SCEC Award Number 16264 View PDF
Proposal Category Collaborative Proposal (Integration and Theory)
Proposal Title Accuracy of Drucker-Prager yield condition in nonlinear attenuation of surface waves
Investigator(s)
Name Organization
Daniel Roten San Diego State University Kim Olsen San Diego State University Steven Day San Diego State University
Other Participants
SCEC Priorities 6e, 6b, 6c SCEC Groups GMP, Seismology, CS
Report Due Date 03/15/2017 Date Report Submitted 06/03/2017
Project Abstract
 We perform 2-D nonlinear simulations of P-SV wave propagation along a cross-section connecting the San Andreas fault with the Whittier Narrows corridor. Two different codes are deployed, a 2-D version of AWP which uses a Drucker-Prager (DP) yield condition, and the Noah2D program which is based on a hyperbolic soil model. Linear simulations are first carried out to verify both methods against each other, and both codes predict horizontal peak ground velocities of ~1.3 m/s in the Pasadena region. In nonlinear simulations using constant cohesions and friction angles throughout the domain, horizontal peak ground velocities predicted by AWP are twice as high as those predicted by Noah2D (1.0 vs. 0.5 m/s, respectively). The discrepancy between the two methods is reduced if nonlinearity is limited to soft sediments, although ground motions produced by Noah2D are still 15-25% lower than those obtained by AWP. Because the DP yield condition overestimates the amount of hysteretic damping with respect to a more realistic hyperbolic model, it was expected that ground motions obtained from AWP would underpredict those obtained from Noah2D. This underprediction was not observed, suggesting that the delayed onset of nonlinearity in the DP plasticity model (with respect to the hyperbolic model) prevails over effects of increased damping. These observations highlight the importance of accurately tracking the stress-strain relationship of soft sediments in wave propagation simulations for deterministic seismic hazard assessment, but suggest that the DP yield condition represents a viable, conservative first-order approximation to nonlinearity.
Intellectual Merit There is a great uncertainty about the level of shaking that must be expected during the next M > 7.5 earthquake on the southern San Andreas fault. While linear simulations predict long-period (> 2 s) ground motions of up to 2 m/s in the Los Angeles basin, simulations for an elastoplastic medium suggest that ground motions could be much lower, depending on the strength of crustal rocks and soft sediments. However, there have been concerns that these elastoplastic simulations based on a Drucker-Prager yield condition underpredict ground motions. The 2-D P-SV simulations performed within this project show that this is not the case.
Broader Impacts Because of the uncertainty associated with the level of shaking during the next large earthquake on the southern San Andreas fault, it is difficult to estimate the physical, social and economic consequences of such event. Improved physics-based ground motion models may help to reduce this uncertainty, and allow society to better prepare for large earthquakes. Computations performed within this project lend credibility to previous scenario simulations, and show that recent modifications in simulation tools lead indeed to more accurate ground motion prediction.
Exemplary Figure Figure 5: Seismograms at Pasadena site (blue rectangle in Fig. 2 and 3) simulated using AWP and Noah2D in the linear and nonlinear case. Numbers above waveforms show peak velocity.