A robust simulation-optimization approach for pre-disaster multi-period location–allocation–inventory planning

Abstract

Natural disasters such as earthquakes always threaten human societies. Therefore, preparedness planning seems to be necessary to reduce casualties and accelerate relief efforts. Decisions in the preparedness phase include determining the optimal location for distribution centers and suppliers, allocating them to affected areas, and determining the amount of inventory of distribution centers. In this research, a robust simulation-optimization approach is presented for the planning in the preparedness phase. Although the demand for relief commodities has been considered as a given parameters. Due to our best knowledge there is no research in which the demand for relief commodities including drinking water, food, blood, medicine, tents and non-drinking water, during the disaster is related to the interactive relations of critical infrastructures of the city affected by the earthquake, and the power of the earthquake. Most of the unplanned occasions are occurred when co-incidences of probabilities are happen together. When a critical urban infrastructure fails due to earthquake it may cause some difficulties in the other infrastructures. Co-incidences may cause some horrible tragedy. First, the critical infrastructures of the city affected by the earthquake are identified and the interactions of these infrastructures are simulated. Considering the interactions of critical infrastructures in a city, various scenarios for earthquake are designed and tested through simulation approach in order to estimate the mean and variance of demand for relief commodities. The stochastic demand behavior which is the output of the simulation approach is assumed as a stochastic parameter of a mathematical model for multi-period location–allocation–Inventory problem. Robust optimization approach is used to deal with uncertainty. The mathematical model determines the location of distribution centers and suppliers, and how they are allocated to the affected areas. The proposed mathematical model is solved using a customized genetic algorithm for a case study in Tehran.