Facility layout problem under uncertainty

The facility layout problem has usually been treated as a deterministic problem and uncertainty regarding the problem parameters has seldom been addressed. In this study, the aim is to investigate di erent ways of dealing with uncertainty to design a facility layout which attains robust and e cient performance under all possible scenarios. For this purpose, seven mathematical models based on the Quadratic Assignment Problem (QAP) formulation have been developed. These formulations cover alternative methodologies existing in stochastic and robust optimization literature such as minimizing maximum cost, expected cost, maximum regret and p-robustness as well as new approaches that combine them in di erent ways. The proposed models are solved using Genetic Algorithms (GA) incorporating operators and local improvement schemes that are specially selected and adapted for the models. As two of the models involve multiple objective functions, a Multi-Objective Evolutionary Algorithm (MOEA) has also been developed. Extensive numerical analysis enables us to compare the performance of these approaches in terms of robustness metrics and to gain important insights into ways of treating the uncertainty issue in facility layout problem.