Estimating Activity Duration in PERT Using the Burr XII Distribution
E. F. Udoumoh
*
Federal University of Agriculture, Makurdi, Nigeria.
I. A. Iwok
University of Port Harcourt, Port Harcourt, Nigeria.
*Author to whom correspondence should be addressed.
Abstract
The problem of parameter elicitation has restricted project managers to the use of few known and sometimes unrealistic input activity distributions in Project Evaluation and Review Technique (PERT). In this paper, the problem of parameter elicitation in PERT with Burr XII activity duration distribution is addressed using classical quantile estimation approach. The proposed method uses any two extreme quantiles alongside the median quantile to develop a procedure for the estimation of activity duration in PERT-Burr XII. Some desirable properties that qualify Burr XII distribution as input duration distribution are also outlined. An adequacy check was demonstrated by comparing the absolute percent errors of the mean estimates from the proposed method with the PERT-Beta and other approximations from the PERT-Beta family. The proposed method presents a relatively simple and accurate procedure for parameter elicitation in PERT. The procedure is therefore recommended for wide application in PERT especially where heavily tailed and right skewed data are involved.
Keywords: PERT, stochastic activity network, burr XII distribution, project completion time, quantile estimation