Bottleneck Detection and Mitigation in Serial Production Systems

Abstract

A variety of analytical models have been proposed to model and analyze serial manufacturing systems. Most analytical models, however, make simplifying assumptions in order to remain mathematically tractable. These analytical formulations do not, however, model the underlying real-world system accurately. Discrete event simulation is one of the primary tools, which provides decision support by capturing the working of complex systems at level of detail and accuracy needed. This thesis analyzes the working of serial production lines characterized by capacitated buffers, stochastic processing times, unreliable machines, rework loops, maintenance and operator issues with the help of discrete event simulation to ascertain its throughput. However, in the past researchers have not been excited about the time it takes to build a complete simulation model. In an effort to fast track the process of model development, a VBA project is undertaken which will dynamically generate a simulation model in Arena 7.01 from an Excel template. We also propose an algorithm which will automatically detect the bottleneck of a serial manufacturing system and provide recommendations to the analyst. These include reallocation of operators, addition of buffers or parallel resources with an objective of increasing the throughput of the system with due economic consideration. Simulation studies are undertaken on different serial manufacturing lines to illustrate the effectiveness of the techniques developed.