Modelling the uncertainty of cost and time in highway projects

Abstract

The construction of highway projects is characterised by cost overruns and time delays, due to the estimation approach and inappropriate analytical tools for predicting uncertainty. This study therefore developed a hybrid intelligent tool that models three sources of uncertainty in linear infrastructure projects: variability, correlation and disruptive events. The developed tool measures the effect of uncertainties on the cost and time of projects, by combining classical and intelligence prediction techniques. The variabilities were modelled using probability distributions; the copula technique modelled the correlations. The Markov processes simulated the occurrence of disruptive events. The adaptive neuro-fuzzy inference system was used to assess the size of the impact of disruptive events on the cost and time of activities. The total project cost and time were simulated by propagating the impact of the three sources of uncertainty in the Monte Carlo simulation environment. The developed uncertainty model was validated against the final cost and time of a highway project. The study found that the accumulated impact of the three sources of uncertainty significantly increased the construction cost and time of infrastructure projects. It concludes that the improvement in the accuracy of cost and time estimation of highway projects depends on a combination of classical and intelligent prediction techniques.