Semi-active vibration control of connected buildings using magnetorheological dampers

Abstract

Response attenuation of seismically excited adjacent buildings connected by a MR damper is studied using semi-active LQR controller design. The modified Bouc-Wen model relating damper force to input voltage/states is considered. Thus, obtaining the input voltage to realize a desired control force is a non-trivial task. The desired control force is obtained using LQR control, and desired voltage predicted based on either a RNN model or a CVL. Results for the 5-storey and 3-storey interconnected buildings (B5-B3) are obtained in terms of peak and RMS responses. These are compared with passive-on control for which a constant saturation voltage is applied to the damper. Percentage reduction in maximum peak [RMS] response, when using LQR-CVL instead of passive on control, is 24[20] for interstorey drift, 18[23] for displacement, and 17[26] for accelerations. Corresponding further percentage reductions of 6[5], 5[5], and À5[4], and reductions in base shear, occur when considering LQR-RNN vis-á-vis LQR-CVL control. Peak accelerations for B5[B3] attenuate [increase] significantly, resulting in a re-distribution and reduction of base shear, when comparing semi-active versus passive-on control. Results show that connection of adjacent buildings using MR damper driven by a LQR-RNN controller provides a promising means of response attenuation.

Shake table tests of structures subject to pounding

“…As already mentioned, to investigate possible beneficial and adverse effects when the two structures are tied together to suppress pounding, the adjacent floors were linked through steel plates and UPE short beams ( Figure 2). It is possible that other coupling devices [29][30][31][32][33][34] could have been more beneficial. Nevertheless, rigid ties are the simplest coupling device, and they allow us to investigate the two extreme cases of the uncoupled and fully (rigidly) coupled structures.…”

“…Nevertheless, coupling of the buildings will modify their earthquake response. The effects of passive or semi‐active coupling devices have been studied in several works as for instance, in the literature …”

“…The effects of passive or semi-active coupling devices have been studied in several works as for instance, in the literature. [29][30][31][32][33][34] In order to gain further knowledge about the effect of pounding on structures and examine issues that have not been addressed in previous studies, an experimental campaign was carried out with two adjacent structures on the main shake table of the French Commissariat à l'Energie Atomique (CEA) in Saclay, France. Several test configurations were considered with various gap values, diverse excitation signals, and increasing excitation amplitude leading to significant damage.…”

Semi-Active Control Using Magnetorhelogical Dampers with Output Feedback and Distributed Sensing

“…It may be noted that the solution of Eqs (12,13,14) would in general yield multiple local minima of the performance index J. Convergence to a particular minima is greatly influenced by the choice of initial stabilizing gain (K 0 ). The present algorithm used to solve the OSOF problem does not yield the global minimum J (although the resulting control is quite effective).…”

“…Chang and Zhou [11] used LQR control with a recurrent neural network model for damper inverse dynamics. Response attenuation of connected buildings using LQR control was studied by Motra et al [12] for clipped/neural-network based voltage laws. Yuen et al [13] used reliability based robust linear controller design with a clipped voltage law.…”