Objectives. Numerous labourers lose their lives or suffer from injuries and disabilities yearly due to the lack of safety enforcement in construction projects and accidents caused by excavation collapses. The identification and ranking of human errors have always been a central concern in civil engineering. Previous studies on excavation work and related risks have focused on retaining structure methods, while human errors may be a significant contributor to accidents and near misses. Methods. This study identified human errors in deep excavation projects using hierarchical task analysis (HTA) and a systematic human error reduction and prediction approach (SHERPA). Results. The fuzzy Bayesian human error assessment and reduction technique (HEART)-5M method was implemented to determine the human error probability (HEP) for all case-study tasks. Critical tasks were obtained as 'drainage system execution', 'water and wastewater pipes', 'gas pipes', 'checking cracks in surrounding buildings' and 'checking soil slippage' with probability levels of 0.46, 0.44, 0.44, 0.37 and 0.37, respectively. Finally, remedial measures were presented for crucial tasks. Conclusions. Six unbiased experts approved the model's desirability. The suggested approach can serve as a valuable guide for all project stakeholders in identifying, evaluating and taking corrective actions in similar projects.
Keywords: Bayesian networks; HEART method; excavation projects; fuzzy logic; human errors; human reliability analysis.