Data-driven determination of collapse accident patterns for the mitigation of safety risks at metro construction sites
Zhipeng Zhou, Yang Miang Goh, Qianqian Shi, Haonan Qi, Song Liu
Abstract
As the most frequent type of hazardous events at metro construction sites, collapse accidents are very dangerous and can bring about multiple injuries and fatalities. This study aims to conduct an accident data-driven investigation on collapses for the reduction of metro construction safety risks. A total of 178 collapse accident cases were collected in the intervening period from 1996 to 2021. Three categories of contributing factors related to time, event and causation were compiled on the basis of textual description of every individual collapse. Cramer’s V and Phi coefficients were analyzed to uncover statistically significant correlations between time-related factors, event-related factors, and causation-related factors. These results provided opportunities for discerning multiple collapse accident patterns from three perspectives of scenario, consequence and causation. Owing to different occurrence process across eleven types of metro construction collapses, specific strategies and countermeasures were devised and proposed for safety management practice for controlling and mitigating collapse risk in metro construction.