EFFECT OF NEAR RESPONSE OF BUILDINGS ON HILLSLOPE DEPARTMENT OF INDIAN INSTITUTE OF TECHNOLOGY ROORKEE NEAR-FAULT GROUND MOTION ON DYNAMIC

Abstract

The immense damage potential due to the near-fault ground motion (NFGM) has been recognized during the damage investigations associated with the 1971 San Fernando, 1979 Imperial Valley, 1989 Loma Prieta, 1992 Landers, 1994 Northridge, 1995 Kobe, and 1999 Chi- Chi earthquakes. The NFGM is composed of high frequencies, representing accelerations, and one or more dominant long-period velocity pulses. It has been further recognized that the large amplitude pulses, primarily related to directivity effect, control the dynamic response of medium- and long-period structures, whereas, the high frequency part of the NFGM plays an important role especially for the response of short-period structures. The impulsive character of NFGM is mainly due to forward-directivity, and fault-normal (FN) component of ground motion is more dominant because of radiation pattern. These aspects of NFGM stimulated researchers to study the characteristics of NFGM and identify its governing parameters that are responsible for observed damages. This study is intended to investigate the near-fault pulsetype characteristics of the three moderate-sized Himalayan earthquakes, characteristics of the extracted near-fault pulses, interpretation of the NFGM response spectra and its comparison with the Indian Seismic (IS) codal spectra, and to analyze and interpret the response of hillslope buildings under the effect of near-fault pulse-type ground motions in the Himalayan region. From seismic safety point of view responses due to NFGMs are compared with the responses obtained from conventional types of seismic inputs. Strong motion arrays were deployed by the Department of Earthquake Engineering, Indian Institute of Technology, Roorkee, to allow measuring the strong ground motion due to moderate and large earthquakes in the Himalaya. These arrays resulted in recording of strong ground motions due to the 1986 Dharamsala earthquake (Mw 5.5) at nine stations, due to the 1991 Uttarkashi earthquake (Mw 6.8) at thirteen stations, and due to the 1999 Chamoli earthquake (Mw 6.5) at eleven stations. Near-Fault Ground Motion (NFGM) spectral characteristics of these three moderate-sized Himalayan earthquakes have been studied from the 33 available strong ground motion recordings. Pulse characteristics of FN components of ground motions in terms of pulse-periods, spectral pulse-periods and pulse-indicators have been extracted adopting wavelet analysis. To allow for pulse detection, standard methodology proposed by Baker (2007) has been adopted. Seven mother wavelets were used in the analysis, and it was found that db4 and db7 mother wavelets were more efficient in extracting the pulsetype characteristics. However, the spectra of long-period pulses extracted using Daubechies mother wavelet of order seven (db7) are closer to the long-period spectral amplitudes of the FN ii components of ground motions at three sites. Comparison of computed pulse-periods of the three earthquakes with the pulse-periods, estimated using available relations between magnitude and pulse-period, showed that computed pulse-periods are on lower side. The estimated pulse-periods by and large conform to the world-wide dataset but are on lower side than the average pulse-periods. It seems that the lower pulse-periods of the Himalayan earthquakes are due to the compressional tectonic environment and thrust-type focal mechanisms. The comparison of peak amplitudes of the velocity pulses estimated using available world-wide relations with the computed amplitudes for the three earthquakes showed lot of variability. NFGM spectra, at Bhatwari and Gopeshwar stations, showed higher spectral amplitudes in the velocity-sensitive and acceleration-sensitive regions compared to Indian codal response spectra. This is attributed to high PGV/PGA ratios. This demonstrated that NFGM leads to widening of acceleration-sensitive region, as a consequence of widening of the acceleration-sensitive region, the structures designed according to IS code as flexible structures shall behave as stiff structures in the near-fault region in the Himalaya. According to the IS code most of the hilly areas in the Himalaya fall in seismic zone IV and V. The building located on hillslope poses special structural problems. In hilly areas, many multistoreyed r.c. framed buildings rests on hillslope. The floors of these buildings generally step-back towards the hillslope and at the same time the building may have setback also, this stepping back of building towards hillslope result into unequal column heights at the same floor level. The buildings resting on hillslope are highly irregular and asymmetric. These buildings are subjected to severe torsion in addition to lateral shears under the earthquake excitation. In the present study to capture the detailed dynamic response of buildings on hillslope, 3D modeling of the building has been carried out. For this purpose dynamic analysis of two special moment resistant frame (SMRF) 3D configurations consisting of Step-Back (SB) and Step- Back-Set-Back (SBSB) models have been conducted. For both types of building models, the number of storeys has been varied from two to five because most of the residential buildings in the hilly regions are low-rise buildings with number of storeys limited to five storeys in majority of cases. Seismic response has been computed adopting six types of seismic inputs at three sites, namely, the recorded near-fault pulse-type ground motion, near-fault fault-normal (FN) component of ground motion, extracted (EXT) pulse of near-fault fault-normal (FN) component using ‘db7’ mother wavelet, residual (RSD) part of near-fault fault-normal component, estimated site-specific ground motion that include near-fault factor, and the IS codal spectra and its compatible ground motion. For computing dynamic responses of the buildings response spectrum method and modal time history method were adopted. iii Response of two types of building models (SB & SBSB) have brought out that as the number of storeys increase from 2 to 5, the response due to extracted (EXT) pulse increases compared to response due to residual (RSD) part of ground motion. This is only true when the seismic excitation is across the slope and direction of floor displacements are observed in the in-plane and out-of-plane on account of the low frequency pulse. However, for all SBSB building models, the response along the slope due to residual (RSD) part is always higher than the response due to extracted (EXT) pulse when the ground motion is applied along the slope because there is no significant variation in the time periods of the models along the slope. At Bhatwari site located in seismic zone IV as per IS code, the responses of all buildings of both types due to codal type ground motion were found to be highly un-conservative compared to pulse type ground motion and estimated site-specific ground motion. At Bhatwari site, for both building forms (SB & SBSB), both response parameters (floor displacements and ground column shear force) due to codal spectra show lowest values compared to those obtained from other seismic inputs. At Gopeshwar site that falls in seismic zone V, for higher storey SB building models, i.e., for 4 and 5 storey building models the response along the slope, and for all the SBSB building models, the response across the slope due to codal type ground motion found highly un-conservative compared to pulse-type ground motion and estimated sitespecific ground motion. The same trend was also observed when the response was considered across the slope for 2, 3, and 4 storey SB building models. At Shapur site, because of smaller magnitude of 1986 Dharamsala earthquake, the recorded pulse-type ground motion, extracted (EXT) pulse, and residual (RSD) part of ground motion showed lower response compared to responses obtained due to estimated site-specific ground motion and Indian seismic codal based ground motion for zone V. With the increasing period of buildings, response due to extracted long-period pulse at three sites is in close agreement with responses due to recorded pulse-type ground motions & FN component of ground motions. This illustrates the possibility of representation of NFGM with an appropriate long-period pulse model. The Indian codal spectra needs modification and the effects of pulse-type ground motion due to moderate, large, and great earthquakes could be incorporated for sites located in the vicinity of active faults.