Earthquake Engineering, Tenth World Conference© 1994 Balkema, Rotterdam. ISBN 90 5410 060 5 The April22, 1991 Limon (Costa Rica) earthquake G. Santana Laboratorio de lngenier{a Sismica, Universidad de Costa Rica, Costa Rica ABSTRACT: The 1992 main Limon earthquake had Ms 7.6, strong-motion duration (at San Isidro) 26.6 s and maximum MM intensity XI. With epicenter near the Caribbean coast, the shapes of its isoseismals differ markedly from the code-specified isoacceleration curves for various return periods, as the latter curves assume that all seismic sources are close to the Pacif­ ic coast. On the other hand, code design spectra are overconservative for long periods, es­ pecially on soft ground. These matters demand a code revision. Although the death toll was moderate, there was considerable material damage especially in the province of Limon, including widespread liquefaction and ground failure, which damaged roads and railways. Main causes for damage to buildings were, as is often the case, insuf­ ficient transverse reinforcement, poor detailing, short columns prone to brittle failure in shear, and soft first story. Storage-tank and bridge failures are also analyzed. The danger of a macroseism within densely populated areas is brought out. 1 INTRODUCTION The main shock had a magnitude Ms 7.6. It occurred at 15:57 local time with epicenter 43 km SE of Port Lim6n at a 10 km depth. Fig 1 shows estimated MM intensities. The earth­ quake caused the failure of buildings, bridges and infrastructure in general and severe damage to roads and industrial facilities. The number of fatalities was less than 100. About 4000 dwellings were destroyed and 12 000 more suf­ fered partial collapse, mainly associated with poor construction. Some 250 small schools were damaged. In Lim6n the water supply system de­ veloped many ruptures and required several weeks for repair. This paper deals with seis­ micity, design practice in the country and di­ rect material losses caused by the earthquake. 2 SEISMICITY AND DESIGN IN COSTA RICA The 1986 Seismic Code (C6digo Sismica de Cos­ ta Rica) uses isoacceleration maps, based on Mortgat et al (1977), for return periods of 50, 100 (fig 2a), 500 and 1000 (fig 2b) yr. The latter reflects the assumption of three source zones, all near the Pacific coast. Santana (1990) gives exam)?l_es_ of dam_age from a recent earthquake with epicenter in Nicoya Gulf off the Pacific coast. Yet the April 22, 1991 event occurred near the Caribbean coast, far from the assumed source zones. Significant events with magnitude greater than 7.0 have occurred along the Caribbean 7033 81i' + 10'1· 9'+ B'+ 86' 15' r I· 85' --$- Epi~cnlcr M' + I· 84' Bl' + t .,. Fig 1. MM intensities of the 1991 main shock (after Rojas, 1991). coast from Nicaragua to Panama (Miyamura, 1982). They include the destructive earth­ quake of 1916 (Ms=7.4) causing much damage in Bocas del Toro and felt strongly in Lim6n, and the 1953 event (Ms=S.S) near Lim6n. That the zoning maps ignore sources near the Car­ ibbean coast is due to imprecision in locat- 86' + ..... I'> as• Fig 2. w 84' 8J' 1 I· + ., NICARAGUA <::> ., I ,. +tl' "' "' ... +10' -$- Ccucenfer (a) 85' f NICARAGUA 14' + 8l' ., t '"' + 85' -$- EpltOI\Ier f 84' (b) Code issoacceleration curves for a) 100 yr return period; b) 1000 yr return period. +10' + 8' ing the foci because of lack of a seismo­ logical network. Rather than using an importance factor the code allows the designer to choose the struc­ ture's life and corresponding exceedance probability. This determines the return pe­ riod. An ordinary structure with 50-yr de­ sign life and 40% exceedance probability has a 100-yr return period. The designer inter- 7034 polates linearly between contours in fig 2a. The code provides response spectra for rock, hard soil and soft soil, and specifies duc­ tility factors of between 1 and 6. (See San­ tana, 1988 on the code, and Sauter, 1989 on seismicity.) The code requires modal analysis for struc­ tures taller than 30 m and allows static analysis for lower structures. In both meth­ ods the peak acceleration is obtained from the isoacceleration maps. For 1-2 story buildings a simpler approach is permitted; the country is divided into three zones (fig 3); base shear coefficients are 0.11, 0.22 and 0.33 for zones I, II and III, respective­ ly. 3 INTENSITY OF DAMAGE The maximum MM intensity was IX in Matina, just north of Highway 32, along Highway 36 on the coast south of Limon and in Panama next to Costa Rica (fig 1). Electric power in Lim6n was interrupted for about 24 hr. No major damage was reported on the main transmission lines. Local distribu­ tion lines experienced many cable breakages. Severe ground fissures forced the closing of Highways 32 and 36, nine bridges suffered severe damage or collapse and significant settlements occurred in bridge approaches where intensities reached VIII and IX. coast uplift of 1.5-2.0 m was observed. This exposed a coral reef in Limon which had been below sea level. Ground failure and liquefaction were re­ ported throughout most of the 12 000-km epi­ central region, in the province of Limon. The area is dominated by a broad plain slop­ ing gently from the Talamanca Mountain Range to the Caribbean. The plain is dissected by several large and many small river valleys that broaden as they app.roach the coast. Most liquefaction occurred in alluvial and fluvial deposits under the river floodplains; also in lagoonal and estuarine deposits under coastal lowlands. About 30% of the highway pavement was disrupted by cracks, scarps and settlements caused by liquefaction (EERI, 1991), and several railways segments were misaligned. The greatest ground-induced dam­ age took place at river crossings, where bridge decks were thrust over abutments, piers shifted riverward and fills settled as much as 2m (Youd at al, 1992). 4 STRONG MOTION D•ATA Accelerograms were recovered from 14 of the 19 permanent stations deployed by the Earth­ quake Engineering Laboratory of the Universi­ ty of Costa Rica (Santana, 1991). Ten of the 14 instruments were on free field or from low-rise structures; the rest on high-rise buildings. The closest strong-motion sta- tion, in San Isidro, on hard ground 73 km from the epicenter (fig l) registered maximum accelerations of 0.20g horizontal and 0.17g vertical (Santana et al, 1991). The maximum free field acceleration recorded was 0.27g, in Cartage, on soft ground 94 km from the epicenter. The Costa Rican Electricity In­ stitute and the Seismological and Vulcano­ logical Observatory maintain a number of ad­ ditional instruments. --At San Isidro the strong shaking (5-95' of the Arias intensity) lasted 26.2 s, much longer than at Presidio during the 1989 Lorna Prieta earthquake and in the 1986 San Salva­ dor earthquake. Comparison of the 5'-damping response spectrum for the strongest San Isi­ dro component with the code design spectrum and with that derived from the Newmark-Riddel criteria (fig 4) shows that the latter spec­ trum is in reasonable accord with the re­ sponse spectrum but that the code overesti­ mates spectral ordinates for long periods. This is more pronounced for the Cartage sta­ tion, on soft ground (fig 5). As a conse­ quence structures with long fundamental pe­ riod, designed according to the code, are overdesigned, especially on soft ground. The situation is brought out in fig 6, which shows ductility demands for single-degree-of­ freedom systems designed for a ductility fac­ tor of 4. Qualitatively the same holds for other design ductility factors. (The ap­ parently excessive demand in very rigid structures is doubtless covered by their overstrength. --Note by the editor.) This helps explain the low-damage incidence to large buildings in san Jose's metropolitan area. a•• + a•• as• ~ NICARAGUA 25 ~ kilometers 14' + a·~ + + 86' ~· 84• ~ Zone I ~ Zone n ISSSI Zone m al' + t '"' +10' + 8' -$- Epicenter Fig 3. Code seismic zones for simplified design. 7035 DESIGN SPECTRUM vs RESPONSE SPECTRUM LIMON- COSTA RICA EARTHQUAKE 1000 ~ ~ § - SITE: SAN ISIDRO DAMPING:5% -· 1---·-.. r,, ,. rL ·, (/ 'I,\'\... \ ::: 100 0 '0 " "' II) a.. DESIGN SPECTRUM vs RESPONSE SPECTRUM LIMON- COSTA RICA EARTHQUAKE 1000 SITE : CART AGO DAMPING: 5% - - 100 ------ 10 / 1 0.01 .- ·-·-· "'· };. r ~ ,, - \ 1----;~-I--M \ // \ \ \ I ~ \ ~~ \ '\ ---- -·- __ \ --Newmark a Riddell \ -- - Response -·- CR Seismic ,I I nl 0.1 Code I 10 Period (s) !\'\ ~ 100 Fig 5. Response and design spectra, Cartage Station. 5 DAMAGE to HOSPITAL IN LIMON The important Dr Tony Facio Castro Hospital built in 1982 was evidently not designed to resist strong earthquakes. A large four­ story wing suffered severe damage. Its re­ inforced concrete frame had end masonry filler walls above the first story. This re­ sulted in a soft story and the ensuing shear distress in the first-story columns. The end walls forming the stairwell were connected to each floor slab through one 20-mm bar which pulled out. The lightweight precast fiber- reinforced panels having styrofoam cores failed in shear and fell through the ground­ floor corridor roof. There was also much nonstructural damage caused by large story drifts and by rain infiltration. This wing had to be closed. The remaining lower rise wings remained operational. 6 DAMAGE TO HOTELS IN LIMON The four-story Las Olas Hotel, built on re­ inforced concrete piers founded on the coral bed that experienced significant uplift dur­ ing the earthquake underwent partial collapse and serious damage. The structure had col­ umns of various lengths. One of the shorter columns failed in shear. The __ 4_00 by 600 mm column was reinforced with four 30 mm longi­ tudinal bars and 12 mm ties at 300 mm. Its clear height was only 950 mm. The longitudi­ nal bars were spliced in the critical region and the ties were much corroded. Another first-story column suffered severe shear dis­ tress owing to the presence of a spandrel beam which resulted in a short column. This hotel had a major discontinuity in its shearwall which ends abruptly at the ground floor. The corresponding columns underwent brittle shear failure, which prevented the rest of the structure from developing its lateral load capacity. The three-story International Hotel failed in the first story and collapsed. The 300 by 600 mm first story columns had embedded plas­ tic pipes and were reinforced with eight 25 mm longitudinal bars and 9.5 mm ties at 300 mm, ending in 90 ° bends which became inef­ fective after concrete-cover spalling. This particular deficiency has been observed in many other earthquakes. The beam-column joints had no transverse reinforcement and were poorly detailed, so the hooked bars from the beams pulled out of the joints. The two-story Ng Hotel experienced shear failures in its poorly detailed short columns and in the shearwalls. The partially in­ filled walls between columns reduced the lat­ ter's clear height, a feature that has caused damage in other earthquakes (Newmark and Ro­ senblueth, 1971; Mitchell et al, 1986; Mit­ chell, Tinawi and Redwood, 1990). 7 DAMAGE TO INDUSTRIAL FACILITIES Lim6n port facilities suffered severe ground damage due to liquefaction of the sand fill. A one-story light structural steel-frame warehouse had permanent lateral deformations in the columns. one section of a ·steel wharf with timber decking collapsed; ita steel mem­ bers were so badly corroded that only a frac­ tion of their sections remained. Lighting poles 12 m tall founded on large footings underwent perman~nt rotation of foundations resulting in a 7 tilt. 7036 DUCTILITY DEMAND vs PERIOD LIMON-COSTA RICA EARTHQUAKE. OUCT.•4 9.----------------------------------. 8 Soft soil, firm soil ond rock profiles for 5 °/0 damping Longitudinal component 500 year return period -~4~--~--------------------------------------1 ::J 03 2 I· o~~~~~~~~~~~,-rT~~~-r~ 0.3 0.8 1.3 1.8 2.3 2.8 Period (s) Fig 6. Ductility demands in a single-degree system designed for ductility factor of 4 according to the code. The Institute Costarricense de Electricidad (ICE) generating plant in Moin was built in 1977. Its capacity is 32 MW. Although there was evidence of severe ground movement and liquefaction, the equipment was undamaged ex­ cept for minor oil leaks and the need for tightening the anchor bolts of a diesel en­ gine due to settlement of its foundation. The main building housing the generators is a 10 m tall one-story light steel structure. Its lateral load resisting system consisted of tension-only bracing in some bents. A number of these 75x75x5 mm angle braces buck- ' led evidencing the poor performance of this type of structural solution. Three adjacent cylindrical oil tanks suf-3 fered damage at their supports. These 80-m tanks had a diameter of about 3 m and overall length of 10.6 m. They were supported on five legs. Two different details were used for the supports. One of the legs typical for two of the tanks suffered weld tearing at the junction with the tank as well as perma­ nent deformations. These supports were not anchored to the strip footing. The end sup­ ports bent severely and the leg caused buckl­ ing of the tank wall. Performance of the tanks having different support details was much better. The legs were braced and an­ chored to longitudinal and transverse support beams, the longitudinal ones in turn anchored to a strip footing. Although the tension­ only braces buckled there was no permanent movement nor damage to the tank walls. The damage to the braces is easily repairable. A warehouse with reinforced concrete frames that was under construction and is part of the ICE facility suffered serious damage due mostly to its short columns brought about by masonry filler walls. These columns failed in shear at their tops and developed flexural hinges -at the bottom of their unsupported lengths. The joints failed in shear due to lack of transverse reinforcement; beam re­ inforcement, bent down at 90~ pulled out. The RECOPE oil refinery in Moin suffered important damage in the plant facilities and oil storage tanks. Structural damage to the process equipment was light. Two fires broke out. Many tanks in the refinery had been filled the day preceding the earthquake. Fifteen of them, ranging in capacity between 560 and 117 644 barrels, failed in various ways. One containing naphtha and diesel oil exploded landing SO m away. Others suffered sloshing, top wall buckling, elephant-foot buckling and float-roof tilting, and a spherical LNG tank that was under construction collapsed. 8 DAMAGE TO BRIDGES Nine bridges in the epicentral region were badly damaged or collapsed. Following the earthquake, road access was impossible to Li­ m6n on Highway 32 and to areas south of Limon on Highway 36. Minor damage was reported in bridges crossing Rivers Tore, Rojo, Escondi­ do, Aguas Claras, San Miguel and Banano. The crossing of River Viscaya on Highway 36 is a two lane bridge with three 22-m simple spans, located about 10 km south of Lim6n. It was designed in 1971. The abutments have vertical and battered piles and the concrete piers rest on piles embedded over the height of the piers. The soil consists of fine sand. Due to very large ground movements one pier collapsed causing loss of support of two spans. A horizontal restraining device held the superstructure together over the other pier. Bridge ends were subjected to large rotations and a tension failure developed in one abutment. The bridge over Bananito River has two sim­ ple spans 25 and 28 m long on skewed sup­ ports. The deck collapsed owing to abutment slumping and rotation. In the village of Bomba, a few kilometers from the Bananito River bridge, a simple-span steel-truss railway bridge crossing the Sana­ no River was very badly damaged owing to failure of the abutment foundations caused by large-scale ground displacements. The bridge had to be closed to traffic. In the eight-span bridge crossing the Chir­ rip6 River on Highway 32 about 30 km west of Lim6n, the six interior spans have haunched continuous girders while the two shorter spans are simply supported. The west-end span collapsed due to loss of support over the first pier, resulting in closure of the bridge for one week. There was no restrain­ ing device over this pier. The structure was temporarily repaired by lifting the col­ lapsed span and placing it on its original support over the pier, and the bridge was re­ opened to traffic. Work was underway to en­ large the pier foundation to accommodate four steel columns providing additional support for the girders. This collapse emphasizes 7037 the need for restraining devices between ad­ jacent simple spans. The continuous spans were undamaged despite a 100 mm transverse displacement of the su­ perstructure on one of the piers. supports of these spans consist of a rocker bearing having transverse sliding capability. Trans­ verse displacement is somewhat restrained by keeper plates welded to the top of the slider and bolted to the rocker assembly below the sliding joint. Keeper plates failed due to the significant transverse displacements. 9 CONCLUSIONS The Limon earthquake caused a considerable disruption in all aspects of life for the af­ fected region. The local infrastructure suf­ fered extensive damage. The impact had to be borne by all levels of society as well as by the government. The cost of repairing or re­ placing all the civil works and housing dam­ aged was higher than estimated and far above the capabilities of public offices. The fast recuperation must be partly credited to na­ tional and international relief organiza­ tions. However, the impact of the earthquake was co~centrated in a small area, about 12 000 km , that is sparsely populated and mostly dedicated to agriculture. Had the epicenter been located in the Central Valley, where two thirds of the country's population lives, the effects would have been much greater. One of the most important lessons of this earthquake is that a new shear zone cutting across the middle of the country and splitting the Cen­ tral Valley in two has been identified. The significance of this finding is enormous and can now be properly documented with an ade­ quate reinterpretation of the historic evi­ dence. Data recovered from this earthquake will doubtless contribute to the further under­ standing of the Central American earthquakes. The strong motion records have already helped interpret the behavior of the building inven­ tory in different parts of the city of San Jos6 and have shed some light in the parame­ ters currently used in the region for the es­ timation of attenuation laws. Finally, the limited impact of the earth­ quake and the rapid recovery should not be taken as indicative of low vulnerability. A false sense of security would help increase the unassessed vulnerability to seismic haz­ ard in Costa Rica and in Central America. REFERENCES Brant, M. 1991. Damage tally. Tico Times, May 3. san Jose, Costa Rica. 104. C6digo Sismica de costa Rica. 1986. Cartage, Costa Rica: Editorial Tecnol6gica de Costa Rica. 104. EERI 1991. Costa Rica Earthquake Reconnais­ sance Report Earthquake Spectra. 7. Sup B. l-127. Mitchel, D, Tinawi, R and Redwood, R G. 1990. Damage to buildings due to the 1989 Lema Prieta earthquake - A Canadian code per­ spective. canadian Journal of Civil Engi­ neering, 17 (5). 813-834. Miyamura, S. 1982. Sismicidad de Costa Rica. San Pedro, Costa Rica: Editorial Universi­ dad de Costa Rica. 190. Mortgat, c P, Zsutty, T c, Shah, H c and Lubtkin, L. 1977. A study of seismic risk for Costa Rica. John Blume Earthquake Engi­ neering Center. Stanford University. Report 25. 463. Newmark, N M and Rosenblueth, E. 1971. Funda­ mentals of Earthquake Engineering. Engle­ wood Clifs, NJ: Prentice Hall Inc. 640. Rojas, w. 1991. Informe sobre los sismos sen­ tides en abri1 de 1991. Red Sismo16gica Nacional. Escuela Centroamericana de Geolo­ gia. Universidad de Costa Rica. Santana, G. 1988. Current seismic design reg­ ulations for Costa Rica. Bulletin of the International Institute of Seismology and Earthquake Engineering. Building Research Institute. Ministry of Construction. 23. Tsukuba, Japan. 95-122. Santana, G. 1990. Sismo de Cobano, 25 de mar­ zo de 1990: efectos sobre suelos y edifica­ ciones. Laboratorio de Ingenieria Sismica. Institute de Investigaciones en Ingenieria. Universidad de Costa Rica. Reports INII 49- 90. 194. Santana, G. 1991. Reports preliminar de los registros de aceleraciones obtenidos duran­ te el sismo del 22 de diciembre 1990. Labo­ ratorio de Ingenieria Sismica. Institute de Investigaciones en Ingenieria. Universidad de Costa Rica. Santana, G, Vargas, W, Sancho, V, Segura, c, Ramirez, A and Sibaja, R. 1991. Registro de aceleraciones del terremoto de Limon, 22 de abril de 1991. Laboratorio de Ingenieria Sismica. Institute de Investigaciones en Ingenieria. Universidad de Costa Rica. Re­ porte INII 4B. 58-91. Sauter, F. 1989. Introduc.cion a la Sismolo­ g!a. Cartage, Costa Rica: Editorial Tec­ nologica de Costa Rica. 271. Youd, T L, Rollins, K M, Salazar, A F and Wallace, R M. 1992. Bridge damage caused by liquefaction during the 22 April 1991 Costa Rica earthquake. Madrid: lOth World Confer­ ence on Earthquake Engineering. 7038