Genetic algorithms for buffer size and work stations capacity in serial-parallel production lines

Recently, many production lines that have complicated structures such as parallel, reworks, feed-forward, etc. are widely used in high volume industries. Among them, the serial-parallel production line (S-PPL) is one of the more common production styles in many modern industries. One of the methods used for studying the S-PPL design is through genetic algorithms (GA). One of the important jobs to use GA is how to express a chromosome. In this paper, we attempt to find the nearest optimal design of an S-PPL that will maximize production efficiency by optimizing the following 3 decision variables: buffer size between each pair of work stations, machine numbers in each of the work stations; and, machine types. In order to do this we present a new GA-simulation based method to find the nearest optimal design for our proposed S-PPL. For efficient use of GA, our GA methodology is based on a technique that is called gene family arrangement method (GFAM) which arranges the genes inside individuals. An application example shows that after a number of operations based on the proposed simulator, the nearest optimal design of S-PPL can be found.