A memetic algorithm for location-routing problem with time windows for the attention of seismic disasters: a case study from Bucaramanga, Colombia

Abstract

Introduction: In recent years, a great part of the population has been affected by natural and man-caused disasters. Hence, evacuation planning has an important role in the reduction of the number of victims during a natural disaster. Objective: In order to contribute to current studies of operations research in disaster management, this paper addresses evacuation planning of urban areas by using buses to pick up affected people after an earthquake. Methodology: The situation is modeled using Location-Routing Problem with Time Windows (LRPTW) to locate emergency shelters and identify evacuation routes that meet attention time constraints. To solve the LRPTW problem, a memetic algorithm (MA) is designed to minimize the total response time during an evacuation. The algorithm is not only validated using instances of literature but also with the assessment of a case study of a seismic event in Bucaramanga, Colombia. Results and conclusions: The main contribution of this article is the development of a memetic algorithm for the solution of the proposed model that allows to solve real-size instances. The hybrid initialization of the MA prevents an early convergence by combining randomness and a heuristic technique. Computational results indicate that the MA is a viable approach for the LRPTW solution. Likewise, a case study is presented for the city of Bucaramanga in order to validate the proposed model. Two scenarios are simulated showing that the management of the time windows (homogeneous or random) directly influences the solution and affects the objective function. From a practical perspective, the location-routing problem must consider other criteria such as the cost of evacuation, including the attention delay cost, and the cost of opening shelters and routing.