Data-theoretic methodology and computational platform for the quantification of organizational mechanisms in probabilistic risk assessment
Date
2017
Authors
Pence, J.
Sun, Y.
Zhu, X.
Mohaghegh, Z.
Ostroff, C.
Kee, E.
Editors
Advisors
Journal Title
Journal ISSN
Volume Title
Type:
Conference paper
Citation
International topical meeting on probabilistic safety assessment and analysis, PSA 2017, 2017, vol.2, pp.1294-1300
Statement of Responsibility
Conference Name
International Topical Meeting on Probabilistic Safety Assessment and Analysis, PSA 2017 (24 Sep 2017 - 28 Sep 2017 : Pittsburgh, US)
Abstract
The quantification of organizational mechanisms remains a vital challenge for Human Reliability Assessment (HRA) and Probabilistic Risk Assessment (PRA). This paper reports on the progress of ongoing research by the authors to (a) identify the social and organizational factors that affect technological system risk, (b) examine how and why these factors influence risk, and (c) specify, in conjunction with PRA, how much they contribute to risk. The authors advance the Socio-Technical Risk Analysis (SoTeRiA) framework that explicitly integrates both the social aspects (e.g., safety culture) and the structural features (e.g., safety practices) of organizations with technical system risk models. While SoTeRiA represents a significant improvement in the integration of organizational factors in PRA, the theoretical details of underlying contributing factors and causal interrelationships for each generic node should be expanded. In addition to the theoretical advancements, a major challenge of quantifying organizational factors for PRA relates to lack of reliable data. This research uses a new approach, highlighting that data is available for organizational factors; however, with a different nature from tabular numerical equipment reliability data. The nature of data for organizational factors is evolving to an 'organizational-level' of analysis, i.e., organizational communications. The Data-Theoretic methodology is a hybrid combination of theoretical conceptualization and data analytics to quantify organizational mechanisms. With this approach, the strength of human interpretation (i.e., systematic interpretation and expert opinion) is utilized along with theoretical constructs to develop adequate causality for guiding new big data analytic algorithms aimed at large unstructured and heterogeneous data of textual organizational communications. This paper highlights the two primary parts of the Data-Theoretic methodology: (1) DT-BASE: Develop factors, sub-factors, and causal relationships in SoTeRiA, equipped with a base quantification grounded on theories and interpretation of information in the literature and reports and (2) DT-SITE: Develop site-specific measurement techniques to quantify the causal models based on site-specific data, utilizing automated data extraction and inference methods
School/Discipline
Dissertation Note
Provenance
Description
Access Status
Rights
Copyright 2017 American Nuclear Society