Methodology for the analysis and quantification of human error probability in manufacturing systems.

The most serious problem for human error estimation is the scarcity of empirical data on human performance for the development and validation of Human Reliability Analysis (HRA) approaches. This issue is strongly evident in manufacturing systems, where the data collection and availability of a meaningful dataset to feed human reliability have severe constraints related to time-resource consuming and accuracy of the collection approach. This paper proposes a beginning taxonomy of human error consequences in order to support data collection in manufacturing systems and to identify experimental human error probability (HEP). This taxonomy is the first step of the methodology for HEP analysis and quantification and for validation of theoretical curves of Simulator for Human Error Probability Analysis (SHERPA) model. A new statistical methodology is able to quantify experimental HEPs, starting from the realistic human error consequences, and to compare them with the SHERPA theoretical human error distributions. This methodology, used for the SHERPA validation, may be useful for assessing current HEP estimation into HRA approaches.