An Estimate of Expected Critical-Path Length in PERT Networks

Abstract

The estimate of expected critical-path length in PERT networks in this paper is usually better than the F ulkerson estimate because of greater flexibility in choosing a path whose length approximates the length of the critical path. As the calculations are based on a model approximating the actual network, the new estimate may fall on either side of the expected critical path, even though the Fulkerson estimate never exceeded the expected critical-path length. A Fortran program was written to calculate the new estimate, whose calculation is more complicated man Fulkerson's estimate. The new estimate, Fulkerson's estimate, and the expected critical-path length are compared in several examples.