Hybrid meta-heuristic approaches for single and multi-objective buffer allocation problems in manufacturing systems

The buffer allocation problem is an NP-hard combinatorial optimization problem involving the determination of the number of buffers in buffer locations required to increase the efficiency of a production line. Researchers in this field have proposed various optimization techniques to solve the problem for different types of production system configurations. The main purpose of this Ph.D study is to introduce efficient and robust hybrid solution approaches for both single and multi-objective buffer allocation problems in serial production lines. In order to solve single objective buffer allocation problem, a hybrid algorithm combining Genetic Algorithm with Simulated Annealing Algorithm -based simulation optimization approach is proposed to allocate of a certain amount of buffers among the buffer areas of a production line so as to maximize the production rate of the system. This approach involves the use of a generative tool and an evaluative tool. The hybrid algorithm is employed as a generative tool to create candidate buffer size configurations. As a performance evaluative tool, discrete event simulation modeling is used to obtain the average production rate of the line. Prior to testing the performance of the proposed hybrid approach, an experimental design study is conducted to identify the best values/scheme for the hybrid algorithm parameters. Moreover, using these best hybrid algorithm parameters, a comprehensive experimental study along with statistical analysis is carried out to investigate the power of the hybridization for various serial line configurations. In the second stage of this Ph.D thesis, multi-objective buffer allocation problem which has two conflicting objectives, i.e. production rate maximization and total buffer size minimization is considered. For this purpose, the generative method is based on an evolutionary algorithm in which Elitist Non-dominated Sorting Genetic Algorithm and a special version of a Multi-objective Simulated Annealing Algorithm are hybridizied. Following the experimental study to identify appropriate values for the hybrid algorithm parameters, a comparative study is carried out to present effectiveness of the proposed hybrid approach on solving buffer allocation problems for various serial line configurations.