Model Research Of Interval Planning For Liner Shipping Hub And Spoke Network Design

The container transportation demand between different ports is an important coefficient that should be considered while design hub-spoke network. Optimal network changes with demand. The fact that once the optimal structure of shipping network is settled, it is difficult to change, and the network optimization problem based on deterministic model cannot adjust according to market changes, which brings risks to decision-makers. And this is why it is necessary to take uncertainty into consideration while optimize hub-spoke network.Stochastic programming, fuzzy programming and robust optimization based on scenarios are commonly used to deal with uncertain linear planning problems, and among which, it is necessary for stochastic programming to get probability distribution, and fuzzy programming affiliation function, robust optimization the probability distribution of scenarios. However, the complicity of port makes it difficult to capture distribution of uncertain coefficients, so the above methods have limitations.Under such situation, this thesis tries to introduce interval to describe container transportation demand to minimize total cost including transit cost and transshipment cost. A mixed integer linear programming model is constructed to deal with hub port location, spoke port allocation and main lines design problems meantime. Risk factor is introduced to transform subject with interval into deterministic function.The model here is NP-hard, and its complicity determined the difficulty of solving it. In this thesis, the combined genetic algorithm with AC is used in this thesis, where GA is implemented to decide hub location, Shortest Path is used to allocate spoke ports and AC is used to optimize mine lines.The algorithm is based on GA, and its effectiveness and feasibility are proved by a case. Which means interval programming can enable decision makers to know risks and consequences through taking uncertainty of data into consideration, and it can reflect information about container demand through getting uncertain information involved.