PSI - Issue 24

12

Vincenzo D’Addio et al. / Procedia Structural Integrity 24 (2019) 510–525 Author name / Structural Integrity Procedia 00 (2019) 000 – 000

521

maximum axis deflection and to visualize an animated 3D precession motion considering the orbits at different sections (Fig. 11).

Fig. 10. Radial displacement in time of a node situated on the rotor surface.

Fig. 11. 3D precession motion on the left and 2D orbit at a specific section on the right.

4.4. Rigid multibody with detailed bearing model All the models and the considerations done so far were related to the whole rotating system dynamics. In addition to those models, an additional model focused on a specific component, the ball bearing, was developed. In particular, a 3D rigid multibody model was set up using the multibody code MSC/Adams in order to estimate the overload produced by the presence of a defect located on the race. This software, in particular, can easily handle body contacts. This model can be useful to evaluate the vibrations introduced by a defect in a reasonable computing time and establish a correlation between their severity and the defect size as well as for diagnostic purposes. As computing time is concerned, compared to the previous simulation, this one took about 2 hours. The detailed CAD model of the bearing was imported in the program, including the real curvature radii of both races and the internal clearance (in particular the maximum value was assumed for simplicity). The most crucial part of this model implementation is the definition of the contacts between the components; the contact type chosen for this application are the following:  Frictional: rolling elements – rings (external and internal)  Frictionless: rolling elements – cage If the evaluation of efficiency or dissipations are not of interest, the precise values of the contact friction coefficients are not important but they must have non-zero values in order to correctly transfer the motion from the inner ring to the rolling elements; μ s = 0,2 and μ d = 0,1 were assumed as the static and dynamic friction coefficient, respectively (they are reasonable values for a dry contact but the real ones should be lower due to the