A Kriging-Based Approach for Simulation of Coupled Multidisciplinary Systems

System analysis of coupled multidisciplinary systems is a key technique of MDO (Multidisciplinary Design Optimization). This work is often computationally expensive since it requires a number of discipline analyses which themselves are also costly. To abate the computation efforts, this paper present a method which evaluate the system level responses by using the discipline level Kriging models. In order to ensure the inter-disciplinary compatibility, the system analysis problem is converted into an optimization problem according to the idea of SAND (Simultaneously Analysis and Design). An infill sampling criterion is derived by considering both the prediction values as well as the prediction errors of the Kriging models. This criterion provided a trade-off between locating the optimum of the associated optimization problem and improving the Kriging models globally. The computational cost was reduced effectively. An electronic packaging problem was taken as a validation example for the proposed method.