Optimization Models And Algorithms For Production Planning And Scheduling Of Multiple Plants

Production planning and scheduling are the core tasks of the enterprise production management. The traditional production planning and scheduling methods were developed for the enterprise with single plant. Along with the developement and extension of enterprise, production organization with distributed multiplant becomes more and more popular. It brings new requirements to production planning and scheduling. How to realize the cooperation of production planning and scheduling between multiple plants, how to implement the coordination of production and transportation, and how to tackle with uncertainty and recover the feasibility of schedules against unexcepted events. The main contribution of this thesis can be summarized as follows.1. A multiplant production planning problem integrating production and transportation is studied. Uncertain demand and capacity are expressed as fuzzy numbers in this problem. A process model, named extended state task network (ESTN), is proposed to model abstractly production, storage, and transportation activities. By adding virtual proportion transformation task and combination transfer task, state task network (STN) is extended as ESTN to describe transfer activity in which several types of material can be transferred simultaneously in one trip. A fuzzy mixed integer linear programming model is presented based on the ESTN. The model is transferred into a parameter planning model. Path relinking algorithm is studied to solve the parameter planning model. Solution quality and diversification based solution update strategy of reference set and decentralization mutation strategy are presented in the path relinking algorithm. The computational experiments are executed to evaluate the performance of the path relinking algorithm.2. Production scheduling pattern integrating the batching and the batch scheduling is investigated. Production scheduling of batch process with parallel machines is studied in the production scheduling pattern. Sequence independent changeover time and sequence dependent changeover time are distinguished in the scheduling problem. Firstly, a fixed charge transportation (FCT) model is formulated to solve the scheduling problem with sequence independent changeover time. Genetic local search algorithm (GLSA) is developed for solving the problem. Intensification local search strategy and escape strategy from local optimal solution are investigated to intensify the capacity of searching global optimal solution. The computational experiments are executed to evaluate the differences of performance between GLSA, permutation encoding genetic algorithm, matrix encoding genetic algorithm, and tabu heuristic search procedure. Secondly, a continuous time integer programming model for the scheduling problem with sequence dependent changeover time is developed. The computational experiments are executed to evaluate the differences of performance between the model and the scheduling model in which the batch scheduling is called after the batching process.3. The cooperation scheduling method syncretizing multipurpose batch process production scheduling and transportation scheduling is proposed. The scatter search algorithm for the production scheduling of multipurpose batch process and the heuristic algorithm for the transportation scheduling are studied. To verify the abilities of decreasing manufacturing cost and guaranteeing due date, the cooperation scheduling method is compared with the independent scheduling method in which the transportation scheduling is executed after the production scheduling. In addition, the computational efficiency is evaluated in the computational experiments.4. Coordination mulitplant rescheduling method is investigated based on multi-agent. Inserting emergent order is the source of disturbance. Performance evaluation index of multiplant rescheduling and coordination algorithms of agents are studied. Affected batch rescheduling (ABR) algorithm is developed in order to solve the rescheduling of multipurpose batch process. ABR shifts only the batches affected by the disturbance. The computational experiments are executed to compare the production stability and disturbance propagation throught plants of the coordination multiplant rescheduling method and the multiplant rescheduling without coordination. ABR algorithm is compared with right shift rescheduling (RSR) on makespan and delay of start time of batch.5. A production planning and scheduling system is designed and developed based on J2EE platform. The models and algorithms presented in this thesis are implemented in the system. The system is used to aid the multiplant production planning and scheduling in Alcohol Company of China Grain Group. The theories and methods in this thesis are verified by the application.