Learning from Construction Failures due to the 2009 L ... - ReLUIS

Learning from Construction Failures due to the 2009 L’Aquila, Italy, Earthquake N. Augenti1 and F. Parisi2 Abstract: An earthquake sequence struck the province of L’Aquila 共central Italy兲 leaving 305 dead, about 1,500 injured, and 29,000 homeless. Hundreds of low-intensity events occurred between January and March, 2009. The mainshock took place on April 6, 2009, and its epicenter was located at about 6 km southwest of L’Aquila town; three stronger aftershocks happened on April 7 and 9, 2009. This paper focuses on actual performance of older and more recently constructed building structures during the earthquake sequence. After the main seismological characteristics of the sequence are described, the most significant observed damages are analyzed and associated with theoretical failure modes for both reinforced concrete and unreinforced masonry buildings. Since older masonry structures were more seriously damaged, the effects of the earthquake are described with more emphasis to ordinary masonry and cultural heritage buildings 共churches, palaces, and castles兲. In conclusion, a number of lessons may be learned from the L’Aquila earthquake sequence. Several features are highlighted and some proposals are given to upgrade the current methods of structural analysis, as well as the existing codes. DOI: 10.1061/共ASCE兲CF.1943-5509.0000122 CE Database subject headings: Earthquake engineering; Losses; Site surveys; Buildings; Structural failures; Italy. Author keywords: Earthquake engineering; Losses; Site surveys; Buildings; Structural failures; Italy.

Introduction In 2009, an earthquake sequence struck the Abruzzo region 共central Italy兲 whose capital town is L’Aquila, which was built between 1240 and 1259 by Federico II, Holy Roman Emperor and King of Sicily, and is positioned approximately 92 km northeast of Rome. Hundreds of low-intensity tremors took place in that area up to the seismic event of March 30, 2009 whose local 共i.e., Richter兲 and moment magnitudes were estimated as M L 4.0 and M W 4.4, respectively. The continuous occurrence of aftershocks, along with the presence of individuals and organizations pretending they had been able to predict time and location of the events based on measurements of radon gas releases from the Earth’s crust, and raising alarms about the next occurrences, resulted in a very high level of concern in the population 共Kerr 2009兲. On April 6, 2009, a strong earthquake hit the L’Aquila province at 03:32:39 a.m. local time 共01:32:39 UTC兲, at about 6 km southwest of the capital town. It was the most destructive event after the M L 6.9 Irpinia earthquake occurred in the Campania region 共southern Italy兲 on November 23, 1980, which killed 2,914 people, injured more than 10,000 people, and left 300,000 homeless. The Italian Department of Civil Protection 共DPC兲 responded after only 3 min to the catastrophic event and the first team was 1 Professor, Dept. of Structural Engineering, Univ. of Naples Federico II, via Claudio 21, 80125 Naples, Italy. E-mail: [email protected] 2 Ph.D. Student, Dept. of Structural Engineering, Univ. of Naples Federico II, via Claudio 21, 80125 Naples, Italy 共corresponding author兲. E-mail: [email protected] Note. This manuscript was submitted on July 31, 2009; approved on January 21, 2010; published online on January 28, 2010. Discussion period open until May 1, 2011; separate discussions must be submitted for individual papers. This paper is part of the Journal of Performance of Constructed Facilities, Vol. 24, No. 6, December 1, 2010. ©ASCE, ISSN 0887-3828/2010/6-536–555/$25.00.

ready to departure for L’Aquila in about 43 min, reaching the affected area at 6:30 a.m. local time 共that is, just about 3 h after the earthquake兲. The volunteers’ mobilization started at about 9:30 a.m. Several groups of rescue workers, technicians, inspectors, and scientists coordinated by DPC went to L’Aquila immediately after the event, in order: 共1兲 to carry out emergency operations; 共2兲 to start preliminary visual inspections on the built environment; and 共3兲 to install mobile accelerometric stations in the affected area. The emergency management activities were complicated by the collapse of the Prefecture building 共Fig. 1兲 so a temporary central control center, named DICOMAC, was established inside a large gymnasium of an army school in Coppito 共outside L’Aquila兲 and seven local management centers were placed in the affected area. About 3,000 volunteers, 2,250 firefighters, 2,000 policemen, 1,500 army personnel, and more than 1,000 technical employees of the Abruzzo Regional office were directly involved in the emergency operations. Twenty-one days after the mainshock, 31 tent cities and field hospitals were installed to provide for care and housing of the homeless, reaching the amount of about 5,700 tents with a capability of 36,000 people. Other inhabitants were temporarily housed in hotels on the Adriatic coast and private residences. The L’Aquila earthquake sequence caused 305 fatalities, about 1,500 injured, and 29,000 homeless. More than 60,000 buildings were seriously damaged or destroyed and about 70,000 inhabitants were evacuated. A total of 81 municipalities were affected by the sequence and 49 of them were classified in the range VI-X of the Mercalli-Cancani-Sieberg 共MCS兲 macroseismic scale, which is 1.2 times the modified Mercalli intensity, more widely used in the United States. Among the villages around L’Aquila, Onna and Paganica 共Italy兲 were reported to be the most damaged ones because they are located on soft alluvial deposits of 300 m, in the Aterno River Valley. A total of 203 and 37 fatalities were reported for L’Aquila and Onna, Italy respectively. Shaking intensities of up to 10 MCS were estimated for Onna, Paganica, and Casteln-

536 / JOURNAL OF PERFORMANCE OF CONSTRUCTED FACILITIES © ASCE / NOVEMBER/DECEMBER 2010

Downloaded 15 Nov 2010 to 143.225.98.192. Redistribution subject to ASCE license or copyright. Visithttp://www.ascelibrary.org

Fig. 1. Global collapse of the L’Aquila’s Prefecture

uovo, Italy; while no more than 6 MCS was evaluated northwest of L’Aquila in neighboring villages placed on bedrock materials, such as Tussio and Monticchio. This damage pattern seems to be consistent also with the extension direction of the fault plane rupture of the mainshock, showing a combination of near-source directivity effects and local 共both stratigraphic and topographic兲 amplifications of the seismic ground motion 关Geo-engineering Extreme Events Reconnaissance 共GEER兲 2009兴.

Main Features of the L’Aquila Earthquake Sequence The L’Aquila earthquake sequence occurred as a result of normal faulting 共which belongs to the dip-slip fault mechanism兲 with southwest dip on a northwest-southeast oriented seismogenetic structure in the central Apennines, a mountain belt formed as a consequence of subduction that runs along the whole Italian peninsula 共Bagnaia et al. 1992兲. The Italian National Institute of Geophysics and Volcanology 共INGV兲 estimated the epicenter location of the mainshock of April 6, 2009 as 42.334° N and 13.334° E, approximately at the following distances from the closest principal towns: 75 km west of Pescara, Italy; 85 km northeast of Rome; 115 km southeast of Perugia, Italy; and 145 km south of Ancona, Italy. Seismologists estimated the Richter magnitude of the event as M L 5.8, with a focal depth of 8.8 km, while strong-motion data inversion resulted in a moment magnitude M W 6.3. The fault rupture plane of the mainshock was found to be rectangular of about 17⫻ 14 km2 and placed at depths between 0.6 and 11.8 km from the surface, with angle of 142° with respect to the north-south direction 共strike兲, slope of 50° 共dip兲, and slip direction vertically oriented with respect to the projection of the rupture plane on the surface 共rake兲. Since several subparallel faults lie in the affected area, the mainshock of April 6, 2009 was

Fig. 2. Epicenter locations of the four main events

followed by tens of low-intensity aftershocks at focal depths ranging from 10 to 12 km, and by three stronger aftershocks with moment magnitude greater than 5 共Table 1兲. Fig. 2 shows the spatial distribution of this earthquake sequence. A seismic event occurred on April 7, 2009 southeast of the mainshock and was characterized by a moment magnitude equal to M W 5.5, while two other events with M W 5.4 and 5.2 happened on April 9, 2009 in the Laga Mountains area, located north of the mainshock 共Ameri et al. 2009兲. In other terms, the earthquake sequence moved first southeast and then northwest of the mainshock, as clearly shown by the measurements of the Italian National Accelerometric Network 共RAN兲 managed by the DPC. This resulted in a progressive release of energy throughout the fault rupture plane. A graphic representation of the rupture plane projection at the surface, as well as the epicenter locations of both mainshock and aftershocks with moment magnitude greater than 5, are given in Fig. 3. The mainshock of April 6, 2009, which seems to occur on the Paganica fault, was recorded by 57 stations of RAN and 142 broadband stations; even 14 of the latter are placed in the Abruzzo region. Strong-motion data were recorded at epicentral distances 共Repi兲 between 4 and 297 km; 23 records were obtained at distances lower than 50 km and four of them were available at distances lower than 10 km 共Table 2 and Fig. 3兲. The latter allowed to preliminarily identify and analyze directivity phenomena among the so-called near-fault effects of earthquakes. Velocity pulses in the fault-normal components of the records induced anomalous 共linear and nonlinear兲 demands on structures with fundamental

Table 1. Strongest Events of the L’Aquila Earthquake Sequence Date April April April April

6, 7, 9, 9,

2009 2009 2009 2009

Origin time 共UTC兲

Lat 共degrees兲

Long 共degrees兲

D 共km兲

ML

MW

01:32:39 17:47:37 00:52:59 19:38:16

42.33 42.27 42.48 42.50

13.33 13.46 13.34 13.35

8.8 15.1 15.4 17.2

5.8 5.3 5.1 4.9

6.3 5.5 5.4 5.2

JOURNAL OF PERFORMANCE OF CONSTRUCTED FACILITIES © ASCE / NOVEMBER/DECEMBER 2010 / 537


Fig. 3. Graphic description of the rupture plane projected at the surface

periods close to one-half of the pulse period 共Chioccarelli and Iervolino 2010兲. Corrected data of waveforms are available at the website of the Italian Network of University Laboratories of Earthquake Engineering 共ReLUIS兲共namely, http://www.reluis.it兲. Peak ground accelerations up to 0.6g were recorded on soft alluvial soils, close to Onna and Paganica, with a maximum value of 0.676g. The site of L’Aquila is quite different from a geological point of view because a stiff conglomerate layer rests on the alluvial soft soil of the valley 共De Luca et al. 2005兲. Also for these reasons, complex local amplifications resulted in a peak ground acceleration of about 0.35g just below the conglomerate layer. An accelerometric station, named AQM 共not included in Table 2兲, recorded 1g or more in the vertical and one horizontal directions, but went off scale above 1g in the other horizontal direction. Its recordings are still under study, so they have not yet been released 关Earthquake Engineering Research Institute 共EERI兲 2009兴. In some cases, the recorded accelerations are larger than those expected by the Italian seismic hazard map, whose discrete values of PGA and spectral shape parameters are listed in Annex B of the Italian building code 关Italian Ministry of Infrastructures and Transportation 共MIT兲 2008兴. In fact, the estimated PGA at L’Aquila is equal to 0.261g and 0.334g at life safety 共return period for ordinary facilities: TR = 475 years兲 and collapse prevention 共TR = 975 years兲 limit states, respectively. In addition, the maximum horizontal accelerations were simultaneous; both EC8 关European Committee for Standardization 共CEN兲 2004兴 and the Italian building code allow to apply the combination of the effects due to an horizontal component of the seismic action with the 30% of the ones due to an orthogonal horizontal component. Records of the L’Aquila earthquake show that this combination rule may lead to underestimate the maximum effects on structures, so the square root of sum of squares criterion is expected to give more conservative results also for linear static analysis, as well as for modal dynamic analysis with response spectrum. The

combination with the effects due to the vertical component should be also revised for design/assessment of structures in the near field. Data processing showed a rather short duration of the seismic ground motion, because 95% of the input energy was released in 10 s or less. Permanent displacements of up to 15 cm were also noted in the records. The RAN’s records were analyzed by several researchers in the frame of ReLUIS network, to get both elastic and inelastic spectra, as well as some peak parameters of engineering interest 共Chioccarelli et al. 2009; Masi and Chiauzzi 2009兲. Polar spectra were developed by Petti and Marino 共2009兲, in order to account for the angle of incidence of the seismic action on building structures. By applying linear baseline correction and Butterworth bandpass filter 共f 1 = 0.1 Hz, f 2 = 25 Hz, order 4兲 to the original records, Cosenza et al. 共2009兲 obtained both elastic acceleration and displacement spectra for several values of equivalent viscous damping. Such researchers pointed out that elastic spectra were characterized by very high pseudo-accelerations in the low-period range, so “rigid” structures such as masonry buildings were subjected to high horizontal forces and, thus, to large strength demands. For epicentral distances lower than 10 km, the maximum value of horizontal spectral acceleration was estimated as 1.72g in the east-west direction for a period of 0.14 s, and as 1.44g in the north-south direction for a period of 0.44 s. Average maximum values of the spectral displacement equal to 10 cm for the horizontal components and 5 cm for the vertical one were computed at the stations AQA, AQG, and AQV listed in Table 2. Much higher values of 24.20 and 10 cm were found for the horizontal and vertical components, respectively, by processing data from the station AQK. It is interesting to note that De Luca et al. 共2005兲 predicted a significant amplification of earthquake ground motion in the lowfrequency range 0.5–0.6 Hz 共which corresponds to a period range of 1.67–2 s兲 related to the presence of a sedimentary basin. On the other hand, the period of vibration of both masonry buildings and RC infilled-frame structures, which are the typical structural types in the city of L’Aquila, did not fall in the frequency range considered by De Luca et al. Hence, the writers believe that severe damage to masonry buildings should be chiefly associated with high acceleration demands at closer distances to the seismic source, rather than with local amplification effects due to soil conditions. As shown by Chioccarelli and Iervolino 共2010兲, the high-frequency content in the seismic input was drastically attenuated with epicentral distance. This resulted in high acceleration demands on masonry buildings located in the near fault 共i.e., at epicentral distances not larger than 10 km兲 and lower demands on similar structures in the far field. Therefore, most part of damage was mainly caused by the occurrence of the mainshock exactly underneath the town of L’Aquila 共Fig. 3兲 and, thus, by the

Table 2. Mainshock Records at Epicentral Distances Lower Than 10 km Record code

Station code

Lat 共degrees兲

Long 共degrees兲

Soil type

Repi 共km兲

PGA 共g兲

GX066 FA030 CU104 AM043

AQV AQG AQA AQK

42.377 42.373 42.376 42.345

13.344 13.337 13.339 13.401

B B B C

4.8 4.3 5.8 5.6

0.676 0.515 0.487 0.373



resulting anomalous accelerations on existing buildings falling in the low-period range. Nevertheless, this evidence was not in contrast with past seismological studies on local amplification effects at low frequencies, because the period range under consideration was considerably different from that associated with the typical buildings located in the affected region. Peak ground velocity and peak ground displacement were also computed and their maximum values were found to be 40.48 cm/s and 11.87 cm, respectively. A preliminary data processing showed that spectral displacement can be seen as an engineering demand parameter more stable than pseudoacceleration. The synthetic aperture radar interferometry, a special technique which merges radar data acquired before and after the earthquake, was applied by the Institute for Electromagnetic Survey of the Environment and INGV based on the radar data from the satellites named Envisat and COSMO-SkyMed of the European Space Agency 共ESA兲 and Italian Space Agency, respectively. These special data allowed to measure surface deformations induced by the mainshock and numerous aftershocks that followed it. Three-dimensional 共3D兲 ground displacement maps from five global positioning system location sites around the affected area confirmed the SAR interferometry measurements. The ground movement was found to be as much as 25 cm between L’Aquila 共9 MCS兲 and Fossa 共7 MCS兲. ESA data sets were made available to everyone 共see http://earth.esa.int/ew/earthquakes/Italy_ April09/兲. This amount of information makes the L’Aquila earthquake sequence to be one of the best recorded normal fault earthquakes.

Field Inspections of Buildings The assessment of the usability of buildings started just two days after the mainshock and the seriousness of the situation led to an estimate of about 60,000 affected buildings. Therefore, also because of the continuous occurrence of aftershocks, it was decided to assess first the less damaged areas and second the historical center of L’Aquila. The latter was entirely evacuated after the mainshock because a lot of buildings were likely to collapse even under very low-intensity aftershocks. The assessment of ordinary buildings was based on visual inspections and relied on simple forms produced by DPC. About 1,500 inspectors were deployed to evaluate about 1,000 buildings daily. Around 65% of the buildings were green tagged 共i.e., ready for occupancy兲, while 27% were red tagged 共i.e., unsafe for occupancy兲. The operations aimed to assess the structural safety of monumental buildings 共churches, castles, and palaces兲 and to preserve them by other tremors were coordinated by the Italian Ministry of Cultural Heritage. Dedicated forms were used for assessing the usability of historical buildings and special units of firefighters operated together with experts to make postearthquake safety operations. About 53% of the monumental buildings were declared as totally unusable, whereas about 23% were found to be ready for immediate occupancy. Also, several international teams of experts visited the affected areas. In the following sections, the findings of the postearthquake reconnaissance field mission carried out by the writers under mandatory of the Ministry of Cultural Heritage are presented, with special reference to both ordinary masonry and cultural heritage buildings. Since RC building structures suffered damages under the L’Aquila earthquake sequence as well, some discussions are made also on the more interesting features of seismic

Fig. 4. Collapse of infill walls in a recently constructed RC building

performance of these more recently constructed facilities.

Damages to RC Buildings The collapse of existing RC buildings in L’Aquila was generally due to story mechanisms caused by the 共brittle兲 shear failure or 共ductile兲 flexural yielding at end cross sections of columns. Extensive cracking and ejection of infill walls were typically observed during the L’Aquila earthquake sequence 共Figs. 4 and 5兲. Such nonstructural elements contributed to resist against the seismic action up to damage occurrence by stiffening the framed structural system, while a story mechanism happened after the collapse of all the infill walls placed at the same floor level 共Fig. 6兲. Among the most important RC buildings in L’Aquila, the San Salvatore Hospital, which was built between 1974 and 1987, and was opened to public in 2000, performed well enough during the earthquake sequence. Just three exterior marginal columns were damaged; Fig. 7 shows the brittle failure of a column. This type of mechanism is likely due to large vertical components of the seismic ground motion, along with the lack of effective stirrups at end cross sections of columns. In this case, the longitudinal steel bars were not confined and, hence, reached the Euler instability. Ma-

Fig. 5. Ejection of infill walls in a RC building



Fig. 8. Main façade of the Duca degli Abruzzi Hotel before the earthquake 共source: Google Street View兲 Fig. 6. Soft story mechanism of an infilled RC frame

jority of damage was to partition walls and to suspended ceiling systems. It is interesting to note that the excitation of the heavy letters constituting the title of the Hospital, supported by exterior walls of the main façade at the top story, caused the collapse of the walls. Therefore, this shows that care should be paid to design nonstructural components based on their interaction with the supporting structures. The Duca degli Abruzzi Hotel, a cast-in-place construction built in the 1970s of the last century, partially collapsed under the earthquake due to the presence of a number of significant structural deficiencies. The structural system was a 3D masonryinfilled frame having pilotis at the ground floor, vertical elements

Fig. 7. Brittle failure of a RC column in the San Salvatore Hospital

with different heights, and poor construction detailing. Therefore, the high irregularity in elevation resulted in a soft story mechanism of the building. Fortunately, no casualties were caused by this accident. The story mechanism did not propagate thanks to a structural joint which split the whole structure in two independent parts. Fig. 8 shows the main façade of the Hotel prior to the earthquake, while Figs. 9共a and b兲 explain well its conditions after the mainshock. The benefit of the structural joint on seismic performance of the building can be clearly understood by observing Fig. 9共b兲 on the left side. The global collapse of the student house dorm of the University of L’Aquila caused seven deaths 共Fig. 10兲. The Prosecutor of

Fig. 9. Duca degli Abruzzi Hotel after the earthquake



Fig. 10. Global collapse of the student house dormitory of the University of L’Aquila

L’Aquila opened a criminal procedure immediately after the event and a team of technical consultants were appointed to investigate the causes of the disaster. However, this building was characterized by high irregularity in plan. Another case of interest in the field of Forensic Engineering is a recently constructed RC building totally collapsed under the earthquake causing 23 fatalities. Fig. 11 shows the area occupied by the building. A particularly interesting case of global collapse is shown in Fig. 12. The whole second level of a four-story RC building “disappeared” under the earthquake. Actually, both the third and fourth floors overlapped perfectly on the first one. Figs. 12共a and b兲 show this particular mechanism in detail, which was probably due to brittle collapse of columns, rather than to the presence of a soft story, because no plastic hinges developed at end cross sections. Figs. 12共c and d兲 show the complete ejection of columns from the second story. This was probably due to: 共1兲 large vertical accelerations of the seismic ground motion and 共2兲 insufficient presence of steel bars into beam-column joints. Also this collapse caused a lot of fatalities.

Damages to Ordinary URM Buildings The older masonry buildings built in the Abruzzo region are typically composed by two- or three-story walls made up of stone

Fig. 11. Global collapse of a RC building

masonry with mortar joints. Instead, more recently constructed houses are made up of rubble stone, clay bricks, or even concrete blocks. In general, it can be stated that just the buildings with poorquality masonry and significant irregularities suffered large damages or fell down. A lot of masonry structures in L’Aquila performed very well under the earthquake, thanks to the better quality of materials 共e.g., bigger size of masonry units, squared stones or clay bricks, and good-quality mortar joints兲, regular distribution of mass and stiffness, and good construction detailing. More than 50% of the masonry buildings were severely destroyed in some little villages such as Onna, Paganica, and Castelnuovo, where multileaf 共called a sacco兲 masonry walls, with an inner core of smaller rubble masonry and mixture of lime and mud mortar, had been typically used without bond stones connecting together the exterior wythes. Due to lower site amplification effects, other towns such as Barisciano, Santo Stefano, and Monticchio, even if they are close to the more damaged ones, experienced smaller failures 关Geo-engineering Extreme Events Reconnaissance 共GEER兲 2009兴. Along with the analysis of damages induced by the L’Aquila earthquake sequence, let us classify theoretical failure mechanisms of masonry buildings. They are typically modeled as assemblage of walls oriented along any direction. The most used and effective methods of structural analysis of masonry buildings under seismic actions are based on the macroelements geometrical discretization of each wall 共Augenti 2004兲. By means of this approach, a masonry wall with openings is divided into three types of two-dimensional elements: 共1兲 pier panels, which are the vertical structural elements between openings; 共2兲 spandrel panels, which are the horizontal structural elements between openings; and 共3兲 cross panels, which link pier and spandrel panels together 共Fig. 13兲. Each macroelement has height H, length B, and thickness s. Structural failures can involve individual panels, for both regular and irregular walls, through the following mechanisms: 1. In-plane 共longitudinal兲 mechanisms, which involve individual panels of each wall and consist of several types of crack patterns a. Crushing failure due to the exceeding of the compressive strength of masonry induced by the vertical component of seismic motion 共Fig. 14兲; vertical cracks develop first throughout the vertical transversal cross section s ⫻ H of the masonry panel 共splitting兲 and finally along the longitudinal cross section B ⫻ H. Fig. 15 shows the total loss of internal integrity of a masonry panel due to a crushing phenomenon which occurred after the splitting failure of the wall. b. Tensile failure due to the exceeding of the tensile strength of masonry induced by the horizontal components of seismic motion; horizontal cracks develop along the horizontal cross section B ⫻ s of the masonry panel; they are not simply visible when the seismic input vanish, because of their closure due to gravity loads. c. Flexural failure, in which compression vertical cracks and tensile horizontal cracks are localized at the toes of the panel due to yielding of masonry; in such cases the macroelement can be fully or partially resistant. Also this type of mechanism is complex to be observed because: 共1兲 tensile cracks close with seismic action reversal and 共2兲 as the seismic ground motion increases, the masonry panel experiences also shear cracks which are more evident than the flexural ones. Figs. 16共a and b兲 show the



Fig. 12. Story crushing in a RC building

flexural cracks of a pier panel and spandrel panel, respectively. Diagonal cracks developed as well in the pier. d. Diagonal shear cracking, in which tensile cracks develop along a diagonal of the longitudinal plane B ⫻ H, from the centroid to the toes. As a result, the masonry panel is divided in two parts that move away from each other. This kind of failure mechanism is shown in Figs. 17共a and b兲 for pier and spandrel panels, respectively. e. Sliding failure, where stepwise or horizontal cracks develop along the longitudinal plane of the masonry panel, which is divided in two parts that slide along their fracture surface. Crack patterns related to these failure mechanisms are quite different for regular and irregular masonry walls. If the contours of

Fig. 13. Geometrical modeling of a masonry wall

the openings are both horizontally and vertically aligned, the masonry wall is called regular and pier panels are simple to be modeled. When the openings are not aligned nor have the same dimensions in horizontal or vertical direction, the wall is called irregular 共Augenti 2006兲. Fig. 18 shows the crack pattern of an irregular wall of a masonry building in L’Aquila. In this case, care should be paid to geometrical modeling of pier panels, because their effective height may significantly change with the orientation of the seismic action. This occurs also for peripheral pier panels because the overlying cross panels are not entirely confined and, thus, may suffer diagonal cracks with average slope angle of 30°–45°. The aforementioned mechanisms are referred to all inner and peripheral panels of masonry walls not well connected to transversal ones. 2. Out-of-plane 共transversal兲 mechanisms, also called first mode mechanisms, which may involve individual panels or panel systems of a given masonry wall; the latter is typically divided in different rocking elements in the vertical transversal plane s ⫻ H. Along with global collapses of masonry structures, a large number of buildings in the historical center of L’Aquila suffered out-of-plane mechanisms due to lack of steel ties or ring beams, or to poor connections between longitudinal and transversal walls. Figs. 19共a and b兲 show outof-plane mechanisms of individual pier panels at the upper story of a building, along with large horizontal displacements of the roof slab. Fig. 20 shows first mode mechanisms involving more panels at several floor levels. In both cases, the



Fig. 14. Crushing of masonry

3.

4.

5.

floor slabs did not constrain the walls along their transversal direction. Combined mechanisms, in which in-plane actions are followed by out-of-plane components of seismic motion that induce rocking mechanisms of panels previously damaged in their own plane 共Fig. 21兲. In other terms, the masonry panel is first damaged by in-plane seismic excitation and second destroyed by out-of-plane components. The probability of occurrence of such collapse phenomena is dramatically increased by aftershocks and successive events of an earthquake sequence. Corner mechanisms, which involve end panels placed at the corners of the masonry building when perimeter walls are well connected together. In such cases, biaxial bending induces the ejection of the corner of the masonry building 共Fig. 22兲. This failure mode does not frequently occur, given that longitudinal and transversal masonry walls are generally built up in separate construction stages, rather than together at the same time. No methods for safety verifications against corner mechanisms are provided by the current seismic codes, so future research is needed to build up theoretical models able to describe this class of failures as well. Global mechanisms, which progressively involve different parts of the building or even the whole structure. A corner building unit which suffered global collapse under the earthquake is shown in Fig. 23. Fig. 24 shows the total destruction of adjacent building units along a main street in L’Aquila, while the collapse of the upper floor level of a three-story building located in Santa Maria a Coppito Place can be seen in Fig. 25.

Fig. 15. Crushing failure after splitting of masonry

After the 1703 earthquake 共M W ⬇ 6.7兲, masonry walls were strengthened by inserting timber ties aimed to connect them and, thus, to obtain a better global behavior under seismic loading. Fig. 26 shows the rupture of these ties induced by the out-of-plane mechanism of a wall and the resulting collapse of the roof. In some cases, RC ring beams realized into existing masonry buildings were found to be harmful and did not prevent local collapses in masonry walls. The theoretical classification of failure modes in unreinforced masonry 共URM兲 building structures given above is consistent with those provided by other researchers, such as Magenes and Calvi 共1997兲 and Calderini et al. 共2009兲. Nevertheless, combined and corner mechanisms had never been observed before the 2009 L’Aquila earthquake sequence, so the writers believe this extended classification will be helpful for both future research and practice in earthquake engineering.

Damages to Cultural Heritage Buildings Cultural heritage suffered large damages due to construction characteristics of the building or to wrong seismic strengthening/ upgrading interventions. The most significant case studies of the historical center of L’Aquila are discussed in the following. Above all, serious damages to three monumental buildings and several churches are shown in detail. Spanish Fort This fortress is one of the most impressive Renaissance castles in Central and Southern Italy. It is located on the highest part of



Fig. 17. Diagonal shear cracking of 共a兲 pier panel; 共b兲 spandrel panel

due to its considerable mass and, thus, to large inertia forces at the top level. Fig. 30 shows the upper floor level from inside the courtyard: we notice the partial collapse of the perimeter wall due to an out-of-plane mechanism. The whole façade experienced rocking rotation toward the courtyard. Fig. 31共a兲 shows the arcade at the ground floor: a stepwise crack on the left column can be seen in Fig. 31共b兲. Fig. 32 points out the colonnade at the first floor of the castle, whose covering barrel vault suffered longitudinal cracks along its length owing to outward rotation of the façade. Fig. 33 shows paving cracking close to columns, whose maximum amplitude was about 25 mm. Fig. 16. Flexural mechanisms of 共a兲 pier panel; 共b兲 spandrel panel

L’Aquila and was built up between 1534 and 1541 under the appointment of Don Pedro de Toledo, the Spanish viceroy of the Kingdom of Naples. Fig. 27 shows the ground floor plan of the castle and the surrounding fosse. This monumental building is currently home of the National Museum of Abruzzo, so large damages to artworks 共e.g., paintings, sculptures, and ancient furniture兲 were caused by structural failures and collapses during the earthquake sequence. The seismic events resulted in many losses including the collapse of the covering at the upper story, which was realized in the 19th century and recently restored. Figs. 28共a and b兲 show the entrance of the castle before and after the mainshock, respectively. Fig. 29 allows to see the whole main façade with the access bridge. A newly constructed RC roofing structure collapsed

Fig. 18. Crack pattern of an irregular masonry wall



Fig. 20. Out-of-plane collapse mechanism of a masonry wall

Fig. 19. Out-of-plane mechanism of an individual story

Fig. 34共a兲 illustrates the crushing of a bearing masonry wall at the first floor, while a partially destroyed wall placed at the second floor can be seen in Fig. 34共b兲. Damaged sculptures and paintings are visible in both images. Part of the second floor collapsed under the mainshock; all the remaining bearing walls and vaults located at the same story and at the first floor were seriously damaged. While little damages to the ground floor were observed, no failures were detected at the underground floor. Severe damages suffered by the building first required urgent shoring systems under the seriously damaged roofing structures, as well as to openings of masonry walls. The latter were closed by means of clay brick masonry. In order to stop outward rotation of the façade and to reduce the probability of occurrence of out-of-plane mechanisms under further aftershocks, the main exterior façade and the one facing the courtyard were connected together by means of metallic tendons. Let us notice that older stone URM structures experienced just large failures, while global or local collapses occurred only in building parts with newly constructed RC slabs, ring beams, and pitched roofs.

very interesting from an engineering point of view. Fig. 35共a兲 shows the whole damaged corner of the building, while Fig. 35共b兲 shows a detail of the detachment between masonry walls. Urgent safety interventions were made by placing a number of tendons along the perimeter ledge of the building and connecting them through metallic plates. Head Office of the University of L’Aquila Carli Palace is currently the Chancellor office of the University of L’Aquila. The earthquake sequence induced huge damages to such a building. However, a diagonal crack on the corner masonry

Library of the L’Aquila Province This monumental building was realized at the end of the 19th century and was designed by Alessandro Mancini in order to take the library of L’Aquila. Among several failures experienced by such a structure, detachment at the corner of perimeter façades is

Fig. 21. Combined failure mechanism of a masonry wall



Fig. 24. Total destruction of inner building units

back to the 19th century. Transept and other large parts of the cathedral fell down under the L’Aquila mainshock of April 6, 2009. Fig. 37 shows the main façade of the church, which experi-

Fig. 22. Corner mechanism due to biaxial bending

pier at the upper story developed and was followed by another one on the opposite building 共Fig. 36兲. This damage pattern was a clear evidence of the seismic action orientation. Cathedral of Santi Massimo e Giorgio This is the cathedral dome of L’Aquila built in 1257 but crumbled down during the 1703 earthquake. The most recent façade dates Fig. 25. Global collapse of a masonry building

Fig. 23. Global collapse of a corner building unit

Fig. 26. Out-of-plane mechanism of a masonry wall with timber ties



Fig. 27. Ground floor plan of the Spanish Fort in L’Aquila

enced serious damages to roofing structures 共triumph arch, transept, and dome兲. Restoration works were ongoing inside the cathedral at the moment of the seismic event. Church of Santa Maria del Suffragio This Baroque church, also known as “Le Anime Sante,” was built after the 1703 earthquake in the ancient market place of L’Aquila, where the Cathedral of Santi Massimo e Giorgio is located too. The Church of Santa Maria del Suffragio became one of the L’Aquila earthquake symbols, because images of the collapsed tambour and dome spread all over the world. Figs. 38共a and b兲 show the church before and after the earthquake, respectively. Also in this case, as a result of many highintensity aftershocks, damage to structures increased a lot leading even to further unexpected collapse phenomena. In Fig. 39, we can observe the partial collapse of the tambour, which was strengthened with circumferential timber ties in the past. To avoid the total loss of the tambour, which was been seriously damaged by the mainshock, some safety interventions were carried out through aerial platforms given the highly dangerous situation. Church of Santa Maria di Paganica Such a Romanesque-Gothic church was built in the 14th century and perhaps is the most damaged religious building in L’Aquila. Figs. 40共a and b兲 show the whole church from the adjacent place and from an aerial platform, respectively. As one can see, the earthquake induced the collapse of the triumphal arch, dome, pitched roof of the central nave, left aisle, and bell tower. All the other structures experienced severe damages as well. Fig. 41 shows the interior of the collapsed church looking toward the altar and apse. Both the collapsed dome and triumphal arch can be seen in Fig. 42; the destroyed roof, along with the damaged chapels of the right aisle, are shown in Fig. 43. The older timber covering structure was substituted with a novel one composed by prestressed concrete 共PC兲 joists and lath bricks resting on a perimeter RC ring beam. Just the older chestnut ties were left on site

Fig. 28. Spanish Fort: 共a兲 before; 共b兲 after the earthquake



Fig. 29. Main façade of the Spanish Fort

and thin metallic bars were been added to eliminate thrust in the roofing structure. The novel RC structure, much heavier and stiffer than the older one, triggered the collapse of the masonry walls of the nave. Observations of the dome’s debris prove that PC roofing structures were realized over it. Also in this case, high-intensity aftershocks resulted in cumulated damage rapidly increasing over time, so the roof shown in Fig. 43 collapsed some days after the mainshock. Urgent safety and protective interventions required also the controlled demolition of some structural components lying in precarious equilibrium conditions. Since the main façade of the church had experienced outward rotation with detachment from longitudinal walls of aisles, temporary connecting elements were installed to avoid the complete out-of-plane collapse. Basilica of San Bernardino da Siena This church was built up in 1472, has a fine Renaissance façade 共Fig. 44兲, and contained the monumental tomb of the saint decorated with wonderful sculptures. Such a facility suffered severe damages during the L’Aquila earthquake, especially to the bell tower and tambour of the octagonal dome 共Fig. 45兲. Church of San Marco This church, dating back to the 15th century, was seriously damaged with particular reference to the covering. Since inspections

Fig. 31. Outward rocking of the colonnade

Fig. 30. Out-of-plane mechanism at the top floor

inside the church were not possible because of the serious hazard conditions, the writers entered the breach formed in the upper part of the wall of the nave by means of a firefighters’ ladder 共Fig. 46兲.



Fig. 32. Horizontal cracks in vaults due to rocking of the colonnade

Fig. 33. Cracking of the floor 共amplitude around 25 mm兲

Damages to interior of the church were observed through this opening. In particular, Fig. 47共a兲 shows damages to masonry arches, while Fig. 47共b兲 highlights the novel covering formed by PC joists and lath bricks, along with a RC tympanum resting on the dome. Such a heavy and rigid element increased the vulnerability of the structure. Finally, a lateral wall of the church experienced rocking mechanism with about 70-cm large outward displacement.

damaged, owing to multiple out-of-plane mechanisms of both the main façade and lateral walls, as shown in Fig. 51. The Church of San Pietro a Coppito, built in the 13th century, suffered large damages to the main façade, vaults of the nave and aisles, apse’s covering, and octagonal-plan bell tower 共Fig. 52兲. The Church of Santa Maria della Misericordia, built between 1528 and 1531, was seriously damaged as well, especially to the right lateral body, because of significant irregularities in elevation 共Fig. 53兲. A lot of less important churches suffered many damages to structural and nonstructural components due to local collapse mechanisms. Fig. 54 shows an interesting example of out-ofplane failure experienced by the tympanum of a church located in the historical center of L’Aquila.

Church of San Silvestro This church was built up by inhabitants of the Castle of Collebrincioni in the first one-half of the 14th century. The bell tower was reconstructed after the earthquakes occurred in the 15th century and the completely restored interior structure is divided into three naves ending with polygonal apses, but without a transept. Fig. 48 shows the main façade of the church along with the bell tower. Since the latter had different modes of vibration with respect to the remaining part of the structure, a large concentration of stress occurred along the tower-nave interface of the church during the mainshock, resulting in large cracks on both external and internal masonry walls 共Fig. 49兲. Serious cracks were detected on other walls too: Fig. 50 shows vertical cracks along the corner between the main façade and the wall placed above the arches of the nave. Significant Damages to Further Churches Among other religious facilities inspected after the mainshock, the Church of San Francesco di Paola was found to be severely

Conclusions and Lessons from Failures The 2009 L’Aquila earthquake sequence was a destructive chain of seismic events that induced high damage over time to a lot of both ordinary and monumental buildings, as well as several RC residential buildings, resulting in many casualties and large economic losses. The information gathered during the field mission and the great amount of seismological and engineering data about the earthquake sequence allow to state that the high level of damage and human losses resulted from a combination of several factors: 1. The so-called near-fault effects of seismic events, along with local amplifications due to complex stratigraphic and topographic conditions, which induced anomalous levels of ground motion at the surface and, thus, very high demands



Fig. 35. Damaged corner in the library of the L’Aquila province Fig. 34. Crushing failure of inner masonry walls

2.

on structures. Source directivity was detected in the records of the L’Aquila mainshock and velocity pulses resulted in larger demands on some building structures. Damages to both URM structures and RC infilled-frame buildings were mainly due to high accelerations at small distances from the seismic source. Indeed, the role of the high-frequency content in the seismic input was extremely important in such conditions. In addition, equal PGA values were recorded along both north-south and east-west directions, so unusual failure modes were observed in masonry structures. The high vulnerability of both older and some more recently RC constructed facilities due to the presence of poor-quality materials, inadequate construction detailing, and wrong repair or strengthening interventions, particularly in the case of cultural heritage buildings.

3.

The collapse of nonstructural components, such as infill walls, internal partitions, and suspended ceilings. On the other hand, both URM and RC buildings with goodquality construction characteristics and structural regularity performed well enough during the earthquake sequence and were found to be ready for occupancy a few days after the mainshock. This means that, in general, both the current seismic design criteria and past construction rules seem to be effective in order to have a good global response under earthquake loading. The following lessons may be learned from the observed damages and collapse phenomena: 1. The very high PGA values recorded close to the seismic source show the importance of near-fault effects also in the seismogenetic structures of Italian Apennines, similarly to what was observed, for instance, in California and Japan.



Fig. 36. Head office of the University of L’Aquila

2.

3.

4.

Given that some building structures are particularly prone to suffer directivity effects, this topic should be analyzed in future studies in the fields of engineering seismology and earthquake engineering. The accelerations recorded close to L’Aquila show that the combination rules of seismic effects suggested by the Italian building code and allowed by EC8 should be revised, in order to get a better estimation of demands on structures. In addition, the role of the vertical component should be carefully taken into account for structures located in the near field. Although failure modes and collapse mechanisms can be generally explained by the current methods of structural analysis, future research is needed to solve some specific features for predicting the complex behavior of structures under directivity earthquakes. Furthermore, some types of failure modes, such as combined and corner mechanisms in masonry buildings, should be analyzed in detail by means of innovative analytical models. Cumulative damage to masonry constructions is another aspect to be studied in future, in order to better predict response under earthquake sequence. Both strength reduction factors and displacement amplification factors of elastic spectra should be specifically defined for masonry buildings

Fig. 38. Church of Santa Maria del Suffragio: 共a兲 before; 共b兲 after the earthquake

Fig. 37. Main façade of the Cathedral dome of L’Aquila JOURNAL OF PERFORMANCE OF CONSTRUCTED FACILITIES © ASCE / NOVEMBER/DECEMBER 2010 / 551


Fig. 41. Church of Santa Maria di Paganica: Internal view

Fig. 39. Collapsed dome of the Church of Santa Maria del Suffragio

Fig. 42. Collapsed dome of the Church of Santa Maria di Paganica

Fig. 40. Church of Santa Maria di Paganica: Global views

Fig. 43. Collapsed roof and chapels in the Church of Santa Maria di Paganica



Fig. 44. Main façade of the Basilica of San Bernardino da Siena

Fig. 47. Damages to arches and vaults in the Church of San Marco Fig. 45. Damages to the octagonal tambour and bell tower

Fig. 46. External view of the Church of San Marco

Fig. 48. Main façade of the Church of San Silvestro



Fig. 51. Damages to the Church of San Francesco di Paola

Fig. 49. Cracks along the height of the bell tower of the Church of San Silvestro

Fig. 52. Damages to the Church of San Pietro a Coppito

Fig. 50. Vertical cracks along the wall intersection

Fig. 53. Damages to the Church of Santa Maria della Misericordia



Fig. 54. Out-of-plane collapse of a masonry tympanum

5.

through damage models able to describe this kind of evolutionary behavior under cyclic loading. Care should be paid to insertions of RC rigid elements in existing URM facilities, because they may radically change the global seismic response. Moreover, added mass may induce not only higher strength demands on the building, but also significant irregularities in plan or elevation, leading to unexpected distributions of the seismic action and, thus, to dangerous localizations of internal forces within the structure.

Notation The following symbols are used in this paper: B ⫽ length of the masonry panel; D ⫽ focal depth of the earthquake; H ⫽ height of the masonry panel; Lat ⫽ latitude of the site; Long ⫽ longitude of the site; M L ⫽ local 共or Richter兲 magnitude of the earthquake; M W ⫽ moment magnitude of the earthquake; Repi ⫽ epicentral distance of the site; s ⫽ thickness of the masonry panel; and TR ⫽ return period of the earthquake.

References Ameri, G., et al. 共2009兲. Strong-motion parameters of theMw = 6.3 Abruzzo 共Central Italy兲 earthquake, Istituto Nazionale di Geofisica e Vulcanologia 共INGV兲, Milan-Pavia Section, Italy, 具http:// www.ingv.it典.

Augenti, N. 共2004兲. Seismic design of masonry buildings, UTET, Turin, Italy 共in Italian兲. Augenti, N. 共2006兲. “Seismic behaviour of irregular masonry walls.” Proc., 1st European Conf. on Earthquake Engineering and Seismology 共CD-ROM兲, Paper No. 86. Bagnaia, R., D’Epifanio, A., and Sylos Labini, S. 共1992兲. “Aquila and Subequan basins: An example of Quaternary evolution in Central Apennines, Italy.” Quaternaria Nova, 2, 187–209. Calderini, C., Cattari, S., and Lagomarsino, S. 共2009兲. “In-plane strength of unreinforced masonry piers.” Earthquake Eng. Struct. Dyn., 38, 243–267. Chioccarelli, E., De Luca, F., and Iervolino, I. 共2009兲. “Preliminary study of L’Aquila earthquake ground motion records V5.20.” 具http:// www.reluis.it典共July 31, 2009兲. Chioccarelli, E., and Iervolino, I. 共2010兲. “Near-source seismic demand and pulse-like records: A discussion for L’Aquila earthquake.” Earthquake Eng. Struct. Dyn., 39, 945-1062. Cosenza, E., Chioccarelli, E., and Iervolino, I. 共2009兲. “Preliminary study of displacements and accelerations at close distances for L’Aquila earthquake V.1.00.” 具http://www.reluis.it典共July 31, 2009兲. De Luca, G., Marcucci, S., Milana, G., and Sanò, T. 共2005兲. “Evidence of low-frequency amplification in the city of L’Aquila, Central Italy, through a multidisciplinary approach including strong- and weakmotion data, ambient noise, and numerical modeling.” Bull. Seismol. Soc. Am., 95, 1469–1481. European Committee for Standardization 共CEN兲. 共2004兲. “Design of structures for earthquake resistance—Part 1: General rules, seismic actions and rules for buildings.” Eurocode 8, Brussels, Belgium. Earthquake Engineering Research Institute 共EERI兲. 共2009兲. “Learning from Earthquakes. The Mw 6.3 Abruzzo, Italy, earthquake of April 6, 2009.” EERI Special Earthquake Rep., Oakland, Calif, 具http:// www.reluis.it典共July 31, 2009兲. Geo-engineering Extreme Events Reconnaissance 共GEER兲. 共2009兲. “Preliminary report on the seismological and geotechnical aspects of the April 6, 2009 L’Aquila earthquake in Central Italy.” 具http:// research.eerc.berkeley.edu/projects/GEER典共July 31, 2009兲. Italian Ministry of Infrastructures and Transportation 共MIT兲. 共2008兲. “Technical code for constructions.” D.M. 14.01.2008, Rome 共in Italian兲. Kerr, R. A. 共2009兲. “After the quake, in search of the science—Or even a good prediction.” Science Magazine, 324, 322. Magenes, G., and Calvi, G. M. 共1997兲. “In-plane seismic response of brick masonry walls.” Earthquake Eng. Struct. Dyn., 26, 1091–1112. Masi, A., and Chiauzzi, L. 共2009兲. “Preliminary analyses on the mainshock of the Aquilano earthquake occurred on April 06, 2009.” 具http:// www.reluis.it典共July 31, 2009兲. Petti, L., and Marino, I. 共2009兲. “Preliminary comparison between response spectra evaluated at close source for L’Aquila earthquake and elastic demand spectra according to new seismic Italian code V.2.00.” 具http://www.reluis.it典共July 31, 2009兲.



Learning from Construction Failures due to the 2009 L ... - ReLUIS

Learning from Construction Failures due to the 2009 L ... - ReLUIS

Suggest Documents

Learning from Failures - PUCPR

Learning from Failures: a Lightweight Approach to

Learning from the failures of others - IngentaConnect.com

Reinforcement Learning: Insights from Interesting Failures ... - CiteSeerX

Learning from failures in communication - UKCIP

Learning Complex Problem Solving Expertise from Failures

Learning from Failures: A Geotechnical Perspective

Learning from Project Failures - Major Projects Association

Failing to Communicate: Learning from Failures in ... - CiteSeerX

DELAYS IN IT PROJECTS DUE TO FAILURES IN THE ...

Learning lessons from the failures of socio-technical ... - CiteSeerX

Learning from Failures from a Geotechnical Prospective

Assessment of pile failures due to excessive ... - J-STAGE Journals

Eliminating Packet Losses due to Network Failures - UCL Department ...

Prediction of Mechanical Shaft Failures Due to Pulsating ... - IEEE Xplore

Weld Failures in Oil Tankers Due to Groundings ... - DSpace@MIT

High Power Transformers Failures due to Flow Electrification: Tools for ...

October 2009 Construction Inspection

UTILIZING TIME TO ENHANCE RECOVERY FROM FAILURES

Testimonies to the L Aquila earthquake (2009) and to ...

From market failures to market opportunities ... - SpringerOpen

Construction Failures and Innovative Retrofitting - MDPI

Title Goes Here - ReLUIS

cinecircolo 2009-2010 - + Pungolo Due