An integrated cell formation and facility layout algorithm to minimize total material handling costs

Most flexible manufacturing models in the literature are dedicated to solving cell formation (CF) and facility layout (FL) separately. CF focuses on grouping machines and parts into a cell by minimizing the bottleneck parts (moves) assuming fixed traveling distances from one cell to another. FL is arranged to improve the total cost, particularly material handling costs. The purpose of this dissertation is to develop an integrated procedure for solving both CF and FL simultaneously. The purposed algorithm consists of two phases. Phase I solves a linear integer program to obtain machine-part grouping given the unit cost of material flow based on a trial facility layout. In turn, the algorithm used to solve the problem in Phase I is an extension of a decomposition procedure by Chankong (15) and test by a few researches (3, 52, 71). By solving the part-grouping subproblem analytically in terms of the variables in the machine-grouping subproblem, the combined machine-part grouping can be solved as a minimin linearly constrained 0-1 program involving only the machine grouping variables. An efficient algorithm is devised to solve this minimin problem. The result yields machine cells and part families that minimize material handling costs by minimizing intercell part flows. From this, a minimum intercell flow matrix is formed for a given facility layout.;Phase II is an integer Quadratic Assignment Problem (QAP) program that is used to find an optimal assignment of machine cells to locations given a set of machine cells and part families. The layout considered here is the linear layout and the objective is to adjust the "flow distance" by finding an appropriate sequence of machine-cell placement in a single row to minimize the overall material flow cost. The result yields unit cost of flow from one cell to another that is based on the given configuration of machine-part cell formation. This result can be fed back to the problem in Phase I and the whole processed is repeated until no further improvement can be made. The algorithm in Phase II, Algorithm II, uses a minimum incremental cost analysis based on dynamic programming (DP) to solve an optimal layout along a straight line. The performance of this algorithm is compared with two standard algorithms: a 2-iteration pairwise exchange method and a greedy heuristic algorithm.;In this dissertation, tests are performed on Phase I and Phase II separately. The algorithms are then integrated to solve the combined CF and FL problem. A systematic procedure is developed to change the traveling distances from Algorithm II for the initial solution in Phase I and to update the volume of flows between cells from Algorithm I for the optimal layout in Phase II. Test results of the overall solution procedure are quite encouraging.