Linearization generally represents a valid approximation in the modelling process of complex dynamic systems, but in several cases it might affect the reliability of the results. Although all systems in nature are nonlinear, accounting for all the existing nonlinearities is practically impossible since the computational time would increase indefinitely. Therefore, in some cases the most relevant nonlinear phenomena should be included. In many engineering systems, significant nonlinear effects are due to the connections between coupled subsystems, such as bolted joints and wire rope isolators. The present research aims at predicting the nonlinear behaviour of coupled systems using a modal substructuring technique. This project focuses on the effects of nonlinear connections on the dynamics of an assembly in which the coupled subsystems can be considered as linear. A first experimental setup has been built at the University of Madison-Wisconsin to measure the dynamic response of the assembly and compare it to the numerical results. A particular focus is given to the realization of the connecting elements between the linear substructures, specifically designed to have a cubic nonlinear behaviour. First results show an interesting correlation between the mode shapes and a good accuracy in the resonance frequency at several excitation levels. However, since some problems arose during the measurement campaign, a new experimental setup needs to be designed. The interfaces between the substructures and the nonlinear elements should be carefully adjusted to avoid loosening of the bolts. Also, the design of the connecting element needs to be updated, for example by employing compliant mechanism, to consider frictionless connections and have a better control on the dynamics of the assembly. Eventually, it could be possible to use real connections and, by measuring their dynamic response in terms of NNMs, perform an experimental substructuring based only on experimental data.
