Study On Theories And Methods Return Value Calculation Of Extreme Environmental Conditions With POT Method For Ocean Engineering

At present the world economy has been met the bottleneck period of resources andenvironment, and limited resource and space on land are under enormous pressure.Many countries have taken developing the marine economy as the nationaldevelopment strategy in the future. But, when humanity is preparing to develop andutilizate marine resources on a large scale, our marine environments are becomingworsen. And many more severe oceanic disasters come more frequently than before.Undoubtedly, it puts towards a high request on design of offshore equipments. How tocorrectly predict the probabilities of their occurrence and strength of the extreme oceanenvironmental conditions is essential for safety and economics of ocean engineerings.In the design of ocean engineerings, the accurate predictability of the returnperiod level of the extreme ocean environmental conditions is a very fatal part. Atpresent, sampling method of annual maximum series was used in these major countries.However the method makes a serious waste of the raw data, and can only get very littleinformation from the raw data. For ensuring the rationality and reliability of the result,the annual maximum series method needs the long-term observation data. Moreover,the univariate probability analysis method is commonly used, and a more objectiveanalyze method-the joint probability method is rarely used. In the univariateprobability analysis method, each environmental load is estimated independently andmakes a simply an addition of the all factors, which results in higher design parametersand construction cost. The paper studies mainly the POT (Peak Over Threshold) modelfrom in two sides: one-dimension and multi-dimension, and its main idea andinnovations are as follows:The paper solve mainly those problems about choosing the threshold value andgrouping in analysing extreme ocean environmental conditions with POT. According toa literature search, all grouping methods are summarized and compared theiradvantages and disadvantages. Sticking to the points in grouping: to discriminate thebig wavs from all waves and to driving adjacent the big waves apart，the new method ofgrouping, what is usable in the POT model, is proposed. By contrasting the traditionalmethods, the method of grouping in the paper is feasible, and POT model GPD andPDS have exactly the same results. Using the new grouping method, POT model can beused in every sea area, not in the sea area affected by only typhoon easily, such as thecompound distribution model. At present P-G compound distribution is used quite frequently in lots of projectsin our country. But in some seas, in which typhoon is barely to be seen, there is thepuzzle of determining the threshold value for P-G compound distribution yet. Byanalyzing the principle of determining the thresholding value of P-G compounddistribution, the new method of determining the threshold value is proposed in thepaper combining the change point thory. The new method has the theoreticalunderpinnings of the mathematical principle, so it is not affected by human’ ssubjectivity. And the proposed method can determine the threshold by a scientificquantitative method too. By comparing the traditional methods with search groups oflong-term databases, The proposed method is proved to be viable. For hindcast data ofocean wave, P-G compound distribution can be used for the return wave height in everydirection with the method of determining the thresholding value, but the P-Ⅲ methodcan not do it.The reseach propose multidimensional compound POT distribution model and itsMonte Carlo Simulation method. First, similar to the relationship between generalizedPareto distribution and generalized extreme value distribution, Multivariategeneralized Pareto distribution model is established. And the paper proposes firstly themultivariate Piossion-Pareto compound distribution. It include these influents of theoccurrences of such storms (or the big wave) in certain sea areas and the theoreticaldistribution of multivariate POT model. The method of deciding the joint threshold isproposed, but needs improved further for overcoming contrived factor. According tothe definition of multivariate generalized Pareto distribution, the Monte CarloSimulation method of multivariate generalized Pareto distribution is set up with therejection method. Combining the multivariate compound distribution, the Monte CarloSimulation method of the multivariate Pareto compound distribution is firstly proposed.Because the method is extended to higher dimensional problems easily, a practical wayis found for multidimensional problems.Taking23years field synchronizing data of wave height and wind speed as anexample, the represent return periods of the design base shear of a oil platform areanalyzed. After deciding the joint threthod of wave height and wind speed, the returnperiod levels of the base shear are found combining bivariate POT model of logistictype. Parameters sensitivity analysis to the bivariate pareto distribution model are done,and it is showed that the model is very stable. By comparing the results of the MonteCarlo Simulation and analytic solution, a good agreement with deviation is demonstrated. So it can be knowed that the peoposed Monte Carlo method is entirelyfeasible. The Monte Carlo Simulation method can be extrended to higher dimensionalproblem.