Identification of bearing supports' force coefficients from rotor responses due to imbalances and impact loads

Experimental identification of fluid film bearing parameters is vital to validate predictions from often restrictive computational fluid film bearing models and is also promising for condition monitoring and troubleshooting. This dissertation presents the analytical bases of two procedures for bearing supports parameter identification with potential for in-situ implementation. Bearing support coefficients are derived from measurements of rotor responses to impact loads and due to calibrated imbalances in characteristic planes. Subsequent implementation of the procedures to measurements performed in a rigid massive rotor traversing two critical speeds provides force coefficients for a novel bearing support comprising a tilting pad bearing (TPJB ) in series with an integral squeeze film damper (SFD).; At a constant rotor speed, the first method requires impacts loads exerted along two lateral planes for identification of frequency-dependent force coefficients. Simulation numerical examples show the method is reliable with a reduced sensitivity to noise as the number of impacts increases (frequency averaging). In the experiments, an ad-hoc fixture delivers impacts to the rotor middle disk at speeds of 2,000 and 4,000 rpm. The experimentally identified force coefficients are in close agreement with predicted coefficients for the series support TPJB-SFD. In particular, damping coefficients are best identified around the system first natural frequency. Bearing stiffness are correctly identified in the low frequency range, but show a marked reduction at higher frequencies apparently due to inertial effects not accounted for in the model.; Measurements of rotor response to calibrated imbalances allow identification of speed-dependent force coefficients. The procedure requires a minimum of two different imbalance distributions for identification of force coefficients from the two bearing supports. The rotor responses show minimal cross-coupling effects, as also predicted by the computational analysis. Presently, the identification procedure disregards cross-coupled force coefficients thereby reducing the effect of false cross-correlations that cause ill-conditioning of the identification matrix. The procedure renders satisfactory force coefficients in the speed range enclosing the first critical speed, and the identified direct force coefficients are in accordance with those derived from the impact load excitations.