A Bayesian-driven Monte Carlo approach for managing construction schedule risks of infrastructures under uncertainty

•A data transformation approach was developed to construct Bayesian network.•Bayesian-driven Monte Carlo approach was developed for schedule risk management.•A case study was conducted to verify the developed approach and tool. The construction of infrastructures has often been challenged by construction delays, which are disruptive and expensive. As the construction of infrastructures is more complex and riskier under uncertainty, the construction schedule risks are becoming more interconnected and occurring in a chain of cascading events in practice. However, past research typically ignored the interdependency between sequence of risk occurrence (e.g., chronological and causal relationships). This research aims to close this gap through developing a novel approach based on Bayesian-driven Monte Carlo (BDMC) simulation for managing these interdependent construction schedule risks of infrastructures under uncertainty. This approach integrates hybrid data processing and analytics methods to (1) construct Bayesian network for identifying risks and risk interdependencies (i.e., causal relationships), (2) conduct risk inference and construction duration prediction involving both of chronological and causal relationships between risks, and (3) identify critical and sensitive risks and provide the most appropriate strategy for risk mitigation. The approach firstly pre-processes the data from risks and risk interdependencies to construct the risk network. It further constructs the Bayesian network using deep-first search (DFS), adapted maximum-weight spanning tree (A-MWST) algorithms and leaky-MAX model. Then the approach is developed for risk mapping based on BDMC simulation and for risk mitigation based on sensitivity and scenario analysis. Finally, a real infrastructure project is selected as the case study to verify this developed approach. Compared to conventional methods, the results show that the developed approach can provide more accurate schedule prediction with least 0.166% error ratio through incorporating interdependent risks into schedule prediction. It is also more informative through proposing effective risk mitigation strategies for delay avoidance and uncertainty reduction, and more convenient in data acquisition and processing when developing a Bayesian network through the developed hybrid data transformation approach converting the risk network into Bayesian network. This research contributes a new way to understand and analyse the interdepend