PSAM 16 - Session M02 Overview

PSAM 16 Conference Session M02 Overview

Session Chair: Johan Sorman (johan.sorman@lr.org )

Paper 1 AG46

Lead Author: Andrei Gribok Co-author(s): Curtis L. Smith curtis.smith@inl.gov

Support Vector Analysis for Computational Risk Assessment, Decision Making, and Vulnerability Discovery in Complex Systems
A primary limitation of modern probabilistic risk assessment (PRA) is that, since the risk scenarios and system vulnerabilities are manually developed by analysts, they critically depend on the analysts’ qualifications, available information about the system, and ability to understand and “discover” the system vulnerabilities (as well as to properly describe them using Boolean logic). In other words, modern PRA is a method of documenting analysts’ discoveries rather than suggesting new, previously unknown risks. The paper describes a method for auto-detecting possible vulnerabilities in system designs, thus revealing previously unseen issues and reducing human error/costs by enabling analysts to focus on critical areas via intelligent, efficient sampling of the system’s parameter space. For existing systems with available fault trees, we developed the proof-of-principle methodology, allowing the proposed novel methodology first stochastically generates large volumes of training data by “rewiring” fault trees of the target system and then learning the most important features of the training data. Rewiring includes randomly changing gate logic and the occurrence of fundamental events (i.e., basic or initiating events) in a fault tree. By rewiring existing target datasets, the training data are skewed toward existing systems, yet still provide the variation needed by generating millions of training examples. The combination of fault tree logic and Boolean variables representing initiating events can be regarded as a configuration usable as input vector for support vector machine (SVM) training. During training, along with classifying each input data vector, the SVM algorithm finds support vectors (SVs) in the training data. By the very nature of the training algorithm, SVM focuses only on those points that are most difficult to tell apart. Because in our case the points are realizations of fault trees, SVM discovers the most similar fault trees from both classes, thus also pointing to the most “vulnerable” configurations. The SVs are the most important input data for separating the two classes (i.e., failure vs. non-failure), and, most notably, represent only a very small portion of the input data. Because in this case the input data are fault tree characterizations, the SVM training produces system configurations that are the “borderline” between failure and non-failure scenarios. These support trees are further scrutinized for insights into the system’s logical vulnerabilities and risks. The primary outcome from this research is a new, broadly applicable methodology in which intelligently guided space sampling methods are used to drastically reduce the number of system configurations needing analyzed. This methodology will enable researchers to auto-detect possible vulnerabilities in system designs, devices, and networks, thus revealing previously unseen issues and reducing human error/costs by enabling analysts to focus on critical areas via intelligent, efficient sampling of the system’s parameter space.

Support Vector Analysis for Computational Risk Assessment, Decision Making, and Vulnerability Discovery in Complex Systems

A primary limitation of modern probabilistic risk assessment (PRA) is that, since the risk scenarios and system vulnerabilities are manually developed by analysts, they critically depend on the analysts’ qualifications, available information about the system, and ability to understand and “discover” the system vulnerabilities (as well as to properly describe them using Boolean logic). In other words, modern PRA is a method of documenting analysts’ discoveries rather than suggesting new, previously unknown risks. The paper describes a method for auto-detecting possible vulnerabilities in system designs, thus revealing previously unseen issues and reducing human error/costs by enabling analysts to focus on critical areas via intelligent, efficient sampling of the system’s parameter space. For existing systems with available fault trees, we developed the proof-of-principle methodology, allowing the proposed novel methodology first stochastically generates large volumes of training data by “rewiring” fault trees of the target system and then learning the most important features of the training data. Rewiring includes randomly changing gate logic and the occurrence of fundamental events (i.e., basic or initiating events) in a fault tree. By rewiring existing target datasets, the training data are skewed toward existing systems, yet still provide the variation needed by generating millions of training examples. The combination of fault tree logic and Boolean variables representing initiating events can be regarded as a configuration usable as input vector for support vector machine (SVM) training. During training, along with classifying each input data vector, the SVM algorithm finds support vectors (SVs) in the training data. By the very nature of the training algorithm, SVM focuses only on those points that are most difficult to tell apart. Because in our case the points are realizations of fault trees, SVM discovers the most similar fault trees from both classes, thus also pointing to the most “vulnerable” configurations. The SVs are the most important input data for separating the two classes (i.e., failure vs. non-failure), and, most notably, represent only a very small portion of the input data. Because in this case the input data are fault tree characterizations, the SVM training produces system configurations that are the “borderline” between failure and non-failure scenarios. These support trees are further scrutinized for insights into the system’s logical vulnerabilities and risks. The primary outcome from this research is a new, broadly applicable methodology in which intelligently guided space sampling methods are used to drastically reduce the number of system configurations needing analyzed. This methodology will enable researchers to auto-detect possible vulnerabilities in system designs, devices, and networks, thus revealing previously unseen issues and reducing human error/costs by enabling analysts to focus on critical areas via intelligent, efficient sampling of the system’s parameter space.

Paper AG46 | Download the paper file. | Download the presentation PowerPoint file.

Name: Andrei Gribok (agribok@utk.edu)

Bio:

Country: USA
Company: Idaho National Laboratory
Job Title: Distinguished Researcher

Paper 2 LA260

Lead Author: Lavínia Maria Mendes Araújo Co-author(s): Isis Didier Lins - isis.lins@ufpe.br, Diego Andrés Aichele Figueroa - diego.aichele@ufpe.br, Caio Bezerra Souto Maior - caio.maior@ufpe.br, Marcio das Chagas Moura - marcio.cmoura@ufpe.br, Enrique Lopez Droguett - eald@g.ucla.edu

A REVIEW OF QUANTUM(-INSPIRED) OPTIMIZATION METHODS FOR SYSTEM RELIABILITY PROBLEMS
Many industrial systems demand equipment with high levels of reliability. Companies and academia have been developing, over the years, mathematical methods and advancing engineering techniques to assist in the maintenance of active and reliable systems. There are different optimization problems in this context, highlighting (1) the redundancy allocation problem (RAP), (2) the reliability allocation problem, and (3) the reliability-redundancy allocation problem (RRAP). Multiple, frequently conflicting objectives are included in system reliability design challenges. Yet, a few are universal, such as maximizing reliability and minimizing costs. Many solving methods have already been applied to these problems, e.g., dynamic, linear, integer, and nonlinear programming, as well as classical metaheuristics based on evolutionary algorithms, such as the Genetic Algorithm (GA). Either way, these methods are modeled according to the specificities of the systems. However, these approaches can be very computationally expensive depending on the problem instances. Meanwhile, quantum computing has gained ground for combinatorial optimization problems. The problems are usually remodeled into the Quadratic Unconstrained Binary Optimization (QUBO) form and are optimized using quantum methods such as the Quantum Approximate Optimization Algorithm (QAOA). It is expected that problems with a high level of complexity can be solved more efficiently using these new methods than classical ones. Optimization methods have attempted to improve their efficiency by adding quantum concepts, as is the case of Quantum-inspired Evolutionary Algorithms (QEA). The QEA has a better diversity and convergence rate compared to other EAs because it uses qubit representation instead of numerical, binary, or symbolic representations. In this context, this paper aims to develop a systematic review of the literature through keyword filtering, article reading, and bibliometric analysis on the application of purely quantum and quantum-inspired methods in system reliability optimization problems, specifically in RAP, the reliability allocation problem, and the RRAP. Our goal is to identify quantum-based techniques’ advantages, limitations, and potential in such a context and suggest a research plan based on the observed literature gaps.

A REVIEW OF QUANTUM(-INSPIRED) OPTIMIZATION METHODS FOR SYSTEM RELIABILITY PROBLEMS

Many industrial systems demand equipment with high levels of reliability. Companies and academia have been developing, over the years, mathematical methods and advancing engineering techniques to assist in the maintenance of active and reliable systems. There are different optimization problems in this context, highlighting (1) the redundancy allocation problem (RAP), (2) the reliability allocation problem, and (3) the reliability-redundancy allocation problem (RRAP). Multiple, frequently conflicting objectives are included in system reliability design challenges. Yet, a few are universal, such as maximizing reliability and minimizing costs. Many solving methods have already been applied to these problems, e.g., dynamic, linear, integer, and nonlinear programming, as well as classical metaheuristics based on evolutionary algorithms, such as the Genetic Algorithm (GA). Either way, these methods are modeled according to the specificities of the systems. However, these approaches can be very computationally expensive depending on the problem instances. Meanwhile, quantum computing has gained ground for combinatorial optimization problems. The problems are usually remodeled into the Quadratic Unconstrained Binary Optimization (QUBO) form and are optimized using quantum methods such as the Quantum Approximate Optimization Algorithm (QAOA). It is expected that problems with a high level of complexity can be solved more efficiently using these new methods than classical ones. Optimization methods have attempted to improve their efficiency by adding quantum concepts, as is the case of Quantum-inspired Evolutionary Algorithms (QEA). The QEA has a better diversity and convergence rate compared to other EAs because it uses qubit representation instead of numerical, binary, or symbolic representations. In this context, this paper aims to develop a systematic review of the literature through keyword filtering, article reading, and bibliometric analysis on the application of purely quantum and quantum-inspired methods in system reliability optimization problems, specifically in RAP, the reliability allocation problem, and the RRAP. Our goal is to identify quantum-based techniques’ advantages, limitations, and potential in such a context and suggest a research plan based on the observed literature gaps.

Paper LA260 | Download the paper file. | Download the presentation PowerPoint file.

A PSAM Profile is not yet available for this author.

Paper 3 AN79

Lead Author: Anders Olsson Co-author(s): Francesco Di Dedda (francesco.didedda@vysusgroup.com) Lovisa Nordlöf (lovisa.nordlof@okg.uniper.energy)

Availability and reliability analysis of Independent Core Cooling at Oskarshamn 3
In unit 3 at Oskarshamn NPP an independent core cooling function has been installed (OBH). Besides incorporating the OBH function in the PSA, an analysis has also been performed of its availability and reliability. As the OBH function is designed to operate during long time periods and under severe conditions, the mission time was extended to 72 hours. Even though a “standard PSA” is supposed to be a realistic assessment, compromise is necessary in terms of the assumptions and simplifications, which may or may not contribute to the results of the PSA. Such assumptions and simplifications are of course an important aspect of the uncertainties. When a single system or function is analysed, the importance of these may be more significant. A crucial part of the analysis was therefore to identify and reduce embedded conservatisms in the PSA. As a prolonged mission time was studied, another important aspect covered in the analysis was the repair of failed components. As the analysis covered both power operation mode and shutdown, different conditions for conducting repair were considered. The presentation will focus on challenges encountered and how they were handled including how the results were evaluated, i.e. was it possible to demonstrate that the OBH function is as reliable as expected.

Availability and reliability analysis of Independent Core Cooling at Oskarshamn 3

In unit 3 at Oskarshamn NPP an independent core cooling function has been installed (OBH). Besides incorporating the OBH function in the PSA, an analysis has also been performed of its availability and reliability. As the OBH function is designed to operate during long time periods and under severe conditions, the mission time was extended to 72 hours. Even though a “standard PSA” is supposed to be a realistic assessment, compromise is necessary in terms of the assumptions and simplifications, which may or may not contribute to the results of the PSA. Such assumptions and simplifications are of course an important aspect of the uncertainties. When a single system or function is analysed, the importance of these may be more significant. A crucial part of the analysis was therefore to identify and reduce embedded conservatisms in the PSA. As a prolonged mission time was studied, another important aspect covered in the analysis was the repair of failed components. As the analysis covered both power operation mode and shutdown, different conditions for conducting repair were considered. The presentation will focus on challenges encountered and how they were handled including how the results were evaluated, i.e. was it possible to demonstrate that the OBH function is as reliable as expected.

Paper AN79 | Download the paper file. | Download the presentation PowerPoint file.

Name: Anders Olsson (anders.olsson@vysusgroup.com)

Bio: Anders Olsson holds a master’s degree in Mechanical Engineering and has since 1995 been working in the nuclear industry. He started at ABB Atom where he mainly performed various thermal hydraulic analysis and worked with structural verification. Since 1999 his main focus has been Probabilistic Risk Assessment where he now has extensive experience in PSA Level 1 and 2 for all operating modes including Human Reliability Analysis. He also holds a position as Vice President in Vysus Group with responsibility for the operation of the nuclear consultancy in Sweden.

Country: SWE
Company: Vysus Group
Job Title: Vice President Nuclear

Paper 4 EM255

Lead Author: Enrique Meléndez Co-author(s): Miguel Sánchez-Perea, msp@csn.es César Queral, cesar.queral@upm.es Marcos Cabezas Sergio Courtin, sergio.courtin@upm.es Rafael Iglesias Julia Herrero-Otero Alberto Garcia-Herranz Carlos París

Standardized Probabilistic Safety Assessment Models: Applications of SPAR-CSN Project
The regulatory activity requires the oversight of licensee performance to be made from an independent position. This position is better served when the regulatory body develops its own methodologies and tools. In particular in the matter of probabilistic risk analysis, even if the licensees’ analyses are subject to peer-review and/or are reviewed by the regulatory body, it is very difficult to manage the large amount of hypothesis and assumptions behind the model. Thus, the development of a PRA model for regulatory use improves the knowledge of the NPP risks and can be seen as an enhancement of the regulatory practice. On this regard, the Spanish Regulatory Body (CSN), in collaboration with the Universidad Politécnica de Madrid (UPM), has developed its own generic standardized model (SPAR-CSN) for 3-loop PWR-WEC designs. The present paper shows an example of the application of the model in an event occurred in a nuclear power plant.

Standardized Probabilistic Safety Assessment Models: Applications of SPAR-CSN Project

The regulatory activity requires the oversight of licensee performance to be made from an independent position. This position is better served when the regulatory body develops its own methodologies and tools. In particular in the matter of probabilistic risk analysis, even if the licensees’ analyses are subject to peer-review and/or are reviewed by the regulatory body, it is very difficult to manage the large amount of hypothesis and assumptions behind the model. Thus, the development of a PRA model for regulatory use improves the knowledge of the NPP risks and can be seen as an enhancement of the regulatory practice. On this regard, the Spanish Regulatory Body (CSN), in collaboration with the Universidad Politécnica de Madrid (UPM), has developed its own generic standardized model (SPAR-CSN) for 3-loop PWR-WEC designs. The present paper shows an example of the application of the model in an event occurred in a nuclear power plant.

Paper EM255 | Download the paper file. | Download the presentation pdf file.

Name: Enrique Meléndez (ema@csn.es)

Bio: Enrique has a degree in Mathematics and a second degree in Physics from the Complutense University in Madrid. At CSN, after an initial involvement with transient and severe accident simulation, he has tackled PRA methods, including PSA model reviews and assessment, performing Accident Precursor Analysis for incidents occurred at Spanish plants, overseeing the development and application of an MSPI indicator for the Spanish regulatory system, and has developed methods and applications for dynamic PSA analyses. He is now project manager for the development of a CSN-sponsored, simplified PRA model for a Spanish NPP.

Country: ESP
Company: CSN
Job Title: Technical Advisor