PSAM 16 - Session M04 Overview

PSAM 16 Conference Session M04 Overview

Session Chair: Yong-Joon Choi (yong-joon.choi@inl.gov)

Paper 1 SO331

Lead Author: Sotaro Kawanishi Co-author(s): Kazuo Furuta (furuta@rerc.t.u-tokyo.ac.jp) Taro Kanno (tkanno@g.ecc.u-tokyo.ac.jp)

A modeling framework for assessing resilience of urban infrastructure systems considering multiple interdependency and uncertainty
1. Introduction Damage to infrastructure during disasters is a major challenge for countries around the world, and various kinds of measures such as seismic retrofitting of infrastructure facilities are taken in place to mitigate it. Simulation-based evaluations are commonly used to determine the most appropriate measure to choose from, and various modeling frameworks have been developed. In many modeling frameworks, the focus is on reproducing infrastructures precisely, but as urban systems are complex and composed of a variety of interacting elements, it is necessary to include not only physical lifeline infrastructures but also important human activities to the society. Against this background, Kanno et al. developed a human-centered modeling framework of urban systems that captures various types of interdependencies underlying urban sociotechnical and socioeconomic systems, by integrating three subsystems: civil life, manufacturing/service industry, and lifeline infrastructure to the model. Following that, Wakayama et al. applied this model to optimize the post-disaster recovery of water supply system and showed the practical feasibility of this model. However, since the simulations only assume specific fixed disaster scenarios, cases with low probability of occurrence but extensive damage could not be considered. Therefore, we developed a modeling framework that can consider both interdependency and uncertainty, by modifying the model of the previous study to enable random scenario generation and applying the Monte Carlo method. In order to evaluate the validity of the proposed framework, we compared two results: with/without simple post-disaster countermeasures. 2. Method 2.1 Overview of the model The simulation model consists of three main subsystems: civil life, manufacturing/service industry, and lifeline infrastructure, and nine different types of interdependencies existing within and between these subsystems are considered. For example, citizens need to use lifelines to live, while lifelines need workers to commute to the lifeline facilities to work properly. Lifelines are expressed as multi-layered networks of links and nodes, with each link having parameters that represent the connection and length. Enterprises and citizen agents are placed on the road network. When a disaster occurs, it is assumed that some lifeline links will be out of service, and repair squads will be dispatched to repair those links. The resilience index of the system is evaluated by calculating the resilience triangle throughout the simulation, where performance of the whole urban system is calculated as a linear sum of each of the subsystems, and the objective is to minimize the resilience index. A brute-force Monte Carlo method is applied to this simulation, and the lifeline links that get destroyed after the disaster are sampled from a uniform distribution for each run. 2.2 Application to the water supply network In this study, we focused on the resilience of water supply system. An additional parameter for whether the link is a main pipe with larger diameter or not was given to the water pipes, and we assumed that only water pipes get damaged when a disaster occurs. Then, we compared the results with and without restoration planning. In case of having a restoration plan, main pipes were prioritized, and when the pipes are the same type, priority was given to the pipe closer to upstream. In case of no restoration plan, the order of restoration was selected randomly. 2.3 Simulation Settings The road network model consists of 2310 nodes and 1951 links, and 6990 citizen agents (each representing 4 people) and 628 enterprises are placed on the network. As for the water distribution network, 4910 pipes are placed, and 13 water repair squads work to repair the pipes after a disaster occurs. 100,000 simulations will be conducted, with each simulation lasting 20 days, with a disaster occurring on the fifth day, when 100 water pipes are randomly destroyed. 3. Results In case of no restoration plan, the average resilience index was 16.1 and the maximum resilience index was 758.2. In the case of having a restoration plan, the average resilience index was 14.4 and the maximum resilience index was 82.5. Over 99% of the simulations in both cases resulted in a resilience index below 50, so the worst cases where the resilience index exceeds 50 could be overlooked without the Monte Carlo method. In addition, while a numerical comparison alone only shows a slight improvement in the average resilience index, a graphical comparison shows a significant reduction in the possibility of an extremely high resilience index. 4. Conclusion We developed a modeling framework based on previous research to enable resilience assessment of urban infrastructure systems considering multiple interdependency and uncertainty. Simulation results showed that the effectiveness of measures can be evaluated from more perspectives than comparing numbers obtained from a simulation with a fixed scenario. The next step will be randomizing the scale of the disaster and adding attributes that affect the failure probability to the pipes. As the number of random factors increases, the variance of the simulation results will increase, and more simulation runs will be needed, so we will apply variance reduction methods to reduce computational effort. After that, we will implement more reality-based pre-disaster and post-disaster measures and comparing the results of their effectiveness, so that the model could be put to practical use. References T. Kanno, T. Suzuki, S. Koike, and K. Furuta, “Human centered modeling framework of multiple interdependency in urban systems for simulation of post disaster recovery processes,” Cognition, Technology & Work, 21(6), pp. 301-316, 2019. Wakayama, K., T. Kanno, Y. Kawase, H. Takahashi, and K. Furuta, “Comparison of the post-disaster recovery of water supply system by GA optimization and heuristics”, Proceedings of the 30th European Safety and Reliability Conference and the 15th Probabilistic Safety Assessment and Management Conference (2020)

A modeling framework for assessing resilience of urban infrastructure systems considering multiple interdependency and uncertainty

1. Introduction Damage to infrastructure during disasters is a major challenge for countries around the world, and various kinds of measures such as seismic retrofitting of infrastructure facilities are taken in place to mitigate it. Simulation-based evaluations are commonly used to determine the most appropriate measure to choose from, and various modeling frameworks have been developed. In many modeling frameworks, the focus is on reproducing infrastructures precisely, but as urban systems are complex and composed of a variety of interacting elements, it is necessary to include not only physical lifeline infrastructures but also important human activities to the society. Against this background, Kanno et al. developed a human-centered modeling framework of urban systems that captures various types of interdependencies underlying urban sociotechnical and socioeconomic systems, by integrating three subsystems: civil life, manufacturing/service industry, and lifeline infrastructure to the model. Following that, Wakayama et al. applied this model to optimize the post-disaster recovery of water supply system and showed the practical feasibility of this model. However, since the simulations only assume specific fixed disaster scenarios, cases with low probability of occurrence but extensive damage could not be considered. Therefore, we developed a modeling framework that can consider both interdependency and uncertainty, by modifying the model of the previous study to enable random scenario generation and applying the Monte Carlo method. In order to evaluate the validity of the proposed framework, we compared two results: with/without simple post-disaster countermeasures. 2. Method 2.1 Overview of the model The simulation model consists of three main subsystems: civil life, manufacturing/service industry, and lifeline infrastructure, and nine different types of interdependencies existing within and between these subsystems are considered. For example, citizens need to use lifelines to live, while lifelines need workers to commute to the lifeline facilities to work properly. Lifelines are expressed as multi-layered networks of links and nodes, with each link having parameters that represent the connection and length. Enterprises and citizen agents are placed on the road network. When a disaster occurs, it is assumed that some lifeline links will be out of service, and repair squads will be dispatched to repair those links. The resilience index of the system is evaluated by calculating the resilience triangle throughout the simulation, where performance of the whole urban system is calculated as a linear sum of each of the subsystems, and the objective is to minimize the resilience index. A brute-force Monte Carlo method is applied to this simulation, and the lifeline links that get destroyed after the disaster are sampled from a uniform distribution for each run. 2.2 Application to the water supply network In this study, we focused on the resilience of water supply system. An additional parameter for whether the link is a main pipe with larger diameter or not was given to the water pipes, and we assumed that only water pipes get damaged when a disaster occurs. Then, we compared the results with and without restoration planning. In case of having a restoration plan, main pipes were prioritized, and when the pipes are the same type, priority was given to the pipe closer to upstream. In case of no restoration plan, the order of restoration was selected randomly. 2.3 Simulation Settings The road network model consists of 2310 nodes and 1951 links, and 6990 citizen agents (each representing 4 people) and 628 enterprises are placed on the network. As for the water distribution network, 4910 pipes are placed, and 13 water repair squads work to repair the pipes after a disaster occurs. 100,000 simulations will be conducted, with each simulation lasting 20 days, with a disaster occurring on the fifth day, when 100 water pipes are randomly destroyed. 3. Results In case of no restoration plan, the average resilience index was 16.1 and the maximum resilience index was 758.2. In the case of having a restoration plan, the average resilience index was 14.4 and the maximum resilience index was 82.5. Over 99% of the simulations in both cases resulted in a resilience index below 50, so the worst cases where the resilience index exceeds 50 could be overlooked without the Monte Carlo method. In addition, while a numerical comparison alone only shows a slight improvement in the average resilience index, a graphical comparison shows a significant reduction in the possibility of an extremely high resilience index. 4. Conclusion We developed a modeling framework based on previous research to enable resilience assessment of urban infrastructure systems considering multiple interdependency and uncertainty. Simulation results showed that the effectiveness of measures can be evaluated from more perspectives than comparing numbers obtained from a simulation with a fixed scenario. The next step will be randomizing the scale of the disaster and adding attributes that affect the failure probability to the pipes. As the number of random factors increases, the variance of the simulation results will increase, and more simulation runs will be needed, so we will apply variance reduction methods to reduce computational effort. After that, we will implement more reality-based pre-disaster and post-disaster measures and comparing the results of their effectiveness, so that the model could be put to practical use. References T. Kanno, T. Suzuki, S. Koike, and K. Furuta, “Human centered modeling framework of multiple interdependency in urban systems for simulation of post disaster recovery processes,” Cognition, Technology & Work, 21(6), pp. 301-316, 2019. Wakayama, K., T. Kanno, Y. Kawase, H. Takahashi, and K. Furuta, “Comparison of the post-disaster recovery of water supply system by GA optimization and heuristics”, Proceedings of the 30th European Safety and Reliability Conference and the 15th Probabilistic Safety Assessment and Management Conference (2020)

Paper SO331 | | Download the presentation PowerPoint file.

Name: Sotaro Kawanishi (sotarokawanishi@gmail.com)

Bio:

Country: ---
Company:
Job Title:

Paper 2 BD285

Lead Author: Brian D. Ehrhart Co-author(s): Benjamin B. Schroeder bbschro@sandia.gov Ethan S. Hecht ehecht@sandia.gov

Quantitative Risk Assessment Sensitivity Study as the Basis for Risk-Informed Consequence-Based Setback Distance Requirements for Liquid Hydrogen Storage Systems
A quantitative risk assessment on a representative liquid hydrogen storage system was performed to identify the main drivers of individual risk and provide a technical basis for revised separation distances for bulk liquid hydrogen storage systems in regulations, codes, and standards requirements. The quantitative risk assessment framework in Hydrogen Plus Other Alternative Fuels Risk Assessment Models (HyRAM+) was used, and multiple relevant inputs to the risk assessment (e.g., system pipe size, ignition probabilities) were individually varied. For each set of risk assessment inputs, the individual risk as a function of the distance away from the release point was determined, and the risk-based separation distance was determined from an acceptable risk criterion. These risk-based distances were then converted to equivalent leak size using consequence models that would result in the same distance to selected hazard criteria (i.e., extent of flammable cloud, heat flux, and peak overpressure). The leak sizes were normalized to a fraction of the flow area of the source piping. The results of this risk assessment sensitivity study were then used to select a conservative fractional flow area leak size of 5% that serves as the basis for consequence-based separation distance calculations. Individually varying risk assessment inputs provided insight into the sensitivity to specific input values and assumptions, and ultimately informed the conservatism of the final fractional flow area. This work demonstrates a method for using quantitative risk assessment to provide a technical basis for determining consequence-based separation distances. Sandia National Laboratories is a multimission laboratory managed and operated by National Technology & Engineering Solutions of Sandia, LLC, a wholly owned subsidiary of Honeywell International Inc., for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-NA0003525.

Quantitative Risk Assessment Sensitivity Study as the Basis for Risk-Informed Consequence-Based Setback Distance Requirements for Liquid Hydrogen Storage Systems

A quantitative risk assessment on a representative liquid hydrogen storage system was performed to identify the main drivers of individual risk and provide a technical basis for revised separation distances for bulk liquid hydrogen storage systems in regulations, codes, and standards requirements. The quantitative risk assessment framework in Hydrogen Plus Other Alternative Fuels Risk Assessment Models (HyRAM+) was used, and multiple relevant inputs to the risk assessment (e.g., system pipe size, ignition probabilities) were individually varied. For each set of risk assessment inputs, the individual risk as a function of the distance away from the release point was determined, and the risk-based separation distance was determined from an acceptable risk criterion. These risk-based distances were then converted to equivalent leak size using consequence models that would result in the same distance to selected hazard criteria (i.e., extent of flammable cloud, heat flux, and peak overpressure). The leak sizes were normalized to a fraction of the flow area of the source piping. The results of this risk assessment sensitivity study were then used to select a conservative fractional flow area leak size of 5% that serves as the basis for consequence-based separation distance calculations. Individually varying risk assessment inputs provided insight into the sensitivity to specific input values and assumptions, and ultimately informed the conservatism of the final fractional flow area. This work demonstrates a method for using quantitative risk assessment to provide a technical basis for determining consequence-based separation distances. Sandia National Laboratories is a multimission laboratory managed and operated by National Technology & Engineering Solutions of Sandia, LLC, a wholly owned subsidiary of Honeywell International Inc., for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-NA0003525.

Paper BD285 | | Download the presentation PowerPoint file.

Name: Brian D. Ehrhart (bdehrha@sandia.gov)

Bio: Brian Ehrhart is a Chemical Engineer at Sandia National Laboratories. Since 2017, he has worked to support technical analyses for safety codes and standards for alternative fuels, particularly hydrogen. His current and past work has focused on assessing risk for hydrogen vehicles, rail, and infrastructure; developing software codes for assessing various fire and thermal scenarios; and working to improve the National Fire Protection Association (NFPA) 2 Hydrogen Technologies fire safety code.

Country: ---
Company: Sandia National Laboratories
Job Title:

Paper 3 S.332

Lead Author: Suyun Ham

Study of Bridge Damage Prediction Leveraging Kinematic Contact Enforcement-based Moving Load and Deep Learning Techniques
This paper presents a study on the development of hybrid models featuring structural health monitoring (SHM) inspection using bridge weigh-in-motion (BWIM) physics-based models. Artificial intelligence (AI) techniques helped improve the structural damage prediction during SHM inspection of bridges. This study introduces a comprehensive assessment of 1) a unique finite element (FE) simulation approach, which leverages the kinematic contact enforcement (KCE) method, verified with the vehicle-bridge interaction (VBI) theory, and 2) machine learning (ML) techniques to identify and automatically predict structural damages from the structural response. The KCE method is a new approach to simulating vehicle motion in a BWIM model, which is used to carry out actual structural responses to motion. Thus, KCE method provides contact conditions between elements (i.e., contact type, material properties, and element moving speed), which enables the realistic vehicle motion and structural response to be simulated. The FE model is designed with four different classes of damages with three different damage locations applied under two different load conditions (i.e., static load and moving load). These responses obtained from the FE simulation are further examined by using the feature selection method, which provides the importance rank for ML models. A prediction model includes: 1) a decision tree, 2) a support vector machine (SVM), 3) backpropagation (BP), and 4) XGBoost. Among these prediction models, the deep learning XGBoost with its assembly decision tree provided the most reliable results. The results in this paper verify that structural damage prediction can be achieved by using ML as well as the BWIM structural response, which provides high accuracy in bridge damage prediction.

Study of Bridge Damage Prediction Leveraging Kinematic Contact Enforcement-based Moving Load and Deep Learning Techniques

This paper presents a study on the development of hybrid models featuring structural health monitoring (SHM) inspection using bridge weigh-in-motion (BWIM) physics-based models. Artificial intelligence (AI) techniques helped improve the structural damage prediction during SHM inspection of bridges. This study introduces a comprehensive assessment of 1) a unique finite element (FE) simulation approach, which leverages the kinematic contact enforcement (KCE) method, verified with the vehicle-bridge interaction (VBI) theory, and 2) machine learning (ML) techniques to identify and automatically predict structural damages from the structural response. The KCE method is a new approach to simulating vehicle motion in a BWIM model, which is used to carry out actual structural responses to motion. Thus, KCE method provides contact conditions between elements (i.e., contact type, material properties, and element moving speed), which enables the realistic vehicle motion and structural response to be simulated. The FE model is designed with four different classes of damages with three different damage locations applied under two different load conditions (i.e., static load and moving load). These responses obtained from the FE simulation are further examined by using the feature selection method, which provides the importance rank for ML models. A prediction model includes: 1) a decision tree, 2) a support vector machine (SVM), 3) backpropagation (BP), and 4) XGBoost. Among these prediction models, the deep learning XGBoost with its assembly decision tree provided the most reliable results. The results in this paper verify that structural damage prediction can be achieved by using ML as well as the BWIM structural response, which provides high accuracy in bridge damage prediction.

Paper S.332 | | Download the presentation PowerPoint file.

Name: Suyun Ham (s.ham@uta.edu)

Bio: Dr. Ham is currently an Associate Professor at the University of Texas at Arlington. He is a director of the Smart Infrastructure and Testing Laboratory (SITL). He finished his Ph. D. in Civil Engineering from the University of Illinois at Urbana-Champaign. Before that, he worked in industries for seven years as a researcher and a structural engineer on projects for more than 100 buildings and infrastructure designs. His research expertise and interests with the infrastructure damage assessment field and deep learning.

Country: USA
Company: University of Texas at Arlington
Job Title: Associate professor