An analysis of direct (mathematical programming) and optimality criteria (indirect search) methods for structural optimization is presented. It is shown that both approaches require many similar calculations in their respective numerical algorithms. Based on this analysis, it is concluded that efficient and general hybrid methods can be developed, exploiting common features of both direct and optimality criteria methods. A numerical example is solved to assess the potential of hybrid methods. Extensions of the hybrid methods to dynamic response problem are discussed.
Read full abstract