Intro to OEE
Susannah Brodnitz /
Aug 25, 2022
Efficient manufacturing is important for saving companies time, money, and energy. Making decisions based on data can improve efficiency, but there’s a lot of data to sort through. Manufacturing equipment contains many sensors, especially in the IIoT space.
Overall Equipment Effectiveness (OEE) was first described by Seiichi Nakajima in the mid-twentieth century as part of his Total Productive Maintenance (TPM) method. OEE is a single value that describes the efficiency of manufacturing as a percentage. An OEE score of 100% means that the quality of what’s being manufactured is perfect, the speed of manufacture is as high as possible, and machinery is always running when it should be.
Measuring OEE over time lets manufacturers find what parts of their processes are causing problems and allows them to make changes strategically. It also lets them correlate changes in practices with changes in OEE to know what works and what doesn’t.
To monitor OEE, companies need to collect time series data from manufacturing equipment. OEE is calculated as the product of Availability, Performance, and Quality.
When machines aren’t on, they can’t make anything. Some shutdowns for equipment maintenance are unavoidable, but limiting avoidable shutdowns is an important part of improving OEE. The two factors that most commonly lower availability scores are breakdowns and product changeovers. As manufacturers track these metrics over time, they can find out what is causing the problems.
If machines are breaking down, manufacturers can be more diligent about maintenance. And if planned stoppages are cutting into scheduled operation time, they can focus on making those stops more efficient.
Measuring availability lets manufacturers find bottlenecks and put their effort where it needs to go, instead of speeding up the wrong parts of the process. It also lets a company coordinate decisions with other factory changes, such as getting new equipment. For example, if its availability score drastically decreases when a company hires new people, it might look into improving new hire training.
Slow machines can also cause inefficiency, as can constantly turning machines off and on again. Performance measures the real world speed of manufacturing compared to an ideal quickest version. Measuring performance lets manufacturers target improvements. In actual calculations, it can be hard to pinpoint exactly how much speed affects efficiency, so sometimes OEE calculations designate inefficiencies that aren’t due to quality or availability as performance issues.
The two main factors that can lower performance scores are machines running slower than they should be and small stops for minor reasons. If a company finds that performance is causing their OEE score to be low, it can focus on speeding up processes. Recording performance over time also lets companies identify if low performance correlates with specific business changes, such as if a change in maintenance frequency speeds up or slows down machines.
Even if all machines are running as fast as possible and are always on when they’re supposed to be, it doesn’t matter if they’re not making good products. Quality measures the percent of products produced without defects. Measuring this quantity separately from availability and performance gives manufacturers greater insights into where to best direct their time and energy to improve efficiency.
If machines produce items with defects, companies either need to dispose or fix them, and either way it’s wasteful. When measuring quality, products that a company eventually fixes are still included in the count of products with defects. This lets companies focus on improving manufacturing so they produce good products in the first place and don’t need to fix them.
OEE is simple to calculate and gives companies a good target number to aim at to compare the efficiency of their manufacturing over time. The data involved in OEE calculations is largely time series data from sensors on manufacturing equipment, and companies need to handle large volumes of it.
Flux has an experimental OEE package that includes functions to help you calculate OEE. The
oee.APQ() function calculates OEE for producing parts using a time window, a string that represents the production state, and counts for good and bad parts. And the
oee.computeAPQ() function calculates OEE using two different input streams for production events and part events, which lets you have different time variables for the production state and the count of good and bad parts. Regardless of which tools a company uses to calculate OEE, it’s important to manage the data involved strategically and with care to get meaningful insights.