By definition a Non-Rejuvinating Event (NRE) is an action that systematically affects condition monitoring value levels, but does not add “life” to (i.e. reduce the physical age of) the asset.
For example, if immediately after greasing the bearing, there is a significant step drop in detected vibration levels, then the greasing activity should be accounted for when building a predictive analysis (that correlates failure probability with high vibration). Otherwise the resulting model could be mislead into thinking that the bearing condition has somehow improved. However, no recorded maintenance or service activity would explain that improvement.
In other words, the analysis will assume that the process of damage accumulation in the bearing “magically” reversed itself at the moment in question. This “counter-trend”, if it occurs frequently, will introduce scatter in the relationship found between vibration and failure probability. If, on the other hand, the model building procedure is made aware that a greasing activity preceded a reduction in vibration, it will disregard any “artificially” lowered readings when generating an age-vibration-reliability relationship. Identifying non-rejuvenating events is a refinement in the model building process. It may improve the predictions of the model by reducing the scatter (standard deviation) in remaining useful life estimation (RULE).
LRCM facilitates the identification and accounting of a work order activity (in the predictive model) as a NRE where a work order action (e.g. a calibration of a sensor) having an “artificial” impact on CBM data, but no effect on the wear state of the equipment.
Note that activities such as greasing, cleaning, filter change can themselves be significant “external” variables included in the failure prediction model. For example an electrical high power transformer will be more reliable it it is cleaned periodically. In this case the work order itself becomes a significant CBM variable with a binary value of 0 or 1.
© 2011 – 2014, Murray Wiseman. All rights reserved.