Neuro-fuzzy model of superelastic shape memory alloys with application to seismic engineering
MetadataShow full item record
Shape memory alloys (SMAs) have recently attracted much attention as a smart material that can be used in passive protection systems such as energy dissipating devices and base isolation systems. For the purpose of investigating the potential use of SMAs in seismic engineering applications a soft computing approach, namely a neurofuzzy technique is used to model dynamic behavior of CuAlBe shape memory alloy wires. Experimental data are collected from two test programs that have been performed at the University of Chile. First, in order to evaluate the effect of temperature changes on the behavior of superelastic SMA wires, a large number of cyclic, sinusoidal, tensile tests are conducted at various temperatures. Second, to assess dynamic effects of the material, a series of laboratory experiments are conducted on a scale model of a three story model of a building that is stiffened with SMA wires and given excitation by a shake table. Two fuzzy inference systems (FISes) that can predict hysteretic behavior of CuAlBe wire have been created using these experimental data. Both fuzzy models employ a total of three input variables (strain, strain-rate, and temperature or prestress) and one output variable (predicted stress). Values of the initially assigned membership functions for each input are adjusted using a neural-fuzzy procedure to accurately predict the correct stress level in the wires. Results of the trained FISes are validated using test results from experimental records that had not been previously used in the training procedure. Finally, numerical simulations are conducted to illustrate practical use of these wires in a civil engineering application. In particular, dynamic analysis of a single story frame and a three story benchmark building that are equipped with SMA damping elements are conducted. Then, an isolated bridge that utilizes a linear rubber bearing together with SMA elements is analyzed. Next, in order to show recentering ability of SMAs, nonlinear time history analysis of a chevron like braced frame is implemented. The results reveal the applicability for structural vibration control of CuAlBe wire whose highly nonlinear behavior is modeled by a simple, accurate, and computational efficient FIS.
Ozbulut, Osman Eser (2007). Neuro-fuzzy model of superelastic shape memory alloys with application to seismic engineering. Master's thesis, Texas A&M University. Available electronically from