Designing an Optimization Model For Dynamic Facility-Location Problem at Post-Disarter area Considering Uncertainty

Indonesia is ranked 38th out of 181 countries most vulnerable to disasters. There were 2,2021 disasters in Indonesia with 2,491,572 people killed and 478,308 facilities damaged by June 2022. After the disaster, post-disaster emergency response management is needed in the form of distribution of logistical assistance. The most basic logistical problem is the problem of facility location. Constraints in the problem of the location of facilities after a disaster occurs, namely the uncertain nature of the disaster, uncertain parameters can affect the performance of humanitarian logistics planning. Parameters under uncertainty environment are demand, facility capacity, cost, transportation time and vehicle. Logistics is also the most expensive activity of any humanitarian aid. Based on the study, it is estimated that the logistics costs for disaster management are around 80% of the total costs in humanitarian assistance. Therefore, a model for optimizing facility location problems in disaster logistics management is needed. The facility location problem model includes determining the dynamic location of emergency facilities and distribution routes for relief items from the facility location to the point of request for disaster victims. This model has two objective functions, namely minimizing the cost and time in the distribution of relief goods to disaster victims. The design of this model can solve the opening of emergency facility locations by using a clustering algorithm in machine learning with a minimum distance objective function, the formation of distribution routes using deep learning algorithms and optimal routes with mixed integer non linear programming to produce two objective functions that minimize the total system cost and time. The results of the design model for dynamic facility location problems in post-disaster areas in uncertainty resulted in reduced costs and time of distribution that had been carried out by the National Disaster Management Agency and had faster computational times in generating route optimization and opening distribution centers.