415.1 N39) 1973 ‘A STUDY OF VERTICAL AND HORIZONTAL EARTH DO_NOT PULL APART 10 PHOTOCOPY! April 1973 WASH—1255 QUAKE SPECTRA UNIVERSITY OF CALIFORNIA Farthquake Engineering Reseorch Center JAN 17 1974 LIBRARY Directorate of U.S. Atomic E n Washington ,D. Licensing ergy Commission C.LEGAL NOTICE ‘This report was prepared as an avcount of work sponsored by the United Siates Government, Nether the Unites States nor the United States Atomic Enerty Commission, nor any of their employees, not any of ther contractors. subcontractors. or their employees, makes. any” warranty xpress or implied, or assumes any legal ability oF tesponsibiily for the accuracy, completeness or usefulness of any information, apparatus, prodct oF process disclosed, or represents that is use would not infringe privately owned nentsWASH—1255 Ii <5. CTR. LIBRARY uc-11 Univ. of Calif,/453 RES. (415) 231-9403 ARI ASTUDY OF VERTICAL AND HORIZONTAL EARTHQUAKE SPECTRA prepared by NATHAN M,|NEWMARK CONSULTING ENGEERING SERVICES * for the DIRECTORATE OF LICENSING UNITED STATES ATOMIC ENERGY COMMISSION UNDER CONTRACT NUMBER AT(49-5) -2667 April 1973TABLE OF CONTENTS 1, INTRODUCTION. .. . 1.1 Object and Scope..... , 1.2 Earthquake Records Studied. ... 2. PROCESSING OF DATA...... 2.1 Time-history....... 2.2 Response Spectra. : 2.3 Amplification Factors...... 2.4 Statistical Analysis... oun & 3. PRESENTATION AND DISCUSSION OF RESULTS. 3.1, Ground Hot fon. 3.2 Response Amplifications....... 4, DESIGN RECOMMENDATIONS AND CONCLUSIONS. 1 General.....-.. 2 Ground Motions 3. Amplification of Ground Mot +4 Design Spectra..... 25 Combined Effects... .0..cccc000 4.6 Recommendations for Future Studies. n REFERENCES APPENDIX ‘A. SEGMENTAL BASELINE ADJUSTMENT. 8. METHODS OF NORMALIZING RESPONSE SPECTRA... 131 1431. INTRODUCTION 1.1 Object and Scope The object of the study, as outlined in the work directive dated 18 August 1971 from the Division of Reactor Standards, U.S. Atomic Energy Commi s: n, was essentially fivefold, namely as follows: (a) Develop vertical and horizontal (two components) response spectra for @ series of 14 strong motion earthquake records, including four San Fernando records, over the range of frequencies of interest, for 0.5, 2, 5 and 10 percent of critical damping. The records which were included as a part of the study are described in Section 1-2 of this report. (b) Compare the vertical and horizontal responses obtained from these spectra. (c) Determine the shape of the spectra in the high frequency range from 6 to 50 hertz, with special emphasis on site-dependent parametric effects. (4) Suggest procedures for combining seismic stresses derived from horizontal and vertical responses. (e) Summarize the above-noted studies in the light of Fecommendations for design. All of the objectives noted have been accomplished with two minor exceptions, namely that after undertaking a comprehensive study of the available ground motion data, it was decided that the ground motion data were generally valid only in the frequency range of 0.95 to 30 hertz and accordingly the response spectra were plotted only for th range. Second, as a result of further studies on the site properties, and based on information that we had available from previous studies, we felt thatit would be difficult to draw general conclusions about site dependence, since only a very few strong motion records on rock were available. Most. of the available records are on alluvium. Nevertheless, some observations pertinent to site dependence were obtained as a part of our studies and comments are presented. Near the end of this investigation it was recognized that the low intensity ground motions were causing a bias in the results. Accordingly studies were made including records corresponding to peak ground accelerations greater than a specified bound to demonstrate the bias. These findings are reported fully. The design response spectrum values reported herein are based on the "bounded" data. The report that follows includes: (1) the description of the methods used in processing the data in Section 2; (2) a presentation and discus: n of the results of the ground motion and special studies in Section 3; and (3) design recommendations and conclusions in Section 4. The procedure employed for adjusting the accelerogram records in this study is presented as Appendix A. The method used for normalizing response spectra 1s glven in Appendix ®. 1.2 Earthquake Records Studied The studies were carried out for the following seismic events: Sen Fernando seismic event Pacoima Dam, 2-9-71, 0600 PST (Record IC Oli1) Castaic, 2-9-71, 0600 PST (Record ID 056) Holiday Inn (First Floor), 2-9-71, 0600 PST (Record IC 048) 15250 Ventura Boulevard (Basement), 2-9-71, 0600 PST (Record IH 115)Other seismic events El Centro, 5-18-40, 2037 PST (Record IA 1) El Centro, 2-9-56, 0633 PST (Record IA 11) El Centro, 4-8-68, 1830 PST (Record IA 19) Hollywood Storage Basement, 7-21-52, 0453 POT (Record IA 6) Hollywood Storage PE Lot, 7-21-52, 0453 PDT (Record IA 7) San Francisco Golden Gate Park, 3-22-57, 11h PST (Record IA 15) Ferndale, 10-7-51, 2011 PST (Record TA 2) Ferndale, 12-21-54, 1156 PST (Record IA 9) Eureka, 12-21-54, 1156 PST (Record IA 8) Hollister, 4-8-61, 2323 PST (Record IA 18) The accelerogram records used in this study were digitized by the Earthquake Engineering Research Laboratory of the California Institute Of Technology; the CIT designation of the earthquake record is indicated in parentheses at the end of each listing. The records are available through the Department of Commerce or the California Institute of Technology.2. PROCESSING OF DATA 2.1 Timeshistory Since the digitized record from the instruments are in terms of acceleration time-histories, the corresponding velocity and displacement are obtained by integration. Adjustments in the records are normal ly required as illustrated by the following example. The acceleration, velocity, and displacement time-histories for a typical record, San Fernando earthquake, NESW component of Castaic, 2-9-71, 0600 PST are shown in Figs. 2.1-2.3. Since the motion duri 1g an earthquake is 2 to and fro type motion, the velocity time-history, Fig. 2.2, is obviously incorrect, and the corresponding displacenent time-history, Fig. 2-3, also is in error. The figure indicates a displacement of 130 in. at 30 seconds. This displacement would have increased further had the computation continued. The errors in earthquake records may arise from any number of sources such as (a) the instrument errors, including effects associated with mounting and instrument housing; and (b) the processing of the record where the i tial conditions (sone motion is required to trigger the mechanism) and the zero acceleration line (baseline) are not known. The errors in the velocity and displacement time-histories arising from integration of the accelerogran are largely associated with the latter category. In order to minimize the record processing errors, the initial conditions are taken as zero, and @ baseline correction is applied to the accelerogram record.‘') fmong various basel ine adjustment procedures, cone which minimizes the square of the error in the velocity is most commonly used. This procedure assumes a polynomial, usually a seconddegree, for the correct acceleration baseline. In some instances the record processing errors are further reduced by using different polynomials for acceleration baseline in different portions of the record. This procedure is explained in detail in Appendix A. Each component of the 14 earthquakes considered in this study was adjusted by the procedure given in Appendix A. A second degree polynomial was used as the acceleration baseline in all cases, with "segmental" adjustments used occasionally as described in Appendix A. The adjusted velocity and displacement time-histories were computed by integrating the adjusted acceleration time-history and they were plotted fon a Calcomp plotter. An examination of the accelerogram records used in this study indicated that the strong motion portion of the record is included within the first 20 seconds of the record. For consistency the computation and plotting of time-histories were carried out for the first 30 seconds of each record, As a general illustration of the effects of adjustment, the unadjusted and the adjusted acceleration, velocity, and displacement ground motion time-histories for a typical record of the San Fernando earthquake, N6QW component of Castaic, 2-9-71, 0600 PST are shown in Figs. 2.1-2.6. 2.2, Response Spectra After the adjusted time-history wes obtained for each record, the time-history was used as a base motion input to a single-degree-of- freedom damped oscillator and the response of the oscillator as 2 function of the frequency of the oscillator was computed. In this way a response spectrum is generated. The computation of response spectra is well (2,3) documented in the literature ‘and is not repeated here. In computingthe response spectra, the adjusted time-history was used because the spectral ordinates computed from the unadjusted record may be in error for low frequencies. The response spectra for 0.5 percent of critical damping for both the adjusted and unadjusted records of the San Fernando earthquake, N69W component of Castaic, 2-9-71, the example used in Section 2.1, are shown in Fig. 2.7. It is seen that adjusting the record affects the spectral ordinates for low frequencies. Although as a general rule the adjustment has little effect for frequencies of interest for nuclear reactors, in order to provide a better basis for comparisons reported later herein, the computations of response spectra were carried out using the adjusted time histories. The response computations were carried out for 38 frequencies having a range of 0.05 to 30 hertz. In computation of the response spectra, the interval in the frequency range over which the computations are carried out influences the shape of the spectra. Generally this influence is not large if small intervals In frequency are used in computations. A comparison between two response spectra for E1 Centro, California, 5-18-1940, NS component, one computed by the California Institute of Technology") using 91 frequencies and the other computed by the authors using 38 frequencies, is shown in Fig. 2.8. In general the agreement between the two spectra is quite good. The difference between the two spectra for frequencies below 0.3 hertz is due to different procedures used for baseline adjustment of the accelerogram. Another comparison between the two spectra for the same earthquake, a coarse frequency interval (38 frequencies), and a fine frequency interval (81 frequencies), both computations being carried out by the authors, is shown in Fig. 2.9.Response computations at higher frequencies were not carried out since the validity of ground motion data for frequencies greater than 30 hertz is questionable. In addition, the response computation at high Frequencies (small periods) require very small integration time steps which are time-consuming computationally. For each component of the record, the response computations were carried out for 0, 0.5, 2, 5 and 10 percent of critical damping. After the results from the response computations were examined, they were plotted on a Calcomp plotter. Plots of the response spectra are included as Figs. 2.10-2.51; the maximum ground motions are shown by three straight-line segnents in each figure, and correspond to the values given in Table 3.1. 2.9. Amplification Factors Since the ground motions for earthquake racords differ from each other, the computed response quantities cannot be compared on an absolute basis. One method of making a meaningful comparative study involves normalization of the response spectra by equating the area under the response spectra between any two frequencies.5) Another method of normalizing the response spectra is to compute the amplification factor -= the ratio of the computed response to the maximum ground motion =» for displacement, velocity, and acceleration at each frequency for the range of interest. With this procedure, the amplification factors can be used to develop design response spectra for a given ground motion. This procedure was followed end is described in Chapter 3. The procedure for normalizing the response spectra is explained in Appendix B.2.4 Statistical Analysis For damping coefficients of 0.5, 2, 5 and 10 percent of critical, and for each frequency, the mean and the standard deviation for both the horizontal and vertical components of the 14 earthquakes were computed. Since there are two horizontal components for each earthquake, the sample size for the horizontal components is twice that for the vertical components. Therefore, on a numerical basis the statistical values for the horizontal components are more reliable than the corresponding values for the vertical components. Initially the statistical studies were carried out as a function of frequency bands; subsequently it was found necessary to carry out the statistical computation at each frequency instead of within frequency bands in order to account accurately for the variation of amplification factors as 2 function of frequency. With the means (or averages) and the standard deviations at various frequencies, the percentile amplifica ns from a normal distribution curve were computed from kt oo where x,, X, and © are, respectively, the percentile amplification, the mean amplification, and the standard deviation at a given frequency. The values of the coefficient c for various probabil ty levels are given in Table 2.1. The percentile anplift ‘ations at each Frequency were used to obtain average amplification coefficients within frequency bands of interest (see Section 3.2). Percentile amplification computations were also carried out by ranking the data. For large sample sizes the two procedures give results which are generally in close agreement with each other.Table 2.1 Normal Deviations from Mean, in 2 Normal Distribution, in Terms of Standard Deviation, as a Function of Cumulative Probability P P deviation from mean standard deviation 0.999 3.09023 0.998 2.87816 0.995 2.87583 0.990 2.32635 0.980 2.08375 0.970 1.88079 0.950 1.64485 0.900 1.28155 0.850 1.03643 0.800 0.84162 0.750 0.67449 0.700 0.52440 0.650 0.38532 0.600 0.25335 0.550 0.12566 Source: Table 11.2, p. 670, "Statistical Methods in Quality Control,” by Dudiey J. Conden, Prentice-Hall, Inc., Englewood Cliffs, New Jersey, 1987.QYOISY D3LSArdbNN - MESN Iib1SbI “47769 “OUNUNYSS NES TAYDICTH SWIL ~ NOFL6HITIIIb I'2°914 ho- 2"0- 10 oo 9 'NOELBYATIIIE 3 nO Be tt 0i QYOI3" T3LSATOBNN - MGB9N JIbLSbO WC61/6/2 *4IIGD “OONGNYS4 NHS “ANDISIH BWIL - ALIQOIIA Z'2°914 938 “BHIL ef &e he 02 at @ h oO + on- oz- 8 a to 2 oz on Mop bed ded tet bear ba tent dg12 QYO93Y O3LSNPOWNN - MGSN ITbLSKD IL61/6/2 **47789 “QONUNYISS NUS “AYDISTH SWIL - LNIW39N1dSIO €'2"9I4 935 “SHIL © ee he oz a1 a e b ° +— ost oo1- os- 0s 001 _—13 QYOI3Y O3LSAFOK ATIbLNIWI3S - MBSN JI61S6I ILGL/6/2 °° 41769 “OONGNYIS NEG "AMOLSTH 3WiL - LN3IWIDWISIO b'2 914 35 “BHIL 6 ee hz oz gt at & b a + + + 06. pe~ at- 0 a oO oz BP tt tO14 1L61/6/2 QYOI3Y GBLSNFGY ATIBLNSWD3S - MGON IIb1ShI “ATTSD “DONGNYIS NUS "AYDISTH WIL - 2LTI019A Geis 235 ‘ANIL oz at a € b a ag: on- o2- oz oh *ALTOO73A 335/NiOYOI3Y GILSNFOY ATIBINSND3S - MEN JIKISKI 1261/E/2 °° 3176 “QONGNYSJ NUS “AMOISTH SWI1 - NOiLbU373906 9'2°9T4 935 ‘3HiL an e b o e we he be 1s s oH 9 "NOT 18Y973998ONT gWHO WITLTYI 40 LNIIWId S°O ~ OYOIIY OSLSNFOwNN MBON JTBLSHI - 1261/6/2 *°41 78D “QONBNYIS NUS WO4 HULIZGS ISNOdSIY YSIS sd *AQNaNDaus a°001 070s ge ol os oe ol sO 20 10 5070 20°0 10°0 16 38/NI *ALEDOT3Aft Su a SISOS eA D3S/Ni ‘411907341O7d IHL NO S3TIN3NOINI QIIHdS ATTWNDI 18 ONISN ‘ONTaWYO ‘WITLIYD 40 1N3D Y3d O'2 - SN - OBI OYINID 73 YO4 BYLI3dS ISNOdS3Y 6'2°9I4 Sd) ‘A3N3NOBUS 0°00! "os oroe alas Ogee Ont 5:0 Oe ER) 200 100 ia 1 20 (s08 14 OP Fyonsayi psso09 2 — lo ° % oe “ALID073A 18 yd » os 938/NI ovor ovoz Ca vos voor1g ONIdWHG WITLIHD LN3IWId OL ¥ *S *e “S* ‘O MhLS UWIODNd - 1L61/6/2 ‘*4I TWD “OONYNYZS NUS - BULIdS 3SNDdSIYO!?’SI4 Sd *,INaNOaH4 o'oot oros ool ors oe o1 so 20 «10 soo Oo 107 oe o's o-oot o-a02 o°o0s D3S/NI *ALTOOT3AONIdWHO TWILLIYD LN3Du3d OL ¥ *S “2 “S* *O 391S BHIDIbd - 1L61/6/2 **4ITND “OONUNY34 NUS - BY1ISdS JSNOdSIY II'2"914 S49 *AIN3NORS o-oor aos 002 ool os O:e 2 Ogle S,0 2:0) 150) = S0,0) 20°70 10°0 Ae o's ovat 20 2 ovoe D3S/NI *ALI9073A S vos j— 4 o-oo! | |e Z| o-o0221 ONIGWbO TWITLIYD LN3IW3d OL ¥ ‘S ‘2 ‘S* ‘O TWITLYBA BWIOGbd - 1L61/6/2 "0S 002 orl ors, "T4ITd9 “OONUNYS4 NUS - HYLIZIS ISNOUSIZIZ"9I4 S49 *AIN3NOIUS oz or sto 2.0 10 so’o 20°0 10" ovoe ovaor 0002 "00s D]S/NI “AL TDO73ASNIdWHO THIILTYI LN3IV3d O1 ¥ “S ‘2 *S* “O S1eN JIb1SKI - 1L61/6/2 **4ITWD “OONYNY34 NUS - HYLI|dS 3SNOdS3Y E12" 914 sd *AON3NOBUS O'001 0"0S oz ool o's , Oe OF SO 270 10 S00 20°0 10°0 of ios on *ALIOOT3A 2 ant 3 4 | o'Or J38/NT o-0e Ns o-oot o-o02SNIdWKO TWITLIYD LN3DW3d O1 ¥ “S ‘2 ‘S* ‘O MBN JIBISKI - IZ61/6/2 ** 411K “OONYNY34 NUS - BYLISdS 3SNDdS3Y bI'Z"914 SJ *AONSQOSUS O°0or 070s a2 orl Ors oz or so 20 10 00 z0°0 100 23 “ovor D3S/NI *LLT9073A o-oorONT dO TWITLTHD LNID3d OI ¥ ‘S ‘2 'S* ‘O WITLYBA JIBLSWI - IZ61/6/2 **417KD “OONUNYS4 NUS - HYLI3dS ISNOdS3Y SIZ 914 Sd0 *AINSNOBYS o'00r 0'0S oro2 orl ors oz or so 20 «10 6s0'0 20°70 10°0 7 sy o's of 4 oor O3S/NI *LLT9073A fo LAS ovoe ovos ovoor By, 1 | 0-00225 ONIdWUO WITLIYD LNIDWId O1 ¥ ‘S “2 *S* ‘O SN NNI ABOIIOH - IZ61/6/2 ‘*4ITU9 “OONUNYI4 NUS - YYLIIdS JSNOdS3Y 912914 Sd9 *AONaNORY 0°001 o'0s O02 0701 ors, oz or sto 270 10 soo 20°70 100 so vos voor "002 D3S/NI *AL19073A26 0-001 M3 NNI AMOINOH - 1Z61/6/2 00S a'0z 0°01 ors ONIdWHO TKITLIYD LN3IW3d O1 ¥ ‘S ‘2 “S* ‘O Sd0 *AON3NORUS oe ol $0 20 10 S00 20°0 “*41789 “QOONUNY34 NUS - YHLISdS 3SNOdS3Y LI'2°914 Ze 3 © oe os ovor ovoz ovos o-oor o'00z “ALTQO73A I3S/NIONTdNBO WWITLIHD LNIDW3d O1 ¥ ‘S ‘2 ‘S* ‘oO WWITLY3A NNT AWOITOH - I1Z61/6/2 * “AITWD “QONUNYIS NUS - BYLIZdS ISNOdSIYSI'Z"914 20°70 10°0 NS 20 o-oor o-a0e 93S/NI *AL19073AONIAWUO TWITLTYD LN3DWad Ol ¥ *S “2 *S* ‘O STIN GAIG BYNINSA ~ IZ61/6/2 *" 41169 “OONYNY34 NUS - WYLISdS ISNOdS3Y 6I'2"914 sd) *AONINOBY o-00t 0's oroz 0701 o's Orc) Ogle 18:0) 20 10 S00 200 10°0 3S/NI *LL190730ONIdWHO TWITLIYD LNIWWId OLY ‘S ‘2 ‘S* ‘0 MBLN GATS BYNLNIA ~ IL61/6/2 ‘*4I7H9 “OONUNYI4 NUS - BYLI3dS ISNOdS3YOZ?"I14 S49 *AONINOHS 0001 0°0s (O;0e8 9301) | 20:5) Once 2030 5:0 20 =10 S00 _ o-oot 0-002 O3S/NI *LL19073AONT dWHO THITLTYD LNIIY3d OL ¥ *S ‘2 *S* ‘0 TWITLYIA GAIA BYALNA - I1L61/6/2 *" 41ND “OONUNYZ4 NUS - HYLIIdS ISNOISIY IZ’ 914 iQ ee nO) 2 SUR0) zoo 100 2"0 0001 0"0s o02 o'ol Ors oe 6 3S/NI *LLEI073A31 ONTdWHG TWIILIYD LNIDWAd O1 % “S ‘2 “*S* ‘O SN - ORGI/BI/S **4ITH3 “OYIN3D 13 - BYLISdS ISNOdSINZe2"II4 S49 *AINSNOHS 0001 0's O02 0701 ors oz or so fle 11,0) | 50,0) 20°70 10°0 os o’cot 0-002 3S/NE *ALI9O73AONIdNHO WWOILIYD LN3I¥3d O1 ¥ 'S ‘2 ‘5° “O M3 - OhEL/BI/S “*41 iD “OYIN3D 43 ~ bYLDSIS SSNOdSIEeS IIS sd9 *AON3NOBS re soo 200 10W 1 20 0001 070s oro2 ool O's a z°0 x | “ a / : Va NS > . A Pb PN SEL A AL arte #0 7 ORR SIO PSP SI PRON _ \. LOSS % < 7 : il SAN re N a a Zs » \Y LPR NE SS ODT [<< ove S / _. \ lA y Y” aa NZ SAW ~ 3 ay . X Sy SY” ‘ DAS/NE *ALIDOT3A33 ONT dWbO WWIILTYD LNIDWId Ol ¥ *S WOILY3A ~ OHEI/8I/S **4I IND “OYLNIO 13 - YYLIaUS S49 *AIN3NOBHS "2g: 3SNOdS, ‘O Bube'2" O14 OOO! 0's o'o2 0701 ors oz ot sto ye DAS/NI *AL13073AONIdWHO THITLTHD LN3DH3d O1 ¥ *S *2 “S* “O SN - SS6I/6/2 **4I1H3 “OYIN3] 13 - BYLI3dS ASNOdSIYUS?T IIS S40. *AIN3NOBES "001 00S oroz2 ool ors oz or sO 270 10 S00 20°0 1070 s0°O D3S/NI *LLI9O73ASNIdWHO WWILLTHD LNIDW3d O1 > Ge we I) *“*4D789 “OYLN3D 73 - BYLIIdS ASNOdS3Y9Z'2"9I 4 z20'0 10°0 \ 35 s0°0 D3S/NI “AL T9030ONIdWYO TWITLTYD LNAIW3d OL ¥ *S ‘2 *S* ‘O TWWITLWIA - 9S6I/6/2 **4TWD “OYLNID 13 - wYLIadS ASNOdS3H22'2° 914 Sd *AONINOSS 0°00I4 O°0S o°o2 ovol ors o2 or so 2010 so"o 20°09 10°0 DBS/NI *ALIIOT3A7 ONTdNBO TWITLTYO LNIIWad O1 YS “2 “St ‘Oo SN - B961/8/h ** 31749 “OYLNID 73 - BYLIIdS ISNOdS3YeZ'Z"9I4 sd “AONaNOS or so R ‘ " "0 10'0 oe ovor 008 o0s o-oor O3S/NI *AL19073ASNTGHO TWIT LTyD AN33U3d O01 ‘5 "2's" +9 M2 - 8961/8/h “4rq3 “OUND 13 - bYLIadS ASNOISHUEZZ"9T 4 Sd9 *AONaNORES O02 orl oe i Ol so a0 93S/NT “4119073439 ONTdWHO TWOTLTHO LNIOW3d OL ¥ “S “@ ‘Ss tO WOILY3A - 8961/8/h **4I7THO “OYINZO 1a - BYLIZdS ASNOdS3YOeS" 14 Sd *AONNOSdS O*00 o'0s o'02 O'Ol os oz of so 20 10 sovo 20°0 100, 93S/NI "41190730ONTdWHO TWOTLIYD IN3DH3d O1 ¥ “S ‘2 *S* ‘Oo SN 1WS@ 91S OMIH - 2S61/12/L ‘*4I7¥9 ‘OOOMATIOH - BULIAdS JSNOdS3Y I¢'S"9T4 Sd9 *AININOIUS "001 o*os 0°02 orol ors oz on sto 20 10 sovo 20°0 to" 0 to O3S/NI *4119073AONTdWHO WIILIWD LN3IW3d Ol ¥ *S ‘2 *S* Sd0_ *ADN3NOBES (Oc1= | Ogle 15,0) ‘Oo M32 LWS8 91S OGMIH - e2S6I/12/L **411W2 “GOOMATIOH - BYLI3dS ISNOdSIWZES II 4 DAS/NI *LLTD073AONIGWKO TWITLIHD 1N3IW3d Ol ¥ *S ‘2 *S* ‘O WILLYIA IWSE 91S GMTH ~ 2S61/12/L **4ITK “OOOMATIOH - BYLIAdS ISNOdSBUEE'2"9I4 S49 *AINENOBUS "ool gos oroz orl ors oz or so 20 170 so°0 z0°0 1070 33S/NI ‘4119073443 ONTdWHO WIILIYD LN3IDY3d OL ¥ “S ‘2 “S" “OD SN 1O1 3d OLS OMTH - 2S61/12/L “SEHD ‘OOOMATIOH - beLIFdS ISNOdSIMHES IIS Sd9_ *AIN3NDIYS o'oot 00s oe ool ors Dice) 02 5:0 20 10 S00 200 too KK “ol “02 0s D3S/NI *ALIOOT3N4 o-oot M3 107 3d 91S OMTH - 2561/12/2 “"4ITH9 “OOOMATIOH - BYLIFdS ISNOSIUGE'S Id "0s 002 ovat os ONT dWHO TWITLTYD LNIIWad O19 “S “2 *S" ‘O 0 SdJ_ *AINSNOBES 2 oo so 270 a vo 20 so oe os ovat ovoz 070s *ALIOOT3AONT dWHO TWILL 1NIIW3d O1 ¥ *S ‘2 ‘S* ‘oO “LUBA 107 3d OLS OMTH = 2561/12/L ‘74189 ‘OOOMATION - 0°00! 070s 05023 030%) 20;9) BYLI3dS ISNOdS3N9E'S II 4 Vo sovo z0'0 1070 RRR RRR Nd oN KK O3S/NI *LL190734ONIdWbO WITLIYD LN3DW3d OL ® *S ‘2 “St ‘oO 30IN Md 3LY9 NOID - LS6I/22/E **4mWO “QISTINKYS NYS - WYLIadS ISNOdSIYZE'Z"OI4 Sd9_*AONINOBUS 07001 o's 0°02 Orr ors oz or so i012 Oe 180.0 20°0 10" 0 soo 20 ovat D3S/NI “41190734 — aeSNTdWHO WWOTLTYD LN3DU3d OT 9 *S cc.Ge 0) “41183 "O9STINY4 NUS - BYLIFdS ISNOdSIUBEZ™II 4 308S Yd 31¥9 NOI - LSBI/22/¢E Sd *AIN3NO o2 or so GO | eat 20°0 10°0 s0°0 20 ovat D3S/NI “4119073448. “LUBA Md 3Lb9 NOID - LSBI/22/E 0001 o's O02 or os ONT dW TWITLIYD LN3IW3d OLY *S ‘2 ‘Ss ‘oO “*41983 “OISTONWYS NUS - bYLI3dS ASNOdS 3U6E'2" 914 Sd “ANAND o2 or so 2:01 11501 = (50,0 200 10°0 so°o O3S/NI "411907340'001 "0s SONI dNuG MORN - ISBI/L/01 ** Si o'o2 o'0l ors oe J LN3IW3d O1 ¥ *S “2 “S* ‘O STWONYSS - bYLISdS ISNOdSIYOH'?"9I4 14 so en 21,0) 280;0) 20°0 10 0 2-0 D3S/NI *LLT9073A50, ONIdWHO TWITLIYD LN3ZIW3d OT ¥ “S ‘2 “S* ‘O MhhS - [S6L/L/O1 **4T18 “JTVONYSS - BYLIFdS ISNOdS3Y Ip'2'914 Sd0, *AINBNDIES "oor 0'0S oo2 orl O's o2 or SO 250,12 0 50:0) 2070 100 so o-oot D3S/NI *ALI9073AO3S/NI “kL 190730 50.0 100.0 10.0 20.0 1.0 2.0 5.0 0.5 0.2 FREQUENCY, CPS 10/7/1951 - VERTICAL IMP ING. - FERNDALE, CALIF., FIG.242RESPONSE SPECTRA +S. 2, 5, & 10 PERCENT CRITICAL oAI a,ONIdWHO TWIILIYD LN3IWad OI ¥ “S “2 *S* ‘O MORN ~ ASB6I/12/21 $40 *AONINOS 0001 00S 002 o'01 ars oz o1 so Ue s0"0 20°0 “41182 “FTBONYS4 - BYLIdS ISNOdSIUEP?"SI4 20 so on os ovor 002 ovos o"n0r D3S/NI *ALIIO73AShHN - RS6I/I2ver * Sd *AQN3NOSY4 SNIdWHO WWOTLTYD LN3OU3d OL ¥ *S ‘est “4ITWD “3TWONYS4 - BYLI3US ASNOdS IU z"914 ont / 53 D3S/NI *4L1907345h SONI dO WWOTLTYI LNIDY3d O1 ¥ *S ‘2 ‘S* ‘oO TWITLYIA ~ WSBI/12/21 **4I WI ‘FTYONYRS - HYLIIdS ISNOdSIUGHT OIG Sd9*29N3NOFUS 0001 o"0s Ce Ogee 0a 1215:0 20 10 soo 20°0 10°0 9001 9]S/NI *ALIDOT3AMIIN - hS6I/12/21 ONTdWHO TWITLIYD LN3IY3d O1 ¥ “S ‘2 “*S* “*4TIW3 “BMBYNA - BHLIdS ISNOdSIUSPe"II4 $d *AON3NOFES wre ot S70 55 “ALTD073A 3S /NIONIdWHO TWITLIYD LN3DWAd OL ¥ “S “2 “S* “O B6LN - ASBI/12/2t “41D “H3YNA - HYLIAdS ASNOdSIY/H2°914 Sd9_ *ADNSNOBES r2 Or St 3S/NI “41190730ONTO TWITLTHD LNIWId O1 ¥ *S ‘2 “S* ‘O WITLYIA = HSBI/12/21 ** 417K ‘wW3UNE - YULIFdS ASNOSIUSYZII4 Sd “LONNOIUS 0°00! 0°0s 0°02 o'ol o's oz ot sto a) 9 so"o z0°0 10 eo, 1o°0 0 ss 20 so D3S/NE “41190734ONTdWHO TWITLIYD LN3IW3d O1 ¥ *S “2 *S* *O MIOS - [961/8/h **4I 1H “W3LSTTIOH - BY1I3dS JSNOdSINEHT IIS sd9 *AIN3NORS 07001 0"0S (O02) 10:01- | 055 oz of S70 270 10 S0"0 20°0 1070, T D3S/NI *ALIIO7RA “ol iOee “os “ool39 oor 00S 0702 SONTdWHO TWITLIYD LN3DWad O1 % “S “2 *S* ‘0 M6BN - I96I/8/h ** S40 *AONINOSY vor ors oe SIH “W3ISITIOH ~ BYLISdS JSNOdSIYOST 914 ot so pei. to so°0 20°0 10°0 vA 2"0 os ovor <—| 0°02 o'os o-oor J3S/NI *ALIIOTBAONIdWHO TWOTLTHD LN3QW3d Ol ¥ *S ‘2 *S* “O JUOTLUIA - I961/8/h **4ITWD “WRIST TIOH - BYLIAdS 3SNDdSIY IST IIS $49 *ADNINOBYS "001 0°0S o°02 Ovo Ors oz ot so 20 170 S00 2070 100, a 8 2 361 3. PRESENTATION AND DISCUSSION OF RESULTS imum values of ground motion obtained from the adjusted acceleration, velocity, and displacement tine-histories for the 3 components of the 14 earthquake records considered are given in Table 3.1 along with a brief site description, It should be recognized that although an attempt was made to characterize the site descriptions as rock, alluvium, or otherwise, these descriptions are not completely dependable and accurate owing to a lack of satisfactory information about the geologic conditions at most of the sites where strong motion instruments have been located. The nondimensional ratio ad/v’, where a, v, and d are the maximum ground acceleration, velocity, and displacement, respectively, is also given in Table 3.1. The ratio ad/v* is at least i part a function of the focal distance of the earthquake and the attenuation of motion in the ground. (©) For very large focal distances, where the ground motion approaches @ harmonic motion, ad/v? approaches 1.0, The ratio increases as the focal distance decreases. The range of ad/v’ for the earthquake records considered in this study Is 1.84 to 17.61 for the horizontal and 2.66 to 30.58 for the vertical components, respectively. The average ad/v* for the 28 ho: ontal components and the 14 vertical components are 5.6 and 10.6, respectively. Among the 42 conponent records considered in this study, only one conponent (vertical component of El Centro, 5-18-40, 2037 PST) has an ad/v* which is extremely high. The ad/v? for this component is 63 percent higher than the next highest value of the vertical components. If this highest ad/v* value is not included, the average ad/v* for the remaining 13 vertical components reduces to 9,1. A summary of the average ad/v* values for horizontal and62 vertical components for alluvium and rock is presented in Table 3.2. Further discussion of the data presented in Table 3.2 and particularly with reference to classification by acceleration level appears in the next section. Since the three components of the ground motion in a given record for a given site have the same focal distance, the difference in ad/v? values probably arises from attenuation of the ground motion and Tocal geological effects on transmission of the ground motions. The ratio ad/v* provides some bound on the relative magnitude of ground motion. For most earthquake records, the ratio ad/v" is between 5 to 15. (8) this ratio is important in selection of ground motions for use in constructing the design spectra as discussed next and later in Chapter 4, However, a knowledge of only the applicable or chosen ad/v* value is not in itself sufficient for constructing the basic ground motion bounds which are required in the construction of design spectra. In addi ion to the characteristic ad/v? values, one needs to know the absolute values of two of the three ground motion quantities (acceleration, velocity, displacement). Knowing two of these three quantities, plus the ad/v value, the remaining quant ity can be calculated. It is convenient for scaling purposes to base the acceleration value on an acceleration of 1.0g. The construction could proceed for any other desired value. Moreover, because of the way i obtained and knowledge arising froma number of years of experience with the construction of spectra from earthquake and other shock-type loadings, it is normally customary to work with the quantities of velocity and63 acceleration, and the relationship between them. One reason for this is that very few good measurements of displacement have been recorded, and the least is known about this quantity. In an effort to arrive at the interrelationship between values of velocity and acceleration, a number of studies were made of the data. The salient features of these studies are recorded in the tables which are described briefly in the following. In Table 3.3 there is presented a summaiy of the ratio of the velocity to the acceleration for the horizontal and vertical components of the records. These values can be obtained directly from the data in Table 3.1. Among the 14 earthquake records considered in this study, the v/e ratios for one record (San Francisco Golden Gate Park, 3-22-57, 1144 PST), both for the horizontal and vertical components, are about 47 percent lower than for the next lowest record's horizontal and vertical components. By not including the three v/a ratios for this record, the average v/a ratios for both alluvium and rock, and rock alone, increase slightly. The summary of the average v/a ratios, based on the data presented in Table 3.3, is presented in Table 3.4. Shown in that table are the various ratios, including: (1) all of the 28 horizontal components for alluvium and rock; (2) 26 horizontal components on alluvium and rock with two components not included, as just discussed; (3) alluvium alone; (4) rock alone; and (5) rock alone with two values not included. Other comparisons are presented for alluvium and rock for all horizontal traces where the maximum ground acceleration is greater than 0.1g, etc, Similar comparisons are presented in Table 3.4 for vertical components.64, In Table 3.5 is shown a summary of the vertical to horizontal v/a ratios. Considerable study indicated that the most representative values could not be obtained by averaging numbers of the type shown in Table 3.4, but instead individual trace data, as shown in Table 3.3, were averaged. On this basis, as noted in Table 3.5, the ratio of the velocity to the acceleration for the vertical traces compared with that of the horizontal traces is on the order of 0.90, and ranges from 0.81 to 0.99. Similarly, in Table 3.3 is shown the ratio of the peak ground acceleration from the vertical traces to that for the horizontal traces; these values are shown in the right-hand two columns of Table 3.3. In order to obtain 2 measure of these ratios, which are needed for scaling purposes for the vertical spectra, studies of the data for various classes of traces based on site characteristics are presented in Table 3.6. Tt will be noted that the values range from 0.40 to 0.72, depending on the acceleration level included and the type of site, etc. On the basis of these data, and reali ng that only a few samples existed for the rock sites, we selected for use later a value of 2/3 as a ratio of vertical to horizontal acceleration. 3.2 Response Amplificat tons Studies of the response amplification in the various ranges of frequencies were made by stud: g plots on the four-way logarithmic plot, with the input ground motion normalized to the following values for all of the earthquake motions studied: (1) A maximum ground displacement of 1.0 inch (2) A maximum ground velocity of 10 in/sec65 (3) A maximum ground acceleration of 1.09 The amplified values for the various frequen for which calculations were made are shown in Figs. 3.1 to 3.16. Separate plots are drawn for vertical and for hoi ontal components for values of damping of 0.5, 2, 5, and 10 percent of critical. The figures are given for the mean (normal distribution) response and the 75 percent level (ranking) of response. For other probability levels the plots would have generally the same characteristics. It can be seen that the amplifications for displacement are for all practical purposes constant over the range of frequencies from about 0.10 hertz to some upper limit of the order of 0.4 or 0.5 hertz. It is also clear from the figures that the amplification factors for acceleration are virtually constant for the range of frequencies from about 2 hertz to about 6 hertz in the horizontal responses and from about 3 hertz to about 10 hertz in the vertical responses, and then decrease fairly uniformly to intersect the ground motion accelerations at frequencies of the order of 30 to 50 hertz, varying with the direction and the damping factors. For the intermediate range of velocities, the horizontal amplifications appear in general to be statistically practically constant in the range of frequencies from about 0.5 to 2 hertz or more, al though there are occasional slight roundings with @ higher value in the central portion, but there is no basis for using other than a simple constant amplification level in this range. For the vertical component, there i evidence of 2 slight drop in amplification factor as the frequency increases in the intermediate range, but the drop is only of the order of about 10 percent over the frequency range for which "velocity66 amplification" is valid, and it therefore appears reasonable for practical Purposes to use also a constant amplification for the velocity range, as well as for the displacement and accelera n ranges. An examination of the response spectra shown in Figs. 2.10-2.51 indicates that for most cases the response spectra for vertical ground motion are flatter (broader) than those for horizontal motions. Careful examina n will reveal that the response spectra for ground motion in the vertical direction approach the ground acceleration at higher frequencies than the case for the horizontal direction. This phenomenon is due mainly to presence of higher frequencies in vertical components of 7) ground motion. ‘/’ Therefore, in constructing a design spectrum, it seems reasonable to amplify the vertical ground acceleration over a larger frequency domain than for the horizontal ground acceleration. ‘An examination of response amplifications, especially the amplifications for velocity and acceleration, indicated that earthquakes with low ground acceleration have greater amplifications than those with @ high ground acceleratt jeaks in an accelerogram reduces the response amplification for high frequencies. This is demonstrated In Figs. 3.17 and 3.18. Figure 3.17 shows the adjusted acceleration-time history for £1 Centro, 5-18-1940, NS component with the first and the second peaks scaled such that the maximum ground acceleration is 1.0g. The Fesponse spectrum for 2.0 percent of critical damping normalized to ground acceleration for this accelerogram is compared with the one without fletitious high peaks, Fig. 3.18, It is seen that for high frequen: the amplification for acceleration is lower for accelerogram with high peaks. The difference between the two curves for intermediate and low67 frequencies arises from normalizing the response spectra to only the ground acceleration over the entire range of frequencies (see Appendix B for further discussion of this point). For this reason, the statistical values of response amplification were computed for two groups of data. One group included all data, whereas the other group included records with peak accelerations greater than 0.1g and 0.059 for the horizontal and vertical components, respectively. The records which were included In the latter group are as follows (see Table 3.1): Pacoima Dam, 2-9-71, 0600 PST (a11 components) Castaic, 2-9-71, 0600 PST (all components) Holiday Inn (First Floor), 2-9-71, 0600 PST (all components) 15250 Ventura Boulevard (Basement), 2-9-71, 0600 PST (al! components) El Centro, 5-18-40, 2037 PST (all components) El Centro, 48-68, 1830 PST (NS component only) San Francisco Golden Gate Park, 3-22-57, 114 PST (all components) Ferndale, 10-7-51, 2011 PST (NHGW and SliW components only) Ferndale, 12-21-54, 1156 PST (N46W and NAWE components only) Eureka, 12-21-54, 1156 PST (211 components) Hollister, 4-8-61, 2323 PST (N89W and vertical components only) The statistical values of response amplifications correspond ing to 50, 75, 90 and 95 percentile levels obtained from a normal distribution curve for 0.5, 2.0, 5.0 and 10.0 percent of critical damping are presented In Tables 3.7 and 3.8 for all 28 (horizontal) or 1% (vertical) components along with corresponding amplification values for those horizontal and vertical components with peak ground accelerations greater than 0.1g and 0.05g, respectively. These amplifications were obtained by first compu68 the mean and the standard deviation of amplifications for isplacement, velocity and acceleration at each frequency for which response computations had been performed. From the mean and the standard deviation at each frequency, amplifications corresponding to various percentile levels were determined. The displacement, velocity and acceleration amplifications y band) were then averaged and are presented in the tables. The frequency bands used for averaging the amplifications, which were established, through examination of the spectra, are as follows: Horizontal displacement 0.2 to 0.4 hertz Horizontal velocity 0.4 to 2.0 hertz Horizontal acceleration 2.0 to 6.0 hertz Vertical displacement 0.1 to 0.3 hertz Vertical velocity 0.3 to 3.0 hertz Vertical acceleration 3.0 to 10,0 hertz Anplification values similar to those just described but obtained from ranking the data are presented in Tables 3.9 and 3.10. The amplifications given in Tables 3.7 to 3.10 indicate that for records with horizontal and vertical ground acceleration greater than 0.19 n amplifications are and 0,05g, respectively, the velocity and acceleral lower than the corresponding values obtained by including all data. In other words, the strong motion data clearly indicate @ decrease in amplification, especially for the velocity and acceleration regions, ‘as compared to the case where low intensity excitation is included.