PSAM 16 - Session W02 Overview

PSAM 16 Conference Session W02 Overview

Session Chair: Curtis Smith (curtis.smith@inl.gov)

Paper 1 CU58

Lead Author: Curtis Smith

Observations and Results from a Benchmark on External Events Hazard Frequency and Magnitude Statistical Modelling
The March 2011 accident at the Fukushima Daiichi nuclear power plant triggered discussions about the natural external events that are low-frequency but high-consequence. To address these issues and determine which events would benefit from international co-operative work, a Nuclear Energy Agency Task Group on Natural External Events (TGNEV) was established. In June 2014, this group reorganised into a Working Group on External Events (WGEV). WGEV is composed of a forum of experts for the exchange of information and experience on external events in member countries, thereby promoting co-operation and maintenance of an effective and efficient network of experts. A recent activity of this group has focused on conducting a benchmark on external events hazard frequency and magnitude statistical modelling. Modelling of these external events is a common practice in hazards and risk assessments found in many countries. Having a valid statistical approach to model these hazards is important. However, current practice indicates a wide variety of approaches being used and an under appreciation of the uncertainties inherent in these types of statistical models. The objective of this activity was to provide a benchmark suitable to explore the application of typical approaches to external hazard representation through a data-informed process. The activity provided two types of benchmarks, one with data and model provided and one with just data provided (a “blind- test”). This paper summarizes the statistical modelling approaches that were submitted for the benchmark and provides observations related to the approaches used.

Observations and Results from a Benchmark on External Events Hazard Frequency and Magnitude Statistical Modelling

The March 2011 accident at the Fukushima Daiichi nuclear power plant triggered discussions about the natural external events that are low-frequency but high-consequence. To address these issues and determine which events would benefit from international co-operative work, a Nuclear Energy Agency Task Group on Natural External Events (TGNEV) was established. In June 2014, this group reorganised into a Working Group on External Events (WGEV). WGEV is composed of a forum of experts for the exchange of information and experience on external events in member countries, thereby promoting co-operation and maintenance of an effective and efficient network of experts. A recent activity of this group has focused on conducting a benchmark on external events hazard frequency and magnitude statistical modelling. Modelling of these external events is a common practice in hazards and risk assessments found in many countries. Having a valid statistical approach to model these hazards is important. However, current practice indicates a wide variety of approaches being used and an under appreciation of the uncertainties inherent in these types of statistical models. The objective of this activity was to provide a benchmark suitable to explore the application of typical approaches to external hazard representation through a data-informed process. The activity provided two types of benchmarks, one with data and model provided and one with just data provided (a “blind- test”). This paper summarizes the statistical modelling approaches that were submitted for the benchmark and provides observations related to the approaches used.

Paper CU58 | | Download the presentation pdf file. Download the presentation PowerPoint file.

Name: Curtis Smith (curtis.smith@inl.gov)

Bio: Curtis Smith, Ph.D., is the Director for the Idaho National Laboratory Nuclear Safety and Regulatory Research Division. Prior to taking a leadership role as Division Director, he led several risk-informed activities for DOE, the Nuclear Regulatory Commission, and NASA. Dr. Smith has been at INL for 31 years and has published over 275 papers, books, and reports on risk and reliability theory and applications. He holds a Ph.D. in nuclear engineering from MIT.

Country: USA
Company: Idaho National Laboratory
Job Title: Division Director

Paper 2 YA220

Lead Author: Yasser Hamdi Co-author(s): Vincent Rebour - vincent.rebour@irsn.fr

OECD/NEA Benchmark: External Events Hazard Frequency and Magnitude Statistical Modelling - IRSN approach
A general feature of external hazards is that they produce off-normal conditions that can impact nuclear installations. Scenarios- and frequencies-based risk analysis allows a more reliable practice by allowing key stakeholders to make risk informed choices rather than simply relying on traditional deterministic estimates of risk, with a brief description of uncertainty. This benchmark is conducted to facilitate an exercise on the estimation of extreme external events. The aim of the benchmark is to better understand the technical aspects and processes used for the characterization of natural hazards. Data and overall objectives for the benchmarking exercise are presented for a hypothetical external event (e.g., precipitation, extreme temperatures, high winds). Our analysis steps, assumptions and modelling results were summarized. Uncertainties are generated by the fact that the provided synthetic data is relatively uninformative. The use of this synthetic data-based approach allows to evaluate the proposed statistical model and add known uncertainty to the data. Two cases are provided: (i) with a given generating model, (ii) a “blind-test case” where only the data is provided. To make the exercise as useful as possible and to present the work in the most comprehensible way, some theoretical and technical background for the selected approaches and the assumptions are provided. After recalling the data provided in the exercise, the underlying theory and assumptions are presented. The results of the proposed statistical models are then presented for each test case together with the associated uncertainties as well as the model adequacy assessment. This work is concluded with a summary of the most essential results.

OECD/NEA Benchmark: External Events Hazard Frequency and Magnitude Statistical Modelling - IRSN approach

A general feature of external hazards is that they produce off-normal conditions that can impact nuclear installations. Scenarios- and frequencies-based risk analysis allows a more reliable practice by allowing key stakeholders to make risk informed choices rather than simply relying on traditional deterministic estimates of risk, with a brief description of uncertainty. This benchmark is conducted to facilitate an exercise on the estimation of extreme external events. The aim of the benchmark is to better understand the technical aspects and processes used for the characterization of natural hazards. Data and overall objectives for the benchmarking exercise are presented for a hypothetical external event (e.g., precipitation, extreme temperatures, high winds). Our analysis steps, assumptions and modelling results were summarized. Uncertainties are generated by the fact that the provided synthetic data is relatively uninformative. The use of this synthetic data-based approach allows to evaluate the proposed statistical model and add known uncertainty to the data. Two cases are provided: (i) with a given generating model, (ii) a “blind-test case” where only the data is provided. To make the exercise as useful as possible and to present the work in the most comprehensible way, some theoretical and technical background for the selected approaches and the assumptions are provided. After recalling the data provided in the exercise, the underlying theory and assumptions are presented. The results of the proposed statistical models are then presented for each test case together with the associated uncertainties as well as the model adequacy assessment. This work is concluded with a summary of the most essential results.

Paper YA220 | Download the paper file. |

Name: Yasser Hamdi (yasser.hamdi@ymail.com)

Bio: Yasser Hamdi specialized in the analysis of the risk associated to natural hazards. He is currently working as a Research Engineer at the Institute for Radiation Protection and Nuclear Safety (IRSN) in France. He has been conducting studies and research on the assessment of the risk associated to natural hazards potentially affecting NPPs safety. He has developed several methodologies and tools dealing with a range of problems in the field of statistics of extremes and Probabilistic Flood Hazard Analysis (PFHA). He also has a great experience in the characterization of hydrometeorological and climatic hazards in a non-stationary context.

Country: FRA
Company: IRSN
Job Title: Research Ingineer

Paper 3 BE120

Lead Author: Beom-Jin Kim Co-author(s): Minkyu Kim, minkyu@kaeri.re.kr Daegi Hahm, dhahm@kaeri.re.kr

Benchmark on External Events Hazard Frequency and Magnitude Statistical Modelling in KAERI
This benchmark study aims to apply statistical modeling for frequency and magnitude estimation based on external event hazard assessment data. Based on the results of this study, it is believed that an approach to the quantification of external event initiating events (IEs) can be formulated and evaluated through the application of an effective statistical model. This study's analysis was based on two cases that considered benchmarks provided by the Organization for Economic Co-operation and Development (OECD). Each case was given a magnitude according to the return period. An appropriate statistical model was applied through regression analysis for each case based on this data. Based on the results, the magnitudes of 500, 5,000, 50,000, and 500,000 years were predicted and presented. The result of this study, statistical analysis was applied to the estimation of two cases presented by the OECD. In any statistical analysis, it is important to understand the characteristics of the data set. For the given problems here, the range of the return period was 10–10,000 years, while that of the magnitude was only 0.4–5.0 meters. Therefore, the coefficient of the synthetic model had a significant influence on the analysis results. This study demonstrates that employing the full extent of the significant figures is important to handle the different ranges of data values. In the future, it is expected that data-based statistical values can be better estimated through various verified statistical models.

Benchmark on External Events Hazard Frequency and Magnitude Statistical Modelling in KAERI

This benchmark study aims to apply statistical modeling for frequency and magnitude estimation based on external event hazard assessment data. Based on the results of this study, it is believed that an approach to the quantification of external event initiating events (IEs) can be formulated and evaluated through the application of an effective statistical model. This study's analysis was based on two cases that considered benchmarks provided by the Organization for Economic Co-operation and Development (OECD). Each case was given a magnitude according to the return period. An appropriate statistical model was applied through regression analysis for each case based on this data. Based on the results, the magnitudes of 500, 5,000, 50,000, and 500,000 years were predicted and presented. The result of this study, statistical analysis was applied to the estimation of two cases presented by the OECD. In any statistical analysis, it is important to understand the characteristics of the data set. For the given problems here, the range of the return period was 10–10,000 years, while that of the magnitude was only 0.4–5.0 meters. Therefore, the coefficient of the synthetic model had a significant influence on the analysis results. This study demonstrates that employing the full extent of the significant figures is important to handle the different ranges of data values. In the future, it is expected that data-based statistical values can be better estimated through various verified statistical models.

Paper BE120 | Download the paper file. |

Name: Beom-Jin Kim (beomjin88@kaeri.re.kr)

Bio:

Country: KOR
Company: Korea Atomic Energy Research Institute (KAERI)
Job Title: Ph.D, Postdoctoral Researcher