SCEC Award Number 15203 View PDF
Proposal Category Individual Proposal (Integration and Theory)
Proposal Title 3D Ensemble Dynamic Rupture Simulations
Investigator(s)
Name Organization
Eric Dunham Stanford University
Other Participants Kenneth Duru, postdoc
SCEC Priorities 3e, 6b, 3c SCEC Groups CS, FARM, DRCV
Report Due Date 03/15/2016 Date Report Submitted 03/14/2016
Project Abstract
 This project is the latest step in a multi-year effort to understand the origin of complexity in the earthquake source process and the expression of that complexity in high frequency ground motion. This is accomplished through simulations of dynamic ruptures on fractally rough faults, with strongly rate-weakening friction laws and off-fault plasticity. Our focus this past year as been on two problems: dynamics of the supershear transition on rough faults (using 2D simulations) and initial work on an ensemble of 3D rough fault dynamic rupture simulations. The supershear study culminated in a publication (Bruhat et al., 2016). The rupture process becomes more complex on rougher faults, manifested through secondary slip pulses, sometimes propagating backward, that coexist with a primary slip pulse. Supershear transitions are initiated at geometric complexities and become more likely on rougher faults. However, sustained supershear propagation is favored on locally smooth fault segments. Our primary accomplishment for the 3D simulations was identifying the most relevant part of parameter space, in terms of initial stresses given the strongly rate-weakening friction law, that produces realistically complex rupture histories and ground motion. Ruptures in our simulations appear to be more complex than those found by other groups modeling dynamic ruptures on rough fault surfaces. We suspect this is because we are using strongly rate-weakening friction laws and have found stress levels that are just above the minimum threshold for self-sustaining propagation. Future work will shift toward production runs to generate the ensemble database and analyses of ruptures and ground motion.
Intellectual Merit Supershear ruptures potentially pose a severe ground motion hazard, motivating studies like ours to determine conditions for both triggering and sustaining supershear ruptures. With dynamic weakening, faults can host ruptures at low stress levels, often much lower than necessary for the classic Burridge-Andrews supershear transition mechanism to apply. Here we demonstrated using numerical simulations how naturally occurring fault geometric complexity can trigger supershear ruptures. We also confirmed an idea in the literature, derived from a limited number of geologic studies of surface traces of faults hosting supershear ruptures, that supershear propagation tends to occur on locally smooth segments. These results, when extended into 3D, could be used to quantify the likelihood of supershear ruptures in hazard calculations. We have taken the first steps toward doing the general problem of dynamic rupture on rough fault surfaces in 3D. Such simulations can be used to quantify stress levels at which faults are expected to operate, and to provide a physically realistic means to generate high-frequency ground motion. An ensemble database of ruptures could be studied to advance kinematic rupture generators and make them more consistent with dynamic source models that are realistically complex.
Broader Impacts The project supported one female PhD student and one postdoc. The postdoc is an applied mathematician and has helped promote the numerical and computational challenges of earthquake modeling within the numerical analysis community. The 3D dynamic rupture simulations have potentially important consequences for how seismic hazard analysis is done, specifically with regard to ground motion modeling at high frequencies. Most current high-frequency seismogram generation methodologies are somewhat ad hoc, but the approach pursued here is grounded on geologic constraints on fault surface roughness and laboratory friction experiments on dynamic weakening. The project also supported development of WaveQLab3D, a new 3D dynamic rupture code, that will be used for a wide range of projects on earth-quake dynamics and ground motion.
Exemplary Figure Fig. 1. From Bruhat and Dunham (2016). Distributions of rupture velocity for all simulations in the Fang and Dunham (2013) ensemble database, focusing on background stress levels close to the minimum threshold for self-sustaining propagation (where natural earthquakes are expected to occur). Supershear velocities emerge on rougher faults and at higher background stress levels. These stress levels are still below the classic Burridge-Andrews stress level, thus requiring an alternative mechanism for the supershear transition.