Strategic Planning of Large-scale, Multimodal and Time-definite Networks for Overnight Express Delivery Services

The rising demand of express delivery service (EDS) and fierce market competition motivate EDS providers to improve service quality by modifying current networks. This project-based dissertation focuses on strategic planning of a large-scale, multi-modal and time-definite EDS network for a top nationwide EDS provider in China, based on its current network. An air-ground Hub-and-Spoke (H/S) network with a fully interconnected/star shaped structure was established to provide trans-city overnight EDS among relatively developed cities in China. The corresponding models are a combination of the hub location problem with fixed cost and the hub set covering problem. The objective function is to minimize the sum of the hub-location fixed cost and transportation cost under the constraints that all demand nodes are covered by their "home" hub. First, the basic model with linear air cost was proposed. Next, the basic model was extended to include air service selection decisions (or aircraft fleet owner-ship decisions) under the consideration of a cost select function for the backbone air service. Finally, two ex-tension models were studied, one to obtain the optimal aircraft fleet composition (Ext.1) and the other under the constraints of current aircraft fleet composition (Ext.2). Due to the large scale of project instances, hybrid genetic algorithms (GAs) were applied to get desirable solutions in an acceptable time period, but without the guarantee of finding optimal solutions. In particular, the overall problem includes three kinds of decisions: 1) hub location decisions, 2) demand allocation decisions and 3) air service selection decisions. A specific algorithm was proposed for each kind of decision, namely, GAs,local search heuristics and integer programming, respectively. These three algorithms were invoked hierarchically and iteratively to solve the original problem. 5 improvement techniques were proposed to different procedures of the original algorithms in order to improve the performance of the algorithms. Computational tests were conducted to evaluate the performance of the proposed algorithms in terms of computational time and solution quality. Tests under small-scale instances with CAB data sets were conducted to evaluate the overall performance of the proposed algorithm by comparing the solutions with the optimal solutions generated by CPLEX. Tests under large-scale instances with AP data sets and project data sets were conducted to evaluate the performance of the proposed improvement techniques. Since neither the optimal solutions nor solutions by other algorithms under large-scale instances were available to serve as benchmarks,the performance of the tailored algorithms and that of the un-tailored simple GAs was compared. Information about the stability of the algorithms with values of the coefficient of variation (CV) and the reliability of the results with T-tests was also provided. The models and the tailored GAs were applied to real-life instances of the project. This study introduces how the input data were collected and modified and how to deal with pertinent problems. By analyzing and com-paring the basic solutions of Ext.1 and Ext.2, the study not only reveals some important features of the net-work, but also arrives at some general conclusions and provided a dynamic aircraft fleet update strategy to guide the implementation of the project. Finally, scenario planning was executed to help decision-makers balance between costs and corresponding decision risks by identifying critical uncontrollable and controllable factors.