Model Reduction and Lumped Models for Jointed Structures

The jointed structures are widely spread but their behaviour is quite hard to model and to simulate. The talk deals with a decomposition in linear and non-linear substructures that allows to make the simulations easier. The authors show that the modal subspace is not optimal to reduce the order of the model. Another choice of reduction subspace is shown to be more optimal as it requires less state-variables and leads to a better correlation with Full Order Models. Finally, the authors show how it is possible to characterize experimentally the joints by loading them in the reduction subspace and to integrate an experimental model in the reduce order simulations of the structure.