Experimental Estimation of Journal Bearing Stiffness for Damage Detection in Large Hydrogenerators

Based on experimental pieces of evidence collected in a set of twenty healthy large hydrogenerators, this article shows that the operating conditions of the tilting pad journal bearings of these machines may have unpredictable and significant changes. This behavior prevents the theoretical determination of bearing stiffness and damping coefficients with an adequate accuracy and makes damage detection difficult. Considering that dynamic coefficients have similar sensitivity to damage and considering that it is easier to monitor bearing stiffness than bearing damping, this article discusses a method to estimate experimentally the effective stiffness coefficients of hydrogenerators journal bearings, using only the usually monitored vibrations, with damage detection purposes. Validated using vibration signals synthesized by a simplified mathematical model that simulates the dynamic behavior of large hydrogenerators, the method was applied to a journal bearing of a 700 MW hydrogenerator, using two different excitations, the generator rotor unbalance and the vortices formed in the turbine rotor when this machine operates at partial loads. The experimental bearing stiffnesses obtained using both excitations were similar, but they were also much lower than the theoretical predictions. The article briefly discusses the causes of these discrepancies, the method’s uncertainties, and the possible improvements in its application.