Data-theoretic methodology and computational platform to quantify organizational factors in socio-technical risk analysis


Pence J., Sakurahara T., Zhu X., Mohaghegh Z., ERTEM M. , Ostroff C., ...More

RELIABILITY ENGINEERING & SYSTEM SAFETY, vol.185, pp.240-260, 2019 (Journal Indexed in SCI) identifier identifier

  • Publication Type: Article / Article
  • Volume: 185
  • Publication Date: 2019
  • Doi Number: 10.1016/j.ress.2018.12.020
  • Title of Journal : RELIABILITY ENGINEERING & SYSTEM SAFETY
  • Page Numbers: pp.240-260

Abstract

Organizational factors, as literature indicates, are significant contributors to risk in high-consequence industries. Therefore, building a theoretical framework equipped with reliable modeling techniques and data analytics to quantify the influence of organizational performance on risk scenarios is important for improving realism in Probabilistic Risk Assessment (PRA). The Socio-Technical Risk Analysis (SoTeRiA) framework theoretically connects the structural (e.g., safety practices) and behavioral (e.g., safety culture) aspects of an organization with PRA. An Integrated PRA (I-PRA) methodological framework is introduced to operationalize SoTeRiA in order to quantify the incorporation of underlying organizational failure mechanisms into risk scenarios. This research focuses on the Data-Theoretic module of I-PRA, which has two sub-modules: (i) DT-BASE: developing detailed causal relationships in SoTeRiA, grounded on theories and equipped with a semi-automated baseline quantification utilizing information extracted from academic articles, industry procedures, and regulatory standards, and (ii) DT-SITE: conducting automated data extraction and inference methods to quantify SoTeRiA causal elements based on site-specific event databases and by Bayesian updating of the DT-BASE baseline quantification. A case study demonstrates the quantification of a nuclear power plant's organizational "training" causal model, which is associated with the training/experience in Human Reliability Analysis, along with a sensitivity analysis to identify critical factors.