Files in this item

FilesDescriptionFormat

application/pdf

application/pdfOSHEA-THESIS-2017.pdf (11MB)
(no description provided)PDF

Description

Title:Spatio-temporal probabilistic methodology to estimate location-specific loss-of-coolant accident frequencies for risk-informed analysis of nuclear power plants
Author(s):O'Shea, Nicholas William
Advisor(s):Mohaghegh, Zahra
Contributor(s):Uddin, Rizwan
Department / Program:Nuclear, Plasma, & Rad Engr
Discipline:Nuclear, Plasma, Radiolgc Engr
Degree Granting Institution:University of Illinois at Urbana-Champaign
Degree:M.S.
Genre:Thesis
Subject(s):Probabilistic
Risk Analysis
Spatio-Temporal
Loss-of-Coolant Accident Frequencies
Abstract:The United States Nuclear Regulatory Commission (NRC) has promoted the use of Probabilistic Risk Assessment (PRA) in nuclear regulatory activities. Since loss-of-coolant accidents (LOCAs) are critical initiating events for many PRA applications, the NRC has taken steps towards the quantification of LOCA frequencies for use in risk-informed applications. This research develops the Spatio-Temporal Probabilistic methodology to explicitly incorporate the underlying physics of failure mechanisms into the location-specific estimation of LOCA frequencies that are required for risk-informed regulatory applications such as risk-informed resolution of generic Safety Issue 191 (GSI-191). The essence of the risk-informed resolution of GSI-191 is that location-specific LOCA frequencies drive the risk. The most recent NRC-sponsored estimations of LOCA frequencies were developed through an expert elicitation approach, provided in NUREG-1829. These estimations provided an implicit incorporation of underlying physics, space, and time. In support of the South Texas Project Nuclear Operating Company (STPNOC) risk-informed pilot project to resolve GSI-191, Fleming and Lydell developed a study which laid the groundwork for the location-specific estimations of LOCA frequencies. This research performs a critical review and a step-by-step quantitative verification of Fleming and Lydell’s methodology and, thus, two key methodological gaps are identified: (a) lack of inclusion of non-piping reactor coolant system components, and (b) lack of explicit incorporation of the underlying physics of failure that lead to the occurrence of a LOCA. To address these gaps, first, this research qualitatively examines the significance of including the contributions of non-piping components into the estimations of LOCA frequencies by conducting industry-academia evidence seeking and screening processes. Then, the Spatio-Temporal Probabilistic methodology is developed that can be used to quantitatively compare non-piping and piping components with respect to LOCA frequencies. The proposed Spatio-Temporal Probabilistic methodology also integrates the following two types of modeling: (1) The Markov modeling technique to depict the renewal processes of components’ repair due to periodic maintenance after degradations; (2) Probabilistic Physics of failure (PPoF) models to explicitly incorporate the failure mechanisms, associated with the location and age of components, into the estimation of LOCA frequencies. PPoF models integrate the underlying mechanisms related to degradation into the Markov modeling technique and, subsequently, into location-specific LOCA frequency estimations. In most of Markov models developed in this area of research, transition rates are developed using solely data-driven approaches and utilizing service data. The main problems with the Markov models with the solely data-driven transition rates are (1) inaccuracy due to insufficient data and (2) the lack of explicit connections with location-specific physics of failure mechanisms associated with transition rates. There is only one existing research that combines the Markov modeling technique with a stress-strength model of erosion corrosion for the piping components of Pressurized Heavy Water Reactors (PHWR); however due to the underlying assumptions of the methodology, this study does not adequately provide explicit incorporation of physical factors associated with locations. The Spatio-Temporal probabilistic methodology is the first research that combines the Markov technique with PPoF models for LOCA frequency estimations and, has four key tasks including:  Task #1: Defining Markov States of Degradation  Task #2: Modeling and Quantification of the Transition Rates of Degradation o Task # 2.1: Developing and quantifying physics of failure causal models o Task #2.2: Propagating uncertainties in the physics of failure causal models to develop Probabilistic Physics of failure (PPoF) models o Task #2.3: Calculating transition rates of degradation based on the output of Probabilistic Physics of failure (PPoF) models o Task # 2.4: Bayesian integration of the estimated transition rate from PPoF models and the ones from solely data-oriented approaches  Task #3: Modeling and Quantification of the Transition Rates of Repair  Task #4: Developing the Time-dependent Distributions of State Probabilities The Spatio-Temporal Probabilistic methodology provides the possibility for explicitly including the effects of location-specific causal factors, such as operating conditions (e.g., temperature, pressure, pH), maintenance quality, and material properties (e.g., yield strength and corrosion resistance) on the probability of LOCA occurrence. This methodology is beneficial, not only for estimation of location-specific LOCA frequencies, but also for incorporation of spatio-temporal physics of failure into Probabilistic Risk Assessment (PRA); therefore, it helps advance risk estimation and risk prevention. The explicit incorporation of failure mechanisms helps more accurately estimate the likelihood of LOCA occurrences, dealing with limited historical data. Additionally, the explicit incorporation of the causal factors enables the use of sensitivity analyses, which allow the physical causal factors to be ranked in order of their risk significance. Ranking of causal factors helps optimize maintenance practices by indicating the most resource-efficient methods to reduce risks. To show the feasibility, the spatio-temporal probabilistic methodology is implemented to examine the effects of Stress Corrosion Cracking (SCC) on the rupture probability of steam generator tubes. This case study demonstrates the comparative capabilities of the methodology by showing the variation in rupture probability based on the selection of Stainless Steel and Alloy 690 materials for fabrication of the expansion-transition region of the steam generator tubes. Although the tasks in this case study are explained based on SCC, which is a dominant mechanism associated with LOCA in nuclear power plants, the Spatio-Temporal Probabilistic methodology can be applied for other failure mechanisms (e.g., wear, creep) and for any high-consequence industry that deals with containment of flowing liquids or gases, such as the oil and gas industry.
Issue Date:2017-04-11
Type:Thesis
URI:http://hdl.handle.net/2142/97327
Rights Information:Copyright 2017 Nicholas O'Shea
Date Available in IDEALS:2017-08-10
Date Deposited:2017-05


This item appears in the following Collection(s)

Item Statistics