Congestion is ever present in most practical situations. We describe a methodology for approximate analysis of open state-dependent M/G/c/c queueing networks in which the service rate is subject to congestion, that is, it is a function of the number of customers in the system. Important performance measurements are easily computed with high accuracy, such as the blocking probability, throughput, expected number of customers in the system (known also as work-in-process), and expected waiting time. The methodology forms a basic building block useful in many practical applications and contexts. Computational results demonstrate that the methodology provides accurate results in many topological configurations as well as in the analysis of network evacuation problems in high-rise buildings.
