Optimization Research On Mobile Facility Planning And Scheduling Under Uncertainty

Mobile facility refers to the certain kind of vehicle that can provide service to the nearby customers when they are stationary as like a traditional fixed facility, and meanwhile can also flexibly move from one place to another like an ordinary vehicle. Given the low costs and the flexibility in changing service locations, mobile facilities have been widely used in many areas, such as cellular service, health care, public service, humanitarian relief service, and etc. The quality of the mobile facility planning and scheduling directly a?ects the performance of the overall performance of the system, however, the associated academic research is still in an early stage. Hence, the optimization of the mobile facilities fleet is an important topic in the field of the logistics and supply chain network design.This dissertation focuses on the optimization research on mobile facility planning and scheduling problem under uncertainty, aiming at applying the methodology of stochastic programming and robust optimization into the studies of the mobile facility routing and scheduling problem with stochastic demand, the stochastic emergency mobile facility redeployment problem and the mobile facility fleet sizing problem under uncertain demand. The two-stage stochastic programming model and two-stage robust optimization model are proposed. By exploiting the inherent characteristics of the problems, an enhanced L-shaped algorithm and a two-level cutting plane algorithm are developed. Extensive numerical experiments are conducted to evaluate the e?ciency of the algorithms and e?ectiveness of the proposed models.Unlike the existing research on deterministic mobile facility routing and scheduling problem, we consider the uncertainty in demand and a two-stage stochastic programming model is proposed for mobile facility routing and scheduling problem. In the algorithm design process, we attempt to improve the traditional L-shaped method by adding lower bound inequalities, disaggregating the optimality cuts and relaxing stopping criteria. A series of numerical experiments are conducted to evaluate computational e?ciency of the algorithm and the e?ectiveness of the proposed improvement approaches, and to quantify the necessity of applying the two-stage stochastic programming method in dealing with the uncertainty in demand.By introducing the concept of multi-period planning of the mobile facility into the field of emergency service system planning field, we formulate a two-stage stochastic programming model for stochastic emergency mobile facility redeployment problem. The influence of the uncertainty in emergency service demand and return time information of the emergency mobile facility over the redeployment decision are considered in the model. The computational experiments are performed to evaluate e?ectiveness of the model and to test how the random return time information a?ects the performance and robustness of the whole system.Regarding to the mobile facility fleet sizing problem under uncertain demand, we formulate a two-stage robust optimization model, in which the uncertainty in demand is modeled as a polyhedral uncertain set. We also propose an enhanced two-level cutting plane algorithm to solve the model. Comparing to the traditional cutting plane method,the new algorithm attempts to improve lower bounding e?ect of the master problem solving process and reduce the computation time of the recourse problem solving process.Numerical experiments are conducted to assess the computational e?ciency of the algorithm and e?ectiveness of the proposed model.