Preliminary Observations on the Southern Peru Earthquake of June 23, 2001

EERI Special Earthquake Report — November 2001

Learning from Earthquakes Preliminary Observations on the Southern Peru Earthquake of June 23, 2001

This report is condensed from the Introduction Only one strong-motion record of post-earthquake surveys of numer- the main event has been recovered ous individuals and teams, most of On June 23, 2001, at 8:33PM UTC, from a station in Moquegua. Field which traveled to Peru with their 3:33 PM local time, a Mw=8.3 surveys and local reports indicate .own funding. They contributed their (USGS) earthquake struck near the that the duration of ground motion ﬁ ndings to this report, one of a coast of south-central Peru along was relatively long (45-60 seconds), series in the Earthquake Engineer- the subduction zone between the with peak ground accelerations ing Research Institute’s Learning Nazca and South American plates, ranging from approximately 0.10g from Earthquakes (LFE) Program. at coordinates 16.2 S, 73.75 W. The near Arequipa to 0.30g near Mo- The publication and distribution of districts of Arequipa, Moquegua, quegua. this report are funded by the Nation- Tacna, and Ayacucho sustained al Science Foundation as part of losses. There were dozens of strong The National Institute for Civil De- EERI’s LFE Program, under Grant aftershocks in the region. The south- fense of Peru reports 77 dead, 68 #CMS 0131895. ern coastline was affected by a tsu- missing, 2,713 injured, 213,430 peo- nami following the main event; the ple affected, 33,570 houses dam- coastal towns of Ocoña and Cam- aged, and 25,399 houses destroyed. ana were particularly hard hit.

Figure 1 Southern Peru earthquake and associated seismicity, June 23 -- July 23, 2001. Offshore epicenters extending from the main-shock epicenter southeast to Ilo probably occurred on or near the main-shock rupture surface. Boundaries of the 1868 rupture are as proposed by Dorbath et al. (1990). Inland shocks occurred on different faults than the main-shock and may have occurred independently of the main shock. (Dewey)

Figure 1 Southern Peru earthquake and associated seismicity, June 23–July 23, 2001. Offshore epicenters extending from the main-shock epicenter southeast to Ilo probably occurred on or near the main-shock rupture surface. Boundaries of the 1868 rupture are as proposed by Dorbath et al. (1990). Inland shocks occurred on different faults than the main shock and may have occurred independently of the main shock. (Dewey)

1 EERI Special Earthquake Report — November 2001

The Earthquake and After- per year. The June 23rd earthquake shock Activity would have substantially reduced elastic strain in the 300-km zone James W. Dewey, U.S. Geological ruptured, but much of the remaining Survey, Denver, Colorado; Juan 700 km of the plate interface that Bariola, JBB S.A.C., Peru ruptured in 1868 and 1877 probably still remains in a condition of high The earthquake resulted from thrust elastic strain accumulated since the faulting on the boundary between late nineteenth century. the Nazca and South American plates. The locations of main-shock One strong-motion record was ob- hypocenter, early aftershocks (Fig- tained at Moquegua. Two other in- ure 1), and moment-tensor solutions struments in the epicentral area of the earthquake imply that fault- apparently malfunctioned. The rupture nucleated at a depth of ap- strong-motion site is situated approximately 30 km beneath the proximately 175 km from the source coastline between Chala and Cam- of maximum slip identiﬁ ed by Kiku- ana, and propagated updip toward chi and Yamanaka (2001), and 100 the Peru-Chile trench and to the km from nearest part of the main- southeast, terminating near Ilo. The shock fault-plane, offshore Ilo, that is causative fault-rupture had an along- suggested by the earthquake’s focal strike length of about 300 km and a mechanism and aftershock distribu- down-dip width of about 100 km. tion.

The average fault displacement was The Universidad Nacional de Ing- several meters, but seismograms of enieria and the Centro Peruano the earthquake imply that slip varied Japones de Investigaciones Sis- greatly along the fault surface. For micas y Mitigacion de Desastres example, a moment-tensor analysis (CISMID) recovered and corrected conducted by Kikuchi and Yamanaka the ground acceleration time histo- (2001) shows maximum slip occur- ries (Figure 2) and the correspond- ring approximately 150 km south- ing acceleration response spectra east of the main-shock hypocenter. (Figure 3) from the Moquequa in- The duration of the overall main- strument recording in east-west, shock rupture was approximately 80 north-south, and vertical directions. seconds (Harvard Centroid Moment PGA values were .295, .220 and Tensor solution). .160g in the east-west, north-south and up directions, respectively. The earthquake struck part of an Larger accelerations were observed approximately 1000-km long section in the approximately normal-to-fault of the western South American plate direction comparatively to the paral- boundary that many seismologists lel-to-fault-direction. The ratio of believed had high potential for great peak-vertical to peak-horizontal ac- earthquakes in upcoming decades celeration was 54%, slightly larger (Dorbath et al, 1990). Most of the than that in previous Peruvian plate-interface thrust zone between events. Chala, Peru (15.9 S), and Mejil- Figure 2 Acceleration time histories lones, Chile (23 S), had been seis- Aftershocks were predominantly of the east-west, north-south, and mically quiescent (deﬁ ning a so- south of the main shock. Fault rup- vertical components of the corrected called “seismic gap”) since the great ture occurred following the same ground-motion recording in the city earthquakes in 1868 (Southern Pe- path shown by aftershocks — north of Moquegua. (The Universidad ru) and 1877 (Northern Chile). The to south directivity as has been com- Nacional de Ingenieria and the zone has been accumulating elastic mon in previous earthquakes in Peru Centro Peruano Japones de Inves- strain as the Nazca plate subducts (Wyss et al., 2000) — so energy- tigaciones Sismicas y Mitigacion de beneath the South American plate focusing effects may have caused Desastres [CISMID]). at a rate of approximately 78 mm larger intensities in towns south of

2 EERI Special Earthquake Report — November 2001

the epicenter such as Moquegua.

Geotechnical Observations

A. Rodriguez-Marek, Washington State University, Pullman; P. Repet- to, URS Corporation, Denver, Col- orado; J. Wartman, Drexel Univer- sity, Philadelphia, Pennsylvania; D. Baures, URS Corporation, Denver; E. Rondinel, Pontiﬁ cia Universidad Católica del Perú; J. Williams, URS Corporation, Denver; J. Zegarra-Pel- lane, Pontiﬁ cia Universidad Católica del Perú. The National Science Foundation sponsored the reconnaissance team, with additional support from Washington State Univer- sity, Drexel University, the Catholic University of Peru, and URS Corpo- ration.

Ground Failure: Considerable damage was done to highways in the regions of Tacna, Moquegua, and Arequipa as a result of landslides, rockfalls, ﬁ ll densiﬁ cation, and liquefaction. Observed ground failure was concentrated along Peru’s primary north-south highway, the Pan- American Highway, and several of the other major roads connecting the cities of Atico, Camana, Arequi- pa, Ilo, Moquegua, and Tacna. Sev- eral instances of relatively minor ground failure were also observed in urban centers of the region.

Fill densifi cation caused considerable damage to bridge abutments and approaches. Roadway damage also occurred at the location of valley fi lls. The fi ll soils typically con- sisted of poorly graded coarse to fi ne sands and non-plastic silts. De- formations were largest along the edges of highway fi lls.

Liquefaction was observed at three bridge sites located in alluvial val- leys. In one instance, liquefaction- Figure 3 Acceleration response spectra obtained from the corrected ground induced lateral spread severely motion along the east-west, north-south, and vertical components of the damaged a concrete bridge and recording in the city of Moquegua. (The Universidad Nacional de Ingenieria several residences located along the and the Centro Peruano Japones de Investigaciones Sismicas y Mitigacion banks of a river (Figure 4). At de Desastres [CISMID]) another site, liquefaction caused a

3 EERI Special Earthquake Report — November 2001

the ridge slopes varied, but were on the order of 35 to 60 degrees.

Engineered Structures: Inspection of three large copper mines located in the region affected by the earthquake indicated relatively minor damage to mining facilities. The only notable exception is the liquefaction fl ow of a heap-leach pad. Local authorities reported no damage to dams in the region. A concrete-face rockfi ll dam at a mine suffered minor damage due to fi ll settlement at a sacrifi cial portion of the crest.

Figure 4 The offset in the corn rows (left to right near middle of photo) is Landslides due to localized displacement induced by lateral spread. (photo: geotechnical reconnaissance team) David K. Keefer, U.S. Geological Survey (USGS), Menlo Park, Califor- localized bearing-capacity failure of that public schools in Peru that were nia. This reconnaissance was car- a concrete column supporting a built in the same time period were ried out by personnel from the large and relatively new bridge. A generally constructed to a similar USGS (David Keefer, Robert Tilling, third site had about a meter of lateral design and using similar construc- and James Smith, who is currently spread, but no apparent damage to tion practices throughout the coun- on assignment to the USAID Offi ce structures. try. Observations of the performance of Foreign Disaster Assistance of school buildings in different loca- [OFDA]); the Instituto Geológico Landslides and slope stability fail- tions permit a comparative evalua- Minero y Metalúrgico del Perú ures occurred at localized cut and tion of the ground motions experi- (INGEMMET — Julio de la Cruz, fi ll sections along highways (Figure enced during the earthquake. Antonio Guzmán, Lionel Fídel, 5). In most cases, slope movements Segundo Nuñez, Marco Rivera, and in the fi ll section were due to fi ll den- Some of the damage to houses and Bilberto Zavala); and the Instituto sifi cation. Rockfalls were observed businesses located at or near the Geofi sico del Perú (IGP — Juan along many of the steep road cuts top of several steep ridges in the Carlos Gómez and Jersy Mariño). in the region. Observed rockfalls towns of Arequipa and Moquegua Funding for USGS participation was ranged from single-block failures to can possibly be attributed to topo- provided by OFDA, and coordination raveling of both bedrock and the graphic effects. The inclinations of and logistical support were provided exposed colluvial material. Some planar and wedge failures within the bedrock were also observed. Natural slopes composed of bedrock or colluvium had very little observed failure, with the exception of displace- ments of a shallow veneer of soil in sandy slopes.

Site Effects: The concentration of damage in some areas of Tacna, Moquegua, and Ilo suggests the in- ﬂ uence of site ampliﬁ cation in the resulting damage levels. Local engineers indicated the presence of sand deposits in areas of concentrated damage, while areas under- lain by stiffer gravel deposits suf- Figure 5 Slope failure along highway between Cuajone and Moqegua. fered less damage. It is noteworthy (photo: geotechnical reconnaissance team)

4 EERI Special Earthquake Report — November 2001 by INGEMMET. landslides typically had volumes of a of the epicenter. Throughout this few to a few hundred cubic meters, area most of the landslides were The earthquake caused landslides and individual landslides are typical- either small rock falls or rock slides throughout an estimated area of at ly spaced a few tens to hundreds of or slumps in highway fi ll. least 25,000 km2 stretching from the meters apart along the susceptible epicenter along the coast south to slopes. Landslides in the better-in- The types of landslides and types of the region between Ilo and Tacna, durated volcanic rocks to the north- materials that produced landslides and inland as far as Arequipa and east of Moquegua were typically in this earthquake are similar to Canderave. The immediate effects smaller and spaced farther apart. those caused by the vast majority of landslides primarily concerned of other earthquakes worldwide highways; repairs were underway at South of Moquegua, in the lower Ilo (Keefer, 1984), but the absence of many localities, and roads were con- Valley, shallow landslides with sizes large landslides outside the small stricted to one traffi c lane at several and spacing similar to those in the zone south of the epicenter con- others. A longer-term hazard from Moquegua formation occurred in the trasts signifi cantly with fi ndings from these landslides derives from their well-indurated but closely fi ssured many other earthquakes of compa- transportation of large amounts of bedrock of the coastal cordillera and rable magnitude. loose sediments into drainage chan- in river-terrace materials (Figure 6). nels; this sediment may exacerbate Landsliding was also severe around Tsunami the damage due to post-earthquake the town of Canderave, east of Mo- fl ooding, brought on by such events quegua, where the largest land- Lori Dengler, Humboldt State Uni- as strong El Niños in coastal areas. slides were rock falls off the scarp versity. International Tsunami Survey of a large, pre-existing deep-seated Team (ITST) members: Sebastian The largest landslides were ob- landslide. Throughout this southern Araya, Lori Dengler, Humboldt State served along a 20-km stretch of region, slumps in highway fi ll were University; Jose Borrero, Matt the Pan-American Highway between also relatively common. Swensson, the University of South- the town of Atico and the village of ern California; Emile Okal, Brandon Cerro de Arena, where a landslide No landslides were observed along Gomer, Northwestern University; damaged a dwelling and restaurant. the Pan-American Highway north of Shunichi Koshimura, National Oce- Many landslides in this area were the epicenter, while between Cerro anic & Atmospheric Administration rock falls, and rock slides in rock that de Arena and the Moquegua area, Pacifi c Marine Environmental Labo- was well indurated but closely fi s- landslides were typically small and ratory (NOAA PMEL); Gustavo Laos, sured. Other landslides were debris spaced far apart. Soil liquefaction Daniel Olcese, Fernando Vegas, slides in colluvium and rotational effects, including lateral spreading Direccion de Hidrografi a y Naveg- slumps in unconsolidated dune sand landslides, occurred in the lower acion, Peru Navy; Modesto Ortiz, and in highway fi ll. These landslides courses of some of the major rivers Centro de Investigación Científi ca caused substantial local damage to that cross the coastal plain south y de Educación Superior de Ense- the highway, and traffi c was still being delayed at sites of repair more than a month after the earthquake.

There were also abundant earthquake-induced landslides inland from the southern end of the fault rupture, in the vicinity of the towns of Moquegua, Ilo, and Canderave. Around Moquegua, many small and moderate-sized falls and slides of rock and soil occurred along stream banks, roadcuts, and other steep slopes composed of material de- rived from the Moquegua formation, which consists largely of weakly cemented clays, sandstones, and conglomerates. The largest of these landslides is estimated to have a Figure 6 Typical rock fall from wall of lower Ilo Valley near Algarrobal. (photo: volume of about 5,000 m3. Other David K. Keefer)

5 EERI Special Earthquake Report — November 2001

nada (CICESE), Encinitas, Mexico; utes after the earthquake extending signifi cant than the tsunami effects. Vasily Titov, NOAA PMEL and the 50 to 100 meters offshore. Most The largest waves (5-8 m) and University of Washington. respondents, upon seeing the water deepest penetration (1-1.2 km) recede, moved to higher ground. (Figure 7) produced by this tsunami The tsunami produced by the earth- The water remained low for “a long coincided with the most-developed quake was recorded throughout the time,” variously described as 15 to beach resort area along the south- Pacifi c basin. Maximum recorded 20 or more minutes. Many expected ern Peruvian coast. La Punta and wave heights (peak-to-trough) were a signifi cant positive wave to follow, the adjacent resort communities 2.57 m in Arica, Chile; 0.7 m in Hilo, but the incoming wave did not signi- have many summer homes, occu- Hawaii; 0.4 m in Crescent City, Cali- fi cantly exceed the ambient high pied by about 5000 residents during fornia; 0.24 m in Sandpoint, Alaska; water level. Several observers de- the peak summer season. 0.55 m on Chatham Island, New scribed cracks opening in the ex- Zealand; and 0.5 m at Hanasaki, posed seafl oor and dirty water with Many people observing the water Japan. A complete tabulation of re- a distinctive odor bubbling up. Three recede self-evacuated. No one recorded tsunami wave heights is or four more oscillations of the sponded to the ground shaking posted at http://wcatwc.gov/06-23- water were observed, the draw- even though all felt the earthquake 01.htm. One hour after the earth- down always apparently larger in strongly. Eyewitnesses described an quake, the Pacifi c Tsunami Warning amplitude. No structures were re- initial draw-down that lasted 15 min- Center declared a tsunami warning ported damaged by the tsunami in utes or more. The initial positive for Peru, Ecuador, northern Chile, these areas. wave was small, but followed by two and Southern Colombia. Later, that destructive waves of near identical was expanded to include all of the The impacts were quite different size. west coast of South and Central within the municipality of Camana. America, Mexico, and French Poly- It has eight districts that include Buildings in the coastal area are of nesia. Mollendo and the adjacent coastal resort communities, farming three general types. In the farming port city of Matarani in Southern regions, and fi shing communities, areas and fi shing villages, weak Peru organized evacuations of low- with a total population of about adobe structures and shacks of lying areas after the water was ob- 53,000 people. Damage was con- bamboo and other light-weight served receding. Port authorities centrated along a 20-km long fl at materials predominate. In the resort directed the Matarani evacuation, coastal beach no higher than 5 m area, most structures are built on and local police directed the Mol- above sea level. The tsunami de- concrete slab foundations with rein- lendo evacuation. stroyed over 2,000 structures, dam- forced columns. Unreinforced bricks aged an additional 1,000, killed 22 are fi lled in to form the walls be- The tsunami was observed by eye- people and left 52 missing, and tween the columns. A number of witnesses from Tanaka, Peru, to fl ooded 2,000 hectares of farmland. hotels, some restaurants, and a few Arika, Chile. Outside the municipal- Estimates of undamaged structures homes were built of reinforced con- ity of Camana, the tsunami was de- within the inundation zone were be- crete. There were no wood struc- scribed by observers as an initial tween 12 and 30. In Camana, the tures in the inundation zone. Adobe draw-down that began 10 to 15 min- earthquake impacts were much less and infi lled wall structures per-

Figure 7 Tsunami inundation and ITST survey lines in the Camana area. (Base map from the Planning Department of the Municipalidad Provincial de Camana, inundation by L. Dengler based on ITST data. Map compiled by Verena Kellner, Humboldt State University.)

6 EERI Special Earthquake Report — November 2001

Gonzalez, Colegio de Ingenieros de Tacna; Antonio Blanco, Dean of the Colegio de Ingenieros del Peru; and Julio Eguia, driver from Autodemas.

Initial surveys indicated that most damage occurred in areas to the east and southeast of the epicenter of the main event and that historic structures and old houses sustained the most severe damage. Adobe houses were most vulnerable. School buildings also performed poorly, except for the few new ones Figure 8 Reinforced column/infi lled brick wall structure in the La Punta area. designed to the 1999 code. Most of Brick walls perpendicular to wave impact were typically blown out. (photo: the damage could be attributed to Sebastian Araya) confi guration problems such as soft stories or short-column effects. Both new and older schools with short formed very poorly. No signs of ado- 1) a tsunami-aware coastal popula- column problems performed poorly. be structures remained. Infi lled walls tion: most of the people interviewed Schools designed according to the were typically blown out on walls knew what tsunamis were, recog- 1999 Peruvian code had isolation perpendicular to the wave direction nized the water draw-down as a details between infi ll masonry walls (Figure 8). Signifi cant scour oc- sign of danger, and self-evacuated; and the concrete framing, and thus curred to structures at the corners, 2) time of year: the earthquake and avoided the short column failures. often undermining the slab founda- tsunami occurred in winter, so the tions. Reinforced concrete structures beach discotheques, hotels, and Moquegua: Moquegua was the city with thicker foundations were most cafes were empty; 3) time of day: hardest-hit by this earthquake. The likely to survive. because the earthquake and tsu- records (see Figure 2) indicate that nami occurred in mid-afternoon peo- the peak ground acceleration was Demographic information on earth- ple could see the water recede; and 0.30g and the response spectra had quake victims compiled by the Insti- 4) ambient tide level: the tsunami a plateau at approximately 0.65g- tuto Nacional de Defensa Civil sug- coincided with a minus-40-cm tide, 0.70g in the range from about 0.15 gests that most of the victims of one of the lowest tides of the year. seconds to 1 second (see Figure 3). ground-shaking damage were the The damage in Moquegua is mainly young or the elderly. However, about Structural Damage to old and new adobe houses. Nu- a third of the tsunami victims were merous adobe buildings collapsed between 30 and 50. No demograph- Eduardo Fierro, Wiss, Janney, Elst- in the downtown area, and many ic information was available for the ner Associates, Inc. (WJE), Emery- others sustained heavy damage to 52 persons listed as missing, but ville, California. Assembled and the point of being almost unstable. anecdotal evidence gathered by the edited by Ayhan Irfanoglu and Cyn- In the area of San Francisco, the hill- reconnaissance team indicated that thia Perry, also of WJE, based on side was nearly totaled due to col- most were farmers working in the photos and text from Peru by Fierro. lapse of new adobe houses. One onion fi elds. The normal Saturday Fierro acknowledges the assistance school had a partial collapse due to work shift ended at 4:00 pm, and of local offi cials and members of the short columns (Figures 9 and 10). people were reluctant to leave be- engineering community in Peru in The cathedral lost one of its vaults, fore their shift ended. providing access, accommodations, and the Church of Belen sustained and information. In particular, he damage to its tower and walls. While the extent of inundation and thanks Jorge Ojeda, Executive Di- the number of structures damaged rector of Autodemas; José Tong, Arequipa: The ground accelerations or destroyed was signifi cant, the Dean of the Colegio de Ingenieros were apparently not very high in number of lives lost was considera- de Lima; Daniel Torrealva and Ale- Arequipa, a city of 700,000 located bly reduced from several other re- jandro Muñoz, Universidad Catolica about 120 miles inland from the epi- cent tsunamis. The difference in del Peru; Javier Pique, Universidad center of the main event. There casualties has several causes: Nacional de Ingenieria; Adolpho are no ground motion records from Arequipa, apparently due to instru-

7 EERI Special Earthquake Report — November 2001

Figure 10 View of the collapse of one of the short columns in the school building shown in Figure 9. (Photo: E. Fierro)

with heavy blocks of low strength sillar supported by steel rails.

Tacna: The University Jorge Basa- dre in Tacna has two campuses, Figure 9 School building in Moquegua. Partial collapse due to failure of short and some damage was observed columns. (Photo: E. Fierro) in both. In the old campus building, some parapet walls at the fourth ment malfunction. However, local founded in 1579. Many traditional fl oor had to be demolished, since people reported that the duration of domed and vaulted structures were they were unstable in the out-of- motion was approximately one min- extensively damaged. Most of the plane direction. On the new campus, ute. Primarily historical buildings and historical monuments are made with a three-story building had moderate older stone masonry structures sus- blocks of sillar (consolidated volca- damage to the infi ll masonry in the tained damage. An estimate of 0.10g nic ash) laid with lime or cement form of diagonal cracking. However, appears reasonable for the peak mortar. Sillar is also used for hous- the infi ll was strong enough to form ground acceleration in Arequipa, ing. A damaged vaulted ceiling in an diagonal compression struts be- based on observed damage pat- older home is shown in Figure 12. tween the beam-column joints, terns. Many of the older roofs were made which cracked the corner-column

The city’s historic cathedral, built in 1612, sustained heavy damage, mainly due to the collapse of one of its towers. The cathedral is situated on the Plaza de Armas in Arequipa’s historic center, a UNESCO World Heritage site. It was damaged by previous earthquakes and ﬁ re, and was substantially rebuilt in the 19th century. Both towers were apparently rebuilt most recently in about 1940. A photo extracted from a video taken during the earthquake shows part of the collapse sequence of the tower (Figure 11). The tower on the left collapsed and fell through the vaulted dome above the main altar of the cathedral.

Many cathedrals and religious monuments sustained damage, such as Figure 11 Cathedral of Arequipa, still photo from sequence taken during the Convent of Santa Catalina, main event. (photo purchased by E. Fierro)

8 EERI Special Earthquake Report — November 2001 joints in shear on the fi rst and second fl oors. The poorly detailed columns of the elevated tank located on the roof (fi fth fl oor) had signifi cant damage but did not collapse. The Colegio Nacional Mariscal Caceres sustained extensive damage to approximately 60 short columns in the typical campus buildings. The school principal was building tem- porary wooden classrooms following the earthquake.

A municipality building outside Tacna suffered severe damage to the short columns and to the few masonry walls that existed. The building had typical poor short-column conﬁ guration without any transverse reinforce- ment, which caused these columns to crush and lose several inches of height. The building is not readily Figure 12 Damge to typical historic dwelling with vaulted roof constructed of repairable and will be demolished. “sillar” blocks. (Photo: E. Fierro)

The Peruvian National Institute for Santa Fe Railroad, Kansas City, or ground water. A number of major Civil Defense reports that distribu- Kansas. rivers fl ow to the Pacifi c Ocean, tion of relief material is ongoing. most of which are dammed or other- Peru may need long-term interna- In general, lifeline damage was less wise diverted at a number of loca- tional fi nancial and technical support than expected compared to that of tions to meet potable and agricultur- for repair and rehabilitation of the other similar magnitude events. This al uses. The government is in the damaged historic structures, which was apparently due to the large dis- process of completing major dams, are an important part of its interna- tances between major population canals, and potable water supply tional cultural heritage. centers and the general absence of systems in each province (state) in lifelines in other areas. southern Peru. Most of the dam- Lifelines age in the upper reaches of this sys- There was greater damage in cities tem was due to landslides blocking Curtis Edwards, Pountney Consult- 300 to 500 km southeast of the origi- exposed canals. There was no re- ing Group, Inc., San Diego, Califor- nal epicenter (Atico) than in cities ported damage to dams, and tunnel nia. The Post-Earthquake Investi- that were closer. The larger cities in systems had not been inspected at gation Committee of the Technical this area of southern Peru are lo- the time of our investigation. Council on Lifeline Earthquake Engi- cated in the upper elevations away neering (TCLEE), a technical coun- from the coastal area. All of these There are major water treatment cil of the American Society of Civil areas had some lifeline damage, plants in Arequipa, Moquegua, and Engineers (ASCE), organized a with the majority of damage in the Tacna. The Arequipa plant overall team of fi ve TCLEE members with area in and around Moquegua. had minor damage, but sloshing support from ASCE to perform a re- Landslides (rockfalls and poorly damaged the tube settlers and the connaissance of the lifelines in the compacted fi lls) caused most of the circular clarifi er effl uent troughs. In earthquake area. The team consist- damage to roads, railroads, and addition, there was a minor chlorine ed of Curtis Edwards (team leader); water systems. There were no re- leak due to unrestrained chlorine John Eidinger, G&E Engineering ports of major lifeline damage in cylinder storage. The Arequipa plant Systems, Oakland, California; Mark areas north of Atico. is at 100% capacity, and operators Yashinsky, Caltrans, Sacramento, indicated that to maintain service to California; Anshel Schiff, Precision Water Systems: The majority of the city, they could not take a clari- Instruments, Los Altos Hills, Califor- land east of the Andes is very dry fi er out-of-service for repairs. There nia; Bill Byers, Burlington Northern- and requires use of imported water was additional damage to the lab- oratory. The Moquegua plant had

9 EERI Special Earthquake Report — November 2001

signifi cant damage to fi berglass baf- fl es in the fl occulation basin, which were not repaired at the time of our inspection. As a result, the clarifi ers were out-of-service, leaving only the sand fi lters and chlorination as the only remaining treatment processes. Tacna has two treatment plants that suffered minor damage. The major damage was to an older partially buried reinforced-concrete clearwell. The columns and roof were heavily corroded and near collapse.

There was minimal damage to water distribution systems in these cities. Breaks of the PVC, asbestos-cement, and cast-iron pipes appeared to be due to shaking and were re- Figure 13 Rock Fall Blocking PeruRail Line. (photo: PeruRail) paired within two weeks after the event. There were no reports of km from the rupture surface, dam- or spread away from the roadway. damage to water storage tanks. age did not correlate with distance, On the Pan-American Highway, a Most of these are constructed of re- indicating that other factors were concerted effort was made to keep inforced concrete, and those in the more important in determining the the road open, even when half the ﬂ atter regions are elevated. One extent of damage. Obvious factors roadway was obstructed due to steel water tank at the Ilo power include depth of cuts, height of ﬁ lls, rockfalls or subsidence. This dam- plant had stretched anchor bolts, but side slopes, and differences in the age continued to impede travel for no other damage or leakage. material affected. months after the earthquake.

Railroads: The railroads of Peru do Highway Systems: There was Highway structures such as bridges not constitute a network, but include widespread damage to the highway and culverts were also damaged separate entities not physically con- system as a result of the earth- due to the use of poorly compacted nected. Two of these sustained ex- quake. Much of this damage was material around abutments and tensive damage: PeruRail, a com- due to the poor performance of wingwalls. This resulted in subsid- mon carrier providing freight and loosely compacted fi lls, and rock ence of the bridge approaches. In passenger service; and the indus- falls and landslides from steep-cut addition, liquefaction caused failure trial railroad of the Southern Peru slopes. The area affected by the of some wingwalls and approaches. Copper Corporation, which trans- earthquake is extremely arid and Most bridges were in use 30 days ports copper concentrate from in- mountainous. The primary highway later, but one older steel truss bridge land mines to a smelter and refi nery system consists of the north-south was closed due to failure of bearing on the coast, and supplies and Pan-American Highway, a well-main- supports (Figure 14). Box culverts equipment to the mines. Major line tained two lane blacktop road. In ad- also performed poorly due to loosely segments, with a combined length dition, there are also many packed- compacted material. An 18’-by- on the order of 700 km, were out of earth and unimproved local roads. 40’-tall single-box culvert remained service for three to seven days. The The roads are cut out of the sides undamaged on the Pan-American most severe damage involved mate- of mountains with fi ll material used Highway, but the roadway over the rial sliding or falling onto the track, or to support the outside shoulder and culvert settled several feet, and the embankment failures. In some cuts, roadway. The uphill slopes were not wingwalls were shattered due to ac- track was buried under as much as 5 closely inspected to remove loose tive pressure from the surrounding meters of material (Figure 13). material or suffi ciently graded to pre- soil. vent slides. The material used for the There was no railroad bridge dam- fi lls was often fi ne silts and sands The severity of damage to the high- age, but few bridges are located in that were poorly compacted. During way system can be reduced in the the affected areas. Tunnel damage the earthquake, material on the up- future by replacing weak fi lls along was minor. Within distances of about hill side fell onto the roadway, and roads and highway structures with 300 km from the epicenter and 150 material on the downhill side lurched 90% compacted material. Slopes

10 EERI Special Earthquake Report — November 2001

only damage reported was the loss of two control-tower windows. This type of damage has been common in U.S. earthquakes. The airport did not close as a result of the damage, and the effects on operations were probably minimal, as there are relatively few operations (takeoffs and landings) relative to airport capacity.

Communications between Lima and Arequipa were disrupted due to damage to microwave transmission towers and break(s) in the land- based optical-fi ber cable that links the two cities. The reported failure of this cable most likely was associated with a damaged bridge used to carry the cable over a river and dry wash, or with slumping of the shoulder of the road that contained the buried cable. The reported failure of Figure 14 Damaged bridge abutment. (photo: Yashinsky) the submarine cable disrupting com- above roads should be carefully in- A new 220/138 kV transmission sub- munications with Chile would be spected for loose material, which station was damaged. Transmission consistent with the design and loca- should be removed or draped to pre- voltages in California, where most tion of the cable of the Pan-Ameri- vent future movement. Catchment earthquake damage to power sys- can Cable System. The damage to walls should be constructed be- tems has occurred, are primarily 115 communication systems could not tween slopes and roads to prevent kV, 230 kV and 500 kV. The perfor- be confi rmed. falling rocks from impeding traffi c. mance of 115 kV equipment has been very good, but damage to There is no natural gas system in Power Systems: Overall perfor- some types of 230 kV switchyard southern Peru. Gasoline and liquid mance of power systems was rel- equipment has been common. fuels are transported by truck. De- atively good. In Moquegua, power Because there is so little 138 kV and liveries suffered minor delays due to was disrupted for about 12 hours; 161 kV equipment, its earthquake highway damage. The port facilities disruption in most other areas was performance is not well-known. The in southern Peru are relatively small limited to a few hours unless a local 1999 Chi-Chi, Taiwan, earthquake in comparison to the main port near feeder was damaged. demonstrated the vulnerability of Lima. There was no report or major 161 kV equipment, and this earth- damage to the ports, tank farms, or The most signifi cant observation quake has now demonstrated the support facilities. was at a new coal-fi red power-gen- vulnerability of some 138 kV equip- erating station built to high seismic ment. It also appears that some of Health Impacts of the Earth- standards. It experienced types of the damaged 220 kV equipment was quake damage not observed in gas- and of the same type that has been oil-fi red plants in the United States. qualifi ed to IEEE (Institute of Electri- Kimberley Shoaf, UCLA Center The anchorages of three of the main cal and Electronics Engineers, Inc.) Public Health and Disaster Relief. boiler structure support columns Standard 693 (1). The ground mo- A team of fi ve social scientists from were damaged or failed, and at least tions that caused the damage were a variety of disciplines traveled to two boiler tubes failed. Systems most likely less than 0.5 g, so the southern Peru in September with used for the lateral restraint of the equipment did not survive a moder- support from EERI’s Learning from boiler were also damaged, but this ate earthquake as defi ned in the Earthquakes Program. has been observed before. The IEEE Standard. bases of two large coal-unloading The earthquake had profound ef- clamshell-scoops cranes mounted Other Lifelines: The only airport fects on the population in four de- on a pier failed, and the cranes fell known to have been affected by the partments: Arequipa, Ayacucho, Mo- into the ocean. earthquake was in Arequipa. The

11 EERI Special Earthquake Report — November 2001

quegua, and Tacna. The impacts hospital in the community of Aplao in include a small number of fatalities, the Province of Castilla was signifi - with a larger number of injuries and cantly damaged and completely out persons whose homes were dam- of operation. Unfortunately, it is the aged or destroyed. In addition, the only hospital providing service to the health care infrastructure in south- provinces of Castilla, Condesuyos, ern Peru was severely damaged. and Caylloma. Travel from these provinces to Arequipa for health Injuries and Deaths: Seventy-sev- care takes as long as 15 hours. en people died as a result of the Plans are currently being studied to earthquake, with an additional 68 replace the hospital in Aplao with a people missing and presumed dead. seismically resistant one, and to es- The majority of both the fatalities tablish one or two other hospitals in and missing persons are from the these isolated provinces. In all, 169 Province of Camana in the Depart- health care centers (hospitals and ment of Arequipa, where the tsuna- health centers) were damaged or mi hit. The low overall death total is destroyed in the four departments. attributed by both offi cials and the population to the fact that the earth- Not only were health care buildings quake occurred on a Saturday after- damaged and closed by the earth- noon when most people were awake quake, but health care personnel and either at home or out in the were affected as well. It was esti- streets and parks. Given the large mated that approximately half of the amount of damage to schools, most health care personnel in Moquegua felt that they were lucky the earth- were victims of the earthquake. quake did not occur when children Similar numbers of affected person- were in school. nel are expected in the other four departments. The deaths from the tsunami are tragic reminders of the need for education and tsunami warning sys- References tems. In spite of the fact that a large tsunami hit the same region Dorbath, L., A. Cisternas, and C. in 1868, local offi cials claimed that Dorbath (1990). “Assessment of the there were no plans or education in size of large and great historical place for tsunamis, because none earthquakes in Peru,” Bulletin of the was ever expected in southern Peru. Seismological Society of America, v. In contrast, Lima and Callao have 80, 551-576. both educational plans and a tsunami warning system in place. While Keefer, D.K., 1984. “Landslides offi cials state that they did not be- caused by earthquakes,” Geological lieve that the area was vulnerable to Society of America Bulletin, v. 95, tsunamis, the local population had no. 4, p. 406-421. heard of previous tsunamis; it is a part of local legend that the elderly Kikuchi, M. and Y. Yamanaka (2001). talk about. EIC seismological note: No. 105, posted on the web site of the Univer- Health Care System: Hospitals and sity of Tokyo Earthquake Information health centers were damaged in the Center (http://wwweic.eri.u-tokyo. earthquake. While some of the dam- ac.jp/EIC/EIC_News/105E.html). age was structural, a great deal of nonstructural and functional damage Wyss M., R. Dungar, J. Bariola, S. created problems in providing health Wiemer, “El Platanal Hydrolectrical care to the affected population. Five Project,” Report to Cementos Lima, hospitals in the region were dam- Lima, Feb. 2000. aged to some extent. The adobe