Monte Carlo estimation of the maximal flow distribution with discrete stochastic arc capacity levels

This article describes a highly efficient Monte Carlo sampling plan for estimating ordinates of the complementary distribution function of the maximal s-t flow in a directed network with source node s, terminal node t and whose arcs assume random capacities. Communication, electric power, transportation, and water distribution systems fit this representation. The article derives point estimates and interval estimates that hold for finite sample sizes, and shows the extent to which the proposed technique saves time, as compared to crude Monte Carlo sampling, in achieving a specified accuracy.