Blog / Calculation
Calculation · Guide

How to Average OEE Across Machines Correctly

OL By OEE Lab Editorial |Updated 1 November 2026 |6 min read
Short answer

Weight each OEE by its scheduled time: Σ(scheduled time × OEE) ÷ Σ scheduled time. This equals the total good output (at the ideal rate) divided by the total scheduled time. A plain average of percentages is wrong when machines or periods differ in length, and so is multiplying averaged availability, performance and quality.

The rule: weight by scheduled time

OEE is a share of scheduled time. To combine several OEE values, you need to know how much scheduled time sits behind each one.

Combined OEE = Σ (Scheduled time × OEE) ÷ Σ Scheduled time
= Σ Good output time ÷ Σ Scheduled time

This is the aggregation rule in Overall Equipment Effectiveness (R. C. Hansen, 2001): weight each product's OEE by its share of scheduled time, and weight each period's OEE by its scheduled time. He warns against taking a plain average.

The two lines of the formula are the same thing. OEE × scheduled time is the good output time at the ideal rate (the "theoretical run time"). Adding these and dividing by the total scheduled time is just pooling.

Worked example: three machines

Example: three machines in a cell. M1 and M2 ran a full 480-minute shift. M3 was scheduled for only 240 minutes.

MachineScheduled minOEEGood output min (sched × OEE)
M148080%384
M248060%288
M324050%120
Total1,200792

Weighted OEE = 792 ÷ 1,200 = 66.0%. The simple average is (80 + 60 + 50) ÷ 3 = 63.3%.

Weighted by scheduled time66.0%
Simple average63.3%
The simple average gives the short M3 run as much say as a full shift, so it pulls the result down by 2.7 points.

Here the gap is small. It grows when the short, poor run is shorter, or when the numbers are further apart.

Why weighting works

Each machine's OEE is its good output time divided by its scheduled time. So its good output time is OEE × scheduled time. That is the number in the last column of the table.

Good output time is measured in minutes, so it can be added across machines even when the machines make different things at different speeds. A bottle filler and a case packer both turn into minutes. Units cannot be added in the same way.

Add the good output minutes, add the scheduled minutes, divide. The weighted formula is just this sum written another way. That is why it also works for products, shifts, weeks and plants.

M1: 384 good of 480 min80%
M2: 288 good of 480 min60%
M3: 120 good of 240 min50%
Pooled: 792 good of 1,200 min66.0%
Bar length is scheduled minutes. The colored part is good output minutes, the grey part is lost minutes. M3 is half as long, so it counts half as much.

Do not multiply averaged factors

Some teams average availability, performance and quality across machines and then multiply the three. This gives a different number, and it is wrong.

Example: two machines with equal scheduled time. M1 has availability 100%, performance 50%, quality 100%, so OEE is 50%. M2 has availability 50%, performance 100%, quality 100%, so OEE is also 50%.

MethodCalculationResult
Average of OEE (correct)(50% + 50%) ÷ 250.0%
Product of averages (wrong)75% × 75% × 100%56.3%

The second method invents 6.3 points that no machine produced. Both machines lose half their time in one way or another, and the loss does not overlap.

If you want to show the factors for a group, pool each one over its own base: availability = total run time ÷ total scheduled time, performance = total ideal time for all units ÷ total run time, quality = total good units ÷ total units. Show them as three bars or three lines. Do not multiply them into a fourth number.

Several products on one machine

The same weighting applies. Example: in one 480-minute shift, product X runs for 300 minutes at 80% OEE and product Y runs for 180 minutes at 50%.

Weighted OEE = (300 × 0.80 + 180 × 0.50) ÷ 480 = (240 + 90) ÷ 480 = 68.75%, so 68.8%. The simple average of 80% and 50% would say 65%.

Hansen adds the case of mixed segments in one period, such as normal running plus rework. His example: 292 hours at 75%, 8 hours at 0% and 12 hours of rework at 65% gives (292 × 0.75 + 8 × 0 + 12 × 0.65) ÷ 312 = 72.7%.

Months to a quarter, or a year

Take a line with these months: January 10,000 scheduled minutes at 72%, February 9,000 at 78%, March 11,000 at 70%.

Weighted: (7,200 + 7,020 + 7,700) ÷ 30,000 = 21,920 ÷ 30,000 = 73.1%. The simple average is 73.3%. Close here, but a short month with an odd result can swing the simple average a lot.

Hansen also advises summarising by completed orders or process changes where possible, because that keeps events intact. Use shifts if you need the crew data to stay together, and fiscal periods only when finance requires them.

Check before you publish a roll-up

  • Same rules everywhere. If one machine uses an accredited ideal rate and another uses last month's average, the combined number mixes two yardsticks.
  • Same excluded-time rules. Planned time must mean the same thing on every machine, or the scheduled time you weight by is not comparable.
  • Same period. Do not combine a full week of one machine with two days of another without weighting.
  • No silent gaps. A machine with no data is not 0% and not 100%. Leave it out, and say so next to the number.
  • Show the coverage. "OEE 66.0% across 3 machines, 1,200 scheduled minutes" is a number someone can check.

Do it in a spreadsheet

Put scheduled minutes in column B and OEE in column C, rows 2 to 4. Then:

=SUMPRODUCT(B2:B4, C2:C4) / SUM(B2:B4)

If your OEE is stored as a whole number such as 80, divide by 100 first or format the cell as a percentage. For a ready layout, see the OEE tracking template.

What a plant number hides

A single plant OEE is a good headline, and a poor tool. It mixes bottlenecks with machines that do not set the pace. Hansen says to apply OEE to bottlenecks first, and to subordinate everything else to them.

Our advice: report the plant figure, but always show the bottleneck figures beside it, and the best and worst result. A plant at 66% with one machine at 50% and one at 80% needs a different plan from a plant where all machines sit at 66%.

To combine machines that work in series, use line OEE vs machine OEE. For the other errors in roll-ups, read OEE calculation mistakes. To start from the single-machine formula, see how to calculate OEE, and to try values, use the OEE calculator.

See it live in Fabrico

Roll-ups go wrong when the inputs are typed in by hand. Fabrico shows OEE per line and shift from the same stops and counts, so the time behind each number stays attached to it. See OEE per line and shift in Fabrico.

Book a Fabrico demo

What people ask next

Why not just take the simple average?

Because a machine scheduled for 4 hours counts the same as one scheduled for 8. The result drifts away from the true plant figure. Weighting by scheduled time gives the same answer as pooling all the good output and all the scheduled time.

Do I weight by scheduled time, run time or calendar time?

Scheduled time. That is the denominator of OEE, so weighting by it keeps the result an OEE. Calendar time is the denominator of TEEP. See OEE vs TEEP.

Can I average availability, performance and quality separately and multiply?

No. The product of averages is not the average of products. The example above shows two machines at 50% each that "average" to 56.3% this way. Average the OEE, and show the three factors as separate bars.

How do I get a line figure from machines in series?

Do not average at all. Measure the line at its output against the bottleneck rate. See line OEE vs machine OEE.

What about machines with very different ideal rates?

The weighting handles it. Each machine's OEE is already a share of its own scheduled time, so a fast and a slow machine can be combined. Do not add up their units. Units are not comparable across machines.

Related guides: OEE for Parallel Machines on One Process Step · How to Calculate Downtime Percentage: Formula · OEE Formula in Excel: Paste-Ready Formulas

← More OEE Lab guides · Open the OEE calculator →

OEE essentials: OEE calculator · What is OEE · Six big losses · Availability · TEEP · Cost of downtime · OEE benchmark · Free OEE tracker · Free OEE software · How to track downtime · How to implement OEE · Light MES vs full MES