Methods and Analysis for Recovery Logistics Networks with Uncertainty and Channel Selection Considerations
MetadataShow full item record
In this dissertation, we develop models and methodologies for effective design and efficient operation of product recovery logistics networks. Recovery networks, employed for recycle-reuse-refurbish-remanufacture purposes, constitute an ever-expanding portion of supply chain networks. For such activities to make business-sense, it is important that the logistical decisions associated with designing and operating underlying networks are made carefully. With this main motivation, we focus on two fundamental problems. First, we consider a generic Closed-Loop Supply Chain (CLSC) network setting under demand and return uncertainty and provide a new model and an efficient solution approach for the associated network design problem. Consideration of uncertainties and their impact on the CLSC network design is a largely ignored area in the literature, thus, this work contributes to closing this gap, in both modeling and solution methodology contexts, as well as in analysis. Second, we consider the specific case of commercial returns, which is quite common in today's business climate, given the generous return policies provided by electronics and department stores as well as retail superstores. In this setting, for operational efficiency and financial effectiveness, it is important for providers to best determine appropriate return channels, i.e., the return channel selection, for commercial products whose values decrease over time. Return channel selection for commercial products is also a largely ignored area in the literature. We first address this problem from an operational efficiency perspective given an underlying network of facilities. In the related models and analysis, we introduce and capture the concepts of channel selection dependence on product and logistics network characteristics. Later, recognizing that the design of an underlying network may be under the control of the provider, we take an integrated design and operation perspective and incorporate the logistics network design into the model to further study dependence of channel selection on network characteristics. In addition to new models and analysis for commercial return logistics, our contributions also include the development of efficient solution algorithms with measurable solution quality. We introduce the problems of interest and their context in today's business environment in the first chapter. In the second chapter of the dissertation, we develop a two-stage stochastic programming model for the generic CLSC network design problem under demand and return uncertainty, represented by a set of scenarios. For the model's solution, we develop a Benders Decomposition (BD) approach that significantly improves computational efficiency via surrogate constraints, strengthened Benders cuts, multiple Benders cuts, and mean value scenario based lower bounding inequalities. In the third chapter, we develop models for the channel selection problem for commercial products under time-value consideration. Based on this model, we analyze the optimal return channel selection strategies under varying underlying logistics network and product characteristics. For this purpose, we utilize real geographical data from the U.S. and product data for Hewlett Packard and Bosch. In the fourth chapter of this dissertation, we develop a Mixed Integer Linear Programming (MILP) model for integrated design and channel selection for commercial product returns under product time-value consideration. For the model's solution, we develop an efficient algorithm based on the Simulated Annealing (SA) approach, benchmarking the quality of solutions against the upper bound obtained by a Benders Decomposition approach. Using this model and the solution approach, we provide an extensive analysis of the relationship between recovery logistics network structure and product characteristics.
Hwang, Sung Ook (2014). Methods and Analysis for Recovery Logistics Networks with Uncertainty and Channel Selection Considerations. Doctoral dissertation, Texas A & M University. Available electronically from