Adaptive Estimation and Management of Cost Contingencies in EPC Projects

General contractors of engineering, procurement, and construction (EPC) projects face diverse, interdependent risks in highly dynamic environments; therefore, reliable estimation and management of project cost contingencies are critical for ensuring success and profitability. This study presents a novel adaptive hybrid Bayesian model (AHBM) for estimating the cost contingency in an EPC project by integrating a hybrid Bayesian network with an incremental parameter learning approach, thus adapting to new information while capturing risk interdependencies, propagation, and uncertainties. Two forms of the incremental parameter learning framework, i.e., sequential Dirichlet-based maximum a posteriori and sequential particle-filter-based maximum a posteriori, are incorporated for discrete and continuous variables, respectively. Then, a dynamic contingency management framework (DCMF) is proposed to allocate the estimated cost contingencies across work packages in an adaptive manner and monitor the contingencies over the project progress, supporting informed and dynamic decision-making. The AHBM and DCMF have been applied to demonstrate their efficiency, validity, and practical utility. The study contributes to the domain knowledge of the cost contingency estimation for the EPC project management. The proposed approach enables the general contractor to execute the EPC project with controlled risks and ensured project success and predictability.