Genetic algorithm approach with pareto optimal solution for multi-objective facility layout problem

Abstract

Facility layout is one of the most important problems for modem manufacturing systems. Facility layout plays a key rolc for companies, and it is an inseparable part of the manufacturing system design process. A good solution for the facility layout problem contribute, to the overall efficiency of operatIons. Traditionally there are two approaches to the facility layout problem. One Is the quantitative approach aiming at minimIzing tbe total materIal handlIng cost and the other one is qualitative approach aiming at maximizing closeness rating score. Various researcbes have been performed on these two approaches separately. In lhis paper both approaches to tbe facility layout problem havc been taken into consideration separately. Subsequently, the research also solved the problem combining tbese two approaches, at the objective function level. In this research work first the genetic algorithm (GA) was developed for the multi-objcctive facility layout problem and found out tbe optimal tacility location for a particular problem considering the two objectiveS; i.e. minimIzing the material handling cost and maximizing the closeness raling score. Then another local search technique, Simulated Annealing (SA) algorithm, was developed to compare the result found in GA. In GA primarily an initial population was created and by tbe crossover operator and mulation process new offspring was generated and if the offspring meet the stopping criteria the result was seleeted for the process. In simulated annealing firslly an initial temperature was selected and then going through the SA process and meeting the stopping criteria the final result was generated. By the two proposed approaches a non domInated soiuti()Il set is found (Pareto optimal) approximately for the multi objective facility layout problem. A Problem of 3x3 matrix was te,ted using both genetic algoritbm and simulated annealing approach. By changing different parameter" of genetic algorithm and simulated annealing the caSe study was examined. From the simulation result it is found that SImulated Annealing always gives betlcr rc~ult lhan the (,enetic Algorithm. In majority of the cases Simulated Annealing lind the department sequence that incurs minimum material handling cost and that d~partment sequence als" maximizes the closeness rating score than the department sequence of Genetic Algorithm. And finally it was found that the SA requires Ie,s computational time tban the GA. So it can be stated that tor finding tbe optimum facility layout the simulated annealing is the best algoritbm for this particular lype of problem.