Blog / OEE
OEE · Guide

OEE for Parallel Machines on One Process Step

OL By OEE Lab Editorial |Updated 2 November 2026 |6 min read
Short answer

Calculate OEE for each parallel machine on its own, then pool them: step OEE = total good output time ÷ total scheduled machine time. A planned spare adds no scheduled time and no loss. If the machines run at different speeds, also report good units ÷ the step's ideal capacity, because that shows how much of the step's possible output you delivered.

Why parallel machines break the usual rules

A step with parallel machines has a property a series line does not. One machine can stop while the step keeps delivering.

So the step has no single "running or stopped" state. It can be fully up, partly up or down, and a single timer on the step will miss the middle case.

That is why you start at the machine. Each one gets its own availability, performance and quality, its own count and its own ideal rate.

Worked example: three presses on one step

Example: three presses share a moulding step for a 480-minute shift. A and B are rated at 10 parts per minute, and C, an older press, at 6.

PressIdeal rateGood partsGood minutesOEE
A10 per min3,84038480.0%
B10 per min3,36033670.0%
C6 per min1,44024050.0%
Pool8,640960 of 1,44066.7%
Good minutes = good parts ÷ that press's own ideal rate. Each press has 480 scheduled minutes.

Pooled OEE is 960 ÷ 1,440 = 66.7%. This follows the scheduled-time weighting in how to average OEE across machines.

Use this number to judge how well the machine minutes were used. It treats a lost minute on C the same as a lost minute on A.

When speeds differ, you need a second number

A lost minute on C does not cost the same output as a lost minute on A. A planner who asks "how much of this step's capacity did we deliver?" needs a different sum.

Combined ideal rate = 10 + 10 + 6 = 26 per min
Ideal capacity = 26 x 480 = 12,480 parts
Capacity use = 8,640 / 12,480 = 69.2%
Capacity use weights each press by its speed, so the fast presses count for more.
Press A80.0%
Press B70.0%
Press C50.0%
Pool, by scheduled time66.7%
Step, by ideal capacity69.2%
Same shift, two honest roll-ups. They match when every machine has the same ideal rate.

The scheduled-time figure is the standard OEE roll-up, and it matches every other roll-up in your report. Capacity use is an added planning number, and plants differ on whether they show it or call it OEE, so write down what you report.

Our advice: show both when speeds differ. If this step sets the pace of a line, capacity use is also what line OEE sees, since line OEE divides the end count by the sum of the rates.

If the machines have different scheduled time, work out ideal capacity machine by machine. Multiply each rate by that machine's own scheduled minutes, then add them up.

Press C sits at 50% in both views. Its weight changes how much it hurts the total, not how it ran.

One of two machines stops: the half stop

This is the case a single timer on the line gets wrong. Picture two fillers feeding one packer.

Filler 150 per minFiller 250 per minPacker120 per min
The bottleneck is the filling step as a whole, at 50 + 50 = 100 per minute. Use that figure as the line rate, not one filler's 50.

Example: filler 2 breaks down for 60 minutes in a 480-minute shift. Filler 1 runs on, so the packer never fully stops.

At machine level the loss is plain: 60 stop minutes out of 960 scheduled machine minutes, or 6.25%. Filler 2's availability takes the hit, with a reason code you can act on.

At line level, a timer on the packer sees no stop at all. The line just ran at half rate for an hour, which costs 30 minutes of output at 100 per minute: 30 ÷ 480 = 6.25%.

The size of the loss is the same in both views. Only the label moves: a breakdown at the machine shows up as a speed loss at the line, and that label will send your team looking for the wrong thing.

So when you read a line's performance factor, check the parallel machines first. A rule for the line itself is in line OEE vs machine OEE: use the end count and the bottleneck rate, where the rate of a parallel step is the sum of the machines scheduled.

Spare machines and partial schedules

Often the plan does not need every machine. A machine left out of the plan in advance is excluded time, so it adds neither scheduled minutes nor loss.

Example: orders for the shift need only A and B. C stays unscheduled, so the pool is (384 + 336) ÷ 960 = 75.0%.

The scheduled time changes during the shift only if the plan really changes. When A breaks and you start C to cover, A's stop is still a loss, and C's time now counts as scheduled time from when it starts.

Keep the idle machine visible somewhere else. Utilization and the TEEP calculator show capacity you own and did not plan to use.

What to measure on the floor

1
Count per machine
One total at the end of the step cannot tell you which machine lost the output.
2
Rate per machine and format
Older or smaller machines get their own proven rate, not the rate of the best one.
3
Stops per machine
Log start, end and reason for each machine, even while the others keep running.
4
Mark shared causes
If a shared feed stops all three, log it on all three, with the same upstream code.
The minimum data that keeps a parallel step honest.

A shared feed that stops three machines for 10 minutes is 30 machine minutes of loss. Tag it as starved so it does not read as three separate faults; blocked and starved covers how.

Where parallel-step numbers go wrong

  • Averaging the percentages when machines had different scheduled time.
  • One ideal rate for all machines. Press C at 10 per minute would read 30% and look broken.
  • Counting only at the end of the step. You get a total and no cause.
  • Counting a planned spare as scheduled. A spare that no one planned to use is not a loss.
  • Skipping the bottleneck question. If the step is not the constraint, its OEE matters less than the one that is. See bottleneck analysis.

To test the numbers for one machine at a time, use the OEE calculator. If you are about to add a machine to the step, read why OEE drops after adding a machine first.

See it live in Fabrico

A parallel step needs stops and counts from each machine, not one total at the end. Fabrico takes stops, speed and counts live from PLCs or IoT sensors, and operators tag the cause at the line. See how Fabrico shows OEE per line and shift.

What people ask next

Should I calculate one OEE for the whole parallel step or one per machine?

Do both, in that order. The machine figures tell you where the loss is, and the pooled figure tells you how the step did. The pooling rule is in how to average OEE across machines.

How do I roll up parallel machines with different speeds?

Use good minutes ÷ scheduled minutes as the OEE roll-up, because each machine's minutes are measured against its own rate. Show good units ÷ ideal capacity beside it for planning. Practice varies, so label which one you report.

How do I find the ideal rate for a line with a parallel step?

Add the ideal rates of the machines scheduled at that step, then compare the total with the other steps. If two fillers at 50 feed a packer at 120, the line rate is 100. See line OEE vs machine OEE.

Does a machine waiting for work count against OEE?

Only if it was scheduled to run. If the plan left it out in advance, the time is excluded, and if it was scheduled but had nothing to run, read how to treat time with no orders.

Is a stop on one of two machines an availability or a performance loss?

At the machine it is availability, because that machine stopped. At the line it reads as performance, because the line kept running at a lower rate. Fix it at the machine, using its reason code.

What if the parallel machines make different products?

Convert each machine's output to good minutes using the ideal rate for that product, and add the minutes. Units cannot be added across products, but minutes can. OEE with mixed products shows the sum.

See OEE per line and shift on your own machines

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.

Book a Fabrico demo

Related guides: Cycle Time vs Process Time, With Takt Time · MTBF for Multiple Machines: Series and Parallel

← More OEE Lab guides · How to average OEE across machines →

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