Overall equipment effectiveness (OEE) is a tried-and-true metric for manufacturers that provides insight on how effective specific pieces of equipment are to meeting a manufacturer’s capacity and throughput goals. But some in industry are questioning the continuing relevance of OEE in light of the advances AI is making to crunch a far wider array of data points beyond OEE’s focus on availability, performance and quality.
To get a better sense of where things stand with OEE amid all the technological changes taking place in industry, Automation World spoke with Jim Toman, senior manufacturing execution system advisor with system integrator Grantek for a recent episode of the Automation World Gets Your Questions Answered podcast.
Automation World: Let's start by first explaining what overall equipment effectiveness is as well as sharing the formula for calculating it.
Jim Toman, Grantek: So overall equipment effectiveness, or OEE, is a measure of how well a manufacturer uses a specific production line or work cell. So basically, there are a few factors that can't be changed in manufacturing, such as how much time there is in a day or perhaps the objective maximum rate at which you can make the product.
But there are other factors that affect the actual rate at which product is made, and those include: availability, which is the amount of time that production happened; throughput, which is the actual rate at which product is made; and quality, which is a measure of how much good product was made versus bad product. There are complexities under each of those measures, but that's pretty much it in a nutshell.
So, we take those three elements and we multiply them together as percentages and that's what gives us OEE. The OEE equation is availability percentage times the performance percentage times the quality percentage.
Automation World: Can you give an example of how to apply the OEE formula for a standard manufacturing line?
Jim Toman, Grantek: I've seen OEE applied in a lot of situations in all three types of manufacturing—discrete, process and continuous. For this example, I'll just take a discrete manufacturing example because it's probably the easiest to grasp.
So, let's say we have a manufacturer that makes snack packs. And let's say that the production line has a filling step, a wrapping step, a sealing step and then maybe weighing and cartoning. For simplicity we'll just assume that this is a single line with no deviations in it. What we would basically do is measure how many packs we fill in a shift, how many were rejected for quality issues, and how much time the line was running versus being down or stopped. We're not going to take into account things like changeover or cleaning time.
For availability percentage, what we would do is take the number of minutes that the machine or the line was running divided by the total minutes that were available for the shift, minus the cleaning and changeover time.
Let's say that the line was down for equipment issues for a couple of hours in an eight-hour scheduled run. If we had that, then we had eight hours of time available. We had two hours down. So, our availability on that was 75%. It's that simple.
Then for the performance percentage, we're going to take the ideal cycle time of the filler. We're going to multiply that by the total number of packs filled and we're going to divide by the running time. If we ran at full speed the entire time, then we would have a performance of 100%. In this case, the ideal cycle time is being met by every pack that we make. So, performance is really good in this scenario.
And then for quality we would simply take how many packs we had to reject for quality issues and then subtract it from the number of total packs to tell us how many good packs were made. We would divide that by the total packs, so if we lost 50 out of 1,000 packs, then our quality number is 95%.
That's basically the long and short of how you get each of the legs for each of the components of OEE. We would then take 75% for availability, multiply that by 100% for performance and multiply that by 95% for quality, which would give us an OEE of 71% for that shift.
So, if you see your OEE go up or you see your OEE go down, you have clarity on whether or not something got better or worse during that production run.
Automation World: While it's easy to understand what OEE tells you based on the formula as you just explained, how do you then apply it to achieve a quantifiable impact in your production operations?
Jim Toman, Grantek: What's interesting is over a lot of years I've seen it used several different ways.
One is that it's commonly used as a success measure. When companies are doing this, they're working to compare one operation versus another—either line to line, shift to shift or plant to plant. And it allows managers to understand where the operation is performing better in one situation versus another, or which particular product is better to place in an area that better handles higher volume versus one that doesn't.
OEE can also be used to plan production and, specifically, production schedules based on past performance so that they can make more reliable commitments to customers.
It also gets used to incentivize production teams. For example, you can show them what their OEE number is over the course of a shift and they know whether they're ahead or behind in meeting the target.
A second thing could be as sort of a canary in the coal mine for equipment performance issues. I often see this when it's an aspect of a maintenance regime—the company might be looking for negative trends in the OEE number. In other words, the OEE result started here and then it's degrading to here and now it's degrading to here and it seems to be going in a trend that isn't positive. OEE gives them the ability to then look into underlying data to try to understand what kind of issues might be developing on that equipment.
I see this a lot at manufacturing sites with very highly mechanized equipment and if they have, for example, increasingly lower speeds that may result in a lower throughput number. Perhaps there's something in maintenance that needs to be done. An investigation needs to be made to find out if a belt needs to be replaced or something needs to be done to prevent a breakdown situation. You can actually make measurable changes to your overall production performance if you keep an eye on those things and try to get ahead of problems as you see them starting to develop.
A third use of OEE, which I think is actually the most interesting, is when companies use OEE and the data that supports it as an enabler for continuous improvement. This is where they might take the data from a particular shift or run, and they would have that available either on a production report at a daily meeting, or perhaps even a weekly meeting or a monthly meeting. And they would look at it to see where they had issues during the course of this time period. Where did we not make as much product as we expected to or where did we have more breakdowns than others or where did we have troubles?
The most effective use of this information is where a company will apply it as part of their general continuous improvement process, such as Lean Six Sigma. Maybe they'll do a five whys investigation or a fishbone analysis. If there was a particular set of incidents that happened over and over again, they'll figure out how to address the problem points and then they might make a change and use OEE to verify that the change had the benefit they were expecting.
Automation World: OEE has become a standard feature on many automation software packages—particularly manufacturing execution systems—but also as standalone OEE software. Can you share some insights into what manufacturers should look for in both types?
Jim Toman, Grantek: You know, this is an interesting marketplace to have been in over the last two to three decades. I mean, at first it was sort of a roll-your-own kind of a thing or people would build a point solution to handle more automated forms of OEE.
But nowadays we have so many different offerings out there and they cover so many different aspects. The way I see it, companies are either looking for something that will help them get quick operational wins or else they're looking for an OEE capability that's part of a larger strategy toward digitalization.
If a company is just looking for operational quick wins and they're not concerned about the greater digital transformation journey, then that's where I see a role for standalone OEE packages. They're the ones that tend to work mostly by mobile app and are usually cloud based so that you don't have to own a server and do a capital setup.
On the other hand, companies that are making more of a digital transformation journey, there are a lot of platforms on the market that can offer OEE as a capability inside of a broader set of capabilities. And some of these have even been templated by integrators to make them easier to install and operate and to get fast time to value when implementing them. They can also provide several advantages, including native integrations to production equipment as well as the ability to share that production information with other applications, such as order management, batch control, statistical process control, track and trace, etc.
So really, it's a decision point that comes not from just the key features, but also looking at it from the standpoint of: Is this being used just to solve a particular problem or are we building this as part of a platform that's going to give us a higher degree of digital transformation down the road.
Access the full OEE podcast discussion with Grantek’s Jim Toman to learn about common mistakes made with OEE and why OEE remains relevant today at https://www.automationworld.com/33037090
