TY - JOUR
T1 - Data-theoretic methodology and computational platform to quantify organizational factors in socio-technical risk analysis
AU - Pence, Justin
AU - Sakurahara, Tatsuya
AU - Zhu, Xuefeng
AU - Mohaghegh, Zahra
AU - Ertem, Mehmet
AU - Ostroff, Cheri
AU - Kee, Ernie
N1 - This material is based on work supported by the National Science Foundation (NSF) under Grant No. 1535167 . Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation . The authors would like to thank all members of the Socio-Technical Risk Analysis (SoTeRiA) Laboratory ( http://soteria.npre.illinois.edu/ ) for their feedback, and especially appreciate the support from Dr. Seyed Reihani, Postdoctoral Research Associates Pegah Farshadmanesh, and Sai Zhang, graduate student Yicheng Sun, and undergraduates Lauren Roby and Nalin Gadihoke. The authors appreciate Dr. Vinh N. Dang from the Risk and Human Reliability Group at the Paul Scherrer Institute (PSI) for his feedback on the initial stage of this research published in Ref. [81] . The authors would also like to thank Russell Hubenak and Fatma Yilmaz from South Texas Project Nuclear Operating Company for their critical review of the training causal model. The DT-BASE computational platform is implementing pgmpy, an opensource Python library for probabilistic graphical model analysis ( https://github.com/pgmpy/pgmpy ). Bayesian is inference performed using OpenBUGS ( http://www.openbugs.net ). Bayesian Belief Network (BBN) is supported by GeNIe Modeler, BayesFusion, LLC, http://www.bayesfusion.com/ .
PY - 2019/5
Y1 - 2019/5
N2 - 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.
AB - 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.
KW - Big data analytics
KW - Causal modeling
KW - Human Reliability Analysis (HRA)
KW - Organizational factors
KW - Probabilistic Risk Assessment (PRA)
KW - Text mining
UR - http://www.scopus.com/inward/record.url?scp=85059217308&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85059217308&partnerID=8YFLogxK
U2 - 10.1016/j.ress.2018.12.020
DO - 10.1016/j.ress.2018.12.020
M3 - Article
AN - SCOPUS:85059217308
SN - 0951-8320
VL - 185
SP - 240
EP - 260
JO - Reliability Engineering and System Safety
JF - Reliability Engineering and System Safety
ER -