100 Framework for the Human Event Repository and Analysis (HERA) System and Its Use to Quantify Human Actions (PSAM-0224)
-
Published:2006
Download citation file:
This paper describes the development and current structure of the Idaho National Laboratory's (INL's) Human Event Repository and Analysis (HERA) system. HERA is a database, worksheet, and related process system for obtaining, compiling, and classifying information from operational experience in nuclear power plants (NPPs) and making that information available in content and form useful to the variety of human reliability analysis (HRA) methods and the human factors discipline in practice.
The weakness of data available for HRA is one of the biggest concerns expressed by practitioners and decision makers. The validity of quantification of human failure event (HFE) probabilities and the development and validation of human performance models used in HRA stand on the footing of the data at their disposal.
Recognizing the many differences between HRA methods, including types of inferences and explanations of human behavior, a goal of the HERA system since its inception has been to provide information designed and qualified to be of value to most methods. The sources of information include both raw, unprocessed information of source documents and additional information related to underlying human performance mechanisms in terms that can be applied directly or easily transformed to support implementation of a variety of HRA methods. The taxonomy and structure of HERA is, thus, designed to accept a variety of activities and to support numerous HRA method implementations.
Beyond supporting the development of insight from observations, the HERA project is working on developing approaches to use observational data to support estimation of quantitative risk metrics for use in PRA. Bayesian methods that have been used widely in estimating risk metrics for equipment studied in PRA are also being considered for use with HERA data. Bayesian methods provide the ability to use data that may be relevant though incomplete. Much of the data that is now or will be soon available for use in HRA may be used through some form of the Bayesian framework, though much is still needed to address important assumptions about human performance and its representation through the Bayesian framework.