A single MTBF value is a snapshot. The real value comes from tracking it period after period for the same machine. A falling MTBF tells you the equipment is wearing or that a recent change hurt it. A rising MTBF tells you a fix worked. Plot it quarter by quarter and the picture is hard to miss.
The trend also guides preventive maintenance, the planned work done to replace or service parts before they fail. If a part tends to fail around a known number of operating hours, schedule its replacement before that point. A practical starting rule is to set the PM interval at roughly half to two-thirds of the observed MTBF for that failure mode, then adjust based on what the parts look like when they come out.
MTBF has a weakness you must respect: with very few failures it is unstable. One failure in 2,000 hours gives an MTBF of 2,000 hours, but a second failure next week cuts it in half. Do not set a long PM interval or declare a machine reliable from one or two events. Wait for more data, and say how many failures the number is based on.
Standard Procedure
- 1Calculate MTBF for each period and plot the values in order.
- 2Identify the dominant failure mode behind the trend from the log.
- 3Set a PM interval at about 50 to 70 percent of the current MTBF for that failure mode.
- 4Inspect replaced parts; lengthen the interval if parts show little wear, shorten it if they are close to failing.
- 5Note the number of failures behind each MTBF and treat low counts with caution.
Lesson Summary
- ✓A downward MTBF trend is a signal to shorten PM intervals and find the failure mode.
- ✓MTBF from one or two failures can change dramatically; label it and wait for more data.