Vibration response characterisation and fault-size estimation of spalled ball bearings

Research efforts have increased to investigate the ability to quantify localised bearing faults, ie spalls. These efforts revolve around extending the useful service life of the bearing after the detection of spalls. A number of studies have investigated a linear correlation between the size of spalls and three geometric points that may be recognised in the vibration response: the entry into the spall; the exit from the spall; and a third impact point between the first two. The time difference between these points, calculated using different signal processing techniques, has been widely exploited for quantifying spall size. Currently, there are two main challenges: the first is to enhance the entry point, which typically has weak excitation; the second is to distinguish the impact and the exit points investigated in the literature based on the spall size. However, for practical applications, there is no prior rough estimation of the fault size (ie small or large) and a method is needed for the interpretation of responses. This paper provides insights into the movement of the rolling element within the spall region and shows that the rolling element strongly strikes the bearing races at a minimum of two points. A new technique is then presented to quantify the spall and determine the inherent scaling factor without comparison to any reference data. The technique is based on evaluating two root-mean-square (RMS) energy envelopes, one for the vibration signal and one for a numerical differentiation of this signal. A geometric scaling factor is then used to give a generalised quantification for the small and large spalls. Serviceable estimations of spall size have been achieved for several seeded faults measured on two dissimilar test-rigs provided by the German Aerospace Centre (DLR) and the University of New South Wales (UNSW).