SCEC Award Number 20093 View PDF
Proposal Category Individual Proposal (Integration and Theory)
Proposal Title Evaluation of the Impact of CyberShake on Risk Assessments for Distributed Infrastructure Systems
Investigator(s)
Name Organization
Yajie Lee ImageCat, Inc.
Other Participants Zhenghui Hu,
Ronald T. Eguchi
SCEC Priorities 4d, 4c SCEC Groups EEII, GM, Seismology
Report Due Date 03/15/2021 Date Report Submitted 11/01/2021
Project Abstract
 Empirical ground motion models (GMMs) are widely used to quantify the spatially correlated ground motion hazard in seismic risk assessments for spatially distributed systems. One weakness of such empirical GMMs is that they are developed from global ground motion datasets. They represent “average” source, path attenuation, and site response characteristics of global earthquakes, and are associated with large variability that reflect a variety of crustal structures and conditions. Such differences in median and variability can lead to poorly centered and wider than necessary distributions of the risk metrics.

We build on a recent probabilistic seismic risk analysis (PSRA) study of the water pipeline network for the City of Los Angeles, where the system-level performance was established as a function of exceedance probability based on a large set of earthquake simulations using empirical GMMs. By repeating this study using the events and simulations from CyberShake15.12, we explore the impact of region-specific simulations on seismic risk assessments of distributed infrastructure.

The system-level risks computed from CyberShake15.12 simulations and empirical GMMs were found to be similar for shorter return periods (< 100-year) but show significant differences in longer return periods. Specifically, system-level risk computed from CyberShake is about 26% lower than that from GMMs at 500-year return period, and about 41% lower at 2,475-year return period, respectively. While a reduction was not unexpected, more work is needed to fully understand the sources of these difference. A careful examination of the regional ground motion characteristics of the two methods is warranted in future studies.
Intellectual Merit In assessing the probabilistic system-level risks of a spatially distributed system, three ground motion characteristics play an essential role in characterizing the seismic hazard inputs: ground motion median, variability, and spatial correlation. Empirical GMMs are developed from global ground motion datasets. The average ground motions in a specific region are often different from those from the global average and are expected to exhibit lower variability. The SCEC CyberShake platform was designed to address the need for regional assessment of ground motions with physics-based earthquake simulations. Provided that the simulations have been properly validated, they should, in theory, include the source, path, and site effects of a specific region.

As the first probabilistic hazard model using physics-based earthquake simulations, the CyberShake platform has been increasingly applied to test risk assessment at a single site for performance-based engineering. While the benefits of incorporating a physics-based understanding of earthquakes in PSHA modeling through models of rupture mechanics, wave propagation, and regional geologic structure have been widely recognized, the research on the use of a physics-based simulation technique for seismic risk assessment of spatially distributed infrastructure, however, has been limited.

The results of this study showed that seismic risk outcomes developed from physics-based earthquake simulations for spatially distributed systems can differ substantially from those obtained from conventional approaches utilizing empirical GMMs. These differences can have significant implications on public policies concerning seismic risks and on estimates of resources needed for agencies to adequately plan and mitigate the risks to meet the system resilience criteria.

Based on the findings from this study, we recommend that a more careful examination of regional ground motion characteristics of CyberShake simulation and empirical GMM methods to understand their impacts on the risk estimates of a spatially distributed system.
Broader Impacts This project has supported the strong collaboration between the scientists who work on the SCEC broadband platform and CyberShake and the industrial practitioners who face crucial challenges in accurately assessing seismic risks and resilience of spatially distributed infrastructure. This work also serves as a proof-of-concept of utilizing the state-of-thee-art earthquake simulation technology to solve real-world risk-related problems and is expected to provide guidance to the research and PSRA user communities.
Exemplary Figure Figure 6. Expected number of repairs as a function of return periods calculated from CyberShake simulations and NGA-West2 empirical GMMs, respectively. The expected number of pipe repairs computed from CyberShake simulations and empirical GMMs are similar for relatively short return periods (< 100-year). However, they show significant differences for longer return periods. At a 500-year return period, the expected number of pipe repairs calculated from CyberShake is about 26% lower than that computed from empirical GMMs. At a 2,475-year return period, the value is about 41% lower.