Geotechnical Earthquake Engineering Ground Motions Steve Kramer Department

Geotechnical Earthquake Engineering: Ground Motions Steve Kramer Department of Civil and Environmental Engineering University of Washington Seattle, WA

Ground Motions PSHA (or DSHA) provides uniform hazard spectrum Codes generally produce design spectrum For linear structural analysis, response spectrum is all you need – no need for individual ground motions For nonlinear analyses time histories are required Nonlinear structural analyses Nonlinear geotechnical analyses Nonlinear SSI analyses All are becoming more common Goal is to identify / create one or more motions that have amplitudes, frequency contents, and durations that are consistent with the ground shaking hazard at the site of interest.

Ground Motions Sources: PEER NGA Database (http: //peer. berkeley. edu/nga/)

Ground Motions Sources: PEER NGA Database (http: //peer. berkeley. edu/nga/search. html) Can search for records with characteristics similar to those controlling hazard at site of interest

Ground Motions Sources: COSMOS Database http: //db. cosmos-eq. org/scripts/search. plx Can search for records with characteristics similar to those controlling hazard at site of interest

Ground Motions Problem: Find ground motion(s) that “match” target spectrum Two common approaches: Simulation – single spectrum-compatible ground motion Scaling – suite of motions with matching ensemble average Required information: Sa Target spectrum Uniform hazard spectrum (UHS) Code spectrum Fundamental period of structure To Intent of analyses for which motions are to be used Mean or median response Mean / median and indication of variability of response T

Ground Motion Scaling Simulation Alter characteristics of motion to “match” target spectrum Two common approaches: Time domain – wavelets (actually, time and frequency domains) Example: RSPMATCH (Norm Abrahamson) Frequency domain – Fourier analysis Example: RASCAL (Walt Silva) Both approaches start with some initial ground motion Important that initial motion has “correct” duration Spectrum-compatible motions are useful for determining the mean or median response of a system. They do not provide direct insight into the variability of that response.

Ground Motion Scaling Simulation Example: Spectrum after matching Initial motion too strong Initial motion too weak After Norm Abrahamson COSMOS workshop presentation

Ground Motion Scaling Simulation Example: Original Modified After Norm Abrahamson COSMOS workshop presentation

Ground Motion Scaling Simulation Example: Original Modified After Norm Abrahamson COSMOS workshop presentation

Ground Motion Scaling Simulation Example: Original Modified After Norm Abrahamson COSMOS workshop presentation

Ground Motion Scaling Simulation Example: After Norm Abrahamson COSMOS workshop presentation

Ground Motion Scaling Simulation Example: Note: In areas where seismic hazards come from multiple sources, different parts of UHS may be controlled by different sources – single motion producing entire UHS may not be physically possible. In that case, use of spectrum-compatible motion may be quite conservative. After Norm Abrahamson COSMOS workshop presentation

Ground Motion Scaling Alternatively, we can identify and scale actual recorded motions for (ensemble average) consistency with a target spectrum Use deaggregation to find representative (mean / modal) values of: • Magnitude • Distance • Style of faulting Candidate motions should be consistent with these characteristics Select consistent motions from database (e. g. PEER NGA database) based on seismological properties • Similar magnitude (within +/- 0. 5 provides reasonable duration) • Similar distance range • Similar spectral shape (or epsilon) • Same style of faulting

Ground Motion Scaling Scale motions by constant factor to “match” target spectrum What constitutes a match? Match is in average sense – average of suite of motions May be defined in terms of SRSS spectra (multi-directional components) Usually need to exceed target over significant period range For structures, typically 0. 2 To – 1. 5 To Lower periods (higher frequencies) covers higher mode response Higher periods (lower frequencies) covers damage-induced softening

Ground Motion Scaling Select large suite of ground motions (50 – 100 or so) Use deaggregation to find representative (mean / modal) values of: • Magnitude • Distance • Style of faulting Select consistent motions from database (e. g. PEER NGA database) based on seismological properties • Similar magnitude (within +/- 0. 5 provides reasonable duration) • Similar distance range • Similar spectral shape (or epsilon) Significant • Same style of faulting period range

Ground Motion Scaling Epsilon To To Scaled negative e motion is too strong Be careful – look for local peaks and valleys in candidate motions prior to scaling Scaled positive e motion is too weak To

Ground Motion Scaling Epsilon Baker (2007) took 382 representative ground motions Computed e values for each, chose 20 highest and 20 lowest Computed spectra after scaling to same Sa(T=0. 8)

Hazard Analysis and Ground Motions Summary Design levels of ground motion determined by seismic hazard analysis DSHA – deterministic PSHA – probabilistic Attenuation behavior is critical Prediction of response spectrum may be sufficient Ground motions may be required Synthetic motions – describe mean/median level of shaking Scaled motions Reflect actual earthquake characteristics Suite of motion required – can account for record-to-record variability Results of site response analyses will be sensitive to ground motion inputs – need to pay careful attention to this issue