In the last decades, the topic of structural control has gained increasing attention in structural design due to the fact that structures are becoming increasingly slender and therefore more prone to vibrations. In the context of passive control, fluid inerter-based control systems represent a novel development with great potential. Their main advantage is that the apparent mass (called inertance) which is able to reduce the displacement and/or acceleration of the structure, is orders of magnitudes higher than its physical mass.

In order to fully explore and optimize the performance of the fluid inerter-based control device, a numerical model that accurately represents the structural behavior is required. In civil engineering applications, the frequency range of interest is usually low, meaning that any non-linear effects that may occur play a major role and cannot be neglected in the numerical model. The selection of model type is intrinsically related to the determination of its associated parameters such that the model predicts the measured performance. Despite the high accuracy of the established numerical model, there may still be a gap between the model and the measurement due to the presence of uncertainties in the parameters. The identification of model parameters in a stochastic setting provides a means to understand and reduce the discrepancies between model and real behavior.

In the present study, the so-called subset simulation method, which has originally been developed for reliability analysis, is used for the stochastic identification of model parameters. It is shown that in this framework, issues such as non-uniqueness of the solution can be addressed and information about the spread of parameter values can be gained to obtain a more realistic model, which represents the real structural behavior more accurately.