Monte Carlo Technique with Correlated Random Variables

TLDR

Monte Carlo simulation for project risk analysis is reviewed, highlighting implementation challenges related to cost distribution types and inter‑cost correlations. The study examines how correlations among construction cost components affect total cost variance and offers guidelines for future research. A multivariate lognormal distribution is employed to generate correlated random numbers for cost components, demonstrated with a large real‑life dataset. The proposed method accurately predicts total cost variance.

Abstract

In this paper the impact of correlation among cost components on the total cost variance of a construction project is analyzed. The Monte Carlo simulation approach for project risk analysis is reviewed and the problems with implementing this technique is described. Specifically, issues related to the type of the underlying cost distributions and existing correlations between cost items are explored. One suggested methodology for generating correlated random numbers in a simulation environment is reviewed. A multivariate lognormal distribution is used to generate correlated random numbers for various construction cost components. A large, real‐life data set is used to show the implementation of the suggested method and the results of the analysis are analyzed. It is shown that the proposed method is capable of predicting the variance of the total cost with good accuracy. Some guidelines for future research in this area are presented.