Monte Carlo Techniques for Stochastic Pert Network Analysis

Abstract

The choice of analytic, approximation, or Monte Carlo methods (and combinations thereof) as means of solving the stochastic PERT problem depends on the activity cumulative distribution functions, network configuration, computing resources available, and desired accuracy. When Monte Carlo methods are Indicated, there exist several techniques for improving accuracy and diminishing computational effort. In this paper the application of five such techniques is described: antithetic variates, stratified sampling, control variates,regression, and conditioned sampling. These techniques may be used separately or in combinations; their existence makes Monte Carlo methods much more attractive in the investigation of stochastic PERT networks.