Operations
Why 85% OEE is the wrong target for a batch plant
Eighty-five per cent OEE is routinely quoted as "world class" in operations decks. In a batch chemical or blending plant, chasing it does more harm than good. The number is not wrong. It answers a different plant's question.
Where the number comes from
Overall Equipment Effectiveness multiplies three ratios: availability, performance and quality. It grew out of total productive maintenance in high-volume, repetitive manufacturing, where a line makes the same part for long runs, the ideal cycle time is fixed, and every minute the line is not running is plainly lost.
In that setting OEE is an excellent measure, and the 85% figure is a reasonable aspiration.
Why it breaks in a batch plant
A batch plant differs on every one of the three ratios.
Availability. Changeovers and cleaning are not breakdowns. In a plant making several products on shared equipment, wash-outs between incompatible products are a designed, necessary part of the process. Count them as lost time and the plant looks inefficient for doing exactly what its product mix requires. Exclude them and you hide the real cost of a poor schedule.
Performance. There is often no single ideal cycle time. Batch duration varies by product, by recipe and by input lot. A "performance" ratio against an average cycle time rewards running the easiest products, not running the plant well.
Quality. First-pass yield in a batch process depends heavily on input variability: a raw material lot slightly outside its usual range can push a batch out of specification with nothing wrong on the line. Folding that into one equipment number obscures where the problem actually is.
The result is a figure that rises and falls with product mix rather than with how well the plant is run. Two well-run batch plants can show OEE twenty points apart, and the lower one may be the better operation.
The other numbers to treat with care
OEE is not the only borrowed benchmark. Two others circulate just as freely:
- An 80:20 split of planned to reactive maintenance. Repeated everywhere and very hard to trace to a defined study. A ratio with no population behind it is not a benchmark.
- 95% on-time-in-full. Common in consumer-goods distribution, with an order profile and lead-time structure unlike most industrial supply.
I do not use any of these as targets. It is better to say so at the start than to be asked later why the numbers look softer than someone else's.
What to measure instead
For a batch plant, these measures track how well the operation is run, rather than what it happened to make that month:
| Measure | Cut it by | Why it works |
|---|---|---|
| Batch cycle time | Product family | Compares like with like |
| Changeover and wash-out time | Product-transition pair | Shows the true cost of the schedule |
| Consumption against standard recipe | Batch, product and shift | Usually the highest-value measure in the plant |
| First-pass quality and rework | Line and input lot | Separates process problems from input problems |
| Laboratory release turnaround | Test type | Often the hidden constraint on throughput |
| Energy per tonne | Line and product family | Comparable only after normalising for mix |
When an outside comparison is worth having
External numbers are useful when they are genuinely comparable. A few are:
- Named peer sets from public credit-rating rationales on listed competitors in the same chemistry, useful for working-capital and inventory days.
- Regulator-notified ranges. In the specific energy consumption values notified for chemical-sector designated consumers under India's PAT scheme, the range runs from 0.262 to 0.997 tonnes of oil equivalent per tonne of product. That nearly fourfold spread is itself the lesson: energy intensity cannot be compared across sites until it is normalised for product mix.
- Government best-practice guides, such as the US Department of Energy's operations and maintenance guidance, useful for the direction of travel and the relative cost of running equipment to failure, never as a target percentage.
Keep a benchmark basis
Whatever you compare against, write down where each number came from: the figure, the source, the date, the population it was drawn from, whether it is comparable to your plant, and what you decided. Include the comparisons you rejected, and why.
It sounds bureaucratic. It is the difference between a performance discussion about the business and an argument about whose number is right.
If you want an outside read on where your operation stands, the free eight-minute Operations Health Check is a quick way in, and a process transformation engagement is how the measures above get built into a plant's routine.
Last reviewed 25 October 2026.