Reverse logistics network design with uncertainty

Over the past few years, recovery of used products has become increasingly important due to economic reasons and growing environmental or legislative concern. Meanwhile, an efficient reverse logistics network is required to process used products returns so as to recover value by reprocessing them and redistributing them in the market. Uncertainty is one of the main challenges face the planning and designing processes of reverse logistics network, the high degree of uncertainty in terms of time, quantity and quality of returned products, and the capacities of different facilities inside the network increase the complexity of reverse network designing problems.With consideration of the factors noted above, this thesis proposes a comprehensive model for reverse logistics planning where many real-world features are considered such as the existence of multi objectives, multi echelons, and multiple commodities reverse logistics network system under uncertainty in terms of product return quantity. In this research, bi-objective two-stage stochastic mixed-integer linear programming model is proposed for designing multi-echelons, multi-commodities, single period reverse logistic network. The proposed bi-objective model includes: (1) minimizing total costs including the sum of fixed, transportation, relocation, collection, inspection, recovery, and disposal costs; (2) minimizing of CO2 emissions from transportation modes inside the network. The problem was formulated in two stage stochastic model in order to find the set of optimal network configurations which achieve the mains goals of the research. The proposed model takes into consideration the quantity of returned product uncertainty, where the inherent risk is modeled by scenarios. To find the set of non-dominated solutions, ε-constraint method is used to obtain a list of Pareto-optimal solutions for the proposed model. Key words: Reverse Logistic Network Design, Two-Stage Stochastic Programming, Multi-Objective Optimization, Augmented Ε-Constraint Method