Models to Improve Efficiency of Disaster Relief Operation

This thesis focuses on relief distribution operations, which is one of the critical activities during disaster response. Immediately after a disaster, relief items such as water, blankets, tents are distributed from relief centers (RCs) for the affected victims. To understand challenges in relief distribution we conducted interviews with several relief agencies and identified design and setup of relief centers, management of large crowds, dynamically changing victim needs due to changing conditions, limited number of volunteers and convergence of donations as critical challenges. In this thesis we develop models that can provide decision support to efficiently address these problems.;We first focus our attention on analyzing relief center (RC) designs, to control the crowd and minimize the waiting time of victims that queue up to receive aid. We analyze current practice using a queuing model and propose a methodology to evaluate alternate designs. Next, we focus on modeling a network of relief centers and analyze the impact of victim mobility on relief distribution performance. We model a network of relief centers distributing aid as a generalized queuing network (G-network) and develop theoretical results and approximations that can quantify the effect of victim mobility on distribution efficiency. Next, we focus on quantifying material convergence, which refers to relief items, from many sources flowing into the affected region in very high volumes and over a short period of time. We propose a transient queuing model to quantify material convergence and derive methods to optimize resource allocation decisions. Next, we conduct simulation studies that can be used to analyze performance of RCs using a transient analysis and identify the optimal relief center designs.;This research provides contributions to both theory and practice. First, we propose queuing models to represent the inherent queuing effects that impact disaster relief operations. We develop solution methodologies to solve these queuing models and determine operational insights. We then apply the developed methodologies to 2015 Nepal earthquake data and provide insights on how to improve disaster relief distribution. We believe that this research will help emphasize critical factors that will improve relief distribution efficiency in the field and provide a basis for decision support.