Phd Degree / Doktora
Permanent URI for this collectionhttps://hdl.handle.net/11147/2869
Browse
2 results
Search Results
Doctoral Thesis Enhancing Earthquake Performance of Civil Structures Via Structural Control(Izmir Institute of Technology, 2021) Şenol, Vedat; Turan, GürsoyIn this study, two different benchmark buildings (3 and 20-story) are employed to attenuate structural responses under seismic disturbances. As control devices, active (actuators), semi-active (Magneto-rheological dampers), passive (Tuned mass dampers and Friction Pendulum Bearings), and hybrid controllers are utilized. The 3-story structure is modeled linearly and employed to apply to different control strategies. Some control algorithms: LQR, PDD-state-feedback, pole-placement, $H_{\infty}$, $ H_2 $, are used with active and semi-active control devices. As passive devices, TMDs and FPBSs are utilized on the nominal-linear model. Thereafter, hybrid controllers are employed: one composed of a TMD and actuator/MRD and one composed of an FPBS and actuator/MRD. A robust controller, $\mu$-synthesis, is employed to control the same linear structure having uncertainties in mass, stiffness, and damping matrices within reasonable ranges. A nonlinearly-modeled 20-story benchmark structure is employed to implement passive and hybrid control strategies. As passive devices, STMD and MTMD setups are employed. Further, a robust control algorithm is used through an actuator serially connected to the STMD. Subsequently, variations caused by nonlinearities are determined. These variations are regarded as uncertainties, and the $\mu$-synthesis is utilized in the design of a robust controller on a truncated linear model. Then, the designed robust control is employed to control the 20-story benchmark structure modeled nonlinearly. The structural responses in both frequency and time domains are discussed. Matlab, Python, and OpenSees framework (Tcl/Tk) were employed to realize all linear and nonlinear simulations throughout the study.Doctoral Thesis Modal Identification of Structures by Using Bayesian Statistics(Izmir Institute of Technology, 2019) Hızal, Çağlayan; Hızal, Çağlayan; Turan, Gürsoy; Turan, GürsoyBayesian Probabilistic approaches in the health monitoring of civil engineering structures has gained remarkable interest during past decades. When compared to the available Operational Modal Analysis (OMA) methods, Bayesian Operational Modal Analysis (BAYOMA) determines a probabilistic range with a most probable value and uncertainty instead of a certain result. For this reason, the most important difference of BAYOMA lies in its capability of uncertainty quantification. Therefore, the modal parameters of a measured structure can be determined based on a probabilistic logic according to various cases (for example single measurement setup, well separated and/or closely spaced modes, multiple measurement setups). Further, the finite element model of the investigated structure can also be updated by a Bayesian approach incorporated with modal identification procedure. Some efficient BAYOMA methods such as Bayesian Spectral Density Approach (BSDA) and Bayesian Fast Fourier Transform Approach (BFFTA) have been presented by various researchers during the past two decades. Despite their efficient and fast solution procedure, the available methods have some critical issues that need to be solved. Most of these problems especially lie in the quantification of posterior uncertainties and some special cases arise in closely spaced modes and/or multiple setup measurement cases. In the literature, solutions for the aforementioned problems have been also presented by various researchers. In the light of the accumulated knowledge in the literature, this study presents a computational framework for BAYOMA and Bayesian Model Updating (BMU). In addition to some improvements to the available methods, new and alternative approaches are presented for BAYOMA and BMU. According to the results, it is seen that the quality of identified modal parameters and updated finite element models increases significantly by the proposed computational procedure.