Improving Risk-Based Decision-Making Using Bayesian-Enhanced EVM

Abstract

This thesis addresses a central challenge in infrastructure project management: despite sophisticated monitoring tools like Earned Value Management (EVM), teams struggle to proactively identify and address root causes of performance deviations. The research investigates why EVM-based systems often fail to deliver actionable insights and explores whether integrating Root Cause Analysis (RCA) with Bayesian-enhanced EVM can bridge this gap. Anchored in a large infrastructure project at Schiphol Airport, the study combines literature review, interviews, Bayesian modelling in GeNIe, and validation with industry stakeholders. The findings present a dual-layer framework: (1) a Bayesian Network linking performance indicators (SPI, CPI) to risk drivers, enabling backward reasoning from deviations to causes, and (2) integration of RCA to reveal systemic factors behind disruptions. Validation confirmed the tool’s conceptual usefulness but highlighted adoption challenges, including dashboard integration, model complexity, and cultural readiness. The research concludes that embedding RCA into Bayesian-enhanced EVM shifts it from a reactive tool to a forward-looking enabler of organizational learning. However, technical robustness must be complemented by openness, accountability, and a culture of learning. Recommendations include piloting in mid-complexity projects, calibration with real data, and building scenario libraries from historical RCA cases. Overall, the work reframes risk management as a dynamic, learning-oriented process, emphasizing that better project outcomes rely not only on better data, but also on diagnosis, culture, and collaborative decision-making