Asset availability is the share of required time that an asset is actually able to run, calculated as availability = uptime ÷ (uptime + downtime). Reliability engineers split it two ways: inherent availability, MTBF ÷ (MTBF + MTTR), which counts only active repair time, and operational availability, MTBF ÷ (MTBF + MDT), which counts every hour of downtime including waiting and delays.
Availability sounds simple until you ask what counts as downtime, and that is where most confusion lives. Does time spent waiting for a spare part count against the machine? Does a planned maintenance shutdown? The answer depends on which availability you mean, and picking the wrong one either flatters your numbers or makes a healthy machine look sick. This guide walks through the base calculation, the crucial split between inherent and operational availability, worked examples, and how availability relates to the "A" in OEE.
What is asset availability?
Asset availability is the probability, or the measured proportion, that an asset is in a state to perform its required function when you need it. Over any period, it is the uptime divided by the total time the asset was required to be available, uptime plus downtime. An availability of 95% means the asset was ready to run 95% of the time it was supposed to be.
Availability is one of the three pillars of equipment performance, alongside reliability (how often it fails) and maintainability (how fast it recovers). Those two feed availability directly: a machine that fails rarely and recovers fast is highly available; frequent failures or slow repairs drag it down. That relationship is the whole reason availability is worth measuring, it is the single number where reliability and maintainability meet.
How do you calculate asset availability?
At its simplest, add up the uptime and the downtime over a period and divide. Suppose a packaging line was scheduled to run 600 hours in a month and lost 42 hours to breakdowns and repairs. Uptime is 558 hours, downtime is 42 hours, and availability is 558 ÷ 600 = 0.93, or 93%.
The same number can be built from failure and repair statistics instead of a raw log, which is where MTBF and MTTR come in. If a machine averages 200 operating hours between failures (its MTBF) and takes on average 4 hours to repair (its MTTR), each failure-and-repair cycle is 204 hours, of which 200 are up. Availability is 200 ÷ 204 = 0.980, or 98%. Both routes measure the same thing; they differ only in what you feed them and, critically, in what you count as downtime.
What is the difference between inherent and operational availability?
The difference is what counts as downtime. Inherent availability counts only the time the asset is actively being repaired. Operational availability counts all the downtime the asset actually experiences, the repair plus the waiting, the parts delays, the administrative time, and often planned maintenance. That is why the two numbers can differ sharply for the same machine.
| Inherent availability (Ai) | Operational availability (Ao) | |
|---|---|---|
| Formula | MTBF ÷ (MTBF + MTTR) | MTBF ÷ (MTBF + MDT) |
| Downtime counted | Active corrective repair only | All downtime: repair, waiting, logistics, admin, often PM |
| Assumes | An ideal support environment, parts and people on hand | The real support environment you actually have |
| Best for | Comparing machine design, spec sheets, warranty claims | Planning, staffing, and real production decisions |
| Tends to be | Higher, the flattering number | Lower, the honest number |
Here MDT is mean downtime, the average total time an asset is down per failure, including everything, not just the wrench time. Inherent availability answers "how good is this machine in a perfect shop?" and is what vendors quote and bodies like the U.S. Defense Acquisition University define, the probability an item operates satisfactorily in an ideal support environment, excluding logistics and administrative delay. Operational availability answers "how good is this machine in my plant, with my spares and my crew?" and is the one that should drive real decisions. If your Ao is far below your Ai, the gap is not the machine, it is your parts, planning, and response.
How do you calculate operational availability step by step?
Operational availability is the one worth computing carefully, because it is the honest one. Work it through like this.
- Define the period and the required time. Pick a window, a month, a quarter, and the hours the asset was required to be available in it. This is your denominator base.
- Count the failures. Total the number of functional failures in the period. This drives MTBF.
- Calculate MTBF. Divide total operating (up) time by the number of failures. If the asset ran 558 hours across 3 failures, MTBF is 186 hours.
- Total all the downtime, not just repair. For each failure, add the response time, the wait for parts and permits, the actual repair, and the return-to-service checks. Sum these across the period.
- Calculate MDT. Divide total downtime by the number of failures to get mean downtime per failure. If 42 hours of downtime came from 3 failures, MDT is 14 hours.
- Compute Ao. Apply MTBF ÷ (MTBF + MDT): 186 ÷ (186 + 14) = 0.93, or 93%. Compare it to inherent availability using MTTR alone to see how much your support system is costing you.
The discipline that makes this possible is honest, timestamped downtime capture with reason codes, the same foundation behind machine downtime tracking. Without it, MDT is a guess, and a guessed availability number persuades no one.
What downtime counts as unavailability?
The number you report is only as good as your definition of downtime, so agree on it before you calculate anything. The safe rule is to count every hour the asset was required but could not perform its function, then classify those hours so you know where they came from. What trips people up is the borderline categories.
- Corrective repair always counts, the machine is down because it failed.
- Waiting and logistics (parts, permits, a free technician) count in operational availability and are excluded from inherent availability. This is usually the largest recoverable chunk.
- Planned maintenance counts in operational availability if you frame availability against required time; inherent availability excludes it. Pick one convention and hold it.
- Idling for no demand the asset was not required, should be excluded from the required-time base entirely, or it will understate a perfectly healthy machine.
- Changeover and setup are typically counted as availability loss inside OEE but may be treated separately in a pure reliability calculation.
Whatever you decide, write it down and apply it identically to every asset and every period. A consistent, slightly imperfect definition beats a theoretically perfect one that changes month to month, because availability is most useful as a trend you can trust.
How is availability different from OEE?
Availability is one of the three factors in Overall Equipment Effectiveness, but it is not the whole story. OEE multiplies availability by performance (are you running at rated speed?) and quality (are you making good parts?). A machine can be highly available and still have poor OEE if it runs slow or scraps product.
There is also a subtle scope difference. The availability factor inside OEE is usually measured against planned production time and counts unplanned stops and changeovers as losses, whereas the reliability-engineering availability above can be framed around required time and total downtime including logistics. Both are legitimate; just be clear which you are quoting. To see how the availability factor rolls into the full metric, work an example in our OEE calculator and read how the pieces fit in OEE calculation and the six big losses.
What is a good availability target?
There is no universal target, a good availability figure depends entirely on the asset's criticality and the cost of it being down. A redundant utility with a warm spare might be perfectly acceptable at 90%, while a single-path bottleneck that stops the whole plant may need 98% or better to protect the schedule. The right target is set asset by asset, which is exactly what an asset criticality ranking is for.
Chasing a high availability number on a low-criticality asset wastes maintenance hours that a critical machine needs. That is why availability targets belong downstream of criticality: rank the assets first, then hold the critical few to tight availability and let the rest run to sensible, cheaper standards. This is the same portfolio thinking behind the equipment reliability maturity ladder.
How do you improve asset availability?
Availability has only two levers, and they map directly to the formula: raise the time between failures, or cut the time lost per failure.
- Raise MTBF (fail less often). Better preventive and condition-based maintenance root-cause elimination of repeat failures, correct lubrication, and operator care all extend the run between breakdowns.
- Cut MDT (recover faster). This is usually the faster win. Stock the right spares, pre-stage kits for known jobs, write clear troubleshooting and repair procedures, and shorten the administrative and waiting time that inflates MDT well beyond the actual wrench time. Much of the gap between inherent and operational availability is recoverable here.
Notice that cutting MDT is largely an information and logistics problem, not a mechanical one. The hours lost waiting for the right part, the right person, or the right procedure are exactly the hours a connected operation removes. That is the layer machine-monitoring platforms like Harmony provide, pulling machine signals, maintenance records, spare-parts status, and downtime reasons into one operational view, so a stop becomes a work order routed to the right person with the right parts already identified, and no rip-and-replace of the systems you already run. See how the platform works or read the CLS case study.