The availability rate is run time divided by planned production time. It is the first of OEE's three factors, and it answers one question: of the time you scheduled the machine to run, what fraction did it actually run? A line planned for 450 minutes that ran 387 has an availability rate of 86%.
The formula is simple arithmetic. What trips people up is the two numbers you feed it, where planned production time starts and stops, and which stops count against run time. Get those definitions straight and keep them stable, and the availability rate becomes the cleanest, most defensible number on your plant scorecard. Get them fuzzy and the rate drifts every shift for reasons nobody can point to. This guide walks the formula, works an example you can check by hand, and lists the edge cases that quietly corrupt the number.
What is the availability rate formula?
Availability rate = Run Time ÷ Planned Production Time. Both are measured in the same time unit, minutes, usually, over the same period. The result is a ratio between 0 and 1, expressed as a percentage. It can never exceed 100%: run time is carved out of planned production time, so the numerator can at most equal the denominator.
Two definitions do all the work:
- Planned production time is the time you intended to make product. It is the shift, minus planned non-production events you decided in advance not to run: breaks, scheduled meetings, planned maintenance windows.
- Run time is planned production time minus every unplanned stop and every changeover. It is the time the machine was actually producing.
The gap between the two is downtime. That is the whole story: availability rate is one minus (downtime ÷ planned production time). It sits at the front of the OEE calculation ahead of the performance rate (speed) and the quality rate (defects), and it is the factor most plants can move first because the biggest losses usually live here.
What counts as planned production time?
Planned production time is scheduled time minus the non-production events you deliberately built into the plan. Start with the shift length, then subtract breaks, scheduled meetings, planned changeover-free maintenance windows, and any block you decided in advance not to run. What remains is the denominator.
The critical discipline is not which rule you pick, it is picking one and never varying it. If Monday's planned production time excludes the 30-minute lunch and Tuesday's includes it, availability moves for a reason that has nothing to do with the machine. Write the exclusion rules down once, post them, and defend them. One quiet way availability gets gamed is shrinking planned production time so the same downtime looks smaller against a smaller denominator; the tell is availability rising while units shipped falls.
Note what planned production time ignores: shifts you chose not to schedule. A line that runs one shift a day has a denominator of one shift, not 24 hours. That deliberate blindness is why availability can look healthy while the plant's capacity utilization is low, the two metrics answer different questions, and TEEP is the one that bridges them.
What counts as run time, and which stops subtract?
Run time is planned production time minus every stop that happened during it. If the machine was scheduled to produce and wasn't producing, that time comes out, no exceptions for how the stop was labeled. The stops that subtract include:
- Breakdowns the machine failed and stopped.
- Changeovers and setups the line was planned to produce and instead was being converted. These count. Excluding them is the single most common way to inflate availability, and it hides one of the six big losses.
- Material and labor starvation no components, no operator, upstream blockage. The machine could run but had nothing to run on.
- Unplanned minor stops longer than your logging threshold a jam that takes six minutes to clear is downtime.
The gray zone is the very short stop. Micro-stops of a minute or two are usually too brief to log by hand, so by convention they fall into the performance rate rather than availability. That is fine as long as you know it is happening; the risk is a manual log that misses two-minute stops and quietly parks the loss where the crew can't see it. Automatic counting from machine signals closes that gap. For a full breakdown of where uptime actually goes, see availability loss analysis.
How do you calculate the availability rate step by step?
Collect two totals for the period, then divide. The procedure:
- Set the period. One shift, one day, one week, whatever you will report consistently. Availability rate is only comparable across periods of the same definition.
- Compute planned production time. Take the scheduled length and subtract your written planned non-production exclusions (breaks, meetings, planned maintenance). This is the denominator.
- Total all downtime minutes. Add every unplanned stop plus every changeover and setup during the period. Pull these from the downtime log not from memory, end-of-shift recall reliably undercounts short stops.
- Subtract to get run time. Run time = planned production time − total downtime.
- Divide and express as a percentage. Availability rate = run time ÷ planned production time.
- Sanity-check against the log. If availability came out at 86% but the downtime log only adds to 20 minutes on a 450-minute plan, a stop went unrecorded. The two must reconcile.
A worked example, checkable by hand
The numbers below are hypothetical a made-up packaging line, chosen so the arithmetic is easy to verify.
| Input | Value |
|---|---|
| Shift length | 480 min |
| Planned breaks | 30 min |
| Planned production time | 450 min |
| Breakdown | 28 min |
| Changeover | 25 min |
| Logged minor stops | 10 min |
| Total downtime | 63 min |
| Run time | 387 min |
Availability rate = 387 ÷ 450 = 86.0%. Read the other way: 63 minutes of downtime on a 450-minute plan is a 14.0% availability loss. That single number tells you nearly a seventh of the scheduled shift produced nothing, and because changeover was 25 of the 63 minutes, setup reduction is the obvious first lever. Punch the same inputs into the free OEE calculator and the availability factor should read 86.0%.
What edge cases trip people up?
The formula is trivial; the definitions are where availability goes wrong. The recurring traps:
- Changeovers excluded. Treating setup as "not really downtime" is the biggest inflater. If the line was planned to produce, changeover time counts.
- A moving denominator. Changing what planned production time excludes from shift to shift makes availability wander. Fix the rules once.
- Confusing scheduled and calendar time. Availability uses planned production time, not 24/7 calendar time. Judging a one-shift line against 24 hours is a TEEP question, not an availability one.
- Availability over 100%. This is impossible and always signals a data error, usually run time booked against the wrong, larger denominator.
- Startup and shutdown ramp. Warm-up and clean-down at shift edges are real stops; decide once whether they are planned exclusions or downtime, then stay consistent.
- The reliability-engineering meaning. Maintenance teams use "availability" as uptime ÷ (uptime + downtime) over calendar time, tied to MTBF and MTTR. That asset availability calculation is related but not identical to OEE's availability rate; see also availability vs reliability.
What is a good availability rate?
The commonly cited world-class target for the availability factor is 90%, which pairs with 95% performance and 99% quality to give the familiar 85% OEE benchmark. Treat those as reference points, not gospel, the 90% figure traces to Seiichi Nakajima's original TPM work and is not an audited industry statistic. A stable 82% that you trust and that is trending up beats a flattering 90% built on excluded changeovers. For the full picture of realistic targets, see what counts as a good OEE score.
For macro scale, the U.S. Federal Reserve's G.17 release put manufacturing capacity utilization at 75.7% in May 2026 about 2.5 points below its 1972–2025 average. That is a broader measure than availability, it counts unscheduled shifts as unused, but it is a plain signal that real plants run well under theoretical maximums by any method. Do not read a public utilization figure as an availability target; read it as a reminder to benchmark your line against its own history rather than against a number that sounds impressive.
The number is only as honest as the log behind it
Availability rate has one input that decides everything: the downtime total. When that total comes from an operator's end-of-shift memory, it undercounts, the 40-minute breakdown gets remembered, the string of short jams does not, and availability reads high for the wrong reason. When it comes from machine signals captured as stops happen, with the operator adding the reason code, the number holds up. That is the practical case for measuring availability at the source rather than reconstructing it later. Harmony reads run state directly from PLCs and sensors, so availability reflects what the equipment did, not what anyone recalled (see the platform or the CLS field results). Once availability is trustworthy, the next move is decomposition: split the loss into breakdown, setup, and minor-stop buckets with availability loss analysis then run the whole thing through the OEE calculator.