Abstract
This paper is devoted to studying scheduling policies in flexible serial lines with two Bernoulli machines and dedicated finite buffers. Priority, cyclic and work-in-process (WIP)-based scheduling policies are investigated. For small scale systems, exact solutions are derived using Markov chain models. For larger ones, a flexible line is decomposed into multiple interacting dedicated serial lines, and iteration procedures are introduced to approximate system production rate. Through extensive numerical experiments, it is shown that the approximation methods result in acceptable accuracy in throughput estimation. In addition, system-theoretic properties such as asymptotic behavior, reversibility, and monotonicity, as well as impact of buffer capacities are discussed, and comparisons of the scheduling policies are carried out.