Damage Detection Of Reinforced Concrete Columns Based On Vibration Measurements

In recent years, structural health monitoring (SHM) has drawn significant attention from the engineering community because of the potential risks associated with structures resulted from aging, misuse, lack of proper maintenance, and, in some cases, climate change impacts. The use of vibration responses has been recognized as a potential means of detecting damage or flaws in a structure. This technique has been widely used in the civil engineering research community to extract structural modal parameters (e.g., natural frequencies, damping ratios, and mode shapes) from vibration measurements. These modal parameters serve as a basis for finite element model updating, structural damage detection, structural safety evaluation, and structural health monitoring. At the same time, parameter identification method in time domain attracts more and more attentions from researchers. The main contents of this thesis include:1. Damage detection for a RC column after quasi-static loading is studied using modal testing data. PolyMax method is employed to extract the modal parameters experimentally based on its impulse responses recorded in its undamaged and various damaged states. These identified modal parameters are presented and compared at different levels of damage. They are then used to identify damage in the column using the finite element model updating procedure based on sensitivity analysis method and neural network method respectively. Obtained results from these two methods are analyzed and compared.2. Vibration based damage detection techniques were employed to detect damage in reinforced concrete (RC) columns exposed to different levels of fire. Modal tests and analysis were carried out for the RC columns in original condition at first. Then, the RC columns were put in fire. Finally, dynamic test on the damaged RC columns was carried out and changes in the dynamic characteristics of the columns before and after fire were examined. Two different methods including Eigen-System Realization Algorithm (ERA) and PolyMax method were employed for extracting modal parameters of the original columns and then after fire. The performance of the proposed methods is discussed. Lastly, the location of the damage is detected by employing the Co-Ordinate Modal Assurance Criterion (COMAC) method, and the damage level is determined by the neural network based model updating method3. Basing on discrete time solution of the state equation, a unique parametric assessment approach can be put forward by the direct use of dynamic responses under impact excitation with two neural networks, neural network emulator(NNE) and a parametric evaluation neural network(PENN). The acceleration time histories obtained from quasi-static tests are then employed to validate the performance of the proposed approach for structural stiffness identification and damage detection.4. Based on the modal parameters obtained from previous experiment, uncertainties in the measured response are considered. Random errors with zero mean value and normal distribution are added in the measured modal data. The probabilities of damage existence of each element are calculated based on the distribution of identified elastic modulus before and after damage.