Stochastic Models For Container Shipping Revenue Management

Since the 1970s, containerized freight transportation has been the standard of the global maritime shipping. Because the container liner shipping industry has the typical characteristics for the application of RM, the research based on revenue management is an important way to strengthen the container liner transportation management. Revenue management(RM) is a method that makes the container transportation enterprises sell the right slot to the right customer at the right price at the right time in order to create the maximal selling revenue of the container slots. It can solve the problems the perishable container slots bring, and the enterprises can price and allocate the capacity better to increase the revenue and profit with it when facing stochastic demand and fixed supply .In this dissertation, overbooking, capacity allocation and pricing are analyzed and studied based on the pratical characteristics of container shipping industry.The main contents of this dissertation are as followed:In chapter 2, an overbooking model without empty container transportation involved under stochastic capacity is constructed and it is studied on how to decide the overbooking level of container position to minimize the expected total cost for discrete or continuous probability distribution of capacity. To solve the empty container transportation problem, an overbooking model with empty container transportation involved under stochastic capacity is constructed and it is proved that the optimal overbooking level declines when considering empty container transportation. The optimal overbooking levels with and without empty container transportation involved are computed when applying the proposed models and the difference between these two optimal levels is gained in numerical examples.In chapter 3, based on revenue management, a stochastic model for multi-liner and multi-segment capacity allocation is proposed. This model adequately considers the two-dimensional characteristic of container shipping, which includes container types with different size and double constraints of volume and weight of a liner ship. It is transformed into a linear integer programming model by robust optimization method. The method is given to get the scenarios and the probability that each scenario occurs. The model is then solved by genetic algorithms. In numerical examples, the optimal and satisfactory allocation solutions of different container types are gained and the research results show the validity of the model and algorithm. Finally, through numerical analysis and discussion, three beneficial characters and management strategies are drawn.In chapter 4, a stochastic model for dynamic capacity allocation with multiple container-types of container shipping two-dimensional revenue management is studied. The difficulty of computing the optimal strategy is analyzed in a special example with two container types. Then the bound of the value function of the model is proved and the two-dimensional problem is decomposed into two one-dimensional problems which are more practical in computing. Based on above, a characteristic problem with four container types is studied with the model established, the solution algorithm put forward and numerical examples given. The analysis result show that the expected marginal value of the twenty or forty feet container slot is not necessarily monotonic and the ratio of the maximal expected revenue of the heuristics to that of the optimal strategy is close to one.In chapter 5, the pricing and allocation model of container shipping based on revenue management is studied considering the loaded and empty containers transportation together. It is proved that the maximum expected revenue is nonincreasing in the loaded container flow imbalance factor. The models are analyzed and discussed by numerical simulation, and the optimized policies of pricing and slot allocation as well as a series of revelations are drawn.In chapter 6, a stochastic model for dynamic pricing with multiple segments and multiple container types of container shipping two-dimensional revenue management is brought forward. The bound of the value function of the model is proved and a more practical heuristic is presented based on the thought of reducing the dimensionality. It is illustrated that a set of critical decision values can be gained when there is one segment and one container type or there are two segments and one container type. The dynamic pricing strategies are analyzed when there is one segment and one container type, there are one segment and multiple container types and there are multiple segments and multiple container types in numerical examples. The analysis result show that the optimal price is monotonic in the first case, but is not necessarily monotonic in the second and third case.