The maximum and sum of two beta variables and the analysis of PERT networks

This paper explores the errors involved in the approximation of both the maximum and sum of two independent Beta random variables, each by another Beta random variable. The principal area of application is the time analysis of PERT networks, and the results are interpreted and discussed in terms of this application. The error was found to be acceptable for the greater part of the range of the parameters involved. The results will help practitioners estimate the mean and standard deviation of the completion time of PERT networks.