Human Risk Analysis
- Hits: 5872
Probabilistic risk analysis gives by default the possibility to demonstrate the interaction between human and plant and the impact to the plant’s safety. Simultaneously, it presents human in greater variability and complexity, compared to the technical components of a system. Behavior modeling proves to be a difficult task for risk analysis and is the main source of uncertainty. This applies mainly to extensive, interdepended actions with involvement of many persons or in situations where decisions or interpretations need to be made.
Due to this problematic nature, a methodology for the description and quantification of human error-behavior is required. The methods THERP nach Swain and ASEP are taken as a basis. The methodology is supplemented with the methods HCR, HCR-ORE, CBDT and FLIM. The systematic approach is supported by a set of form sheets, which ensure the complete assessment of every relevant aspect. Simultaneously, these forms are the base of a data model for an appropriate software implementation.
The analyzed human tasks are categorized according to organization (planed or unplanned), according to technical criteria (indirect operative actions, actions before an accident, actions that cause an accident, actions during an accident) and according to psychological criteria (knowledge-based actions, rule-based actions, skill-based actions). Performance shaping factors include available time, stress, difficulty/education, staff redundancy, document quality, design of human-engine interface and the job hierarchy of the person involved.
The evaluation concept includes the following steps:
• System analysis in order to identify the relative personnel actions and their involvement in the fault trees and event trees
• Fixing of a personnel action, implementing an action analysis and division into action steps and recovery possibilities
• Fixing of an action step; Classification in the categories diagnostically, rule-based and recovery; Evaluation with the proper forms
• HRA-tree creation according to the results of the previous steps
• Computer-aided analysis
• Estimation of insecureness, to the extend needed
• Fitting of results in fault and event trees