Subscribe and listen to AW’s podcast!
Subscribe and listen to the Automation World Gets Your Questions Answered podcast!
Listen Here

Which Production Metrics Should I Track?

How manufacturers should assess their operations to figure out which metrics to track with MES software.


Quick hits:

  • Measuring the bigger problems facing your operations may prove too complex to begin with.
  • Why measuring easy-to-capture metrics could be more beneficial than a business objective-oriented KPI strategy.
  • The importance of standardizing collected data to get the most benefit from MES metrics tracking.

Related to this episode:


Automation World, David Greenfield: So getting back to the specific reader question behind this podcast is, you know, how should a manufacturer go about assessing their operations to figure out which key metrics they should be tracking with MES software?

Grantek, Sam Russem: My take on it is that there's really kind of three schools of thought when you come to trying to find where you're going to start that MES journey, right? It's that you find your your biggest problems and complaints, and you go and try to tackle those. So you go and try to that's option one. Option two is you find those low hanging fruit to try to show that you can get something working. And three, which is a lot of kind of how I was trained a while back was really that you start with these overall business objectives, and then cascade your goals down to figure out what that MES and KPI strategy is going to be. I think that there's issues with all of these strategies in isolation, right? If you're If you're looking for the biggest problem, you're probably also finding the most complex problem, because if you could solve it, you would have solved it already, right? So that can really lead to a very complex design and expensive and risky implementation. On the other hand, if you go to the other extreme and take that lowest hanging fruit, sure, you might be able to show something, but you might not actually get any value out of it, right. And then what's the point, and business objectives like sounds like a really great starting point, and I wouldn't say it is probably the best practice. But what I've seen happen a lot of times is companies maybe without as strong of a strategic direction, or maybe don't have kind of that clear goal and business objectives laid out, it can really spiral into more of like a business management consulting activity, and you kind of get lost in in what you were trying to do in the first place. So I think it's probably pretty obvious what I'm trying to get to here, which is, I think that the right answer is a balance between these three, right, you want to find a problem that is really going to that people are complaining about, it's going to provide value, but not your hardest one, right. And you're going to want to make sure that this is all aligned to some type of overall business goal that you can set a goal post around, I want to reduce scrap by 10%, or increase throughput by 4%, or something like that, you should have some type of business metric that you're trying to do. And then you collect all your pain points and opportunities that can really drive that. And that's where you start to kind of build up where you might want to apply this MES technology to get started.

David Greenfield, Automation World: One thing that I'm thinking about here, and this may be one of those, it depends kind of answers. Because absolutely, you know, because every MES piece of software is going to be different. You know, but once a user is determined which metrics they want to track, how is this typically implemented in the MES software? I guess, at a general, you know, high level?

Grantek, Sam Russem: Yes. So actually, I'm gonna go back a little bit to kind of how I answered the first question, which is, I really kind of recommend breaking things out into three letter two to three layers, kind of like a model view controller type thing, if you're used to that type of programming terminology. But like, it's your connectivity, your compute and your display, right? Your connectivity piece, how do I get all of these data from all of these systems, that's going to change all the time, because you probably have lots of different systems that are going to want to get your data in different ways. So like, that needs to be its own core, because it's going to change at a different rates, and for different applications more than the other sides of things, then you have your compute. So once you've kind of modeled all of this data, hopefully you're bringing it in, and the common format. Now what you do with that data can become more standard. And when it gets to visualization, then you want things to be very standard, right? The more you can reuse, the better. So I do think you're gonna help to break into those ideas, how am I going to connect to all of this equipment? How am I going to get all of that data? And then who do I send it to? And I also think that as you break things down, that way, you'll kind of intentionally modularize your code to make it easier to kind of improve and change these things down the line as you do you need to roll out and expand.

David Greenfield, Automation World: And speaking about changing things, you know, as you move forward, you know, how difficult is it to add or adjust the metrics that you track in MES? If adjustments do need to be made over time based on what you've learned?

Grantek, Sam Russem: Yeah, well, if you follow what I just kind of described, hopefully not that bad. But it does, it can get messy all the time. I mean, there's plenty of times that I've walked into plants and other people that Grantek have walked into plants that have MES systems that have a lot of like hard coded data, right? So if you don't know, right, we're talking about hard coding something we're talking about, like when you code very specifically to a problem at hand, and not for flexibility and change in design in the future, like so people are gonna make a chart that shows you a temperature over time, but only works with this one temperature probe, and it won't translate over to other temperature probes and things like that. Right. So if you're designing your code with a proper modularity with that flexibility from the beginning, hopefully small changes are easy to implement. Of course, this is going to be an always depends type question. You know, if I'm riding a bicycle today, I'm not going to be able to turn it into a Ferrari tomorrow. But there's still work with small tweaks, we should be able to do it easy, easy, easily if it is well designed.

Test Your Machine Learning Smarts
Take Automation World's machine learning quiz to prove your knowledge!
Take Quiz
Test Your Machine Learning Smarts
Discover New Content
Access Automation World's free educational content library!
Unlock Learning Here
Discover New Content