Buffer Allocation Optimization For FFS Based On Queuing Network Model

With the development of modern industry, the degree of personalized customization of product is more and more higher, which make the ETO enterprises become more and more important. As there are a lot of random factors in the product process of the ETO enterprises, which will bring some problems, such as the manufacturing process is instability, the date of delivery will be lagged behind, and so on. Now for the date of project delivery, the one of core competences, the more effective management method is the prediction method of delivery based-on workload control, but if want to predict the date of delivery accurately, the first step is confirm the boundary of production load, that is to say allocate the buffer capacity of every stage become the most common and the key factor. So an effective research and optimization method of buffer allocation problem is urgently and meaningful.Consider the main random factors in the production process of ETO enterprises, the paper will apply workload control and queuing network modeling theory, combine the simulation modeling method, treat the multistage flow-shop and every stage has more than one device as an object, research its buffer capacity allocation and optimization, the details as follows:Firstly, the paper introduces the common workload control methods and their limitations. Then expounds the relationship between workload delimited and buffer allocation from analysis the characteristics of flow-shop with one single type job, after that analysis and select the goal functions of buffer allocation problem.Secondly, the paper analysis all the states of general node i in the model, and write down the CTMC states transfer chart and lists the states transfer balance equations, at the same time expounds the method of states space classification and the solution of states equations, and infers the calculate method of the statistics index. Then developed a procedure based on Visual Studio and Mat lab to solve the states transfer balance equations, and verify its validity from setting up examples. Thirdly, in order to extend the solution to more large and more complex manufacturing systems, the paper structured a eight-stage simulation model and every stage has more than one device based on eM-Plant simulation software, and designed a group of examples to optimize the buffer capacity of the model, and illustrate the simulation statistic is an effective solution of BAP via contrast the simulation results and the results in the literature.Finally, the paper design a series of examples for the eight-stage flow-shop via two heuristic methods, and on the premise of satisfied the system preinstall rejected rate, treat the average lead time and the average utilization rate of equipment as objective functions respectively, optimized the buffer capacity of every stage in the model, compares and analysis the numerical results and the simulation results, and verified the system performance.This paper using the queuing network modeling and workload control theory, built a flexible flow-shop buffer capacity allocation and optimization model in the random environment, and introduces the corresponding numerical algorithm and the simulation calculation process, designs and develops the corresponding software to provide reasonable parameter settings for the workload norms of the shop buffer allocation and optimization which based on workload control.