Research On Optimizing Planning Models And Algorithms Of Car Rental Fleet Based On Revenue Management

Currently, the application of revenue management has been involved in many industries. The application in car rental industry include mainly four aspects, i.e., demand forecasting, booking, fleet planning and pricing, in which fleet planning is a core team of content. Rental fleet planning is the vehicle logistics optimization process by the car rental company, it can be divided into strategic fleet planning and tactical fleet planning, and before the strategic fleet planning, there needs to segment the pools. Optimizing planningof rental fleet is to optimize the allocation of fleet resources, and reduce the vehicle logistics costs as well as improve the efficiency of logistics operation. As the fluctuation of demand at car rental locations, the customers'random behavior, the distinct types of cars, and the upgrade supply policy, fleet planning in car rental industry is one of the most difficult problems in the application of revenue management.This paper focuses on several key issues in the process of optimizing rental fleet planning, including the pool segmentation, strategic fleet planning and tactical fleet planning. The main work and research results are as follows:According to the pool segmentation principle and three constraint conditions, this paper puts forward a pool segmentation method based on the P-median model, and designs a heuristic algorithm to solve the question, which can obtain the pool segmentation results and regional logistics center position. The model need to adjust the ownership of some sites and can get the satisfactory solution when the demand or the number of leasing sites change.The main tasks of the strategic fleet planning are described, and relevant mathematic model is proposed according to the aim of minimum logistics cost, which can guarantee the actual demand in each pool to be meet. The number and their routes of Car/auto shipping truck is optimized by the vehicle routing problem with split deliveries and pickups, which can meet the demand and make sure the minimum logistics costs.Several common construction methods and improving methods are described, and a three-phase heuristic algorithm to solve the vehicle routing problem with split deliveries and pickups is proposed based on the existing algorithms. First, according to the deliveries and pickups demands of task points, the number of vehicles can be confirmed. Second, visit task points which can delivery and pickup fully at the same time with the sufficient vehicles. And then, the remaining tasks of deliveries and pickups are accomplished by the remaining vehicles, in the light of the dynamic relations among deliveries, pickups and the remaining capacity of currently vehicle. At last, improve the existing routes and get the ultimate vehicle routes. The results of examples show that the algorithm can significantly reduce the transportation costs than the existing algorithms.From the angle of logistics costs, vehicle using costs and company revenue, and based on the nested demand characters of leasing sites, this study builds one-stage models to optimize resource allocation for multi-site and multi-type in a pool and puts forward solution methods for the circumstance of fixed demand and stochastic demand, respectively. By the aid of car use cost, a dummy balanced transportation problem is given for stochastic demand, which can obtain the detailed allocation amount at each site.Optimization model is set up with an objective of minimum operation cost. The fleet planning among rental locations is abstracted as a time-space network, and constraints are obtained according to the supply policy and flow balance at each node. Aimed at the characters of model, original problem is divided into two kinds of sub-problem by the use of Benders decomposition, and corresponding algorithm is proposed subsequently. A numerical example based on one week demonstrates the effectiveness of the proposed model and algorithm. The results indicate that the model and algorithm could be a promising way to improve the management quality of fleet planning.