A data-driven penalty-reward methodology for performance assessment of risk control systems

Abstract

The frequency and severity of industrial accidents (both process and occupational) are dependent on the effectiveness of risk control systems (RCSs). Despite its importance, the research in this domain is scarce. In this study, a data-driven decision support methodology to evaluate the performance of the RCSs, based on penalty and reward policy, is proposed. The performance influencing factors (PIFs) are identified from incident records and prioritized based on static (expert opinion) and dynamic (data-driven) weights. The static weight is computed using fuzzy best-worst method. The penalty and reward criteria are designed and they correspond to weakness (WI) and effectiveness (EI) index of RCSs, respectively. The values of WI and EI provide information about the performance of RCSs in three levels. The first level provides the overall performance of the RCSs. The second level reports the performance of the individual RCS. The third level provides information about the factors influencing the performance of the RCS. The proposed methodology is applied to process safety incidents reported in a steel manufacturing plant. It is found that the RCS “inspection and maintenance” has poorest performance. The proposed methodology can help industries to objectively assess and monitor the performance of the RCSs, identify the significant factors affecting the performance, and enable management to make informed decisions to improve the safety and safety management system.