Eathqushe Engineering, Toh Word Conterence ® 1862 Baker, Rotterdam. ISBN 90 5410 0605 Seismic hazard analysis in the Philippines using earthquake occurrence data G.L.Molas & EYamazaki Instn of Indusrial Science, The University of Tokyo, Jepan Y-Tomatsu Techicol Research Insitute. Nishimatsu Construction Co. id. Japan ABSTRACT: The seismic hazard in the Philippines is evaluated from historical earthquake data using a new ‘computer program called the Seismic Hazard Mapping Program (H-MaP), The seismic hazard is given in terms of the expected peak ground acceleration and the expected acceleration response spectrum. Regions of high seismic hazard are identified. These include Central Luzon which suffered heavy damage during the 16 July 1990 earthquake. The design levels ofthe seismic force of the Philippines ae compared with those of Japan and are found to be considerably lower. Long period structures are found to be more vulnerable to damage. The collection of, ‘wong ground motion records from Philippine earthquakes is necessary for more realistic design levels for the Philippines. From the seismic hazard maps, a seismic zoning map based on the expected maximum accelerations is proposed. 1 INTRODUCTION ‘The 1990 Luzon earthquake disaster exacted a heavy ‘oll in the human, economic, and social resources ofthe Philippines ata time when she was struggling for devel- opment. The first two authors joined the investigation teams ofthe Architectural Intte of Japan (AU) and of the Japan Society of Civil Engincers (JSCE) which surveyed the damage of the earthquake a month after it ‘occurred ‘The earthquake affected a wide area and many seismic hazards like building and bridge collapse, landslides, and liquefaction were observed. Building collapse was the main cause of human casualties. Although there were two isolated cases of building collapse near the epicenter, heavy damage was observed in some areas Which are more than 100 kilometers from the epicenter. Since no strong motion records of the main shock were taken it is difficult to quantify the main causes of the structural collapse ‘To address the need to mitigate earthquake disasters in the Philippines, aseismic hazard stady using earthquake ‘occurrence data was made. It is hoped that this work ‘can help engineers o beter understand the magnitude of hazard or rise faced by natural and man-made structures in he Philippines. For engineering purposes, earthquake monitoring data in the Philippines is severely lacking. Even the seismic design provisions were adapted from the United States. But without actual data, it is very difficult, Impossible, to assess the Tevel of safety that is being ‘sed for design, This study aims to assess the seismic hazard in the lippines and its implications to the seismic design Pk 601 code. By using earthquake occurrence data, the task of Identifying seismic sources and their parameters is avoided. Due to the lack of data, this task has inherent “uncertainties. It is acknowledged that the simple model used in this study has limitations, but the general trend in the regional seismic hazard can be immediately scen. It is believed that this work can help engineers make ‘more sound decisions regarding seismic design. 2. EVALUATION OF SEISMIC HAZARD FOR ONE, SITE There are two general types of seismic hazard methodologies. The first method uses the historical uake occurrence as pioneered by Kawasumi (1951) and applied recently by Tomatsu and Katayama (4988), The second is based on probabilistic principles pioneered by Cornell (1968). The probabilistic approach assumes seismic sources as being points, lines, OF dipping planes. Seismic parameters such as seismic activity and the maximum probable magnitude are assigned i each source. Duc to the complex nature of the seismicity in the nes and the uncertainties in identifying the loca- tions of seismic sources and in assigning seismic ‘parameters, the method based on historical earthquake data was chosen for this study. ‘The seismic hazard is ‘based primarily on the assumed relationship of the mean annual occurrence rate, v, of a peak ground motion equal oor greater than a given value as: log v=a+b logy awhere y_ is the peak ground motion, and a and b are regession consianis. . "To evaluate the seismic hazard ata given site, earth- quakes within a given epiceatral distance from the ste te chosen, For each eartiquake, the value ofthe peak {ground motion Is estimated using atteavation laws. A fegresion analysis is then performed to determine the values of a and b If the cccurresce of peak ground motion values is ‘assumed to be & Poisson thea the probability of ‘Eoceurtences of the peak ground motion in f years is tiven by: niga whet C) “The probability of no occurrence in years is then: PO.n=e™ ® ‘The value ofthe pea ground motion for a given noa- exceedance probability, Q, and me period, f, i then given by: logy ={log(-in Q/1)-0)/6 o For this type of analysi exceedance probability and the retum period of peak ‘round motion is given by: Te-1/InQ © ‘Thus he 10% probability of exceedance in $0 years is equivalent afetun period of about 475 yeas, ‘ua eaalogues from the Pilippne Istiute of Vol- ancogy and Seismology Pivots), the Iterations Scismologieal Center (SC) and the U.S. Geological Survey (USOS) were collected and analyzed. for completeness usiog the method proposed by Stepp (1992) The culalogue is deemed fo be complete for & sven magnitude category ifthe slope of the standard Acvinion of the mean aanual occurence rae for he Period is close to the slope of the model line. The Catalogue iaken fom be USGS dais et was found 1 be amplete for magnitudes 5.0 107.0 from 1963 191990 (Gigue ) Thas, ds dataset was sed forthe analy ‘The plo of epiceters ofthe data weds given in Figure 2 “The attenuation law for peak ground acceleration (PGA) used in the analysis was developed by McGuire (97H) as: 20.482 POM (4 25713 a where ate pk round acleraon ing. te Rassiode, andr te bypocenval dance a T the acceler ote accleraon response spectrum of « 3% arp yt eaten nw Wed a ee by Katayama (1982) as: erst SAT 0.05) = fur fa -foo © sg ances Bats foatess iS etcen eoatees eat james wate Host IIH tin, T (ear from 1950) Figure 1, Standard deviation of the mean annual occur rece rte with respect to the sample period. enters of the data used in the Figure 2. Plot of 3-190). analysis (USGS, 1 where fs fas and foc are the weighting factors forthe magnitude, epicentral distance, and ground condition, respectively. The weighling faciors are dependent on the peciod, 7. In the absence of stvong ground motion records, 2 survey of fallen objects in the affected area can give rough estimates of the maximum accelerations expeti- ‘enced at diferent sites, The upper and lower bounds of the maximum accelerations ‘can be estimated by calculating the aspect ratio of simple objects like tombstones. During the damage survey of the JSCE, fallen objecis with simple shapes were used to estimate the maxlimum accelerations. ‘These included gateposts, highway markers, furnitures, and religious satues. A summary ofthe results is given in Figure 3. 6082Eber icfes>> Baaenttomtooo ‘Stone Dine tom Fal) Figure 3. Estimated maximum accelerations from field survey and theoretically calculated resuls. 3 REGIONAL SEISMIC HAZARD ‘Anew computer program called the Seismic Hazard Mavping Program (H-MaP) was developed and used to compute the scismle hazard parameter. The selamlc hecard value was calculated for 13,230 points Ia the Philippines and this can be dsplayed as color taps or sescaow moot, The pane nib cog are the retum period o (zound motion an probability of exceedaace for a given ime period. By tring color codes, the regional differences inthe selamle hazard ean be Immediately soca. Preliminary analysis showed that as the ite belog ‘evaluated goes nearer a seiamle source (represeaed by Sense cusier of earthquakes, the value ofthe expected PGA becomes very lage. This was caused by the non- linear tendency ofthe reladonship between the PCA and the occurrence rate in the logeog scale (Figure 4), Since the concern of seismic hazard analysis are the regions of high PGAs, a cutoff scheme. for the repession a a mean annual occurrence rate of 2.0 was oe Figure 5 isa contour map ofthe 100-year peak ground acceleration (PGA) in eas? for the country. By ‘comparing the hazard map with the plot of earthquake ‘ploenters ofthe data used, It can be seen that higher Stismic hazard areas follow a band corresponding 1 the farthquake generators Inthe couniry. The highest Seismic hazard can be found in regions where weveral Shallow earthquakes have occured. These regions, with 00-year PGAs greater than 200 emv/s2, are ia Cental leuzon, Southern Luzon, and the peninsular portion of Western Mindanao, Central Luzon experienced heavy damage during the Luzon earthquake. To check I is High seismic hazard is caused by selsnic scuvty related toibat earthquake, the analysis was repeated forthe data cxcling the 1990 earthquakes. Iwas found thatthe ‘egion of high selmlc hazard el exits, Figure 6 shows the hazard in terms of the 100-year acceleration response spectrum forthe period T= 0.7, Peak pound wedenon enhectes) Figure 4. Regression of the mean annual occurrence Fale and the PGA for a site near a selsmic source. Figure 5, Contour map of 100-year peak ground accel- zation (PGA) ing. 4 CONSIDERATION OF EARTHQUAKE FORCE IN DESIGN ‘The Philippine selanic design provisions are patemed after the Uniform Building Code (UBC) of the United States. Unless dynamic analysis is performed, the Tera! seismic force applied tothe siructure is evaluated by calculating the base shear using the modified seismic ‘coefficient method as: Base Shear /W = Z- C- S-1- K © ‘where W isthe weight ofthe structure; Z, the zoning factor; C, the response factor; S, the soll facto; I, theFigure 6. Contour map of 100-year acceleration re- sponse spectrum for period, T=0.7e, and 5% damping. Importance facior; and K, the structural type factor. “The seismic coefficient is dependent on the fundamental period ofthe sructure andthe fundamental petiod of the ground, The selsmic coefficient for Ordinary structures on rock slies for different fundamental periods of the structure and the seismic coefficents for the Japanese bullding and bridge codes sxe ploued in Figure 7. ‘When the coefficients are compared, it can be seen thatthe eeismie coefficient for the Philippines is consid- cerably lower than that for Japan. ‘This i especially rue Tor structures with fundamental periods from 0.25 10 about 1.25 where the difference is from 0.05 10 about 0.06. For short period structures, the response of the suructure is close 0 the PGA of the ground movon, In {his case, the maximumn value ofthe seismic coefficient 0.14, ‘in longer period structures, the seismic coefficient for the Philippine code decreases drastically. By taking the hazard mape in terms of the acceleration response spectrum for several periods, the values of the seismic coefficients and the expected respoase canbe ‘A survey of buildings in an area in Baguio Cy ‘revealed that most of the buildings which were severely damaged or collapsed are those with seven to nine stories, A summary of the percentage of damaged bulldings with respect to the numberof stores is shown, in Figure & Figure 9 shows the distribution of the return periods for exceeding the design seismic coefficients. Ttean be seen that there are some regions in the country which have a high probability (Le., low retum petiod) of ‘the design seismic coefficient. "For these 030 02s ous 00} 0.05 000 00 05 10 15 20 25 Structure period, T (see) Figure 7. Design horizontal seismic coefficients for the Prilippine building code and the Japanese building and bridge codes (rock sts). Horizontal seismic coe 3.0 © Negligible wo slight = Moverate = Severe: ~~ Percentage of damaged buildings oseysseusss foo 4s 6 TD Number of sores Figure 8. Percentage of damaged buildings in Bagulo CCiy with expect tothe number of stories. teplons, an increase in the. value of the seismic coefficiensis indicated. 5 SEISMICZONATION Results ofthe previous sectlon showed that by using the current design codes, structures built In some regions of the country may be under-designed based on the expected elsmle forces. One of these regions is Cental Luzon where heavy damage was experienced during the Luzon. For these regions, a higher seismi ‘coefficient shouldbe used. ‘From these results, a new seismic zoning map based ‘on the 100-year PGA Is proposed (Figure 10). Table 1 fives the mean and standard deviation ofthe expected POA for Zones? to 4. Zone 1 is historically considered szelsmle and the PGAs were not calculated for this, region, Table 2 gives the mean and standard deviation forthe retum period for exceeding the current design selsmle coefficients. Zone 4 corresponds to a retum period of about 40 years; Zone 3, from 40 to 200 years: 4nd Zone 2, more than 200 years.ean Fed ‘ous = Figure 9. Return periods for exceeding seismic coefl- cient of the Philippine code = 1.0, K= 1.0, rock si iy taking the mean of Zone 3 a8 a reference, he 2000 facors canbe comped by noraaiing the ean of le expected PGA wil respect to Zone 3. The resulng {oe fects are given ia Table 3. shuld be noted, however, that the values ia ‘Teles I and 2 were computed from the elie analysis area which also Includes the sea. Curenly, «scheme for escalating ‘ly forthe land portion Isbin sted. ‘igure 11 isthe curentseisnle zoning map. While the Cutent zoning map bas lee zones the proposed ronaton has fours Zone 4, which represents We highest expected PGAS, is assigned the highest value ofthe Sato tn : ‘By comparing the two zone maps, several obeervatons can made. For boi tape, Zone 1, which ‘eprsetsssiamie region ate county, are identical. ‘region tn Ceatal Visayas ia ssgned as Zone 2 in toh imaps, Por the cureat map, most of the remalalng reps aiid te 3 i" ne main contradiction fe the penisslar pon Westen Minas whieh i aged Zoue ae sew map. is asigned as Zone? i Be cute map. Some portions of Northen Luzon also have the same sation vean lo be noticed tht de capt, Metz Manila, i assigned as Zone 2. Manila bas many medium and highrise stetues therefore, it must be cautioned that the proposed seismic zone map is for short period Suustures on rock sites. As shown In Figure 6, te Sinsibation of the selami hazard is diferent for long vid suctures. A study ofthe response spectrum for Philippine eartiquakes is seeded fora possible change inhe shape of design specu. ‘Table 1. Mean and standard deviation of the 100-year PGA for zones 210 4. ‘ZONE Mews cout) Susdard deviation Gl. “66.7 19. 3 1422 268 4 2699 614 ‘Table 2, Mean and standard de tion of the retuen of exceeding the design selsmie coefficiens. ZONE ‘Standard deviation (year) 2 2446.1 4379.5 3 ' 436 4 264 79 ‘Table 3. Zone factors for selsmic coefficient comput ton, ORE ane Frio, 1 Os 2 07 3 10 4 1s. 6 CONCLUDING REMARKS Anew computer program was developed to calculate the seismic hazard in the Philippines using historical ‘earthquake occurrence. The selamic hazard was calcu- lated in terms ofthe expecied peak ground acceleration ‘and the expected acceleration response spectrum. Regions with high seismic hazard were identified. ‘These included Central Luzon which was heavily dam- aged by the 1990 Luzon earthquake. "A comparison of selseaic coefficients ised for design showed that the design levels in the Philippines are ‘considerably lower than that in Japan. Based on the Ihazard maps and the seismic coefficients, anew seismic zoning map is proposed, The new hazard map has four Zones and the value of the highest zone factor, Z, is given as 1.5. The new zoning map is based on the expected PGA on rock sites; therefor, it is applicable for shon-pesiod structures, ‘Several bulldings with seven to nine stores in Baguio City experienced severe damage or collapse, This is smllar to damage in Manila during the 41968 and 1970 earthquakes. The seismic design provisions of the Philippines are pattemed after the UBC code in the United States. This stresses the need {o develop seismic provisions based on data from the Philippines. ‘The attenuation laws used in this study are based on data from the United States and Japan. ‘This was neces- ‘say because of the lack of data inthe Philippines. ‘Because of these observations, a review of the design ‘yalues specified by the design code and the shape ofthe design is needed. For a beter esdimate of the seismic hazard and the seismic forces applied 10 structures during earthquakes, song ground motion records In the Philippines should be collected andFigure 10. Proposed seismic zoning map for the ‘Prllippines based on the expecied peak ground uccelera- tion analyzed. Thus, the collection of strong ground motion fecords should be given « high priotty to improve earthquake disaster midgation inthe Philippines. REFERENCES ‘AU Fiss Investigation Team. 1990. Quick report on the 1990 Luzon earthquate (in Japanese). Comell, CA, 1968. Engineering seismic risk analysis. ‘Bul. Seis. So. of Am. 58:1583-1606, Japan International Cooperative Agency (JICA). 1990. ‘Report of expert team of Japan Disaster Relief Team UDR onthe earthen te Pale of ly 16, Kawasumi, H. 1951. Measures of earthquake danger ‘and expectancy of maximum jatensity throughout Japan as inferred from the eelemi activity in hision- cal times, Bull. Earthquake Research Institue 29, University of Tokyo. Katayama, T. 1982, An engineering prediction model of sceeleration response spectra and its application to seismic hazard mapping, Earthquake Engineering ‘and Structural Dynars 10:149-163. , RK. 1977, Seismic design spectra and ig procedures using hazard analysis based iret on oscillator response. Fil. Jour. Earthquake Engineering and Structural Dynanis 5211-234. ‘Stepp, J.C. 1972. Analysis of completeness of the earth- ‘Quake sample in the Puget Sound area and its effect on statistical estimates of earthquake hazard. Proc. Intl. Conf. Microzonation for Safer Construction 23897-5909. Figure 11. Current seismic zoning map in the Philippines (Villaraza, 1991). Tea, Yan KauyanaT. 1988, An elne pari compuir program CERISA-G> and is implication 0 felumie macro-ronaton of Japan. Proc. tk WCEE Saai-ias. Vilaaze, CM, 1991. A study on the seismic zoning of the Philippines. Proc. th Indl. Conf. on Selsmie Zonaton 3511-518 6036