You are on page 1of 10
Site-Dependent Seismic Response Including Recent Strong Motion Data RB. Seed University of California at Berkeley, USA S.W. Chang Washington State University, Pullman, USA. SE. Dickenson Oregon State University, USA. ID. Bray University.of California at Berkeley, USA ABSTRACT: This paper presents a brief summary of recently completed studies of site-dependent seismic site response incorporating, the wealth of strong motion data provided by recent earthquakes. The empirical data, results of back analyses of various strong motions recording sites, and analyses of the response of sites to various design levels of shaking are combined to develop recommendations for site classification, prediction of site-dependent amplification, and site-dependent design spectra. The adequacy of current US. building codes and provisions in addressing site-dependent site response is assessed in light of the strong motion data from these recent earthquakes. 1 INTRODUCTION In recent years, the importance of site effects on seismic site response has been repeatedly demonstrated during earthquakes such as Mexico City (1985), Armenia (1988), Loma Prieta (1989), the Philippines (1990), Northridge (1994), and Hyogo-ken Nanbu (1995). This paper presents a brief overview of recently completed studies on the seismic response of (2) soft cohesive sites (b) deep, stiff cohesive sites, and (c) deep, stiff cohesionless soil sites which incorporate the wealth of empirical data and analytical results, principally from the Loma Prieta and Northridge Earthquakes (Chang, 1996 and Dickenson, 1994), The results of these studies were used to develop recommendations for site classifications and site- dependent design spectra for code-based design. The resulting recommendations are then compared with the design levels recommended by the 1994 Uniform Building Code (UBC) and the 1994 National Earthquake Hazards Reduction Program (NEHRP) Provisions. Maps of the areas affected by the Loma Prieta and Northridge earthquakes are presented in Figure | and Figure 2, respectively. The locations of strong motion stations are shown, along with a simplified overview of the regional geology. In Figure 1, soft and deep cohesive soil sites are primarily located 207 along the San Francisco Bay margins; deep stiff soil sites of interest are generally located in the Bast Bay (Oakland) area, The soil sites in Figure 2 are predominantly deep stiff soil sites. 2 SOFT AND DEEP COHESIVE SOILS. Strong motion records were obtained at ten soft and/or deep cohesive soil sites throughout the San Francisco’ Bay region during the Loma Prieta earthquake for moderate levels of shaking (Age = 0.14g to 0.33g). Dickenson (1994) back-analyzed these sites and developed one-dimensional site response models using both equivalent linear (SHAKE90) and fully nonlinear (MARDESRA) analysis methods. SHAKE90 is a slightly modified version of the original SHAKE (Schnabel et al., 1972), and MARDESRA is similar to DESRA-2 (Lee and Finn, 1978) except that the dynamic properties of the soil are represented by the Marti Davidenkov (Martin, 1975) model. The predictive capabihues of these methods can be excellent, as illustrated in the following analysis of Treasure Island, one of the ten soft and/or deep cohesive scil sites of interest. The generalized soil profile for Treasure Istand (TD, shown in Figure 3, indicates that the site consist, cf loose sandy fill and loose silty sand [nook avo grati0w Resi £OH, Acceleration Duriss | Figure 1: Overview of Regional Geology and Recorded Peak Horizontal | the 1989 Loma Prieta Earthquake (Seed et al, 1990). | Nt 228 LEGEND: eset ‘Presoene ati and aie rae deposits (Pad witesnd ttt tt, ‘oka, ad cys basen cs Figure2: Overview of Regional Geology and Locations of Strong Motion Stations that Recorded Motions from the 1994 ?-ethridge Earthquake. underlain by a significant thickness of soft Young San Francisco Bay Mud, a Holocene marine clay. Below the Bay Mud, the site is underlain to a depth of about 90 meters by older materials, stich as dense sands and stiff to hard silty clays. The rock motion recorded at Yerba Buena Island (\"BI), located approximately 2 km from the TI instrument, was 229 used as the input motion. Figure 4 presents the shear wave velocity profile at TI developed from downhole shear wave velocity measurements and regional correlation studies (Dickenson, 1994). Figure 3 also shows the acceleration time histories recorded at TI and YBI, along with the calculated response spectra (5% damping) of the sptema, scent Figure 3: Soil Response at Treasure Island during the 1989 Loma Prieta Earthquake (Seed and Dickenson, 1992). recorded motions. By comparing the response spectra, it can be observed that A,,.. Was. amplified by a factor of two, and that spectral values were results of the SHAKE and MARDESRA analyses of the TI site are presented in Figure 5. Both methods provided excellent agreement with the actual strong amplified by factors of up to four to five. The — motion recording, ‘SHEAR WAVE VELOCITY (nse) me a TREASURE ISLAND | 12 Transverse Component = & 1.0 2 - é 2 & A Eres auere a 4 08 Stade 3.8% z 3 comping = 8 z . ’ - 8 ' é S 8 S a & Figure 4: Soil Profile and Shear Wave Velocity Profile at Treasure Island (Dickenson, 1994) 230 PERIOD, s Figure 5: Results of Equivalent Linear and Fully Nonlinear Analyses for Treasure Island (Seed and Dickenson, 1992). me 3 GENERAL SITE CONDITIONS Seed and Dickenson (1992) performed additional studies to evaluate the general seismic site response characteristics of various idealized sites. The idealized soil profiles for “soft clay” sites, shown in Figure 6, were subjected to varying levels of shaking (varying A,.., magnitude, and frequency content), SHAKE90 analyses (more than 100) were performed for motions with Aniow between 0.103 and 0.30g. A lesser number of fully nonlinear GENERALIZED PROFLE Idcalized Site Conditions Analyzed For “Soft Nonlinear Methods. —— 3 (Domping = 5 =) Bol so ts te B4 NORMALIZED SPECTRAL ACCELE nN 9° Pn PERO, = (2) Eaput Motions: Seuyea = 01-03 g Figue 8: SITE CONDITIONS CONSIDERED FILL THICKNESS oS (a)=0,208 ‘SOFT CLAY THICKNESS a (©)=0,5,20,40,208 Samo COLDER GAY ALLUVUM OL BAY ALLUVIUM é (VERY STIFF, PRIMARILY (©) =0,50, 100, 400% aoe A mae roca” 010 020, 0.50, 0.659 : M<65-06) gon Clay” Sites Using Equivalent Linear and Fully MARDESRA analyses were performed for Anne between 0.30g and 0.50g. The resulting nonlinear variation of A,,., on soft or deep cohesive soil versus Acct is presented in Figure 7. Both the empirical data available and the results of the analyses are included in this figure. The mean and mean + one standard deviation normalized response spectra calculated for the idealized sites for (a) low to moderate and (b) high levels of rock input motion are presented in Figure 8, illustrating the nonlinearity of response to Maximum aceeleration on rock, Figure 7: Variation of A. on Soft or Deep Cohesive Soil vs. o E (Damping = 5%) = Hae en ore de a NORMALIZED SPECTRAL ACCELERATION, AarerAus w 2 S PERIOD. = () Input Motions: aparmas = 05 - 0.65 g Mean and Mean + 1 Standard Deviation Response Spectra for San Francisco Bay Mud Sites at (a) Low to Moderate and (b) High Levels of Input Rock Motion. 231 | differing levels of excitation. These curves were used to develop proposed -hormalized response spectra for soft and deep cohesive soil sites similar to those presented in Figure 10 (Dickenson, 1994). Chang (1996) performed similar studies of deep, stiff soil sites using the same general approach. Both equivalent linear (SHAKE91) and fully nonlinear analyses (D-MOD; Matasovic, 1993) were performed for strong motion recording sites from the Loma Prieta and Northridge Earthquakes to better calibrate these analytical tools. Additional analyses ‘were then performed for hypothetical sites subjected to differing levels of excitation. These analyses, along with the available database of strong motion records, were then used to develop recommendations for evaluation of site-dependent amplification and spectral response. This, in tum, resulted in modification of the global site-dependent response recothmendations of Seed and Dickenson (1992), with the principal modifications applying to deep, stiff cohesionless and deep, stiff cohesive site conditions. 4 RECOMMENDED DESIGN VALUES ‘The results of these calibrated analytical studies, as well as the empirical strong motion data, have next been further extended to consider a broader suite of site conditions. The resulting recommendations for evaluation of site-dependent scismic response for all site conditions are presented in Table 1 and Figures 9 and 10. These are based on the available strong- motion database (not just from the Loma Prieta and Northridge Earthquakes), on both equivalent linear and fully nonlinear analyses of response to varying levels and characteristics of excitation, and judgment. The site-dependent amplification relationships presented in Figure 9 are set at “mean” levels, and they do not incorporate any explicit allowance for variability. As illustrated in two examples in Figures 7 and 11, such variability can be pronounced. Similarly, as illustrated in one example in Figure 12, variability of site-dependent response spectra can be very pronounced. Taken together, these two sources of variability lead to a significant likelihood that spectral response at any given single site will strongly exceed the “mean” spectral levels, predicted based on averaging numerous sites (and responses) within a class of site characteristics. It is the authors’ experience that some additional 232 conservatism is often applied when developing design recommendations based on response analyses for a particular site, and it is our recommendation that such conservatism is warranted. Accordingly, the response spectral values of Figure 10 are set at approximately a mean plus one-half standard deviation level. Approximate mean levels can be developed by dividing all spectral response levels (at ‘T>0 seconds) by about 1.15, and mean plus one full standard deviation levels can be developed by multiplying all spectral response values (at T>0 seconds) by about 1.15. Evaluation of design level response for a given site proceeds in three steps as follow. First, the peak horizontal ground acceleration (Ava) that would oceur if the site had been a competent rock site (site class A) is evaluated, based on regional seismicity. This is then modified to estimate the mean estimate site-specific peak acceleration (A...) for the actual site conditions using Figure 9." The appropriate normalized spectrum from Figure 10 is then scaled to this value of A,,, (at T= 0 seconds). 5. DISCUSSION An illustration of the adequacy of the “fit” provided by these recommended site classification amplification relationships and spectra, as well as the variability and degree of conservatism, is presented in Figures 11 and 12. Taking “deep, stifi” soil sites as a class, Figure 11 presents a plot of both recorded and calculated amplification factors. The recorded data are situated by plotting recorded peak ground acceleration (for deep, stiff soil sites) vs. Avot based on regional attenuation pattems developed for the earthquakes in question (Chang, 1996). The calculated values were developed using both equivalent linear and fully nonlinear analyses methods previously calibrated and verified by means of back analyses of recordings made at so'l sites with good subsurface characterization data (Chang, 1996). As shown in Figure 11, there is considerable scatter to the data, but the recomraided amplification curve from Figure 9 (for site class C3) provides good, representative median values. Figures 12(a) and (b) present the 5% damped elastic response spectra for these same sites and records, along with recommended spectral s!apes from the 1994 UBC provisions (site classes 8, and S) and the 1994 NEHRP provisions (site cliss D), along with the earlier recommendations of Seed and See metres Tablel: + Proposed Site Classification System. General Description Site Characteristics ‘Very hard rock Ws Gv) > 5,000 fusin top 50 R Roy Competent rock with lille oro soil] 2,500 fs = V, (rock) = 3,000 fs, and ; | 7 A, and/or weathered rock veneer. Heoil+weathered rock < 40 ft with Vs > 800 fs | | (all but the top few feet3). 1 | AB, | Soft, fractured and/or weathered rock. | For both AB and AB3: | AB 40 ft < Hsoilt weathered rock $ 150 ft, and | ‘SHA, very shallow soll over rock V5 800 fs (in all but the top few feet3). and/or weathered rock. ‘Deep, primarily cohesionless 7 soils. | No “soft clay” (eee Note 5), and B ‘soil $300 ft) Hohesive soil < 0.2 Heohesionless soil | ‘Medium depth, stiff cohesive soils | Hail soils $2008, and | Bl. B, __ | andlor mix of cohesionless with stiff | Vs (cohesive soils) > 500 fs | cohesive sols; no “soft clay”. (ee Note 5). ‘Medium depth, stiff cohesive soils | Same as Bp above, except i and/or mix of cobesionless with stiff | 0 ft 300 ft i oil | | Deep, Saf CobESIve soils andVor mix oF | Hyoit > 200 Rand | ccohesionless with stiff cohesive soils; | Vs (cohesive soils) > 500 fs | ; no “soft cla i call ‘Soff, cohesive soil at small to moderate | 10 ft = Hyoft clay < 100 ft, and | levels of shaking Amaxrock £0.25 & | Slt cahesve wea Tsong | 10 A= Hag cay = TD, and i | D D, _| levels of shaking. 0.25.8 < Amaxsock $0.45 g, or i | (0.25 g< Amix rock £0.55 g and M<7-1/4) | E, | Very deep, soft cohesive soil Tisoft clay > 100 ft (see Note 5) | k Toft clay > 10 fend either: | Soft, cohesive soil and very stron y By) Bakar sotmd vena ‘Amax rock > 0.55 8, oF | ‘Ammax rock > 0.45 g and M> 7-1/4 I E, _| Very high plasticity clays. Ne aan ree | | if Highly organic and/or peaty sols. H.> 20 ft of peat and/or highly, organic soils | bl @ ‘Sites likely to suffer ground failure due | - ea f F, _| either to significant soil liquefaction or | Liquefaction and/or other types of ground al E other potential modes of ground failure analysis required. I | instability [aT oat tented doph of soils of the peor pes where | | 2. Vpeseismic shear wave velocity (fse) at small shear strain (shear strain ~ 10°%) | | 3. tfurace wils are cohesonless,V,may be less than 800 fac i top 10 fe. | | 4, *Cohesionless sols" = soils with less than 30% "fines" by dry weight, "Cohesive sil” = sols with more than 30% "fines" by | | gy weight, and 15% < PI (fines) < 90%. Soils with more than 30% fines, and PI (fines) < 15% are considered “sity” soils | herein, and these should be (conservatively) treated as "cohesive" soils for site classification pusposes in this Table ‘oft Clay" is defined as cohesive soil with: (a) Fines content >30%, (b) Plfines) 220%, 500 fs Site-specific geotechnical investigetions and dynamic site response analyses are strongly recommended for these conditions Response characteristics within this Class (E) of sites tends to be more highly variable than for Classes A, through D, and the response projections herein should be applied conservatively in the absence of (strongly tecommended) site-specific studies. | 7. Site-specific geotechnical investigations and dynamic site response analyses are required for these conditions. Potentially significant ground failure must be mitigated, and/or it must be demonstrated that the proposed structureifaility can be engineered ts satisfactorily withstand such ground failure and () Ve i I i 233 Dickenson (1992). As shown in these figures, scatter or variability is high, and the UBC and NEHRP provisions may be significantly exceeded for any given individual site and motion. ‘The original spectral shape proposed by Seed and Dickenson (1992). was judged to be slightly overconservative at T ~ 0.8 to 1.2 seconds, and so was reduced in this range to produce the spectral shape now recommended in Figure 10. 6 CONCLUSIONS : Recent earthquakes have provided a large number of sirong motion recordings that allow us to update the state-of-practice in estimating site-dependent design response spectra. ‘The empirical data from these earthquakes and back-analyses of the response of numérous sites have been combined to develop recommendations for classification of sites based on response characteristics, and for site-specific evaluation of (a) amplification and (b) elastic response spectra, for use in simplified code-based seismic design. Early versions of this work were incorporated in the recently revised 1994 NEHRP Provisions treatment of site effects on-response, but the now completed results presented herein lead to recommendations of stronger response levels for some site classes than those currently embodied in the NEHRP or UBC code provisions. REFERENCES: Building Seismic Safety Council 1994. NEHRP Recommended Provisions for Seismic Regulations for New Buildings. Chang, S.W. 1996. Seismic Response of Deep Stiff Soil Deposits, dissertation submitted in partial satisfaction of the requirements for the degree of Doctor of Philosophy in Civil Engineering of the University of California at Berkeley. Dickenson, $.E. 1994. Dynamic Response of Soft and Deep Cohesive Soils During the Loma Prieta Earthquake of October 17, 1989, dissertation submitted in partial satisfaction of the requirements for the degree of Doctor of Philosophy in Civil Engineering of the University of California at Berkeley. International Conference of Building Officials 1994, Uniform Building Code, Vol. 2, Structural Engineering Design Provisions. 234 =, Lee, MK.W. & Finn, W.DL. 1978. DESR4.» Dynamic Effective Stress Response Analysis of Deposits with Energy Transmitting Bounce, Including Assessment of Liquefaction Pc 4 Soil Mechanics Series No. 36, Dept. of Cj, Engineering, University of British Column, ‘Vancouver, Canada, 60 p. 7 Martin, P.P.. 1975. Non-Linear Methods ji, Dynamic Analysis of Ground Response, thesis presented to the University of California x Berkeley, in partial fulfillment of requirements for the degree of Ph.D. in Engineering. Matasovic, N. 1993. Seismic Response of Composie Horizontally-Layered Soil Deposits, Phy, Dissertation, Civil and Environmental Engineering Department, University of California, Log Angeles, 483 p. Schnabel, P.B., Lysmer, J. and Seed, HB. 197), SHAKE: A Computer Program for Earthquake Response Analysis of Horizontally Layered Sites, Report No. EERC/72-12, University of California at Berkeley, December. Seed, R.B. & Dickenson, SE. 1992. Site-Dependen Seismic Site Response, Proceedings , Second Annual CALTRANS Research Workshop, Sacramento, California, March. Seed, R.B., Dickenson, S.E., Riemer, MF., Bray, I.D,, Sitar, N., Mitchell, K., Idriss, 1M., Keyen, R.E., Kropp, A., Harder, L.F., and Power, MS. 1990, Preliminary Report on the Principal Aspects of the October 17, 1989 Loma Prieta Earthquake: Report No. UCB/EERC-90/05, University of California at Berkeley, 124 p. Stewart, J.P., Bray, J.D., Seed, R.B., and Sitar, N. 1994. Preliminary Report on the Principa Geotechnical Aspects of the January 17, 19% Northridge Earthquake, Report No. UCB/EERC 94-08, University of California at Berkeley. o7 5% damping 8 g Effective Peak Ground Acceleration Maximum acceleration (soil), g 8 00 01 02 03 04 05 06 O7 ‘Maximum acceleration (rock), g Period (T, seconds) Figure 9: Proposed Site-Dependent Relationship Figure 10: Proposed Site-Dependent Response Between Mean Aj, ad Anuy fOr ‘Spectra (with 5% damping). “Competent” Rock Sites. 9 ayia een Gero ep et tas cat | eset SHARE catattety 08100 Maximum acceleration (sol aN Ln, ange of MHA on foc ve. MHA on so frosted dung tama Peta (is, 991) ere 00 01 02 03 04 OS 06 o7 08 OB 10 Maxiium acceleration (ck), Figure 11: Relationship for Aza, VS. Anus act for deep stiff soil sites based on available empirical data from the Loma Prieta and Northridge Barthquakes and calculations using both equivalent linear and fully nonlinear site response methods (modified from Chang, 1996). 60 ‘SHAKES | 65 | | 1 . cael Record data (Lome Prat, Neve) | 50 E 1904 URC Sete i 45 1904 USC 53 ete Dicansoreod (1042) C ste 40 35 30 25 20 15 : 10 05 . 0.0 Normalized Spectral Acceleration O0e 0G) ee 20 |] 26) 20 Period (seconds) Figure 12(a): Calculated Normalized Response Spectra from Oakland and Los Angeles Deep, Stiff Sites Compared to Current Design Spectra (Chang, 1996). ‘rerage SHAKE, D-MOD and corded aut] | 1904 NEHRP Sto Type DickonsoevSond (1894 ste Recommanded design spectum 5% damping 4 Normalized Spectral Acceleration i oo Extsitissitrreetiritivietiid i | . 00 05 10 15 20 25 30 age Period (Seconds) \ Figure 12(b): Average of Normalized Response Spectra from Oakland and Los Angeles Sites Recommended Design Spectrum for Deep Stiff Soil Sites (Chang, 1996). 236

You might also like