Nonlinear viscoelastic isolation for seismic vibration mitigation

TL;DR

This paper evaluates nonlinear viscoelastic damping in seismic base isolation, demonstrating its robustness across various excitation spectra and its potential to outperform linear isolators in unpredictable seismic conditions.

Contribution

It introduces a nonlinear base isolation model with viscoelastic damping and shows its improved robustness over linear systems for seismic vibrations.

Findings

Nonlinear damping exhibits non-monotonic bell-shaped behavior.

Poorly damped vibrations can be triggered by steep damping decreases.

Tuned nonlinear isolators outperform linear ones across diverse seismic spectra.

Abstract

The aim of this paper is to assess the effectiveness of nonlinear viscoelastic damping in controlling base-excited vibrations. Specifically, the focus is on investigating the robustness of the nonlinear base isolation performance in controlling the system response due to a wide set of possible excitation spectra. The dynamic model is derived to study a simple structure whose base isolation is provided via a Rubber-Layer Roller Bearing (RLRB) (rigid cylinders rolling on rigid plates with highly damping rubber coatings) equipped with a nonlinear cubic spring, thus presenting both nonlinear damping and stiffness. We found that, under periodic loading, due to the non-monotonic bell-shaped viscoelastic damping arising from the viscoelastic rolling contacts, different dynamic regimes occur mostly depending on whether the damping peak is overcome or not. Interestingly, in the former case, poorly damped self-excited vibrations may be triggered by the steep damping decrease. Moreover, in order to investigate the robustness of the isolation performance, we consider a set of real seismic excitations, showing that tuned nonlinear RLRB provide loads isolation in a wider range of excitation spectra, compared to generic linear isolators. This is peculiarly suited for applications (such as seismic and failure engineering) in which the specific excitation spectrum is unknown a priori, and blind design on statistical data has to be employed.