10.2 Ground Motion and Seismic Ground Deformation Characterization

The regional seismic hazard assessment and ground-motion characterization should follow the procedures provided in applicable seismic guidelines and code provisions (e.g., PEER TBI 2017, ASCE 7). These procedures include the application of Probabilistic and Deterministic Seismic Hazard Analyses (PSHA, DSHA) incorporating specific seismic source models (e.g., UCERF, USGS NSHMP 2014 or 2018) and ground motion models (GMMs). The GEOR may use updated, widely adopted models in PSHA and DSHA in site-specific analysis. The ground-motion characterization should address pertinent issues such as near-fault effects, basin effects, and dynamic soil response (site effects). Embedment and base averaging effects may be accounted for, as applicable.

The selection and modification of ground motions (acceleration time series) should be consistent with recommendations found in the applicable codes and standards.

The subsequent sections address ground-motion characterization at the surface and at depth.

10.2.1 Ground-Motion Characterization at Surface

• For Site Classes A, B, and C, the ground-motion development may be based on Vs30 measured from the ground surface using ground motion models (GMMs). The resulting ground-surface acceleration response spectra (MCER and DE) should be checked against minimum code requirements.

• For Site Class D determined based on Vs30 measured from the ground surface, the ground motion may be developed using site response analyses

or GMM’s, as determined by the GEOR and approved by the geotechnical members of the EDRT. The resulting acceleration response spectra should be checked against the minimum code requirements. Consideration of site response analysis is warranted because of the breadth of the Vs30 values defining Site Class D soil profiles (i.e., Vs30 of 600 to 1,200 ft/sec) and the range in anticipated ground-surface motions for the wide variety of soil conditions represented by Site Class D sites.

Other factors influencing the decision to perform a site response analysis include: (1) depth to material with shear wave velocity equal to or greater than 1,200 ft/s, (2) depth to bedrock defined as the Site Class B/C boundary (2,500 ft/sec), and (3) the trend of site-specific Vs with depth (i.e., the site period).

• For Site Classes E and F, site response analysis using methods suitably calibrated by the GEOR should be performed and the design spectrum calculated at the ground surface should be in conformance with the applicable Building Code requirements. For sites where (1) surficial soil (e.g., liquefiable fill and soft Bay mud) are removed through basement excavation and foundation installation or (2) ground improvement is used to bypass liquefiable or soft soil, the GEOR should evaluate whether the site could be reclassified as site class D with concurrence with the geotechnical members of the EDRT.

th the applicable Building Code requirements. For sites where (1) surficial soil (e.g., liquefiable fill and soft Bay mud) are removed through basement excavation and foundation installation or (2) ground improvement is used to bypass liquefiable or soft soil, the GEOR should evaluate whether the site could be reclassified as site class D with concurrence with the geotechnical members of the EDRT.

The number and characteristics of ground motions, variation in shear-wave velocity profile, and variation in soil shear modulus reduction and material damping curves used in site response analysis should be adequate to capture the potential variation in surface ground motion in a realistic and defensible manner.

The following procedure is suggested for consideration by the GEOR:

After a thorough review of site-specific geotechnical and geophysical data, evaluate the applicability of GMMs (i.e., Vs30-based estimation) for approximating the dynamic response of the soil profile.

If site-specific aspects of the soil profile are not reasonably approximated by the “average, characteristic Vs profile” implied by the GMMs, ground response analysis should be considered. The ground-surface motions developed through ground response analysis should be checked against minimum code requirements.

Ground Motion Characterization Commentary: The level of analysis required for establishing surface, or near-surface, ground motions should reflect site-specific factors such as stratigraphy, geotechnical characteristics and properties of the soils, depth to bedrock, the trend of Vs from the ground surface to competent bedrock, and the amplitude of the bedrock motions (e.g., MCER, DE). Methods of analysis can be generalized as consisting of (1) numerical dynamic site response analyses, (2) estimation using current GMMs that include regression terms for Vs30 (e.g., NGA-West2 GMMs), and (3) simplified, code-based site class designation and site coefficients (Fpga, Fa, and Fv), which are required as a check on the ground motions developed using methods 1 or 2. The applicability and suitability of site response analysis and GMMs for the development of design-level ground surface motions should be evaluated prior to adoption on a project-specific basis for all Class D sites. The potential range of representative ground-surface motions anticipated at Class D sites due to the inherent variability of subsurface conditions and dynamic response of soil profiles falling under this general Vs30based classification in the San Francisco Bay Area necessitates critical evaluation of the procedures applied for developing design ground motions. It is suggested that the GEOR engage the geotechnical members of the EDRT as soon as practical after pertinent site-specific geotechnical and geophysical data have been collected to identify the appropriate method of developing ground-surface motions prior to analysis. The following suggestions are deemed pertinent to local practice and provided for demonstration and guidance. For sites containing soft-to-medium stiff fine-grained soils (e.g., BM), numerical ground response analysis is preferred and considered the primary method for developing ground-surface motions. This suggestion also applies to sites with lower Vs30 (600 ft/s to 900 ft/s). In this situation, methods 2 and 3 are performed as checks on the results of the numerical site response analyses. For stiffer soil profiles with Vs30 in the upper range of the Site Class D category (900 ft/s to 1,200 ft/s), methods 2 and 3 may be acceptable for characterizing ground motion. Many sites in the San Francisco Bay Area are underlain by dense sand and stiff clays that contribute to Vs30

values in the range of 1,000 to 1,200 ft/s. The development of ground motion for these sites may be based on site-specific Vs30 measured from the ground surface using GMMs.

Ground Response Commentary: Dynamic ground response analyses are routinely performed in practice using equivalent-linear and nonlinear models. The strengths and limitations of both methods of analysis have been addressed in the technical literature, and one of the primary differences in the two approaches is simulation of moderate- to large-strain behavior in cyclic loading. The combination of soft or medium stiff soil (i.e. BM or other marine deposits) and liquefiable sands that are prevalent in San Francisco, and the

nonlinear models. The strengths and limitations of both methods of analysis have been addressed in the technical literature, and one of the primary differences in the two approaches is simulation of moderate- to large-strain behavior in cyclic loading. The combination of soft or medium stiff soil (i.e. BM or other marine deposits) and liquefiable sands that are prevalent in San Francisco, and the

strength of design-level cyclic loading leads to highly nonlinear soil behavior. Therefore, nonlinear models that have been suitably calibrated are preferred over the equivalent linear model; however, equivalent linear site response analysis results are often used for comparison with nonlinear site response analysis results. Numerous computer programs have been used to perform nonlinear site response analysis on local projects. The GEOR may select the preferred model for the project. It is suggested that the GEOR provide documentation supporting calibration of the proposed model for analysis of similar soil profiles subjected to ground motions that are similar in nature to the design-level motions required for the project. Irrespective of the model used on the project, the results of the dynamic response analysis should be reviewed by the geotechnical members of the EDRT. The slope of bedrock in the vicinity of the site should be evaluated and the GEOR, with approval from geotechnical members of the EDRT, should determine whether a two-dimensional site response analysis is required.

For sites at which lateral and vertical variability of the soil profile and depth to bedrock is significant enough to result in dual Site Class designations, two-dimensional or three-dimensional site response analysis may be required to develop an appropriate ground motions for design. The required check against code-based ground motions should be provided for both Site Classes, and the proposed design motions presented to the geotechnical members of the EDRT for review.

10.2.2 Site Response and Ground Motion Characterization at Depth of Interest

Time series selected and modified by the GEOR for use in structural dynamic analyses by the structural engineer of record (SEOR) should be representative of the ground motions at the depth of interest for the structural model. The depth of interest is a function of the modeling approach implemented by the SEOR. Primary considerations for the ground motions used in dynamic structural analyses are well presented in numerous documents (e.g., NEHRP 2015, NIST 2011, NIST 2012). In most cases, ground surface motions should not be used in structural models for buildings with multiple basement levels. Therefore, the acceleration response spectrum used as the basis for modification of time series should be developed using either (1) calibrated ground response analysis allowing development of the acceleration response spectrum at the appropriate depth, or (2) validated simplified methods that account for foundation embedment effects. The latter would be required, for example, on projects for which the ground surface motions were developed using GMM’s and trends in the motions with depth are not provided.

loped using either (1) calibrated ground response analysis allowing development of the acceleration response spectrum at the appropriate depth, or (2) validated simplified methods that account for foundation embedment effects. The latter would be required, for example, on projects for which the ground surface motions were developed using GMM’s and trends in the motions with depth are not provided.

The design team, with review by the geotechnical members of the EDRT, should determine whether ground response analysis should be performed using ground motions corresponding to MCER, DE (or both), and possibly the Serviceability Level earthquake (SLE).

For sites where (1) surficial soils (e.g., liquefiable fill and BM) are removed through basement excavation or foundation installation, or (2) ground improvement is used to bypass liquefiable and soft soil, the GEOR, with concurrence of the geotechnical members of the EDRT, should evaluate whether the site could be reclassified for the sake of ground-motion comparison to code-based requirements based on a representative 30 m (100 ft) time-averaged interval velocity that is computed using site-specific Vs data over a depth range deemed appropriate for configuration of the basement, foundation, or ground treatment.

Ground Motion Characterization Commentary: For a surface foundation, the energy transmitted to the structure is applied through soil in contact with the base of the foundation. For embedded structures, the basement walls may be in contact with liquefiable soil or soft clayey soil over a certain depth and then in contact with competent soil down to the lowest elevation of the basement walls. In this case, the presence of soft or liquefiable soil may be ignored and Vs30 could be evaluated from the surface of competent soils. The rationale behind this is; while ground motion within soft or liquefiable soil may be higher than ground motion within the competent soils, the energy transmitted to the structure from these layers is relatively small due to their low stiffness (i.e., the product of ground-motion intensity and soil stiffness controls the amount of energy transmitted to the structure from each layer). However, seismic earth pressures should consider the effects of soft soil against basement walls.

10.2.3 Kinematic Soil-Structure Interaction (KSSI)

KSSI analysis may be performed using (1) simplified methods accounting for base averaging and embedment effects (e.g., NIST 2012), or (2) finite element or finite difference kinematic SSI analysis. It should be noted that the provisions of ASCE 7-16 (Chapter 19) provide a maximum allowable reduction of ground motion due to combined (base averaging and embedment) kinematic SSI effects when performing nonlinear response history analyses. Per ASCE 7-16 Section (19.2.3), the site-specific response spectrum modified for kinematic SSI shall not be less than 70% of Sa as determined from the design response spectrum and MCER response spectrum motions developed using the code-based approaches. When using the simplified method (Chapter 19, ASCE7-16) for evaluation of ground motion with embedment effects, the Vs30 computed from the ground surface (as opposed to from the bottom of the basement) should be used. To compute the ground motion reduction due to embedment effects, the average shear wave velocity over the height of the basement should be used.

If finite element or finite difference kinematic SSI analysis is performed (1) the ground motion near the boundary of the model should be similar to those obtained from one dimensional site response analysis, and (2) kinematic ground motion should meet ASCE 7 requirements.

Commentary: If soil conditions at the boundary of the FEM model vary from those at the site, the ground motion calculated at the boundary may be compared with results of one-dimensional finite element or finite difference site response analysis using soil profile at the boundary.

10.2.4 Development of Ground Motion Time Series

If ground acceleration time series are used (i.e., performance-based design approach), seed motions should be selected based on the controlling earthquake scenarios (e.g., magnitude, site-to-source distance, significant duration (D5-75, D5-95), Arias Intensity, peak ground velocity (PGV), and period of pulse for forward-directivity motions), and the Vs30 at the recording station. The percentage of seed motions that have near-source (directivity) characteristics can be defined from deaggregation of the regional seismic hazard (PSHA) for the 2,475-year average return period and across the structural period range of interest, identification of the primary seismic hazards, and the amplitude of the motions from the predominate seismic sources relative to the uniform hazard (NEHRP 2015, NIST 2011).

If spectral matching of seed motions is performed, care should be exercised not to eliminate or unreasonably elongate the pulse period.

Ground motions with velocity pulse characteristics should be rotated and oriented along fault normal (FN) and fault parallel (FP) directions. Furthermore, the modified motions in FN and FP directions should be rotated again based on the orientation of the building axis relative to the causative fault. Seed motions that do not exhibit near-fault effects (i.e., without the forward-directivity or fling step) may be used in a random orientation.

stics should be rotated and oriented along fault normal (FN) and fault parallel (FP) directions. Furthermore, the modified motions in FN and FP directions should be rotated again based on the orientation of the building axis relative to the causative fault. Seed motions that do not exhibit near-fault effects (i.e., without the forward-directivity or fling step) may be used in a random orientation.

Commentary: Applying seed motions that do not exhibit near-source effects in a random orientation deviates from ASCE 7 requirements but is judged to be appropriate. However, care should be taken that the mean spectra for each direction of response meets ASCE limits so as to avoid design that do not meet minimum strength criteria in any direction.

It is recommended that orbit plots at structural periods of interest be made before and after spectral matching and before and after rotation of ground motion along the building axis to confirm that the appropriate orientation of ground motion is used in the structural dynamic analysis.

For structures on continuous foundations with plan dimensions of greater than 400 feet, effects of wave passage and incoherency of ground motion on design ground motions should be evaluated and addressed.

10.2.5 Seismic Slope Stability and Soil Liquefaction Hazards

The potential for and consequences of liquefaction or cyclic degradation of soils should be evaluated using current and widely adopted methods of analysis. The evaluation of liquefaction hazard should be based on standard semi-empirical methods.

If potentially liquefiable soil layers are present below the foundation level, the effects of soil liquefaction (strength loss, settlement and down-drag loads acting on deep foundations) and potential for lateral spreading should be evaluated. The GEOR should review published maps and reports regarding potential for soil-liquefaction-induced ground settlement and lateral spreading at the site and in its vicinity.

Commentary: Existing reports include Lawson Report on the 1906 Great San Francisco Earthquake (1908), Harding Lawson Associates, City and County of San Francisco Soil Liquefaction Report (1992), GHD-GTC Port of San Francisco Seawall Stability Report (2016), and Port of San Francisco, Seawall Resiliency Study currently underway.

For sites underlain by BM, the potential for seismically induced slope deformation should be evaluated, and mitigation measures should be identified.