Research On Order Priority Sequencing Of Re-entrant Manufacturing System

With the development of market economy and the diversification of marketsdemands, Make-to-Order(MTO) production is accepted by more and moremanufacturing companies instead of traditional Make-to-Stock production, especiallyfor the reentrant production system, represented by semiconductor manufacturing lines.Based on this background, the characteristic of semiconductor manufacturing lines andthe difference between it and Flow-shop as well as Job-shop is given. Detailed analysisof its model description and scheduling research is also provided in this paper.Typical scheduling policies of order sequencing for common MTO companies aresummarized. And their advantages and disadvantages are studied in the background ofsemiconductor manufacturing lines. By using Simul8software, a model with essentialcharacteristics of semiconductor manufacturing line is modeled for simulation analysis.Firstly, research on the manufacturing lines performance affected by the ordercombination variation is conducted. Simulation results show that order combinativerelease policy is better. Secondly, discussions on the manufacturing lines performancevariation caused by the on-line order amount are carried out.The hierarchical order priority sequencing algorithm based on Analytic HierarchyProcess(AHP), with a two-order structure, is proposed in this paper to calculte the orderpriority. Order acceptance/rejection control by the workload of bottleneck is taken at thefirst step, its necessity and effectiveness are proved by the related simulationexperiments; and the second step is to calculate the order priority using AHP.Application of this algorithm is showed by a detailed example using MATLAB.As to the aspect of the application and dynamic improvement of above mentionedalgorithm, a global optimum release policy, named Dynamic ClassifiedWork-in-Progress (shortly for DC-WIP), is proposed for order combination and releasecontrol of the semiconductor manufacturing lines. Firstly, we set the maximum steadystate release rate for each order base on the Theory of Constraint (TOC). Next thefeedback control of DC-WIP is used to balance the manufacturing lines. DC-WIPrelease rule controls the WIP allocation according to the maximum production rate of corresponding bottleneck machine. Once there is an order arriving at or leaving themanufacturing system, the optimal WIP level will be reset according to the DC-WIP.The simulation result also proves its effectiveness.In the end of this paper, a order management and scheduling scheme based onmulti-agent is constructed, which turns the order scheduling problem into a Multi-AgentSystem(MAS) management problem. Order release is controlled by bothrelease/rejection control and priority sequencing. Additionally, hot order scheduling isalso taken into consideration in this management and scheduling scheme.