Assignment allocation and simulated annealing algorithms for cell formation

Abstract

In this paper a nonlinear mathematical programming model is developed for cell formation that identifies part families and machine groups simultaneously with no manual intervention or subjective judgement. The objective of the model is minimization of the weighted sum of the voids and the exceptional elements. Changing weights for void and exceptional elements aids the designer with a systematic generation of different solutions, i.e., forming large loose cells or small tight cells. An assignment allocation algorithm (AAA) and a simulated annealing algorithm (SAA) are developed to solve the model. AAA and SAA compare favorably with many well-known procedures for the problems tested. AAA is less computer-intensive and hence large problems with 400 parts and 240 machines were solved with AAA in less than a minute on Sun Sparc station. However, AAA is sensitive to the initial machine grouping solution input to the algorithm. SAA gives consistent results but requires more computational time.