Definition
MTBF is the ratio of total accumulated operating time to the number of failures observed, for items that are repaired and return to operation. With a constant failure rate, MTBF is the inverse of that rate. The lower the failure rate, the higher the MTBF.
Where the constant rate comes from
The life of many components follows the so-called bathtub curve: early failures at the start, then a long stage with an approximately constant rate and, at the end, wear-out failures. MTBF makes sense above all in the middle stage. In that stretch failures are considered random, and the constant-rate model is reasonable.
What it does not mean
- It is not service life. Equipment with a long MTBF can have a short wear-out life, and vice versa. MTBF describes the frequency of random failures during normal operation, not when wear-out begins.
- It is not a guaranteed time. With a constant rate, a significant fraction of units fail before reaching a time equal to the MTBF. It is a statistical average of a population.
- It does not apply to a single unit as a prediction of its failure date.
How it is obtained
It is calculated from real field data, from accelerated life tests, or from prediction models that add up the failure rates of each component. Predicted values depend on assumptions about temperature, electrical stress and environment, so they can differ from those observed. Two MTBF figures can be compared only if they were obtained with the same method and under the same conditions.
How to interpret it
MTBF is useful for comparing designs, sizing spare parts and estimating availability together with mean time to repair. For a set of items in series, where the failure of any one stops the system, the failure rates add up, so the MTBF of the system is lower than that of any of its parts. With redundancy the opposite happens.
What an engineer sees in practice
In a fleet of equipment, the maintenance log makes it possible to calculate the actual MTBF: total operating time divided by the number of failures. If the value falls over time, there are signs of wear-out. If it is very different from the declared value, it is advisable to review the operating conditions before blaming the design.