Baharmand et al. A Framework for Shelter Location Decisions by Ant Colony Optimization Short Paper – Decision Support System Proceedings of the ISCRAM 2015 Conference - Kristiansand, May 24-27 Palen, Büscher, Comes & Hughes, eds. A Framework for Shelter Location Decisions by Ant Colony Optimization Hossein Baharmand University of Agder, Norway hossein.baharmand@uia.no Tina Comes University of Agder, Norway tina.comes@uia.no ABSTRACT Earthquakes frequently destroy the homes and livelihoods of thousands. One of the most important concerns after an earthquake is to find a safe shelter for the affected people. Because of large numbers of potential locations, the multitude of constraints (e.g. access to infrastructures; security); and the uncertainty prevailing (e.g., number of places required) the identification of optimal shelter locations is a complex problem. Nevertheless, rapidly locating shelters and transferring the affected people to the nearest shelters are high priority in crisis situations. In this paper, we develop a framework based on Ant Colony Optimization (ACO) to support decisions-makers in the response phase. Using the same framework, we also derive recommendations for urban planning in the preparedness phase. We demonstrate our method with a case focusing on the city of Kerman, in Iran. Keywords Crisis Management, Shelter Planning, Location Decision, Ant Colony Optimization. INTRODUCTION A quick look at recent earthquakes over the last decade depicts that most of them have occurred in Asian countries leaving massive casualties (Hamada, 2014). According to (UNHABITAT) 1 , shelter is one of the priority needs in humanitarian crises worldwide. Locations of shelters or camps will determine the logistics processes, and can have a major influence on the longer term economic and social stability (Rashed & Weeks, 2003). Locating these safe places before the crisis can play a significant role in improving the crisis management programs as well as reducing injuries (Givechi, Attar, Rashidi, & Nasbi, 2013). However, because of the uncertainty in predicting earthquakes and their consequences, such as the behavior of the population, finding an optimal combination of shelters before the disaster is a complex problem (Anping, 2010). BACKGROUND The well-known crisis management cycle includes four stages: prevention, preparedness, response, and recovery. Because of the time pressure and limited capacity in the aftermath of a disaster, we propose to identify potential shelter locations in the Preparedness stage. During the response, information about the locations needs to be updated (e.g., safety, accessibility), and the evacuation to the operational shelters needs to be executed. 1 http://mirror.unhabitat.org/content.asp?typeid=19&catid=286&cid=868