Continuous-flow simulation of manufacturing systems with assembly/disassembly machines, multiple loops and general layout

Performance evaluation methods are important to design and control manufacturing systems. Approximate analytical methods are fast, but they may be limited by the restrictive assumptions on the system. On the contrary, simulation has not specific limitations in its applicability, but the time to model and analyse a manufacturing system can increase as the level of detail addressed by the model increases. The main contribution of this study is presenting a computationally efficient methodology to simulate single-part continuous-flow manufacturing systems with assembly/disassembly machines, multiple loops, general layout and general inter-event time distributions. By using graph theory, a new method is presented to identify the machines causing slowdown, blocking and starvation in a general layout and determine the time before the occurrence of a state transition for each machine and the time before the fulfilment or depletion of each buffer. By advancing the time clock to the next event-time accordingly, the number of discrete events needed to be simulated is decreased compared to a discrete-event simulation with discrete flow of parts. As a result, the proposed method is on average 15 times faster than DES methods in the analysis of discrete-flow systems, and 110 times faster on average in the analysis of continuous-flow systems. The low computational time of the proposed method allows to simulate systems under general assumptions and in a very short time.

Automated summary

The main contribution of this study is presenting a computationally efficient methodology to simulate single-part continuous-flow manufacturing systems with assembly/disassembly machines, multiple loops, general layout, and general inter-event time distributions. By using graph theory, a new method is presented to identify the machines causing slowdown, blocking, and starvation in a general layout and determine the time before the occurrence of a state transition for each machine and the time before the fulfillment or depletion of each buffer. The proposed method is significantly faster than traditional methods, making it suitable for analyzing both discrete-flow and continuous-flow systems.