Automated process discovery from event logs in BIM construction projects

Abstract To fully understand how a construction project actually proceeds, a novel framework for automated process discovery from building information modeling (BIM) event logs is developed. The significance of the work is to manage and optimize the complex construction process towards the ultimate goal of narrowing the gap between BIM and process mining. More specifically, meaningful information is retrieved from prepared event logs to build a participant-specific process model, and then the established model with executable semantics and fitness guarantees provides evidence in process improvement through identifying deviations, inefficiencies, and collaboration features. The proposed method has been validated in a case study, where the input is an as-planned event log from a real BIM construction project. The process model is created automatically by the inductive mining and fuzzy mining algorithms, which is then analyzed deeply under the joint use of conformance checking, frequency and bottleneck analysis, and social network analysis (SNA). The discovered knowledge contributes to revealing potential problems and evaluating the performance of workflows and participants objectively. In the discussion part, as-built data from the internet of things (IoT) deployment in construction site monitoring is automatically compared with the as-planned event log in the BIM platform to detect the actual delays. It turns out that the participant playing a central role in the network tends to overburden with heavier workloads, leading to more undesirable discrepancies and delays. As a result, extensive investigations based on process mining supports data-driven decision making to strategically smooth the construction process and increase collaboration opportunities, which also help in reducing the risk of project failure ahead of time.