Empty Dynamic Optimization Models And Methods Of Configuration

Empty car distribution is a big issue that attracts more and more attention in the railway industry. In America, the percent of empty of total car kilometers is up to 45%. In our country, there are approximately 500,000 freight cars, and the average ratio of empty car kilometers to loaded car kilometers is about 30%, and the max ratio even perhaps 60%. Obviously, empty car distribution plays an important role in the efficiency of railcar utilization, the degree of meeting freight demand, and the modernization of railway operation. Because of the compositive influence of many factors, including management mode, network capacity, flow path, planned task, train connection and freight demand, as a common puzzle, empty car distribution tends to be so complex that it needs to be studied further in theory and practice.In recent 30 years, dynamic empty car distribution has been a popular topic in many developed countries. However, it hasn't received more attention in our country because of the characteristics of large-scale railway network, limited transport capacity, and planned economy. In practice, the scheme of dispatching empty cars changes from time to time following the fluctuation of freight demand and train connection. So, the scheme, laid out by static planning method and historical experience, is unscientific and unhelpful for proper allocation and maximum utilization of empty car resources.Based on researches on the problem at home and abroad, this dissertation constructs a series of optimization models of dynamic empty car distribution, which are close to our railway operating features and have a great value in application.There are five main aspects, which are studied in detail in the dissertation presented as follows:1. This dissertation designs a service time-space network combined with train formation plan and train schedule for the first time, which makes for modeling dynamic empty car distribution. In the range of railway administration (or company), empty service time-space network with discrete period and empty-and-load service time-space network with continue time are both designed. Through dividing time axis into several period segments, the first one not only reduces the size of network successfully, but also increases the flexibility for modeling. It's not the same that the second one describes each incident with detailed time, which realizes the exact optimization to train connection and train makeup. In the range of the whole railway network, OD time-space network with balanced transportation is created in order to optimize technical plan. The network has not relation to transportation of freight cars on the way, but refers to the situation of them in the origin and destination.2. This dissertation constructs two optimization models of dynamic empty car distribution for fulfilling daily and shift transport plan, which involve discharging empty cars, assigning empty cars and loading freights. One is the multi-stage optimization model of empty car distribution (MSMED), and the other is the model of optimizing empty and loaded car operation dynamically according to train schedule (MELOD). Because of MSMED neglecting the movement of loaded cars, it is approximate in forecast for empty car supply and evaluation of transport capacity. Moreover, some parameters in MSMED are so flexible that they can change with detailed situation of operation every day. Not only can MSMED make...