On Estimating the Probability Distribution Functions in PERT-Type Networks-Revision.

Abstract

This study deals with the problem of approximating the probability distribution function of the project duration in probabilistic activity networks. It describes a procedure that has been developed, programmed and tested, using activity networks of real life projects as well as randomly generated ones. The procedure allows the activity duration to have any of the following distributions: Uniform, Triangular, Normal, Exponential, Gamma, Beta or any discrete distribution presented in a finite set of ordered pairs. The computational experience indicates that the resultant probability distribution function (pdf) is very close to the actual pdf, the latter is obtained through extensive Monte Carlo sampling. In fact, computational experience shows that the pdf obtained by Monte Carlo sampling converges to the approximate pdf as the sample size increases. The procedures is programmed in FORTRAN and the CPU time for any moderate size project (i.e., G(N,A) < or = G(60,200)) is less than half a minute on AMDAHL V-7. (Author)

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jun 01, 1980
Accession Number
ADA088296

Entities

People

  • Bajis Dodin

Organizations

  • North Carolina State University

Tags

Communities of Interest

  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Accuracy
  • Convolution
  • Data Science
  • Discrete Distribution
  • Distribution Functions
  • Fourier Transformation
  • Guarantees
  • Information Science
  • Intervals
  • North Carolina
  • Operations Research
  • Probability
  • Probability Distribution Functions
  • Probability Distributions
  • Random Variables
  • Sampling
  • Standards

Readers

  • Computer Science.
  • Statistical inference.