ANALYSIS OF STOCHASTIC NETWORKS
Abstract
The purpose of this research was to: (1) develop a general purpose GERT Simulation Program, (2) investigate resultant distributions of GERT networks containing all AND nodes to determine validity of the PERT normality assumptions, and (3) investigate analysis of AND nodes through analytical and simulation methods for distribution of the equivalent time parameter. The Simulation Model is a fast, flexible, user-oriented computer program requiring only one input card for each activity in the network. Five probability density functions are available for use in describing the distribution of activity durations. Program output includes a criticality index on each activity and node analysis on specified nodes, including mean, variance, probability of realization and histograms of node realization times. Various GERT networks were simulated to provide examples of the use of the Simulation Model and verify results through statistical tests. Analysis of resultant distributions of PERT networks verified the PERT normality assumptions utilizing approximation formulas for mean and variance of activity durations. The merge bias correction procedure utilized to produce better approximations of network realization times was found not applicable to general solution of GERT networks containing AND nodes.
Document Details
- Document Type
- Technical Report
- Publication Date
- May 01, 1966
- Accession Number
- AD0635234
Entities
People
- William W. Mcdonald
Organizations
- Arizona State University