Probabilistic-Based Cascading Failure Approach to Assessing Workplace Hazards Affecting Human Error

Abstract

To minimize human errors, the propagation mechanics of hazards that can lead to human error must be clarified. Accordingly, the objective of this study was to quantify the effects of hazards on human errors and determine the associated propagation paths using a probabilistic-based cascading failure approach. To demonstrate this process, an interdependence network was developed using actual safety-inspection data acquired from elevator installations. A bootstrap approach was then utilized to obtain the critical capacity of each hazard in the cascading failure process. Subsequently, simulations using different hazards at different load levels were performed, and the shielding rates of the hazards were calculated to demonstrate the propagation mechanics. These results provide new insights into the denoising ability of the cascading failure approach when assessing the propagation process of safety hazards that can be activated by nontriggered antecedents. This research contributes to analysis of the propagation of safety hazards by utilizing an improved assessment approach, providing a basis for a safety-management plan that can more effectively control/rectify critical hazards in construction.