Building Trust in AI for Project Controls: A Governance-Embedded Data Engineering Framework

Abstract

Capital projects in the energy sector continue to experience systemic cost overruns and schedule delays. While AI-driven predictive analytics promise improved foresight, their effectiveness is often limited by fragmented data landscapes, inconsistent reporting practices, and weak governance. This paper introduces a data engineering framework for AI-driven project controls, showing that reliable forecasting in capital projects depends more on structured, governed data preparation. The framework unifies engineering, operational, and financial data through structured harmonization, entity resolution, temporal alignment, and governance-embedded quality controls. The approach was developed and validated using data from capital projects, demonstrating that disciplined data preparation is foundational to achieving reliable AI-driven forecasting in complex project environments.