Performance of Buildings During the January 26, 2001 Bhuj Earthquake Rakesh K. Goel, M. EERI Department of Civil and Environmental Engineering California Polytechnic State University San Luis Obispo, CA 93407 rgoel@calpoly.edu Abstract Background A strong earthquake of magnitude 7.7 (USGS revised) struck the Kutch area in Gujarat at 8:46 AM (local time) on January 26, 2001. The most damaging earthquake to strike India in the last five decades, it has led to a large loss of life and property. Nearly 20,000 persons are reported to be dead and over 150,000 injured; the numbers are expected to rise as more information becomes available. The estimated economic loss is reported to be about US $5 billions. A week after the event, an EERI reconnaissance team was in the field for post-earthquake investigation. The building group co-coordinated by Dr. Rakesh K. Goel of Calpoly, San Luis Obispo, CA, USA, and Dr. C. V. R. Murty of Indian Institute of Technology at Kanpur, India focused on building performance in Ahmedabad and other regions of Gujarat affected by the January 26 event; other members of the group were Alok Goyal and Jaswant Arlekar. In addition, Rainer Metzger of PSM Consulting, and a team of three engineers from Degenkolb Engineers led by Jeff Soulages, contributed to the efforts of the EERI team. Several other teams, which are in the field or will be visiting at a latter date, would also be invited to supplement the efforts of the EERI team. This abstract summarizes the findings on damage to engineered buildings in Ahmedabad. A detailed report and damage to non-engineered structures will be presented in a comprehensive report latter. During the January 26, 2001 earthquake, numerous mid- to high-rise residential buildings collapsed in the city of Ahmedabad leading to several hundred causalities and significant financial loss. The city of Ahmedabad lies about 300 km (400 km by road) east of the epicenter of the January 26 event and falls in the seismic Zone III (IS: 1893-1976) of India (Figure 1). The lateral design forces for this region are about 4 to 6% of total weight of the building, depending on the foundation type and soil conditions. Given that the horizontal accelerations recorded in Ahmedabad during the earthquake event are about 10% of gravity (Figure 2), the buildings may be expected to deform slightly into the inelastic range. However, the extent of damage observed was significantly more than expected in such a moderate seismic region. Following is a brief summary of the reasons that contributed to this unexpected damage in residential construction. It is useful to note the significant length of the ground shaking; noticeable accelerations were recorded for a duration of nearly two minutes. The shaking in the city of Ahmedabad consisted Goel-1
of low levels of motion for about first 30 seconds followed by a strong shaking phase lasting about another 30 seconds, and then a gentle shaking again at the end (Figure 2). Discussions with residents of many apartment complexes indicated that many people were able to run out during the initial low-level shaking phase. The building collapses occurred a little after the onset of the strong shaking phase. Therefore, the initial gentle shaking appeared to have served as a warning signal and may be credited with saving many lives. Figure 1. Map of India showing Seismic Zones (IS: 1893-1976). Figure 2. Accelerations recorded at the basement of the passport building in Ahmedabad during the January 26, 2001 Bhuj earthquake (Source: University of Roorkee Web Site). Goel-2
Structural System The typical residential construction in Ahmedabad consists of reinforced concrete moment resisting frame system. The frame at the ground floor is open while frames at the upper floors are filled with un-reinforced brick panels. This type of lateral load resisting system leads to what is commonly known as a soft-story system (Figure 3). Most buildings also have overhanging covered balconies at higher floors (Figure 4); the overhangs were observed to be about 5 feet. The columns at the ground floor may not align with the columns at the upper floors giving rise to vertical discontinuities in the lateral load resisting system. Figure 3. Example of a soft-story apartment building (Photograph by Goel). Figure 4. Example of overhang at upper floors in residential construction (Photograph by Goel). The above-described lateral load resisting system occurs because of two factors. First, the open ground floor is needed to provide car parking; the buildings are usually built on very small land lots with little room for open parking. Second, the Floor Surface Index (FSI) used by the local municipal corporation for residential construction permits the land developers to cover more area Goel-3
at upper floor than the ground floor. The FSI only counts the area of within the column footprints at the ground floor. Therefore the developers are tempted to design the lateral load resisting system with only two to three columns in a frame on the ground floor with a beam overhangs on both sides. The upper floors may or may not continue these columns. But at least two floating columns are added, one on each end of the cantilever beam, starting from the first floor and running the entire height of the building. The columns for low-rise residential buildings, up to ground plus four stories (G+4), rest on shallow isolated footings located about 5 feet below the ground level. For taller buildings, say up to ground plus ten stories (G+10), the column foundations are still isolated footings located at a depth of 8 to 10 feet; sometimes tie beams are used. The column sizes for low-rise buildings (G+4) are about 9 inch by 18 inch with ties consisting of mild steel smooth No. 2 bars at a spacing of about 8 to 9 inches. The beams tend to be much deeper to accommodate large spans and overhangs giving rise to strong-beam, weak-column construction. The columns sizes for taller buildings (G+10) tend to be a little bigger, usually 12 inch by 24 inch with No. 3 deformed steel ties at a spacing of 8 to 9 inches. The beam size in taller buildings may be similar to the column size. The ties always end up with a 90-degree hook. The most residential buildings appear to be designed primarily for gravity load; there are some indications that the lateral loads may not have been properly considered in design of these buildings. There is insufficient confining steel to provide required ductility in the lateral load resisting system, and column reinforcement is spliced just above the beam level, with often insufficient development length. Failure of Residential Buildings in Ahmedabad Most residential buildings in Ahmedabad suffered some type of damage during the earthquake. Much of the damage was in the form of cracking of infill wall panels at the ground floor level. However, nearly one hundred 1 residential buildings collapsed during January 26, 2001 event. Since for the ground motion experienced in the city, buildings with sound design and construction should not have experienced any structural damage (although some nonstructural damage may be expected), the damage appears to be due to a combination of factors. Based on the post-earthquake field investigation, following appear to be the technical reasons for the observed damage. Soft-Story System. As described previously, a large number of residential buildings in the city have open ground floors leading to soft first story. Besides the elevator core, there are few walls, if any to provide lateral resistance. The upper floor frames are usually filled with un-reinforced brick masonry forming a very stiff lateral load resisting system. Most of the collapses and significant damage occurred in this type of soft-story buildings. It is well known from observations after past earthquakes in California and as well as after the recent Turkey earthquake that this type of building construction is highly vulnerable to earthquakes. Nearly all the deformation occurs in the columns in the soft-story, with rest of the building going for a ride during the earthquake. If these columns are not designed to accommodate these large deformations, they may fail leading to catastrophic failure of the entire building, as was the case in most of the buildings that collapsed in the city. An example of such failure is shown in Figure 1 The number of buildings collapsed or near collapse is approximate. The exact number will become available after the local authorities have completed a block-by-block survey of the entire building stock in the city. Goel-4
5, where half of the building (in the foreground) collapsed. Since there were several causalities, most of the building debris was removed by the time the EERI team reached the site. Figure 5. Half of the building with soft-story collapsed during the January 26, 2001 Bhuj earthquake (Photograph by Goel). Insufficient Confinement. As described previously, the columns typically had very light confinement, in the form of No. 2 or No. 3 ties with a spacing of 8 to 9 inches and 90-deree hooks. With large deformation and resulting extreme shear demands that occur in ground floor columns of lateral load resisting system with soft first story, the provided confinement was insufficient. As a result ground floor columns failed in brittle shear mode, which led to catastrophic failure in most collapsed buildings. Many buildings that did not collapse showed signs of significant shear cracking in the columns, as evident in crossing (diagonal) cracks in columns of building shown in Figure 4. Insufficient Shear Core Details. In soft-story structures, the shear wall around the elevator is usually relied upon to carry much of the lateral load at the ground floor level. However, insufficient amount of shear wall core coupled with poor quality of construction, improper reinforcement detailing, and insufficient connection to the rest of the building contributed to poor performance of many buildings. As evident from Figure 6, the shear core is usually connected to the rest of the building by floor slabs only with no or very few beams. The anchorage of the reinforcing steel from slab to the elevator core is insufficient. As a result, the slabs just pull out clean from the core leaving the stiff elevator core standing. In some situations, this may have proved to be blessing in disguise as it may have saved the rest of the building from collapsing. The shear core itself is about 4 to 6 inches thick with very light reinforcement consisting of two layers of mesh formed with of No.3 or 4 bar at vertical and horizontal spacing of about 18 inches. This detailing may have been insufficient to transfer the lateral load at the ground floor, as evident from the severe cracking present in the shear core of the building shown in Figure 7. The shear cracks in the elevator core of most building were observed only at the ground floor, an indication that the soft first story system imposed large demands on the shear core at this floor. Goel-5
Figure 6. Half of the building pulled away from the elevator core that formed the shear core in the building (Photograph by Goel). Figure 7. Severe shear cracking in the ground floor shear core (Photograph by Goel). Poor Quality of Material. Although the factors listed above contributed to failure of many buildings, they alone cannot be responsible for the building collapse. In many instances, only one out of several similarly constructed buildings in the same apartment complex collapsed; others while suffering significant damage did not collapse. This indicates that poor quality of material may have been one of the key factors that caused collapse of many buildings; the earthquakes are known to be unforgiving in finding structural defects (or weak links) wherever they may exist. Figure 8 shows bottom of a columns in a partially collapsed building. The concrete disintegrated within the reinforcement cage; when touched, the concrete appeared to have very little cement content. Also note that the 90-degree hooks have opened up which leads to little or no confinement of the concrete. In many situations, the concrete cover to reinforcement was noted to be less than half-inch; most of the cover was provided by plaster used to smooth the columns Goel-6
surface. It is also worth noting that most of the water supply in the outer part of the city is through ground water which is salty in taste. Usually the same water is used in preparing the concrete for construction. Therefore, the presence of salts may have also affected the concrete quality. Figure 8. Poor quality of concrete (Photograph by Goel). Figure 9. Part of the building shown in Figure 6 fell on the neighboring building indicating presence of torsional motions (Photograph by Goel). Plan Asymmetry. In many situations, the buildings had significant asymmetry in the plan. The asymmetry usually resulted from uneven distribution of mass, which occur due to modifications/additions after the building had been constructed. For example, there is some anecdotal evidence that in one of the buildings that collapsed during the earthquake event, owner of the building had built a swimming pool on one corner of the building. Since the type of residential building construction found in the city tends to be torsionally very flexible at the ground floor (usually shear wall core and a few columns provide the torsional rigidity), even slight eccentricities between the centers of mass and rigidity may lead to significant torsional motions. As a result, columns on one side of the building may experience excessive deformations. If not designed to accommodate these excessive deformations, they may fail leading to the buildings collapse. The excessive deformations due to torsion may have Goel-7
contributed to failure of at least one building. There is some physical evident to support this conclusion as half of the building shown in Figure 6 fell on the neighboring building shown in Figure 9, which would indicate that the collapsed portion was moving more than the portion that was left standing due to torsional motions. The direction of the building motion appears to be consistent with that expected due to eccentricity created by the swimming pool on one corner of the building. Soil Conditions. The localized soil conditions also contributed to the collapse of many buildings. A thick alluvial deposit along the Sabarmati River underlies the City of Ahmedabad. Although a cursory analysis of location of building collapses would indicate no particular pattern, a careful analysis reveals that most of the buildings that collapsed lie along the old path of Sabarmati River. This becomes apparent when location of collapsed buildings are plotted on the city map and compared with the satellite image of the city. Note that the building collapse pattern west of the Sabarmati River is closely aligned with the old path of the river, visible in the satellite image as a small loop of faint thick white line just west of the present river path. The south part of the city, especially the Mani Nagar area, where additional collapses were observed falls between two lakes, indicating the presence of either poor soil conditions or possibly construction on non-engineered fills. While the evidence presented in Figure 10 is strong, it would be useful to further verify these conclusions with filed testing. Figure 10. Building damage pattern and satellite image of the city of Ahmedabad showing old path of Sabarmati river (Courtesy Mr. S. K. Pathan, ISRO, Ahmedabad). Goel-8