The Pisco, Peru, Earthquake of August 15, 2007

EERI Special Earthquake Report — October 2007

Learning from Earthquakes The Pisco, Peru, Earthquake of August 15, 2007

EERI team leader Eduardo A. Fierro Crete, and Erick Ortega and Pablo ple were killed and 1,090 were in- of BFP Engineers, Inc. (Bertero, Broncano of Direccion de Hidrografia jured. The majority of the damage Fierro, Perry), in Berkeley, Califor- y Navegacion of the Peruvian Navy. and casualties occurred in Chincha nia, visited the area affected by the Alta, Ica, and Pisco (Figure 1). Most The teams visited the following cities: earthquake August 18-24, 2007. of the buildings destroyed were Pisco, Chincha, Canete, Ica, San Professor Marcial Blondet of the adobe housing. Hospitals, schools, Andres, Tambo de Mora and Lima. Pontificia Universidad Católica del and other medium-to-large public All cities were within 97 km of the Perú (PUCP) in Lima participated buildings were also damaged. Most epicenter, with the exception of Lima, as an in-country member of the of these buildings are built using located about 145 km away. The EERI team, providing critical assis- reinforced concrete frames and infill EERI team joined up with the PUCP tance in the field. Other members brick masonry rigidly attached to team led by Nicola Tarque, who start- of the EERI team included Donald the frames. ed the site inspection on August 17. Ballantyne of ABS Consulting, Inc., The PUCP team included Luis Carlos Widespread communications and in Tacoma, Washington; Mikael Fernandez and Jesus Carpio from power outages occurred in the Gartner of Nabih Youssef & Asso- PUCP, and Eric Hulburd from Stan- area. The Pan American Highway, ciates in Los Angeles, California; ford University. the Carretera Central, and other Adolfo Matamoros of the University main transport routes suffered of Kansas; Kim Shoaf and Alina The EERI team and the publication of heavy damage due to landslides Dorian of the Center for Public this Learning from Earthquakes re- and liquefaction. A small tsunami Health and Disasters at the Uni- port were supported by the National resulted in wave heights of 100 cm versity of California at Los Ange- Science Foundation through grant at Callao and La Punta, Peru; 36 les; and Hope Seligson of MMI #CMII-0131895. NSF supported the cm at Arica, 28 cm at Coquimbo, Engineering in Huntington Beach, GEER team and the tsunami survey and 17 cm at Valparaiso, Chile. California. EERI collaborated with team through grants #CMII-0323914 and CMS-0646278, respectively. many other investigators in the Seismological Aspects field, including Liliana Pinto and Kent Yu of Degenkolb Engineers, Introduction The Pisco earthquake was the re- Mario Rodriguez of UNAM, San- sult of the subduction process be- The Mw 8.0 Pisco earthquake struck tiago Pujol of Purdue University, tween the Nazca plate and the at 6:40 p.m. local time. The epicenter and Arturo Schultz of the University South American continental plate. was about 45 km west-northwest of of Minnesota, representing The This subduction process results in Chincha Alta, or about 145 km south- Masonry Institute. a high rate of seismicity along the southeast of Lima. At least 519 peo- Adrian Rodriguez-Marek of Wash- ington State University led the Geo- Figure 1. Engineering Earthquake Recon- Overview naissance (GEER) team and also map and served as a member of the EERI Paracas team. His team included Brady Cox peninsula of the University of Arkansas; detail map Jorge Meneses of Kleinfelder, Inc., showing in San Diego, California; Viviana reconnais- Moreno and Manuel Olcese of sance track PUCP; Rodolfo Sancio of Golder with tsunami Associates in Houston, Texas; and survey loca- Joseph Wartman of Drexel Univer- tions, gener- sity in Philadelphia, Pennsylvania. alized runup A tsunami reconnaissance team heights, and was composed of Hermann Fritz earthquake (team leader) of the Georgia In- location. stitute of Technology, Nikos Kalli- geris of the Technical University of (Background satellite imagery: Europa Technologies/Google Earth)

EERI Special Earthquake Report — October 2007

Figure 2. East-west spectral accelerations recorded by CISMID stations (from Lazares et al., 2007). All stations are located at a range of 96-111 km from the fault plane. The Rimac and San Figure 3. Acceleration, velocity, and displacement Isidro stations are located on dense, stiff gravel deposits (re- time histories for the La Molina record (stiff soil). ferred to as Lima Conglomerate); the La Molina station is located on shallow soil overlying the denser Lima Conglomerate; ground surface projection of the the Callao station is located on soft soil. fault plane. The earthquake was recorded at 16 coasts of Peru and Chile. The earth- had a moment magnitude (Mw) of 8.0 stations within 150 km of the fault quake was an interface event and (USGS, Harvard CMT), and the hypo- plane. All of the stations except for occurred on a previously identified center was located at 13.76S and two in Ica are located approximate- seismic gap between the rupture 76.97W, at a depth of 39 km (USGS). ly 100 km to the north in the vicinity areas of the 1974 Lima and the Fault rupture propagated south from of Lima. A pattern of higher PGAs 1996 Nazca earthquakes (Tavera the hypocenter (Tavera et al., 2007). at soil stations was observed; addi- and Bernal, 2005). The earthquake The finite fault solution of Ji and Zeng tional evidence of site effects is (2007) indicates a seen by the distinctly different spec- rupture plane with tral shapes observed in the four a strike of 324º CISMID stations (Figure 2), which and a dip of 27º, are located in three different soil having approxi- types. Site effects are manifested mate dimensions by significant amplification over of 190 km along period bands that are compatible strike and 95 km with the general site description along dip. The (i.e., shallow, stiff soil deposits at La locations of after- Molina, and soft soils at Callao). shock hypocen- Figure 3 shows the ground motion ters match this record from La Molina. It shows two fault plane (Tav- phases of strong ground motion, era et al., 2007). which correspond to slip along two The shallow dip asperities in the finite fault solution and large width of of Ji and Zeng. The overall shape of the fault plane the acceleration time history is simi- implies that the lar for other ground motion stations. cities most affect- Figure 4. Approximate area of seaward lateral displace- ed by the earth- All the strong motion records had ment of a marine terrace in Canchamana (background quake (Pisco and long durations of strong shaking. image from Google Earth). Ica) lie within the For example, for the La Molina rec-

EERI Special Earthquake Report — October 2007

Figure 5. Approximately 0.9 m of liquefaction-induced settlement at a one-story residence in Tambo de Mora. Figure 6. Maximum vertical offset of approximately 3.0 m along the interface between the marine terrace deposit ord, the 5-95% significant duration and the Cañete formation in the Canchamana lateral slide. (Trifunac and Brady, 1975) is 101.3 seconds and the bracketed dura- embankments damaged by lateral not firmly established, but appeared tion (duration of ground motion spreading, toppling of power poles to be influenced by the presence of higher than 0.05g) is 98.6 seconds. founded in liquefied soil, rupture of human-made works at a bend in the A duration of approximately 100 water and sewer lines, and disruption Pan American highway. The west- seconds is significantly longer than to port facilities. The liquefaction at ward slope of the marine terrace the duration expected for earth- each of these sites was confirmed by surface was approximately 1.6 to quakes of similar magnitude from either the presence of sand boils at 2.1 %. A clear vertical offset (scarp) shallow crustal events (the attenu- the ground surface or wet sand ejecta of variable magnitude was observed ation relationships of Abrahamson in open cracks. The spatial distribu- along much of the geological inter- and Silva [1996], developed using tion of observed and reported lique- face between the marine terrace data from shallow crustal events, faction features along the Peru cen- and the Cañete formation (Figure 6). predicts a median duration of 48 tral coast ranged from Villa, just south In addition to the main scarp at the seconds for a Mw 8.0 earthquake). of Lima and approximately 90 km interface, numerous extensional north of the rupture plane, to Paracas ground cracks developed across Geotechnical Aspects in the south (the southern limit of our the marine terrace parallel to the The Pisco earthquake caused a reconnaissance), approximately 24 coastline. In general, the cracks significant amount of soil liquefac- km from the rupture plane, and Ica in were largest (both in terms of hori- tion in the mesoseismal zone which the east, about 40 km from the fault zontal and vertical displacement) resulted in considerable damage to plane. Within this region, liquefaction near the scarp and became smaller urban areas and to the transporta- was widespread wherever the ground towards the coastline. Some crack tion infrastructure. In addition to liq- water table was shallow. The north- widths were as large as 1 m and uefaction, there were a significant ernmost liquefaction site, in the Villa sand ejecta was found inside many number of landslides in road cuts neighborhood of Lima, occurred in a of them. and natural terrain. Landslides were swampy area. responsible for the three-day clos- The massive liquefaction-induced Tsunami ing of the east-west Highway 24A. seaward displacement of a marine The earthquake was centered 35 km The earthquake spawned a wide terrace in Canchamaná, 2.5 km north offshore from Chincha Alta on Peru’s variety of liquefaction failures. No- of Tambo de Mora, was 1 km wide central coastline. A number of near- table liquefaction-induced damage by 3 km long (Figure 4). The eastern by villages felt the earthquake, in- included a massive lateral spread boundary of the lateral spread was cluding all those affected by the tsu- extending over approximately 3 km defined by the interface between a nami in any way. Unfortunately, the (Figure 4), a 400 m-long slope fail- Holocene marine terrace and the closest tide gauge — located 55 km ure induced by liquefaction at the Pleistocene Cañete formation. The south of the epicenter at the Puerto toe, spectacular one- to two-story southern boundary appeared to be General San Martin — was dam- building foundation failures result- defined by the interface of the marine aged by the earthquake and unable ing in up to 0.90 m of settlement terrace with a Holocene alluvial sand to record the tsunami. Peruvian (Figure 5), numerous highway deposit. The northern boundary was Coast Guard personnel at countless

EERI Special Earthquake Report — October 2007

Figure 7. Debris on the shore at the fishing village of Figure 8. Paracas marina tsunami height measured Lagunilla, which was completely destroyed by the tsunami. based on a floating dock stuck in an uplifted position.

outposts were alerted by the earth- ly, the tsunami peaked in a mostly un- (Sieh, 2006), and that critical facili- quake and triggered evacuations inhabited desert area. At Lagunilla, ties such as hospitals, schools, and in affected fishing villages in the the tsunami flooded up to 2 km inland prisons should not be located in tsu- short time window of 10-20 minutes over extremely flat terrain, with 4 m nami hazard and flood zones. This between the earthquake and the runup at the inundation limit and 5-6 is of particular importance for Peru, arrival of the tsunami. However, m runup at the village on the shore- given the fairly frequent moderate- warnings did not reach Lagunilla, line (Figure 7). Similar runup heights size tsunamis such as the 2001 on the south coast of the Paracas and flow depths during the 2004 In- Camana tsunami in southern Peru, peninsula, where three of the sev- dian Ocean tsunami resulted in much the 1996 Chimbote tsunami in en inhabitants were washed away higher death tolls in Sri Lanka (Liu et northern Peru, and the 1974 tsu- and their bodies recovered up to al., 2005) and Somalia (Fritz and Bor- nami in the same area as this event. 1.8 km inland. Although the resi- rero, 2006). dents of Lagunilla had felt the Adobe In Paracas, there was damage to the earthquake, they were unaware of dock, a fixed platform supported by The earthquake caused enormous the post-earthquake tsunami haz- piles embedded in sand with a float- damage to earthen buildings in the ard, so the event failed to trigger a ing platform at the end. The sand affected area. Adobe construction spontaneous self-evacuation. supporting the dock liquefied and is widespread in Peru; most houses The International Tsunami Survey the piles lost bearing capacity and over 50 years old are made of Team measured local flow depths sank differentially into the sand up to adobe and, although confined and tsunami heights, maximum run- approximately 30 cm, while the masonry is becoming the preferred up, and inundation distances; re- floating dock rose with the tsunami construction technique for families corded structural damage; and inter- and became stuck when the water who can afford it, adobe is still the viewed eyewitnesses per estab- level went down (Figure 8). only viable alternative for many fam- lished methods (Synolakis and Okal, ilies. On the Peruvian coast, most To the north of the Paracas penin- 2005). Eyewitnesses described be- adobe houses are one story high. sula, the tsunami runup was roughly tween one and four main waves When the house has two stories, halved, with characteristic runup with an initial recession correspond- the second story is usually built with heights of 3 m. Boats were washed ing to a leading depression N-wave quincha, a series of wooden frames into the streets in Pisco, and the Tam- (Tadepalli and Synolakis, 1996). filled with crushed cane, covered bo de Mora prison partially flooded in More than 50 transects were record- with mud and plastered with mud or Chincha Alta. The prison compound ed from the waterline to the inunda- gypsum. Since it does not rain on walls had collapsed in the earth- tion limit and adjusted for tide levels most of the Peruvian coast, roofs quake, so luckily the 600 prisoners upon tsunami arrival. are horizontal and flat, and consist were gone prior to the tsunami arrival. of wooden joists supported directly The tsunami runup distribution This earthquake demonstrates that on the adobe walls and covered by peaked south of the Paracas penin- community-based education and wooden planks or a layer of crushed sula with a 10 m runup, and sus- awareness programs are particularly cane, sometimes also covered by tained a runup in excess of 7 m essential to help save lives in locales straw mats (esteras) and then plas- along 5 km of coastline. Fortunate- at risk from near-source tsunamis tered with mud.

EERI Special Earthquake Report — October 2007

Figure 9. Street in Pisco three days after the earthquake. The house at the far right, undamaged, was made of confined masonry. Figure 10. Two collapsed adobe houses in Pisco. The Wall thickness and height depend That more people roof joists of the house on the left are supported on the mostly on the age of the dwelling: were not killed quincha partitions. The roof joists of the house on the older houses have thicker, taller in their adobe right were supported on the façade wall, and thus col- walls (up to 0.80 m thick, 4 m high; houses is attrib- lapsed with the façade. Notice the undamaged confined with slenderness ratios of 5-6). utable in part to masonry house at far left. Contemporary houses have thinner, the fact that Peru- shorter walls (typically 0.25 m thick, vians usually run out of their houses walls probably would have been 2,5 m high, with slenderness ratios when they feel an earthquake. The helpful to prevent this type of of 8-10 or higher). Interior walls and team spoke with a number of people damage (Figure 10). partitions are made of adobe or who watched their houses collapse City block effect: In Pisco, and in quincha. The walls are not provided after they got out of them. other towns, the seismic separation with additional reinforcement to In Pisco, closest to the epicenter, more joints between buildings do not ap- withstand seismic forces. In short, than 80% of the adobe houses col- pear to be common practice. Build- adobe walls are massive, weak and lapsed or sustained heavy damage ings within a block influenced the brittle. Since they are massive, they (Figure 9). Apparently, the adobe performance of their neighboring attract large inertia forces during blocks and mortar in the Pisco and structures due to direct contact. It seismic shaking, which they are un- surrounding areas were made with appeared that the weakest aligned able to resist because the masonry sandy soil, which did not have suffi- building system in a block would is weak, and brittle failure occurs cient clay to provide good adhesion typically fail, for example, adobe without warning. The duration of between mortar and adobe blocks. building fronts failed when aligned shaking in this earthquake — about with brick and/or “confined masonry 100 seconds — also contributed to The most common type of failure ob- systems.” The failure of the adobe the many collapses. served in adobe houses was due to structures (typically one story) may the formation of have served as a dissipater of vertical cracks at energy for the rest of the block. the corners of the façade walls as Reinforced adobe: In 1999 the a result of out-of- Catholic University of Peru (PUCP) plane shaking, fol- reinforced 19 adobe houses in dif- lowed by the col- ferent locations of the country. The lapse of the walls reinforcement consisted of bands onto the street, of welded wire mesh nailed to the and sometimes walls and covered with a cement- the collapse of the sand mortar. Figure 11 shows the roof (especially house that was built in Guadalupe, a small town near Ica. It did not suf- Figure 11. This undamaged adobe house was reinforced if the roof joists were supported fer any damage during the earth- with bands of wire mesh covered with sand-cement mor- quake, whereas neighboring unrein- tar. Vertical bands were placed at each wall intersection, on the façade wall). A top collar forced houses were damaged or and horizontal bands at roof level. Notice the collapsed collapsed. This reinforcing system unreinforced wall at the right. beam joining all

EERI Special Earthquake Report — October 2007

Figure 13. Confined masonry with soft stories, relevant irregularities, or bad detailing, collapsed or showed severe damage. Figure 12. San Luis Church. The lateral wall and its heavy abutments were torn down, presumably by the dynamic Schools: The typical old-style thrust of the collapsing vault. school building in Peru consists of a reinforced concrete frame with in- should be used with caution be- concrete in which joists in one direc- fill panels. In some instances, the cause, although shaking table tests tion are formed by infilling the slab infill panel is rigidly attached to the have shown that the strength of with hollow clay bricks. frame and forms short columns that the adobe walls is significantly were severely cracked in the earth- Confined masonry with soft stories, increased, the mode of failure is quake. In other buildings there was relevant irregularities, or bad detail- brittle. an attempt to separate the infill pan- ing, collapsed or showed severe el from the frame with a 1-inch Churches: Many beautiful old damage (Figure 13). Modern and separation filled with styrofoam. churches in Peru are built with ado- well-designed confined masonry Unfortunately, a 1-1.5 inch layer of be walls and a quincha vault cover- resisted the earthquake with little or stucco was placed monolithically ing the nave. Most churches were no damage, including the confined on both sides of the walls and the damaged during the earthquake. masonry made with hollow clay tile column, rendering the separation The most dramatic case was that of that was thought to perform poorly. ineffective and causing short col- the San Clemente church in Pisco, umns that were severely damaged where the vault collapsed suddenly Engineered Structures in shear. and killed around 160 people who were attending a funeral service. Engineered buildings in the area typi- New school buildings in Peru are The priest was saved because he cally use RC systems with structural built with a combination of frames was under the dome, which had walls or frames. Hollow clay partitions and 3-foot shear walls spaced been repaired with reinforced con- are used; when they were isolated every 15 feet. The infill walls in the crete after a previous earthquake. from surrounding frames, they were longitudinal direction have a self- Figure 12 shows a side view of the damaged in the out-of-plane direction. supporting frame and are separated church of San Luis, in a small town When they were not isolated, they from the frame by a 1-inch elasto- just south of Cañete. were damaged in the in-plane direc- meric material. No stucco is allowed tion. Most of the damage to engi- over the joint. These schools be- Confined Masonry neered buildings happened to those haved very well, and no cracking with RC beam-column frames. Fur- was observed anywhere in the Most residential housing in urban thermore, cases of RC frame con- concrete frame or walls. areas of Ica, Pisco, and Chincha struction were observed where the Alexander Von Humboldt in Pisco: is constructed using confined brick structural system had no damage, but This campus, built in 1997, consists masonry, typically one to two sto- partitions either collapsed or had se- of four buildings forming a ‘U’ shape ries in height. This type of construc- vere damage. This behavior of frame in plan. The lateral force-resisting tion uses masonry with vertical RC buildings was observed in several system of the buildings is typically tie-columns that confine brick walls school buildings and hospitals in made of concrete moment frames and RC bond-beams along walls Pisco and Ica. Building construction with unreinforced masonry walls. A at floor levels. The floor system with structural walls behaved well one-inch seismic joint was used to typically consists of cast-in-place during the earthquake.

EERI Special Earthquake Report — October 2007

Figure 15. Shear failure of short columns next to corridor.

and back, respectively. This resulted in severe shear failure of the columns Figure 14. Extensive masonry wall in the front, which were on the verge damage. of losing gravity-carrying capabili- ties (Figure 15). The windows next Figure 16. Damaged building at the separate the structures at all levels. to the short columns fractured and University San Luis Gonzaga due It appears that no seismic joint was fell on the floor. The front masonry to short columns. used to separate the main class- infill in the first-story also experienced room building and the adjacent two- the moderate diagonal cracking. The story east wing. Pounding damage in the center, separated with a partial brick infill created the short was observed throughout the height 2-inch seismic joint at all levels. column condition, and caused the of the buildings with significantly Exterior moment frames of each center columns at the second story to more damage on the top floors. Al- wing typically showed no sign of experience shear cracking. The water though compressible foam was stress except significant pounding table at the site is approximately four used in the seismic joint, grouting between the wings and the elevator feet below the ground surface and was found to infill the top. The main tower. The infill walls at the corridor signs of sand boiling were observed lateral system of all the buildings experienced extensive damage of in several locations behind the build- suffered insignificant damage, but architectural finishes. Mechanical ing complex. the unreinforced masonry walls pipes and medical equipment and suffered significant damage during The University San Luis Gonzaga in supply racks were not seismically the earthquake (Figure 14). Most of Ica: Many buildings were damaged braced, resulting in extensive non- these walls need to be demolished here due to the presence of short structural damage. As a result of and replaced. Several partial brick columns (Figure 16). Some of the extensive damage to architectural infill walls next to the corridor com- building had the infill panels isolated and mechanical components, the pletely collapsed. from the frames. These panels were main hospital was abandoned. supported by a frame for the out of The Instituto Superior Tecnologico San Jose Hospital in Chincha: The plane forces. The connection of this Estatal Pisco in Pisco: Designed in second largest hospital in the re- frame to the main frame was inade- 1995, this classroom building con- gion consists of mostly one-story quate. With the bars anchored only sists of concrete frames with mason- structures. After a seismic assess- 4-5 inches into the main frame, the ry fill with floor and roof concrete ment of the hospital was completed out-of-plane forces made these frame slabs. In the transverse direction, about three years ago, the old hos- collapse or nearly so. it has five lines of reinforcement pital buildings are being replaced, concrete frames. Both the two end Hospitals: The Hospital Regional De including many adobe buildings. At frames and the middle frame have Referencia Ica: The largest hospital in the time of the earthquake, at least two bays with complete masonry the region, it comprises approximate- three new reinforced CMU build- infill. The brick infill in the front is ly 16 buildings, most of which were ings were near the completion of about one and half times higher designed and constructed in 1964. construction, and they performed than that in the back, resulting in an The main hospital consists of three very well. Several adobe buildings aspect ratio of about two and four four-story wings in a T configura- used for storage, as a morgue, and for the columns in the front tion and a five-story elevator tower to house the emergency generator

EERI Special Earthquake Report — October 2007

Figure 17. Damage to Pisco Hospital. Figure 18. Damage to a Pisco Hospital building with window openings at one end and a door opening at the other. experienced partial collapse. The hospital had to borrow a new emer- supplies were destroyed, and the one-story rectangular buildings gency generator for emergency hospital was without oxygen, elec- constructed about 74 years ago treatment. Other existing engi- tricity or water for a couple of days. (Figure 17). neered structures built in 1989 The patients stayed outside for The only original construction that performed well, except that one four days after the quake and then did not experience some structural corridor structure experienced sig- moved into the newly constructed damage was the kitchen/mechani- nificant damage due to pounding buildings. cal equipment building. This is due because the contractor used the Pisco Hospital: The Pisco Hospital to its lateral force system of mul- grouting to infill the seismic joint. has about 18 structures, two of tiple lines of concrete shear walls The mechanical pipes, mechanical which were reinforced concrete in each direction. After the earth- and medical equipment, and supply wall buildings and near the comple- quake, the hospital salvaged medi- racks were typically not seismically tion of construction. The majority of cal beds and medical equipment braced, so a majority of the medical the original hospital buildings were and supplies and moved them into the newly constructed buildings. Figure 18 shows the structural damage to an original building with window openings at one end and a door opening at the other end. Significant torsional response of this building caused the wall piers adjacent to the windows to suffer more damage, resulting in the out- of-place collapse of the windows and the walls. Figure 19 shows damage to a reinforced concrete building.

Lifelines Bridges: The two major roadways crossing the epicentral region are the Pan American Highway, which runs north-south along the Pacific Coast, and the Carretera los Liber- tadores, which runs approximately east-west, linking Pisco to Ayacu- cho. These two roads intersect at the town of San Clemente. The Figure 19. Damage to a reinforced concrete building at Pisco Hospital. Carretera runs along the Pisco

EERI Special Earthquake Report — October 2007

Other damage to the superstructure consisted of wide shear and flexural cracks in the diaphragm beams of piers 1 and 2. Due to the type of support used, the transverse motion of the bridge girders induced significant demands on the diaphragm beams. When the reconnaissance team first arrived it was open in one direction, but was closed several days later. Highways: The Pan American High- way from Lima to Pisco and Ica was closed immediately following the earthquake due to a slide at a point just north of Chincha where the highway climbs, moving away from the coast. Other sections of the highway suffered from liquefaction/ cracking, but probably remained passable immediately following the event. Highway 24a from Pisco to Figure 20. Evidence of liquefaction was observed at three piers of the Hua- Arequipa was closed for three days mani Bridge along the Pan American Highway and in the surrounding area. due to at least two large rockfalls and numerous smaller ones. River and has several minor and 2, and between piers 3 and 4 Ports: Small piers were slightly to bridges, mostly crossing irrigation had segments of superstructure sup- moderately damaged in Pisco and channels or small creeks. Because ported on expansion joints at both Paracas. The most significant dam- the epicentral region has very low sides. Permanent displacements on age was at the Port of San Martin. precipitation, river beds are relative- the order of 4 inches in the transverse It has a wharf about 1000 m long ly shallow and the bridge structures direction were observed at the expan- (Figure 21). The front (water side) are not very large. The bridges sion joints of both of these segments. along the Pan American Highway are of older construction, while the bridges in the Carretera los Liber- tadores consist mostly of single- span recently constructed precast concrete bridges with spans on the order of 70 feet. In general, most of the bridge structures in the epicentral region performed well. Damage was observed only at the site where liquefaction was also observed. The Huamani Bridge along the Pan American Highway near the town of San Clemente used pier walls with rela- tively low height to width ratios, and superstructures with several expansion joints. Evidence of liquefaction was observed at three piers and in the area surrounding these piers (Figure 20). The bridge had a total of five spans with lengths ranging from approximately 70 feet to 90 Figure 21. The fill behind a sea wall settled about 1 m at the Terminal Marí- feet. The spans between piers 1 timo General San Martín.

EERI Special Earthquake Report — October 2007

was broken so Tambo de Mora had no water. Wastewater: There is a wastewater system in Pisco. The collection system was heavily damaged and will probably need to be replaced. Their treatment plant was undamaged, but no sewage could flow to it.

Human Impacts As of August 31, 463 deaths were estimated for the City of Pisco alone, with 340 listed by name in a local newspaper (El Libertador, Ano XVII, No. 193, Setiembre 2007). This total does not include injured that were transferred to Lima or Figure 22. Temporary phone booths were installed for civilian use in Pisco. other hospitals, and 42 reported missing. 20 m is pile-supported. The fill be- were many broken poles, poles that Most are assumed to have resulted hind a sea wall settled about 1 m toppled in liquefiable soils, and drops from building collapse. Of these, and pushed the wall towards the that were cut when buildings col- 148 happened in the collapse of the piles, slightly bending them. A ware- lapsed, particularly in Pisco. Cathedral de San Clemente located house just behind sea wall has pile- at the Plaza de las Armas, where Communications: The hardwired supported walls. The floor and sur- only two survivors were pulled from system failed, but cell systems were rounding area settled about 60 cm, the debris by search and rescue generally functional. In the main town with extensive evidence of liquefac- teams. Another significant building square of Pisco, temporary phone tion. collapse was at the Embassy Hotel, booths were installed for civilian use where at least 15 people (MSNBC. There is no gas (Figure 22). After the earthquake Gas/Liquid Fuels: COM) were killed when the bottom distribution system; gas is distrib- severed the electrical power grid, two stories of the 5-story structure uted in bottles. There is a gas line the Pisco hospital started its back-up collapsed. Pisco residents reported coming down from the Andes; the generators and allowed residents to that the mass burial of the dead pro- gas is liquefied at a facility in Pisco charge their cellular phones. The hos- ceeded rapidly, but that the families and loaded onto ships in an off- pital also operated by cell phone to of the dead were responsible for shore facility. There was no damage coordinate their post-disaster efforts providing caskets. Many families observed at any of these facilities. and coordinate their needs with the could not afford caskets and, in Peruvian Department of Health. Electric Power: Hydro-generated any event, casket suppliers could electricity comes down from the Water: Pisco gets its water from a not meet the demand, so some Andes to a substation in Indepen- infiltration gallery 30 km east of town. bodies were buried wrapped only in dencia. There was no apparent Damage to the pipeline (AC and rein- blankets. damage to the substation. Power is forced concrete) along the transmis- There are more data available on then distributed at 60 KV from this sion line into town reduced the flow earthquake-related injuries. The substation through seven circuits from 300 lps to about 170 lps. Water main building at the tertiary care serving Pisco, Ica, and Paracas. was distributed from the operations hospital in Pisco collapsed, requir- On the circuit serving Paracas/San center by tank truck. The Pisco ing transport by ambulance of 192 Andreas, three poles broke. The system had one 1400 m3 concrete significantly injured patients to other two transformers at this substation elevated tank where the operations hospitals in Lima and Ica. Epidemio- (60 KV> 10 KV) rolled on their center is located, and one 4500 m3 logical data collected by the Ministry tracks, resulting in broken insula- ground-level tank. The elevated tank of Health indicate a total of 2,615 in- tors where the wire connections support structure had spalling at the juries between August 15 and Sep- were fixed. The building housing joints. The ground level tank ap- tember 3. The Ministry of Health’s the substation had only slight struc- peared to be OK. Tambo de Mora lead physician deployed to Pisco tural damage. Substations were gets water from Chincha Baja. The indicated that they were currently back on line within a day. There transmission line connecting the two treating the third wave of injuries,

10 EERI Special Earthquake Report — October 2007

the camp as of September 8. There were 30 on-site latrines, but no shower facilities. In contrast, the self-organized Albergue Tropical in San Andres (south of Pisco), housing 2500 people from 585 families, appeared to have little external assistance. Two crates of hospital goods remained unopened. Wet rations were provided once a day by a private company (Figure 24), but otherwise food supplies were limited. Power was connected on September 9. The camp had centralized water supply, 23 latrines, but no shower facilities. Temporary housing is being provided by the government, to residents that are able to clear debris from their plots. Two types of temporary housing were observed: Figure 23. The largest shelter camp, Albergue Bolognesi, in the city of Pisco. lightweight wood (Figure 25), or metal frame (Figure 26). The associated with clean-up and debris of Health, the Brigada Medica Cu- metal frame structures have two removal. The first treatment peak bana, and the Salvation Army were rooms, with no plumbing, pow-er, or took place on the day after the providing assistance. This camp was sewage hook-ups provided. Those earthquake; the second was 5-7 well-organized, with numbered tents observed had no floors, and were days after the earthquake. laid out in rows and “named” streets open to the elements at the back. for walkways (Figure 23). Camp Post-event medical care in Pisco maps, rules of conduct and an orga- was being provided at several lo- Acknowledgments nizational chart were posted by the cations: the tertiary care hospital The EERI team is grateful for sup- camp’s entrance. Water was available (operating out of their new emer- port provided by Julio Vargas, pro- from a central distribution point. Wet gency department building, under fessor at the Pontificia Universidad rations were being distributed, and it construction and not occupied at Católica del Perú (PUCP), and appeared that residents had no other the time of the earthquake), two Eduardo Ismodes, dean of engi- means of preparing meals. Power temporary hospitals operated by a neering at PUCP. The GEER team had reportedly just been supplied to total of 60 Cuban medical profes- sionals from the Brigada Medica Cubana located at the two largest formal shelter camps (Albergue Grau and Albergue Bolognesi), as well as Ministry of Health medical personnel at these and other shelter camps. As of September 9, six formal shelter camps were in operation in the city of Pisco. The number of people occupying each camp varied con- siderably, as did the available ser- vices at each location. The largest camp, Albergue Bolognesi, was reportedly sheltering 470 families and approximately 3,000 people, and Figure 24. Wet rations were provided once a day by a private company at the the Peruvian military, the Ministry self-organized shelter camp Albergue Tropical in San Andres (south of Pisco).

11 EERI Special Earthquake Report — October 2007

Lives, Phil. Trans. R. Soc. A 364 (1845): 1947-1963. Synolakis, C. E., and E. A. Okal, 2005. Perspective on a decade of post-tsunami surveys: 1992-2002, in: Tsunamis: Case studies and recent developments, ed. by K. Satake, Adv. Natur. Technol. Haz- ards 23:1-30. Tadepalli, S., and C. E. Synolakis, 1994. The Run-Up of N-Waves on Sloping Beaches, Proceedings: Mathematical and Physical Sci- ences 445 (1923): 99-112. Tavera, H., and Bernal, I., 2005. Figure 25: Lightweight wood type of temporary housing. Spatial distribution of rupture areas and seismic gaps in the western coast of Peru, Volumen Especial national Conference of Earthquake is thankful for the help of Dr. Jorge No. 6, Alberto Giesecke Matto, Engineering, Centro Peruano Ja- Alva-Hurtado, Dr. Carlos Zavala, Sociedad Geologica Del Peru, ponés de Investigaciones Sismicas and Dr. Xenon Aguilar of CISMID p. 89-102, in Spanish. (the Peru-Japan Center for Seismic y Mitigación de Desastres (CISMID), Research and Disaster Mitigation), Universidad Nacional de Ingeniería, Tavera, H., Bernal, I., and Salas, H., and Dr. Ronald Woodman and Her- Lima, Peru. In Spanish. 2007. The August 15, 2007 (7.9Mw) Pisco Earthquake, Ica Department, nando Tavera of the Peruvian Geo- Liu, P.L.-F., P. Lynett, J. Fernando, Peru (Preliminary Report), Boletín physical Institute. EERI gratefully B.E. Jaffe, H. M. Fritz, B. Higman, R. Sociedad Geológica, Instituto Geo- acknowledges Nabih Youssef’s Morton, J. Goff, and C. E. Synolakis, físico del Perú, August 2007, in generous $1,000 donation to this 2005. Observations by the Interna- Spanish. reconnaissance effort. tional Tsunami Survey Team in Sri Lanka, Science 308 (5728): 1595. Trifunac, M.D., and Brady, A. G., References 1975. A study of the duration of Sieh, K., 2006. Sumatran Megathrust strong earthquake ground motion, Abrahamson, N. A. and Silva, W. J., Earthquakes: From Science to Saving BSSA (4), 139-162. 1997. Empirical response spectral 66 attenuation relations for shallow crustal earthquakes, Seismological Research Letters 68 (1): 94-127. Fritz, H. M., and J. C. Borrero, 2006. Somalia field survey of the 2004 Indian Ocean Tsunami, Earthquake Spectra 22 (S3): S219-S233. Ji, Z. and Cheng, Y., 2007. Prelimi- nary Result of the Aug. 15, 2007 Mw 8.0 Coast of Central Peru Earthquake, Web report (http:// earthquake.usgs.gov/eqcenter/ eqinthenews/2007/us2007gbcv/ finite_fault.php). Lazares La Rosa, F., Almora, J.P., López Vásques, J., Cabrejos Hurtado, J. J., and Piedra Rubio, R., 2007. Analysis of accelero- grams, Ica earthquake of August 15, 2007, Presentation in Inter- Figure 26: Metal frame type of temporary housing.