Abstract
MMFLA model is the abbreviation of Multiechelon Multiactivity Facility Location Allocation model. This model was proposed in 1984 by Mahmoud, Mohamed Moustafa of University of Pennsylvania for solving plant location problems. This model is used for two echelons (input source ↔ plant ↔ market) where we input raw materials, manufacture products in multiplex production lines and ship them to markets. MMFLA model is defined as the objective function shown in (1) and relaxed and subdivided problems are solved by an approximation algorithm. However, since exponentially larger amount of computation for MMFLA model is needed as the number of variable is increased, which we call NP completed problem, and the algorithm for solving by MMFLA model is too complex, its application to real world problem is limited to special cases : the number of plant, market, input source, product and input are 6, 5, 5, 4 and 4, respectively. And it is necessarily desirable that a new algorithm much more sophisticated and extended MMFLA, for reducing computation time as well as expanding variables is urgently needed to be developed for the application to real cases. The purpose of this paper is to propose a new algorithm for MMFLA model based on Genetic Algorithm (GA) and to show our proposed algorithm applicable to more practical problems by extending the models size and reducing execution time of computation.
