An inspection is like the skin of an onion.
As you peel back each layer you will spend more money and time, but in return, you will get more information. For example:
- oil analysis – least expensive, least intrusive, no downtime – some information
- diesel trap – more cost, more intrusive, some downtime – more information
- tear down / overhaul – much more cost, much more intrusive, much more downtime – much more information
Our objective as maintenance engineers is to get valuable information from the least costly and least intrusive inspections. We do this by developing predictive models. Therefore, each time we perform an intrusive inspection, we must take full advantage of that informational opportunity. On the CMMS work order, we must reference our RCM knowledge base for each failure mode that we find. We must also record their event types (PF1 or FF2 or S3). This will make the work order data useful for analysis. Through analysis of a sample of life cycles ending in failure or suspension we may develop our predictive models.
Furthermore, we must also use the opportunity afforded by the work order to update our RCM knowledge base. If we discover failure modes during the inspection that we hadn’t anticipated in the initial RCM analysis we must add them to the RCM knowledge base. We should also correct and clarify the effects and other fields of the knowledge record. All this will be done quickly and inexpensively because all the information is fresh in our minds at the moment we close the work order.
The processes of referencing and updating the knowledge base should be integrated into the day-to-day maintenance routine and culture. By recording PFs we will obtain larger samples for analysis. (A sample is a set of failure mode life cycles.) The larger the sample:
- the lower the standard deviation in our RULEs (remaining useful life estimations)
- the more predictive confidence, and therefore:
- the lower the cost of maintenance
- the lower the downtime
- the higher the availability
A deliberate process of continuous improvement will thus improve our predictive models. With good models we will get the most information from the least costly and least intrusive inspections. We track the standard deviation in the remaining useful life estimation (RULE) in order to measure how well our CBM programs are performing. We must continually improve their performance wherever it is cost effective to do so.
1 Potential Failure: An imminent (confirmed) failure that has not yet resulted in major functional loss.
2 Functional Failure: A function has been lost.
3 Suspension: A renewal of a part or failure mode for any reason other than failure.
© 2011 – 2018, Murray Wiseman. All rights reserved.