Abstract

Many manufacturers are moving towards green manufacturing. One of the actions for environment friendly manufacturing is collection of end-of-life products (EOL). EOL products are transported to the proper facilities for reprocessing or proper disposal. Movement of collected products is performed through reverse logistics networks. Reverse logistics networks may be designed independent of forward logistics networks, or as integrated networks, known as integrated forward and reverse logistics (IFRL) networks. Recent research shows that IFRL networks are more efficient than independent networks. In this work, we study a number of IFRL networks. We present a comprehensive mathematical model to represent an assignment and location-routing IFRL network. Afterwards, this model is decomposed into a number of sub-models that represent different IFRL networks. For each network we develop a solution methodology to solve practical size problems.
Two sub-models based on the comprehensive model are presented to design two IFRL location-routing networks. The first network considers decision on the location to establish a disassembly plant. The second network considers decisions on the location to establish a manufacturing facility. For both networks, routing decisions are assigning customers to vehicles, and establishing vehicles’ routes. We develop two heuristic methods to solve the models. The heuristics are able to reach optimal or near optimal solutions in reasonable computational times.
The vehicle routing problem with simultaneous pickup and delivery and time windows (VRPSPD-TW) is studied in this work. We use a sub-model of the comprehensive model to represent the problem. Classic heuristics and intelligent optimization or metaheuristics are widely used to solve similar problems. Therefore, we develop a heuristic method to solve the VRPSPD-TW. Results of the heuristic serve as initial solutions for a simulated annealing (SA) approach. For most tested problems, the SA approach is able to improve the heuristic solutions, and reach optimal solutions. Computational times are reasonable for the heuristic and SA.
We also study the multi-depot vehicle routing problem with simultaneous pickup and delivery and time windows (MDVRPSPS-TW). A sub-model of the comprehensive model represents the problem. The network considers assignment of customers and vehicles to depots, assignment of customers to vehicles and routing of vehicles within customers’ time windows. We develop a 2-phase heuristic and a SA approach to solve the problem. Heuristic solutions serve as initial solutions for the SA approach. SA is able to reach optimum or near optimum solutions. Computational times are reasonable for the heuristic and SA