A DOE Based Approach to Multi-Response Optimization

This paper presents a DOE (Design of Experimentation) based approach to optimize systems with numerous responses (y's) and design variables (x's). The approach first conducts a large scale DOE to identify important variables or critical X's and their main effects on the system responses; and then develops “trend of effect” for each important variable. The trend of effects will be used to determine an optimal design, which improves those responses that do not meet design requirements and meanwhile maintains response performances that have already met requirements. A successful application on a vehicle road NVH CAE system will be presented.