Field testing and model updating of typical RC buildings for damage identification

A great portion of existing building stock of Turkey consists of reinforced concrete (RC) structures. A significant amount of these RC structures, which were constructed several years ago, do not satisfy the requirements of the modern seismic design codes. Although majority of these structures are legally engineered structures, even some of them are not, due to the lack of an official control mechanism, these structures mostly do not fulfil the requirements of the seismic design codes, which were in force during their contruction time. Accordingly, these buildings have basic structural weaknesses such as low concrete strength, insufficient reinforcement detail, etc. and represent a significant risk in case of any seismic action. Thereby, the condition assesment of the existing structures after an earthquake is crucial for determining the residual performance of the building. Up to the present, condition assessment of the buildings after a major earthquake has been immediately carried out based on visual inspection. Although this is a required and an essential method for making decision the current situation of the buildings, some damages could not be identified through visual inspection. For example, internal cracking or slip of the lap-splice can be hidden from view, however, their effects on the load bearing capacity of the building can be crucial and vital. From this point of view, the vibration signiture of the building, which is obtained through St-Id methods, can provide significant information about the condition of the buildings. In this dissertation, it is aimed to present an automated damage identification algorithm through field testing and model updating of typical RC buildings. For this purpose, first, a series of field testing was carried out for quantifying the influence of different types and extents of structural damage on the dynamic characteristics of typical existing RC buildings. In the context of the field testing, vibration-based tests were performed before and after quasi-static reversed lateral loading cycles as well as at certain damage levels. Since the test buildings (the buildings TB-1 and TB-2) are representative of large number of substandard RC buildings, the test results would be useful for the seismic performance assessment of existing RC buildings. At the end of the field tests, it was found that the modal frequencies decreased, and the modal damping ratios increased significantly up to the elastic limit. However, the progressing of damage beyond the elastic limit was not accompanied with a similar significant change in the modal frequencies and the modal damping ratios. Furthermore, despite a significant degradation in lateral load bearing capacity, as approximately 20%, induced by quasi-static lateral loading at around 3% lateral d.r., no intensified structural damage could be observed by visual inspection in case of building TB-2. However, the forced vibration test results presented that the modal frequencies and the damping ratios changed due to lateral loading in comparison with its undamaged state. These findings indicate a serious risk about the missing the structural damages during visual inspections after earthquakes. Hence, utilization of dynamic testing accompanying with visual inspection could be a promising method for more realistic assessment of structural damages after seismic events. Second, an automated damage identification algorithm was proposed using the results of the experimental study. A model based automated algrorithm relies on the updating of Finite Element Model (FEM) of the test buildings was introduced. Different than the majorty of the damage identification methods introduced in literature, two existing RC structures accompanied with different extend of damage were considered. By means of the presented damage identification algorithm, first the initial FEM of the buildings were successfully updated via changes stiffness, unit weight and damping parameters and the baseline model, which was taken as reference for damage identification, were obtained. Subsequently, the revision in the two-step algorithm in terms of updating the damping parameters provided to obtain the mathematical description of the buildings. Significantly, curve fitting clearly became a useful and practical part of the model updating algoritm and found to be a promising tool in case of the dynamic measurements including noisy components. Followingly, the FRF based model updating algorithm, which was introduced by updating the initial FEM of the buildings, was adopted for damage detection, localization and quantification using experimentally collected data from the building TB-1 and TB-2. At the end of the application of algorithm, it was found that the first step of the algorithm (Step 1) provided a successful damage detaction, localization and quantification process. Although the second step (Step 2) also provided to match the experimental FRFs and FEM FRFs in terms of their amplitudes, it is basically have not any contribution for Level 3 damage identification purpose. Subsequently, the damage identification algorithm was computed by considering a simplified FEM of the test building TB-2 to examine the efficiency of utilizing a simple model. At the end of the process, a damage detection was provided in terms of story level not member level. Although this can not be classified as a Level 3 damage identification since it did not address the damage location, it could be a practical and rapid approach for assessing a multi-story building ater earthquake. Third, the Level 3 damage identification method was extended in order to predict the serviceability of the structure after damage. For this purpose, the nonlinear model of the building TB-1 was established, and the model was then subjected to the same lateral loading reversals with the quasi-static tests. Following to calibrating the nonlinear model with regard to experimental base shear force-1st story drift relationship, a set of dynamic data was generated from the model for both undamaged and damaged case. After verifying the dynamic results obtained from field tests (and updated ones) and estimated through generated signal are matching, the performance based assessment results of the building, which were obtained using ASCE 41-13, were compared with the vibration based damage identification results. By this way, the damage identification algorithm was not only examined in terms of efficiency and validity, but also extended to investigate the residual capacity of damaged structures. Most importantly, an additional step for algorithm was suggested so that nonlinear moment-rotation based damage state of the member can be obtained through vibration data. Based on the literature survey, variation of dynamic characteristics with respect to increasing seismic structural damage and followingly a damage identification process was for the first time presented for the full-scale RC structures. Furthermore, the utilization of nonlinear moment-rotation based damage criteria for vibration based damage detection could provide practical and result-oriented assessment tool for engineers. Both experimental and analytical results present the importance of utilization of the vibration based methods for condition assessment of RC structures.