PSAM 16 - Session F02 Overview

PSAM 16 Conference Session F02 Overview

Paper 1 BE100

Lead Author: Marco Behrendt Co-author(s): Matthias G. R. Faes, matthias.faes@kuleuven.be Marcos A. Valdebenito, marcos.valdebenito@uai.cl Michael Beer, beer@irz.uni-hannover.de

Capturing epistemic uncertainties in the power spectral density for limited data sets
In stochastic dynamics, it is indispensable to model environmental processes in order to design structures safely or to determine the reliability of existing structures. Wind or earthquake loads are examples of these environmental processes and may be described by stochastic processes. This type of process can be characterised by means of the power spectral density (PSD) function in the frequency domain. With the PSD function governing frequencies and their amplitudes can be determined. For the reliable generation of such a load model described by a PSD function, the uncertainties that occur in time signals must be taken into account. In this paper, an approach is presented to derive an imprecise PSD model from a limited amount of data. The spectral densities at each frequency are described by intervals instead of relying on discrete values. The advantages of the imprecise PSD model are illustrated and validated with numerical examples in the field of stochastic dynamics.

Capturing epistemic uncertainties in the power spectral density for limited data sets

In stochastic dynamics, it is indispensable to model environmental processes in order to design structures safely or to determine the reliability of existing structures. Wind or earthquake loads are examples of these environmental processes and may be described by stochastic processes. This type of process can be characterised by means of the power spectral density (PSD) function in the frequency domain. With the PSD function governing frequencies and their amplitudes can be determined. For the reliable generation of such a load model described by a PSD function, the uncertainties that occur in time signals must be taken into account. In this paper, an approach is presented to derive an imprecise PSD model from a limited amount of data. The spectral densities at each frequency are described by intervals instead of relying on discrete values. The advantages of the imprecise PSD model are illustrated and validated with numerical examples in the field of stochastic dynamics.

Paper BE100 | Download the paper file. | Download the presentation PowerPoint file.

Name: Marco Behrendt (behrendt@irz.uni-hannover.de)

Bio:

Country: DEU
Company: Institute for Risk and Reliability, Leibniz Universität Hannover
Job Title: Research assistant

Paper 2 BI52

Lead Author: Marius Bittner Co-author(s): Marco Behrendt behrendt@irz.uni-hannover.de Jasper Behrensdorf behrensdorf@irz.uni-hannover.de Michael Beer beer@irz.uni-hannover.de

Epistemic uncertainty quantification of localised seismic power spectral densities
The modelling and quantification of seismic loadings such as earthquakes to improve the safe design of structures is a challenging task. In particular, the unpredictable nature of earthquake characteristics like amplitude, dominant frequencies, and duration pose a great risk especially for sensitive structures like power plants, oil rigs, high-rise buildings, or large-span structures. The analysis, understanding and evaluation of those seismic characteristics and their influence on safe structural design is especially important for regions prone to earthquakes. The tectonic mechanisms leading to seismic underground waves are complex but measurements of earthquakes and their mechanical causes on surfaces are available manifold. A new procedure is presented herein for describing uncertainties in the power spectral density (PSD) function of seismic loadings and utilises the novel approach of Sliced-Normal distributions to describe multivariate probability density functions over frequency and amplitude. This representation enables analysts of stochastic dynamic systems the usage of a compact description for PSD functions and to reduce epistemic uncertainties on specific regions. This newly formed PSD function can be used in the simulation of seismic loads via spectral representation or other spectral-based stochastic process generators and is a subsequent development of the already introduced relaxed PSD function.

Epistemic uncertainty quantification of localised seismic power spectral densities

The modelling and quantification of seismic loadings such as earthquakes to improve the safe design of structures is a challenging task. In particular, the unpredictable nature of earthquake characteristics like amplitude, dominant frequencies, and duration pose a great risk especially for sensitive structures like power plants, oil rigs, high-rise buildings, or large-span structures. The analysis, understanding and evaluation of those seismic characteristics and their influence on safe structural design is especially important for regions prone to earthquakes. The tectonic mechanisms leading to seismic underground waves are complex but measurements of earthquakes and their mechanical causes on surfaces are available manifold. A new procedure is presented herein for describing uncertainties in the power spectral density (PSD) function of seismic loadings and utilises the novel approach of Sliced-Normal distributions to describe multivariate probability density functions over frequency and amplitude. This representation enables analysts of stochastic dynamic systems the usage of a compact description for PSD functions and to reduce epistemic uncertainties on specific regions. This newly formed PSD function can be used in the simulation of seismic loads via spectral representation or other spectral-based stochastic process generators and is a subsequent development of the already introduced relaxed PSD function.

Paper BI52 | Download the paper file. | Download the presentation PowerPoint file.

Name: Marius Bittner (bittner@irz.uni-hannover.de)

Bio: Graduated 2018 with a M.Sc. in “Computational Methods in Engineering” at the University of Hannover. During the master absolved a semester abroad at the International Joint Research Center for Engineering Reliability and Stochastic Mechanics, Tongji University, China. Since 2019 working at the Institute for Risk and Reliability at the University of Hannover as research assistant and pursuing a PhD in “Efficient reliability analysis for complex high dimensional stochastic dynamic systems”. Since 2021 joined the International Research Training Group 2657 “Computational Mechanics Techniques in High Dimensions” hosted by the University of Hannover and the ENS Paris-Saclay.

Country: DEU
Company: Institute of Risk and Reliability, Leibniz University Hannover
Job Title: Research assistant

Paper 3 DY330

Lead Author: Gyunseob Song Co-author(s): Man Cheol Kim - charleskim@cau.ac.kr

An estimation method for heat pipe cascading failure frequency for micro modular reactor PSA
Initiating event analysis is one of essential elements in probabilistic safety assessment (PSA) to estimate core damage frequency (CDF) or large early release frequency (LERF) as risk metrics. As several types of reactors have been developed which are so called Generation Ⅳ reactor, it is expected that several types of initiating events which are not considered in traditional reactors exist. The frequency of an initiating event is generally estimated using historical data from operating reactors. However, there is no observed experience about expected initiating events for developmental stage reactors and hence methodologies to estimate frequencies of expected initiating events should be developed. In Generation Ⅳ reactors, heat pipe is widely considered as heat removal system because of its passive property. Even if a heat pipe is designed to have enough capability in normal operating conditions, the performance of a heat pipe may depend on heat load. Therefore, it is possible that individual failures of heat pipes cause cascading failure of other heat pipes due to the increase in heat load resulted from failed heat pipes. When the cascading failure of heat pipes occurs, the heat removal capability of the system decreases and hence the integrity of the reactor can be threatened. Therefore, the analysis should be performed to estimate occurrence frequency of heat pipe cascading failure. Previous researches on estimating cascading failure frequency of heat pipes mainly use Monte Carlo analysis based on defined failure probability within a time step which depends on increased heat load due to failure of adjacent heat pipes. However, the time step size is restricted by computational cost and hence the defined failure probability may be underestimated because the failure of adjacent heat pipe cannot be instantly reflected. Furthermore, the frequency is simply estimated as a point value, the ratio of the number of cascading failures to the number of simulations. In this research, we model the failures of heat pipes as a stochastic process and approximate the analytic distribution for occurrence time of heat pipe cascading failure. Then, the estimated frequency of heat pipe cascading failure event is mathematically calculated.

An estimation method for heat pipe cascading failure frequency for micro modular reactor PSA

Initiating event analysis is one of essential elements in probabilistic safety assessment (PSA) to estimate core damage frequency (CDF) or large early release frequency (LERF) as risk metrics. As several types of reactors have been developed which are so called Generation Ⅳ reactor, it is expected that several types of initiating events which are not considered in traditional reactors exist. The frequency of an initiating event is generally estimated using historical data from operating reactors. However, there is no observed experience about expected initiating events for developmental stage reactors and hence methodologies to estimate frequencies of expected initiating events should be developed. In Generation Ⅳ reactors, heat pipe is widely considered as heat removal system because of its passive property. Even if a heat pipe is designed to have enough capability in normal operating conditions, the performance of a heat pipe may depend on heat load. Therefore, it is possible that individual failures of heat pipes cause cascading failure of other heat pipes due to the increase in heat load resulted from failed heat pipes. When the cascading failure of heat pipes occurs, the heat removal capability of the system decreases and hence the integrity of the reactor can be threatened. Therefore, the analysis should be performed to estimate occurrence frequency of heat pipe cascading failure. Previous researches on estimating cascading failure frequency of heat pipes mainly use Monte Carlo analysis based on defined failure probability within a time step which depends on increased heat load due to failure of adjacent heat pipes. However, the time step size is restricted by computational cost and hence the defined failure probability may be underestimated because the failure of adjacent heat pipe cannot be instantly reflected. Furthermore, the frequency is simply estimated as a point value, the ratio of the number of cascading failures to the number of simulations. In this research, we model the failures of heat pipes as a stochastic process and approximate the analytic distribution for occurrence time of heat pipe cascading failure. Then, the estimated frequency of heat pipe cascading failure event is mathematically calculated.

Paper DY330 | |

Name: Gyunseob Song (dyrnfmxm678@cau.ac.kr)

Bio:

Country: KOR
Company: Chung-Ang university
Job Title: