Estimating a Project's Completion Time Distribution Using Intelligent Simulation Methods

Other researchers have indicated the theoretical problems of using the PERT techniques commonly described in elementary operational research and operations management texts. Monte Carlo simulation has often been suggested as an alternative means of analyzing PERT networks but simulating a network of real world proportions is often assumed to be prohibitively expensive. This paper measures the accuracy and cost of standard PERT and simulation methods for real world sized problems. In addition, several computational heuristics are described and tested that indicate that simulation is a viable alternative. The results indicate that intelligent simulation of PERT networks is considerably more accurate than standard PERT analysis and is definitely not cost prohibitive.