An evaluation of stress-softening behaviour of HDNR bearings on the seismic response of base-isolated structures

In the last few decades, high damping natural rubber (HDNR) bearings have been extensively employed for seismic isolation of bridges and buildings because of their low horizontal stiffness and high damping capacity, which allows shifting the vibration period of the isolated structure away from where the earthquake input has the highest energy content and at the same time controlling the motion of the system. In HDNR material, filler is added to the natural rubber in order to improve its properties such as stiffness and dissipative capacity. The addition of the filler induces also a stress-softening behavior, known as “Mullins effect”. This effect makes the response of HDNR bearings path-history dependent and thus may influence the seismic performance of isolated systems. Published literature has suggested that the initial “virgin” properties of the material are eventually recovered. Accordingly, current seismic codes make the assumption that “Mullins effect” is a reversible phenomenon. The present work aims at studying the consequences of such strain-history dependent behavior on the seismic response of structural systems isolated with HDNR bearings. In particular, the first part of the paper reports a wide experimental campaign carried out on a large number of virgin rubber samples in order to better investigate some aspects of the stress-softening behavior of filled rubber, such as the direction-dependence and the recovery prosperities, and to characterize the stable and transient response under different strain histories. Test results are used to define a model for simulating the behavior of HDNR bearings in shear, which is an advancement in the description of both the stable and the transient behaviors. The proposed model has been used to analyze the seismic response of a simplified isolated structure modeled as a S-DOF (single degree of freedom) system under ground motions with different characteristics and by considering two different conditions for the bearings: one assuming the virgin (or fully recovered) rubber properties and the other assuming the stable (or fully scragged) rubber properties. The obtained results show that, except for the special case of near-fault (NF) ground motions, the differences between the responses are limited although not negligible, whereas for NF records, the assumption of the virgin (or fully recovered) condition significantly reduces the effect of this type of motion on isolated structures.