For a continuous process there is no cycle, so use a rate. Availability is run time over scheduled time, performance is actual output over ideal rate times run time, and quality is on-spec output over total output. The simplest check is good output ÷ ideal rate ÷ scheduled time.
What is different about continuous processes
A continuous process makes product without a cycle. Think of a dryer, a kiln, a mixing line, a paper machine or a blending line. Material flows in, product flows out, and you measure it in tons, liters or meters, not in parts.
So the usual OEE words need a small change. There is no "cycle time" and no part count. You use a rate: tons an hour, liters a minute. The ideal rate takes the place of the ideal cycle time.
The three factors still apply. The data comes from flow meters, scales, tank levels and the control system instead of a part counter.
The three factors for a continuous line
| Factor | What to count | On a continuous line |
|---|---|---|
| Availability | Run time ÷ scheduled time | Trips, unplanned stops, feed shortages and grade changes are stops. A planned shutdown scheduled ahead can be excluded. |
| Performance | Actual output ÷ (ideal rate × run time) | Running below the best accredited rate, for any reason except a scheduled decision to run slow. |
| Quality | On-spec output ÷ total output | Off-spec product, transition material and product that has to be reworked or downgraded. |
The three factors for a continuous process.
(good output in tons, ideal rate in tons per hour, scheduled time in hours)
This is the simplest method in Overall Equipment Effectiveness (R. C. Hansen, 2001). It needs three things: scheduled time, good output and the ideal rate. Use the loss records afterwards to explain where the missing percentage went.
Set the ideal rate
The ideal rate is the best rate the line was designed for and has proven it can hold, for that product and format. It is not an average. And it is not a hope.
- Take the highest accredited rate for the product family.
- If you do not know it, use the rate from the best 4 hours of the last 400, a rule from Hansen.
- Cap the rate factor at 100%. If a day runs above the ideal rate, the ideal is too low. Re-accredit it and mark the change.
- Keep one ideal rate per product family. If a difficult grade runs slower, leave the rate and note the loss. That keeps the cost of that grade visible.
For more on this, see ideal cycle time. The idea is the same, with a rate instead of a time.
What counts as a stop, and what does not
| Event | How to count it |
|---|---|
| Trip or unplanned shutdown | Stop (availability) |
| Feed or utility shortage | Stop (availability), caused outside the line |
| Grade or product change | Stop or transition time (availability). The off-spec material made during it is a quality loss. |
| Running at 80% because of a raw material problem | Speed loss (performance) |
| Running slow because there are no orders | Not a loss if it is a schedule decision. Better: run at the ideal rate for fewer hours and exclude the rest as headroom. If you must run slow, record run hours × (1 - actual rate ÷ ideal rate) as excluded time. |
| Planned shutdown, set at least a week ahead | Excluded time (Hansen) |
| Cleaning or CIP between runs | Stop, if it is part of the normal run. Be consistent. |
How common events are counted. Decide once and apply the same rule to every line.
The line between "slow because of a problem" and "slow because we chose to" is the one to watch. If a plan says the line runs at 70% rate because demand is low, that is a schedule decision. If the line runs at 70% because a pump cannot keep up, that is a loss. Write down which one it was.
Worked example: one week on a dedicated line
Example: a dedicated process line with an ideal rate of 20 tons an hour. The week has 168 hours.
| Item | Hours |
|---|---|
| Calendar time | 168 |
| Planned maintenance shutdown, set two weeks ahead (excluded) | 8 |
| Scheduled time | 160 |
| Trips | 6 |
| Grade changes | 4 |
| Feed shortage | 2 |
| Total stops | 12 |
| Run time (160 - 12) | 148 |
Example week, hypothetical numbers.
During the 148 hours the line averaged 18.5 tons an hour, so it made 148 × 18.5 = 2,738 tons. Of this, 98 tons were off-spec, so good output was 2,640 tons.
| Factor | Calculation | Result |
|---|---|---|
| Availability | 148 ÷ 160 | 92.5% |
| Performance | 2,738 ÷ (20 × 148 = 2,960) | 92.5% |
| Quality | 2,640 ÷ 2,738 | 96.4% |
| OEE | 0.925 × 0.925 × 0.964 | 82.5% |
| Check with 3 numbers | 2,640 ÷ 20 = 132 h ÷ 160 h | 82.5% |
Both routes agree.
- OEE (good output)82.5%
- Stops7.5%
- Running below the ideal rate6.9%
- Off-spec product3.1%
The biggest loss here is stops, 12 hours or 7.5% of scheduled time. But running below the ideal rate costs 11.1 hours, nearly as much.
Check that: 148 h run time minus 2,738 ÷ 20 = 136.9 h of ideal time gives 11.1 hours, or 6.9% of 160. A report that only tracks stops would miss it.
The same week against the calendar gives a different number: TEEP = 132 ÷ 168 = 78.6%. The planned shutdown is excluded from OEE but counts in TEEP. See OEE vs TEEP.
Where to find 85%, 90% and 95%
Hansen sets world-class bars by process type, building on Nakajima's TPM figures: 85% or more for discrete and batch processes, 90% or more for dedicated discrete and non-dedicated continuous processes, and 95% or more for dedicated continuous processes.
The example line is a dedicated line, so the bar is 95% or more. It is 12.5 points below.
Do not treat the bar as a target for next month. Use it to size the gap and set your own steps. See how to set OEE targets and is 85% OEE realistic.
How to measure the data
- Output: a flow meter totalizer, a weigh scale, a tank level change or a counter. Use one source and check it against the stock.
- Run state: from the control system: running, stopped, tripped. Trips need a timestamp and a cause.
- Rate: the output divided by the run time, per hour or per shift.
- Quality: the lab or in-line analyzer result, linked to the time or lot. The result may come late, so link it back to the period it came from.
Data delay is the usual problem with quality. If a lab result takes hours, the OEE for the shift is final only after the result comes in. For batch-type steps in a continuous plant, see OEE for batch processes. For data options, see OEE data collection methods and PLC signals for OEE.
Common mistakes
- Using the nameplate rate. If the line has never held it, performance will always look poor.
- Not capping the rate factor. A value over 100% means the ideal rate is too low.
- Mixing units. Tons in one report and liters in another. Pick one unit per line.
- Counting slow running as a stop. The line was running. Put it in performance so you can see it.
- Hiding off-spec product in the grade change. It is a quality loss. Count it.
- Treating a decision to run slow as a loss. Write it as unscheduled time, so OEE is not blamed for a market decision.
What to do next
Pick one line. Write down the ideal rate, the scheduled time and the rules for slow running and grade changes. Use the three-number method for a week, then split the missing percentage into losses with the six big losses calculator. Start with the largest loss and use the steps in how to improve OEE.
Continuous lines need run state, rate and output from the process. Fabrico reads stops, speed and counts live from PLCs or from IoT sensors, and operators tag the cause at the line, so trips, grade changes and slow running show up as separate losses. See machine connectivity for OEE.
What people ask next
Does OEE make sense for a continuous process?
Yes. The question is the same: how much of the best possible output did you get from the time you planned to run? The units change to tons or liters, and the ideal rate takes the place of the ideal cycle time.
How do I define the ideal rate when there is no cycle?
Use the best rate the line has held, for the product family, in good quality. If you do not know it, use the best 4 hours of the last 400. Cap the factor at 100% and re-accredit if the line goes above it.
How do I treat a deliberate slowdown?
If it is a planned decision, such as low demand, treat it as unscheduled time and note it. If it is a response to a fault, such as a pump or a raw material problem, it is a speed loss. Always write down which one it was.
How do I count start-up and shutdown?
A planned shutdown set at least a week ahead is excluded time in Hansen's method. An unplanned trip is a stop. Off-spec material made during start-up counts as a quality loss. See start-up losses.
What if the product is reprocessed?
If off-spec product goes back through the same line in scheduled time, treat it as new input. Its time, rate and quality count in the OEE of that period. If it is reworked off the line, the line OEE does not change, but it is still a cost.
Related guides: OEE vs Process Reliability: How They Differ · OEE in the Chemical Industry: What to Count
← More OEE Lab guides · Open the six big losses calculator →