Research On The Method Of Capacity Optimization Configuration Of Facilities For The Order-driven Equipment Manufacturing System

Equipment manufacturing is a strategic industry which provides equipment for other industies of national economy. However, the challenge from customization and the shorter due date is given to the production management of equipment manufacturing. One important reason is that the exact capacity configuration of facilities is difficult. It is not only to consider the fluctuation of orders and the limit of due date, but also to aviod the waste caused by over-investment of facilities. And the optimization of all kinds of facilities configuration needs to satisfy the demand of production under uncertain manufacturing environment with the lower investment of facilities for equipment manufacturing enterprise. So the capacity configuration problem(CCP) under the dual constraints of the average throughput and makespan is studied in this paper, which is against the background of equipment manufacturing with uncertainty.The CCP of equipment manufacturing can be abstracted as a stochastic and nonlinear integer programming problem. Because of the flexibility of processing routing, uncertain manufacturing environment and stochastic orders in equipment manufacturing, the relationship between the parameters including the average throughput and makespan and the variables consisting of the facility number and the maximum buffer capacity can’t be expressed by close form. And the solving, evaluation and analysis of CCP become very difficult. An approximate optimization method integrated mathematical analysis with simulation optimization is presented to approach the best object value by improving upper and lower bound. The main works of this paper are following:Firstly, a simulation system of equipment manufacturing consisting of multi-productions, multi-processes, multi-routings and multi-types of facility is created by abstracting typical procedure of order-driven equipment manufacturing. The experiments of simulation are designed based on the anaylsis of facts which enfluence the performance indicators of manufacturing system. The relationship between the parameters including the average throughput and makespan and the variables consisting of facility number and buffer capacity is analyzed preliminarily by the experiments of simulation. And the research of the the approximate optimization method is based on the resluts of simulation experiment.Secondly, the introduction of the approximate optimization method is presented. An optimization model is developed for CCP of equipment manufacturing. And the detail of the approximate optimization method is described including the acquisition of the initial upper bound and lower bound of the best object value and the steps of improvement of the bounds. The proof of the relational conclusion is presented afterward.Thirdly, considering the flexibility of processing routing in equipment manufacturing, the optimization of CCP is studied with only one constraint of throughput first of all. The flexibility of processing routing is described by network flow gragh, and a mathematic model of CCP is developed by regarding the average throughput of system as a constraint. The initial lower bound of the best object value is abtained by solving the model. A simulation system is established with the configuration of facilities corresponding to the initial lower bound of the best object value. The initial upper bound of the best object value is gained by the heuristic method of simulation optimization. And the best object value is got by improving its upper and lower bound continuously.Fourthly, according to the aforementioned research idea, the further study on the optimization of CCP is advanced with the dual constraints of the average throughput and makespan. The optimization of CCP is modeled approximatively by relaxing constaints of he average throughput and makespan. After solving the model, the initial lower bound of the best object value under this condition is acquired, and the initial upper bound of the best object value can be obtained by the heuristic method of simulation optimization in stage. The upper and lower bound are improved sustainedly to gain the best object value. The optimization results from the two aforementioned cases are analyzed and contrasted lastly.Fifthly, the optimization of CCP is researched deeply facing the stochastic arrival of orders and the uncertain manufacturing environment. The manufacturing system of equipment manufacturing with uncertainty is described by queueing network. To avoid the curse of dimensionality in parameter caculation of queueing network, an approximate mathematic model is built based on the relax constraint including performance parameters of manufacturing system, and the model is solved to get the initial lower bound of the best object value. The sequence of upper bound is obtained by the method of simulation optimization based on genetic algorithm(GA). Using the performance parameters from simulation experiment to substitute into the mathematic model, the sequence of lower bound is gained subsequently. And the iteration will not be ended untill the upper and lower bound are very close to the best object value.Finally, a software platform for the optimization of CCP is developed based on optimization method presented by this paper to applicate to a large enterprise of equipment manufacturing. And the decision supports to the capacity configuration of facilities and the evaluation of layout from the project programming new plant of the enterprise is supplied from the software platform. The background of the project is introduced first. The software platform is designed based on analysis of operation flow, function design, database design and design of frame structure etc. The development and application of the software platform is presented lastly.Based on the above research and application, the optimization of CCP under uncertain environment for the equipment manufacturing enterprise is tried to solve in this paper. The relevant research conclusions of this paper can be used to support the equipment manufacturing enterprises to program their facilites configuration in middle or long term.