How to calculate OEE : method and best practices

Calculating availability : accounting for all downtime

Availability measures the proportion of time during which the equipment is actually running.

In practice, the most common mistake is to focus only on major downtime. However, the most significant losses often come from the accumulation of short stops.

Consider a real-world example. Over an eight-hour shift, a machine experiences one hour of major downtime due to a breakdown. In addition, several short stops occur throughout the day, totaling forty-five minutes.

If only major downtime is recorded, availability will be overestimated. When all interruptions are taken into account, the actual loss becomes much more significant.

This situation is common when data relies on manual reporting. In contrast, automated monitoring provides a precise view of all downtime events.

In this context, vibration monitoring helps better characterize downtime related to mechanical issues.

Calculating performance : measuring real production output

Performance reflects the ability to produce at the expected rate. It is often overestimated because slowdowns are not always visible.

A machine may run continuously without ever reaching its nominal speed. Operator adjustments, flow variability or upstream and downstream constraints can gradually reduce output.

Over the course of a day, these deviations can result in a significant production shortfall.

For example, a machine designed to produce 100 units per hour may actually average 90 units, without any obvious downtime. Over eight hours, this results in a loss of 80 units, equivalent to nearly one hour of production.

To properly assess performance, it is necessary to continuously compare actual output with expected output.

Production order tracking provides this level of visibility.

Calculating quality : focusing on usable output

Quality represents the proportion of conforming products. It should not be limited to rejected parts alone.

In many cases, reworked or repaired parts are not included in quality calculations. However, they consume time and resources, and therefore impact overall performance.

Consider a scenario where 1,000 parts are produced, 50 are rejected and 40 require rework. If only rejected parts are considered, quality appears acceptable. When rework is included, actual performance is significantly lower.

A meaningful OEE calculation must therefore consider the full scope of usable output.

A complete OEE calculation example

Let’s look at a realistic production scenario.

A production line is scheduled to run for eight hours. During this time, it experiences one hour of major downtime and forty-five minutes of short stops, resulting in six hours and fifteen minutes of actual operating time.

The theoretical production rate is 100 units per hour, leading to an expected output of 800 units. In reality, 720 units are produced.

Out of these, 670 are conforming, while the rest are rejected or require rework.

In this case :

• availability is approximately 78%,
• performance is 90%,
• quality is 93%.

The overall OEE is around 65%.

This result reflects a combination of losses across all three dimensions rather than a single issue.

Why OEE is often miscalculated

In most cases, discrepancies between theoretical OEE and actual performance stem from how data is collected and used.

When data relies on manual input, downtime causes are often imprecise. Operators may lack the time or tools to accurately classify events, and short stops are rarely tracked in detail.

In addition, the distinction between different types of production time is not always clear. Planned time, operating time and productive time are often confused, leading to inaccurate calculations.

Finally, insufficient data granularity makes it difficult to identify root causes. An aggregated OEE figure provides an overview, but not the insights needed for improvement.

As a result, the indicator may appear consistent while remaining difficult to act upon.

Improving OEE accuracy through better data

The reliability of OEE depends primarily on the quality of the underlying data.

An indicator based on estimates cannot effectively support decision-making. Conversely, precise and continuous data allows for accurate loss identification and targeted action.

Automated data collection plays a key role here. It provides consistent and objective information without relying on manual input.

For example, energy consumption monitoring can reveal operational drifts that are not visible in standard production metrics.

By combining this data with production and downtime information, it becomes possible to gain a much more accurate view of performance.

Calculating OEE is only the first step

Calculating OEE is necessary, but not sufficient.

An indicator, no matter how accurate, only creates value if it leads to concrete action. The goal is not to produce a number, but to identify losses, understand their causes and prioritize improvement initiatives.

For a broader perspective on OEE and how to optimize it, you can refer to our dedicated guide :
OEE : definition and optimization of equipment performance

Conclusion

Calculating OEE may seem straightforward, but doing it properly requires rigor and precision.

The main challenges do not come from the formula itself, but from the quality of the data and how it is interpreted.

A reliable OEE highlights real losses and supports effective decision-making. An approximate indicator, on the other hand, creates a false sense of control.

This distinction is what turns OEE into either a simple metric… or a powerful operational tool.