Even 30 years after the Nowlan & Heap RCM report, there still remains a lot of confusion over the principles and applications of P-F intervals, and Predictive Maintenance (PTIVE) in general.
This post is aimed at highlighting some of these modern myths, but space will probably prevent me from providing a detailed response. Feel free to carry on the conversation in the comments.
Myth #1: The six failure curves dictate whether PTIVE tasks can be applied.
Incorrect.
Whether a failure mode is likely to occur after time, or totally randomly, PTIVE tasks are based on the time between the warning signs of failure (Potential failure) and the Functional Failure.
This can still apply even if the failure curve itself is random. Bearings are the obvious example here although there are many others.
Bearings fail randomly as they are complex assets with no dominant failure modes. Yet once they have started to fail there is a definite time period between this point and the functional failure.
Myth #2: P-F Interval is the time period between the potential failure and catastrophic failure.
Incorrect.
P-F interval is the time period between the potential failure (P) and the functional failure (F), this difference is not a trivial matter.
When determining the effectiveness of a PTIVE task we always need to consider what happens at functional failure first.
This is the case all through RCM and is also a defining principle of Run-to-Failure (RTF) tasks. When we run to failure we run to functional failure, not to destruction.
Myth #3: Frequency of inspection is determined by the criticality of the asset.
Incorrect.
This is one of the great myths of PTIVE task assignment. In fact I regularly see tables for determining PTIVE task frequencies listing the method and the task frequency as determined by the criticality of the asset.
This is not only wrong but dangerous. The frequency of inspection is determined wholly and solely by the P-F Interval and the probability of detection (PoD). Nothing else. Period.
If you throw that out the window and start setting frequencies based on criticality alone then you run the considerable risk of either missing the warning signs of failure, or of over maintaining and wasting precious maintenance resources.
Criticality, or relative importance as determined by the consequences of failure, tells you if you should do something or not. That's all.
(Note: Criticality should not be confused with risk, in that it is not a measure so much of frequency as it is of solely the consequences of a failure)
Myth #4: PTIVE maintenance is not useful for time based failures.
Incorrect.
This is an extremely common line of thought. Although not often expressed it seems to underline the thinking of most people when the confront time based failures.
The immediate reaction is often to go straight for time based maintenance, because after all this is a time based failure mode and thats what these things are for, right?
In reality, time based failures can also be managed using the principles of PTIVE maintenance. But instead of measuring, recalling or recording the point of potential failure ("P") you get to establish it.
The case is pretty complex, but in certain types of large power transformers the degradation of the insulation is an indicator of both life position and remnant life.
This is a very specific case so please don't take these as figures for you to use.
With a transformer life of, say, 20 years. And a corresponding level of depolymerisation, the measure of degradation of the insulation, we can then establish which point we wish to use as the potential failure.
In this example we picked a point that represented the P-F interval of ten years, which when we determined the probability of detection using Furan analysis, then allowed us to determine the most effective testing frequency.
Predictive maintenance is almost always a more effective and applicable means of managing failures, regardless of whether the failure is time based or random.
This post is aimed at highlighting some of these modern myths, but space will probably prevent me from providing a detailed response. Feel free to carry on the conversation in the comments.
Myth #1: The six failure curves dictate whether PTIVE tasks can be applied.
Incorrect.
Whether a failure mode is likely to occur after time, or totally randomly, PTIVE tasks are based on the time between the warning signs of failure (Potential failure) and the Functional Failure.
This can still apply even if the failure curve itself is random. Bearings are the obvious example here although there are many others.
Bearings fail randomly as they are complex assets with no dominant failure modes. Yet once they have started to fail there is a definite time period between this point and the functional failure.
Myth #2: P-F Interval is the time period between the potential failure and catastrophic failure.
Incorrect.
P-F interval is the time period between the potential failure (P) and the functional failure (F), this difference is not a trivial matter.
When determining the effectiveness of a PTIVE task we always need to consider what happens at functional failure first.
This is the case all through RCM and is also a defining principle of Run-to-Failure (RTF) tasks. When we run to failure we run to functional failure, not to destruction.
Myth #3: Frequency of inspection is determined by the criticality of the asset.
Incorrect.
This is one of the great myths of PTIVE task assignment. In fact I regularly see tables for determining PTIVE task frequencies listing the method and the task frequency as determined by the criticality of the asset.
 |
| Things don't fail faster just because they are more important... obvious isn't it. |
This is not only wrong but dangerous. The frequency of inspection is determined wholly and solely by the P-F Interval and the probability of detection (PoD). Nothing else. Period.
If you throw that out the window and start setting frequencies based on criticality alone then you run the considerable risk of either missing the warning signs of failure, or of over maintaining and wasting precious maintenance resources.
Criticality, or relative importance as determined by the consequences of failure, tells you if you should do something or not. That's all.
(Note: Criticality should not be confused with risk, in that it is not a measure so much of frequency as it is of solely the consequences of a failure)
Myth #4: PTIVE maintenance is not useful for time based failures.
Incorrect.
This is an extremely common line of thought. Although not often expressed it seems to underline the thinking of most people when the confront time based failures.
The immediate reaction is often to go straight for time based maintenance, because after all this is a time based failure mode and thats what these things are for, right?
In reality, time based failures can also be managed using the principles of PTIVE maintenance. But instead of measuring, recalling or recording the point of potential failure ("P") you get to establish it.
 |
| Life based failures of transformer windings |
This is a very specific case so please don't take these as figures for you to use.
With a transformer life of, say, 20 years. And a corresponding level of depolymerisation, the measure of degradation of the insulation, we can then establish which point we wish to use as the potential failure.
Predictive maintenance is almost always a more effective and applicable means of managing failures, regardless of whether the failure is time based or random.