Nonlinear dynamic model upgrading and updating using sine-sweep vibration data
Dynamic modelling is a core activity in mechanical engineering towards supporting system and control design. It is typically carried out in a computational environment, involving idealizing assumptions of diverse kinds. The most notable assumption commonly adopted in the field is the excitation-to-response linearity of the mechanical vibrations. This common practice contrasts with the day-to-day experience of test engineers, who are ever more confronted with nonlinearities when dynamically testing modern mechanical structures. A nonlinear behaviour may result from various physical mechanisms, the most recurrent being the existence of dynamic boundary conditions in the direct vicinity of structural joints and interfaces. In this paper, a coherent set of techniques is described to locate, characterize and model nonlinearities using sine-sweep vibration data, with the purpose of upgrading a pre-existing linear numerical model into a reliable nonlinear one. A constant thread in this set of techniques is the analysis of sensor measurements in phase space. The presented tools are illustrated using acceleration time histories measured at multiple forcing amplitudes on a full-scale F-16 aircraft. Nonlinear stiffness and damping elements, modelling a loosened attachment at one of the aircraft wing tips, are identified and introduced in a linear finite-element model, leading to accurate response predictions in a strongly nonlinear regime of motion.