If you operate trucks, you probably follow the preventive maintenance (PM) frequency recommended by the vehicle manufacturer. Like we used to say: No one ever got fired for following the OEM manual.
But what do you do with weird, special or a one-of-a-kind piece of equipment?
MTBF
Let's say we have a complete and accurate computerized maintenance management system (CMMS) that is chock full of data. One of the basic CMMS metrics is MTBF (mean time before failure) - the average time that a component works without failure.
On the surface, it seems obvious that the failure rate should influence the PM frequency.
If the MTBF is 14 months, and we have a good deal of data, we can calculate the variability of this MTBF by looking at the standard deviation (SD), which for our purposes we will say is 1.5 months.
Statistics tells us that if the MTBF is 14 months and the SD is 1.5, then by the time 12.5 months (MTBF minus SD) have passed, 15.9 percent of the original components will have failed.
We also know from statistics that at 11 months (MTBF minus two times the SD), there will be about 1.9 percent failures.
PM Periods
If this is true then, we are golden. We can program our PM frequency by our allowable failure rate.
If things would just wear out - like tires, saw blades or sanding blocks - we could run PM from statistics alone.
If we could tolerate some failures, we might pick an annual PM interval. If the equipment is more critical or there are safety considerations, we might go to a quarterly PM. At a quarterly PM frequency we be could confident that we will catch any problems before failure.
Not Just Wear
Unfortunately, the mechanism of failure is only rarely a "wear out," aside from the common wear components mentioned previously. With those components, if the asset is used the same each day, we indeed could take the MTBF approach to PM frequency.
In reality, an asset will perform well, but then suddenly have something happen. Its performance deteriorates and finally, it fails.
The things that commonly happen to cause failure are dirt, looseness, lubrication problems, overloading and abuse. Once these issues start, failure will occur. The speed of the failure is related to the speed of operation of the equipment and the tolerances needed to operate.
By way of illustration, it will take a short time for a 16,000 rpm turbine to fail from the point of the bad event. On the other hand, a 50 rpm propeller on a cruise ship might take years to fail.
Event-To-Failure Interval
In the world of theory, a MTBF of 14 months means that on average a component that is kept clean, tight, lubricated and not abused will last about 14 months. But in the real world, we know that on average components experience adverse events that lead to their demise sooner.
What we don't know is how long it takes for a particular component to go from the adverse event to failure. It could be a year, in which case the annual PM might very well catch it.
But what if the event-to-failure interval is one month? Then, even a quarterly PM will most likely miss the symptoms. In fact, we would have to go to a two-week PM frequency to insure we would catch it.
So, a 14-month MTBF might require a two-week frequency or a one-year frequency and the 14 months doesn't really matter.
Best Practice
Here is the key takeaway: When it comes to PM, the best activity is the one that eliminates the bad event in the first place. Keeping the equipment clean, tight and properly lubricated, and keeping operators from abusing the equipment, will go a very long way to making the equipment more reliable.
Any inspection that catches a problem is too late because the damage is done. Truly effective preventive maintenance is where the critical deterioration doesn't even happen in the first place.
