Structural Optimization Tool using Genetic Algorithms and Ansys

When it comes to solving nonconvex, discontinuous, or discrete problems in Structural Optimization (e.g. maximizing flrst eigenfrequency of a structure), the use of computationally expensive Genetic Algorithms (GA’s) gets interesting. GA’s are stochastic optimization algorithms based on natural selection and genetics. In contrast to traditional gradient-based methods, GA’s work on populations of solutions which evolve typically over hundreds of generations. A tool is presented, which applies GA’s to solve typical problems in structural optimization, integrating ANSYS on a UPF (User Programmable Features) level to evaluate the objective function (fltness values of GA individuals). To overcome e‐ciency limits, the method is implemented for parallel evaluations on a workstation cluster. The performance of the software tool is shown by two real world applications, the frequency optimization of a complex machinetool frame and the weight minimization of a fuel cell plate.