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.
44 6.1 General The field and experimental programs completed as part of this project were developed primarily to address data gaps within the best soil liquefaction case history availableâthe WLA site. The WLA site was shaken by two earthquakes in 1987 (the first one did not liquefy the site; the second one did, but some of the PWP records are the subject of controversy) and was re-instrumented in 2004 at a nearby location. There is a wealth of recorded information at the re-instrumented site, including PWP records from the 2012 M 4.9 Brawley Swarm event. However, information required for effective-stress calculations by means of modern software was lacking. The supplemental geotechnical exploration program at the WLA is documented in this study. The program was designed to fill in several data gaps, including site characterization data at both the original WLA site and the re-instrumented WLA site, and to generate (i.e., recover by sonic drilling) soil for the centrifuge modeling (herein referred to as the centri- fuge experiment). Filling the data gaps allowed for additional interpretations of SM and PWP recordings at the site, several of which are also presented in this chapter. Detailed information about site-specific investigation at the WLA site, including a descrip- tion of site conditions, geology, and material deposition history, and the results of standard in-situ and laboratory testing, is provided in Appendix B-1. Information about advanced labo- ratory testing of intact samples of the WLA site silty sand is provided in Appendix C-2. Infor- mation about advanced laboratory testing of remolded samples from the WLA site is presented in Appendix C-3. Information about the centrifuge experiment is presented in Appendix C-1. Also presented in this study is information relevant to the improvement of the Treasure Island case history. This information is presented and explained in detail in Appendix D-1. It includes the results of advanced laboratory testing performed on soil specimens retrieved from the same, well-characterized geologic unit at a site nearby, with similarity further established based on soil index properties and Vs. 6.2 WLA Site â Supplemental Characterization and Data Interpretation 6.2.1 General The WLA case history only partially relies on a large amount of published data (e.g., OYO suspension logging; large amount of data from cone penetration test sounding and standard penetration test sounding). A significant portion of the information used herein was generated during this project and is presented in Appendices B-1, C-1, and C-2. To facilitate review of this C H A P T E R 6 Field and Experimental Programs
Field and Experimental Programs 45  report, select information is presented in Figures 6-1 and 6-2, which are reduced-size versions of Figures 2 and 10, respectively, from Appendix B-1. Select26 locations of past and current geo- technical instrumentation, current investigations, and past and current interpretations of site conditions are included in the profile. The study-specific centrifuge experiment was initially designed to replicate, as closely as rea- sonably possible, the response of the upper 7 meters of the WLA site in three recorded events, including the 1987 M 6.2 Elmore Ranch and M 6.6 Superstition Hills earthquakes and the 2012 Brawley Swarm, as represented herein by one of its largest events (i.e., by the 2012 M 4.9 Hovley event; see Table 4-8). The fictitious model boundaries (i.e., the centrifuge model domain) are delineated in Figure 6-2(a) by the magenta box. This domain is within the boundaries of the site exploration effort. Soils used for the centrifuge experiment were recovered at WLA from within the prototype domain. Remolded samples prepared by wet tamping in an oversized mold27 are characterized in subsequent sections of this report, as well as in Appendices C-1 and C-3. This information was used to develop the study-specific centrifuge experiment case history. 6.2.2 Field Exploration, Sampling, and Routine Laboratory Testing Prior to the field exploration, a wealth of information was available for the WLA site. This information included borehole logs with SPT sounding logs (standard samplers were used), seismic cone penetration test (sCPT) sounding logs [Figure 6-1; results from the Youd et al. (2004) study are available up to 32 m b.g.s.], OYO suspension logs of in-hole shear wave velocity sound- ings (to 100 m b.g.s.; Youd et al., 2004), and the results of detailed geologic mapping. Information from past site exploration efforts was the basis for planning and execution of the supplemental site exploration program. The current program included mud-rotary drilling, sonic drilling, and sCPT soundings, which were performed between the previously investigated and instrumented locations. With the exception of soil classification of the top layer in the profile (Unit A in Figure 6-2), results of previous investigations and interpretations were confirmed. Unit A was classified as âsilty clayâ by Bennett et al. (1984), then reclassified to âsandy lean clay to lean clayâ by Youd et al. (2004), and given that both materials are present in the unit, it is herein classified as just âclayâ [Unified Soil Classification System (USCS) classification: CL]. Processed and interpreted information is presented in Appendix B-1 in the form of a stand- alone, practice-oriented site exploration report. The site exploration report includes a detailed description of the site and site geology, a narrative summary of previous investigations (includ- ing GIS database with aerial presentation of the data), and shear and compressional wave velocity profiles that include the results of soundings taken at the time of this writing as well as past soundings [including OYO suspension logging by Youd et al., 2004, and cross-hole and spectral analysis of surface waves (SASW) sounding by Cox, 2006]. Borehole logs taken at the time of this writing (energy calibration was performed), sCPT logs, logs of PWP dissipation tests (performed using CPTu), and a log of a slug test are included in the appendices. Results of routine laboratory testing, including location of test specimens on plasticity charts (multiple samples) and grain size distribution charts, are provided for Units A and C. Field pro- cedures for recovery of intact ring and bulk samples are presented as a narrative. Details about recovery and transport of intact samples are provided in Appendix C-2, where the results of advanced laboratory testing are also included. 6.2.3 Interpretation of Site Conditions The interpretation of the WLA site data along the profile developed through three areas studied (the original WLA site, the re-instrumented WLA site, and an area in between that
46 Seismic Site Response Analysis with Pore Water Pressure Generation: Guidelines was simulated in a centrifuge experiment) is provided in Figure 6-2. It shows the authorsâ assessment of site stratigraphy and composition as amended by the site exploration program documented in Appendix B-1. Alluvial soils in the profile that encompasses both previously investigated and instrumented WLA sites and the currently investigated site generally consist of, from top to bottom: ⢠Upper fine-grained stratum (Unit A). Light brown, brown, and reddish gray soft to stiff clay to sandy silt that extends from the ground surface to a depth of approximately 2.1 m. Locally, this layer extends to a depth of 3.7 m. Plasticity of this material changes with depth and across the site. Depending on the sample recovery location, it was classified as lean clay (CL), sandy silt (ML), or fat clay (CH) in the past. Fat clay appears to be the dominant material in the profile. ⢠Coarse-grained stratum (Unit B). Light brown and brown medium-dense silty sand (SM) to poorly graded sand that extends from the bottom of Unit A to depths ranging from approximately 6.4 to 7.0 m b.g.s. (a) (b) Note: âGLA (2021)â refers to Appendix B of this study (field exploration and site characterization). Figure 6-1. Select results of in-situ testing at the WLA site before and after the 1987 M 6.6 Superstition Hills earthquake: (a) CPT cone tip resistance; (b) normalized and standardized SPT blow counts. The critical layer is highlighted.
Notes: T.D. = total depth; GLA = Geo-Logic Associates, Inc. Figure 6-2. Plan view of the WLA site with indicated borehole, sounding, and instrumentation locations.
48 Seismic Site Response Analysis with Pore Water Pressure Generation: Guidelines ⢠Lower fine-grained stratum (Unit C). Light brown and brown lean clay to sandy lean clay; encountered from the bottom of the coarse-grained stratum to depths of 7.0 to 7.6 m b.g.s. ⢠Lower coarse-grained strata (Units DâG). Sands and silts with clay interbeds. Silts dominate, but the presence of medium-dense sand is significant, especially from 17 to 25 m b.g.s. WLA site stratigraphy, as shown in Figure 6-3, is approximately horizontally layered and rela- tively uniform laterally. A site with such uniform horizontal layering is ideally suited for 1D site response evaluations. This was confirmed by a review of the horizontal to the vertical spectral ratio (HVSR), as recorded in the 1987 M 6.2 Elmore Ranch and 1987 M 6.6 Superstition Hills earthquakes. The Unit A and Unit B soils for centrifuge testing were recovered from the area shown with the magenta box in Figure 6-3a. Soil recovered from Unit A (upper fine-grained stratum) and remolded in the centrifuge was spun for 105 prototype years. It should be noted, however, that this spinning does not over-consolidate tested soil, which remains normally consolidated. Over- consolidation effects from seasonal wet/dry cycles, cementation by mineral deposition (e.g., evaporation), and other environmental effects were not replicated. The influence of these effects was, however, identified in a gross sense by processing of CPTu data and from results of consoli- dation tests. 6.2.4 Interpretation of Advanced Laboratory Testing Results Available interpreted information about the nonlinear drained response of liquefiable Unit B (silty sand) from advanced laboratory and in-situ testing conducted in prior studies is presented in Figure 6-4 in the form of modulus reduction and damping curves. The advanced laboratory testing included CyDSS, CyTX, and RC testing. The results of in-situ testing (Vibroseis shaker) were available for modulus reduction only. Results of back analysis of WLA site data (modulus reduction only) are also included in Figure 6-4. References are provided within the figure. With the exception of results obtained by in-situ testing with excitation generated by Vibroseis shaker by Cox et al. (2009), both modulus reduction and damping data fall within a relatively narrow range. Ordinates of the WLA site silty sand modulus reduction and damping as evaluated in this study are provided in Table 6-1 at commonly used strain levels. The results of the RC tests correspond to a drained test condition. The results obtained by in-situ testing with excitation generated by the Vibroseis shaker also correspond to a drained test condition. Most CyTX and CyDSS tests are conducted undrained. (A constant-volume CyDSS test is also undrained.) Strain-controlled CyDSS tests that were performed on intact and remolded samples of the WLA site silty sand during this study (Appendices C-2 and C-3) are undrained. Therefore, modulus reduction and damping have been evaluated from either the first or second cycle loops where PWP buildup is relatively small. 6.2.5 Interpretation of Strong-Motion Data Strong-motion record pairs at a downhole array can be used to infer the average shear stress- shear strain response of the soils between the two sensor depths (e.g., Matasovic, 1993; Zeghal and Elgamal, 1994; Elgamal et al., 1995). An interpretation for the middle of Unit B [2.9 m b.g.s.; see Figure 6-3(a)] subject to the Superstition Hills northâsouth (NS) acceleration history is shown in Figure 6-5. This interpretation was made assuming that the shear strain distribution is uni- form between two in-hole SM instruments. A pair of acceleration histories was double-integrated to obtain the corresponding pair of displacement histories. For a known distance between the WLA site SM instruments, calculated displacement histories were converted into an average shear strain history. The corresponding shear stress histories were calculated based on the shear beam
Notes: T.D. = total depth; AMSL = above mean sea level. Figure 6-3. Northâsouth (i.e., longitudinal) soil profile through WLA site.
50 Seismic Site Response Analysis with Pore Water Pressure Generation: Guidelines (a) (b) Figure 6-4. Interpreted modulus reduction and damping curves for the WLA site Unit B (silty sand). Cyclic Shear Strain (%) Normalized Shear Modulus (-) Equivalent Viscous Damping (%) 0.0001 0.000316 0.001 0.00316 0.01 0.0316 0.1 0.316 1 0.999 0.997 0.992 0.974 0.923 0.792 0.545 0.275 0.107 1.01 1.04 1.14 1.44 2.31 4.54 8.73 13.32 16.18 Table 6-1. Modulus reduction and damping curves for WLA silty sand. theory. Similar calculations can be performed from an average acceleration history and known unit weight of soil by means of Newtonâs second law (force = mass à acceleration, where force is shear stress for unit area and mass is calculated from unit weight). A plot of in-situ stressâstrain loops evaluated following the procedure outlined previously for the 3.1 to 7.0 m depth interval at the WLA site from shaking recorded in the NS direction during the M 6.6 Superstition Hills event is shown in Figure 6-5. Note that evaluated loops do not exhibit stress- or strain-controlled behavior. The dilative behavior, however, may be inferred subsequent to liquefaction beyond approximately 0.75% shear strain. Similar evaluations of the in-situ hysteretic response of soil profile to strong ground shaking were performed for the
Field and Experimental Programs 51  critical layer of the WLA site in the 1987 Elmore Ranch earthquake and for the critical layer of the Owi Island site in the 1985 Chiba-Ibaragi earthquake. These evaluations are provided in Appendices E-2 and E-4, respectively. Comparison with the results of comparable numerical modeling (i.e., calculated hysteretic loops) is provided in Appendix E-4. 6.3 Centrifuge Experiment 6.3.1 General With two exceptions (the Port Island, Japan, and the WLA site in the Hovley earthquake case histories; see Table 4-8), the recorded case histories considered in this study are in response to relatively weak input motions (i.e., in-hole or bedrock outcrop motions with PGA ⤠0.2 g). With two exceptions (the WLA site in the Elmore ranch earthquake and the Owi Island, Japan, case history), the weak motions were sufficient to trigger soil liquefaction but were not large enough to induce large-strain responses. The PGA recorded at the Port Island site was relatively high (0.58 g recorded by an in-hole accelerometer at a depth of 16 m b.g.s.), but no PWP transduc- ers were installed at the site at the time of the earthquake. Centrifuge experiments with laminar boxes excited with PGA in excess of 0.4 g are available (see Appendix A-3 for a list of such experi- ments). However, almost exclusively, these experiments were performed using clean sand rather than the silty sand more commonly found in nature. Given these data gaps, a study-specific centrifuge experiment was designed. This experiment included excitation with a high PGA (PGA ⥠0.4 g), abundant instrumentation to record seismi- cally induced PWP, and a layered specimen built from clay and silty sand in a laminar box. The experiment was a 35 g experiment (i.e., the model was accelerated to 37.8 g with adjustment for soil settlement simulation). It was performed at the facilities of UNH. The details of the employed centrifuge, shake table, and mounted laminar box are provided in Appendix C-1. The study-specific centrifuge experiment was designed to replicate, as closely as reasonably pos- sible, the upper 7 meters of the WLA site. The soil profile at the WLA site is shown in Figure 6-3(a), where the approximate domain of the centrifuge experiment is indicated by the magenta box. The centrifuge model of the WLA site is schematically represented by a cross-section through Figure 6-5. WLA site Unit B (silty sand) â in-situ stressâstrain loops in the middle of the WLA site critical layer during the 1987 Superstition Hills earthquake.
52 Seismic Site Response Analysis with Pore Water Pressure Generation: Guidelines the laminar box presented in Figure 6-6. It includes a layer of fat clay28 with a thickness of about 3.1 m underlain by a layer of silty sand with a thickness of about 3.9 m. The target depth of the groundwater table was 1.5 m (approximately in the middle of the clay layer). Relevant information about input (i.e., target) acceleration histories and their counterparts achieved at the base of the laminar box is provided in Table 6-2 and further in Appendix C-1. For the first two motions listed in Table 6-2, comparison of the target spectrum and spectrum achieved upon processing to accommodate for the response of the hydraulic actuator at the base of laminar box is presented in Figure 6-7. Similar spectral plots for other motions used in this study are provided in Appendix C-1. Record (Component)(1),(2) Target PGA Achieved PGA Target AI Achieved AI Elmore Ranch (NS) 0.078 g 0.079 g 0.10 m/s 0.11 m/s Superstition Hills (NS) 0.172 g 0.22 g 0.90 m/s 1.11 m/s Hovley (EW) 0.239 g 0.17 g 0.07 m/s 0.14 m/s Sitka MO (NS) 0.68 g 0.65 g 6.37 m/s 2.37 m/s Olympia WHTL (EW) 0.68 g 0.57 g 2.79 m/s 7.51 m/s Notes: (1) Herein, the record name corresponds to a seismic event as recorded at the case history site in-hole, except for the Sitka Magnetic Observatory (MO) and Washington Highway Test Laboratory (WHTL) in Olympia, where a ground surface record was applied at the base of the laminar box. (2) The larger of the two orthogonal components was selected. NS = northâsouth (component); EW = eastâwest (component). PGA here also refers to acceleration at the base of the laminar container; AI = Arias intensity (index of motion energy; the inherent property of ground motion). Table 6-2. Motions, abbreviations, and comparison of target and achieved parameters. Note: LVDT = Linear variable differential transformer. Figure 6-6. Centrifuge model of the WLA site (top 7.0 m of the WLA site; dimensions shown are in the prototype scale; dimensions of the actual container are 178 3 356 mm in the plan view and 241 mm in height).
Field and Experimental Programs 53  As presented in Table 6-2, consistent with the spectral plots in Figure 6-7(a), the best agree- ment between recorded and target histories, as expressed in terms of spectral ordinates, PGA values, and AI values, is for the Elmore Ranch acceleration history. The same agreements for the Superstition Hills record are poor. This is especially notable in Figure 6-7(b), which reveals that the hydraulic actuator introduced a spike in spectral period at the fundamental period of the considered soil profile (Ts = 0.29 s). Poor agreement between achieved and target spectral ordinates (see Appendix C-1), PGA values, and AI values for the Hovley acceleration history is expected. This is because this acceleration history is essentially a pulse motion (Ds = 5.2 s; see Table 4-8), and pulse motions are difficult to match with a hydraulic actuator. For the last two motions [the July 1977 M 7.68 Alaska earthquake, Sitka Magnetic Observatory (MO) record and the 1949 M 7.1 Western Washington earthquake (Olympia), Washington Highway Test Laboratory (WHTL) record] scaled up to the upper limit of motions that can be applied at the base of the aforementioned laminar box, which is 0.68 g, a poor match was also achieved. However, this poor match is irrelevant herein because these motions were not selected and scaled up to replicate an actual case history. They were selected to generate a new case history that simulates the response to large-magnitude and high-energy events. The centrifuge model was shaken at the base of the laminar box, in succession, by the suite of motions listed in Table 6-2. The first part of the centrifuge test, when excess pore water pressure mainly accumulates, is considered a surrogate29 of an undrained CyDSS test and is used, as such, for comparison. An interval of at least 150 minutes (in prototype scale) was allowed between motions to allow for full dissipation of excess PWP. This centrifuge experiment was a partial success. The primary goal of the experiment (to create a case history of liquefied silty sand subject to very strong ground shaking) was achieved, but not for all motions. The Olympia WHTL record (i.e., the record with the highest AI; allied last in the sequence) is the only motion that liquefied the model. Relevant information in the experi- ment was recorded, and this information was sufficient to create the seventh case history (listed in Table 4-7). Discussion on how the response to the Olympia WHTL record was affected by the earlier shaking is provided in Appendix C-1. The comparison of the 1987 WLA site surface response in the field to the centrifuge experi- ment is presented in Section 6.4. (a) (b) Figure 6-7. Comparison of in-hole and base of the laminar box acceleration histories: (a) Elmore Ranch earthquake; (b) Superstition Hills earthquake.
54 Seismic Site Response Analysis with Pore Water Pressure Generation: Guidelines 6.3.2 Advanced Laboratory Testing 6.3.2.1 General The most significant of these data gaps is a lack of results of undrained laboratory testing of the WLA site silty sand. Results of this type of advanced laboratory testing are required for proper calibration of advanced CMs. Information generated by advanced laboratory testing was also required to build a centrifuge model that served as a basis for the development of a supple- mental case history of strong ground shaking (see an outline of a 35 g centrifuge experiment in Table 4-7). To cover a range of possibilities of advanced CMs, both strain- and stress-controlled undrained CyDSS tests were performed. To accommodate the requirements of standard engineer- ing practice, testing was performed in accordance with ASTM D8296, âStandard Test Method for Consolidated Undrained Direct Simple Shear Test Under Constant Volume with Load Control or Displacement Control.â The undrained cyclic response of silty sand is of primary interest in this study. The undrained cyclic response of clay is of secondary interest. Given the rapid application of shaking at the base of the laminar box, one may infer that a dynamic centrifuge experiment is an undrained test. However, unlike an undrained CyDSS test, dissipation of excess PWP occurs during shaking of the laminar box, continues after cessation of shaking, and may even revert to zero at the end of the centrifuge experiment. Therefore, the centrifuge experiment is closer to the field condi- tions than its undrained CyDSS counterpart. Like in any geotechnical project, intact samples are tested first. Disturbed and remolded samples are tested only if intact samples are not available or, as in this study, where testing of remolded soil is required to numerically replicate the results of the centrifuge experiment. Therefore, both intact (i.e., relatively undisturbed) specimens and remolded specimens were tested. Intact specimens of WLA silty sand were recovered from a mud-rotary borehole by a Shelby tube. A relatively large quantity of soil was required for the centrifuge experiment. Such a quantity could not be recovered from a mud-rotary borehole. Therefore, another borehole was advanced nearby through sonic drilling. Soil from that borehole was used both for the centrifuge experiment and to prepare remolded samples for advanced laboratory testing. The drilling techniques, intact sampling equipment, and sampling procedures are described in detail in Appendix C-2 along with the related procedures for specimen transport and extrusion. The procedures for preparation of remolded specimens of the same soil, and of clay soil from the same site, are described in Appendix C-3. Detailed testing protocols are also presented in these appendix sections. 6.3.2.2 CyDSS Testing â Intact Specimens (Strain- and Stress-Controlled) Because the number of available intact samples of the WLA site Unit B (silty sand) was small (two), the strain-controlled testing on the first intact sample was performed in stages of increas- ing shear strain amplitude. The staged uniform strain-controlled loading sequence is schemati- cally shown in Figure 6-8(a). It was executed as shown (i.e., testing was not stopped between stages to allow for dissipation of excess PWP). Cyclic PWP (excess PWP) was measured during CyDSS testing. The silty sand specimen liq- uefied [i.e., normalized excess PWP ratio ru reached 0.9530] at the cumulative cycle number 50. As shown in Figure 6-8(a), the cumulative cycle number 50 was logged after 10 cycles of loading at 1% single-amplitude shear strain. The stressâstrain response of the first intact sample from the WLA site Unit B (silty sand), plotted over the entire range of applied strain [Figure 6-8(a)], is presented in Figure 6-9(a). The plot of a stressâstrain response inferred from in-situ records (see Section 6.2.5 for explanation)
Field and Experimental Programs 55Â Â (a) (b) Figure 6-9. Comparison of stressâstrain responses in Unit B (silty sand; critical layer): (a) strain-controlled CyDSS test; (b) inferred from in-situ recordings. (a) (b) Figure 6-8. Strain-controlled CyDSS test on intact sample of WLA site Unit B (silty sand; critical layer): (a) applied shear strain history; (b) normalized PWP ratio (ru) history.
56 Seismic Site Response Analysis with Pore Water Pressure Generation: Guidelines is shown in Figure 6-8(b). Even though, theoretically, both plots represent similar boundary and confining conditions and are representative of the same geologic unit, only a qualitative comparison is possible. Inferred stressâstrain loops appear to show dilatancy, but the results of advanced laboratory testing performed at similar (cyclic) shear strain levels do not. Stress-controlled testing was performed on the second available intact sample of the WLA site Unit B (silty sand). The PWP and stressâstrain responses are shown in Figure 6-10. As shown in Figure 6-10(a), this intact sample liquefied in the first31 stress cycle [cyclic shear stress ratio (CSR) = 0.2]. The results of strain-controlled testing, performed in a manner presented herein and shown in Figures 6-8 and 6-9, can be used to develop parameters of both simple and advanced effective- stress CMs used in this study (i.e., MKZ, UBCSAND, PM4SAND, and UCSDSAND3), and also for many other CMs. However, the same test results cannot be used to develop modulus reduc- tion curves, and they also cannot be used to develop PWP model parameters of simple con- stitutive models. Similarly, the results of the stress-controlled testing can be used to develop parameters of advanced CMs, cannot be used to develop modulus reduction curves, and cannot be used to develop PWP model parameters for the simple constitutive models. While undrained strain-controlled testing is arguably more representative of in-situ condi- tions, soil dilation is easier to spot in the results of stress-controlled testing. In general, practicing engineers prefer stress-controlled undrained testing to undrained strain-controlled testing. One of the reasons for this preference may be that the instructions for development of advanced CM parameters from the results of stress-controlled testing are more detailed (a better explanation is provided) than the instructions for advanced CM parameter development from the results of strain-controlled testing. 6.3.2.3 CyDSS Testing â Remolded Specimens (Strain-Controlled) Strain-controlled CyDSS testing of remolded samples was performed in support of the cen- trifuge experiment. It was performed in stages of increasing shear strain amplitude. The staged loading sequence is schematically shown in Figure 6-11. The loading sequence follows its intact sample counterpart except for (i) the largest applied strain was 0.5%,32 and (ii) excess PWP was allowed to dissipate between stages (valves were open at the end of each stage). This type of staged testing is suitable for development of material parameters for both simple and advanced CMs. (a) (b) Figure 6-10. Stress-controlled CyDSS test on intact sample of WLA site silty sand: (a) normalized PWP response; (b) normalized stressâstrain response.
Field and Experimental Programs 57  Remolded specimens from WLA site Unit A (clay) and Unit B (silty sand) were tested. Tests were performed in a research laboratory (UNH). The procedure for preparation of remolded specimens of silty sand and of clay soil and other testing details are provided in Appendix C-3. The results of undrained testing of remolded clay soil are presented only in Appendix C-3. Excess PWP generation was measured during each stage of CyDSS testing. The excess PWP was normalized with the initial vertical effective stress. The plot shown in Figure 6-12(a) shows that ru â 0.90 is reached after approximately 20 to 25 cycles of uniform shear strain with an amplitude of 0.5%. This is, along with the flattening of the excess PWP curve, an indication of soil liquefaction. The excess PWP and stressâstrain responses shown in Figure 6-12 do not correlate well. After 50 cycles of 0.5% uniform shear strain (i.e., after significantly more straining than occurred in the 1987 Superstition Hills earthquake), consistent with the results of strain-controlled CyDSS testing of intact samples, silty sand should fully liquefy, and the effects of soil liquefaction should be more pronounced in the corresponding stressâstrain response shown in Figure 6-12(b). Soil strength and stiffness should be degraded to their residual values, and the plot should resemble a stressâstrain plot as shown in Figure 6-9(a). The results of staged, undrained, strain-controlled CyDSS testing with PWP dissipation between stages can be used to develop modulus reduction and damping curves. The modulus reduction and damping data are typically evaluated from second cycles (i.e., when stressâstrain loops are fully closed and excess PWP is still relatively low). The modulus normalization to Gmax is best based on on-sample shear wave velocity measurement by bender elements (BEs). (a) (b) Figure 6-11. Strain-controlled CyDSS test on remolded samples of WLA site silty sand â applied shear strain history (PWP was allowed to dissipate between stages).
58 Seismic Site Response Analysis with Pore Water Pressure Generation: Guidelines An attempt to generate modulus reduction and damping curves from the results of the undrained strain-controlled CyDSS testing performed on remolded samples of the WLA site silty sand (see Figures 6-11 and 6-12, and Appendix C-3) is presented in Figure 6-13. e counter- part generic modulus reduction and damping curves developed by EPRI (1993) (range of con- ning stress corresponds to the testing conditions) and Vucetic and Dobry (1991) (PI = 0) are shown in Figure 6-13 for reference. Figure 6-13 shows that the results of CyDSS testing of remolded samples of the WLA site silty sand do not correspond to their expected counterparts. is is another indication that there are problems with the subject testing and that the particular CyDSS test results are not usable. Similar observations have been made on the results of CyDSS testing of remolded samples of clay. e likely cause of observed (poor) behavior is system compliance of the UNH CyDSS testing device. (a) (b) Figure 6-12. Stress-controlled CyDSS test on remolded sample of WLA site silty sand (Unit B): (a) normalized PWP response (stage with cyclic shear strain 5 0.5%); (b) stressâstrain response over 50 cycles (discarded test results). (a) (b) Figure 6-13. Results of strain-controlled CyDSS tests on remolded WLA site silty sand compared to generic curves: (a) modulus reduction, (b) damping (discarded test results).
Field and Experimental Programs 59  6.4 WLA Site â Comparison of Field and Physical Modeling Data The secondary goal of the centrifuge modeling program was to replicate the 1987 WLA site case history (two records) in a laminar box. Therefore, the laminar box was shaken at its base by motions targeting/attempting to replicate the M 6.2 Elmore Ranch and the M 6.6 Superstition Hills earthquake records. The comparison of the surface response in the field and in the centri- fuge experiment is shown in Figure 6-14 for the Elmore Ranch NS record and in Figure 6-15 for the Superstition Hills NS record. A relatively good agreement between comparable ground surface responses (i.e., accelera- tion response spectra) is shown in Figure 6-14 for the first 1987 event (M 6.2 Elmore Ranch earthquake), in which liquefaction did not occur at either the WLA site or in the centrifuge experiment. Figure 6-14. WLA site â record from the centrifuge experiment compared to its field counterpart (Elmore Ranch earthquake, NS component). Figure 6-15. WLA site â record from the centrifuge experiment compared to field counterpart (Superstition Hills earthquake, NS component).
60 Seismic Site Response Analysis with Pore Water Pressure Generation: Guidelines A relatively poor agreement between comparable ground surface responses (i.e., acceleration response spectra) is shown in Figure 6-15 for the second of the 1987 events (M 6.6 Superstition Hills earthquake). The silty sand (i.e., sand from Unit B) liquefied in the field, but the same silty sand remolded in the laminar box did not liquefy. One may argue that the agreement between two spectra is poor because remolded silty sand did not liquefy. However, the main reason for the poor agreement is that motion applied at the base of the laminar box does not match its counter- part in the field. This is evident in Figure 6-7(b), where the match between the field acceleration history and the one that was applied at the base of the laminar box is poor. It is possible to develop valuable insights from the centrifuge experiment that simulates the WLA response in the 1987 earthquakes. For example, the controversy about the WLA site being a true 1D site (i.e., concerns about whether WLA responds in a 1D mode were voiced as early as 1987) is resolved by a good agreement between the field and laminar box response in the 1987 Elmore Ranch earthquake (see Figure 6-14). This confirms that the WLA site, at least in the Elmore Ranch event, responded in a 1D mode. Pore water pressure records from the centrifuge experiment with the 1987 Elmore Ranch earthquake may partially compensate for the lack of such records in the actual event.