Empirical methods arise from the concept that to prevent a failure from occurring unexpectedly, the warning condition must be inspected at an interval that is less than the period of decay (the PF Interval). What happens in practice is that analysts estimate PF intervals based on their experience of actual or similar equipment. The task interval is determined by applying a safety factor of two; or perhaps three, if the consequences of the failure are very high.

The reasons why empirical methods are widely used and effective are as follows:

• It is far more simple and practical to ask the maintenance fitter "How long will that bearing last once it starts to make a noise?" and inspecting the bearing for noise inside that interval, than collecting years of failure history on that bearing, deciding all the various costs, determining the shape of failure pattern.
• Shop floor maintenance personnel are typically not trained in statistics or complex mathematics and statisticians lack the understanding of the plant. Experience shows that a far better result is produced if the shop floor person is provided with a sensible but simple method rather than trying to train the statistician in the nuances of the plant.
• Empirical methods yield much faster results than statistical methods.
• Empirical methods can be easily applied without using computers whereas most statistical methods require software packages to run them.
• Statistical methods often produce ridiculous outputs particularly if the input data is suspect.

In conclusion, it remains a puzzle why engineers persist with statistical methods where the empirical approaches are far quicker and much more reliable.

Figure 1  Setting Condition Monitoring Intervals