ABSTRACT
ABSTRACT: This work proposes a robust optimization criterion of mechanical parameters in the design of linear Tuned Mass Dampers (TMD) located at the top of a main structural system subject to random base accelerations. The dynamic input is modelled as a stationary filtered white noise random process. The aim is to properly consider non-uniform spectral contents that happen in many real physical vibration phenomena. The main structural system is described as a single linear degree of freedom, and it is assumed that uncertainty affects the system model. The problem parameters treated are described as random uncorrelated variables known only by the estimation of their means and variances. Robustness is formulated as a multi-objective optimization problem in which both the mean and variance of a conventional objective function (OF) are minimized simultaneously. Optimal Pareto fronts are obtained and results show a significant improvement in performance stability compared to a standard conventional solution.
Engineers, physicists and, in general, scientists dealing with real phenomena usually have to deal with uncertainty which often raises serious theoretical and computational difficulties. With the aim of reducing these complications that frequently make many problems irresolvable, standard methods have been developed in structural analysis assuming implicitly that all involved parameters are deterministically known. This remains is an oversimplification of real conditions because parameters are only partially known. Due to economical and technical reasons, only few problem parameter measurements are available if not simply intrinsically uncertain. The uncertainty of structural problems may afflict many involved factors, such as dynamic loads intensity, material mechanical parameters and geometrical configurations, all commonly considered as deterministic in standard analysis.
