Modeling and identification of magnetorheological (mr) dampers for civil engineering structures

This thesis aims to develop a robust dynamic hysteretic model capable of capturing the nonlinear response of a Magnetorheological (MR) damper using an innovative parametric identification methodology for the purpose of implementing these devices in large scale civil engineering structures for vibration mitigation under earthquake, wind, and/or man-made dynamic loads, Semiactive devices comprising MR dampers are typically designed for linear structures. Civil structures, however, are designed to yield, thus behaving nonlinearly during extreme dynamic loading. Because they cannot inject mechanical energy into the controlled system, semiactive devices are inherently stable and well suited for application to structures with potential to behave nonlinearly. Additionally, the low power requirements of semiactive devices ensure that during extreme events, when external power may not be available, the semiactive device can continue to fully function using an alternate power source. In addition to the controllability, stability (in a bounded-input bounded-output sense), and low power requirements inherent to semiactive devices, MR fluid dampers with their large temperature operating range and relatively small device size have the added benefits of: (i) producing large control forces at low velocities with very little stiction; (ii) possessing a high dynamic range (the ratio between maximum force and minimum force at any given time); and (iii) having no moving parts, thus reducing maintenance concerns and increasing the response time. This dissertation starts by developing a fundamental understanding of MR fluids physical properties and basic principles, the MR dampers mechanical characteristics, the dynamic models used to represent its nonlinear phenomena, and their application in full-scale civil engineering structures. These devices are highly nonlinear and their accurate modeling is important for effective simulation and control system design. The use of the laws of physics for such device modeling is complex, and models that combine a physical understanding of the device along with a black-box description are used instead. The system identification plays a key role in order to link the modeling and control of a system. A hysteretic model based on the normalized Bouc-Wen model is proposed to test its effectiveness in a small-scale controllable friction MR damper and a large-scale MR fluid damper. A sensitivity analysis is carried out due to a large uncertainty on the identified viscous friction coefficient when the viscous friction term is smaller than the dry one. For this reason, this identification methodology is modified appropriately. Experimental results with two classes of MR dampers show a good level of accuracy between the predicted and measured responses.