Die-Face Design Optimization Using Response Surface Method with Space Mapping Technique

Conventional optimization algorithms are effective in solving general engineering problems, but lose their power in highly non-linear metal forming process. Our approach shows that a polynomial based on response surface method can be successfully utilized to eliminate computational noise and help to converge to a global optimum. Surrogate models in space mapping are used to determine response surfaces with their associated gradients, and then a fine model is applied to correct the design point for the next iteration. Thus space mapping makes it possible to reduce the total computation time to reach optimum solution. The proposed methodology has been realized with FE simulations, and examples demonstrate that it is efficient, accurate and robust for metal forming die-face design optimization.