Modeling of the lateral load resistance of masonry infilled frames with innovative steel ties

This research aimed at modeling a cyclic load-deformation hysteresis relationship, captured from experimental results of a reinforced concrete (RC) frame with and without an infill masonry wall. In-plane behavior of masonry walls plays a major role in the overall cyclic loading response of an RC frame and the lateral load resistance, which are important design aspects. The out-of-plane behavior of a masonry wall is the most frequently encountered failure mode under seismic loads. In order to increase the out-of-plane stability of the infill wall, innovative steel ties were installed in the masonry wall and their contribution of the masonry infill was studied. To simulate the RC frame behavior for different tie configurations, respective behavior of RC frames under cyclic loading were studied, and two main simulation strategies were conducted. First, nonlinear spring models were utilized to model the nonlinear behavior of joints regions in the RC bare frames under incremental cyclic loading. To simulate the effect of infill walls with or without steel ties, a diagonal pin jointed nonlinear spring was added to the RC frame to account for the corresponding rigidity of the wall. This modeling strategy relies on analysis of the specimen section, mainly its geometry and material properties, and does not correlate the experimental results to the analysis outputs. Second strategy was to calibrate a nonlinear plastic spring based on experimental results to simulate its hysteresis behavior under lateral cyclic loading. Understanding the linear or nonlinear relationship between load and deformation in structural materials or structural frames is important for an accurate simulation. Therefore, modeling load-deformation relationship based on experimental results could be a viable approach to predict load-deformation behavior of a similar frame under a loading pattern. To reduce the experimental data size recorded with measuring devices or Linear Variable Differential Transformers (LVDTs), regular and irregular data resampling technics were implemented. Hysteresis models to simulate the cyclic response of an RC frame were reviewed, then simulation strategies were implemented to obtain a best fit to the experimental results. The difference between analytical and the experimental results was then studied. A Genetic Algorithm (GA) was used to fit the simulated results to experimental results through minimization of the disagreement between simulated and experimental results. Genetic Algorithm seeks he best parameters to describe the experimental results. This method could be used to train models to predict the capacity and performance of frames, under different loading patterns. The appendix includes the simulated hysteresis results and demonstrates how the simulated model can fit the experimental results in close agreement. Three hysteresis models have been used to represent the experimental results. 1-Kenematic hysteresis model, 2-Takeda hysteresis model 3-Pivot hysteresis model. Since Kinematic and Takeda models rely on their backbone to represent the cyclic data, the backbone was extracted from experimental cyclic results and was assigned to these models. However, Pivot model has five more degradation parameters that was obtained through optimization while minimize the deviation between simulated and experimental results. While fitting simulated results to experimental results, Pivot hysteresis model, in comparison with Kinematic and Takeda model, well presented the experimental results. At the end of the thesis we modeled cyclic lateral excitation of the bare frame and infill frame with Pivot hysteresis model. The backbone load-deformation curve of such hysteresis model was calculated from analytical finite element modeling. Then, the cyclic parameters of the hysteresis model were obtained and assigned to the model using the second strategy. By analytically obtaining the backbone load deformation curve, material and geometry characteristics of the specimen is considered. It can also be assumed that Pivot hysteresis model parameters are not significantly varied while geometry and material characteristics of the frame are changed. Therefore, Pivot hysteresis parameters that have been captured from cyclic excitation behavior of a frame can be assigned for almost any frame that has approximately the similar geometry and material characteristics to that frame. In this research application of MATLAB has been used as the programming platform and SAP2000 is utilized as the finite element structural solver. To facilitate obtaining the analytical results, an Application Programming Interface (API) has been implemented to utilized the functionality of SAP2000 from MATLAB. This allowed us to take advantage of the state-of-the-art functionalities of SAP2000 from MATLAB as oppose of developing such solver from the ground up. In addition, in this research GA toolbox has been utilized as a convenient GA platform. The novelty of this research is the implementation of new strategies to model and predict the performance of frames or materials under any cyclic pattern by use of experimental results.