Development of an Operational Framework for Cellular Manufacturing System Using Hybrid Genetic Algorithm

Abstract

Cellular Manufacturing System (CMS) is the applied form of Group Technology (GT) in the manufacturing sector. CMS has the flexibility of job shop and efficiency of flow shop, besides having reduced material handling cost, reduced work in-process inventory and reduced setup time, thus brings about major decrease in the manufacturing cost and increase in quality of the final product. Though, having many benefits, it is a challenging task to get the benefits associated with CMS for the complex, real world problems. Most of the researchers have focused on the cell formation component of the CMS, which comprise of grouping parts into families and clustering machines (required to process these parts families) into cells. However, the operations related information such as processing time and sequence are rarely considered. The use of optimum manufacturing resources can be materialized through effective and efficient scheduling system. Therefore, current research focuses on the cell formation problem and the cell scheduling was also executed to complete the operational aspects of CMS and make it more useful for shop floor managers. All the benefits associated with CMS can be achieved through solving cell formation problem along with cells scheduling. In order to develop a comprehensive operational framework for CMS, two sub-frameworks were developed in this research study. First sub-framework was developed for solution of Cell Formation Problem (CFP) through a novel hybrid Genetic Algorithm (hybridization of Genetic Algorithm with Simulated Annealing and employing shake-off). The second sub-framework, which is a hybrid Genetic Algorithm-similar to CFP, was developed for solution of cells scheduling problem (solved as job shop scheduling problem as each cell is considered like a job shop). The effectiveness of each sub-framework was separately validated through testing of a number of problems from literature. The testing results of 35 CFP problems produced 24 best results with overall mean of 66.20. The proposed approach not only achieved the best solutions consistently, with minimum computational time, but also improved the results of two problems (problems 29 and 33), reporting it for the first time. The testing of 38 job shop scheduling problems produced best optimal results for all tested problems (i.e. matched the best results reported in literature so far).