Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
85  8.1 Overview The basis for development of guidelines for effective-stress SRA is presented in Chapter 4. The corresponding research approach is presented in Chapter 5, and the field and experimental programs conducted to support this approach are presented in Chapter 6. Numerical modeling as presented in Chapter 7 builds on the information presented in Chapters 5 and 6. The introduc- tory parts of Chapter 7 serve as the basis for development of guidelines for effective-stress SRA presented later. Sections 7.2.2 and 7.2.3 are important parts of the guidelines presented herein. The case histories evaluated herein encompass a relatively broad range of site conditions and ground shaking. The results of these evaluations support the guidelines presented herein. The information generated and processed herein supports several elements of the seismic design of highway facilities since it includes protocols for ESA, including protocols for model selection, site exploration, standard and advanced laboratory testing, and others explained later in this chapter. Information presented herein is neither intended nor sufficient to update or change other, well-established overall protocols for seismic design, such as protocols for per- formance of seismic hazard analysis, development of design ground motions, accounting for uncertainty in both ground motion and soil properties, and performing total stress SRA. The guidelines provided herein do not include, and do not discuss, AASHTO and other code- based requirements for interpretation of the results, such as comparison of calculated values to code-mandated minimum requirements. 8.2 Guidelines for Effective-Stress SRA 8.2.1 General Guidelines for conducting effective-stress SRA involve a sequence of steps. The following dis- cussions summarize key aspects of each step, as applied to effective-stress SRA. Reference is made to appendices that demonstrate the application of these steps. Levels of investigation or evaluation for each step will depend on the complexities of site conditions, levels of anticipated earthquake shaking, and application of results. It will be up to the geotechnical team and numeri- cal modelers carrying out the SRA to decide how much detail is required for a particular site. 8.2.2 Analysis Type Assuming that the SRA approach has been selected for evaluation of the site effects (see Sec- tion 2.2 for discussion of other options) and that 1D SRA is appropriate for the project, the numerical C H A P T E R 8 Guidelines
86 Seismic Site Response Analysis with Pore Water Pressure Generation: Guidelines modeler should evaluate (i) whether ESA is required or if TSA will suffice, and (ii) if TSA will suffice, whether nonlinear analysis is required, or if conventional equivalent-linear analysis will do the job. The first question should be resolved by screening. Several screening options to evaluate soil liquefaction potential from the results of SPT, Becker penetration test (BPT), and CPT sounding or based on the results of shear wave velocity measurement are outlined in, for example, Idriss and Boulanger (2008; 2010). In general, if the site is liquefiable, then ESA is warranted. Note also the requirements for evaluation of âSite Class Fâ sites of ASCE 7-22 and comparison of TSA and ESA spectra shown in Figure 7-9. ESA should not be used as the sole means to evaluate whether a site is liquefiable. The second question should be resolved by performing total-stress SRA using the equivalent- linear approach. As outlined in Section 7.2.3, equivalent-linear analysis should be performed first and regardless of whether nonlinear or nonlinear ESA will follow. The results of this analy- sis should be evaluated to (i) further define critical layer(s) previously identified by the initial screening analyses, if any, and (ii) to review calculated peak shear strain in the profile, including shear strain in the critical layer. Additionally, shear stress from this analysis may be used for evaluation of soil liquefaction potential using the simplified method (Seed and Idriss, 1982, as updated by Idriss and Boulanger, 2008). If the results of total-stress SRA show that, for any given motion, peak shear strain exceeds 1.0%, then nonlinear analysis is required (see discussion in Section 3.2 on the limitations of the equivalent-linear approach). If, at any location within the profile, the average of calculated peak strains exceeds 0.5%, then nonlinear analysis is recommended. This recommendation is intended to account for uncertainty in material properties in the profile (e.g., shear wave velocity and unit weight) and for uncertainty associated with evaluation of depth to bedrock. 8.2.3 Software and Constitutive Models A process of software screening and selection is outlined in Chapter 4, with required attri- butes of nonlinear effective-stress SRA software summarized in Section 4.2.2 (Table 4-1). Details are provided in Appendix A-1. A process of screening and selection of constitutive models is also outlined in Chapter 4, with required attributes of constitutive models summarized in Sec- tion 4.3.2 (Table 4-3). Additional discussion is provided at the end of Appendix A-2. This software screening and selection project step does not necessarily need to be performed first. However, it is highly recommended to have the numerical modeler involved at the early stages of the project, including at the project proposal stage, which includes commitments for use of specific software and constitutive models. 8.2.4 Site Exploration The requirements for site exploration would depend on the specific codes and references cited in the contract and interpreted as relevant by parties to the contract. For example, a contract may call for AASHTO load and resistance factor design (LRFD) and its reference codes, which would then incorporate various FHWA manuals. The minimum requirements are provided in the latest version of the AASHTO Manual on Subsurface Investigations (AASHTO, 2022). Requirements related to performing site exploration in support of total-stress nonlinear SRA are provided in Stewart et al. (2014). Considerations for planning and execution of a site exploration program that supports ESA vary from site to site. Every project is unique. Engineers planning and performing ESA may
Guidelines 87  benefit from the example site exploration program for the WLA site that is presented in detail in Appendix B-1. This example program contains information on how to plan and perform site characterization for effective-stress SRA, plan and perform in-situ testing, recover intact samples, and interpret and present the results. Additional information on how to commission and perform advanced laboratory testing (including how to prepare for a centrifuge experiment) is provided in Appendices C-1, C-2, and C-3. 8.2.5 Site Characterization Detailed information on how to perform and document characterization of a liquefiable site is provided in Appendix B-1. The concept of a critical layer (i.e., a liquefiable soil layer that likely affects the overall response of the site the most) is further explained in the same appendix. If present, such a layer merits detailed characterization. This may involve supplemental in-situ testing and, where possible, advanced laboratory testing. Note that it is possible to have more than one critical layer. Documentation and interpretation of data at the level of detail presented in Appendix B-1 is not required for most sites. The presentation in Appendix B-1 is based on an extensive site exploration program performed in the past and is more suitable for a study such as this one. Examples of less detailed and more focused site characterizations are provided in Appen- dices B-2 through B-4. 8.2.6 Advanced Laboratory Testing and Interpretation of the Results To limit uncertainties in the 1D effective-stress SRA, advanced laboratory testing of select site soils (e.g., soils recovered from the critical layer) is warranted. The focus on the advanced labora- tory testing should be on obtaining information required for conducting element tests. Updates of the WLA site case studies (the 1987 Elmore Ranch and Superstition Hills earth- quakes) and of the TI case study provide guidelines for the development of modulus reduction and damping curves for silty sands, and for generation of information for performing an element test on these materials. Specifics are provided in Appendices C-2 and C-3. These appendices also include detailed information on how to specify advanced laboratory testing (i.e., undrained CyDSS testing on silty sand and clay). Appendix C-2 includes detailed information on how to recover and transport intact samples of silty sand obtained from below the groundwater level. Interpretation of the results is presented in this report and in Appendices C-2, C-3, and D-1 through D-3. An example of how to spot a bad test result or a test result that is unusable for model calibration is presented in Section 6.3.2.3 (Figure 6-13) and in Appendix C-3. 8.2.7 Element Testing Element testing is essentially a numerical simulation of advanced laboratory testing (drained or undrained). In nonlinear total-stress or effective-stress SRA, fitting of modulus reduction and damping curves (CM sub-model testing) using element testing is performed first. This test- ing is theoretically required for every layer in the profile. However, for many generic soil/CM pairs, this information is readily available. For example, one may fit the Vucetic and Dobry (1991) modulus reduction and damping curves shown in Figure 3-1 and save the CM sub-model parameters for future uses. Once the CM sub-model testing is complete, the numerical modeler may proceed with development of equivalent-linear and nonlinear models for system testing (see Section 8.2.7).
88 Seismic Site Response Analysis with Pore Water Pressure Generation: Guidelines Fitting of undrained stressâstrain and excess PWP responses (e.g., results of CyDSS testing) is recommended for the critical layer (or occasionally for critical layers) only. This layer (or layers) is identified during the site characterization program by inspection or screening (e.g., using the latest update of the simplified method). A trade-off between better fitting of stressâstrain response and better fitting of excess PWP is commonly observed. In this event, a good fit in undrained stressâstrain response is preferred to achieving good fit in its excess PWP counter- part. This type of element testing requires estimates of shear wave velocity, relative density, and undrained shear strength [i.e., parameters that are typically established in the next step (SRA Model Development)]. Therefore, this type of element testing is typically performed after com- pletion of the system testing (see Section 8.2.7). Examples of successful and not-so-successful element testing are provided in Sections 5.3.2 and 7.2.2. Details are provided in Appendices D-1 through D-3. 8.2.8 SRA Model Development SRA model development starts with proper selection of boundary conditions. Basic input parameters (i.e., parameters of the soil profile, including stratification, depth to groundwater, and standard soil parameters such as unit weight and shear wave velocity) should be interpreted and established in accordance with the principles of sound geotechnical engineering. These basic input parameters form a basis for development of the SRA model. The process of SRA model development is presented in Appendices E-1 through E-7 for the seven case histories considered herein. As discussed in Section 4.4, these case histories encom- pass a range of site conditions and earthquake loadings. Discretization of the soil profile into layers of approximately uniform thickness is impor- tant when dynamic analysis is performed. The numerical modeler should be cognizant of this requirement, which is not consistent with the requirements for conventional FEM and FDM analyses, which allow for variable layer thicknesses that typically increase with depth. The modeler should consider recommendations for the minimum layer thickness provided in Section 3.3 and the requirements for the maximum layer thickness that can be calculated using Equa- tions 3 and 4. An example of discretization of a soil profile for dynamic analyses is presented in Appendix E-3 for the re-instrumented WLA site. 8.2.9 System Testing System testing is calibration of the analysis at the soil profile level. It entails development of a reference model, which is usually an SRA model developed for use with well-known and verified SRA software such as software based on SHAKE. The calculated spectral ordinates, and preferably calculated peak accelerations, shear strains, and shear stresses within the profile, are compared to their counterparts calculated by SHAKE to confirm similarity or to identify errors in the model assembly, if any. The second part is performed to improve overall system perfor- mance, if desired or required. The system testing is described in detail in Section 7.2.3. Examples are provided in Appendices E-1 through E-7. Presumably, both total- and effective-stress procedures will be improved over time, and then some of the calibration steps recommended and required in Section 7.2.3 would not be needed. 8.2.10 Analysis Analyses presented in this report focus on evaluation of case histories with significant PWP buildup, not on design projects from engineering practice. However, the same assumptions,
Guidelines 89  methods, and principles employed to evaluate case histories may be used and applied in design projects. These include items presented and discussed in Sections 8.2.2 through 8.2.10 and items not discussed herein such as development of design basis, seismic hazard analysis, and develop- ment of design ground motions. Sensitivity studies may be required for some sites (projects). These studies may be required to understand variation in results with, for example, shear wave velocity, modulus reduction and damping, PWP (model) parameters, and in some cases, earthquake motions. 8.2.11 Interpretation of the Results and Interim Reporting A general discussion of data presentation and a presentation of the results is provided in Section 1.6. Guidelines for the evaluation of success of site response analysis are provided in Table 5-3. Additional guidelines and discussion are provided here. Seismic hazard analysis, including development of seismic hazard parameters and design ground motions, should be properly documented. The suggested format45 for presentation of acceleration response spectra of input motions (i.e., abscissa in logarithmic scale) is shown in Figure 7-7. Examples of presentation of SRA results in this format are provided throughout this report. Geotechnical (i.e., soil) parameters for SRA should be presented in the form of a representative plan view (e.g., Figure 6-2) and soil profile (e.g., Figure 6-3). Soil columns should be developed at representative locations within the soil profile (e.g., Figure 6-1). The shear wave velocity profile should also be presented (see examples in Appendices B-1 through B-4 and E-1 through E-7). As noted in Section 8.2.5, documentation and interpretation of data at the level of detail pre- sented in Appendix B-1 is not required at most sites. An example of the presentation of the results of SRA at the ground surface is shown in Figure 7-9. Additional presentation, not shown in Figure 7-9, includes the comparison of the results of ESA with and without excess PWP dissipation. The design engineer should consider the possibility that drainage does or does not occur and that this effect is difficult, though pos- sible, to fully capture with all software considered herein [see, e.g., Figures 7-10(c) and 7-10(g)]. Therefore, both results with and without excess PWP dissipation should be presented to the owner, and then an informed decision should be made on the approach for design. Options include enveloping of the effective-stress response as well as not considering the response with excess PWP dissipation, given the uncertainty in knowing whether dissipation of excess PWP will be impeded by interlayering or percent fines within a layer. Presentation of site response within the soil profile (see Section 7.3.3) is technically not required for most engineering evaluations. However, presentation of plots of calculated accel- eration, shear strain, and shear stress as a function of depth provides valuable insight into site response to strong ground shaking. This presentation may facilitate validation of the results, may provide input for supplemental evaluations (e.g., for evaluation of liquefaction-induced lateral spreading; see Yang and Kavazanjian, 2021), and may facilitate regulatory approval of the design. In particular, it is recommended to present calculated excess PWP response for layers with ru ⥠0.5. A suggested figure format for these data is shown in Figure 7-11. Further discussion and examples of presentation of the results are provided in Section 7.3.3. The FHWA guidance document for seismic design of highway facilities (FHWA-NHI-11-032: LRFD Seismic Analysis and Design of Transportation Geotechnical Features and Structural Foundations) will be posted at the FHWA site in 2024. The results should be evaluated in accor- dance with this guidance document and the latest edition of AASHTO specifications and should conform to the mandated minimum ground surface spectra.
90 Seismic Site Response Analysis with Pore Water Pressure Generation: Guidelines 8.2.12 Peer Review and Final Reporting Peer review is an important part of engineering evaluation and the design process. Obtain- ing a qualified peer reviewer or a team of qualified peer reviewers is recommended for larger or complex projects. The peer reviewers should be senior professionals with knowledge of AASHTO and other codes who have hands-on experience in performing nonlinear effective-stress SRAs. It is recommended that the peer reviewers be involved at all stages (i.e., during the entire project), beginning with reviewing plans for field and laboratory testing and including modeling approach and setup, selection of ground motions, interpretation of data from the SRA, and review of a draft report. 8.2.13 Use of Information Presented Herein for Seismic Design of Highway Facilities Information presented herein is a relatively small, mostly geotechnical contribution to an overall, comprehensive process of seismic design of highway facilities used by the FHWA. Updated information for seismic design of highway facilities will be posted at the FHWA site in 2024 (FHWA-NHI-11-032: LRFD Seismic Analysis and Design of Transportation Geotechnical Features and Structural Foundations). Currently, the FHWA-NHI-11-032-GEC 3 by Kavazanjian et al. (2011) is available for download. This document, however, focuses on the geotechnical aspects of design of highway facilities. Information on characterization of ground motions will be presented in the 2024 document. 8.3 Additional Observations The numerical modeling program was performed in accordance with protocols presented in Sections 7.2.2 and 7.2.3. Relatively good agreement between measured and calculated stressâ strain and excess PWP response (element testing) and recorded and calculated ground surface response and response within the soil profile (evaluation of case histories) confirms its validity. The results summarized in this report and presented in detail in appendices provide additional insight into effective-stress site response analysis that may benefit numerical modelers. These insights include: ⢠If a numerical modeling program is performed in accordance with protocols presented in Sections 7.2.2 and 7.2.3, ground surface and in-profile responses of selected case histories can be matched with selected software and CMs. The achieved accuracy is well within the accuracy required for this type of engineering evaluation. ⢠All of the nonlinear effective-stress SRA software considered herein can do the job. The same appears to be true when analyses are performed with the same software but using different CMs, including simple CMs. Therefore, no preference is given or can be implied to be given to any of the software and CMs considered herein. Ease of use varies significantly. ⢠Advanced undrained laboratory testing of liquefiable sandy silt performed on intact samples is useful and should seriously be considered for major projects. Whenever possible, it should be performed for critical layers. Critical layers should be identified by screening, which may rely on the results of simplified analysis. Engineering judgment is a significant part of the screening process. The results of numerical modeling performed for case histories for which the results of advanced laboratory testing of silty sand were not available are encouraging. For these case histories, relatively good agreement between recorded and calculated response has been achieved with sets of generic material parameters. ⢠The final results of nonlinear effective-stress SRAs presented herein are, in most part, based on careful adjustments of CM parameters to achieve a best fit of recorded ground surface
Guidelines 91  response. For the WLA site, an attempt was made to also achieve the best fit between mea- sured and calculated undrained CyDSS response (i.e., element testing). Neither is possible for forward applications where no such model-data comparisons can be performed. It is for this reason that consistent and validated parameter selection protocols are important. For impor- tant and critical projects, use of best-estimate properties is not sufficient. For such projects, sensitivity studies might be required. 8.4 Additional Conclusions Based on experience and consistent with the findings of this study, additional conclusions can be derived, and recommendations can be made when considering PWP generation and dissipa- tion in 1D models: 1. Effects of PWP generation and dissipation are important to capture in the analysis. In almost every case history considered herein, closer agreement with respect to the compari- son of recorded and calculated response is achieved when PWP generation and dissipation is included in the analysis. 2. The impact of PWP dissipation on calculated site response, as calculated with the PWP dissipation models, is secondary to its PWP generation counterpart. PWP redistribution through cracking, which cannot be captured by nonlinear effective-stress models, may be important. Guidance on how to evaluate surface manifestation(s) of soil liquefaction (i.e., when layer thickness is such that manifestations are expected) is provided in Ishihara (1996). If results of evaluations indicate that manifestation is expected, excess PWP redis- tribution through cracking is expected. 3. The ability to capture soil dilation is, arguably, an important aspect of CMs and has been identified as a required attribute for CMs in Table 4-3. However, results of evaluations documented herein reveal that good agreement between recorded and calculated response can be achieved with simple CMs [i.e., with CMs that cannot simulate dilation (e.g., the MKZ model)]. 4. Uncertainty related to key soil properties can be significant. For example, even when a site exploration program is carefully planned and executed, the range in qc and (N1)60 can be significant (see, e.g., range of values for the critical layer at the WLA site in Figure 6-1 and broad range in Vs presented in Appendix B-1). In general, if key soil properties are mea- sured, the corresponding uncertainty is lessened. If they are estimated from correlations, the uncertainty implicit to the correlation should be carried through the analysis to see how it affects the results. Recommendations on how to treat uncertainty in material parameters and depth to bedrock are provided in the commentary of ASCE/SEI (2022). Model parameters for accounting in uncertainty in shear wave velocity are summarized in Stewart et al. (2014). 5. In general, practicing engineers prefer stress-controlled undrained testing to strain- controlled undrained testing. While strain-controlled testing is arguably more representa- tive of in-situ conditions, soil dilation is easier to visualize in the results of stress-controlled testing. Information for development of advanced CM parameters based on the results of undrained testing is better explained in the corresponding CM documentation. 6. There is other software, and there are other CMs, that can do the job. It is recommended that such numerical codes be considered by DOTs if the modeler can replicate the results of one of the case histories presented in this report that has similar characteristics to the project site. 7. When possible, evaluation of soil liquefaction potential at a given site should not be based solely on the results of nonlinear ESA. Analysis with the simplified method should be per- formed first. The results from a total-stress SRA may be used for evaluation of soil liquefac- tion potential. This may include the calculated peak shear stress profile from equivalent-linear analysis (if calculated shear strains do not exceed 0.5% to 1%) or from nonlinear TSA.
92 Seismic Site Response Analysis with Pore Water Pressure Generation: Guidelines 8. If required, limit the reduction of up to one-third of the code-mandated spectrum (e.g., AASHTO, 2020). 9. When attempting to replicate case histories with centrifuge testing or a shake table, the achieved motions (i.e., motions modified by hydraulic actuator and applied at the base of the laminar box) should closely match the target motions. For example, for the 1987 M 6.6 Superstition Hills earthquake shown in Figure 6-7(b), the match between recorded and achieved surface response is not good enough. 10. Instrumentation of the centrifuge experiment should be redundant because sensors spun to high acceleration levels may malfunction (e.g., see Appendix C-1 for sensors that malfunc- tioned in the centrifuge experiment).