Metamodel-Assisted Simulation-Based Optimization of a Real-World Manufacturing Problem

Many real-world manufacturing problems are too complex to be modeled analytically and in these settings simulation-based optimization is a highly valuable tool. It has shown to be a useful technique for systems improvement and has been applied to address a wide range of industrial optimization problems. Simulation runs are, however, often computationally expensive and to enhance the efficiency of the optimization process the incorporation of computationally efficient metamodels has been suggested, so called metamodel-assisted simulation-based optimization. This paper presents metamodel-assisted simulation-based optimization applied to a multi-objective real-world manufacturing system. The problem is about finding optimal buffer levels in a complex production line in order to minimize product lead times and maximize system throughput. A steady-state evolutionary algorithm is used for the optimization.