EFFECT OF DIFFERENT ACTIVITY DURATION DISTRIBUTION ON PROJECT DURATION PREDICTION PROBLEM
Adepeju A. Opaleye1 & Oliver E. Charles-Owaba2
1 &2 Department of Industrial and Production Engineering,
University of Ibadan, Nigeria
Email: opaleye_adepeju@yahoo.com
ABSTRACT
Project scheduling researchers mostly rely on beta or triangular distribution for modelling activity duration. This reliance is associated with the underlying assumption of the traditional PERT model. Contemporary researches in this area have however partly refuted this concept and many different distributions proposed without empirical justification. In this study, availability of historical project activity duration data in construction industry was investigated and associated statistical distribution empirically determined. Effects of these distributions on the project completion time were also experimentally investigated. It is shown that about 49.23% of activities durations observed exhibited lognormal distribution and a significant difference in predicted project duration exist due to varying activity duration distribution.