23rd International Towing Tank Proceedings of the 23rd ITTC – Volume II 619 Conference
The Specialist Committee on Prediction of Extreme Ship Motions and Capsizing Final Report and Recommendations to the 23rd ITTC
1. INTRODUCTION Dr. J.O. De Kat MARIN, The Netherlands Professor A. Francescutto 1.1. Membership, meetings and University of Trieste, Italy organisation Professor J. Matusiak Membership: The Committee appointed Helsinki University of Technology, by the 22nd ITTC consisted of the following Finland members: Meetings: Seven Committee meetings Professor D. Vassalos (Chairman) were held as follows: Universities of Glasgow and Strath- Shanghai, China, September 1999 clyde, UK Launceston, Australia, February 2000 Osaka, Japan, October 2000 Dr. M. Renilson (Secretary) Glasgow, Scotland, UK, May 2001 Australian Maritime College, Australia, Trieste, Italy, September 2001 and QinetiQ, Haslar, UK Heraklion, Greece, October 2001 Mr. A Damsgaard Glasgow, Scotland, UK, February 2002 Danish Maritime Institute, Denmark (Editorial meeting) Professor H.Q. Gao Organisation: The following working China Ship Scientific Research Centre, groups were established and chairmen ap- Mr. D. Molyneux pointed: Institute for Marine Dynamics, Canada Benchmark Testing for Intact Ship Sta- bility (Umeda) Professor A. Papanikolaou Benchmark Testing for Damaged Ship National Technical University of Ath- Stability (Papanikolaou) ens, Greece Guidelines for Experimental Testing of Professor N. Umeda Intact Ship Stability (de Kat) Osaka University, Japan Guidelines for Experimental Testing of Damage Ship Stability (Damsgaard) In addition, the following corresponding Questionnaire (Molyneux) members contributed greatly to the work of Symbols and Terminology (Frances- the committee: cutto) 23rd International 620 The Specialist Committee on Prediction of Exteme Ship Motions and Capsizing Towing Tank Conference
Liaisons: The following Committees and 2. BENCHMARK TESTING FOR organisations have been contacted: Loads and INTACT SHIP STABILITY Responses; Manoeuvring; Waves; IMO (Re- vision of 1966 ICLL, Intact Stability, Harmonisation Group); WEGEMT; CRN; 2.1. Introduction SNAME Technical Panel; EU Thematic Net- work − SAFER EURORO; SRA of Japan − This chapter describes results of the ITTC Panel RR71; COREDES. benchmark testing of intact stability. For these tests, a container ship and a fishing vessel were selected and their hull forms, captive test 1.2. Tasks from the 22nd ITTC data and results of capsizing model experi- ments were provided in advance. On this ba- Coordinate a comparative study of sis, eight research organisations submitted mathematical models for the prediction numerical results. Comparisons between nu- of intact and damage stability in waves. merical and experimental results revealed that The mathematical models will be com- some numerical models are able to predict pared to the results of benchmark tests extreme motions qualitatively, including cap- for two test ships, Ships A and B, as sizing due to parametric resonance and due to specified in Section 7.2 of the report of broaching. Moreover, the importance of sev- the Stability Committee of the 22nd eral factors necessary for capsize prediction is ITTC. noted by mutual comparisons of the numerical studies. Present the guidelines for experimental testing of intact and damage stability, as List of Participating Organisations given in Appendix A of the report of the Stability Committee of the 22nd ITTC, Ship A-1: in the format defined in the ITTC Qual- ity Manual. Flensburger Schiffbau Gesellschaft1 Symbols and terminology should agree (Ms. Heike Cramer) with those used in the 1999 version of Helsinki University of Technology the ITTC S&T List; if necessary, new (Prof. Jerzy Matusiak) symbols should be proposed. Maritime Research Institute Netherlands (Dr. Jan O. de Kat) 1.3. Contents of the 23rd ITTC Report Osaka University (Prof. Naoya Umeda) The following chapters detail the tasks Technical University of Malaysia (Dr. undertaken by the Committee: Adi Maimun) Chapter 2: Benchmark Testing for Intact Universities of Glasgow and Strath- Ship Stability clyde, The Ship Stability Research Cen- Chapter 3: Benchmark Testing for Dam- tre (SSRC) (Prof. Dracos Vassalos) age Ship Stability Chapter 4: Guidelines for Model Testing University of Tokyo (Prof. Masataka Fu- of Intact and Damage Stability jino). Chapter 5: Questionnaire Chapter 6: Symbols and Terminology Chapter 7: Conclusions and Recommen- dations 1 The computer program at FSG was originally Chapter 8: References and Nomenclature developed at Universitat Hamburg. 23rd International Towing Tank Proceedings of the 23rd ITTC – Volume II 621 Conference
Ship A-2: (1996). Here the ship model capsized mainly due to parametric resonance in the lower Helsinki University of Technology speed region. The second set was carried out (Prof. Jerzy Matusiak) with a 1/15 scaled model of a 135 gross ton- Memorial University of Newfoundland nes purse seiner (Ship A-2) at the seakeeping (Prof. Don Bass) and manoeuvring basin of the National Re- search Institute of Fisheries Engineering Osaka University (Prof. Naoya Umeda) (NRIFE) by Umeda et al. (1999). In these Universities of Glasgow and Strath- tests, the model capsized mainly due to clyde, The Ship Stability Research Cen- broaching in the higher speed region. The tre (Prof. Dracos Vassalos) principal particulars and body plans of these ships are shown in Table 2.1 and Figures 2.1 This order is not related to the code used and 2.2. In the experiments each ship model in this report. was self-propelled and free from any re- straints, steered on a specified course by using an auto pilot in regular following and quarter- 2.2. Background ing waves. The angular velocities and angles were measured using an optical gyroscope, The trend towards adopting performance- and were recorded on an onboard computer. based criteria in favour of rules-based criteria The reference system used in this report is aiming at safety improvement at sea continues shown in Figure 2.3. unabated at the International Maritime Or- ganisation (IMO), the rule making body of the Table 2.1 Principal particulars of the test United Nations. To facilitate this process, ships. model experiments and numerical simulations tools need to be developed and validated. Items Ship A-1 Ship A-2 However, a standard numerical prediction LPP (length) 150.0 m 34.5 m technique for capsizing has not yet been es- B (breadth) 27.2 m 7.60 m tablished. Therefore, the 22nd ITTC (ITTC, D (depth) 13.5 m 3.07 m
1999) organised a specialist committee for Tf (draught at FP) 8.5 m 2.50 m this purpose and planned benchmark testing T (mean draught) 8.5 m 2.65 m of numerical predictions with selected data T (draught at AP) 8.5 m 2.80 m from free running model experiments. This a chapter summarises the results of these Cb (block coefficient) 0.667 0.597 benchmark tests and highlights the importance kyy/LPP 0.244 0.302 of a number of factors to the numerical pre- (pitch radius of gyration) diction of ship capsizing. xCG 1.01 m 1.31 m (longitudinal position of aft aft centre of gravity from mid- 2.3. Framework of ITTC Benchmark ships) Testing GM (metacentric height) 0.15 m 1.00 m
TE (natural roll period) 43.3 s 7.4 s In the intact benchmark testing pro- 2 2 AR (rudder area) 28.11 m 3.49 m gramme, two sets of free running model ex- D (propeller diameter) 5.04 m 2.60 m periments were utilised. The first set was car- P ried out with a 1/60 scaled model of a 15000 TE (time constant of steering 1.24 s 0.63 s gross tonnes container ship (Ship A-1) at the gear) seakeeping and manoeuvring basin of the KR (proportional gain) 1.2 1.0 Ship Research Institute by Hamamoto et al. KR TD (differential gain) 53.0 s 0.0 s 23rd International 622 The Specialist Committee on Prediction of Exteme Ship Motions and Capsizing Towing Tank Conference
Among several hundreds of model runs, four runs were selected for each ship for the purpose of ITTC benchmark tests as described in Tables 2.2 and 2.3. Here the nominal Froude number, Fr, and the auto pilot course from the wave direction, χc, are control pa- rameters and the wave height, H, and wave length, λ, are the wave parameters. The initial values of ship motion were specified based on measured data except for the sway velocity, which was assumed to be zero because of Figure 2.2 Body plan of Ship A-2. measurements limitation.
For ships A-1 and A-2, the captive model X experiments, e.g. resistance test, self- propul- sion test, propeller open test, circular motion ROLL tests (CMT), roll decay test and so on, were G carried out mainly in NRIFE’s seakeeping and Y manoeuvring basin using an X-Y towing car- PITCH riage. These data together with hull offset data RUDDER and the above mentioned initial values were YAW provided to the participating organisations prior to undertaking any numerical simula- Z tions. Figure 2.3 Reference system.
The numerical predictions are firstly re- 2.4. Results quired to qualitatively agree with the corre- The ITTC benchmark test programme for sponding model experiments. Thus, the quali- intact stability commenced in March 2000 tative nature of the results obtained from ex- with numerical results submitted by March periments and numerical calculations are 2001. Numerical prediction methods used by overviewed in Tables 2.4 and 2.5. This in- the participating organisations are outlined in cludes capsize, non-capsize, harmonic roll, Umeda (2001) with numerical results shown sub-harmonic roll, surf-riding and broaching. in Figures 2.4 to 2.6 together with the experi- Here as a judging criterion of broaching the mental results. In agreement with the partici- proposal of Umeda (1999) is used. That is, pating organisations the results have been pre- broaching is a phenomenon in which both the sented anonymously throughout this bench- yaw angle and yaw angular velocity increase mark programme. despite the application of maximum opposite rudder angle. Lack of qualitative agreement between numerical and experimental results identified with shading.
Table 2.2 Calculated conditions for Ship A-1. χ H/ λ λ/L Fr c PP degrees (a) 1/25 1.5 0.2 0 (b) 1/25 1.5 0.2 45 Figure 2.1 Body plan of Ship A-1. (c) 1/25 1.5 0.3 30 (d) 1/25 1.5 0.4 30
23rd International Towing Tank Proceedings of the 23rd ITTC – Volume II 623 Conference
Table 2.3 Calculated conditions for Ship A-2. using available databases instead of the cap- tive test data provided. χc H/ λ λ/LPP Fr degrees Table 2.4 Overview of qualitative results (a) 1/10 1.637 0.30 -30 for Ship A-12. (b) 1/10 1.637 0.43 -10 (c) 1/8.7 1.127 0.30 -30 Experiment A B C (d) 1/8.7 1.127 0.43 -30 (a) cap (s) cap (s) cap (h) no roll (b) (s) (s) (s) N/A 2.5. Discussion (Tables 2.6-2.7 & Figures 2.4-2.6) (c) (h) (h) N/A N/A (d) cap (h) cap (h) N/A N/A Ship A-1 D E F G For Ship A-1 all the participating organi- sations used 6 degrees of freedom (DOF) (a) (s) cap (s) cap (h) cap (h) models. However, only Organisation-A sub- (b) (s) (h) (h) (s) mitted results that qualitatively agree with the (c) (h) cap cap (h) experiments. (d) (h) cap cap cap (h) Organisation-A calculated radiation and diffraction forces using a strip theory and The method used by Organisation-B is dealt with manoeuvring forces by the MMG based on a conventional seakeeping approach. model, utilizing a body coordinate system. It That is, heave, pitch, sway and yaw are as- evaluated the Froude-Krylov forces, including sumed to be linear around the averaged roll restoring moment in waves, by integrating course. This organisation reported that this incident wave pressure up to the instantaneous method is not able to deal with ship runs at water surface. With this numerical model, Froude number greater than 0.3. capsizing with sub-harmonic rolling in case Organisation-D proposed a method to (a) and capsizing with harmonic rolling in avoid this limitation of the seakeeping model case (d) were well predicted. by a two-stage approach. Here the motions are Organisation-G also shows similar agree- assumed to be the sum of linear parts with ment but numerical results in case (a) predict hydrodynamic memory effects and nonlinear capsizing with harmonic rolling, which was contributions. This means that linear motion not observed in the corresponding experiment. was calculated around the instantaneous head- The method used here is almost the same as ing angle instead of the auto pilot course. that of Organisation-A except for radiation However, agreement between predictions us- and diffraction modelling. ing this calculation and experiments is not satisfactory. This may be partly because the Organisation-E has problems in the pre- initial values used in order to take memory diction of the heading angle. In some cases effects into account are different to those the ship course changes to bow sea and then a specified. completely different situation occurs. This model is different from the above two organi- Organisation-F is a unique example where sations in a number of ways. The radiation diffraction forces were ignored, but the nu- forces were calculated using a a strip theory 2 with hydrodynamic memory effects. The ma- Here (h) and (s) mean harmonic and sub- noeuvring forces, roll damping moments, re- harmonic roll motions, respectively and cap sistance and propulsion forces were estimated indicates capsizing. 23rd International 624 The Specialist Committee on Prediction of Exteme Ship Motions and Capsizing Towing Tank Conference
merical results do not agree well with the ex- Table 2.5 Overview of qualitative results perimental results, particularly the case (b) for Ship A-23. concerning calculated pitch motion ampli- tudes. exp. A B C D non- non- non- non- non- (a) Following successful application to cap cap cap cap cap seakeeping predictions, Organisation-C at- tempted to apply CFD to the present problem. surf surf surf Here the Euler equation was solved by a finite (b) broach broach cap cap non- difference method with fully nonlinear free cap cap cap non- non- non- non- non- surface and body surface conditions. How- (c) ever, it can provide a solution only for case (a) cap cap cap cap cap without lateral motions. If the specified initial (d) cap cap cap cap cap values for lateral motions are input, even for case (a) the calculation process failed. In addi- Organisation-B applies a 6 DOF model in tion, it cannot deal with cases (b), (c) and (d), which radiation and diffraction were calcu- in which the desired heading angles are not lated with a 3D Green function for zero for- zero. This fact demonstrates that the CFD ap- ward velocity. Here the change of roll restor- proach is not yet appropriate for practical use ing moment due to waves was taken into ac- in capsize prediction. count but the hydrodynamic lift due to wave fluid velocity was ignored. The hydrodynamic Ship A-2 memory effect was included in this calcula- tion, although the initial values were not ex- For Ship A-2, only Organisation-A ob- actly as specified. While the predictions of tained qualitative agreement with experi- mean yaw angle for cases (a), (c) and (d) are ments. Here a 4 DOF model was used by as- better than those from the other organisations, suming that heave and pitch motions trace the predicted rudder angle for case (b) is their static equilibria, which are calculated as smaller than the corresponding experimental the limit of solution sets of a strip theory at results. zero encounter frequency. The manoeuvring forces were estimated using the MMG model Organisation-D also takes memory effects and the wave-induced forces, including hy- into account but with a strip theory. Like or- drodynamic lift due to wave fluid velocity, ganisation-B, the initial conditions are differ- were calculated using Ohkusu’s slender body ent from those specified. This organisation theory. The wave effects on roll restoring predicts stable surf-riding in the case (b). This moment and manoeuvring forces were ig- may be a result of some shift of the stable nored as higher order terms. As a result, this equilibrium point towards a wave crest or in- organisation succeeded in predicting capsiz- accuracy of hydrodynamic lift due to wave. ing due to broaching associated with surf- As a whole, the four participating organi- riding as well as periodic motions. sations predicted the results relatively well Organisation-C used a method that is al- compared to experiments for Ship A-2, the most the same as that of Organisation-A but obvious exception being broaching. the nonlinear terms in the manoeuvring mod- els, deriving from the Froude-Krylov and ra- diation forces were added. As a result, for case (b) it predicted capsizing without surf- riding and with a smaller rudder angle com- pared to the results from the experiment and 3 Here surf and broach mean surf-riding and those predicted by Organisation-A. broaching, respectively. 23rd International Towing Tank Proceedings of the 23rd ITTC – Volume II 625 Conference
2.6. Factors Affecting Prediction Hydrodynamic memory effect Accuracy It is well known that the linear transient As mentioned above, the mathematical motions of a ship with frequency-dependent models for capsizing prediction involve a hydrodynamic forces can be calculated using number of factors without clear guidance in the convolution integral for hydrodynamic place on which of these should be taken into memory effect. However, it is not so clear for account in which case. Mutual comparisons capsizing prediction whether the hydrody- among the organisations do not easily clarify namic memory effect should be taken into ac- the importance of each particular factor be- count or not. This is because an extreme mo- cause more than two factors are often differ- tion leading to capsizing is nonlinear and the ent between the organisations. Therefore, this hydrodynamic forces acting on a ship running report reviews comparative studies of numeri- in following and quartering seas do not sig- cal simulations with and without each particu- nificantly depend on the encounter frequency. lar factor for Ships A-1, A-2 or indeed other ships. Hamamoto & Saito (1992) carried out a 6 DOF vs. 4 DOF or 1 DOF comparative study for a container ship in fol- lowing seas with and without the memory ef- Although all organisations submitted re- fect in heave and pitch motions. They con- sults with 6 DOF models for Ship A-1, many cluded that no significant difference exists if theoretical studies with 1 DOF models can be the added mass and damping coefficients are found for capsizing due to parametric rolling. calculated for the natural frequency of heave Munif (2000) estimated the capsizing bounda- and pitch motions. Matusiak (2001) investi- ries for Ship A-1 with a 1 DOF model, a 4 gated this problem and concluded that mem- DOF model ignoring heave and pitch motions ory effects can improve agreement with ex- (4 DOF A model), a 4 DOF model with static periments for Ship A-1. Here it is noteworthy equilibria of heave and pitch motions (4 DOF that exact calculation with memory effects B model) and a 6 DOF model. Here the first should be carried out from the start of the three models were obtained by simplifying the waves. Thus the present benchmark testing, 6 DOF model. As a result, the following con- which does not specify the initial conditions clusions were drawn: of fluid motions, is not appropriate for this The 1 DOF model overestimates capsiz- purpose. ing danger. The difference between the 4 DOF A Manoeuvring coefficients model and the 6 DOF model can be sig- nificant. In following and quartering waves, predic- The results from the 4 DOF B model are tion of manoeuvring coefficients is important in reasonable agreement with those from because hydrodynamic lift is dominant. The the 6 DOF model and the experiment. first question here is whether the effect of The small difference between the 4 DOF nonlinear terms of manoeuvring forces on B model and the 6 DOF model derives capsizing prediction is important or not. For from the fact that the natural frequency Ship A-2, Umeda et al. (2000) produced time of heave and pitch motions is far from domain simulations with and without these the encounter frequency with the ship non-linear terms and concluded that the effect running in following and quartering seas of non-linear terms is negligibly small. This is (Matsuda et al., 1997). This conclusion because the sway velocity and yaw angular suggests also that coupling effects of velocity non-dimensionalised with the higher heave and pitch on the extreme roll mo- forward velocity are not large even during the tion are not very important. process of broaching. 23rd International 626 The Specialist Committee on Prediction of Exteme Ship Motions and Capsizing Towing Tank Conference
The next problem is wave effect on the Hydrodynamic lift due to wave fluid ve- linear manoeuvring coefficients. This problem locity has been discussed for many years but its ef- fect on capsizing prediction has not yet been Very small effect of wave-making does fully investigated. Hashimoto & Umeda not mean small effect of incident waves as a (2001) tackled this problem with Ship A-2. ship behaves like a lifting surface with a time- Their main conclusion is that the effect of varying angle of attack due to wave fluid ve- waves on the derivatives of manoeuvring locity and ship forward velocity. Within the forces can be important with respect to sway assumption of small wave steepness, this hy- velocity but it is not significant with respect to drodynamic lift can be calculated as an end yaw angular velocity. term of slender body theory or strip theory, which represents trailing vortices as a line Nonlinearity in yaw doublet shed from the aft end (Umeda, 1988). For a 3D theory, it is necessary to include free In seakeeping theory, ship motions, such vortex layers shed from the hull surface. as yaw, are often linearised around the inertial Comparison between calculations with and system moving with the averaged speed and without the hydrodynamic lift due to wave course of a ship. In the field of manoeuvring fluid velocity for Ship A-2 can be found in on the other hand, ship motions are described Umeda (2000). The results indicate that pre- with a body fixed coordinate system. Hama- diction of broaching is largely affected by this moto & Kim (1993) introduced a horizontal term. body coordinate system, which is body-fixed but not allowed to roll. Cramer (2001) re- Roll damping moment ported the limitation of linearisation of yaw motion with an inertial coordinate system. Roll damping moment consists of wave- making, eddy-making, lift and friction com- Radiation and diffraction ponents, the main non-linearity deriving from the eddy-making component. However, as the In following and quartering seas, the en- experimental work of Umeda (2000) showed, counter frequency of a ship with forward roll damping can be regarded as linear when speed is generally low and hence the wave- the Froude number is greater than 0.2. Since making effect is not so significant. In this re- eddies are shed away at high speed, the eddy spect, it is difficult to predict pitch and heave making component disappears. In addition, motions near zero encounter frequency be- the wave-making component is not significant cause of divergence of the 2D added mass. because of the low encounter frequency and Matsuda et al. (1997) solved the problem by the friction component is generally small. calculating the limit of the solution set of strip Therefore, roll damping relating to this theory for the zero encounter frequency and benchmark testing scheme consists of mainly confirmed that the new method explains the the lift component, which is linear and de- experimental results. In case of a 3D theory at pends on forward velocity. A comparison of very low encounter frequency, it is essential to predictions of broaching boundary using em- use the Green function with both forward pirical methods of calculating the lift compo- speed effect and frequency effect taken into nent was presented by Ikeda et al. (1988) for account. This is because the wave related to the Ship A-2, indicating that the predicted re- frequency, the k2 wave, disappears at the zero sults depend on the selection of empirical encounter frequency and only the wave related methods. Because of this, roll decay tests with to forward speed, the k1 wave, remains. forward velocity were carried out for the Ship A-2 to obtain reliable results for use in the benchmark study. 23rd International Towing Tank Proceedings of the 23rd ITTC – Volume II 627 Conference
Wave effect on roll restoring moment model tests were provided in advance. How- ever, since such data are not always available, While all organisations took into account an accurate prediction method is still desirable. the wave effect on roll restoring moment to simulate parametric resonance, some organi- Wave irregularity and short-crestedness sations ignored this effect for broaching pre- diction. This is because the wave effect esti- The applicability of numerical models to mated on the basis of hydrostatics can be too realistic seaways, that is, short-crested irregu- large in case of high forward velocity. Results lar waves, should be examined in the future. from captive model experiments (Umeda & Although capsizing model experiments for Yamakoshi, 1986) for a small trawler indicate Ships A-1 and A-2 were carried out in both that the measured metacentric height in waves long-crested and short-crested irregular waves is smaller than the calculated value, derived (Umeda et al., 1995), the benchmark testing by using hydrostatics. It is, therefore, neces- programme deals with only the case of regular sary to develop theoretical or empirical meth- waves. The experimental results indicate that ods for a more accurate prediction of the wave capsizing danger is least in short-crested ir- effect on roll restoring, particularly for higher regular waves, followed by long-crested and forward speeds. finally regular waves. However, numerical simulations in time domain for capsizing in Roll-yaw coupling short-crested irregular waves are very limited. Only recently Sera & Umeda (2001) executed When a ship runs in calm water with a numerical calculation in both short-crested constant heel angle, sway force, yaw moment and long-crested irregular waves with a 1 and roll moment act on the hull in addition to DOF model, and confirmed the qualitative conventional manoeuvring forces and mo- conclusion, from the experiments, that wave ments. In this benchmarking scheme, these short-crestedness reduces capsizing danger. data from the captive tests for Ship A-2 at NRIFE were provided in advance. However, such data are not always available and em- 2.7. Concluding Remarks pirical or theoretical methods have not yet been established. Renilson & Manwarring As a result of benchmark testing of intact (2000) reported a comparison in predictions stability, it was found that numerical models of broaching boundary with and without the can qualitatively predict capsizing due to pa- roll-yaw coupling for a trawler and the re- rametric resonance and due to broaching in sults indicate that the prediction without the the limited cases tested here. In the case of roll-yaw coupling can underestimate the parametric resonance, a 6 DOF model, hydro- danger of broaching. statics for the wave effect on roll restoring moment, strip theory for wave radiation and Resistance and propulsion diffraction and experimental data of manoeu- vring forces, hull resistance, propulsion force It is well accepted that predictions of hull and roll damping are used. In the case of resistance and propulsive performance have broaching, a 4 DOF model, slender body the- been the most crucial issue at ITTC. This is ory for the added mass and hydrodynamic lift also the case in the prediction of capsizing of due to wave fluid velocity and experimental intact ships. This is because surf-riding, which data of manoeuvring forces, hull resistance, can trigger off broaching and then capsizing, propulsion force and roll damping are used. depends on hull resistance and propeller thrust However, minimum requirements for accurate in addition to wave-induced surge force. In the modelling of intact ship capsize have not yet benchmarking study, data from calm-water been fully established. 23rd International 628 The Specialist Committee on Prediction of Exteme Ship Motions and Capsizing Towing Tank Conference
Table 2.6 Comparison of numerical prediction methods for Ship A-1. Manoeuvring Organisations DOF Radiation Roll Restoring Damping exp. hydrostatics in A 6 strip theory (non-linear) waves (linear in sway, hydrostatics in B 6 heave, pitch, strip theory ignored waves yaw) (coupled with CFD CFD C 6 ignored fluid motion) (non-linear) (non-linear) (two stage hydrostatics in D 6 strip theory exp. (linear) model) waves hydrostatics in E 6 strip theory empirical wave hydrostatics in F 6 empirical empirical waves hydrostatics in G 6 strip theory exp. (non-linear) waves Hydrodynamic Organisations Roll Damping Froude-Krylov Diffraction Lift due to Hull Resistance Wave experimental+ A empirical non-linear strip theory ignored experimental forward speed B empirical linear strip theory ignored experimental non-linear CFD C no viscous effect ignored no viscous effect (cfd) (non-linear) D empirical non-linear strip theory ignored experimental E empirical non-linear strip theory cross-flow model empirical F empirical non-linear ignored ignored empirical experimental+ slender body G empirical non-linear end term experimental theory at ω=0 forward speed Propeller Hydrodynamic Specified Initial Organisations Rudder Force Incident Wave Thrust Memory Effect Conditions A experimental experimental linear ignored yes B adjusted ignored linear ignored yes C adjusted ignored linear included no D experimental empirical linear included no E empirical empirical linear included no F empirical empirical linear ignored yes G experimental experimental linear ignored yes
23rd International Towing Tank Proceedings of the 23rd ITTC – Volume II 629 Conference
Table 2.7 Comparison of numerical prediction methods for Ship A-2.
Manoeuvring Roll Roll Froude- Organisations DOF Radiation Diffraction Damping Damping Restoring Krylov slender exp.+ static slender body experimental empirical hydrostatics A 4 heave & linear body theory theory at (linear) forward in calm water pitch at ω=0 ω=0 speed effect 3D theory 3D theory （Green empirical + hydrostatics non- （Green B 6 ignored function at tuned in waves linear function at Fr=0） Fr=0） exp.+ static slender experimental empirical hydrostatics non- C 4 heave & strip theory body theory (non-linear) forward in calm water linear pitch at ω=0 speed effect two experimental hydrostatics non- D 6 stage strip theory empirical strip theory (linear) in waves linear model Hydro- Hydro- Specified dynamic Hull Propeller Rudder Incident dynamic Organisations Initial Con- Lift due to Resistance Thrust Force Wave Memory dition Waves Effect A end term exp. exp. exp. linear ignored yes B ignored empirical tuned empirical non-linear included no C ignored exp. exp. exp. linear ignored yes D ignored exp. exp. empirical linear included no
To improve quantitative prediction accu- rately predict ship capsize. For wider valida- racy, it is essential that the contribution of tion studies, it is desirable to execute bench- several factors should be further investigated mark tests concerning the prediction of cap- through comparative studies of capsizing pre- sizing boundary curves as shown by Munif dictions with and without these factors being (2000) for Ship A-1 and by Umeda & Hashi- accounted for. These should include wave ef- moto (2002) for Ship A-2. This is because fect on manoeuvring coefficients and roll re- capsizing boundaries can be complicated as a storing moment, roll damping prediction, ra- result of system nonlinearities. Furthermore, diation and diffraction and resistance at higher practical application to ship design, operation speeds with use of captive model experi- and regulation necessitates the extension of ments. It is noteworthy that stability predic- the current predictive capability to more real- tion should be based on accurate predictions istic seaways. of seakeeping, manoeuvring, resistance and propulsion, subjects which have been the fo- This, in turn, warrants the undertaking of a cus of considerable effort by the relevant new benchmark testing of numerical codes ITTC technical committees over many years. with relevant experimental data from the most This alone offers ample justification of the advanced model basins having multi- problems encountered in attempting to accu- component wave makers. 23rd International 630 The Specialist Committee on Prediction of Exteme Ship Motions and Capsizing Towing Tank Conference
Experiment Calculation Experiment Calculation
Pitch(degrees) Pitch(degrees) Pitch(degrees) Pitch(degrees) 10 10 10 10 5 5 5 5 0 0 0 0 0 10 20 30 0 10 20 30 0 20 40 60 0 20 40 60 -5 t(s) -5 t(s) -5 t(s) -5 t(s) -10 -10 -10 -10 Roll(degrees) Roll(degrees) Roll(degrees) Roll(degrees) 90 90 90 90 60 60 60 60 30 30 30 30 0 0 0 0 -300 10 20 30 -300 10 20 30 -300 20 40 60 -300 20 40 60 -60 t(s) -60 t(s) -60 t(s) -60 t(s) -90 -90 -90 -90 Yaw(degrees) Yaw(degrees) Yaw(degrees) Yaw(degrees) 45 45 60 60 30 30 40 40 15 15 20 20 0 0 0 0 -150 10 20 30 -150 10 20 30 -200 20 40 60 -200 20 40 60 -30 t(s) -30 t(s) -40 t(s) -40 t(s) -45 -45 -60 -60 Rudder(degrees) Rudder(degrees) Rudder(degrees) Rudder(degrees) 15 15 15 15 10 10 10 10 5 5 5 5 0 0 0 0 -50 10 20 30 -50 10 20 30 -50 20 40 60 -50 20 40 60 -10 t(s) -10 t(s) -10 t(s) -10 t(s) -15 -15 -15 -15 (a) (b)
Experiment Calculation Experiment Calculation
Pitch(degrees) Pitch(degrees) Pitch(degrees) Pitch(degrees) 10 10 10 10 5 5 5 5 0 0 0 0 0 10 20 30 40 0 10 20 30 40 0 5 10 15 20 0 5 10 15 20 -5 t(s) -5 t(s) -5 t(s) -5 t(s) -10 -10 -10 -10 Roll(degrees) Roll(degrees) Roll(degrees) Roll(degrees) 90 90 90 90 60 60 60 60 30 30 30 30 0 0 0 0 -300 10 20 30 40 -300 10 20 30 40 -300 5 10 15 20 -300 5 10 15 20 -60 t(s) -60 t(s) -60 t(s) -60 t(s) -90 -90 -90 -90 Yaw(degrees) Yaw(degrees) Yaw(degrees) Yaw(degrees) 60 60 60 60 40 40 40 40 20 20 20 20 0 0 0 0 -200 10 20 30 40 -200 10 20 30 40 -200 5 10 15 20 -200 5 10 15 20 -40 t(s) -40 t(s) -40 t(s) -40 t(s) -60 -60 -60 -60 Rudder(degrees) Rudder(degrees) Rudder(degrees) Rudder(degrees) 15 15 15 15 10 10 10 10 5 5 5 5 0 0 0 0 -50 10 20 30 40 -50 10 20 30 40 -50 5 10 15 20 -50 5 10 15 20 -10 t(s) -10 t(s) -10 t(s) -10 t(s) -15 -15 -15 -15
Organisation-A Organisation-E Organisation-G
(c) (d) Figure 2.4 Experimental results and numerical results for Ship A-1 from three organisations. 23rd International Towing Tank Proceedings of the 23rd ITTC – Volume II 631 Conference
Experiment Calculation Experiment Calculation
Pitch(degrees) Pitch(degrees) Pitch(degrees) Pitch(degrees) 10 10 10 10 5 5 5 5 0 0 0 0 0 10 20 30 0 10 20 30 0 20 40 60 0 20 40 60 -5 t(s) -5 t(s) -5 t(s) -5 t(s) -10 -10 -10 -10 Roll(degrees) Roll(degrees) Roll(degrees) Roll(degrees) 90 90 90 90 60 60 60 60 30 30 30 30 0 0 0 0 -300 10 20 30 -300 10 20 30 -300 20 40 60 -300 20 40 60 -60 t(s) -60 t(s) -60 t(s) -60 t(s) -90 -90 -90 -90 Yaw(degrees) Yaw(degrees) Yaw(degrees) Yaw(degrees) 45 45 90 90 30 30 60 60 15 15 30 30 0 0 0 0 -150 10 20 30 -150 10 20 30 -300 20 40 60 -300 20 40 60 -30 t(s) -30 t(s) -60 t(s) -60 t(s) -45 -45 -90 -90 Rudder(degrees) Rudder(degrees) Rudder(degrees) Rudder(degrees) 15 15 15 15 10 10 10 10 5 5 5 5 0 0 0 0 -50 10 20 30 -50 10 20 30 -50 20 40 60 -50 20 40 60 -10 t(s) -10 t(s) -10 t(s) -10 t(s) -15 -15 -15 -15 (a) (b)
Experiment Calculation Experiment Calculation
Pitch(degrees) Pitch(degrees) Pitch(degrees) Pitch(degrees) 10 10 10 10 5 5 5 5 0 0 0 0 0 10 20 30 40 0 10 20 30 40 0 5 10 15 20 0 5 10 15 20 -5 t(s) -5 t(s) -5 t(s) -5 t(s) -10 -10 -10 -10 Roll(degrees) Roll(degrees) Roll(degrees) Roll(degrees) 90 90 90 90 60 60 60 60 30 30 30 30 0 0 0 0 -300 10 20 30 40 -300 10 20 30 40 -300 5 10 15 20 -300 5 10 15 20 -60 t(s) -60 t(s) -60 t(s) -60 t(s) -90 -90 -90 -90 Yaw(degrees) Yaw(degrees) Yaw(degrees) Yaw(degrees) 60 60 60 60 40 40 40 40 20 20 20 20 0 0 0 0 0 10 20 30 40 0 10 20 30 40 -20 -20 -200 5 10 15 20 -200 5 10 15 20 -40 t(s) -40 t(s) -40 t(s) -40 t(s) -60 -60 -60 -60 Rudder(degrees) Rudder(degrees) Rudder(degrees) Rudder(degrees) 15 15 15 15 10 10 10 10 5 5 5 5 0 0 0 10 20 30 40 0 10 20 30 40 0 0 -5 -5 -50 5 10 15 20 -50 5 10 15 20 -10 t(s) -10 t(s) -10 t(s) -10 t(s) -15 -15 -15 -15
Organisation-B Organisation-C Organisation-D Organisation-F (c) (d) Figure 2.5 Experimental results and numerical results for Ship A-1 from four organisations. 23rd International 632 The Specialist Committee on Prediction of Exteme Ship Motions and Capsizing Towing Tank Conference
Experiment Calculation Experiment Calculation
Pitch(degrees) Pitch(degrees) Pitch(degrees) Pitch(degrees) 20 20 20 20 10 10 10 10 0 0 0 0 0 5 10 15 20 0 5 10 15 20 0 5 10 15 20 0 5 10 15 20 -10 t(s) -10 t(s) -10 t(s) -10 t(s) -20 -20 -20 -20 Roll(degrees) Roll(degrees) Roll(degrees) Roll(degrees) 90 90 90 90 60 60 60 60 30 30 30 30 0 0 0 0 -300 5 10 15 20 -300 5 10 15 20 -300 5 10 15 20 -300 5 10 15 20 -60 t(s) -60 t(s) -60 t(s) -60 t(s) -90 -90 -90 -90 Yaw(degrees) Yaw(degrees) Yaw(degrees) Yaw(degrees) 60 60 60 60 40 40 40 40 20 20 20 20 0 0 0 0 -200 5 10 15 20 -200 5 10 15 20 -200 5 10 15 20 -200 5 10 15 20 -40 t(s) -40 t(s) -40 t(s) -40 t(s) -60 -60 -60 -60 Rudder(degrees) Rudder(degrees) Rudder(degrees) Rudder(degrees) 45 45 45 45 30 30 30 30 15 15 15 15 0 0 0 0 -150 5 10 15 20 -150 5 10 15 20 -150 5 10 15 20 -150 5 10 15 20 -30 t(s) -30 t(s) -30 t(s) -30 t(s) -45 -45 -45 -45 (a) (b)
Experiment Calculation Experiment Calculation
Pitch(degrees) Pitch(degrees) Pitch(degrees) Pitch(degrees) 20 20 20 20 10 10 10 10 0 0 0 0 0 5 10 15 20 0 5 10 15 20 0 5 10 15 20 0 5 10 15 20 -10 t(s) -10 t(s) -10 t(s) -10 t(s) -20 -20 -20 -20 Roll(degrees) Roll(degrees) Roll(degrees) Roll(degrees) 90 90 90 90 60 60 60 60 30 30 30 30 0 0 0 0 -300 5 10 15 20 -300 5 10 15 20 -300 5 10 15 20 -300 5 10 15 20 -60 t(s) -60 t(s) -60 t(s) -60 t(s) -90 -90 -90 -90 Yaw(degrees) Yaw(degrees) Yaw(degrees) Yaw(degrees) 90 90 60 60 60 60 40 40 30 30 20 20 0 0 0 0 0 5 10 15 20 0 5 10 15 20 -30 -30 -200 5 10 15 20 -200 5 10 15 20 -60 t(s) -60 t(s) -40 t(s) -40 t(s) -90 -90 -60 -60 Rudder(degrees) Rudder(degrees) Rudder(degrees) Rudder(degrees) 45 45 45 45 30 30 30 30 15 15 15 15 0 0 0 5 10 15 20 0 5 10 15 20 0 0 -15 -15 -150 5 10 15 20 -150 5 10 15 20 -30 t(s) -30 t(s) -30 t(s) -30 t(s) -45 -45 -45 -45
Organisation-A Organisation-B Organisation-C Organisation-D (c) (d) Figure 2.6 Experimental results and numerical results for Ship A-2 from four organisations. 23rd International Towing Tank Proceedings of the 23rd ITTC – Volume II 633 Conference
3. BENCHMARK TESTING FOR number of incoming regular beam waves and DAMAGE SHIP STABILITY the damage ship capsizing boundaries in ir- regular beam seas for varying KG values.
3.1. Introduction Lack of uniformity in the interpretation of the initial guidelines and specifications (Pa- This chapter highlights the results of the panikolaou & Spanos, 2001) led to specifica- benchmark testing on damage stability for tions being revised (Vassalos, Umeda and Pa- Ship B-2 of the 22nd ITTC (ITTC, 1999). The panikolaou, 2001) and resubmission of results benchmark test programme commenced in (Papanikolaou, 2001; Jasionowski, 2001). March 2000, inviting ITTC member organisa- tions and other qualified research institutions Details of the experimental results and to express their interest in participating at the ship characteristics are reported by launched study. Based on this call, five or- Jasionowski & Vassalos (2001). ganizations submitted numerical results. Despite the fact that finally only a small The selected ship for this investigation is a number of organisations were able to partici- Ro-Ro Passenger vessel, which has been pate in the benchmark study and that only one model tested in the Denny Tank (the test tank ship could be benchmark tested during, it is of the Ship Stability Research Centre) accord- felt that the final outcome enabled the draw- ing to the Model Test Method of IMO ing of some important conclusions. Even if SOLAS ’95 Resolution 14 (Vassalos and some numerical results cannot be considered Jasionowski, 2000). The condition addressed satisfactory, compared to model experiment in the study concerns the midship damage as results, the study clearly identified advances described in SOLAS (1997). Wave conditions and gaps in the present state of knowledge on were set in the benchmark guidelines (Umeda the prediction of extreme ship motions and & Papanikolaou, 2000) as well as the range of capsizing of damaged ships leading to rec- tests in regular and irregular seas, as shown in ommendations on further studies concerning Table 3.1. In particular it was requested to in- specifically identified problem areas. vestigate the intact ship performance for a
Table 3.1 Overview of benchmark test results.
Participants × results available; − results not available P1 P2 P3 P4 P5 GZ curves (intact/damaged) × × × × × Simulated free roll decay curves (intact) × × × × × Simulated free roll decay curves (damaged) − × × × × Simulated frequency roll response curves × × × × × (intact/damaged RAOs), constant wave height Simulated roll response curves (intact), × × × × constant wave slope (1:25) − Simulated survivability boundaries × × × × × (0% capsize, 100% capsize for KG=12.892 m) Wave, roll, heave, water on deck time series for KG=12.892 m × × × × × and Hs=4.0 m, 4.25 m and 4.5 m (5 runs) 23rd International 634 The Specialist Committee on Prediction of Exteme Ship Motions and Capsizing Towing Tank Conference
List of Participants. The following five or- Radiation and wave diffraction forces are ganisations participated in the ITTC ship B-2 calculated by a number of approaches, all in damage stability benchmark study: the framework of potential flow theory, as in- dicated in Table 3.2. Flensburger Schiffbau Gesellschaft, FSG (Ms. Heike Cramer) Maritime Research Institute Netherlands Additional information for the calculation (Dr. Jan O. de Kat) of damping forces is shown in Table 3.3. National Technical University of Athens (Prof. Apostolos Papanikolaou) The floodwater is generally considered as Osaka University (Prof. Naoya Umeda) independent variable masses moving inside Universities of Glasgow and Strath- the flooded compartments and interacting clyde, SSRC (Prof. Dracos Vassalos). with the ship. Modelling of floodwater motion can be generally described by the modelling of the floodwater free surface condition, as 3.2. Software employed shown in Table 3.4. All the software employed is non-linear time domain codes. Six degrees of freedom Table 3.4 Floodwater free-surface model- (DOF) models are used by all participants, ling. except for Participant 3 where a three DOF model is used (sway, heave and roll). Participant Approach P1 Plane and free movable (when Table 3.2 Potential theory approaches to period away from natural); ship motions employed by participants. Glimm’s equations (when period close to natural) Participant Approach P2 Plane and free movable P1 Strip theory, 6 DOF P3 Plane and horizontal P2 3D source panel theory, 6 DOF P4 Plane and free movable P3 Newly modified strip theory, 3 DOF P5 Plane and horizontal P4 3D source panel theory, 6 DOF Water ingress/egress through the damage P5 Strip theory, 6 DOF opening is commonly based on hydraulic models following application of Bernoulli’s Table 3.3 Modelling of damping forces. dynamic pressure head equation. All partici- Participant Approach pants used empirically determined coefficients to take into account the actual water in- P1 Non-linear roll damping according gress/egress flow through the specified dam- to P. Blume age opening. P2 Equivalent linear roll damping estimated from the available intact ship roll decay measurements P3 Non-linear roll damping according 3.3. The Test Ship to Ikeda P4 Adaptive-linear roll damping The general particulars of the test ship in according to Ikeda full and model scale are given in Table 3.5. P5 Non-linear (equivalent linear & The model scale is 1:40. quadratic) roll damping based on a modified Ikeda’s approach 23rd International Towing Tank Proceedings of the 23rd ITTC – Volume II 635 Conference
Table 3.5 Main ship particulars analysis of records of free rolling tests in damaged condition. Full Model Scale Scale Pitch radius of gyration: iyy/LPP = 0.217 (iyy = 872 mm). Radius iyy was estimated LOA 179.00 m 4,475.0 mm by analysis of free pitching experiments LBP 170.00 m 4,250 mm in air (Vassalos and Jasionowski, 2000). B 27.80 m 695.0 mm Yaw Radius of Gyration: izz/LPP = 0.238 T 6.25 m 156.3 mm (iyy = 960 mm). Radius izz was assumed DCARDECK 9.00 m 225.0 mm to be 10% greater than pitch radius of Displacement 17,300 gyration. This increase can be justified 270.3 kg (even keel) tonnes by the fact that for models in a damaged condition the mass of superstructures is Intact KG 12.89 m 322.0 mm normally absent and hence the lateral (above BL) mass distribution of the model is ex- Intact Design 2.63 m 65.8 mm pected to be greater than the vertical GM one. Figure 3.1 depicts the damaged case con- The ship was studied in intact and damage sidered in the benchmark study with the GZ condition. Some details of the model charac- curves in intact and damaged cases given in teristics pertaining to these conditions are Figures 3.2 and 3.3, respectively. given next.
Metacentric height (intact): GM = 65.76 3.4. GZ curves mm. This is determined by inclining ex- periment. The accuracy of hydrostatic calculations of Roll radius of gyration, ixx/B= 0.235 the numerical codes used in the benchmark (ixx= 163 mm). The roll radius of gyra- depends on the discretisation of the ship’s ge- tion ixx was estimated by analysis of the ometry. This explains the differences ob- free roll decay measurements for the in- served in Figures 3.2 and 3.3, which should be tact condition and given to the partici- borne in mind when analysing the predicted pants. This radius refers to the inertia of ship responses by the participating organisa- ship structural mass (derived from the tions, as any inaccuracy in the geometry and relevant decay measurements by ac- ship hydrostatics will affect the estimated counting for the hydrodynamic added stiffness (restoring ability) of the ship and inertia) and is a characteristic constant hence her natural frequencies. In the intact of the model, for the specific loading ship case the observed differences are minor. condition. However, in the damaged case, the GZ curve by Participant 3 shows higher initial stiffness Intact natural roll period: T = 2.056 ni for the flooded ship, whereas the range of sta- seconds. This period was determined by bility computed by Participants 2 and 5 is no- analysis of records of free rolling tests in ticeably lower. Only Participants 1 and 4 intact condition. properly capture the hydrostatic properties of Damaged natural roll period: Tnd = 2.300 the benchmark ship over the entire stability seconds. This period was determined by range.
23rd International 636 The Specialist Committee on Prediction of Exteme Ship Motions and Capsizing Towing Tank Conference
Figure 3.1 Midship Damage Case (Vassalos & Jasionowski, 2000).
1.1 0.4 1
0.9 0.3 0.8
0.7 0.2 0.6 Participant 1 0.5 0.1 Participant 2 0.4 Participant 3 0.3 0 Participant 4 0 5 10 15 20
0.2 [m] lever Righting
Righting lever [m] lever Righting Participant 5 Participant 1 0.1 -0.1 Participant 2 0 Participant 3 0 102030405060 -0.1 -0.2 Participant 4 -0.2 Participant 5
-0.3 -0.3 Heel angle [deg] Heel angle [deg]
Figure 3.2 Computed GZ curves for the in- Figure 3.3 Computed GZ curves for the tact ship. damaged ship. 23rd International Towing Tank Proceedings of the 23rd ITTC – Volume II 637 Conference
3.5. Free Rolling Simulations (b) Inaccurate representation of the floodwa- ter dynamics and its coupling with ship The results of free rolling simulations are motion shown in Figure 3.4. Simulation of free roll (c) Possible inconsistencies in the available decay in the intact condition presents no diffi- experimental data (clarification of ex- culty, as generally good agreement with the perimental conditions and way of analysis experiments was achieved by all participants. of data). However, similar attempts to simulate the free roll response in the damaged condition were As a first step towards improving this less successful. Results presented by all par- situation, it would seem necessary to under- ticipants show a distinctive overestimation of take a new benchmark study in the future for the natural roll frequency. In this respect, at least another ship case and to perform addi- three possible sources for this discrepancy tional experimental verifications of free roll may be cited: tests in damaged conditions. (a) Lack of understanding of the complete hydrodynamics of the damaged ship
PRR1, KG=12.892m, Free roll decay, Intact condition
8 EXPERIMENT
6 Participant 1
4
2 g] de [ 0 Not available 0 20406080100120 Roll Roll -2
-4
-6
-8 Time [s]
PRR1, KG=12.892m, Free roll decay, Intact condition PRR1, KG=12.892m, Free roll decay, Damage condition
8 6 EXPERIMENT EXPERIMENT 6 4 Parti ci pant 2 Participant 2 4 2
2 0
g] 0 20406080100120 de [ 0 -2 0 20406080100120 Roll Roll -2 [deg] Roll -4
-4 -6
-6 -8
-8 -10 Time [s] Time [s] 23rd International 638 The Specialist Committee on Prediction of Exteme Ship Motions and Capsizing Towing Tank Conference
PRR1, KG=12. 892m, Free roll decay, Intact condition PRR1, KG=12.892m, Free roll decay, Damage condition
8 6 EXPERIMENT EXPERI MENT 6 4 Participant 3 Participant 3 4 2
2
g] 0 g] de
de 0 20406080100120 [
0 [ 0 20406080100120
Roll Roll Roll -2 -2
-4 -4
-6 -6
-8 -8 Time [s] Time [s]
PRR1, KG=12.892m, Free roll decay, Intact condition PRR1, KG=12.892m, Free roll decay, Damage condition
8 6
EXPERIMENT EXPERIMENT 6 4
Participant 4 Participant 4 4 2
2
g] 0 g] de [
0 de 0 20406080100120 [ 0 20406080100120 Roll Roll -2 -2 Roll
-4 -4
-6 -6
-8 -8 Time [s] Ti me [s]
PRR1, KG=12.892m, Free roll decay, Intact condition PRR1, KG=12.892m, Free roll decay, Damage condition
8 6
EXPERIMENT EXPERIMENT 6 4 Participant 5 Participant 5 4 2
2 0
g] 0 20 40 60 80 100 120 g] de [
0 de
[ -2 0 20406080100120 Roll Roll Roll Roll -2 -4
-4 -6
-6 -8
-8 -10 Time [s] Time [s]
Figure 3.4 Free roll decay for the intact and damaged ship (measured and simulated). 23rd International Towing Tank Proceedings of the 23rd ITTC – Volume II 639 Conference
3.6. Ship Performance in Regular Waves more apparent in the simulation of the damaged ship roll RAO. The results of the benchmark study for the intact and damaged ship roll response in regu- In particular, none of the participants ob- lar beam waves, with tained the natural frequency of the damaged ship close to the value derived experimentally. ¾ constant wave height (H = 1.2 m, 2.0 m w In fact, the predicted natural frequency of the and 2.4 m) and damaged ship is quite inconsistent among the ¾ constant wave slope, namely constant participants: Participants 2, 4 and 5 predict a wave height to wavelength ratio equal to slight decrease of this frequency in relation to 1/25, the natural frequency of the intact ship, a are presented in this section. The purpose of trend shown also in the damaged model this particular study was to gain insight into experiments, but not to the extent measured in the modelling of roll motion by the partici- these experiments. Participant 1 is even pre- pants in normal operational and extreme con- dicting an increase of the intact natural fre- ditions (significant wave height up to 20 m). quency, whereas the predicted natural fre- quency by Participant 3 remains practically Roll Response Amplitude Operators unchanged. (RAOs) for Constant Wave Height Excitation. Based on the results presented in Figure 3.5, 4 The above results suggest that the pre- the following can be concluded : dicted hydrodynamic added moment of inertia Simulation of intact ship roll RAO for by all participants deviates significantly from constant wave height monochromatic the experimental value. wave excitation: All participants have generally accomplished the simulation Regarding the predicted peak values of the of the basic intact ship roll RAO suc- damaged ship roll response, practically all cessfully. participants predict higher roll amplitudes in- Some differences in the predicted peak dicating a clear underestimate of roll damp- values of ship response, occurring at the ing. natural roll frequency (more pronounced in the simulation by Participant 1), are The experimentally measured damaged due to the differences in the modelling ship roll response indicates the existence of a of roll damping, which is based on semi- second resonance frequency at approximately empirical coefficients and approaches twice the main roll resonance frequency. This that need further improvement. phenomenon is predicted by Participants 4 and 5 but at much higher frequencies and for Simulation of damaged ship roll RAO higher secondary resonance peak values. for constant wave height monochro- matic wave excitation: The simulation The above conclusions call for additional of the basic damaged ship roll RAO research in this area and a reassessment of the could not be predicted satisfactorily by damaged ship roll RAO results in the future, the participants. The reasons given ear- when more experimental and numerical lier concerning free rolling simulation benchmark results become available. apply equally here and become even
4 Note that experiments were performed in wave heights equal to 1.2 m and 2.4 m, whereas numerical simulations were under- taken in the range between 1.2 and 2.4 m. 23rd International 640 The Specialist Committee on Prediction of Exteme Ship Motions and Capsizing Towing Tank Conference
PRR1, KG=12.892m, Roll RAO, Intact condition PRR1, KG=12.892m, Roll RAO, Damage condition 18 18
16 Experi ment, (Hw=1.2m) 16 Experi ment, (Hw=1.2m) 14 Experi ment, (Hw=2.4m) 14 Experi ment, (Hw=2.4m) 12 Participant 1 (Hw=2.0m) 12 g ] Participnt 1, (Hw=2.0m) ]
/m 10
/m 10 g g de [ de 8 [ 8 Roll Roll Roll Roll 6 6
4 4
2 2
0 0 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 ω [rad/s] ω [rad/s]
PRR1, KG=12.892m, Roll RAO, Intact condition PRR1, KG=12.892m, Roll RAO, Damage condition 18 18
16 Experiment, (Hw=1.2m) 16 Experiment, (Hw=1.2m)
14 Experiment, (Hw=2.4m) 14 Experiment, (Hw=2.4m)
12 Participant 2, (Hw=2.4m) 12 Participant 2, (Hw=2.4m) g ]
]
/m 10
/m 10 g g de [ de 8 [ 8 Roll Roll 6 6
4 4
2 2
0 0 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 ω [rad/s] ω [rad/s]
PRR1, KG=12.892m, Roll RAO, Intact condition PRR1, KG=12.892m, Rol l RAO, Damage condition 18 18
16 Experiment, (Hw=1.2m) 15 Experi ment, (Hw=1.2m) 14 Experiment, (Hw=2.4m) Experi ment, (Hw=2.4m)
12 Participant 3, (Hw=1.2m) 12 Participant 3, (Hw=1.2m) g ] ]
/m 10 /m g g 9 de [ de 8 [ Roll 6 Roll 6
4 3 2
0 0 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 ω [rad/s] ω [rad/s] 23rd International Towing Tank Proceedings of the 23rd ITTC – Volume II 641 Conference
PRR1, KG=12.892m, Roll RAO, Intact condition PRR1, KG=12.892m, Roll RAO, Damage condition 18 18
16 Experi ment, (Hw=1.2m) 16 Experiment, (Hw=1.2m)
14 Experi ment, (Hw=2.4m) 14 Experiment, (Hw=2.4m)
12 Participant 4, (Hw=2.0m) 12 Participant 4, (Hw=2.0m) g ]
]
/m 10
/m 10 g g de [ de 8 [ 8 Roll Roll Roll 6 6
4 4
2 2
0 0 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 ω [rad/s] ω [rad/s]
PRR1, KG=12.892m, Roll RAO, Intact condition PRR1, KG=12.892m, Roll RAO, Damage condition 18 18
16 Experiment, (Hw=1.2m) 16 Experi ment, (Hw=1.2m)
14 Experiment, (Hw=2.4m) 14 Experi ment, (Hw=2.4m)
12 Participant 5, (Hw=1.2m) 12
g Participant 5, (Hw=1.2m)
] ] 10 /m /m 10 g g de [ de 8 [ 8 Roll Roll 6 6
4 4
2 2
0 0 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 ω [rad/s] ω [rad/s] Figure 3.5 Free roll decay for the intact and damaged ship (measured and simulated).
Roll Response Amplitude Operators case and no experimental data were available (RAOs) for Constant Wave Slope Excitation. to cross check the numerical predictions. The This study was restricted to only the intact ship results are presented in Figures 3.6 and 3.7.
PRR1, KG=12.892m, Roll RAO, Constant wave slope 1:25, Intact condition PRR1, KG=12.892m, Roll RAO, Constant wave slope 1:25, Intact condition 5 5 Participant 1 Participant 1 4.5 4.5 Participant 2 Participant 2 4 Participant 4 4 Participant 4 3.5 Participant 5 3.5 Participant 5
3 3
2.5 2.5
2 2
Roll Ampl / kA [-] / kA Ampl Roll 1.5 [-] / kA Ampl Roll 1.5
1 1
0.5 0.5
0 0 00.511.522.53 00.511.522.53 λ/L [-] λ/L [-] Figure 3.6 Roll response in regular beam Figure 3.7 Roll response in regular beam waves with constant wave slope over λ/L. waves with constant wave slope over ω. 23rd International 642 The Specialist Committee on Prediction of Exteme Ship Motions and Capsizing Towing Tank Conference
Based on the derived results, the following highly nonlinear ship motion problems is not can be stated: yet satisfactory. Overall, the results produced by the par- ticipating organisations differ significantly. 3.7. Ship Performance in Irregular Waves Results for large wavelength to ship length ratios (λ/L) appear satisfactory but this is not The prediction of the damaged ship per- the case for small λ/L ratios. Only results by formance is assessed on the basis of analysis Participants 2 and 4 appear to agree closely of the numerically simulated time series for throughout the frequency range considered in the exciting wave, the ship motion response the study. and the amount of floodwater, in comparison Predictions in the resonance region deviate with the corresponding time series of model substantially indicating clearly the differences experiments, simulation of the damaged ship of the semi-empirical damping models used survival boundaries and finally identification by the various participants for the extreme of critical wave heights. Detailed results of motion amplitude predictions. this study (experimental and numerically simulated time series) are given by In conclusion, it appears necessary that a Jasionowski & Vassalos (2001). In this sec- more comprehensive study should be carried tion only sample results are presented (Figures out in the future to investigate the relationship 3.8 to 3.19) and discussed. between the damping models used by the benchmark study participants. Unfortunately, Two representative runs per participant, experimental measurements were not avail- one for survival and one for capsizal cases, able to enable a more thorough evaluation of are presented, all corresponding to a signifi- the employed numerical procedures for the cant wave height excitation of 4.0 m. Wave intact, large amplitude and large slope motion elevation, heave and roll motions as well as studies. The only apparent result from this the amount of water accumulated on the car comparison is that numerical modelling of deck are shown.
Figure 3.8 Experimental measurements – damaged ship model (survival case).
23rd International Towing Tank Proceedings of the 23rd ITTC – Volume II 643 Conference
Figure 3.9 Experimental measurements – damaged ship model (capsizal case).
Figure 3.10 Numerical simulations by Participant 1 – damaged ship model (survival case).
Figure 3.11 Numerical simulations by Participant 1 – damaged ship model (capsizal case). 23rd International 644 The Specialist Committee on Prediction of Exteme Ship Motions and Capsizing Towing Tank Conference
Figure 3.12 simulations by Participant 2 – damaged ship model (survival case).
Figure 3.13 Numerical simulations by Participant 2 – damaged ship model (capsizal case).
Figure 3.14 Numerical simulations by Participant 3 – damaged ship model (survival case). 23rd International Towing Tank Proceedings of the 23rd ITTC – Volume II 645 Conference
Figure 3.15 Numerical simulations by Participant 3 – damaged ship model (capsizal case).
Figure 3.16 Numerical simulations by Participant 4 – damaged ship model (survival case).
Figure 3.17 Numerical simulations by Participant 4 – damaged ship model (capsizal case). 23rd International 646 The Specialist Committee on Prediction of Exteme Ship Motions and Capsizing Towing Tank Conference
Figure 3.18 Numerical simulations by Participant 5 – damaged ship model (survival case).
Figure 3.19 Numerical simulations by Participant 5 – damaged ship model (capsizal case). 23rd International Towing Tank Proceedings of the 23rd ITTC – Volume II 647 Conference
A visual comparison between the numeri- lower frequencies, closer to natural roll cally predicted and experimentally measured frequency of the ship. time series shows a rather unsatisfactory level The predicted roll response spectra by of agreement. Indeed, none of the numerical Participants 1, 2, 4 and 5 indicate under- time series matches qualitatively the experi- estimation of roll damping, with reso- mental values and only some of the numerical nant roll amplitudes significantly higher results agree qualitatively (Participant 1 and to than the experimental values. some extent Participants 4 and 5). The roll response predicted by Participants 2 and 3 Participants 2 and 3 predict considerably displays again noticeably higher amplitudes, higher roll spectral densities, partly as a possibly due to inaccurate roll damping mod- consequence of the predicted higher els. peak roll values and partly due to the shift of the peak frequency of the excit- A Fourier spectral analysis of the calcu- ing wave spectrum. lated time series records enables a better un- derstanding of the differences between the A noticeable peak appears in the roll numerical simulations and the response char- motion spectrum by Participant 5 at a acteristics derived by physical model tests frequency of about 1.0 rad/s unlike the (Figure 3.20) as outlined next: experimental results. Participant 2 failed to reproduce exactly The spectrum of heave response derived the exciting wave spectrum characteris- by Participant 3 shows a peculiar second tics, and hence used a wave spectrum lower peak around 0.6 rad/s. with its peak slightly shifted towards
23rd International 648 The Specialist Committee on Prediction of Exteme Ship Motions and Capsizing Towing Tank Conference
Figure 3.20 Spectral analysis of experimental and numerical ship responses. 23rd International Towing Tank Proceedings of the 23rd ITTC – Volume II 649 Conference
3.8. Survivability Boundaries SURVIVE/CAPSIZE BOUNDARIES 6 The reported survivability boundaries show consistently high accuracy by most the 5
participants. Both, the “lower” and “upper” 4
Exp boundaries distinguishing between sea states P1 P2 3 P3 Hs [m] Hs leading to ship survival, marginal survivabil- P4 P5 ity or capsize were predicted with a spread of 2 approximately ±0.5 m in Hs. Only Participant 3 fails to predict the capsizal boundary (no 1 capsize predictions for the specified condi- 0 tions). It must be mentioned here, that the simulation time is of importance in establish- Figure 3.21 Comparison between predicted ing the boundary consistently, as the longer and measured survival/capsizal boundaries. the duration of the simulation, the lower the boundary tends to be. This, however, was clearly defined as a benchmark test constraint 3.9. Concluding remarks and some variation between the participants is Considering the relatively low number of noticeable. The demonstrated accuracy in pre- the benchmark study participants and the com- dicting the critical sea states seems quite satis- plicate nature of the benchmark study problem factory from the point of view of application being addressed it is felt that the main objec- of such information to practical survivability tives of the present study have been met, assessment procedures. An overview of rele- though a similar study should be repeated in vant results is shown In Table 3.6 and de- the future with the aim to alleviate the effects picted in Figure 3.21. In compiling this table of some of the identified weaknesses. It has the ship survival boundary, for the particular been ascertained that at the present state of damage case and sea state tested, is identified knowledge, model experiments remain indis- on the basis of zero capsize events occurring pensable for assessing the survivability of for five consecutive numerical simulations damaged ships in waves, though theoretical- using different irregular wave realisations in numerical prediction methods can greatly con- each case. On the other hand a capsizal tribute to the assessment of the survivability of boundary is identified on the basis of all five damaged ships in waves. runs leading to a capsize event.
Table 3.6 Experimentally observed and 4. GUIDELINES FOR MODEL numerically simulated capsize events. TESTING OF INTACT AND DAMAGE STABILTY Experiment P1 P2 P3 P4 P5 The purpose of these guidelines is to pro- Hs=3.50m - - - - - 0 vide ITTC member organisations, intending to Hs=3.75m - - 0 - 0 1 undertake intact and damage stability model tests in waves, with a sound basis for carrying Hs=4.00m 0 0 2 0 1 3 out these tests. The derived guidelines are Hs=4.25m 3 0 3 0 2 4 based on those presented to the 22nd ITTC, upgraded to reflect member organisation ex- Hs=4.50m 5 2 5 1 3 5 perience. The full recommended guidelines Hs=4.75m - 5 - 1 3 - for model test on intact stability are included in the ITTC Quality Manual as Procedure 7.5- Hs=5.00m - - - 2 5 - 02-07-02.5 and on damage stability as Proce- dure 7.5-02-07-02.6. 23rd International 650 The Specialist Committee on Prediction of Exteme Ship Motions and Capsizing Towing Tank Conference
5. QUESTIONNAIRE Four of the organisations described ex- periments on intact and damaged ship models that fitted within the operating practice de- 5.1. Questionnaire on Model Experiments scribed by the 1999 ITTC Guidelines. Two of these organisations had experience with ex- The questionnaire was formatted with the treme motions, but not capsize. intention of validating the ITTC guidelines for Intact and Damaged model experiments, pre- For experiments in oblique waves with sented at the 1999 ITTC. forward speed, three organisations used bat- tery based power systems onboard the model The following organisations replied to the for intact extreme motions or capsize experi- questionnaire: ments and data was handled in one of the fol- Institute for Marine Dynamics, National lowing ways: Research Council of Canada Collected and stored on the model MARIN Transferred to storage unit by telemetry Naval Ship Research and Development Transferred to storage unit by cable. Centre Osaka University, Department of Naval For capsize/extreme motions, all three or- Architecture & Ocean Engineering ganisations used a free running model, pow- QinetiQ ered by a model propeller, with autopilot to The Universities of Glasgow and Strath- keep the model on course, once the experi- clyde, SSRC ment had started. A summary of the questionnaire is given For the damaged case, all four organisa- in Table 5.1. The full questionnaire, including tions supplied power to the model and trans- details of how to fill it out and submit it, can ferred the data to storage via an umbilical ca- be found on the web site of the committee: ble. The models were allowed to drift under www.ssrc.na-me.ac.uk/ittc/scexcap/ the action of waves. Scaling of openings through which water would flow was based Table 5.1 Summary of questionnaire on on geometry. No allowance was made for model test procedures. scale effects. Some notable exceptions to the guidelines were: 1. Past experience Damaged stability model experiments at 2. Model design, construction and outfit one organisation were carried out with Construction material Uppermost limit of accurate modelling length and scale requirements outside Factors considered in scaling flooded com- the guidelines. partments, deck, etc. For one organisation, damaged stability Scaling flow through pipes and openings Modelling of flooded spaces experiments were on simplified hull Model scale forms and permeability of under deck Instrumentation and equipment spaces was 100%. Model Power supply systems Two organisations conducted damage Data recording Ballasting stability tests using a much more sophis- Control of models ticated procedure than that recom- 3. Experiments mended for the typical Ro-Ro ferry case, Initial conditions and this was to study damage stability of Start of data acquisition warships. For these tests, the experiment Distance from wave maker Wave types used starts with an intact, self- propelled Model speed measurement model. The preparation and procedures 4. Data analysis for this part of the experiment fit within those described in the ITTC guidelines 23rd International Towing Tank Proceedings of the 23rd ITTC – Volume II 651 Conference
for intact model experiments. At some 5.2. CFD Survey point in the experiment, the damage is The following organisations replied to the simulated and the model floods. At this questionnaire: point the requirement is for the model to MARIN behave as a damaged ship and flooding Memorial University of Newfoundland, is monitored with water depth sensors Department of Naval Architecture and and video cameras. The internal struc- Ocean Engineering ture of the model is much more complex National Technical University of Ath- than a Ro-Ro model, due to the more ens, Department of Naval Architecture complex internal structure of the ship. and Marine Engineering Also, the requirement to have the model W. S. Atkins Consultants Ltd. self-propelled at the start of the experi- Universities of Glasgow and Strath- ments brings in extra challenges for clyde, Department of Naval Architecture model construction. Model scale for and Marine Engineering. these tests was 1:24 giving model lengths between 5 and 6 metres. The The survey is summarised in Table 5.2. preferred approach is to use a com- pletely free running model, with teleme- Table 5.2 Summary of survey on the use of try to transfer data from the model to a CFD. shore station. 1. What was the original motivation for you in- For intact stability tests most organisations corporating CFD into ship motion prediction use models ranging in length from 3 to 5 m, software? although there are cases of using models un- 2. Give a brief description of the ship motion der 2 m. prediction code used 3. Briefly describe the CFD code used In conclusion, the analysis of the small 4. Give a brief description of how you inte- sample of questionnaires suggests that there are grated the CFD code with the ship motions no major flaws in the ITTC guidelines for code model tests as proposed in 1999. The commit- 5. Flow types modelled by CFD component tee was lucky in that much of the work done by 6. Linking Issues between ship motions code researchers in this area is well documented in and CFD code the literature and the performers have accepted 7. Post Processing the academic tradition of exchanging informa- 8. State of Development of Codes tion freely on techniques and processes. 9. Validation 10. Special Limitations Based on the results of the questionnaire, 11. Future Developments each organization has slightly different proce- dures, but all appear to be well founded on 12. References good experiment practice. The challenge for The survey focused on combining codes to the future is to ensure that the guidelines ap- predict ship motions in waves with CFD, to ply to a wider range of ships than Ro-Ro fer- model the combined motion of ship and fluid ries, and should also consider ships where trapped on the deck or within a ship’s hull. It portions of the ship are flooded in normal op- should also be recognised that CFD has been erations (cruise ships, landing platform docks, used for other areas relevant to extreme mo- ships with moon pools, etc). tions including capsizing, such as modelling steep breaking waves, and also for seakeep- ing. Using CFD presents a means of overcom- ing the classical assumptions of linear theo- ries, combining forward speed with motion in 23rd International 652 The Specialist Committee on Prediction of Exteme Ship Motions and Capsizing Towing Tank Conference
waves and non-viscous flow. Beck & Reed roll tanks, the volume of water was fixed. (2001) give a good overview of the develop- Two methods kept the CFD and ship mo- ment of numerical seakeeping predictions tions codes separate, but the methodology from their earliest beginnings to the latest un- for linking them together varied. In one steady RANS approaches. They also point out case, the exchange between the codes the huge computational requirements for us- was by a data file. Another file controls ing CFD based seakeeping predictions, and which program runs, but each code runs the level of uncertainty in the results. As such, completely separately on different com- CFD methods for full ship motion predictions puters. In the other case, the motion code are a long way from practical application in was considered as an external subroutine engineering situations. by the CFD code. This was possible be- cause the code was designed for analys- CFD can also be applied to the reduced ing sloshing in spacecraft, and so it was problem of predicting viscous roll damping possible to include externally computed (Salui et al., 2000). This has always been a body motions as input to the problem. severe limitation for potential flow methods, The advantage of this approach is that the which cannot realistically predict roll motion. CFD code needs no modification. Most work with CFD in this area has focused For one organisation, the CFD code was on forced rolling in calm water, to obtain pre- integrated into the ship motion code as a dictions of the added mass and damping coef- subroutine. Using linear motions to com- ficients directly from the numerical procedure, pute the ship motions is a simpler (and hence eliminating the empirical assumptions. faster) process. However, it is limited to Three of the replies are related to the clas- ‘operational’ rather than extreme mo- sical assumptions of CFD methods, and one tions. Synchronising time steps between was a simplified approach. For the three ‘clas- the two commercial CFD codes and the sical’ CFD approaches, the following obser- ship motion codes was an issue in both vations were made. cases. In the Ro-Ro case, using files to Three methods use a 6DOF time domain transfer the data makes the coupling ex- code for predicting ship motions, linked to plicit, but the limitation of very small a CFD code to predict the motion of water time steps within the CFD code makes on the deck of the ship. Two ship motion this solution acceptable for the applica- codes have some linear and non-linear tion considered, but this may not be so elements in the hydrodynamic coefficients for other cases. In the fishing boat case, and restoring moments. The third method even though the motions code is treated uses a linear response time domain as a subroutine by the CFD code, the two method as the basis for its motion code. can function independently and the time All three CFD codes used variations of step in each one can be varied internally the ‘Volume of Fluid’ technique, where to obtain solutions. The CFD code uses a the grid is fixed (body fitted) and the free non-inertial reference frame. The re- surface is tracked either by grid elements quired ‘gravity’ components are interpo- that are partially full or by special free lated at whatever time step the CFD code surface cells. In one case studying a needs to obtain a solution. The limitation flooded Ro-Ro, the method included of the non-inertial reference frame is fluid entering and leaving the deck area, reached when the fluid reaches the top of based on the relative motion given by the the boundary. This limitation is less criti- ‘seakeeping’ code and matching the in- cal for an enclosed tank than for water ternal fluid height with the external trapped on the deck, when at this point height at the damage opening. In another the water would spill out, which is not case studying a fishing vessel with anti- modelled by the CFD code. For one or- 23rd International Towing Tank Proceedings of the 23rd ITTC – Volume II 653 Conference
ganisation, the CFD code adjusts the time This model is derived from simple rigid body steps internally to reach a solution, but motion consideration. Finally, an artificial co- since the two codes are combined the in- efficient is introduced to represent damping of ternal clock is consistent. floodwater motion. An ad hoc value of 0.15 is adopted for this coefficient derived for simple All methods have been partially validated box-shaped compartment from comparisons against experimental data (Bass & Cumming, with experimental data. With the geometric 2000; van Daalen et al., 2000; Woodburn et information about the tank stored in a data- al., 2001). base, the model is complete. The method has The two replies, also studying Ro-Ro fer- been validated against experiment data ries, took a different approach, which was out- (Jasionowski & Vassalos, 2001). side the classical definition of CFD. In one case The major advantage of the lumped mass a 6DOF non-linear time domain motions code approach is that it is computationally more has been expanded to 9DOF by considering the efficient than the traditional CFD methods, fluid as a lump mass. Hydrodynamic coeffi- and provided that there is not a significant cients related to ship-wave interaction are cal- amount of sloshing of the fluid, the prediction culated externally with a frequency domain of ship motions, including the dynamics of panel method. Calculated quantities are then floodwater is of acceptable accuracy. Flood- transferred to the time domain by the Impulse ing and draining issues are handled with em- Response Function technique. Flooding and pirical methods. draining of compartments can be modelled. The method has been validated against experi- Using CFD for predicting water motion on ment data (Papanikolaou et al., 2000). the deck and its effect on ship motions was the focus of the survey. Other approaches that In the other case, the underlying equations have been used in the past are to use either of the ship are derived from conservation of lumped mass (as one survey respondent did) linear and angular momentum applied to rigid or to use potential flow but for non-linear bodies, extended to include the internal fluid waves (Huang & Hsiung, 1996). In this case, mass in six degrees of freedom. The Froude- Euler’s equations were used for non-linear Krylov and restoring forces and moments are shallow water flow on the deck, with a flux integrated up to the instantaneous wave eleva- split based on the characteristic directions of tion, the radiation and diffraction forces and motion. The method was employed for two moments are derived from linear potential and three-dimensional decks. Whilst this is flow theory and expressed in time domain not CFD in the classical sense, it is a valid based on convolution and spectral techniques, approach for tackling the problem, provided respectively. The hull asymmetry due to ship that the volume of water on the deck is con- flooding is taken into account by a “database” stant or changing very slowly with time. approach, whereby the hydrodynamic coeffi- cients are predicted beforehand, and then in- An area that is similar to the extreme mo- terpolated during the simulation. The correc- tions and capsize problem is sloshing of fluid tion for viscous effects on roll and yaw modes within tanks on a ship. Related to this, Cariou of motion is applied based on empirical meth- & Cassela (1999) give a comparison of nu- ods. The second order drift and current effects merical results for 11 different CFD codes. are also catered for, based on parametric for- This paper summarizes the time and space do- mulations. Fluid sloshing has been modelled mains, viscosity, compressibility free surface by a free mass point moving due to the and wall condition for each of the codes, and acceleration field and restrained geometrically compares results for specified cases. The first by predetermined potential surfaces of centre was a simple two-dimensional problem, and of buoyancy for given amount of floodwater. compared the surface elevation of the fluid in 23rd International 654 The Specialist Committee on Prediction of Exteme Ship Motions and Capsizing Towing Tank Conference
an oscillating rectangular tank for a range of 7. CONCLUSIONS AND amplitudes and periods. The second case was a RECOMMENDATIONS 3D problem, consisting of an LNG cargo tank, excited for a combination of pitch and roll, again for a range of amplitudes and periods. 7.1. General Technical Conclusions An alternative to the grid based CFD ap- As the maritime industry progressively proaches is Smoothed Particle Hydrodynam- moves towards performance based criteria to ics (SPH). In this method the fluid is idealised address safety issues, there is wide scope and as elemental particles, and the tracks of these a major opportunity for member organisations particles are followed during the computation. to benefit from these developments. The method was originally developed for as- trophysics problems, but has since been ap- However, the severe limitations identified plied to free surface problems. The method in the existing numerical models for predict- can give a good prediction of breaking waves, ing ship capsizing and extreme motions needs but according to Beck & Reed (2001), de- to be addressed as summarised below: tailed comparisons of pressures and fluid ve- locities for simplified problems are not avail- 7.1.1 Prediction of Intact Ship Capsizing able. The method has been used to obtain and Extreme Motions some interesting results for the sloshing prob- At this stage there is a limited number of lem and post-breaking behaviour of waves. numerical models for predicting intact ship Naito & Sueyoshi (2001) presented results for capsizing and extreme motions with a range an SPH prediction of water moving on the of different levels of sophistication and pa- deck of a flooded Ro-Ro ferry. rameters. Only a few of these models consis- tently agree qualitatively with all the extreme motions and modes of capsize identified in 6. SYMBOLS AND TERMINOLOGY free running model experiments. None of the A number of international organisations models does so quantitatively. have been involved in ship stability for many More work is required to improve the years and have developed their own symbols agreement between physical and numerical and terminology for the field. model tests results. As these organisations have been using their symbols and terminology for a number 7.1.2 Prediction of Damaged Ship Capsizing of years, it is felt that it is now not practical to At this stage there is a limited number of develop a single set of symbols and terminol- numerical models for predicting damaged ship ogy to be used by all working in this field. capsizing. Unlike the case for intact ships, However, the need has been identified, fol- most models can consistently predict the cap- lowing a comprehensive review by Frances- size boundaries obtained from model experi- cutto (2002), for taking immediate action on ments in realistic sea conditions, albeit for the some items that are directly linked to stability specific damage scenario and mode of capsize and recommendations to this effect are put considered in the benchmark tests. forward. Moreover, a table comparing the symbols used by ITTC; ISO 12217; HSC There are, however, fundamental differ- 2000; and IMO has also been prepared and ences in the way these models handle flood- submitted to the Symbols & Terminology water/ship dynamics, none of the models giv- Group (Francescutto, 2002). This report is ing good agreement, either qualitatively or also posted in the web site of the committee. quantitatively, with results from physical model experiments. 23rd International Towing Tank Proceedings of the 23rd ITTC – Volume II 655 Conference
Before confidence can be gained in these Boat”, STAB 2000. models, wider application to different damage scenarios and ship types is required. Beck, R.F., and Reed, A.M., 2001, “Modern Computational Methods for Ships in a 7.1.3 Guidelines and Procedures Seaway”, SNAME Annual Meeting, 2001. Those member organisations involved Cariou, A., and Cassela, G., 1999, “Liquid with experimental testing of intact and dam- Sloshing in Ship Tanks: A Comparative age stability generally follow the guidelines Study of Numerical Simulations”, Marine recommended by the 22nd ITTC as appropri- Structures Vol. 12, No. 3, pp. 183-198. ate, justifying the adoption of procedures based on these guidelines as refined by the Cramer, H., 2001, “Effect of Non-Linearity in work undertaken by this committee. Yaw Motion on Capsizing Prediction” Proceedings of the 5th International Work- 7.1.4 Symbols and Terminology shop on Stability and Operational Safety The rationalisation of symbols and termi- of Ships, Trieste, Italy. nology for stability has not been made possi- ble, as international organisations directly van Daalen, E.F.G., Kleefsman, K.M.T., Ger- dealing with the subject are too ingrained in rits, J., Luth, H.R. and Veldman, A.E.P., the use of symbols different to those adopted 2000, “Anti-roll Tank Simulations with a by the ITTC. Notwithstanding this, it is rec- Volume of fluid Based Navier-Stokes ommended that steps are taken to communi- Solver”, 23rd Symposium on Naval Hy- cate widely the use and encourage the adop- drodynamics, Val de Reuil, France. tion of the Greek symbol “φ” to denote heel/list/roll angle. Francescutto, A., 2002, “A Critical Review of the Symbols and Terminology Relevant to Ship Stability”, Department Naval Archi- 7.2. Recommendations to the Conference tecture, Ocean and Environmental Engi- neering, University of Trieste, Italy. Adopt the Procedure “Seakeeping – Model Tests on Intact Stability” 7.5-02-07-02.5. Hamamoto, M., and Enomoto, T., 1996, Adopt the Procedure “Seakeeping – Model “Model Experiment of Ship Capsize in Tests on Damage Stability” 7.5-02-07-02.6. Astern Seas − 2nd Report”, J. Society of Naval Architects of Japan, Vol. 179, pp. 77-87. 8. REFERENCES AND NOMENCLATURE Hamamoto, M., and Kim, Y. S., 1993, “A New Coordinate System and the Equations Describing Manoeuvring Motions of a 8.1. References Ship in Waves” J. Society of Naval Archi- tects of Japan, Vol. 173, pp. 209-220, (in Bass, D., and Cumming, D., 2000, “Numeri- Japanese). cal and Experimental Investigation of Wa- ter Trapped on Deck on a Small Fishing Hamamoto, M., and Saito, K., 1992, “Time Boat”, STAB 2000. Domain Analysis of Ship Motions in Fol- lowing Waves” Proceeding of the 11th Bass, D., and Cumming, D., 2000, “Numeri- Australian Fluid Mechanics Conference, cal and Experimental Investigation of Wa- Hobart, Vol. 1, pp. 355-358. ter Trapped on Deck on a Small Fishing 23rd International 656 The Specialist Committee on Prediction of Exteme Ship Motions and Capsizing Towing Tank Conference
Hashimoto, H., and Umeda, N., 2001, “Impor- merical Analysis of Violent Free Surface tance of Wave Effects on Manoeuvring Flow on an Flooded Car Deck Using Par- Coefficients for Capsizing Prediction”, ticle Method”, STAB2001 5th Interna- Proceedings of the 5th International Work- tional Workshop on Stability and Opera- shop on Stability and Operational Safety tional Safety of Ships, Trieste,Italy,. of Ships, Trieste, Italy. Papanikolaou, A., Zaraphonitis, G., Spanos, Ikeda, Y., Umeda, N., and Tanaka, N., 1988, D., Boulougouris, E., and Eliopoulou, E., “Effect of Forward Speed on Roll Damp- 2000, “Investigation into the Capsizing of ing of a High-Speed Craft” Journal of Damaged RO-RO Passenger Ships in Kansai Society of Naval Architects, Japan, Waves”, STAB2000. Vol. 208, pp. 27-34 (in Japanese). Papanikolaou, A., and Spanos, D., 2001, ITTC, 1999, “Specialist Committee on Ship “Benchmark Study on the Capsizing of a Stability. Final Report and Recommenda- Damaged Ro-Ro Passenger Ship in tions to the 22nd ITTC “Proceeding of the Waves” Draft Final Report, May 2001. 22nd ITTC, Seoul, Korea and Shanghai, China, Vol. 2, pp. 399-431. Papanikolaou, A., 2001, “Benchmark Study on the Capsizing of a Damaged Ro-Ro Jasionowski, A., 2001, “Detailed Analysis of Passenger Ship in Waves”, Revised Final the Revised Damage Benchmark Results”. Report, October 2001. University of Strathclyde − The Ship Sta- bility Research Centre, October 2001. Renilson, M.R., and Manwarring, T., 2000, “An Investigation into Roll/Yaw Coupling Jasionowski, A., and Vassalos, D., 2001, “De- and its Effect on Vessel Motions in Fol- tailed Analysis of the Final Revised Dam- lowing and Quartering Seas” STAB2000, age Benchmark Results”, University of Vol. A, pp. 452-459. Strathclyde − The Ship Stability Research Centre, December 2001. Salui K.B., Sarkar, T., and Vassalos, D., 2000, “An Improved Method for Deter- Matsuda, A., Umeda, N., and Suzuki, S., 1997, mining Hydrodynamic coefficients in Roll “Vertical Motions of a Ship Running in Motion using CFD Techniques” Ship Following and Quartering Seas”, J. Kansai Technology Research, Vol. 47, No. 4. Society of Naval Architects, 227 : 47-55, (in Japanese). Sera, W., and Umeda, N., 2001, “Effect of Short-Crestedness of Waves on Capsize of Matusiak, J., 2001, “Importance of Memory a Container Ship in Quartering Seas” J Ja- Effect for Capsizing Prediction” Proceed- pan Institute of Navigation, Vol. 104, pp. ings of the 5th International Workshop on 141-146, (in Japanese). Stability and Operational Safety of Ships, Trieste, Italy. SOLAS, 1997 “Consolidated edition 1997- Annex 5: Resolutions of the 1995 SOLAS Munif, A., 2000, “Numerical Modelling on Conference. Model test method” IMO , Extreme Motions and Capsizing of an In- Resolutions of the Conference of Contract- tact Ship in Following and Quartering ing Governments to the International Seas”, Doctor Thesis, Osaka University, Convention for the Safety of Life at Sea, Japan. 1974, adopted on 29 November 1995.
Naito, S., and Sueyoshi, M., 2001, “A Nu- Umeda, N., and Yamakoshi, Y., 1986, “Ex- 23rd International Towing Tank Proceedings of the 23rd ITTC – Volume II 657 Conference
perimental Study on Pure Loss of Stability tative Aspects of Nonlinear Ship Motions in Regular and Irregular Following Seas”, in Following and Quartering Seas with STAB86. High Forward Velocity” Journal of Marine Science and Technology, in press. Umeda, N., 1988, “Application of Slender Body Theory to Lateral Ship Motions with Woodburn, P., Galagher, P., and Letizia, L., Both Free Surface and Free Vortex Layers”, 2001, “Fundamentals of Damaged Ship Bulletin of National Research Institute of Survivability”, RINA, Spring Meeting, Fisheries Engineering, No. 9, 185-203. 2002.
Umeda, N., Hamamoto, M., and Takaishi, Y., Vassalos, D., and Jasionowski, A., 2000, 1995, “Model Experiments of Ship Cap- Stockholm Agreement Water on Deck size in Astern Seas” J. Society of Naval Model Experiments for Passenger/Ro-Ro Architects of Japan, Vol. 177, pp. 207-217. Vessel, Final Report, PSBG-RE-004-AY. University of Strathclyde − The Ship Sta- Umeda, N., Matsuda A., and Takagi M., 1999, bility Research Centre, February 2000. “Model Experiment on Anti-Broaching Steering System”, J. Society of Naval Ar- Vassalos, D., Umeda, N., and Papanikolaou, chitects of Japan, Vol. 185, pp. 41-48. A., 2001, “2nd Revised Guidelines for ITTC committee on the Prediction of Ex- Umeda, N., 2000, “Effects of Some Seakeep- treme Ship Motions and Capsizing ing/Manoeuvring Aspects on Broaching in Benchmark Tests”, June 2001. Quartering Seas”, Contemporary Ideas on Ship Stability, Elsevier Science Publica- tions (Amsterdam), pp. 423-433. 8.2. Nomenclature COREDES Committee on Research and Umeda, N., Munif A. and Hashimoto H., 2000, Development in European “Numerical Prediction of Extreme Motions Shipbuilding and Capsizing for Intact Ships in Follow- CRN Co-operative Research, Navy ing/Quartering Seas”, Proceeding of the 4th Osaka Colloquium on Seakeeping Per- JSNAJ Journal of the Society of Na- formance of Ships, Osaka, Japan, 368-373. val Architects of Japan RINA Royal Institution of Naval Umeda, N., and Papanikolaou, A., 2000, “Re- Architects vised Guidelines for ITTC committee on SNAME Society of Naval Architects the prediction of Extreme Ship Motions and Marine Engineers and Capsizing Benchmark Tests”. STAB Int. Conf. on Stability of Ships and Ocean Vehicles Umeda, N., “Final Report on the Benchmark WEGEMT European Association of Tests for Intact Ships”, ITTC SCEXCAP Universities in Marine Tech- Progress Report, December 2001. nology and Related Sciences Umeda, N., and Hashimoto, H., 2002, “Quali- tative Aspects of Nonlinear Ship Motions 9. ACKNOWLEDGEMENT in Following and Quartering Seas with High Forward Velocity” Journal of Marine The committee would like to thank Science and Technology, in press. wholeheartedly all the organisations and indi- viduals who contributed to this work over the Umeda, N., and Hashimoto, H., 2002, “Quali- last three years. 23rd International Towing Tank Proceedings of the 23rd ITTC – Volume III 745 Conference
The Specialist Committee on Prediction of Extreme Ship Motions and Capsizing Committee Chair: Prof. Dracos Vassalos (University of Strathclyde) Session Chair: Ir. George F.M. Remery (MARIN)
I. DISCUSSIONS erating conditions. With deterministic seakeeping tests in longcrested regular and irregular seas the wave characterictics at the I.1. Discussion on the Report of the 23rd position of the cruising ship, i.e. wave eleva- ITTC Specialist Committee on tion, pressure distribution as well as accelera- Prediction of Extreme Ship Motions tion and velocity fields in space and time are and Capsizing: Deterministic analysis known. Thus, these data can be related to the of extreme motions ship behaviour, and the physical phenomena, i.e. the mechanism of large roll motions with By: Günther F. Clauss, Technical University subsequent capsizing are investigated as of Berlin, Germany cause-effect chain. Using these sophisticated data our non-linear numerical model has been Congratulations for this excellent report. validated. International benchmark tests are extremely important, though they are very time- As an application of our numerical tool we consuming and work-intense for organizers evaluated the motion behaviour of selected and participants. Especially when regarding ships in arbitrary seas, long-crested and short- the development of rules and regulations to- crested. Simulation results are presented in wards direct analysis it is indispensable to polar plots which give the limiting wave evaluate the quality of available numerical height for capsizing of a vessel depending on tools. The current benchmark test reveals its speed and course. As shown in Figure I.1.1 some deficiencies of the applied tools, how- the most critical regions of resonance motions ever, it also shows its limitations in interpret- as well as parametric resonance are identified. ing the results due to the selection of the ref- Only a change of trim by 1 m to stern reduces erence ships and operating conditions. the capsizing risk considerably (comparing left hand and right hand diagram). Conse- As presented in the Group Discussion B.2 quently, the assessment of the seakeeping be- “IMO Standards and ITTC” our contribution haviour of a floating structure requires a “Evaluation of Capsizing Risk by Determinis- highly complex procedure combining non- tic Analysis of Extreme Roll Motions” (au- linear numerical simulation methods validated thors: G.F. Clauss, J. Hennig and H. Cramer) by deterministic seakeeping tests. As a result, deals with recently built ships, and most in- safer ships can be designed and loading condi- vestigations are carried out with speeds and tions optimized, improving ship operation and stability characteristics which reflect real op- navigation significantly.
23rd International 746 The Specialist Committee on Prediction of Extreme Ship Motions and Capsizing Towing Tank Conference
Figure I.1.1 Polar plot with limiting wave heights for a RO-RO design-based on nonlinear calcu- lation methods.
I.2. Discussion on the Report of the 23rd These waves tend to be steeper than the ITTC Specialist Committee on preceding and following waves. Prediction of Extreme Ship Motions Wave steepness increases with wave and Capsizing: Characteristics of height, with large waves showing greater extreme waves in the model variability in steepness than smaller waves. experiments and calculations Even in deep water storm waves are sig- By: Julian Wolfram, Heriot-Watt University, nificantly non-linear with crests eleva- United Kingdom tions much greater than trough depths.
My compliments to the Committee on In shallow water the situation is more se- their report and the very extensive amount of vere with increased steepness and asymmetry. work it represents. Extreme ship motions tend Just of the coast near Venice a platform in 18 to be associated with extreme wave condi- m of water has recorded a 12 m high wave. tions, so I will preface my questions by a few This platform is passed by fishing vessels comments drawn from observations of wind- regularly. driven storm waves in the North Sea: My concern is that the characteristics of Most big waves occur near the middle of extreme waves are not captured in the physi- a group of waves of above average cal model experiments and numerical calcula- height. tions the Committee describes. 23rd International Towing Tank Proceedings of the 23rd ITTC – Volume III 747 Conference
If non-linear responses are being consid- imply that the selection of Ships A-1 and A-2 ered should we use a non-linear wave input? as well as their operational conditions are in- appropriate. Ship A-1 was newly designed by If so is 2nd order sufficient? a professional design office especially for this series of experiments to enable all hull design Do the committee think wave steepness of details to be published. Ship A-2 is a real ship 1 in 25 be sufficient, given that in short- currently operating in Asian Waters with a crested seas wave steepness above 1 in 5 can number of similar ships having sunk in waves. occur in deep water? The loading conditions chosen are specified to marginally comply with the current IS Code. Although experiments were performed in I.3. Discussion on the Report of the 23rd regular, long-crested irregular and short- ITTC Specialist Committee on crested irregular waves, only regular waves Prediction of Extreme Ship Motions were selected for the bench mark testing sim- and Capsizing: Design and operation ply to ensure the maximum number of partici- of vessels to avoid capsize pating organisations.
By: Neil Bose, Memorial University of New- Although the Committee appreciates Prof. foundland, Canada Clauss’ invaluable data, it will be even more constructive and useful if he were to publish The committee has worked on a very im- comparisons between his experiments and portant subject. But I can’t help thinking that numerical simulation as time series in near while it is important that we improve on mod- future. This could significantly contribute to elling and accuracy of the actual capsize proc- further progress in numerical modelling to- esses, what is much more important is how wards practical use. Presenting information on can we make ships that are less susceptible to encountering dangerous situations in follow- capsize. Capsize of vessels nearly always re- ing/quartering seas has been debated exten- sults in loss of life. Has the work of the Com- sively in IMO in the past with a number of mittee led to the identification of factors that organisations presenting information during lead to capsize and ways in which the designs this debate in the form presented by the dis- and operations of vessels can be improved to cusser. More over, based on the findings of avoid capsize? studies and experiments aiming to identify and understand the associated problems guid- ance to master to avoid the postulated danger- II. COMMITTEE REPLIES ous situations have been formulated.
Finally the Committee would like to re- II.1. Reply of the 23rd ITTC Specialist mark that deterministic transient waves are Committee on Prediction of Extreme useful to validate numerical models but are Ship Motions and Capsizing to not appropriate to use for direct stability as- Günther F. Clauss sessment itself as an alternative tool to the IS code. Thank you for your written contribution. This is because the occurrence probability The Committee has conceded that addi- of such waves cannot be evaluated and the tional tests using other ships in other scenar- direct stability assessment requires capsizing ios, particularly for damage stability, would risk or capsizing probability to be compared certainly add value to the benchmarking work with acceptable risk at the final stage. already undertaken. This, however, does not 23rd International 748 The Specialist Committee on Prediction of Extreme Ship Motions and Capsizing Towing Tank Conference
II.2. Reply of the 23rd ITTC Specialist quately captures the essential wave effects Committee on Prediction of Extreme pertinent to predicting extreme ship motions. Ship Motions and Capsizing to Julian Having said this, the committee follows ongo- Wolfram ing developments in nonlinear wave and loads modelling. The Committee appreciates the comments made by Prof. Wolfram. Before answering his questions directly, we would like to start with II.3. Reply of the 23rd ITTC Specialist some general comments. First, extreme ship Committee on Prediction of Extreme motions (or capsize) are not always associated Ship Motions and Capsizing to Neil with extreme wave conditions. For example, Bose parametric rolling, loss of stability in a wave group or broaching may occur in waves of Over the recent past, there has been an evi- moderate steepness; equally, high speed craft dent shift from prescriptive to performance- may lose stability and capsize even in calm based safety standards whereby safety, for ex- seas. Furthermore, it is not only the wave ample, against capsize is assessed on the basis of height or steepness that is important, the wave numerical simulations and/or model experi- period plays an essential role in giving rise to ments. In response to this, the International critical conditions. The committee concurs Maritime Organisation has adopted “Equivalent with Professor Wolfram on the importance of Design and Approaches” and “Safety Equiva- wave groups. Research is currently underway lence” to facilitate the introduction and promote attempting to correlate wave groupiness and the implementation of first principles (scientific) damaged ship capsize in critical conditions, approaches to assessing ship safety. Building on whilst efforts are underway to quantify this these developments, philosophies pertaining to effect experimentally and numerically for in- “Design for Safety” are being pursued rigor- tact ship stability in short-crested seas (see ously, rendering safety progressively more a life- committee report section 2.6). In this sense cycle design issue rather than a constraint. In the work performed in the Waves Committee this respect and within this fast evolving safety is of particular relevance. environment, the work of the Committee is thought to be important in three ways: firstly, in Concerning the question on the wave the validation of numerical tools and establish- slope used and linearity of the wave input, the ment of experimental procedures for assessing committee would like to respond as follows. safety against capsizing without which perform- Whilst a 1:25 wave slope was used in the con- ance-based standards would not be developed or tainer ship (ship A-1) model experiments, implemented; secondly, for enhancing under- which given the low GM for this ship proved standing of the mechanisms of capsize and of to be more than adequate for the intended the key design and operational parameters influ- purpose, it was necessary to use wave with a encing the occurrence/avoidance of capsize for steepness of up to 1:8.7 in the purse seiner use by designers, operators and regulators; tests (ship A-2 with a GM of 1 m) for capsize thirdly, by providing the basis for measuring to occur. These tests indicate that it is not safety performance, i.e. quantitative levels of steepness alone that is the determining factor. safety, vital to the development of “Safety Regarding the wave nonlinearity, physical Equivalence” approach. Having said this, safety model experiments do capture these effects against capsizing is but one of the many factors (the nominal in the ship A-2 tests was 1:7, affecting the safety of a ships at sea and many of whilst the measured steepness was 1:8.7). The these as well as the role of capsize safety within numerical models typically rely on linear the safety realms are important issues outside the wave theory (regular waves or Gaussian ir- remit of this Committee. regular waves) in the belief that this ade-