Optimal Multiple Module Design - An Extended Simplex-like Method

Multiple Modular Design (MMD) is the method of designing a set of standard modules to meet demands for di erent functions. This work aims to nd an optimal set of modules given the demands. In this paper, we start with Evans' nonlinear programming model of MMD. By exploring the special structure of the formulation, we develop several properties of optimal solutions. With these properties, we develop a heuristic algorithm, called The Extended Simplex-like Method. In this algorithm, the objective value monotonically decreases during each iteration until a KKT point of the problem is found. Conceptually, a global optimal solution can be found by enumerating all KKT points with a special property. The example and computational experience show that the Extended Simplex-like Method can solve the problem to similar accuracy but with much less computational e ort than previous methods.