Abstract:

Gradient based optimization algorithms require gradient of the response functions to find search directions. These gradients are usually computed using finite difference method which becomes computationally very expensive for large number of design variables. The objective of the thesis therefore was to develop an efficient scheme that reduces the overall time required for the optimization cycle, using Ansys and Matlab. In the present work, as an alternate, the semi analytical method is implemented to compute gradients of displacements and stresses. The stiffness matrices required for semi analytical method are extracted from Ansys and gradients are then computed in Matlab. The efficiency of the implemented method is realized considering three case studies of size optimization.