Abstract: The storage tank is an important component of liquid rockets. However, the finite element model of traditional lightweight design, contains a large number of rib details, and it is difficult to achieve the rapid design with a large number of grids. Firstly, based on the load-bearing characteristics of slender bars, the periodically arranged bars were equivalent to anisotropic cylindrical shells containing axial stiffness and shear stiffness. With equivalent stiffness method, each layer of shell was rigidly connected to the skin to achieve stiffness superposition, and the model size could be reduced. Secondly, according to the design criteria for storage tank wall panels, for the optimization of its structural parameters, the existing structure was designed to reduce weight with sequential quadratic programming algorithm. As facing new design requirements, the multiple island genetic algorithm was adopted to provide optimal solutions for different structure forms in one round of optimization. It's verified the weight of storage container can be reduced by 15%-20%. By comparing with the fine model of the entire storage tank, their deviation is within 5%. The above results indicate the proposed structural simulation and optimization modeling method has good application prospects for the future design of rocket storage tanks.