MTBF, MTTR, and availability are indicators used to understand how often repairable assets fail, how long they take to return to an available state, and what share of time they remain capable of performing the required function. Together, they help separate two problems that are often confused: failing too often and taking too long to restore service.<\/p>\n\n\n\n
MTBF \u2014 Mean Time Between Failures \u2014 represents the average time between failures within a defined population or history. MTTR, in ABNT NBR 5462:1994, corresponds to mean restoration time: the expected interval between the occurrence of a failure and returning the item to an available state. Availability, in turn, expresses the ability of the item to be in a condition to perform a required function, considering reliability, maintainability, and maintenance support.<\/p>\n\n\n\n
These concepts appear simple, but they are often used incorrectly. MTBF is not service life. MTTR should not automatically be confused with actual repair time. And operational availability cannot be calculated reliably without knowing which times are included in the numerator and denominator.<\/p>\n\n\n\n
For Maintenance Engineering, the value of these indicators is not in producing an attractive dashboard number. It lies in diagnosing whether performance loss results from recurring failures, poor maintainability, logistical delays, lack of spares, design deficiencies, insufficient redundancy, or inadequate data governance.<\/p>\n\n\n\n
MTBF is mainly used for repairable items and represents the average interval between failures considered relevant within a period or observation set. In operations, a common estimate is accumulated operating time divided by the number of observed relevant failures, provided the failure criterion, asset boundary, and period are consistent.<\/p>\n\n\n\n
This caveat is decisive. If an organization changes what it considers a failure midway through the historical series, the indicator loses comparability. If planned shutdowns are included as failures in one month and excluded in the next, MTBF stops representing technical behavior and starts reflecting a registration rule.<\/p>\n\n\n\n
NBR 5462 distinguishes failure, fault state, operating time, time between failures, and relevant failure. This vocabulary is useful because it forces the organization to explicitly define which event enters the calculation.<\/p>\n\n\n\n
An operational way to estimate MTBF is:<\/p>\n\n\n\n
Estimated MTBF = accumulated operating time \/ number of relevant failures<\/strong><\/p>\n\n\n\n This expression does not turn any average into a universal statistical parameter. It summarizes observed history under defined criteria. Reliability forecasting, especially for small populations, variable regimes, or non-stationary behavior, may require more robust statistical methods.<\/p>\n\n\n\n An MTBF of 20,000 hours does not mean the equipment \u201cwill last 20,000 hours.\u201d That interpretation confuses the average between events in repairable items with time to failure or service life.<\/p>\n\n\n\n NBR 5462 treats service life as the interval from first availability until the failure intensity becomes unacceptable or the item is considered non-recoverable after a fault. It also distinguishes time to failure and time between failures. These are different concepts.<\/p>\n\n\n\n A repairable system may operate for many years, accumulate several failures, and have MTBF calculated over the intervals between them. Service life concerns the item’s lifecycle, not merely the average recurrence of events.<\/p>\n\n\n\n This distinction also matters when comparing suppliers. A catalog MTBF value may be calculated by model or testing under specific conditions and should not be interpreted as a guarantee of field service life.<\/p>\n\n\n\n The calculation starts by defining the boundary. It is necessary to know whether the indicator belongs to a component, equipment item, line, system, or fleet. Mixing levels produces results of little value.<\/p>\n\n\n\n Consider a repairable asset that accumulated 4,800 operating hours during the analyzed period and experienced 4 relevant failures according to the organization’s criterion. The observed MTBF would be:<\/p>\n\n\n\n MTBF = 4,800 h \/ 4 = 1,200 h<\/strong><\/p>\n\n\n\n This result means that, in that data set, the average operating time per relevant failure was 1,200 hours. It does not mean that the next failure will occur exactly after 1,200 hours.<\/p>\n\n\n\n Periods without observation, regime changes, replacements, and modifications also need to be recorded. If the asset was significantly redesigned, mixing pre- and post-change data may hide the effect of the change.<\/p>\n\n\n\nMTBF is not service life<\/h2>\n\n\n\n
How to calculate MTBF with field data<\/h2>\n\n\n\n