Simulation run allocation for optimization of order-up-to level configuration in a complex logistics

A complex logistics (CL) is often consisted of a number of entities such as manufacturers, suppliers, wholesalers and retailers. Under periodic review policy, simulation has become an efficient tool to optimize the order-up-to level of each entity in a CL where the customer demand of retailers is fuzzy. However, the number of order-up-to level design or configuration in a CL is often very large, which brings very high total simulation cost. Therefore, it is necessary to allocate simulation run time for each design to reduce the total simulation time and improve simulation efficiency. Based on the existed CL simulation optimization tool (CLCSim), a new simulation optimization architecture considering simulation run allocation (CLCSim-SRA) is proposed by comparing the main simulation run allocation methods. Application of CLSim-SRA is proved by an illustrative example. Effects of the two simulation run allocation methods, Equal Allocation method (EA) and Proportional To Variance method (PTV), on CLCSim-SRA are discussed and illustrated by several experiments. After that, several conclusions are given.