Optimization of Block Layout for Grocery Stores

Abstract

Because of the proportion of the retail industry worldwide of the global economy and the lack of analytic approaches in the literature to retail facility design, this dissertation addresses block layout in grocery stores with the participation of Migros, a major Turkish retailer. The goal is to develop an effective method for solving realistic grocery store block layout problems which consider revenue and adjacency. The main difference between previous research and this proposed research is the integration of stochastic simulation and optimization together to solve the layout problems modeled specifically for grocery stores. Actual data and insights from Migros are also incorporated into the methodology. This project is divided into two main parts. The first part is the revenue assessment of store layout using stochastic simulation considering impulse purchase rates and customer traffic patterns. Using market basket data from Migros a methodology is developed to identify groups of departments where the products are often purchased concurrently. This, in turn, is used to estimate the impulse purchases of customers. In the second part, optimization is used by considering limited space requirements, unit revenue production and department adjacencies. As an optimization tool, because of the strong neighborhood structure and complexity of the problem, a tabu search algorithm is used and compared with two simple constructive heuristics. The candidate layouts generated in this step are evaluated by discrete event stochastic simulation. As we have multiple objectives to maximize, a bi-objective model is formulated for the store with the concurrent objectives of revenue maximization and adjacency satisfaction. A set of non-dominated designs are generated by the tabu search for a decision maker to consider further. This approach is both effective and pragmatic for optimal design of grocery store block layouts.