A Unified Continuous-Time Markov Chain Framework for Modeling and Comparative Performance Evaluation of Finite-Buffer Production Systems

Finite-buffer production systems are widely used in manufacturing, where performance depends on the interaction among machine reliability, buffer capacity, and maintenance policies. This study proposes a unified Continuous-Time Markov Chain (CTMC) framework for modeling and comparatively evaluating alternative finite-buffer production systems. Three configurations are considered: a serial production line with unreliable machines, a workstation-based system with parallel machines, and a condition-based maintenance system incorporating multi-state machine degradation and preventive maintenance. For each configuration, the state space, transition structure, infinitesimal generator matrix, and steady-state probability distribution are derived to estimate throughput, work-in-process inventory, cycle time, and total operating cost. Numerical experiments investigate the effects of buffer capacity, workstation parallelization, maintenance policies, degradation severity, reliability parameters, and financial factors. Under the baseline setting, the conventional serial configuration achieves a throughput of 0.8300 products/min, while workstation parallelization increases throughput to 1.6840 products/min, corresponding to a 102.9% improvement. The condition-based maintenance configuration achieves 0.8173 products/min while accounting for equipment deterioration and preventive-maintenance interventions. Overall, the framework enables the consistent evaluation of operational and economic trade-offs and supports production planning, maintenance optimization, and manufacturing system design.