The design of the structural supports has always been practically important in engineering applications. In addition to holding a structure properly, supports can also be utilized to improve the structural performances. In this study, by using modified finite element method (MFEM) and Imperialist Competitive Algorithm (ICA), the maximum of bending moment was minimized. In this paper both elastic and rigid supports are taken into account. As compared to other design optimization methods, ICA is robust, more efficient, and requiring fewer number of function evaluations, while leading to better quality of results. Appling the modified finite element method not only reduces computational cost and increases convergence rate, but also reach the global optimum position of supports. Three classical examples are given to demonstrate the validity and capability of the proposed optimization procedure for finding the global support positions. Results show that support position optimization by using present method, can reduce the maximal moment significantly, and deserves more investigation.