Operational capacity allocation for unreliable module-based assembly systems

This paper discusses the operational capacity allocation (OCA) problem in an unreliable module-based assembly (MOBA) system. MOBA systems stem from pervasive applications of agile manufacturing. A MOBA system consists of a number of assembly stations in series, each of which is fed by a sub-fabrication line. Each assembly station performs one assembly operation which requires one workpiece pulled from the upper stream station and one module supplied from its sub-fabrication line. The inter-station buffers are small. The assembly stations are subject to random breakdowns and repairs. OCA for throughput maximization is a key issue in the design and operation of a MOBA system. We develop a set of rules for allocating a finite amount of operational capacity of the production system so that its throughput is maximized. Simulation results show that this policy is robust and applicable to systems with more than three stages and with general processing times. Through numerical experiments, we discuss the impacts of the system parameters on the optimal allocation policy. We then propose the concept of soft and rigid throughput bottlenecks and explore criteria for identifying the soft throughput bottleneck of a system. We believe our work provides a basis for a promising approach for OCA in complex and unreliable assembly systems.