|Abstract or Summary
- For manufacturing and service industries to stay competitive in this rapidly changing, globalized world, one of the most important operations, and one that many businesses struggle to plan and manage efficiently, is material handling. There has been a large amount of research on various vehicle routing problems (VRPs) in recent decades, but relatively less work on certain unique types of VRPs that characterize some specific industries.
This thesis identifies an interesting and new type of VRP with unique characteristics in order to serve the needs and goals of the electronics industry. The problem deals with two types of customers; some require only delivery, while the others require both delivery and installation. There are two different types of vehicles in this problem: delivery vehicles and installation vehicles. The delivery vehicle and (if needed) the installation vehicle are allowed to visit each customer only once. There is an additional constraint to provide the guaranteed service quality, which is measured by the amount of time between the delivery and the installation. A customer must be visited by an installation vehicle within the predetermined maximum allowable time after the visit by a delivery vehicle. Therefore, it is required that both types of vehicles be synchronized for the customers requiring both delivery and installation. The problem under consideration is more complicated than other, traditional VRPs that have been studied widely in the past, since in this case two different types of vehicles must be synchronized.
A mixed-integer nonlinear programming (MINP) model for this problem is formulated. A hierarchical approach using a genetic algorithm (GA) is proposed to solve the problem effectively. Various examples are tested to show the effectiveness of the proposed hierarchical approach. To demonstrate the robustness of the proposed approach, partial factorial experimental design varying the parameters of characteristics in the problem is performed using the Taguchi method. In the hierarchical approach, an algorithmic limitation is conjectured in which the attempt to find the best solution tends to dwell on the local optimal solution, searching only part of the entire solution space. In order to tackle the limitation of the hierarchical approach and search the entire solution space effectively and efficiently for a global optimal solution, an endosymbiotic evolutionary algorithm (EEA), which concurrently considers subproblems having cooperative interactions, is considered. Various test examples are solved using the EEA, and this method’s efficiency and effectiveness are shown by comparing the computational results with the ones from the hierarchical approach. A set of problems is solved by the MINP model, the hierarchical approach using a genetic algorithm, and the EEA, and solutions from the three approaches are compared.