Human Error Identification and Analysis for Shield Machine Operation Using an Adapted TRACEr Method

Abstract

This paper investigated shield machine operation (SMO) errors involved in shield tunneling construction accidents based on the Technique for the Retrospective and Predictive Analysis of Cognitive Errors (TRACEr). Human errors are classified and identified at a coarse-grained task level in the TRACEr framework, which could cause failures to completely identify and analyze the human errors in a given accident. This motivated us to propose an adapted TRACEr to overcome the limitation. The adapted TRACEr incorporates hierarchical task analysis (HTA) to decompose a task into combinations of activities, which helps describe human errors at a fine-grained activity level. The connection between the added activity level and the cognitive functions was constructed according to the Phoenix method. Based on the adaptation, an activity-oriented structure of human error taxonomies was developed, and a corresponding retrospective analysis procedure that focuses on identifying errors under various construction operational situations was proposed. Based on the adapted TRACEr, SMO errors were identified and analyzed. The error taxonomies of SMO were developed, and 72 accidents were retrospectively analyzed to identify and code the errors. Data mining techniques were applied to analyze the fine-grained SMO error data to reveal the main manifestations of SMO errors and the hidden associated rules for their cognitive failures. Consequently, several targeted cognitive-based human error mitigation strategies were proposed, showing the application potential of the adapted TRACEr as a human error management tool in the construction industry.