No abstract

This paper presents the motivation development and an application of a unique methodology to solve Industrial Optimization problems, using existing simulation software programs. In order to make this a reality, (1) a unique way of communicating with simulation programs, (2) parallel execution of multiple Designs dictated by an extensive library of Design of Experiment plans - sampling the Design Space with the minimum amount of function evaluations while capturing the Design behavior - (3) a Stochastic Interpolation Response Surface Modeling technique and (4) Non-Linear Programming techniques are the essential elements of this methodology. Finally an application from the automotive industry demonstrate the methodology.

No abstract

A study of the condition number of matrices appearing in the resolution of mixed boundary value potential problems through the boundary element method (BEM) is presented here. It is shown how this condition number varies with the size of the domain under consideration. An algorithm is provided giving the best scale factor, which is the one leading to the smallest possible condition number.