Vous êtes sur la page 1sur 8

Proceedings of OMAE Specialty Conference on Integrity of Floating Production, Storage& Offloading (FPSO) Systems August 30 - September 2, 2004, Houston, TX

OMAE-FPSO04-0073
Analysis of wave induced loads on a FPSO due to abnormal waves
Carlos Guedes Soares Unit of Marine Technology and Engineering, Technical University of Lisbon, Instituto Superior Tcnico, Lisboa, Portugal Email: guedess@mar.ist.utl.pt Guenther F. Clauss Technical University Berlin Ocean Engineering Section Berlin, Germany Email: clauss@ism.tu-berlin.de Nuno Fonseca Unit of Marine Technology and Engineering, Technical University of Lisbon, Av. Rovisco Pais, 1049-001 Lisboa, Portugal Email: nfonseca@mar.ist.utl.pt Christian E. Schmittner Technical University Berlin Ocean Engineering Section Berlin, Germany Email: schmittner@ism.tu-berlin.de Ricardo Pascoal Unit of Marine Technology and Engineering, Technical University of Lisbon, Instituto Superior Tcnico, Lisboa, Portugal Email: rpascoal@mar.ist.utl.pt Janou Hennig Technical University Berlin Ocean Engineering Section Berlin, Germany Email: hennig@ism.tu-berlin.de

ABSTRACT The paper presents an analysis of structural design wave loads on a FPSO. The vertical bending moment at midship induced by rogue waves are compared with rule values. The loads induced by deterministic rogue waves were both measured in a seakeeping tank and calculated by an advanced time domain method. Two procedures are used to calculate the expected extreme vertical bending moment during the operational lifetime of the ship. The first one relies on a standard linear long term prediction method, which results from the summation of short term distribution of maxima weighted by their probability of occurrence. The short term stationary seastates are represented by energy spectra and the ship responses by linear transfer functions. The second procedure to calculate the design wave loads is a generalisation of the former, however it accounts for the nonlinearity of the vertical bending moment. It uses nonlinear transfer functions of the bending moment sagging peaks which depend of the wave height.

Thus various hydrodynamic methods were developed and calibrated with experiments and various statistical descriptions of climatic wave variability were developed. Methods for long term estimation of the design loads were developed in which the information about the probability of occurrence of the sea states was combined with the ship response to each of these excitations. Analysis of accidents indicated that in several cases the existence of a specific wave with very extreme proportions was critical and often responsible for major damage or even loss of the structure. Suggestions have also been made that the approach to assess design loads should be revised and that design situations should be based on the survival to the so-called freak, rogue or abnormal waves (Faulkner, and Buckley, 1997). This has led to queries about the nature of these waves and about their predictability, both in terms of probability and of the geometric characteristics of their profile (Guedes Soares et al., 2003). Techniques have also been improved to represent the large waves in model testing tanks and to include them in sequences of random waves (Clauss et al., 2004a). So, being possible to reproduce these situations in tank and being able to calculate them with numerical methods, it is natural to conduct models tests to validate the calculations against test data. A model test program has been conducted in a seakeeping tank to measure the motions and load effects on a FPSO hull

INTRODUCTION

The traditional approach to estimate the design loads for ships and offshore structures was to estimate the linear or eventually the nonlinear wave induced load effects on the hull and to extrapolate them to values of low probability of occurrence.

Copyright 2004 by ASME

subjected to rogue waves and also to severe seastates (without rogue waves). In Clauss et al (2004b) the model tests are described and some comparisons with predictions of the vertical bending moment at midship are provided. This paper complements the referred one by providing additional results of the motions and bending moments experienced by the model and comparing them with the results of calculations. A nonlinear time domain seakeeping program is used to estimate the loads induced by selected time series of wave elevation, which include a rogue wave. Rogue waves are extremely high and steep waves and in fact the existing numerical methods are still not validated for these conditions. The experimental results are analysed and used to validate the seakeeping program. Additionally the expected extreme vertical bending moment during the operational lifetime of the ship is estimated using two methods. The first is based on a standard linear long term prediction method, which results from the summation of short term distribution of maxima weighted by their probability of occurrence. The second one is a generalisation of the former, however it accounts for the nonlinearity of the vertical bending moment. It uses nonlinear transfer functions of the bending moment sagging peaks which depend of the wave height. Finally all calculated and measured loads are compared with rule values 2 EXPERIMENTAL PROGRAM Figure 2 Side view of the FPSO midship vertical position of the connecting elements can be changed (photo: measuring device at waterline level) The seakeeping tests have been performed in head seas, and the New Year Wave is generated to investigate the corresponding impact. Furthermore, the encountering position of New Year Wave and FPSO is varied. The transfer functions are obtained by using deterministic wave packets (Clauss, Khnlein 1995). The motions of the model are registered using a dedicated video system. 2.2 Wave Generation To simulate rogue wave sequences experimentally the target position xtarget of the deterministic wave/structure interaction is selected. At this location, the target wave train is either specifically designed or reproduced from an existing wave registration like the New Year Wave. This wave train is transformed upstream to the position of the wave maker, which requires an adequate wave propagation model (Clauss et al. 2004a). On the basis of linear wave theory the specified amplitude distribution of the target wave train is given as Fourier transform F(,xtarget) with circular frequency as a function of wave number k. Adaptation of the phase spectrum to the wave maker location x0 gives the Fourier transform in x0:
F( , x 0 ) = F ( , x target ) e
i (t k ( x target x 0 ))

2.1 Setup of Equipment For the experimental investigation a wooden model of the selected FPSO design has been built at scale 8 = 1:81 (Fig. 1). The model is cut into three segments with intersections at Lpp/4 and Lpp/2 measured from bow. The main dimensions of the FPSO are L = 280.78m, B = 46m, T = 16.67m, displacement = 174000t, CB = 0.87, and the bending moments are measured at Lpp/4 and Lpp/2.

(1)

For reducing the number of time steps until the wave maker starts to operate, the wave train is shifted by time tshift, i.e.
F( , x 0 ) e i tshift

(2)

Figure 1 Segmented model of the FPSO during tests in the wave tank (scale 1:81) At the intersections the connecting steel elements on both sides are equipped with strain gauges for the registration of vertical bending moments. These moments consist of the contributions caused by vertical forces and by surge loads. Due to the fact that the entire system acts as a dynamic, oscillating system the vertical position of the connecting elements is essential for the registration. To investigate this effect, the measuring devices can be attached to the model at different heights (Fig. 2). Measurements have been carried out both with connecting elements at deck level and at water line level.

For extremely high wave trains like the New Year Wave the associated wave propagation is modelled by a semiempirical nonlinear method which is based on the fact that short and high wave groups with strong nonlinear characteristics evolve from long and low wave groups which are characterized by linear principles (Clauss and Khnlein 1997, Khnlein et al. 2002). As the total energy of the wave is invariant during its metamorphosis, the initial Fourier transform of the linear wave train is introduced as "wave information" and selected as the backbone of wave propagation. This wave information along with the adequate celerity at each time step give the nonlinear phase characteristics of the wave train. The detailed shape of the

Copyright 2004 by ASME

resulting wave train is developed at each time step considering the temporary steepness of the wave. This iteration principle can be used either for backward transformation (wave generation) as well as for forward transformation (calculation of the moving reference frame wave train at cruising ships (Clauss and Hennig 2003). From the wave train at the position of the wave maker the corresponding control signals are calculated using the hydrodynamic transfer function (relating wave board motion to wave elevation), the geometrical RAO (which considers the flap type), and the electric-hydraulic RAO. This control signal is used to generate the specified wave train which can be measured at the target position in the tank. Fig. 3 compares model test results (tank dimensions: length 80m, width 4m, water depth 1.5m, piston type wave maker) with the New Year Wave registration.

zero for the zero speed case), and convolution integrals of memory functions. The convolution integrals represent the effects of the whole past history of the motion accounting for the memory effects due to the radiated waves. The vertical forces associated with the green water on deck, which occurs when the relative motion is larger than the free board, are calculated using the momentum method (Buchner, 1995). The mass of water on the deck is proportional to the height of water on the deck, which is given by the difference between the relative motion and the free board of the ship. The methodology to produce the correct wave field and the associated wave exciting forces which are consistent with a deterministic wave record measured at one point fixed in space is presented in Fonseca et al. (2001) and Guedes Soares et al. (2003). 3.2 Transfer Functions The transfer functions of heave and pitch motions have been obtained from Fourier analysis of the experimental records in irregular waves corresponding to several significant wave heights (Hs) and peak wave periods (Tp). Figure 4 presents the amplitudes of the experimental transfer functions together with numerical results. Both heave and pitch amplitudes are normalised by the wave amplitude. The agreement between experimental data and numerical results is reasonable. For heave one can observe that the experimental transfer function for low frequencies is not correctly obtained from the Fourier transform of the time signals, since the normalised amplitudes are significantly smaller than one. The reason is probably related to the low energy levels of both the seastate and response for low frequencies. The transfer function is obtained from the quotient between the response spectrum and wave spectrum. If the energy is small for both spectra at the low frequency range, then the quotient is more prone to be affected by measuring errors. The same type of behaviour was observed in the experimental data of the Octoplus floating platform (Pascoal et al., 2004). Regarding pitch, the numerical results overestimate the experimental data for the wave frequencies around 0.50 rad/s. A larger dispersion of the experimental transfer function is observed at the low frequency range. The reason is believed to be the same as presented before for heave. Figure 5 presents comparisons between the experimental and calculated vertical bending moment at midship transfer function amplitudes. The graphs are in a scale of wave frequencies and the amplitudes are normalised by the wave amplitude. The experimental results are obtained from Fourier analysis of the measured records in the New Year Waves. In fact the wave is always the same but the large crest (approx. 18m) occurs at different positions along the ship longitudinal direction. The numerical results compare well with the experiments, although the numerical peak of the transfer function occurs at a slightly higher frequency than the experimental one. At the low frequency range the numerical amplitudes underestimate the experimental ones. Again it is believed that the numerical predictions are correct and the experimental data is affected by inaccuracies associated with small energies of the wave and response

Figure 3 Comparison of model wave (measured at scale 1:81) with the registered New Year wave presented as full scale data 3. COMPARISONS BETWEEN EXPERIMENTAL DATA AND NUMERICAL RESULTS 3.1 Numerical Model The calculations are done using a time domain nonlinear strip theory code. The method assumes that the nonlinear contribution for the vertical bending moment is dominated by hydrostatic and Froude-Krilov forces, thus these components depend of the instantaneous hull wetted surface. Radiation and diffraction forces are linear. Additionally green water loads on the deck, which contribute to the calculation of motions and global loads, are represented by the momentum method. A detailed presentation of the method is given by Fonseca and Guedes Soares (1998a, 1998b) The exciting forces due to the incident waves are decomposed into a diffraction part and the Froude-Krilov part. The diffraction part, which is related to the scattering of the incident wave field due to the presence of the moving ship, is kept linear. Since this is a linear problem and the exciting waves are known a priori, it can be solved in the frequency domain and the resulting transfer functions be used to generate a time history of the diffraction heave force and pitch moment. The Froude-Krilov part is related to the incident wave potential and results from the integration at each time step of the associated pressure over the wetted surface of the hull under the undisturbed wave profile. The hydrostatic force and moment are calculated at each time step by integration of the hydrostatic pressure over the wetted hull under the undisturbed wave profile. The radiation forces, which are calculated using a strip method, are represented in the time domain by infinite frequency added masses, radiation restoring coefficients (which are

Copyright 2004 by ASME

spectra at low frequencies. The comparisons between experimental records and time domain simulations presented in the next section, confirm that the numerical model is able to represent very well the experimental data.
Heave
1.2 1.0 0.8
3(m/m)

Hs=14.5, T p=16 Hs=13, T p=13 Hs=13, T p=18 Hs=12, T p=12 Hs=10, T p=10

relatively large number of harmonics that are used to represent with good accuracy around 10 minutes of the measured wave elevation (full scale). The good agreement of wave elevations can be observed in the first plot of figure 6. If the objective is only to assess the maximum ship responses due to the large wave crest, then a simulation of around 60s is enough and in this case it is possible to reduce considerably the number of harmonic components. The benefit of using smaller duration time traces reflects on the reduced computational effort.
numerical experimental
16.0

0.6 0.4 0.2 0.0 0.20

Numerical

wave(m)
0.30 0.40 0.50 0.60 0.70 0.80

8.0 0.0 -8.0 400 450 500 550 600 650

(rad/s)

t(s)

700

5.0

Pitch
1.0 0.8
5 (deg./m)

Hs=13, T p=13 Hs=13, T p=18 Hs=12, T p=12

heave(m)

Hs=14.5, T p=16

2.5 0.0 -2.5 -5.0 400 6.0 450 500 550 600 650

0.6 0.4 0.2 0.0 0.20 0.30 0.40 0.50 0.60

Hs=10, T p=10

t(s)

700

Calc. linear
pitch(deg)

3.0 0.0 -3.0 -6.0 400 450


5.0E+006

500

550

600

650

t(s)

700

0.70

0.80
vbm(kNm)

(rad/s)
Figure 4 Transfer function amplitudes of heave and pitch motions
Transfer Function of VBM at midship
7.0E+05 6.0E+05 5.0E+05 VBM (kNm) 4.0E+05 3.0E+05 2.0E+05 1.0E+05 0.0E+00 0 0.2 0.4 0.6 0.8 1 frequency (rad/s) wave 2895 wave 2415 wave 2575 wave 3735 wave 2815 wave 2975 Num. - IST

2.5E+006 0.0E+000 -2.5E+006 -5.0E+006 450 500 550 600 650

t(s)

700

Figure 6 Measurements and simulations of the wave at midship, heave, pitch and the vertical bending moment at midship. NYW crest occurs at midship. Regarding the FPSO responses to the waves, the agreement between measured and simulated heave, pitch and vertical bending moment at midship is very good. The dashed and continuous lines are not coincident, and one would not expect them to be, however the numerical model is able to represent well the magnitudes as well as the phases of the responses. One should note that the spatial distribution of the simulated wave elevation is extrapolated from the wave elevation measured at one point fixed in space, assuming linear superposition of wave harmonic components. This in fact seems to be a good approximation for the incident wave field, even for a severe seastate, if the objective is to calculate the ship responses. 3.4 Influence of wave crest position One parameter that might influence the maximum wave structural loads to an abnormal wave is the position in space, with respect to the ship, where the abnormal wave is created. To investigate this effect, nine different positions of the NYW crest along the longitudinal direction of the ship were used. With respect to midship, the positions in space where the large wave crest is generated are: 3 / 2 L pp , L pp , 1/ 2 L pp ,

Figure 5 Transfer function amplitude of the vertical bending moment at midship 3.3 Comparisons of time histories Figure 6 presents comparisons between measured and simulated wave elevations at midship and the FPSO responses. Dashed lines represent measured signals and the continuous lines represent calculated time histories. The large wave crest, which had an elevation of approximately 16m, occurs around t = 670s and is generated at midship. For the simulations presented through-out the paper, 200 harmonics are used to represent the measured wave elevation and also the related exciting forces. This is a

Copyright 2004 by ASME

1/ 4 L pp , 0. , 1/ 4 L pp , 1 / 2 L pp , L pp , 3 / 2 L pp . Figure 7

presents measured (dashed lines) and simulated (continuous lines) time histories of the vertical bending moment at midship for the nine waves, where the arrows relate the graphs with the positions where the large wave crest occurs. The length between perpendiculars is 259.8m and the results are in kNm. The simulated hogging bending moments compare very well with the measured ones for all waves. The largest measured sagging moments occur half of a cycle before the large wave crest passes at midship, meaning when the crest is at the bow. Interestingly the simulated sagging moment that occurs half of a cycle after the crest passes at midship tends to be significantly larger than the measured one. It is possible that this is related with the viscous effects associated with the large relative vertical motions between the hull and the waves. Figure 8 presents a comparison between experimental and numerical maximum hogging and sagging peaks which are associated with the large crest passing through the ship. Results are presented as function of the large crest position along the ship, where nine points correspond to nine waves. From the observation of the graph, the first conclusion is that the position where the large wave crest is created has a small influence on the maximum vertical bending moment at midship. Secondly one observes that the measured bending moment has a nonlinear characteristic since sagging peaks are in average 25% larger than the hogging peaks. This means that in order to predict with accuracy the wave induced vertical bending moment on this FPSO one has to apply a nonlinear computer code, such as the one used for the present study. Finally the numerical results compare remarkably well with the experimental peaks, including the mentioned asymmetry effects. The asymmetry of the vertical bending moment is mostly related to the shape of the hull. The FPSO under study has a hull with vertical sides under the design waterline and some

flare above the waterline. The bow flare above the waterline is small and similar to the flare of tankers, and for this reason the nonlinear effects on the vertical bending moments associated to the bow flare are small. For ships with large bow flare the nonlinear effects on the vertical bending moment are expressed mainly by the asymmetry of the sagging and hogging peaks, where the ratio sagging / hogging can reach values of 2-3. For this FPSO, both the experimental data and the numerical results show ratios of around 1,25.
6.0E+06 4.0E+06 2.0E+06

VBM (kNm)

0.0E+00 -2.0E+06 -4.0E+06 -6.0E+06 -8.0E+06 -500 -300 -100 100 300 500

Exp. - Sag Max Exp. - Hog Max Num. IST - Sag Max Num. IST - Hog Max

Position of NYW crest with respect to midship (m)

Figure 8 Comparison of experimental and numerical maximum hogging and sagging peaks, which are associated with the large crest passing through the ship. Results presented as function of the large crest position along the ship. Length between perpendiculars is 259.8m The shape of the hull above the static waterline is taken into account in the calculation of the buoyancy type forces (but not in the radiation and diffraction forces). Bow flare slamming loads are not taken into account also. Bow flare slamming (vertical) forces are believed to be relatively small for several reasons: the flare of the hull is small, the time records of the measured vertical bending moment do not include the typical slamming loads (large peaks if small time duration), and finally the numerical results without slamming effects compare very well with the experimental data.

5.0E+006 2.5E+006 0.0E+000 -2.5E+006 -5.0E+006 700 700 700 650 650 650 650 650

t(s)

650

Figure 7 Measured (dashed lines) and simulated (continuous lines) time histories of the vertical bending moment at midship for different positions of the NYW crest along the longitudinal direction (indicated by the arrows). Results in kNm.

Copyright 2004 by ASME

4. CALCULATION OF DESIGN WAVE BENDING MOMENT 4.1 Calculation of Long Term Distributions In this section design wave induced vertical bending moments are estimated using linear and pseudo nonlinear long term distributions and assuming an operational life of 20 years in the North Sea. The resulting values are compared with the maximum moments due to the New Year Wave. The procedure for calculating long-term cumulative probability distributions of wave induced linear responses is a well defined and accepted one. The basic assumption of the method is the linearity of the ship responses with respect to the exciting waves and the Gaussian properties of the seastate. With these premises linear spectral analysis may be applied to derive simple statistics of the seastate and ship responses. The first step in the calculation procedure is to determine the response transfer function, or amplitude response as function of the wave frequency. Then the response spectrum is obtained from the linear operation on the input wave spectrum by the transfer function. The seastate is modelled as a stationary, zero mean, Gaussian process and because the responses are linear, the response process will be described by the same model. This together with the assumption of a narrow band process implies that a Rayleigh distribution describes the amplitudes or the peaks of the processes (Longuet-Higgins, 1952). The former describes the short term model. If the predictions are made for periods of longer duration, then the seastate can not be considered stationary and the solution is to consider a succession of periods of stationary seastates during which the process is defined by an uncertain value of the variance. Therefore the Rayleigh distribution becomes conditioned on the value of the variance and the marginal distribution must be obtained by integrating over all values of the variance, weighted by their probability density function. The probability of exceeding a maximum of the response in a long-term is readily obtained (Fukuda, 1967, and Guedes Soares and Moan, 1991). In order to introduce the nonlinear asymmetry of the vertical bending moment on the calculation procedure, Guedes Soares (1993) and Guedes Soares and Schellin (1998) proposed a generalization of the former method. The new method uses form functions that transform the linear transfer functions to nonlinear transfer functions associated with different seastates. The form function depends of the wave height and frequency and may be represented by:

represented by the form factor 2 , is easily introduced in the long term distribution calculation. For the present investigation, a variant of the later procedure has been applied to estimate the maximum expected bending moment during the lifetime of the FPSO. The difference is that all response spectra are computed using directly calculated nonlinear transfer functions, instead of using form functions. The nonlinear transfer functions are computed with a nonlinear time domain code. 4.2 Comparison of wave Design Bending Moments The two procedures described in the previous section were applied to calculate the long term distribution of the wave induced vertical bending moment at midship on the FPSO. It is assumed that the ship operates in the North Sea and always in head waves. The wave climate statistics used are based on visual observations (Hogben, da Cunha and Olliver, 1986), and the scatter diagram corresponds to area 8 from the Global Wave Statistics data base.
10
0

Wave Induced Bending Moment Linear Nonlinear

10 Exceedance Probability

-2

10

-4

10

-8

-8 P=10

10

-10

4 5 M [kNm]

Linear
7

10

Non Linear
8 x 10 9
6

-6

Figure 9 Long term distribution of the wave induced bending moment at midship Figure 9 presents the long term distributions of the wave induced vertical bending moment at midship, using the linear transfer function and the nonlinear sagging transfer functions. The graph includes also the design values as required by DNV rules. The horizontal line represents the level for the probability of exceedance of 10-8, which corresponds to an operational life of approximately 20 years. Usually the probability levels considered for the design of ship structures is 10-8, therefore this is the reference value considered for the purpose of comparison with the structural loads induced by the rogue wave. Presently there is no basis to assess the probability of occurrence of rogue waves, and for this reason calculations for this level of probability are not included. Table 1 compare calculated and rule design wave bending moments together with the maximum measured wave bending moment in the New Year Waves. One observes that the calculated design bending moment from long term distributions are significantly larger than the rule values. The nonlinear sagging design moment is 47% larger than the rule value, while the linear one is 31% larger. Discrepancies of this type between rule values and long term

NL ( , H s ) = 1 ( ) 2 (H S )
where function of the wave height only. The functions 1 and 2 are calculated comparing linear transfer functions with pseudo nonlinear transfer functions of the sagging moments. The frequency dependent correction can be applied directly to the linear transfer function before the long term calculations are performed. The nonlinearity of the vertical bending moment with respect to the wave height,

is a function of the frequency only and

is a

Copyright 2004 by ASME

DNV Rules sagging

DNV Rules hogging

predictions were found also for conventional ships by Guedes Soares (1999). When comparing the rule values with the maximum moments measured in the New Year Waves one concludes that the rule values are around 10% larger indicating that the ship would resist to the global vertical bending moment induced by this abnormal wave. Table 1 compares the rule design wave bending moments (DNV rules) with the maximum measured wave bending moment in the New Year Waves. One concludes that the rule values are around 10% larger indicating that the ship would resist to the global vertical bending moment induced by this abnormal wave.
Sagging (kNm) Linear ( p = 10 8 , 20 years ) Nonlinear ( p = 10 8 , 20 years) Rules (DNV) New Year Wave Hogging (kNm) Sagging normalised by rule value 1.30 1.45 1.00 0.91 Hogging normalised by rule value 1.38 1.00 0.85

Clauss, G., and Khnlein, W., 1997. Simulation of design storm wave conditions with tailored wave groups, Proceedings of the 7th Int. Offshore and Polar Engineering Conference (ISOPE), pp. 228237. Clauss, G., and Hennig, J., 2003. Deterministic Analysis of Extreme Roll Motions and Subsequent Evaluation of Capsizing Risk, Proceedings 8th International Conference on the Stability of Ships and Ocean Vehicles (STAB 2003), pp. 5767. Clauss, G., Hennig, J., Schmittner, C., and Khnlein, W., 2004a. Non-linear Calculation of Tailored Wave Trains for the Experimental Investigation of Extreme Structure Behaviour, Proceedings 23th International Conference on Offshore Mechanics and Arctic Engineering (OMAE2004), paper OMAE2004-51195. Clauss, G., Schmittner, C., Hennig, J., Guedes Soares, C., Fonseca, N. and Pascoal, R., 2004b, Bending Moments of an FPSO in Rogue Waves, Proceedings 23rd International Conference on Offshore Mechanics and Arctic Engineering (OMAE2004), paper OMAE2004-51504. Faulkner, D. and Buckley, W. H., 1997, Critical Survival Conditions for Ship Design, Proceedings International Conference on Design and Operation for Abnormal Conditions, RINA, paper no. 6, pp. 1-25. Fonseca, N. and Guedes Soares, C., 1998a, Time-Domain Analysis of Large-Amplitude Vertical Motions and Wave Loads, Journal of Ship Research, Vol. 42, n 2, pp. 100113. Fonseca, N. and Guedes Soares, C., 1998b, Nonlinear Wave Induced Responses of Ships in Irregular Seas, Proceedings 17th International Conference on Offshore Mechanics and Arctic Engineering (OMAE98), ASME, New York, paper OMAE 98-0446. Fonseca, N., Guedes Soares, C. and Pascoal, R., 2001, Prediction of Ship Dynamic Loads in Ship in Heavy Weather, Proceedings of the Conference on Design and Operation for Abnormal Conditions II (RINA), London, United Kingdom, pp. 169-182. Fukuda, J., 1967, Theoretical Determination of Design Wave Bending Moments, Japan Shipbuilding and Marine Engineering, Vol. 2, No. 3, pp. 12-22. Guedes Soares, C., 1993, Long-Term Distribution of NonLinear Wave Induced Vertical Bending Moments, Marine Structures, Vol. 6, pp. 475-483. Guedes Soares, C., 1999, On the uncertainty in long-term predictions of wave induced loads on ships, Marine Structures, Vol. 12, pp. 171-182. Guedes Soares, C., and Moan, T., 1991, Uncertainty in the Long Term Distribution of Wave Induced Bending Moments for Fatigue Design of Ship Structures, Marine Structures, Vol. 4, pp. 294-315. Guedes Soares, C. and Schellin, T. E., 1998, Nonlinear Effects on Long-Term Distributions of Wave-Induced Loads for Tankers, Journal of Offshore Mechanics and Arctic Engineering, Vol. 120, n 2, pp. 65-70. Guedes Soares, C., Fonseca, N and Pascoal, R., 2003, An Approach for the Structural Design of Ships and Offshore

7.3 8.1 5.6 5.1

7.3 5.3 4.5

Table 1 Design wave bending moments CONCLUSIONS The paper presents an analysis of structural wave loads on a FPSO induced by rogue waves. The motions and loads induced by a deterministic rogue wave were both measured in a seakeeping tank and calculated by a nonlinear time domain method. The objective was to assess if such waves should be used as additional conditions to be considered in determining design wave loads on FPSOs. Comparisons between experiments and simulations show that the numerical model is able to represent remarkably well the wave induced bending moment at midship in this highly nonlinear wave, including the asymmetry of the sagging and hogging peaks. The variation of the maximum bending moment with the position in space where the abnormal wave is created was investigated and found to have a small influence. The vertical bending moments measured in the New Year Wave are around 10% bellow the rule values indicating that the ship would resist to the global vertical bending moment induced by this rogue wave. ACKNOWLEDGMENTS This work has been performed in the scope of the research project Rogue Waves Forecast and Impact on Marine Structures (MAXWAVE), partially funded by the European Commission, under the programme Energy, Environment and Sustainable Development (Contract no. EVK3:2000-00544). REFERENCES Clauss, G., and Khnlein, W., 1995. Transient wave packets an efficient technique for seakeeping tests of selfpropelled models in oblique waves. Proceedings Third International Conference on Fast Sea Transportation, pp. 11931204, Vol. 2.

Copyright 2004 by ASME

Platforms in Abnormal Waves, Proceedings of the Maxwave Workshop on Rogue Waves Forecast and Impact on Marine Structures, Geneva, October 8-10. Guedes Soares, C., Cherneva, Z., and Anto, E.M., 2003, Characteristics of abnormal waves in North Sea storm sea states, Applied Ocean Research, in press. Haver, S., Anderson, O., 2000, Freak Wave: Rare Realization of a Typical Population or Typical Realization of a Rare Population?, Proceedings of ISOPE 2000 10th International Offshore and Polar Engineering Conference, pp. 123-130. Hogben, N., da Cunha, L.F. and Olliver, H.N., 1986, Global Wave Statistics, Brown Union London. Khnlein, W., Clauss, G., and Hennig, J., 2002. Tailor Made Freak Waves Within Irregular Seas, Proceedings 21st International Conference on Offshore Mechanics and Arctic Engineering (OMAE2002), paper OMAE2002-28524. Longuet-Higgins, M. S., 1952, On the Statistical Distribution of the Heights of Sea Waves, Journal of Marine Research, Vol. 11, n 3, pp. 245-266. Pascoal, R., Guedes Soares, Facon, G., Ptri, F., Vach, M., 2004, Hydrodynamic Analysis and Motions of the Octoplus Platform, Proceedings 23rd International Conference on Offshore Mechanics and Arctic Engineering (OMAE2004), paper OMAE2004-51574.

Copyright 2004 by ASME

Vous aimerez peut-être aussi