OMAE2005-67123

download OMAE2005-67123

of 8

Transcript of OMAE2005-67123

  • 8/7/2019 OMAE2005-67123

    1/8

    Proceedings of OMAE 200524 th International Conference on Offshore Mechanics and Arctic Engineering

    Halkidiki, Greece, June 12-17, 2005

    OMAE2005-67123

    VALIDATION OF NUMERICAL MOTION SIMULATIONS BY DIRECT COMPARISONWITH TIME SERIES FROM SHIP MODEL TESTS IN DETERMINISTIC WAVE

    SEQUENCES

    Gunther F. ClaussOcean Engineering Section

    Technical University BerlinGermanyEmail: [email protected]

    Janou Hennig

    Ocean Engineering Section

    Technical University BerlinGermanyEmail: [email protected]

    Heike CramerBasic Design Department

    Flensburg ShipyardGermany

    Email: [email protected]

    Kay-Enno BrinkDepartment Seakeeping and Manoeuvring

    Hamburg Ship Model BasinGermany

    Email: [email protected]

    ABSTRACTShip safety under normal and severe weather conditions is

    determined by ship design, approval, and operation. Numeri-cal simulation of ship motions has proven to be a valuable toolfor ship design evaluation not only for accident investigationsand studies addressing fundamental stability related phenomena,but also during the design process. A sufcient number of sim-ulations provide a data basis for polar plots to judge the shipssituation in a sea state dened by signicant wave height andcharacteristic period with respect to load case, encounter angleand ship speed. Application of data provided by numerical toolshas to take into account the validity range of the model and hasto be validated sufciently by model test data.

    For providing useful validation data, the exact correlation of wave excitation and ship motion in model testing is indispens-able. In the framework of the German research project S IN SEE ,this is achieved by a fully automated test procedure with a freerunning ship model in combination with deterministic genera-tion of tailored realistic wave sequences and their transformationto the moving reference frame of the cruising ship. The resultant

    Address all correspondence to this author.

    wave train can be directly correlated with time series of motions registered by an optical system and forces. In this paper, thesemethods are applied to investigate pre-simulated seakeeping sce-narios in the model basin and compare the results directly to thesimulation results.

    INTRODUCTIONNumerical motion simulations are a valuable tool for ship

    design, investigation of accidents and operational guidance.Most of them have improved signicantly during the last yearsand are considered routinely in the design process and for thebasic investigation of ship safety. However, model tests are stillrequired and quite accepted by the classication societies whilenumerical simulations are not as the required quality of numeri-cal simulation tools is not agreed upon. Thus, two questions withregard to a routine numerical safety assessment of ship safetyhave to be answered:

    Is the numerical tool able to model the basic mechanism(qualitative validation)?

    How good are the quantitative results of the method?

    1 Copyright c 2005 by ASME

  • 8/7/2019 OMAE2005-67123

    2/8

    Before being accepted as reliable means to replace model testsand/ or regulations, a given motion simulation tool has to be val-idated.

    In this paper, a numerical motion simulation method appliedand further developed by the Flensburg Shipyard is presented asan example of a state-of-the-art tool for the simulation of shipmotions in realistic seas.

    A new type of validation method is introduced which al-lows to compare time series of model test and simulation resultsdirectly as the reproducible time-space correlation of ship andwave encounter is given by the generation of deterministic wavesequences and a fully automated deterministic capsizing test pro-cedure.

    As a result, time series from numerical simulations andmodel tests in all degrees of freedom are compared in the movingreference frame. These comparisons are based on the exact mod-elling of wave sequences both in the numerical and experimentalsimulation.

    VALIDATION OF NUMERICAL MOTION SIMULATIONSThe numerical motion simulation tool Rolls applied in the

    framework of the present project S IN SEE is based on a method-ology developed in the 80ies by Kroeger [1] and Petey [2] at theInstitut fuer Schiffbau, University of Hamburg, based on work by Soeding, [3], [4], and others.

    Rolls simulates the motion of intact and damaged ships intime domain in all six degrees of freedom in regular waves andirregular long or short crested seaways. For four motions, namelyheave, pitch, sway and yaw, response amplitude operators (RAO)

    are used, calculated linearly by means of strip theory. The surgemotion is simulated assuming a hydrostatic pressure distribu-tion under the water surface for the determination of the surge-inducing wave forces. The roll motion is simulated non-linearlywith the righting arm hs in the seaway being determined for ev-ery time step using Grims effective wave as modied by Soed-ing [3].

    The underlying theoretical model was described in previouspublications, e. g. [5]. It allows for very short simulation timesand is basically able to predict the following mechanisms of largerolling angles and capsizing very well:

    Roll resonance in general Parametric resonance Loss of stability phenomena Combinations of the above

    At the same time sway and yaw dominated phenomena can onlybe roughly judged based on the underlying linear RAO.

    Numerical tools need to be validated in order to evaluatetheir reliability and applicability. For capsizing investigation thisis for practical reasons based on model test data. For validationthe following methods are commonly used:

    Comparison of amplitudes and phases in regular waves Analysis of statistical results in irregular waves Investigation of basic ability to model the phenomenon qual-

    itatively

    A validation by direct comparison of time series is quite fre-quently used for regular waves (e. g. when comparing the non-linear roll motion leading up to a capsize, ITTC-Benchmark andothers) but not for irregular waves. Such comparisons are usuallynot considered for mainly two reasons:

    1. The exact correlation of time and position for both wave ex-citation and ship motion is not available in standard test set-ups.

    2. Even if the exact correlation over time is available, a di-rect comparison of measured and simulated time series isextremely complicated when all six degrees of freedom areconsidered. All small deviations in the ship reaction might

    add up over time in a way that the situation is not compara-ble anymore. E. g. a small deviation in the surge motion willchange the wave trains encountered by the vessel as such.

    DETERMINISTIC CAPSIZING TESTS AND WAVE SE-QUENCES

    At the Hamburg Ship Model Basin (HSVA), a deterministiccapsizing test procedure is implemented. The detailed aspects of this technique have been covered by numerous papers [6], [7],[8], [9], only the basic features of the entire method are summa-rized here:

    Computer control of the entire test run High accuracy due to control and measurement by an opticalsystem

    Free running ship model with individual propulsion Generation of deterministic wave sequences

    A sketch of the HSVA towing tank and its main dimensions isgiven in Fig. 1. A double ap wave maker is used for wave gen-eration. The deterministic wave generation and test technique al-lows to generate all kinds of deterministic wave sequences suchas rogue waves within a realistic sea state or wave tank realiza-tions of observed wave records in a model basin [10], [11]. Thistechnique is used here to bring the desired wave sequences with

    exact time-space correlation to both the model basin and the nu-merical simulation.

    INVESTIGATED SHIP MODELThe investigated ship is a twin screw RO-RO design. The

    full scale data are Lpp = 190 .29 m, B = 26 .50 m, T = 7.35 mThe tested conditions correspond roughly to stability values(GM) around the current limits as given by IMOs Code on Intact

    2 Copyright c 2005 by ASME

  • 8/7/2019 OMAE2005-67123

    3/8

    Figure 1. Test setup for computer controlled capsizing tests at the Ham-

    burg Ship Model Basin.

    Stability. Next to validation purposes one main target of S IN -SEE is the evaluation of the current safety limits as given by therules and regulations and the development of direct assessmentsof ships intact safety (e. g. [12]).

    BRINGING THE SIMULATED WAVE SEQUENCE INTOTHE MODEL BASIN (QUALITATIVE COMPARISON)

    In order to get an idea of the possible dangerous scenariowhich causes large roll angles or capsizing to a given ship, sus-pectable situations are simulated in Rolls and investigated withrespect to interesting results. An example for this is given inFig. 2.

    The chosen sea state is a JONSWAP sea with T P = 11 .5 sand H s = 10 m which is represented as energy equivalent spectralcomponents at the beginning of the simulation. The simulationresults in large roll angles due to parametric excitation in severalwave sequences at a speed and course of v = 8 kn and = 10 (GM=1.24 m). In the simulation, the ship capsizes in a wavesequence. This scenario has to be realized in the deterministiccapsizing test.

    Therefore, the ship model (scale 1:34, with according loadcase) sails at same course and model speed in the model basinwhich is ensured by the deterministic test technique. The wavesequences are realized according to the previous simulation. Inthe simulation, the wave sequence from the chosen sea state(t , x0) characterized by T P and H s is represented by its

    Figure 2. Simulated scenario of parametric rolling of the RO-RO ship ina JONSWAP wave sequence.

    Fourier coefcients A j and the initial phase j0 at x0:

    (t , x0) =n/ 2

    j= 0

    A j cos ( jt + j0). (1)

    These spectral wave data from Rolls refer to the beginning of thesimulation which is dened x0 = 0 and t 0 = 0 both for head and

    following seas. At this position the ship speed is v = 8 kn and theencounter angle between ship and waves for an uni-directionalsea state is = 10 .

    During the simulation, linear wave theory serves as wavemodel which means that the amplitude A j remains constant atany position in time and space during the simulation and thephase is adopted according to linear dispersion:

    (t , x0 + x) =n/ 2

    j= 0

    A j cos ( jt + j0 k j x) (2)

    2j = k j g tanh (k jd ). (3)

    x is calculated from the ship speed at the given position and thesurge motion s from the numerical simulation:

    x(t ) =t max

    t = 0

    v(t )t + s(t ). (4)

    The full scale wave train from the numerical simulation is givenby inverse Fast Fourier Transform as target wave train (Fig. 3

    3 Copyright c 2005 by ASME

  • 8/7/2019 OMAE2005-67123

    4/8

    top). The appropriate target wave is transformed to model scale(1:34, Fig. 3). The sea state has the initial shape as in the simula-tion and the encounter time is the time the wave train needs fromthe test beginning to arrive at the encounter position. The waveshave to be generated as long as the ship sails through the towingtank, plus the time the waves need to the stop position.

    In head seas, the position of encounter between ship andwave which is equivalent to the simulation beginning, has to bedened. In a test tank the position most opposite with regard tobeginning of measurement is chosen. At HSVA, the position isxtarget = xe = 215 m from the wave maker. At this position theship model reaches the desired course and speed in calm water.This is where the wave train at the beginning of the simulationrefers to.

    In following seas, the ship either tends to sail out of the waveeld or the waves may overtake the ship model depending ongroup velocity of the waves and the ship speed. Thus, it has to beensured that at the beginning of the simulation enough waves arein the tank as the simulation starts with this condition as well. Inour example, 100 s are simulated before the ship encounters thetarget wave train. At HSVA, the target position where the modelreaches target course and speed and the registration starts, is atxe = 67 m, and here the target wave train appears after 100 s plusthe time the wave needs to reach this position starting from thewave maker.

    Now the modied non-linear approach [10] is applied to ob-tain the wave train at the position of the wave maker. For thetransformation zero values are added in order to deal exactly withthe desired wave train wrap around would give mistakes in thewave tank simulation. After this manipulation the wave train has

    to be transformed to the target position again to get the new en-counter time t e of ship and wave. This value is put into the mastercomputer of the towing carriage to ensure that the model will beat xe at time t e after the start of the test.

    EXECUTION OF THE DETERMINISTIC CAPSIZINGTEST

    The corresponding control signals for driving the wavemaker (signals for upper and main ap of double ap wave makerat HSVA) are calculated (Fig. 3). Now the experiment is started,and the wave train is generated. The wave train needs the time t eto reach the position xe . The automated test procedure as men-

    tioned above ensures that the model reaches this target positionin time.

    The wave train is veried by wave probe registrations duringthe test. In the example from Fig. 3, the generated wave train isregistered at a stationary wave probe close to the wave maker andtransformed to x = 125 m (compare target wave train).

    The ship position is measured during the test. Thus, the sta-tionary wave train can be transformed to the moving referenceframe of the ship model to obtain the wave as experienced by the

    ship. The resultant (measured) roll motion can be directly com-pared to this wave train (Fig. 3 bottom). The motion simulationis carried out for the same load condition as in the model test,GM=1.24 m.

    RESULT OF PRE-SIMULATIONThe resultant roll motion can be compared to the predicted

    time series. Visual (qualitative) comparison of numerical simula-tion and model test proves that the method matches the underly-ing basic mechanism, Fig. 4 [13]. As the motions are simulatedbefore the corresponding model test, the given qualitative valida-tion scheme might prove to be trust-enhancing.

    BRINGING THE MODEL TEST CONDITION INTO THENUMERICAL SIMULATION (DIRECT COMPARISONFOR VALIDATION)

    The exact representation of the simulation results in themodel test is not possible since the track of the ship model inthe tank differs slightly due to disturbances or other slight varia-tions. Therefore, wave and ship encounter differently (compareFig. 9) and the following procedure is applied to re-model thetest conditions in a new numerical simulation run for direct com-parison.

    As a rst step, the measured x-position of the ship model ismodied with respect to starting position and instant. The rstmeasuring point in the time series is dened as instant t = 0 andassigned to starting position x = 0 instead of the measured realstarting position (Fig. 5):

    xsim (t sim ) := xmeas (t meas t 0) xmeas (t meas = 0). (5)

    Accordingly, the measured wave train at an undisturbed sta-tionary position xmeas = x0 has to be transformed to the startingposition of the ship model and the time axis has to be shifted(Fig. 6). Now, the wave train is calculated at full scale and repre-sented by 50-100 Fourier coefcients after adequate reduction of time data to reduce the simulation duration.

    Based on the deterministic capsizing test procedure the ex-act ship position is known at any time step of the model test.

    For verifying the modications with the simulation results, thewave train is calculated at the position of the ship. The chosentime window the simulation duration has to be a multiple of the longest occurring wave in the simulation, T sim = 2/ min

    max / 1.56.Now, the simulation is run with the given wave spectrum and

    x data. In the simulation the vessel is usually simulated in all sixdegrees of freedom. But as said before, even small disturbances(especially regarding the x-t correlation) will lead to a variation

    4 Copyright c 2005 by ASME

  • 8/7/2019 OMAE2005-67123

    5/8

    1200 1300 1400 1500 1600 170010

    0

    10

    t [s]

    [m]

    wave train used in numerical simulation (full scale)

    190 200 210 220 230 240 250 260 270 280 290 300

    0.2

    0

    0.2

    t [s]

    [m]

    target wave train for model test at x = 125 m (model scale 1:34)

    50 100 150 200 250 300

    0.2

    0

    0.2

    t [s]

    [m]

    target wave train transformed to position of wave maker

    50 100 150 200 250 30010

    0

    10

    t [s]

    U [V]

    control signal for upper flap of wave maker

    50 100 150 200 250 30010

    0

    10

    t [s]

    U [V]

    control signal for main flap of wave maker

    50 100 150 200 250 300

    0.2

    0

    0.2

    t [s]

    [m]

    registration at stationary wave probe close to wave maker (x = 5.19 m)

    190 200 210 220 230 240 250 260 270 280 290 300

    0.2

    0

    0.2

    t [s]

    [m

    ]

    wave train at x = 125 m, calculated from registration at x = 5.19 m

    180 190 200 210 220 230 2400

    50

    100

    t [s]

    x [m] ship position in xdirection

    180 190 200 210 220 230 2400.30.20.1

    00.10.20.3

    t [s]

    [m]

    wave at ship (moving reference frame)

    180 190 200 210 220 230 24050

    25

    0

    2550

    t [s]

    []

    capsizing of the RoRo vessel

    Figure 3. Experimental realization of JONSWAP wave sequences fromsimulation (see Fig. 2).

    Figure 4. Qualitative comparison of numerical simulation and experi-mental realization of capsizing due to parametric excitation (compareFig. 3).

    5 Copyright c 2005 by ASME

  • 8/7/2019 OMAE2005-67123

    6/8

    0 50 100 150 200 2500

    10

    20

    30

    40

    50

    60

    70

    80

    measured x-position of ship model

    x-position for simulation

    t [s]

    x [m]

    Figure 5. The measured track data (x-coordinate) of the ship model isreferred to the starting position in the simulation.

    0 50 100 150 200 250-0.4

    -0.2

    0

    0.2

    0.4measured wave train at x = 5.19 m

    t [s]

    [m]

    0 50 100 150 200 250-0.4-0.2

    0

    0.2

    0.4wave train at starting position of ship, x = 67 m

    t [s]

    [m]

    0 50 100 150 200 250-0.4

    -0.2

    0

    0.2

    0.4time-shifted wave train at x = 67 m

    t [s]

    [m]

    Figure 6. Transformation of wave train according to modied ship trackdata: The blue graph according to the track data of same colour in Fig. 5is registered and transformed to the starting position of the ship (greengraph). After time shift the wave train (red graph) corresponds to themodied red x-position of the ship.

    in the x-position versus time compared to the model tests andthus change the wave train encountered by the ship which in turnmakes a direct comparison of simulation results and model testresults difcult to impossible. Thus it was decided to guide theship regarding the longitudinal position over time exactly as mea-

    sured in the model experiment consequently the wave train en-countered in the simulation corresponds well to the wave trainfrom the experiment.

    The initial conditions were chosen such that the rst tworoll peaks seem sufciently well represented. Variations of theseidentied initial conditions were also tested to identify the de-pendency and susceptibility between the roll response and initialconditions.

    DISCUSSION OF RESULTSThe results for two different model test runs with slightly

    different starting conditions are given in Figs. 7 and 8. The mea-sured stationary wave sequence from a JONSWAP sea state withT P = 11 .5 s and H s = 10 m (model scale 1:34) is transformed tothe moving reference frame of the ship sailing at average speedv = 10 .4 kn and course angle = 3 . The course changes closeto the wall (due to z-manoeuvre in the model tests) were disre-garded in the simulation.

    Fig. 7 shows the model test result for this wave sequence incomparison with results from two numerical simulations in thesame wave train but with slightly different initial conditions. Thetwo simulation runs show very similar results up to the actualcapsize which occurs with a difference of one wave crest. Thecomparison with the model test results shows a good agreementin the rst wave group. Further in time the simulation seemsto damp the roll motion too fast. A further investigation into themodel test data shows, that the model performed a course changeat this time. This inuences the development of the roll motionfor quite some time, while the roll response in the wave group

    leading to the actual capsize shows a good agreement betweenmodel test and simulation again around 350 s. Following the nextroll peak the tank model caught water on the upper deck whichleads to a rather sudden roll damping until the water ooded outagain. Water on deck was not included in the simulation. Never-theless the vessel is not safe in the model test either and capsizesshortly afterwards.

    In Fig. 8 can be observed that the ship model encounters thewave sequence slightly differently compared to the previous testrun. Again the roll response agrees well comparing the modeltest results and the numerical simulation of the same sequence inthe rst wave group. After approximately 100 s, a course changeis performed by the ship model which in this case leads to a rather

    sudden decrease of the roll angle which is not met in the numeri-cal simulation as rudder forces were disregarded. Neverthelessthe roll response in the fatal wave group (300 s ff) correspondswell.

    A comparison of the two model test runs shows the repro-ducibility of the capsizing scenario applying the deterministicwave sequences and model test technique (Fig. 9). The vesselencounters both wave groups under almost identical conditionsand therefore rolls and capsizes accordingly. The comparison

    6 Copyright c 2005 by ASME

  • 8/7/2019 OMAE2005-67123

    7/8

    0 50 100 150 200 250 300 350 400 450-10

    0

    10

    [m]

    stationary wave sequence from model test

    0 50 100 150 200 250 300 350 400 4500

    1000

    2000

    x [m]

    x-position of ship from model test

    0 50 100 150 200 250 300 350 400 450-10

    0

    10

    [m]

    wave sequence at ship model

    0 50 100 150 200 250 300 350 400 450

    -50

    0

    50

    t [s]

    []

    measured (blue) and simulated (red) roll motion

    Figure 7. Comparison of time series from numerical and experimentalsimulation of capsizing due to parametric excitation.

    0 50 100 150 200 250 300 350-10

    0

    10

    [m]

    stationary wave sequence from model test

    0 50 100 150 200 250 300 3500

    1000

    2000

    x [m]

    x-position of ship from model test

    0 50 100 150 200 250 300 350-10

    0

    10

    [m]

    wave sequence at ship model

    0 50 100 150 200 250 300 350

    -50

    0

    50

    t [s]

    []

    measured (blue) and simulated (red) roll motion

    0 50 100 150 200 250 300 350-20

    0

    20

    t [s]

    []

    measured course angle

    Figure 8. Comparison of time series from numerical and experimentalsimulation of capsizing due to parametric excitation.

    also shows the inuence of the course change and the impact of slight differences on the roll motion in between the wave groupswhere the period of encounter does not match the ships naturalperiod of roll in one case the roll motion is enforced in theother case damped. Nevertheless the capsize in the second wavegroup is independent of this history.

    Concluding, the roll responses (and the capsize) are reliably

    0 50 100 150 200 250 300 350 400 450

    -10

    -5

    0

    5

    10

    [m]

    wave sequence at ship model

    0 50 100 150 200 250 300 350 400 450

    -60

    -40

    -20

    0

    20

    40

    60

    t [s]

    []

    roll motion

    Figure 9. Comparison of two different capsizing test runs with repro-ducible conditions.

    predicted by the numerical simulation as long as resonance andstability loss effects dominate the roll motion while the effect of the course change as well as the water on deck leads to differ-ences between model test results and numerical investigations.This is not surprising as rudder forces and water on deck werenot accounted for in the simulation. Thus, we can summarize:

    1. Capsizes due to resonance and stability variations/loss were

    reliably predicted, not only qualitatively but also quantita-tively.

    2. When investigating intact safety, the numerical tool needsto be validated and the theoretical limits of the theory mustthoroughly be evaluated regarding their potential inuenceon quantitative results.

    SUMMARY AND PERSPECTIVESDeterministic wave sequences and capsizing experiments

    provide a reliable means to investigate phenomena endangeringships and deliver a basis for detailed validation and improvement

    of numerical tools.Numerical tools can provide a basis for the direct assess-ment of ship safety in various seas, not only qualitatively but alsoquantitatively as long as the tools are validated with respect totheir reliability and applicability and the restrictions of the theo-retical model are thoroughly accounted for.

    Future developments within S IN S EE will consider furthervalidation cases and aim at the inclusion of steering forces indirect comparisons.

    7 Copyright c 2005 by ASME

  • 8/7/2019 OMAE2005-67123

    8/8

    ACKNOWLEDGMENTThe authors are indebted to the German Federal Ministry

    of Research and Education (BMBF) for funding the projectS IN S EE .

    REFERENCES[1] Kr oger, P., 1987. Simulation der Rollbewegungen von

    Schiffen im Seegang . PhD thesis, Universit at Hamburg.[2] Petey, F., 1988. Abschlussbericht zur Erweiterung des

    Vorhabens Leckstabilit at im Seegang. Tech. rep., Institutf ur Schiffbau der Universit at Hamburg.

    [3] S oding, H., 1987. Ermittlung der Kentergefahr aus Bewe-gungssimulationen. Schiffstechnik, 34 , pp. 2839.

    [4] S oding, H., 1987. Simulation der Bewegungen intakter undlecker Schiffe. 23. Fortbildungskurs, IFS.

    [5] Cramer, H., and Kr uger, S., 2001. Numerical CapsizingSimulations and Consequences for Ship Design. In STGSummermeeting Gdansk, Poland.

    [6] Clauss, G., and Hennig, J., 2002. Computer ControlledCapsizing Tests with Tailored Wave Sequences. In OMAE2002 - 21st Conference on Offshore Mechanics and ArcticEngineering. OMAE2002-28297.

    [7] Hennig, J., Brink, K.-E., and Kuehnlein, W., 2003. Inno-vative deterministic seakeeping test procedures. In STAB2003- 8 th International Conference on theStabilityof Shipsand Ocean Vehicles.

    [8] Clauss, G., and Hennig, J., 2004. Deterministic Analy-sis of Extreme Roll Motions and Subsequent Evaluation of Capsizing Risk. In International Shipbuilding Progress,

    vol. 52, Issue 2.[9] Clauss, G., Hennig, J., Brink, K.-E., and Cramer, H., 2004.A New Technique for the Experimental Investigation of Intact Stability and the Validation of Numerical Simula-tions. In International Workshop on the Stability of Shipsand Ocean Vehicles.

    [10] Clauss, G., Hennig, J., Schmittner, C., and K uhnlein, W.,2004. Non-linear Calculation of Tailored Wave Trainsfor the Experimental Investigation of Extreme StructureBehaviour. In OMAE 2004 - 23rd International Con-ference on Offshore Mechanics and Arctic Engineering.OMAE2004-51195.

    [11] Clauss, G. F., Hennig, J., and Schmittner, C. E., 2004.

    Modelling Extreme Wave Sequences for the Hydrody-namic Analysis of Ships and Offshore Structures. InPRADS 2004 - 9th International Symposium on PracticalDesign of Ships and Other Floating Structures.

    [12] Cramer, H., Kr uger, S., and Mains, C., 2004. Assessmentof Intact Stability Revision and Development of StabilityStandards, Criteria and Approaches. In 7th internationalworkshop on stability and operational safety of ships.

    [13] Cramer, H., Reichert, K., Hessner, K., Hennig, J., and

    Clauss, G. F., 2004. Seakeeping Simulations and SeawayModels and Parameters Supporting Ship Design and Oper-ation. In PRADS 2004 - 9th International Symposium onPractical Design of Ships and Other Floating Structures.

    8 Copyright c 2005 by ASME