Loading time and operating time are two adjacent buckets in the OEE time model. Loading time is the planned production time a machine is scheduled to run; operating time is what is left after downtime is subtracted. The gap between them is availability loss, and availability equals operating time ÷ loading time.
These are not interchangeable words for "run time." They are specific rungs on the Nakajima time ladder that turns a plant's total calendar hours into the fully productive hours OEE rewards. Confuse the two and your availability number is wrong, because availability is defined as the ratio between them. This post walks the whole cascade, pins down where loading time and operating time sit, shows the arithmetic with real numbers, and explains why the choice of what counts as "planned" quietly decides your OEE.
What is the difference between loading time and operating time?
Loading time is the time a machine is scheduled and expected to produce; operating time is the portion of loading time when the machine was actually running. Loading time is total calendar time minus planned shutdowns, no-demand hours, planned maintenance, breaks, meetings. Operating time is loading time minus unplanned downtime, breakdowns, setups, changeovers, material waits.
The distinction is the whole basis of the availability metric. Availability = operating time ÷ loading time. If a shift schedules 8 hours, plans a 30-minute meeting and two 15-minute breaks, then loses 45 minutes to a breakdown and 30 minutes to a changeover, the loading time is 7 hours and the operating time is 5.75 hours. Availability is 5.75 ÷ 7 = 82 percent. Notice the planned hour never enters the availability penalty; only the unplanned 75 minutes does. That is the point of separating the two buckets.
Where do loading time and operating time sit in the OEE time model?
They sit as the second and third rungs of the six-loss time cascade, right where availability is measured. The model developed by Seiichi Nakajima for Total Productive Maintenance takes total calendar time and narrows it in four steps, each removing one family of loss.
- All time every hour on the calendar, 24/7.
- Loading time (planned production time), all time minus schedule loss: hours with no demand, planned maintenance, and unstaffed shifts. This is the OEE denominator.
- Operating time loading time minus availability loss: breakdowns, setups, and changeovers. Availability = operating time ÷ loading time.
- Net operating time operating time minus performance loss: minor stops and reduced speed. Performance = net operating time ÷ operating time.
- Fully productive time net operating time minus quality loss: scrap, rework, and startup rejects. Quality = fully productive time ÷ net operating time.
OEE is the last bucket divided by the first meaningful one: fully productive time ÷ loading time, which is algebraically the same as availability × performance × quality. That is why loading time is the anchor of the whole OEE calculation. Move the loading-time line and every downstream percentage moves with it.
By the numbers. The OEE time model traces to Seiichi Nakajima's work on Total Productive Maintenance at the Japan Institute of Plant Maintenance, which frames the losses as the six big losses that separate calendar time from fully productive time (Overall equipment effectiveness). Vorne's OEE standard defines Planned Production Time (loading time) as the OEE denominator and availability as Run Time divided by Planned Production Time (OEE.com, OEE factors), with the six big losses mapped to the same cascade (OEE.com, six big losses).
How do you calculate operating time from loading time?
You subtract every minute of unplanned downtime from loading time. The routine below keeps the two buckets clean so availability comes out honest.
- Fix the calendar window. Pick the shift or day and write down its total clock time. For a single 8-hour shift, all time is 480 minutes.
- Subtract schedule loss to get loading time. Remove planned, unstaffed, or no-demand time: breaks, planned meetings, scheduled maintenance, and any hours the line was not expected to run. What remains is loading time, the OEE denominator.
- List the downtime events. Capture every unplanned stop during loading time: breakdowns, setups, changeovers, material and labor waits, and any stop long enough to count as a stop rather than a minor stall.
- Subtract downtime to get operating time. Loading time minus the summed downtime is operating time, the minutes the machine was actually running.
- Divide for availability. Availability = operating time ÷ loading time. Keep the answer with the underlying minutes so anyone can audit which stops were called planned versus unplanned.
The trap is step two. Whether a changeover counts as planned schedule loss or unplanned downtime changes both loading time and availability, and there is no universal rule, only a plant standard you apply consistently. Setups are usually treated as availability loss because SMED can shrink them; a genuinely fixed sanitation window is usually schedule loss. Pick a convention, write it down, and never move the line to flatter a number.
What does loading time versus operating time look like in numbers?
A worked shift makes the two buckets concrete. The table walks an 8-hour shift down the cascade so you can see exactly where each rung lands and which loss caused the drop.
| Rung | Minutes | Loss removed at this step |
|---|---|---|
| All time (one shift) | 480 | |
| Loading time | 420 | 60 min planned: breaks + meeting |
| Operating time | 345 | 75 min downtime: breakdown + changeover |
| Net operating time | 318 | 27 min minor stops + speed loss |
| Fully productive time | 300 | 18 min scrap + rework |
From this one shift, availability = 345 ÷ 420 = 82 percent, performance = 318 ÷ 345 = 92 percent, quality = 300 ÷ 318 = 94 percent, and OEE = 300 ÷ 420 = 71 percent. Every one of those ratios is anchored to loading time. If you had wrongly folded the 60 planned minutes into the denominator as if the line should have run 480, availability would read 345 ÷ 480 = 72 percent and OEE would collapse to 63 percent, punishing the line for hours it was never asked to run.
Why does the loading-time boundary decide your OEE?
Because loading time is the denominator of both availability and OEE, so every judgment call about what counts as planned time shifts the score. A plant that classifies long changeovers as planned schedule loss will report higher availability than an identical plant that calls them downtime, even though the machines behave the same. Neither is lying; they drew the loading-time line in different places.
This is also why OEE and TEEP differ: TEEP measures fully productive time against all calendar time instead of loading time, exposing the schedule loss that OEE deliberately excludes. If you want to know how much of the machine you actually run, watch capacity utilization and utilization rate, which use calendar or scheduled time as the base rather than planned production time. Keep the loading-time convention fixed across lines and shifts, or benchmarking becomes meaningless. The six big losses only line up cleanly when everyone agrees which rung each loss falls on.
What are the common mistakes with loading time and operating time?
Most loading-time and operating-time errors come from putting a loss on the wrong rung or from letting the boundary drift between shifts. Each one bends availability without anyone touching the machine, which is why they survive so long: the number looks plausible.
- Counting no-demand hours in the denominator. If the line had no orders, those hours belong to schedule loss, not loading time. Leaving them in the denominator drags availability down for idleness the machine could not control, and confuses OEE with utilization.
- Hiding downtime as planned. The opposite move: reclassifying a chronic breakdown as "planned maintenance" so it leaves loading time. Availability climbs, but the loss is still there, now invisible. This is the most common way a plant fools itself.
- Letting short stops fall through. A two-minute stall that no one logs stays inside operating time even though the machine was down. Enough of them and operating time is quietly overstated, which is really a performance-loss problem masquerading as good availability.
- Inconsistent changeover treatment. One shift calls a changeover downtime, the next calls it planned. Availability swings shift to shift for a reason that has nothing to do with the equipment.
The cure is a written classification standard applied identically everywhere, plus a downtime record accurate to the minute. When the rungs are stable, the six-loss cascade lines up and cross-line benchmarking finally means something.
How does automatic downtime capture keep the two buckets honest?
The hard part is not the arithmetic; it is knowing exactly when the machine stopped, for how long, and whether the stop was planned. When operators reconstruct downtime from memory at end of shift, short stops vanish and the operating-time bucket is quietly inflated, which flatters availability. Direct machine monitoring timestamps every state change, so the line between loading time and operating time is drawn by the machine, not by recollection. The same feed distinguishes planned from unplanned stops when reason codes are attached at the station.
That is where a connected floor earns its keep: availability computed from source-level state changes, not estimated from a clipboard. See how the pieces fit in machine downtime tracking and the timing metrics in mean time metrics for production run the math on your own line with the OEE calculator and see a live example in the CLS case study.