New PEER Report 2020/14: "Probabilistic Seismic Hazard Analysis and Selecting and Scaling of Ground-Motion Records"

December 15, 2020

PEER has just published Report No. 2020/14: "Probabilistic Seismic Hazard Analysis and Selecting and Scaling of Ground-Motion Records," a report for the "Quantifying the Performance of Retrofit of Cripple Walls and Sill Anchorage in Single-Family Wood-Frame Buildings" Project. It was authored by Silvia Mazzoni, University of California, Los Angeles; Nicholas Gregor, N. Gregor Consulting; Linda Al Atik, L. AlAtik Consulting; Yousef Bozorgnia, University of California, Los Angeles; David P. Welch, Stanford University; and Gregory G. Deierlein, Stanford University.

Visit the PEER publications page to download a free color pdf of the document.


This  report  is  one  of  a  series  of  reports  documenting  the  methods  and  findings  of  a  multi-year,  multi-disciplinary  project  coordinated  by  the  Pacific  Earthquake  Engineering  Research  Center  (PEER)  and  funded  by  the  California  Earthquake  Authority  (CEA). The  overall  project  is  titled  “Quantifying the Performance of Retrofit of Cripple Walls and Sill Anchorage in Single-Family Wood-Frame Buildings,” henceforth referred to as the “PEER–CEA Project.”

The  overall  objective  of  the  PEER–CEA  Project  is  to  provide  scientifically  based  information  (e.g.,  testing,  analysis,  and  resulting  loss  models)  that  measure  and  assess  the  effectiveness of seismic retrofit to reduce the risk of damage and associated losses (repair costs) of  wood-frame  houses  with  cripple  wall  and  sill  anchorage  deficiencies  as  well  as  retrofitted  conditions that address those deficiencies. Tasks that support and inform the loss-modeling effort are: (1) collecting and summarizing existing information and results of previous research on the performance   of   wood-frame   houses;   (2)   identifying   construction   features   to   characterize   alternative  variants  of  wood-frame  houses;  (3)  characterizing  earthquake  hazard  and  ground  motions  at  representative  sites  in  California;  (4)  developing  cyclic  loading  protocols  and  conducting  laboratory  tests  of  cripple  wall  panels,  wood-frame  wall  subassemblies,  and  sill  anchorages to measure and document their response (strength and stiffness) under cyclic loading; and (5) the computer modeling, simulations, and the development of loss models as informed by a workshop with claims adjustors.

This  report  is  a  product  of  Working  Group  3  (WG3),  Task  3.1:  Selecting  and  Scaling  Ground-motion records. The objective of Task 3.1 is to provide suites of ground motions to be  used by other working groups (WGs), especially Working Group 5: Analytical Modeling (WG5) for  Simulation Studies.  The  ground  motions used in the numerical  simulations are intended to  represent seismic hazard at the building site. The seismic hazard is dependent on the location of the site relative to seismic sources, the characteristics of the seismic sources in the region and the local soil conditions at the site. To achieve a proper representation of hazard across the State of California, ten sites were selected, and a site-specific probabilistic seismic hazard analysis (PSHA) was performed at each of these sites for both a soft soil (Vs30 = 270 m/sec) and a stiff soil (Vs30=760  m/sec).  The  PSHA  used  the  UCERF3  seismic  source  model,  which  represents  the  latest  seismic source model adopted by the USGS [2013] and NGA-West2 ground-motion models. The PSHA was carried out for structural periods ranging from 0.01 to 10 sec.

At   each   site   and   soil   class,   the   results   from   the   PSHA—hazard   curves,   hazard   deaggregation,  and  uniform-hazard  spectra  (UHS)—were  extracted  for  a  series  of  ten  return  periods,  prescribed  by  WG5  and  WG6,  ranging  from  15.5–2500  years.  For  each  case  (site,  soil  class, and return period), the UHS was used as the target spectrum for selection and modification of  a  suite  of  ground  motions.  Additionally,  another  set  of  target  spectra  based  on  “Conditional  Spectra”  (CS),  which  are  more  realistic  than  UHS,  was  developed  [Baker  and  Lee  2018].  The  Conditional  Spectra  are  defined  by  the  median  (Conditional  Mean  Spectrum)  and  a  period-dependent variance. A suite of at least 40 record pairs (horizontal) were selected and modified for each return period and target-spectrum type. Thus, for each ground-motion suite, 40 or more record pairs were selected using the deaggregation of the hazard, resulting in more than 200 record pairs per target-spectrum type at each site. The suites contained more than 40 records in case some were rejected by the modelers due to secondary characteristics; however, none were rejected, and the complete set was used.

For the case of UHS as the target spectrum, the selected motions were modified (scaled) such that the average of the median spectrum (RotD50) [Boore 2010] of the ground-motion pairs follow  the  target  spectrum  closely  within  the  period  range  of  interest  to  the  analysts.  In  communications  with  WG5  researchers,  for  ground-motion  (time  histories,  or  time  series)  selection  and  modification,  a  period  range  between  0.01–2.0  sec  was  selected  for  this  specific  application for the project. The duration metrics and pulse characteristics of the records were also used in the final selection of ground motions. The damping ratio for the PSHA and ground-motion target spectra was set to 5%, which is standard practice in engineering applications.

For the cases where the CS was used as the target spectrum, the ground-motion suites were selected and scaled using a modified version of the conditional spectrum ground-motion selection tool (CS-GMS tool) developed by Baker and Lee [2018]. This tool selects and scales a suite of ground motions to meet both the median and the user-defined variability. This variability is defined by  the  relationship  developed  by  Baker and  Jayaram  [2008].  The  computation  of  CS  requires  a  structural period for the conditional model. In collaboration with WG5 researchers, a conditioning period of 0.25 sec was selected as a representative of the fundamental mode of vibration of the buildings of interest in this study. Working Group 5 carried out a sensitivity analysis of using other conditioning periods, and the results and discussion of selection of conditioning period are reported in  Section  4  of  the  WG5  PEER  report  entitled  Technical  Background  Report  for  Structural  Analysis and Performance Assessment.

The WG3.1 report presents a summary of the selected sites, the seismic-source characterization model, and the ground-motion characterization model used in the PSHA, followed by selection and modification of suites of ground motions. The Record Sequence Number (RSN) and the associated scale factors are tabulated in the Appendices of this report, and the actual time-series files can be downloaded from the  PEER  Ground-motion database Portal (