Reactive maintenance ultimately results in a plant closing unless it’s replaced with a better way. Traditionally, that way is preventive maintenance (PM). But PM is a system based on time. That usually is not even actual usage time, but calendar time. One consequence is you fix things that “don’t need fixing” while still reacting to failures this approach doesn’t prevent.
Predictive maintenance (PdM) has widely been touted as the solution to the gaps left by PM. But PdM is often done with a PM perspective. In a PM, you check various characteristics to see if they are in spec. This often translates PdM-wise into we’re going to check X equipment with our thermographic camera, insulation resistance tester, vibration analyzer, etc., and if it’s good we won’t need to fix it.
Predicting impending failures requires trending and analysis, not just collecting pass/fail data. For example, on a given cable the insulation resistance tests show a relatively slight decrease in resistance each time. But on the latest two tests, the decrease is significantly larger than the trend: schedule replacement.
