Transient Performance Evaluation, Bottleneck Analysis and Control of Production Systems

The focus of production system is to analyze, improve, and control the flow of products in the manufacturing process. In the field of study, the major difficulties are unreliable machines and finite buffers capacities, which lead to nonlinear and stochastic mathematical models. Extensive results on production systems have been derived for steady state operations, while their transient performance and properties are also of practical importance but paid with significantly less attention. In this dissertation, we study the problems of transient performance evaluation, bottleneck analysis, and production control of serial lines, closed lines and assembly systems. Specifically, in the framework of finite production run-based serial lines with Bernoulli/ geometric machine reliability model, we derive mathematical model and analytical formulas to evaluate the performance measures of small systems. Then, we propose computationally efficient algorithms based on decomposition and aggregation for large systems, to approximate the systems performance measures with high accuracy. For closed lines and assembly systems, based on Markovian analysis, we develop the mathematical models and propose approximation methods for transient performance evaluation. For serial lines with Bernoulli machines and with operation control, mathematical models for the system under consideration are derived and analytical methods are developed for calculating the system transient performance. One effective and efficient approach of analyzing transient performance of serial lines, closed lines and assembly systems is present in this dissertation. Bottlenecks, theoretic properties and control of the systems are studied under the transient analysis. Extension of the results to systems with exponential or non-Markovian models, adaptive control of machines, continuous improvement of systems, etc. can be further studied in future work