Stochastic Model for Project Performance Control

Abstract

A common problem, throughout various industries, is expansion of projects beyond their originally allocated resources (time and money). A recent report of American Society for Quality shows, that less than 35% of projects could be considered successful, i.e. they both met the project objectives and were completed for the time and cost they were agreed to. Traditional project management practices include project planning & control from a pure deterministic standpoint. A typical project schedule is based on a series of deterministic assumptions regarding the duration of each one of the activities that together, make up the project as a whole. The same is true about a budget, as the sum of deterministic estimates of each cost component. The proposed model for project performance control from schedule and budget points of view creates a planning & control framework, deployed in parallel to (but not instead of) the traditional methods. The model is based on the procedure of Probability Encoding developed by Lockheed-Georgia Company [1] (in the framework of NASA projects - end of 1980's). The utilization of this model is based on an iterative process that occurs at pre-defined Control-Points or Gates along the project timeline. Each iteration includes a process of collecting duration and cost estimates, given by a multi-disciplinary team of experts. Those estimates are encoded into corresponding distribution functions, and the model is developed to reflect the inter-relations and dependencies between project activity packages. Using a Monte-Carlo simulation, two distributions for the total cost and for the potential finish dates are characterized. These distributions reflect a whole range of potential project outcomes (in terms of schedule and cost), i.e. the uncertainty associated with the cost estimate as well as with the expected finish date.