Analytical Modeling Of Series-parallel Production Lines

The modeling and performance evaluation of production system have significant values in industry because of the increased market size and increased competition among enterprises. This dissertation concentrates on the research of series-parallel production lines in the following aspects.In this dissertation, we address the problems for modeling of production system and point out the characteristics of the system performance. First we model and analyze a two-stage manufacturing system with multiple independent unreliable machines at each stage and one finite-sized buffer between the stages. This model is exact and is constructed by means of a Markov processes. Then we define the state space of the Markov processes and generate the transfer rate matrix of this Markov processes. We further point out that the transfer rate matrix can be transformed into a matrix form fitting the properties of a Quasi-Birth-Death (QBD) process, and use the Matrix-Analytic (MA) method to calculate the system state probability and throughput. Numerical results demonstrate that the new multi-state model and the MA solution method are accurate and fast in analyzing manufacturing systems with parallel machines.An approximate decomposition method is presented for the analysis of the general series-parallel production lines. First a two-stage analytical model with multiple independent unreliable machines at each stage is introduced as the basic component in the system. Then an approximate decomposition method is presented to analyze the system throughput. Numerical experiments are given to show the accuracy and reliability of the decomposition method.Comparing with the Witness simulation model, the proposed analytical model in this dissertation is accurate and reliable. On this basis, we use this model to analyze and discuss the effects of system parameters on system performance, such as machine quantity, processing rate, failure rate and repair rate etc. And the results indicate that: (1) the throughput of production system will increase while the buffer capacities increase; (2) the WIP will increase while the buffer capacities increase; (3) the system performance improves with the improvement of machines? performance; (4) the machines at the bottle neck station have more influence on system performance than the machines at other stations. And these indicate that the model we proposed have a great application prospect.