Buffer allocation in asynchronous serial production systems with Bernoulli machines during transients

Some asynchronous production systems can never be truly balanced. This phenomenon demands the incorporation of asynchronism and unreliability in buffer allocation problems (BAPs). This paper reports a BAP in an asynchronous serial production system with Bernoulli machines. A profit-based integer programming model is formulated to maximise the throughput and minimise the buffer capacity simultaneously. Because the production run is finite in the current mass customisation environment, a transient analysis is used to estimate the system performance. Piecewise closed-form expressions are derived for a two-machine case based on Markovian structures. For M > 2 machine lines, an efficient recursive algorithm based on aggregation is developed. Considering the nonlinear characteristic of this BAP, a dynamic-objective particle swarm optimisation with neighbourhood searching is developed. This approach is verified by a case from an assembly plant. The results demonstrate that the programming method can increase the system profit by 1.85%. Copyright © 2021 Inderscience Enterprises Ltd.