Numerical models comparison for fluid-viscous dampers: Performance investigations through Genetic Algorithm

Abstract

Fluid-viscous dampers played a crucial role in the protection of new or existing buildings against external actions as earthquakes and winds. In the last decade, several investigations have been conducted aiming to develop accurate numerical models. However, none has been focused on a comprehensive comparison between the most used fluid-viscous damper models considering the variability of their parameters in a mass-production series.In this paper, an identification procedure has been performed by comparing nine different existing literature models with the objective of evaluating their ability to match experimental loops of mass-produced fluid-viscous devices, both in terms of accuracy and robustness. Indeed, the model that is most effective for reproducing the characteristic of a specific specimen may not be representative (i.e., showing larger parameters variability) of the mass production of the same device type. For this purpose, dynamic tests have been developed in the laboratory and the experimental outputs have been adopted as the target function of the procedure. The identification scheme has been designed by implementing an optimization procedure via Genetic Algorithm. Results demonstrate how differential laws better fit the experimental cycles with respect to algebraic ones, and also show how few models in the series can offer a high level of both accuracy and robustness.