← Back to Sessions

Sensitivity to Optimization for Scenario Maps for Performance Evaluation of Distributed Infrastructure

Session: Seismic Hazard Analysis for Critical Infrastructure

Type: Oral

Date: 4/19/2021

Presentation Time: 10:15 AM Pacific

Description: 

Evaluation of the seismic performance of distributed infrastructure requires estimates of the spatial dependence of earthquake ground motion from spatial correlation models of the ground motion. Previous spatial correlation models (e.g., Jayaram & Baker, 2009) use global datasets to develop correlation models that are ergodic and are then applied to ergodic ground motion models (GMMs). One challenge facing optimization of scenario maps is the tradeoffs between the number of selected maps, the annual occurrence probabilities of the candidate earthquake sources and the ground-motion epsilon. The smaller the number of the selected map the larger the scale factors of the annual occurrence probabilities of the candidate earthquake sources and ground-motion epsilon values must be to match the target hazard at the individual sites. The scale factors of the annual occurrence probabilities of the candidate earthquake sources and ground-motion epsilon also trade-off between each other. The larger the scaled annual occurrence the less the ground motion epsilon needs to adjust for the suite of maps to meet the target hazard.

While the results from the optimization process for scenarios are analytically correct, the results may be unrealistic. For example, the scaled annual occurrence of the selected event may be on an order of five plus higher than the original event; or the scaled ground-motion epsilon is creating unrealistically high-ground motions. With this any value that optimization for scenario maps provided for risk calculations is negated.

In this study, we perform a rigorous analysis of the sensitivity of the optimization to the three key inputs: number of selected maps, the annual occurrence probabilities of the candidate earthquake sources and the ground-motion epsilon. For the results, we compare the scenario maps computed using alternative sets of constraints on the optimization and outline a process for optimizing for realistic scenario hazard maps.

Presenting Author: Nicolas Kuehn

Student Presenter: No

Authors