Combining balancing, sequencing and buffer allocation decisions to improve the efficiency of mixed-model asynchronous assembly lines

When the production of a variety of models is considered in assembly lines, the more suitable control system is asynchronous: the pace of the line is not imposed but depends on the actual time needed in each work centre to complete a product. In asynchronous lines blocking and starvation phenomena are responsible of performances deterioration. Three popular techniques able to limit these phenomena are: balancing the workload among stations by tasks assignment, allocating buffers between stations, optimizing of the sequence of models entering the line. These three techniques correspond to the Mixed-model Assembly Line Balancing Problem (MALBP), the buffer allocation problem (BAP) and the sequencing problem (SP). The presented approach is able to simultaneously solve these three problems, which are strictly connected, and is based on a Genetic Algorithm procedure that uses discrete event simulation to evaluate the fitness function of individuals. Tests conducted on a set of benchmark instances allow quantifying the effect on line efficiency of sequencing and buffer allocation decisions separately, with the aim to assess which of the two techniques is more effective and to investigate whether if combining sequencing optimization and buffer allocation gives additional advantages in terms of increasing line efficiency.