Shape design optimization of cylindrical tank using b-spline curves

Abstract

In this paper, a novel tank design algorithm is proposed for enhancing roll stability of fuel tank shape considering the multi-objective genetic algorithm, sum of weighted cost function and Pareto solution. Two conflicting objective functions, fluid c.g. height and overturning moment, are well thought out in the present optimization processing. The proposed novel algorithm is presented based on using b-spline function to optimize tank cross section. This algorithm receives tank capacity as an input, and then a modified cross-section is proposed with the aid of third order b-spline functions using 8 control points, initially. Consequently, with averaging and optimizing, range of control points are modified and the numbers of control points as well as the order of b-spline functions are enhanced into fourth order with 10 control points. The optimized results of Pareto solution are compared with sum of weighted cost functions to explore the best method for fuel tank design. The comparison indicates that both solutions are in a good agreement although the Pareto solution indicates a faster convergence. In this respect, investigation of certain operators such as selection, crossover, mutation and elitism has been carried out to provide a proper design. The results show that, the mutation rate seems to be taken in the range of 4-6% and the number of individuals in each generation should be at least 60. Comparing the GA results with conventional method, demonstrates that the optimized tank shape reveals more roll stability while its rollover threshold is approximately 6% higher than conventional tanks.