The problem of service and capacity allocation in state-dependent M/G/c/c queueing networks is analyzed and algorithms are developed to compute the optimal allocation c. The model is applied to the modeling of pedestrian circulation systems and basic series, merge, and split topologies are examined. Also of interest are applications to problems of evacuation planning in buildings. Computational experiments assert the algorithm's speed, robustness, and effectiveness. The results obtained indicate that the pattern of the optimal capacity surprisingly repeats over different topologies and it is also heavily dependent upon the arrival rate. Additional computational simulation results are provided to show the accuracy of the approach in all configurations tested.
