OEE for a line of several machines: which machine to measure and how to read the result

This article is for production managers, manufacturing engineers, continuous improvement leads and plant managers running lines made of several linked machines (machining, assembly, filling, packaging). It answers a question that comes up as soon as you want to track OEE for a line: should you measure every machine, the last one or just one, and how do you read a figure that depends on all the others?

Why line OEE is not the average of its machines

On a line in series, each machine depends on the one before and the one after. A filler can show good individual OEE while standing still part of the time because the labeller downstream is blocked. Conversely, a machine can show poor OEE simply because it is faster than the rest of the line and waits all the time.

Averaging machine OEE therefore produces a number with no physical meaning. What matters to the plant is the number of good parts leaving the end of the line, compared with what the line could have produced at the rate of its most constrained machine. That is the definition to use for line OEE, and it should be written down so everyone calculates it the same way.

Which machine to measure: the bottleneck first

The machine that sets the pace of the line is the bottleneck: the one with the lowest real capacity. Every minute lost on the bottleneck is lost for the whole line; a minute lost elsewhere may be absorbed by a buffer. Three points to watch:

• The theoretical bottleneck is not always the real one. The lowest nominal rate is not enough: a faster machine that stops often can become the true constraint. Only stop measurement over several weeks settles the question.
• The bottleneck can move. Depending on product, format or crew, the constraint sometimes shifts to another machine.
• Measuring the bottleneck alone is not enough to act. To understand why it stops, you need to see the state of its neighbours, especially starvation and blocking.

In practice, the most robust rule is to measure at least the bottleneck and its immediate neighbours, and every machine on the line when possible. Line OEE is calculated on good parts at the line exit, against the bottleneck rate; machine OEE is used for diagnosis.

Starvation, blocking and buffers: reading induced stops

On a line in series, part of a machine’s stops do not belong to it. They must be separated from its own stops.

Buffers (accumulation tables, conveyors, stock between stations) decouple machines. A well sized buffer absorbs one machine’s micro stops without stopping the next. A buffer that is too small spreads every stop across the line; one that is too large hides problems and ties up work in progress.

The useful reading fits in three sentences. If the bottleneck is starved, the priority is upstream. If it is blocked, the priority is downstream. If it stops by itself, the priority is the bottleneck.

An illustrative reading

For example, imagine a fictitious four machine packaging line: filler, capper, labeller, case packer. The figures below are invented for illustration. The labeller has the lowest nominal rate: it is the bottleneck. Over one week, its measured OEE is 62 %. Breaking down its time shows 9 % starved, 4 % blocked, and the remaining losses come from its own stops and slow running.

In this example, the labeller’s starvation comes mostly from the capper’s micro stops, which the small accumulation table between the two machines cannot absorb. Yet the capper shows a higher individual OEE than the bottleneck, which could have suggested it was not an issue. The most profitable action plan therefore targets not the labeller, but the capper’s micro stops and the buffer between them.

The most common calculation errors

1. Averaging machine OEE to get line OEE: the figure matches no real production.
2. Multiplying machine OEE values together: on a line with buffers, the same losses are counted several times and the result becomes artificially low.
3. Using the fastest machine’s rate as the line reference: the line always looks underperforming, without showing where to act.
4. Coding starvation and blocking as breakdowns of the machine that suffers them: maintenance gets sent to a healthy machine.
5. Excluding induced stops from line OEE: the line shows a good figure while actual output lags.
6. Counting good parts at the line entry instead of the exit: scrap at downstream stations disappears from the calculation.

What automatic measurement brings

TEEPTRAK measures stops, micro stops and rates on every machine in real time with a module wired into the electrical cabinet and a line side sensor, with no PLC connection and about one hour of installation per machine. OEE is available within 48 hours, machine by machine, so you can see the bottleneck’s own stops and those of its neighbours on the same timeline. Across more than 700 plants in over 30 countries, TEEPTRAK customers see 30 % fewer unplanned stops on average.

Further reading

• Bottleneck analysis with OEE
• Performance tracking: measurement pitfalls that distort decisions
• Eliminate micro stops: Pareto, top root causes and targeted SMED

FAQ

Should every machine on a line be equipped? It is ideal for diagnosis. At a minimum, measure the bottleneck and its two neighbours, to separate its own stops from starvation and blocking.

Is line OEE calculated on the last machine? It is calculated on good parts leaving the line, against the bottleneck rate. The last machine is the right counting point only if no scrap appears after it.

How do you know the bottleneck has moved? Track starvation and blocking on every machine: the bottleneck is rarely starved or blocked, and its own stops weigh most on line output.