Key takeaways
Short answer: Equipment availability is a single machine uptime against its scheduled time. Plant availability is whether the whole plant can actually make sellable product. The two diverge because availability does not multiply cleanly across a line, buffers absorb some losses while the bottleneck amplifies others.
A line where every machine is 95% available can still have low plant availability if those stoppages land on the constraint. See also bottleneck vs constraint .
Equipment availability is asset-level: run time divided by scheduled time for one machine. It feeds the Availability term of that asset OEE and is the right lens for maintenance, because it pinpoints the single weakest machine. It says nothing, on its own, about whether the plant met its number.
Plant availability is system-level: can the plant, as a connected whole, produce sellable output? It accounts for how machines depend on one another, a stoppage upstream starves downstream, a stoppage at the bottleneck caps everything. It is the number closest to the question executives actually ask.
A four-station line runs at 95% availability per machine. Naively that sounds like 95% plant availability, but it is not.
If the bottleneck (station 3) is one of the 95%-available machines and has no buffer in front of it, every minute it is down is a minute the whole line loses, while the other three machines downtime is partly hidden by buffers.
The arithmetic of dependency means the line might deliver only 82% of sellable output despite every machine looking healthy in isolation.
Availability does not multiply intuitively across connected steps. Buffers absorb some downtime so it never reaches the customer; the bottleneck amplifies downtime so a small stop there costs the whole plant. A non-bottleneck at 90% may cost nothing, while the constraint at 98% sets the ceiling. This is why a dashboard of green per-machine numbers can sit above a red plant number.
Use equipment availability to find and fix the weak asset; use plant availability to know whether that fix actually moved sellable output. Improving a non-constraint machine raises its number and changes nothing downstream, a classic way to feel busy while output stays flat. Always check the fix against the constraint.
1. Averaging machine availability to get plant availability. The average ignores dependency and the bottleneck.
2. Improving non-constraint assets. Effort and cost with zero throughput gain.
3. Ignoring buffers. Buffer changes can shift plant availability without any machine changing.
4. Two disconnected numbers. The executive view and the maintenance view must reconcile through the constraint.
OEE is computed per asset; rolling it to plant level requires weighting by where the constraint sits. That is precisely why plant availability and equipment availability tell different stories, and why constraint-aware roll-up is what makes a plant-level OEE number meaningful.
Fabrico measures availability per asset and helps you see where losses actually cost throughput, so improvement targets the constraint rather than the easiest machine. Book a demo to see asset-level and line-level views together.
Effectively yes, with a single asset the distinction collapses. It only matters across connected steps.
Plant availability, it maps to sellable output and revenue.
Equipment availability, it finds the weakest asset to work on.
Only if it is the constraint or feeds it; improving a non-bottleneck adds cost without throughput.
Weight by the bottleneck and account for buffers, a simple average will mislead.
Login
Blog