MTTR, MTBF, MTTF and availability, sorted out

MTBF is how often it fails. MTTR is how long it takes to fix. MTTF is how long it lasts before it is replaced rather than repaired. Availability is what you get when you put the first two together. Four measures, four questions, and constant confusion between them.

One example, all four

A chilled water pump ran 8,000 hours last year and stopped 4 times. Repairs took 3, 5, 4 and 8 hours.

• •MTBF: 8,000 divided by 4 equals 2,000 operating hours between failures
• •MTTR: 20 hours of repair divided by 4 repairs equals 5 hours
• •Availability: 2,000 divided by 2,000 plus 5, which is 99.75 percent
• •MTTF does not apply here, because the pump is repaired rather than replaced. It applies to the seal inside it.

How availability falls out of the other two

Availability equals MTBF divided by MTBF plus MTTR. That single line explains why two very different maintenance strategies reach the same availability: you can fail rarely and take ages to fix it, or fail often and recover in minutes.

It also tells you which lever to pull. If MTBF is 2,000 hours and MTTR is 5, doubling MTBF barely moves availability. Halving MTTR barely moves it either. When availability is already high, the way to move it is to attack whichever number is furthest from where similar equipment sits.

Which to use when

• •Choosing between two machines to buy: MTBF, and ask over what population and period
• •Deciding what spares to hold: MTTR, because spares are the biggest lever on it
• •Planning replacement of components: MTTF
• •Reporting to somebody outside maintenance: availability, because it means something to them

The mistake that matters most

Reporting MTBF to a board. It is an internal engineering measure and it invites a target that gets hit by reclassifying small failures as adjustments. Report availability and the cost of unavailability. Keep MTBF and MTTR for the people who can act on them.