In batch production, OEE is measured per batch against the recipe's ideal cycle and batch size; in continuous production, OEE is measured against a nameplate rate over a run. Both use the same formula, Availability × Performance × Quality, but they define ideal cycle time, the unit of output, and the treatment of changeovers and cleaning differently, which is why the frames are not interchangeable.
Pick the wrong frame and the number lies in predictable ways: a batch process forced into a continuous rate shows fake speed losses during charging and discharging, and a continuous process chopped into artificial batches hides real micro-stops. This comparison lays out each frame, works an example of both, and shows where they diverge. For the underlying method either way, start with the OEE calculation then use the OEE calculator.
What is the difference between batch and continuous production for OEE?
The difference is whether output flows or arrives in discrete lots. Continuous production, a bottling line, a paper machine, an extruder, runs a steady stream against a nameplate rate, and OEE asks how close the stream stayed to that rate. Batch production, a mixer, a reactor, a fermenter, an oven load, produces one lot at a time through a fixed recipe of charge, process, and discharge, and OEE asks how many good batches you completed versus how many you could have.
The same plant often contains both. A bakery mixes dough in batches and then bakes it in a continuous oven; a beverage plant blends syrup in batches and fills bottles continuously. That is why the frame has to be chosen per process, not per plant.
How do you measure OEE in continuous production?
In continuous production you measure OEE against a nameplate rate over the run. Ideal cycle time is the best sustained speed the process can hold for that product, bottles per minute, meters per minute, kilograms per hour, and Performance is how close the actual output came to that rate over the run time. Availability captures stops and changeovers; Quality captures off-spec product, usually counted per unit or per unit of mass.
A hypothetical continuous example: a filler with a 480-minute shift, 30 minutes of breaks (450 planned), 54 minutes of stops and a format change (396 run time), a nameplate of 500 bottles/min, 178,200 bottles filled, and 174,636 good. Availability = 396 ÷ 450 = 88.0%; Performance = 178,200 ÷ (396 × 500) = 90.0%; Quality = 174,636 ÷ 178,200 = 98.0%; OEE = 77.6%. The whole story is the gap between the actual stream and the nameplate stream. This is also why throughput and OEE move together on continuous lines.
How do you measure OEE in batch production?
In batch production you frame OEE per batch. The ideal cycle time is the recipe's minimum time to charge, process, and discharge one batch at standard conditions; Performance is that ideal batch time divided by the actual batch time, averaged across the batches you ran; Availability is the planned batches you could have run versus the time actually available; and Quality is good batches (or good mass) over total, since an off-spec batch is frequently rejected whole.
A hypothetical batch example: a reactor with 420 minutes of planned production time, an ideal batch cycle of 90 minutes, and 4 batches completed in the shift. If the batches actually took 90, 105, 95, and 100 minutes (390 total run time; 30 minutes lost to a charging delay), Availability = 390 ÷ 420 = 92.9%. Performance = (4 × 90) ÷ 390 = 360 ÷ 390 = 92.3%. If one of the four batches was off-spec and rejected, Quality = 3 ÷ 4 = 75.0%. OEE = 0.929 × 0.923 × 0.750 = 64.3%. One bad batch dominates the score, which is exactly the signal a batch process should send.
| Factor | Continuous frame | Batch frame |
|---|---|---|
| Ideal cycle time | Nameplate units/min at best sustained speed | Recipe minimum time to charge, process, discharge one batch |
| Unit of output | Individual units or mass in a stream | Whole batch (or mass per batch) |
| Performance | Actual stream ÷ nameplate stream | Ideal batch time ÷ actual batch time |
| Quality | Good units ÷ total units | Good batches ÷ total batches (often whole-lot) |
| Changeover / CIP | Availability loss between runs | Availability loss between batches or campaigns |
Where do batch and continuous OEE definitions diverge?
They diverge at three points. First, ideal cycle time: a rate for continuous, a fixed recipe duration for batch. Force a per-minute rate onto a reactor and the charge and discharge phases read as speed loss even when the batch ran perfectly. Second, the unit: a stream of units versus whole lots, which changes Quality from a defect rate into an accept/reject decision that swings the score hard. Third, cleaning and changeover: on continuous lines these are clean breaks between runs; on batch lines, clean-in-place between batches can be most of the non-producing time, and where you draw the planned-versus-unplanned line matters more.
How do you choose the right OEE frame?
Decide it process by process with this order:
- Ask whether output flows or arrives in lots. A steady stream is continuous; discrete charges through a recipe are batch. This single question sets the frame.
- Define ideal cycle time to match. A best sustained rate for continuous; the recipe's minimum charge-process-discharge time for batch.
- Set the unit of output. Individual units or mass for continuous; whole batches (or mass per batch) for batch, so a rejected lot lands where it should.
- Place CIP and changeovers. Decide once whether each clean sits inside planned production time, and hold that rule, it is the same discipline as the six big losses.
- Handle ramp and transitions honestly. Startup and grade-change ramps on continuous lines are Performance loss, not free time; account for them rather than trimming the run.
- Never blend the two into one number. A plant OEE that averages a reactor and a filler is a number nobody can act on.
What about hybrid lines that do both?
Hybrid lines measure each stage in its own frame and connect them through the buffer between. A bakery mixes dough in batches, holds it briefly, and bakes continuously; the mixer gets a per-batch OEE and the oven gets a rate-based OEE, and the proofer between them is where a batch stumble becomes a continuous starve. Trying to force one frame across the whole line hides exactly the coupling you most need to see. When you compare hybrid lines or decide where to hold inventory, the make-to-stock vs make-to-order choice and OEE vs TEEP both matter, because scheduled downtime looks different in each frame. For food and beverage plants specifically, food manufacturing software ties the batch and continuous stages into one view.
Which frame gives a lower number, and why does it matter?
Batch frames often produce lower OEE because a single rejected lot can knock Quality down hard, while continuous frames spread a defect rate across many units. That is a feature, not a flaw: batch processes carry more value per lot, so the metric should punish a lost batch. The point is not to compare a batch OEE against a continuous one, that comparison is meaningless, but to compare each against its own history.
- ISO 22400-2:2014 defines OEE and its time-states precisely for exactly this reason: comparability. Two sites using the standard reach the same number from the same facts, whichever frame fits (ISO 22400-2).
- The 85% world-class figure comes from Nakajima's 1980s TPM work and is a discrete-manufacturing reference point, not a rule for batch chemistry, see what a good OEE score is.
- For context, U.S. manufacturing capacity utilization was 75.7% in May 2026 per the Federal Reserve's G.17 release (Federal Reserve G.17), a different metric, but proof that real plants of every type run below theoretical maximums.
Choose the frame that fits the process, hold it steady, and read the trend. Harmony computes OEE from the machine's own signals in whichever frame the process needs, rather than from end-of-shift estimates (see the platform and the CLS case study). Then run your inputs through the OEE calculator.