On a bakery line, measure OEE at the constraint, which is often the oven, with an ideal rate for each product based on bake time and oven load. Also log the dough that a stop ruins upstream, because a short stop can waste product already mixed and proofing, and at the oven that loss looks like an unexplained gap.
What makes a bakery line different
Most lines can pause, and the work simply waits. A yeast-dough line cannot, because dough keeps fermenting whether the line is ready for it or not.
So a stop in one place can spoil product somewhere else, some time later. A stop log taken alone misses this, and your OEE method has to catch it.
For the wider picture across food plants, see OEE and downtime in food and beverage. This page stays with the bakery line itself, from the mixer to the cooler.
Where to measure: find the pace-setter
On many bread and roll lines the oven sets the pace, since it holds a fixed load for a fixed bake time. Check yours, because a slow cooler or packer can also be the limit, and bottleneck analysis shows how.
A batch process. Watch batch timing, since dough made too early waits and goes off spec.
Short stops and weight rejects. These run as discrete machines.
Fixed dwell, so a full proofer cannot wait for long. Watch for over-proof loss.
Usually the pace-setter. Bake time and load set the ideal rate, and empty belt is lost capacity.
A long stop here can back up to the oven, which cannot simply hold a bake. Code it as blocked, plus any product lost.
For a line like this, line OEE is the headline, and machine data explains it. The batch process guide helps with the mixer, and the continuous process guide helps with a tunnel oven.
Set one ideal rate per product
A tunnel oven's rate depends on the product, since each has its own bake time and belt loading. So the ideal is a rate per product, proven on your oven.
Use time, not raw counts, to combine products in a day. Otherwise a day of rolls looks far better than a day of loaves, and OEE by product has more on that.
| Product | Good units | Ideal per min | Good minutes |
|---|---|---|---|
| White loaf | 18,000 | 100 | 180 |
| Seeded loaf | 9,000 | 90 | 100 |
| Rolls | 20,000 | 250 | 80 |
| Total | 47,000 | 360 of 480 = 75.0% |
Worked example: one product, one shift
Example: white loaf runs alone on a 480-minute shift, and the line stops for 72 minutes in total. The oven turns out 36,720 loaves, and 35,251 pass inspection and go on to packing.
- Availability: 408 / 480 = 85.0%.
- Performance: 36,720 / (408 x 100) = 90.0%, with empty belt and slow make-up as the usual causes.
- Quality: 35,251 / 36,720 = 96.0%.
- OEE: 35,251 / (480 x 100) = 73.4%.
We do not print a bakery OEE benchmark, because we found no neutral published figure for it. Use your own line's baseline, and how to set OEE targets shows how to build from there.
Count the dough a stop ruins
Now the bakery-specific part. A stop log records the stop, but the dough it ruins shows up later, as a gap at the oven with no stop attached.
Example: the oven loader jams for 15 minutes. The oven misses 1,500 loaves, and 1,200 pieces at the end of the proofer over-proof and get thrown away, which is another 12 minutes at 100 a minute.
If the oven band itself stops, the loaves inside can over-bake too, so count those as well.
So weigh or count dough waste at each point, and tie it to the stop that caused it. If you only count at the oven exit, those 1,200 pieces arrive as 12 minutes of empty belt: OEE still falls, but as a performance loss with no reason next to it, and quality looks better than it is.
To put the loss where it belongs, count dough pieces at the divider as total units, or log scrap in units at the proofer. The result also feeds your production yield, and the scrap rate guide shows how to record it.
Product changes: code them by type
Not all product changes are equal. A single average will hide the difference, so split them by what actually has to happen.
| Change type | Example time | What takes the time | Who can shorten it |
|---|---|---|---|
| Same dough, new weight | 8 min | Divider and pan settings | Crew, with preset values |
| New oven profile | 25 min | Empty belt, oven settling | Planner, through sequence |
| Allergen change | 70 min | Full clean, verification | Only the waiting around the validated steps |
Sequence is often the cheapest fix. Running products in order of oven temperature and allergen profile can cut the longer changes, and changeover frequency shows the balance to strike.
Count all of it as availability loss, since changeovers count in OEE even when planned. Then use how to measure changeover time and SMED to shorten it.
Your food safety plan owns the cleaning steps, so any change to them goes through your food safety or quality team first. Sanitation downtime shows how to time the waiting around those steps.
Other bakery losses to code
- Oven gaps. An empty stretch on the belt is lost capacity even though the oven is running, and it counts as performance loss. Tally them at the oven exit, and use mean time between stops to see how often the belt runs empty.
- Morning start. If heat-up falls inside planned time, count it as a loss and keep the rule the same each day. See start-up losses.
- Mixer timing. A batch that finishes early waits, and a batch that finishes late starves the line. Log both, with the batch number.
- Packing stops. A long wrapper fault can fill the cooler and force product off the line or a stop at the oven, so code the cause at the wrapper. See upstream and downstream stops.
- Rework dough. Dough sent back into a later batch has not passed first time, so it does not count as good, however useful it may be.
Common mistakes
- Treating bake loss as a quality loss. Moisture lost in the oven is normal, and it belongs in yield. Quality is about loaves that fail.
- Taking the rate from the oven's nameplate. Use the proven rate for each product, since a nameplate says nothing about your bake time.
- Averaging OEE across products. Add good minutes, then divide by planned minutes.
- Logging only stops over 5 minutes. A 3-minute stop can still ruin dough, so tag the stops that cause waste.
Where to start
Begin with one oven and one product family for two weeks. Keep an hourly tally of oven output, a stop log and a daily dough waste weight.
The OEE calculator checks your math, and the OEE tracking template holds the shift records.
Bakery lines often mix old ovens with newer make-up machines. Fabrico takes stops, speed and counts live from PLCs or, on older machines, external IoT sensors, and operators tag the cause at the line. See how Fabrico connects ovens and older machines.
What people ask next
How do you calculate OEE for a bakery oven?
Set an ideal rate per product from oven load and bake time, then divide good units by the ideal output of the planned time. 35,251 good loaves over 480 minutes at 100 a minute gives 73.4%.
Where should OEE be measured on a bakery line?
At the pace-setter, which is often the oven, with a final good count after inspection. Keep data for the mixer, proofer and packer to explain the result. See line OEE vs machine OEE.
Is dough waste a quality loss in OEE?
It depends on where you count. Counted at the divider, ruined dough is a quality loss. Counted only at the oven exit, it shows as empty belt, a performance loss with no visible cause, so log the waste with the stop behind it.
How do I count an allergen clean in OEE?
It stays inside planned time as an availability loss, with its own reason code. Sanitation downtime covers how to time it without touching validated steps.
Does bake moisture loss lower OEE?
No. Normal moisture loss is part of the process, so it belongs in production yield. OEE counts units, and a loaf that passes inspection is good.
What is a good OEE for a bakery?
We found no neutral published benchmark for bakery lines, so start with your own baseline. The 85% figure usually traced to Nakajima is a general reference, and is 85% OEE realistic explains how to treat it.
Fabrico is a light MES for OEE. It takes stops, speed and counts from PLCs or IoT sensors, with add-on AI cameras counting output where a machine has no usable signal, and operators tag the cause at the line.