modifications on oil outflow the DAMAGE calculationsshould be coupled with outflow analysis.The next step <strong>in</strong> the study is to carry out aprobabilistic analysis of damage <strong>in</strong> ground<strong>in</strong>g scenariosdef<strong>in</strong>ed by probability density functions of the <strong>in</strong>putparameters. This analysis will be coupled with outflowcalculations. The objective of the study will be to providea tool for comparative studies between vessel designs.5 COLLISION MODELS5.1 Background <strong>and</strong> PlanIn 1979, the Ship Structure Committee (SSC) conducted areview of collision research <strong>and</strong> design methodologies[18]. They concluded that the most promis<strong>in</strong>g simplifiedcollision analysis alternative was to extend M<strong>in</strong>orsky’sorig<strong>in</strong>al analysis of high-energy collisions by <strong>in</strong>clud<strong>in</strong>gconsideration of shell membrane energy absorption. Asimple <strong>and</strong> fast model is important <strong>in</strong> probabilisticanalysis because thous<strong>and</strong>s of different scenarios must berun to develop statistically significant results.A more recent review of the literature <strong>and</strong> of theapplicability of available methods for predict<strong>in</strong>g structuralperformance <strong>in</strong> collision <strong>and</strong> ground<strong>in</strong>g was made <strong>in</strong> the1997 International Ship <strong>and</strong> Offshore Structures Congress(ISSC 97) by Specialist Panel V.4 [9]. Their report states:“Knowledge of behavior on a global level only (i.e., totalenergy characteristics like the pioneer<strong>in</strong>g M<strong>in</strong>orskyformula) is not sufficient. The designer needs detailedknowledge on the component behavior (bulkheads,girders, plat<strong>in</strong>g, etc.) <strong>in</strong> order to optimize the design foraccident loads.”The approach taken <strong>in</strong> this study is to progressively<strong>in</strong>crease the complexity of SIMCOL (Simplified <strong>Collision</strong>Model), the Ad Hoc Panel’s basel<strong>in</strong>e collision model,until results with sufficient accuracy <strong>and</strong> sensitivity todesign characteristics are obta<strong>in</strong>ed. In order to assess themodel’s consistency <strong>and</strong> sensitivity, the model is applied<strong>in</strong> a series of collision scenarios with a range of tankersizes <strong>and</strong> designs. SIMCOL results are compared toresults obta<strong>in</strong>ed us<strong>in</strong>g three other collision models:DAMAGE [10,19], ALPS/SCOL [20,21], <strong>and</strong> a TechnicalUniversity of Denmark (DTU) model. The DTU model<strong>and</strong> ALPS/SCOL were also further developed dur<strong>in</strong>g thisstudy. Determ<strong>in</strong><strong>in</strong>g the ideal trade-off between simplicity<strong>and</strong> sufficiency is a primary concern of this process.5.2 Models Included <strong>in</strong> the Current StudyThree types of collision were identified for <strong>in</strong>vestigationby the panel:• Low energy puncture at a po<strong>in</strong>t• Low energy rak<strong>in</strong>g puncture• Penetrat<strong>in</strong>g <strong>Collision</strong> – right angle <strong>and</strong> obliquesufficient to penetrate outer hull with significantdamage extend<strong>in</strong>g <strong>in</strong> at least 2 directions(penetration, horizontal, vertical)Work thus far is focused on models for penetrat<strong>in</strong>gcollision.Each collision model must consider two verydifferent aspects of the collision event:• External problem – <strong>in</strong>cludes the globalcharacteristics of the strik<strong>in</strong>g <strong>and</strong> struck ships <strong>and</strong>their motion before, dur<strong>in</strong>g <strong>and</strong> after collision;• Internal problem - <strong>in</strong>cludes the <strong>in</strong>ternal deformationmechanics <strong>and</strong> structural response of the struck ship<strong>and</strong> the strik<strong>in</strong>g ship bow dur<strong>in</strong>g collision.The four models studied by the panel take differentapproaches to solv<strong>in</strong>g <strong>and</strong> coupl<strong>in</strong>g these two problems.Although not a formal validation, comparison <strong>and</strong>agreement between these models provides useful <strong>in</strong>sight<strong>in</strong>to their performance <strong>and</strong> <strong>in</strong>creases confidence <strong>in</strong> thevalidity of their results.5.2.1 Simplified <strong>Collision</strong> Model (SIMCOL) [22]SIMCOL uses a time -doma<strong>in</strong> simultaneous solutionof external dynamics <strong>and</strong> <strong>in</strong>ternal deformation mechanicssimilar to that orig<strong>in</strong>ally proposed by Hutchison [23].Figure 12 shows the SIMCOL simulation process. Theexternal sub-model uses a three degree-of-freedomsystem for ship dynamics shown <strong>in</strong> Figure 13. The<strong>in</strong>ternal sub-model determ<strong>in</strong>es react<strong>in</strong>g forces from side<strong>and</strong> bulkhead structures us<strong>in</strong>g detailed mechanismsadapted from a 1981 Rosenblatt study [24,25]. Itdeterm<strong>in</strong>es the energy absorbed by the crush<strong>in</strong>g <strong>and</strong>tear<strong>in</strong>g of decks, bottoms <strong>and</strong> str<strong>in</strong>gers us<strong>in</strong>g theM<strong>in</strong>orsky correlation [26] as modified by Reardon <strong>and</strong>Sprung [27].Go to the next time step:i = i+1At time step iBased on current velocities, calculate thenext positions <strong>and</strong> orientation angles of theships, <strong>and</strong> the relative motion at impactCalculate the change of impact locationalong the struck ship <strong>and</strong> the <strong>in</strong>crement ofpenetration dur<strong>in</strong>g the time stepCalculate the average reaction forces dur<strong>in</strong>gthe time step by <strong>in</strong>ternal mechanismsCalculate the average accelerations of bothships, the velocities for the next time step,<strong>and</strong> the lost k<strong>in</strong>etic energy based on externalship dynamicsNoMeet stopp<strong>in</strong>gcriteria ?YesCalculate maximumpenetration <strong>and</strong> damagelengthFigure 12 - SIMCOL Simulation Process8
ybeg<strong>in</strong>n<strong>in</strong>g of time step nStrik<strong>in</strong>g ShipP 4,n+1, P5,n+1 P1,n P1,n+1G 2Strik<strong>in</strong>g ShipG 1θ2φξP4,n, P5,nend of time step nφ ′ nP3,nαP2,nP2,n+1side shellStruck Shiplθ1Struck ShipP3,n+1damaged area dur<strong>in</strong>gtime step nNote: The positive direction of angle is alwayscounterclockwise.xFigure 14. Sweep<strong>in</strong>g Segment MethodηFigure 13. SIMCOL External Ship DynamicsV.U. M<strong>in</strong>orsky conducted the first <strong>and</strong> best known ofthe empirical collision studies based on actual data [26].His method relates the energy dissipated <strong>in</strong> a collisionevent to the volume of damaged structure. Actualcollisions <strong>in</strong> which ship speeds, collision angle, <strong>and</strong>extents of damage are known were used to empiricallydeterm<strong>in</strong>e a l<strong>in</strong>ear constant. This constant relates damagevolume to energy dissipation. In the orig<strong>in</strong>al analysis thecollision is assumed to be totally <strong>in</strong>elastic, <strong>and</strong> motion islimited to a s<strong>in</strong>gle degree of freedom. Under theseassumptions, a closed form solution for damaged volumecan be obta<strong>in</strong>ed. With additional degrees of freedom, atime-stepped solution must be used.Crake <strong>and</strong> Brown developed SIMCOL Version 0.0 aspart of the work of SNAME Ad Hoc Panel #3 [6,28].Based on further research, test runs <strong>and</strong> the need to makethe model sensitive to a broader range of design <strong>and</strong>scenario variables, improvements were progressivelymade by Chen <strong>and</strong> Brown at Virg<strong>in</strong>ia Tech [22]. Asweep<strong>in</strong>g segment method, shown <strong>in</strong> Figure 14, is addedto the model <strong>in</strong> SIMCOL Version 1.0 to improve thecalculation of damage volume <strong>and</strong> the direction ofdamage forces. Models from the Rosenblatt study [24,25]are applied <strong>in</strong> Version 1.1 assum<strong>in</strong>g rigid web frames. InVersion 2.0, the lateral deformation of web frames is<strong>in</strong>cluded as shown <strong>in</strong> Figure 15. In Version 2.1, thevertical extent of the strik<strong>in</strong>g ship bow is considered.Table 1 describes the evolution of SIMCOL over thecourse of the study thus far.<strong>Design</strong> data required for the strik<strong>in</strong>g ship <strong>in</strong>cludesbow half-entrance angle, bow height, length, beam, draft<strong>and</strong> displacement.Scenario data required <strong>in</strong>cludes strik<strong>in</strong>g <strong>and</strong> struckship velocity, collision angle, <strong>and</strong> longitud<strong>in</strong>al location ofimpact <strong>in</strong> the struck ship.Web frames act<strong>in</strong>g as a vertical beamdistort <strong>in</strong> bend<strong>in</strong>g, shear or compressionStrike at webframeAnalyze each shellseparatelyconsistent withweb deformation.Strike betweenweb frameAnalyze each shellseparately withnodes consistentwith webdeformation.Figure 15. Web Deformation <strong>in</strong> SIMCOL 2.0 [24,25]5.2.2 DAMAGE 4.0 [19]The DAMAGE 4.0 collision module solves the externalproblem uncoupled from the <strong>in</strong>ternal problem, <strong>and</strong> appliesthe calculated absorbed energy to plastic deformation ofthe struck ship. <strong>Structural</strong> components, motions, massesetc. are described <strong>in</strong> ship coord<strong>in</strong>ate systems local to eachship <strong>and</strong> <strong>in</strong> one global coord<strong>in</strong>ate system. Degrees offreedom <strong>in</strong> DAMAGE <strong>in</strong>clude strik<strong>in</strong>g ship surge <strong>and</strong>struck ship sway <strong>and</strong> yaw.The follow<strong>in</strong>g assumptions are applied <strong>in</strong> DAMAGE4.0:• Both ships are perpendicular before <strong>and</strong> dur<strong>in</strong>gimpact, i.e. only right angle collisions are considered.• The forward motion of the struck ship is assumed tobe zero.9