Application of multiobjective optimization in selection of printed wiring boards

This paper presents an application of multiobjective optimization in optimal selection of number of signal layers in printed wiring board (PWB). The criteria considered in optimization are: cost, electrical performance, and reliability. The reliability of system is influenced by the operating temperature, which is determined by the power dissipated in parts placed on the board, ambient temperature, and cooling technology. In evaluating cooling and calculation of junction temperature both heat conduction through the board and forced air cooling are considered. Designers have to deal with multiple, conflicting criteria because of variety of requirements that packaging should meet. However, sophisticated methods developed for multiobjective optimization are not yet used in packaging design. This paper describes some of the methods used in solving multicriteria optimization and gives an example of application to design of PWB. The weighting method is applied to find the optimal solution first. The graphical form of optimality of different weights for different criteria is presented. The distance based methods, which include l/sub 1/-distance, l/sub 2/-distance, l/sub (infinity)/-distance, and the geometric distance method, are also applied to determine the optimal number of signal layers. The results from different methods are compared and discussed.