A Tale of Three Cities

Abstract

To our knowledge, no single building has experienced three successive strong ground motions from major earthquakes that occurred nearby. In Turkey this happened in a roundabout way when three identical buildings designed to serve as the provincial directorate offices for the Ministry of Public Works and Resettlement (MPWR) underwent such an experience over a time span of 11 years in three different cities that were hit by major earthquakes. The provincial offices are designed and constructed according to template designs under the Ministry’s own supervision. For ease of access and security, the Ministry’s strong motion recording stations that are part of the national network are located adjacent to these buildings. This study examines the performance of three of these standard ground-plus-four-story reinforced concrete (RC) frame buildings that were subjected to strong ground motions in different cities of Turkey. Bidirectional nonlinear dynamic analyses of 3D analytical models are performed. The principal focus of these nonlinear analyses is to assess whether the analytical model of the buildings could indicate column-beam damage consistent with that observed at the sites after the earthquakes. Our results illustrate that nonlinear time history analyses are capable of indicating the occurrence of shear failure in captive columns, but they overestimate the global damage level for all buildings. Introduction Investigating the response of structures during the earthquakes has been a useful tool to improve methodologies for design and analysis of structures. An example is the study on the structural performance of a RC building: the extensive field and analytical investigation of the Olive View Hospital Medical Treatment and Care Facility which suffered severe damage during the 1971 San Fernando earthquake (Mahin 1976). Observed structural damage was compared with the predictions made through linear and nonlinear dynamic analysis of the mathematical models. Another similar research (Kreger and Sozen 1989) was made on the Imperial County Services Building of El Centro in California which was severely damaged during the October Graduate Research Assistant, Department of Civil Engineering and Earthquake Engineering Research Center, Middle East Technical University, TR-06531, Ankara Professor, Department of Civil Engineering and Director, Earthquake Engineering Research Center, Middle East Technical University, TR-06531, Ankara Proceedings of the 9th U.S. National and 10th Canadian Conference on Earthquake Engineering Compte Rendu de la 9ième Conférence Nationale Américaine et 10ième Conférence Canadienne de Génie Parasismique July 25-29, 2010, Toronto, Ontario, Canada • Paper No 167 15, 1979, Imperial Valley Earthquake. Measured response of the building was presented and a hypothesis was developed for the prediction of identified column failures. However, research examining the response of identical RC frame buildings subjected to different strong ground motions has not been carried out. This paper investigates the structural performance of typical branch office of MPWR. This is a ground-plus-four-story RC frame building constructed in the 1980s in different regions of Turkey. All buildings suffered damage with varying degrees of severity during the March 13, 1992 Erzincan, November 12, 1999 Düzce ̧ and May 1, 2003 Bingöl earthquakes. During these events, three-component strong ground motion data were recorded in one-story buildings adjacent to the main buildings in Bolu and Bingöl and in a onestory building two kilometers away from the main building in Erzincan. The building in Bolu sustained severe damage that we judge to represent a “life safety” level of performance while those in Erzincan and Bingöl sustained lighter damage corresponding to slightly more than “immediate occupancy.” After the Düzce earthquake, a careful recording of the damage distribution was performed for the building located in Bolu. The fortuitous combination of known input motions for the buildings and their design drawings permitted us to respond to the obvious question that begs to be answered: given the tools of current computational performance assessment technology, is the damage in each of these buildings within reach of our ability to predict them by proper modeling? In the paper, the observed structural damage is compared with those predicted in bi-directional nonlinear dynamic analyses of 3D analytical models. Description of the Strong Ground Motions The strong ground motions used in this study were recorded by stations of the Turkish national strong-motion network. The processed data and the seismological features of the motions have been obtained from (TUBITAK 2009) which is the first systematic compilation and uniform processing on strong motion data recorded by the Turkish national strong motion network with detailed geophysical and geotechnical site measurements for all stations. The station information and important seismological features of the ground motion data used in this paper are given in Table 1. Table 1. Seismological features of the strong ground motions. March 13, 1992 Erzincan Earthquake November 12, 1999 Düzce Earthquake May 1, 2003 Bingöl Earthquake Station Location Meteorology Building, Erzincan Ministry of Public Works and Resettlement, Bolu Ministry of Public Works and Resettlement, Bingöl Depth (km) 22.6 10.4 10.0 Rjb* (km) 3.3 8.0 2.2 Fault Type Strike-slip Strike-slip Strike-slip Vs,30** (m/s) NI**** 294 529 Mw*** 6.6 7.1 6.3 Longitudinal PGA (g) 0.488 0.754 0.556 Transverse PGA (g) 0.412 0.821 0.282 Vertical PGA (g) 0.243 0.204 0.481 Longitudinal PGV (cm/s) 78.2 52.3 34.4 Transverse PGV (cm/s) 108.4 66.0 21.8 Longitudinal PGD (cm) 29.5 12.5 10.2 Transverse PGD (cm) 34.4 10.5 5.1 *Rjb : Joyner-Boore distance, **Vs,30 : Average shear-wave velocity of the upper 30 m. soil layer ***Mw: Moment magnitude, ****NI: Not investigated Description of the Case Study Building The case study building is main part of the typical branch office of MPWR which is a five-building complex designed and constructed in 1980’s. Other than this building there are four other facility buildings separated by seismic joints in the same compound (Fig. 1). The main building is a ground-plus-four-story RC structure 20 m by 13.2 m in plan. Story height is 3.8 m. for the ground floor and 3.2 m. for the other floors. The building is rectangular in shape with three bays in both perpendicular directions (Fig. 2). (a) (b) Figure 1. a) Plan of the building complex b) General view of the case study building Beams in the exterior frames have an unusual depth of 1.2 meters with 0.3 m width. Dimensions of the beams in the interior frames are 0.3 m by 0.7 m in the longitudinal direction (Fig. 2, Axis A-F) and 0.3 m. by 0.6 m. in the transverse direction (Fig. 2, Axis 1-4). There are eight rectangular columns oriented in the longitudinal direction, five in the transverse direction and three L-shaped columns on the corners. Except for the L-shaped columns, sizes of the columns and their longitudinal reinforcement decrease progressively from the lower to upper stories but dimensions of the beams and amount of the longitudinal reinforcement do not vary with height. Case study building Five story Case study building One story building Enclosed Garden Office building Four-story Two story building Seismic Joint Seismic Joint One story building Figure 2. Ground Floor Plan The slab thickness is 15 cm at each level. The peripheral masonry infill walls are 26 cm in thickness. The infill walls separating office rooms from corridors are 19 cm thick and those separating office rooms from each other are made up of 9 cm masonry. The amount of masonry walls is less at the ground and top floor than the other floors. Analytical Models 3D nonlinear analytical models of the buildings in Bolu, Erzincan and Bingöl are carried out. In all models, distributed plasticity is utilized through fiber analysis approach (Perform 3D 2005) to simulate the nonlinear and bi-axial flexure behavior of the columns. Shear hinges with Vx and Vy interaction are defined at the column ends to limit the shear strength where the flexure-shear strength is larger. Beam members are composed of elastic elements with effective stiffness. Bi-linear moment-curvature relationship and elastic-perfectly-plastic shear hinges are defined at both ends. The contribution of slip deformation to the yield displacement is taken into account in beams by introducing members with reduced effective stiffness to the model. However, due to high level axial load and aspect ratio, the slip of the reinforcing bars is neglected in the columns (Elwood and Eberhard 2009). The infill walls are also taken into consideration. Eigenvalue and nonlinear dynamic analyses are based on structural stiffness after separation. The in-filled frames are modeled as equivalent diagonally braced frames that are represented by diagonal compression struts. Masses are concentrated at the mass centers of each floor. P-delta effects are included. Following customary practice, T sections are utilized for beam sections and the effective flange width values are considered as 1/5 of clear span length (ASCE 2007) of the beam. Rayleigh damping was utilized in the analytical model with 5 percent damping ratio specified for the first and fourth modes (Chopra 1995). Other assumptions about the material and loading are summarized in Table 2. 1 2 3 4 A B C D E F