FEA Based Weight Optimization of Semi-Submersible Structure Considering Buckling and Yield Strength

Abstract

Semi-submersible structure is widely used to drill and produce oil and gas in the ocean. This structure is very sensitive to weight increase in terms of payload and stability. As the detailed design progresses, design changes such as lower shape change due to the increase in weight of the upper equipment often lead to delivery delay. Therefore, it is important to secure the weight margin by optimizing the substructure at the initial design phase. There have been many researches on the weight optimization of ship structures in terms of strength. Since the strength assessment procedures of ship are relatively simple, it is possible to repeat the strength assessment process and optimize its weight in terms of strength. However, semi-submersible structure is complicated compared to ship, therefore it requires more complex procedures than conventional methods of ships strength assessment. This strength assessment process of semi-submersible is not currently fully automated. Thus, the offshore structure including semi-submersible structure has been optimized not considering the strength but considering the motion and stability. In order to perform the optimization considering the strength of the structure, it is necessary to automate the strength assessment process. For this reason, necessary processes related to stress such as stress scanning, mapping, and combination is automated as a preliminary study of optimization. The process for strength assessment such as generating panel and assigning panel information is automated. Based on the developed automatic strength check system, weight optimization considering the strength of semi-submersible structure is carried out. The weight is set as an objective function, and the buckling and yield strength were set as constraints. The plate thickness and beam sections are set as design variables. In order to reduce the number of design variables and to exclude solutions of unrealistic beam sections, design variables are discretized. In addition, the steepest descent method is selected as the optimization algorithm to minimize the analysis time. The number of FE analysis is reduced by using the equation for analytically estimating the stress change. In case of optimizing the entire model at once, there are too many design variables to deal with. In the optimization problem, when the number of design variables increases, the optimum point may not be reached exactly. To solve this problem, the design variables of each optimization step are reduced by independently performing optimization for each plane. For the column model of the semi-submersible structure, the optimization using the method presented in this paper is performed to confirm the convergence of the optimal solution.