Applying Bayesian networks to model uncertainty in project scheduling

Abstract

Risk Management has become an important part of Project Management. In spite of numerous advances in the field of Project Risk Management (PRM), handling uncertainty in complex projects still remains a challenge. An important component of Project Risk Management (PRM) is risk analysis, which attempts to measure risk and its impact on different project parameters such as time, cost and quality. By highlighting the trade-off between project parameters, the thesis concentrates on project time management under uncertainty. The earliest research incorporating uncertainty/risk in projects started in the late 1950’s. Since then, several techniques and tools have been introduced, and many of them are widely used and applied throughout different industries. However, they often fail to capture uncertainty properly and produce inaccurate, inconsistent and unreliable results. This is evident from consistent problems of cost and schedule overrun. The thesis will argue that the simulation-based techniques, as the dominant and state-of-the-art approach for modelling uncertainty in projects, suffers from serious shortcomings. More advanced techniques are required. Bayesian Networks (BNs), are a powerful technique for decision support under uncertainty that have attracted a lot of attention in different fields. However, applying BNs in project risk management is novel. The thesis aims to show that BN modelling can improve project risk assessment. A literature review explores the important limitations of the current practice of project scheduling under uncertainty. A new model is proposed which applies BNs for performing the famous Critical Path Method (CPM) calculation. The model subsumes the benefits of CPM while adding BN capability to properly capture different aspects of uncertainty in project scheduling.