Building empirical model through simulation - A case study

Simulation and design of experiment are commonly used in engineering design and optimization today. In this work, we apply the methodology of the design of experiment and response surface method to finite element analysis for the building of an empirical model. This eliminates the need for expensive resources to carry out the actual experiments. The developed empirical model enables us to obtain global optimization of a design. To further reduce the resources needed for the development of the model, Draper-Lin small-composite design is used, as it requires the minimum number of "experimental run". However, the design will not produce an accurate model. We present a modification method to improve the model developed, making it accurate for the optimization of our device design.