Dissertation Title: Commodity Shipment Planning in Incomplete Hub Location-Routing Problem
Abstract
Transportation systems for commodity, mail and passenger shipment use hub network due to its economy of scale. Hubs are facilities provided three main functions: collecting, sorting and distribution.
This dissertation considered the incomplete hub network in order to find the optimal location of the hubs, flow shipments through routes and the departure time between the origins and destinations. The main purpose of this research is to minimize the transportation costs, the fixed costs of locating hubs and the costs of establishing hub to hub, non-hub to non-hub and hub to non-hub links. In order to reach this purpose, two incomplete hub location-routing mathematical models have been suggested. According to their variable types and tightening of models, one of them has been selected for the flow shipment planning problem. Furthermore, the uncertainty of the transportation time and volume of flows has been considered as stochastic environment. In addition, another objective function is added to the model in order to increase the volume of flow sent on its collection date. All mathematical models have been tightened by the valid inequalities and preprocessing to reach the optimal solution quickly. Furthermore, for bi-objective models an e-constraint-based approach has been used.
Computational studies on well-known data sets, besides the study of the interstate ground mail network have been preformed to evaluate the models. The costs of this mail network have been reduced by 5.53% when suggested solution has been used. Results show the network tends to employ fewer hubs when the discount factor has been increased. Moreover, this increasing leads more transportation and less fixed cost. However, generally, total costs have been decreased in this case. By increasing in service level, the number of hub to non-hub links has been decreased. Also, this change causes less total cost. Besides, in the stochastic model, the number of Pareto solutions is more than the deterministic model.
High performance of the proposed valid inequalities and preprocessing is another important outcome of this research. The prformance has been investigated on CAB and AP datasets. The computational time has been averagely decreased by 49.46% for the first model of the hub location-routing in an incomplete network by using these tools. Moreover, for the second model of the hub location-routing in an incomplete network, the computational time has been improved averagely 17.93%. The computational time has been reduced by around 65.90%, when preprocessing and valid inequalities have been used for mathematical model of commodity shipment planning in an incomplete hub location-routing problem.
Keywords: Hub, Incomplete network; Location; Routing; Departure time; Valid inequality.
Defense Date: 2015-July-28
