A New Technique for Estimating the Distribution of a Stochastic Project Makespan

Introduction

Scheduling is a core project management area, and finding project's duration (makespan) is one of its important objectives (Kerzner, 2006; PMI, 2008; Zwikael et al., 2006). This paper presents a practical technique that estimates the makespan distribution of a project in a more accurate and reliable way than PERT (Project Evaluation and Review Technique), and is consistent with Monte Carlo Simulation results.

One of the most common techniques for project scheduling under uncertainty is PERT, which was originally proposed by Malcolm et al. (1959). While it is still a prevalent planning and controlling tool in project management (Ash & Pittman, 2008; Adler & Smith, 2009), it has been critiqued that PERT provides inaccurate information about the project completion time (e.g., Klingel, 1966; Shogan, 1977; Schonberger, 1981; Dodin, 1985; Schmidt & Grossman, 2000; Dodin, 2006; Hahn, 2008; Kirytopoulos et al., 2008). Alternatively, Monte Carlo simulation has been a practical tool for evaluating the makespan distribution of a stochastic project (e.g., Van Slyke, 1963; Burt & Garman, 1971; Sullivan et al., 1982; Iida, 2000; Demeulemeester & Herroelen, 2002; Kirytopoulos et al., 2008). However, the simulation computational intensity and the time consumed by a simulation, motivated research for searching alternatives (Adlakha & Kulkarni, 1989; Cohen & Zwikael, 2008; Salaka & Prabhu, 2008). While several research directions had been pursued (as discussed in section 1.1.) there is still a need for a practical technique that would replace PERT and would give consistent results with Monte Carlo Simulation.