, L 2009a. DDeP S for Ship and Infinit e/ Finit e Ice Interaction. Rep ort by

BM T Fleet Tec hno logy Limit ed. submitted to Research & Produ ct Development,

Ame rica n Bureau of Shipping. Jan . 2009 .

Daley. e.G .. Liu, L 20 10. Assess me nt of Ship Ice Loads in Pack Icc. Intern ati onal

Co nfe rence and Exhibitio n on Perform an ce of Shi ps and Structures in Ice. Anchorage.

A laska. USA Se ptember 20-23 . Paper No IC ET ECI-II0- 14 L 20 10.

Daley. e.G. Yu, II .. 2009b. Ass ess me nt of Ice Load s on Stern Region s of Ice Class Ships.

Intern at ional Co nference on Ship and Offsh ore Technolo gy: Icc C lass Ships. 28 - 29

Se ptembe r 2009 . Busan, Korea .

149
Dcrradji- Aouat , A.. 2003 . Multi- Surface Failure Criterion for Sa line ice in the Brittl e

Regime. co ld Region s Scie nce and Tec hno logy . vol. 36. pp. 47 -70. 2003.

DNY. 20 11. Ships for Naviga tion in Icc. Rules for C lass ifica tion of Ships - Ne wb uildings

- Specia l Serv ice and Type - Additiona l Class

Frede rking. R.. 1998. Th e Pressur e Area Relation in the Defi niti on o f Icc Forces.

Proceedings o f the Eighth ( 1998) Intern ation al Offshore and Po lar Eng ineering

Confere nce. Mon tre al. Ca nada. May 24- 29. 1998 .

Freder king. R.. 1999. The Loca l Pressure-Area Relation in Ship Impact with Icc.

1h
Proceedin gs of the 15 Intern ational Co nfere nce on Port and Ocea n Eng inee ring und er

Arctic Conditions. POAC' 99. Yol.2 . pp. 687-696 . lIe lsinki. Finla nd. August 23-27 . 1999 .

Gag no n. R.. 200 4. Physical Mode l Experi me nts to Assess the Hydrod ynami c Interaction

Between Floati ng G lacia l Ice Masses and a Tra nsiting Ta nker. jo urna l o f Offs hore

Mec hanic s and A rctic Engineering.

Gag no n. R. and Derradji- Aou at, A.. 2006. First Res ults o f Nume rica l Sim ulations o f

1h
Bergy Bit Co llision wit h the CCGS Terr y Fox Icebreaker. the 18 International

Sympos ium on Icc. Sapporo . .Japan.

150
Hallquist., .1.0 .. 2006 . LS-DYNA Theo ry Manu al 2006 . Livcmore Software Tec hnology

Co rporatio n (LSTC) .

Hulme, A.. 1982. The Wave Forces Acting on a Floating hemisphere Undergoing Forces

Period ic Osci llations. .I. Fluid Mcch.. 121. pp. 443-4 63.

lACS. 20 IO. Requir ement s Co ncerning Polar Class. Lond on : International Asso ciation o f

Class ifica tion Soc ieties

Kendr ick. A.• Daley. CG .. 2000a. Deriv ation and Use of Formulatio ns for Framing

Design in the Polar Class Unifie d Requirements. Prepared for lACS Ad -hoc Gro up on

Polar Class Ships Tra nsport Canada. 2000 .

Kendrick. A.. Daley. CG .. 2000 b. Unifi ed Requ irements Load Model - 'Sy nthesis

Approac h' , Prepared for lACS Polar Rules Harm onization. Semi-Permanent Workin g

Gro up. Prepared by. AMA RK Inc. Montrea l. and Memorial University of New found land.

St. John ' s.

Kendrick, A.. Daley, CG . Quint on. B.. 2009. Scenario-Based Assessment of Risks to Icc

Class Ship s. Off shore Techn ology Conference. Houston . Texa s. USA. 4-7 May 2009.

OTC 20002 .

151
Kenn ed y. L 20 12. Personal co mmunications.

Korsm eyer, F. '1'.. and Sc lavo unos, P. D.. 1989. The Large-T ime Asy mptotic Expa nsion

of the Impul se Respon se Function for a Floa ting Body. Appl. Ocea n. Res.. 11(2). pp. 75-

88.

Kwak, M.. Choi • .1.• Park. S.. Kang• .1 •• 2006. Strengt h Assess me nt fo r Bow Structure of

Arctic Ta nker ( I07 k) und er Ship-Ice Interaction . Daewoo Shipbuilding & Mari ne

Eng ineer ing Co .• LT D.

Lee . S.. Lee. I.. Baek, Y.• Co uty, N.. Goff S.. and Q ucnc z.. .1.• 2007. Membr ane-Type

Carrier Side Co llision wit h Iceber g Effec t of Impact Co nd itions on Sructura l Respon se

Throug h Se nsit ivity A nalysis

Liap is. S. .1 .• 1986. T ime -Doma in Ana lysis o f Ship Motion s. Phd thesis. Unive rsity of

Michigan .

Liu, Z.. Amda hl. .1•• and Lose t, S.. 2009. Numer ica l Simulation of Co llision between

Ships and Icebergs. Proceed ings of thc zo" Intern ational Conference on Port and Ocea n
Engineering und er A rctic Co nditions. Lulca, Swe de n. Jun e 9- 12. 2009 .

152
LSTC. 2007a . LS- DYNA Keyword User's Manu al Volume I. Versio n 97 1. 200 7.

Livermo re Software Tec hnology Corpora tion.

LSTC. 2007b . LS-DYNA Keyword User's Manua l Vo lume II. Vers ion 97 1. 2007 .

Livermore Software Tec hno logy Corpora tion.

LSTC. 20 10. LS- DYNA Lectur e Notes. 20 10. Live rmo re Soft ware Tec hno logy

Cor poration.

Masterson. D.M.. Frederking. R.M.W .. Wright. B.. Kam a. t.. and Madd ock. W.P.. 2007.

A Revised Ice Pressure-Area Curve. Recent developm ent of Offshore Engi nee ring in

Co ld Region s. Yue (cd.). POAC-07. Dalian. China. Jun e 27-30 . 2007 .

Myhre. S.. 20 10. Analysis of Accide ntal Iceberg Impacts wit h Membrane Tank LNG

Carriers. Master 's thesis. orwegia n University of Sc ience and Tec hnology.

Prins. H. .I.. 1995. Time-doma in Ca lculations of Dri ll Forces and Mome nts. PhD thesis.

Del ft Universi ty of Tec hnology .

Popov. Yu.. Fadd eycv, 0 .. Khcisin, D.. and Yalovlcv, A.. 1967. Strength of Ships Sa iling

in Icc. Sudos troe nie Publ ishin g House. Leningrad. Tec hnica l Translation. U.S. Arm y

Foreign Sc ience and Tec hnology Center. FSTC -HT-23-96 -68.

153
Quinton. B.. 2009. Progressive Damage to a Ship' s Structure Due to Ice Loading.

master' s thesis. Memorial University of Newfoundland.

Sierevogel, L 1998. Time Domain Calculations of Ship Motions. PhD thesis. Delft

University of Technology.

Storti. M.. and D' Elia, J.. 2004. Added Mass of an Oscillating Hemisphere at Very-Low

and Very-High Frequencies, Transactions of the ASME, 1048/Vol. 126. 2004.

Wang. B.. Yu, H.. Basu, Roo Lee. II.. Kwon. .J..Jeon, B.. Kim. .I.. Daley. CiG.. and

Kendrick. A.. 2008a. Structural Response of Cargo Containment Systems in LNG

Carriers under Ice Loads. ICETECII 2008 Conference. Banff , Canada. .Iuly 20-23. 2008.

Wang. B.. Yu, II.. Basu, R.. 2008b. Ship and Ice Collision Modeling and Strength

Evaluation of L G Ship Structure. Proceedings of the ASME 2t" International

Conference on Offshore Mechan ics and Arctic Engineering. EstoriL Portugal, .Iune 15-20.

2008.

Wang. .I. and Derradji-Aouat. A .. 20 10a. Numerical Prediction for Resistance of

Canadian Icebreaker CCGS Terry Fox in Level Ice. Report from the National Research

Council's Institute for Ocean Technology. St. John 's, Canada.

154
Wang. J. and Derr adji- Aou at . A.. 20 10b. Ship perf ormanc e in Brok en Ice Floes -

Prelim inary Numerica l Simulations. Report from the Na tional Research Co uncil's

Institut e for Oc ean Te chn olo gy. St. John' s, Canada .

Wang. J. 20 11. Person al co mmunications.

155
Appendices

156
 Appendix A: STe PS2 Cluster S pecifications

T he STe PS2 cluste r used in th is resear ch has spec ifica tions as follo ws :

Pr ocessors
# ofC PUs 2
CPU tv e Intcl(R) Xcon(R) E5520
Corcs pcrC PU 4
CPU Frequency 2.27G Hz
CPU Max Turbo Frequencv 2.53 GHz
CPUCachc 8MB
CPU Address Sizes 40 bits hysical, 48 bits virtual
QPI Spccd 5.86GT/s
Instruction Sct 64-bit
Hyper thread ing Yes and Enabled
Mcmo ry
Total M emory 32G B
Mcm ory per C f'U 16 GB
Memorv Slots 8 crC PU(a Il 8 fillcd)
DIMM Sizc 2G B
Typc DDR 3 ECC
Mcmorv Fre ucncv 800 MHz
Storaue
Arrav I
RAID RAID 5
Number of disks 5
To ta l Storaue 584G B
Sto rage per disk 146G B
Disk Ty pe SAS
Disk Spccd 15000 RPM
RAID Co ntro ller Hardw are
Arra ' 2
RAID RAID 5
Number ofd isks 3
Total Storaue 600 GB
Storaue pcr disk 300G B
Disk T c SAS
Disk Speed 15000 RPM
RAID Controller Hardware
O ne rat inz Svs te m
o crarinu Svstcm RHEL Server 5.4 (Tikanua)
l. inux Kernel 2.6.18- 164
Architecture Intc l x86 64

157
 Ta ble A-2: Co mpute Node Ha rdw ar e
Pr ocessor s
Same as head node exc ent Hvper threadina not Enabled
Me mory
To ta l Mem orv 24G B
Mernorv cr CPU 12 GB
Mem orv Slots 8 crC PU(6of 8 fi lled)
DIMM Sizc 2G B
Tv pe [)DR 3 ECC
Memorv Fre ucnc 1067 Mll z
Sto ra ae
RAID RAIDO
Number of disks 4
Tota l Sto raue 584 GB
Storac e cr disk 146 GB
Disk Typc SAS
Disk S ccd 15000 RPM
RAID Contro ller Ha rdware
O pera ting S stc m
Operating System Sa me as head nod e

158
 Appendix B: DYNA's Keyword File of the Final Model

This appendix is the keyw ord file of the fina l model. The keyword file o f the geo met ric

model is not includ ed here du e to its enor mo us size .

*KEYWO RD

*T IT LE

Ship Ice Collision Mod el

$===============================================================

$ I J Mo de l Geo metry

$===============================================================

*INC LU DE geo.k

$===============================================================

$ I] EXECU T ION CONTROLS

$===============================================================

$*CONT ROL _M PP_ DECOM POS ITION_S I-IOW

*CO NT RO L_TE RM INAT IO N

$# endtim endcyc dtmin ende ng endmas

*CONT RO L_T IMESTE P

$# dtinit tss lac isdo tslimt dt 2m s Ictm ero de msl st

0.7

$# t2 msf dt2ms lc imscl

159
*DEFIN E_ CU RVE

$ lcid sidr scla sclo offa offo

1000

absc issa ord ina te

4 . 15E-6

1000 4.1 5 E-6

*CO NT RO L_E NE RG Y

$# hgen s lntc n rylen

*C ONT RO L_ M PP_ DECO M POS IT IO N_ M ET HO D

$# nam e

RCB

*CO T ROL _M PP_ IO_ O D3 DUM P

*C O NT RO L_M PP_ IO_ O DUM P

$===============================================================

$ I I O UT PUT CON T ROLS

$===============================================================

*DAT A BAS E_G LST AT

$# dt bi nary leu r ioop t

0.1

*DAT A BAS E_ M ATS UM

160
$# dt binary leur ioopt

0. 1

*DATABASE_RCFORC

$# dt binary leur ioopt

0.0 1

*DATABASE_BI ARY_D3I' LOT

$# dt Iedt beam npltc psctid

0. 1

$# ioopt

*DATABASE_B1NARY_D3TIIDT

$# dt Iedt beam npltc psetid

0. 1

$===============================================================

$ 1J IPART_IDISECTION_IDII'vlAT_IDIEOS_ IDII-IG_1D1
$===============================================================

*MAT_ RIGID

$# mid pr couple m alias

7850 2.0E+ II 0.3

$# cm o con l con2

$# leo or al a2 a3 vI v2 v3

161
$---------------------------------------------------------------------------- ---

*PART

$# titlc

rigid ship

$# pid scci d mid eos id hgid grav adpopt tm id

*SECT ION- SHELL

$# scc id cl forrn shrf nip propt qr/irid icomp sctyp

0.833333

$# t l t2 13 t4 nloc ido l' cdgsc t

0.03 0.03 0.03 0.03

*I'ART

$# title

deep web frame

$# pid scci d mid cos id hgid grav adpopt tmid

15 15

*SECTION- SHELL

$# sec id el form shrf nip propt qr/irid icomp sctyp

15 0.83333

$# t l t2 t3 t4 nloc idol' cdgs ct

0.03 0.03 0.03 0.03

162
*PART

$# title

load carryi ng str inge rs

$# pid sec id mid eosi d hgid gray adpo pt tmid

17 17

*SECTION_SHELL

$# sec id elform shrf nip propt qr/irid icomp sety p

17 0.83333

$# t1 t2 t3 t4 nloc ido l' edgse t

0.03 0.03 0.03 0.03

*PART

$# title

bulkhead

$# pid sec id mid eosi d hgid gray adpo pt tm id

25 25

*SECTION_SHELL

$# secid elfor m shr f nip propt qr/irid ico mp sety p

25 0. 83333

$# t 1 t2 t3 t4 nloc ido l' edgse t

0.03 0.03 0.03 0.03

*PART

$# title

163
deck

$# pid seeid mid eosid hgid gray adpopt tmid

31 31

*SECTION_SHELL

$# seeid elform shrf nip propt qr/irid ieomp setyp

31 0.83333

$# t 1 t2 t3 t4 nloe idol' edgset

0.03 0.03 0.03 0.03

*PART

$# title

hull

$# pid seeid mid eosid hgid gray adpopt tmid

70 70

*SECTION_SHELL

$# seeid elfor m shrf nip propt qr/irid ieomp setyp

70 0.83333

$# t1 t2 t3 t4 nloe idol' cdgse t

0.03 0.03 0.03 0.03

*PART

$# title

bow

$# pid seeid mid eosid hgid gray adpopt tmid

164
73 73

*SECTION - SHE L L

$# secid el for rn shrf nip propt qr/i rid ico mp sety p

73 0.83333

$# t 1 t2 t3 t4 nloc idol' cdgse t

0.022 0.022 0.022 0.022

*PA RT

$# tit le

lee

$# pid sec id mid eos id hgid gray ad po pt tmid

170 170 170

$-------- ------ ------- ------- ------- --- --------- --- --- ----- ----------------- ----

*MAT_A DD- EROS ION

$# m id exc l m xpr es mn ep s effeps vo leps numfi p

170 0.9

$# m npres sigp 1 sigv m mxeps epss h sigt h im pulse failtrn

*MAT_C RUS I-IABL E- FOAM

$# m id pr lcid damp

170 900 5.0E +9 0.003 170 8.0E +8 0

*DEF INE _CU RVE

$# Ic id sidr sfa sfo offa offo datt yp

165
170

$# al 01

1.5e6

0.02 1.5e6

0.03 1.5e6

0.04 1.5e6

0.05 1.5e6

0.065 1.5e6

0.075 3e6

0. 1 3e6

0.5 3e6

0.8 3e6

0.89 3e6

*SECT ION_SO LID

$# sec id el for rn

170

$-------------- ------------------------- -------- ---------------- ----------------

*PA RT

$# tit le

rigid ice

$# pid secid mid eos id hgid gray adpopt tmid

171 171 171

166
*MAT_ RIGID

$# mid pr couple m alias

171 900 5.0 E+9 0.003

$#c mo con I con 2

$# leo or al a2 a3 vI v2 v3

*SECTION_S O LID

$#secid clfor m

171

$-------------------------------------------------------------------------------

*SET _ PA RT_ LlST

$# sid dal da 2 da 3 da4 so lver

MECI-I

$# pidl pid2 pid 3 pid4 pid 5 pid 6 pid 7 pid8

IS 17 25 73

*SET]A RT_ L1ST

$# sid dal da2 da3 da4 solver

M ECH

$# pidl pid 2 pid 3 pid4 pid5 pid 6 pid 7 pid8

170 171

*SET]A RT_ LIST

167
$# sid dal da2 da3 da4 solver

MECI-I

$# pidl pid2 pid3 pid4 pid5 pid6 pid7 pidS

170

*SET PART_LIST

$# sid dal da2 da3 da4 solver

MECl-1

$# pidl pid2 pid3 pid4 pid5 pid6 pid7 pidS

15 17 25 31 70 73

$===============================================================

$ [ I CONTACTS

$===============================================================

*CONTACT_INTERIOR

$# psid

*CONTACT_AUTOMATIC_S INGLE_SURFACE

$# eid title

$# ssid msid sstyp mstyp sboxid mboxi spr mpr

$# fs fd de vde pcnchk bt dt

I.OE20

$# sfs sfm sfst sfm t fsf vsf

16S
$# so ft sofscl lcidab rnaxpa r sbopt depth bsor t frefrq

0.1 1.025

*CON TACT_ FO RCE _T RAN SDUCE R

$# eid title

$#ssid msid sstyp mstyp sbox id mboxi spr mpr

$#fs I'd de vde penehk bt dt

1.0 E20

$# sfs sfm sfs t sfm t [~ f vsf

$===============================================================

$ [ ] BC's + IC's + BODY LOAD + FORCE FIELDS

$===============================================================

*CONS T RA INE D_E XT RA_N O DES_NO DE

$# pid nid

500000

500001

500002

500005

50000 6

*DEF INE_COO RDINATE _N O DES

169
$# cid nl n2 n3 flag dir

500 000 50000 1 50000 2 I x

*E LEME NT_ MASS_ PA RT_ SET

$# psid add finmass

3. I E+6

$---- -----------------------------------SHIP------------------------------------

*IN ITIA L_ V ELOC ITY_G EN ERAT ION

$# sid/pid styp omega vy ivatn icid

$# x c yc ny phase iridid

$= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =

*DEF INE_CU RVEJUNCT ION

$# lcid sidr sfa sfo offa off o dattyp

$# function

-682821934*DZ(50 000 0)

*LOA D_N O DE_ PO INT

$# nid dol' lcid sf cid ml m2 1113

500000

*DE FIN E_CU RVEJU NC T ION

$# lcid sidr sfa sfo off a ofTo dattyp

170
$# function

-2.9E+07.29*AX(500000)

*LOAD_NODE_POINT

$# nid dol' lcid sf cid 1111 1112 1113

500000

*DEFINE_CU RVEJUNCTION

$# lcid sidr sfa sfo offa 01'1'0 dattyp

$# functio n

-1.8E+0722.13*AY(500000)

*LOAD_NODE_POINT

$# nid dol' lcid sf cid 1111 1112 1113

500000

$------------------------------------------------------------------------------

*DAMPING] ART_MASS_SET

$# psid Icid sf flag

10

$# st x sty sry

0.30964 0.0832 0.4639 0

$# lcid sidr sfa sfo offa o ffo dattyp

171
10

abscissa ordinate

$# al 01

100

*END

172