The importance of reassessing MES value after deployment
One of the hot technologies in manufacturing is manufacturing execution systems, and one of the hot questions is how you measure value for the investment in MES.
One of the hot technologies in manufacturing is manufacturing execution systems, and one of the hot questions is how you measure value for the investment in MES. Manufacturing Summit panelists David Frede of GE Fanuc, Michael Gay of Rockwell Automation and Bob Ziegenfuse of Advanced Automation Associates tackled that question and others from attendees in a discussion moderated by editor Bob Vavra:
VAVRA: I really wanted to start off with something that is really important for our readers at the plant level.We've talked a lot about buy-in.We've talked a lot about unlocking this information down there. There's a tremendous amount of knowledge on the plant floor that really hasn't been unlocked in any meaningful way that I can see, before MES has come along.
What kind of feedback are you folks seeing as you're going out and talking to plant managers and plant engineers and plant floor personnel both in terms of what they expect from MES and, really, what they can contribute to the entire enterprise operation?
FREDE:I'm just thinking back through some conversations -- is that plant managers are under a lot of stress right now, especially in the larger, multi-plant companies, that they have to perform or shut down the plants or reduce what they make.
So they're looking for return on investment, on reducing labor, reducing energy costs, reducing inventories.In fact, what we find is, a lot of the people from the plant level, if you go from a plant level perspective on up, you've got a lot of these plant managers competing with the corporate IT people for project money.Corporate IT has got a long history of doing a great job of justifying their projects, and they're asking the vendors to help them justify their projects and applications.
GAY:One of the things that I see consistently when I go out and talk with clients about their issues is, the topic of MES never comes up.It's not a topic that plant managers or Bob that works on line 3 has a concept of as a solution. All they know is that when they make something, it's not made right because the plant manager tells them every day in the morning meeting that, "You consistently haven't met your target."So you really have to go in and discover what it is he needs to know to do his job better.So he can tell you exactly what the problems are.He is the guy that has the answers to all the issues, and that's Bob on line 3. The plant manager knows that from an enterprise perspective, he's being told day in and day out to improve your yield, reduce your manpower, take raw material out of the plant, to just reduce your carrying costs on a daily basis.
So they kind of come at it from two separate areas.But they don't really know that MES can solve the problem for them.So I think one of the big things that they're looking for from us, again, just from a supplier of products and solutions is, "Give it to me in something I understand."
Instead of talking about MES, one of the areas that I bang heads consistently with is IT because they're very technically focused.They're buying solutions.They're talking to their vendors.They want to push a solution down to the plant level.But if you push something down to a guy that you don't put it in tangible results with the red, green and yellow, forget it. It's not going to be successful.
So those are the things that we're seeing."Don't talk to me about technology. Talk to me about actionable data, information and solutions that make sense to me as an operator."
VAVRA:We talked for a long time about plant workers being told to make things better without ever defining what "better" meant. It seems like MES, one message to get out to the plant floor is, "Here's a way, as you just said, to make that more tangible, to give you raw data that is actionable rather than just theoretical."
GAY:It's data versus information. I'm sure everyone has probably heard that, too, but you have got to put it in tangible results.
VAVRA:And, Bob, you talked about that as well.It's very much a process of getting that buy-in but also working with the plant floor folks to help them understand that this is not just another software solution being thrown at them, this is about productivity and really tangible results.
ZEIGENFUSE:I can touch on that from two angles.We do a lot of shop floor control.One of the huge advantages of doing that is, the process machinery doesn't talk back.It doesn't express an opinion.You can move the equipment in the motions it was designed to move, no more, no less, and that's sort of a finite boundary.
Then you take a very fixed environment like that and move a process of doing an application where there are many departmental silos, some of which don't talk to each other, some of which are competing with each other, some of which have metrics that are way outdated, there's the human element, then, that does talk back.It does express an opinion.If you don't get their buy-in, it won't work, much like a piece of machinery breaking down, it just won't work.
The second point I'd like to make along those lines is, most of our clients, when we talk to them at all about anything above the shop floor layer, whatever label you want to put on that, they feel like they already have some systems in place, and the information they gather insults their intelligence.So they're wondering why they should invest any more money on those types of applications when there's already really, pretty much a myth that they throw away most of the reports that they get, or it insults their intelligence to the point they won't use them.
The reality is, is that be it the way their systems are currently designed, the normalization of data, the making the data, I believe you said, within context, there's a lot of effort required in getting real-time information up to some level of a database to produce a report that is often skipped.And, therefore, we have to go through a process of proving to them, if they design the data from the shop floor up to a database correctly, they're going to get information that doesn't insult their intelligence; and then, they're more willing to move forward with additional applications at that point.
VAVRA:It seems like the value proposition is different depending on who you're talking to at various stages of the organization.Yet, it ought to be the same value proposition if you are able to get all those silos broken down.
VAVRA: Michael, you sound like you're doing a lot of this.Explain that value proposition at various levels of an organization. How do you get through some of those silos?
GAY:I think a lot of it has to do with, again, putting it in the context that people understand.So it's a lot of the discussions of trying to discover the value because I don't know how to make a pound of cheese, as an example.But I certainly can understand what their problems are, and I can help them apply the information that's already in their process.
They've got control systems.They have manual processing.They have paper.You know, if I had a dollar for every Excel spreadsheet I've seen in a plant, I'd be a millionaire.
So it's more of a contextual discussion of getting it down to the people that are actually pressing the buttons and putting the product into the application or making the widget or creating the recipe itself, actually writing the recipe.Those are the people that really understand what the hurdles are.
The plant manager understands the financial aspect as far as how it's hurting him, and the financial guy in the back room that's running the numbers, he's the one that knows the value.So you have to kind of take all three of those pieces to make the equation to really get that message across.
I mean, you have to have the operational folks kind of by themselves, no management there, to tell you what really the problems are.Then you get to the line supervisor type people who will tell you, because the plant manager is not sitting there, they will tell you exactly what's going on.Then the plant manager, if the VP of operations is not in the room, he will tell you what's going on, too.So it's really keeping them apart to really get the truth.
Then you bring them all together at the end, and you sell them the whole picture, and they're all kind of looking at each other like, you know, the truth comes out, and there it is, yeah.
VAVRA:And it's a surprise in a lot of places?
FREDE:It is.You see some tap dancing going on.
VAVRA:You talk about the guys on the floor who have this information.Ninety percent of the time, they have the solution.They simply aren't able to drive it anywhere to where it's going to make some sense.They need to be able to drive some data and some information up through the organization.
I think for us, one of the things we've written so extensively about MES, we really think this is an opportunity for the plant floor to get credit for all the things they're doing right.It reveals a lot more going right on the plant floor than going wrong.
GAY:It's a collective solution. It's kind of like the prime directive.If everyone kind of has the same marching orders and they are all doing the same thing, then it could be a much better solution.
VAVRA:We talked a little bit about MESA.This is an organization where the market for MES is still very fragile. There are some people at MESA -- and I know because I've been working with them %%MDASSML%% who don't even necessarily think that MES is the right term.
But in any event, you've got all the major vendors in the MES space all sitting there, working together collaboratively to try and raise the profile of MES as a concept and as -- for lack of a better term -- a product line, but also as a solution. What are you seeing coming out of that organization?
FREDE:I was just at the conference last year.They are coming to grips with the term "MES," and they're not sure if that's the right term for the space we're doing it in.At GE, we're actually terming it "production management."We think it encompasses MES, which is more execution-based.Where we're seeing a lot of gains is not necessarily in executing but getting good manufacturing intelligence and information to act upon.
To Michael's point, I think the biggest transformation from more of a controls area to production management or MES, whatever you want to call it, to the business layer, it's now, you're not going with product features and functions.You have to actually go and be a business consultant first and talk to every aspect of the business, get all the issues on the table, and get a consensus of buy-in.Once you figure out what you're going to go solve, worry about the products and technologies to apply to solve that particular business set.So that's where we're trying to raise up to.
VAVRA:Well, Bob, you're on the integration side.Does that pretty much line up with what you guys are seeing at your end of the business?
ZEIGENFUSE:That's correct.Again, we work a lot with our clients to try to unconfuse them relative to MES because, quite frankly, it confused us because if you look at most of the definitions and most of the positioning out there as relative to selling product versus trying to be academia, educating the marketplace, so all the information out there, to me, has a product or a company skew or slant to it.
So even within our own organization, I had to break it down into, these are the components of successful implementations or solutions.We can call them many things, but this is what they're really doing at the end of the day, and go through the process of explaining it with a client.
And then, whatever terminology the client wants to use is just sort of what we stick with because that's what they have comfort with.
VAVRA:Different solutions for different organizations?
ZEIGENFUSE:Different solutions. Even different solutions within the same organization, within the same plant, on two different lines. The process equipment, you would think, would be standard from one line to another or one site to another, but there are a lot of subtle variations, and you have to allow for some adjustments.Even though your dataset is standard and can move up, what the users see and do every day, they need to be allowed to do some variation.
A good example of that is even on a simple thing as a delay code, some plants have some real strong cultures in place.So we have a standard corporate delay code, but we allow the plants or each site to have their individual delay codes underneath that.So that they can sort them by terminology they're used to, corporate gets the information, everybody is happy.
FREDE:The large corporate rollouts that we/re doing, they struggle with the standardization aspect. They want the same software across all the plants, but there's a certain level because every plant is different, and even machine to machine is different.So they're struggling with that layer of where you start to standardize the terminology.
VAVRA:But they all want the same thing at the end of the day.They all want a measurable ROI. Is that where you begin the discussion, or is that the end result of all of the things that you go to present to somebody on this subject when you start to talk about MES at the plant?
FREDE:I can talk about how we do it.We do a similar assessment process where you've got to go and find out all the issues at the company from all levels and all stakeholders. Once you go back and do the final presentation or meeting, you'll have what falls out of consensus on a few of those issues that they want to go tackle. That's the time when you will be asked to start doing some feedback ROI analysis.
VAVRA:Is that fairly low-hanging fruit for most organizations?
FREDE:Well, once you get them all to agree to go after one thing, yeah, then it becomes much more low-hanging fruit.The biggest issue is if you can only get one stakeholder and go down his path, and somebody else doesn't like it, you will just be tied up.
VAVRA:Do we have any questions out in the audience at the moment?
FROM THE FLOOR One of you made reference to the ERP system and SAP.I guess the questionis, you know, the Oracles and the SAPs of the world, they only go down so far. They operate at a level (where) the information is not usable from a production floor. Is the effort that you're talking about here an effort to standardize the interface into the ERP systems and the billing management systems, or is it a separate entity that discretely uploads and downloads?
FREDE:You know, there's definitely a continuous effort to standardize the interface.SAP has recently partnered up with a lot of the MES vendors in different vertical markets to go extend their solution down to the plant floor.
I sold the ERP systems before, and it was quite a mind-altering experience for myself, coming from the plant floor, to ERP systems because the terminology is the same in some cases but have totally different meanings.As an example -- and here is where I think one of the big misses between ERP and MES and the plant floor is -- in the one particular one that I was involved with, we would have an automatic data input module.I was amazed when I first came on board.I was, like, this is great, an ERP company that has a tie with automation.And it wasn't true. What they did is, they put a terminal on the plant floor for someone to type in numbers. That was their automatic data input.But from an MES and a plant floor perspective, that was completely manual.
So people are buying things that they think they have, and that's why we're not seeing the ROI results that we're seeing and why I think now you're seeing IT spend start to fall into the production management space, and that's why you're seeing Oracle and SAP starting to really dive into that space heavy.
FROM THE FLOOR:Is Oracle also going?
FREDE:I don't know of anything official.I'm pretty sure they're active in trying to figure out that space, with SAP leading the
way. I'm sure you'll start to hear some more stuff from Oracle down that same path.
FROM THE FLOOR:I guess I'm looking at something even more detailed.Corporate is running PeopleSoft, driving numbers down.They're looking at everything day-to-day.I'm running Wonderware on the plant floor to control the processes.I've got PLC drives -- all kinds of stuff running out there in basically corporate never-never land, running seven different processes and 12 different process lines.They're unique in every single way. How do you go about integrating something like that? Just kind of take me through on a high level, you know, your steps to do that. I see this MES as this piece that ties it together.You know, we're struggling with corporate seeing our numbers, and our numbers aren't going in the right direction, but we're not getting the information we need to fix the problems that we know we have; you know, trying to find the root causes of things like that.
GAY:I'm going to try to combine both of your questions.I think they kind of interrelate because from an ERP perspective, you hear a lot about the SAPs of the world talking about how they can integrate the plant floor and get that information off automatically.
When I mentioned before about myself banging heads a lot with the IT guys, it's because they come at it from that enterprise level.They're trying to say, "We can get that information off the plant floor," but, again, it's a lot of this manual data entry on the plant floor, and it's, like, "Hey, guys, you want to know how many pounds of raw product that just went into that vessel?It's right there.I'll tell you.It's right there in that register, in that PLC, and I can get it up to you."
So a lot of what happens, though, in our discussions from a grass roots level of driving the value on the plant is, is, the plant can come up with all these great ways of getting information, and one of the benefits is integration SAP or integration ERP because all the information is already in MES.All I gotta do is roll it up to you at the end of the day, and you've got your information that you want.
So that's, from an SAP perspective, my take on ERP getting done in the manufacturing level, is, there's no chance.I mean, I think that people like GE, like Rockwell, like Siemens, they have a much better handle of what's going on in the plant floor, and they can do a much better job of sending the information up.
Plants all have a different take on what information is valuable, and they all have their own kind of semi-language.But in that respect, if you're going to connect together 12, 13 or 14 or even 500 different enterprise plants around the world, you've got to have some kind of standardization on the communication of the metrics that go from the plant level up to enterprise, and that's the only way you can do it because it just gets too confusing at the end of the day.
So there may be some translation on the plant floor that makes sense to, you know, line 3, Bob, operator.But at the end of the day, it still rolls up to some metric that enterprise uses.
ZEIGENFUSE:I can answer that specifically.The example I talked about was 35 plants that had three business operations; and under each operation, they had maybe three different processes operated differently. We went through a bit of a design effort to make each one of those "datasets" run through a translator that then translated it to the corporate standard dataset so that each of those data were derived as close to the same way as possible, depending on the process.There was still real-time information that the operators were able to use that was sometimes different than that corporate dataset but was what the plant needed to actually run.
Another fantasy that most clients have is that you can define the dataset, and you're going to design it, and you're going to be done.Since I mentioned that, there's a continuous process associated with close of design gaps, and that being, the request for data never changes.So you have to have a methodology behind how you're going to collect all data.
We prefer to do that in an event-driven fashion versus time-based because then, you can get real-time information that isn't time-skewed.No matter what event is thrown at you in the future, it doesn't destroy your whole data acquisition methodology.That's what hurts most of the people who go with a simpler time-based solution.Once they start getting a lot of requests over the next year or two, their whole methodology breaks down, and it's a big rework project to get it right.
VAVRA:Do you find a need to change the terminology at the plant floor level?Do you need to find some standardization down there so that from plant to plant, you're all talking the same language at a specific function level?
ZEIGENFUSE:To me, that's a long-term battle versus any one particular victory in that the most successful way we have seen our clients do that is, they augment the design team with what I'll call a process redesign team.That's consists of operational folks, whom I will call respected peers in the operations.So that they're more in charge of when the design team says, "Hey, we need a standard here," the process redesign team then works with operations and the silos and has respected peers, and they'll try to create a consensus there that they will all move forward on, and the respected peers help derive that solution.
Then a big part of that is even giving them a look-and-feel or a demo of what they might be agreeing to because nobody really reads specifications anymore. So the look-and-feel for acceptance is a very important part of that process.So if you think about how that works, you've got a design team that uncovers needs, a process team that's going to provide solutions to those teams; and the common tool we tend to use is look-and-feel of the actual applications to gain consensus.
That's quite an effort to manage, and that's why implementing these systems there are difficult, not to be taken lightly.
FROM THE FLOOR:I think this is perhaps for Michael. In your approach to working with your customers, how rigorous is the activity you do in terms of pre- and post-measurement, and do you come back after a period of time to reevaluate, reassess?
I think you had the slide that was showing human factors as the largest risk area. I would presume it's also the largest constraint to adoption.
FROM THE FLOOR:And, as such, it's kind of like seeing is believing.So there's nothing that beats the real hard-core return-on-investment type of calculations that KPIs don't necessarily get at.So I guess the question is, really, how rigorously do you go after measurements for success?
GAY:In Europe, if you are a plant manager and your budget is $100 million. If you're going to go out and spend, say, a million dollars on a process improvement project which is surrounding MES and automation projects, your budget for next year is going to be trimmed by a million dollars.
So one of the things is that the plant managers -- and they're our toughest customer -- is, we have to show them that there's going to be value in this plant for that guy to sit down and sign his name at the bottom line to spend a million dollars next year because he's going to have a million less dollars to spend.
Rockwell Automation wouldn't be in business, and GE and Siemens and no one else would be in business very long if we made promises like, "We'll save a million dollars next year" if the million dollars wasn't there.So we spend a lot of time making very conservative judgments on what the value is going to be. “Underpromise and overdeliver" is the mantra that we consistently use.
It requires a customer to be very educated on the process of MES and what it can do for you.One of the best salespeople that we ever have in very successful MES installations is our own internal champion, the guy that signs up for the value, that guy understands how it's going to return value to his plant.He becomes an internal sales guy within the organization.He's the one that's out there educating the people on the plant floor how to use the information to make better decisions and to drive up that value.So it's in our best interests to make sure that we do deliver that value, so we're very rigorous in that regard.
FROM THE FLOOR:Will you make performance guarantees?
GAY:We've done it before.No customer has ever taken us up on it because what if the upside is so great they have to share some of the money?
FREDE:Execution of those are very difficult.
GAY:It's very difficult to do. You know, when it comes right down to it, the bean counters don't want to.
VAVRA:We have talked about those plant floor champions, though.But those are the people at the end of the day who are being asked to make some kind of a commitment in terms of either improvement absent the data, or they're being asked to cut staffs, to curtail the operations. So there's an incentive to find those champions out there.Once you do, it seems like this is a project that they're willing to embrace at the plant level very enthusiastically.But at the CFO level, it's a different discussion.It's a different process.But at the plant level, you're talking about people that they can see and know and will have to deal with if they go away.
ZEIGENFUSE:I'd like to add something to that.What we have seen, again, from a perspective of the most successful approaches is, you need to design into the system those ROIs that you're looking for. So those are numbers that your systems are going to produce, all right?Because, remember, we stated earlier that those reports insult your intelligence.Well, they insult ours as well.So when you tell us that you have a certain efficiency, and when we look at how you acquire that, I'd have to tell you I couldn't sign a contract around that because they fill in what you want to hear in most cases.
So sometimes, we have to first put in what we call enabling technology to be the system of truth in terms of what are you really doing today, and then we can sign up for what we will do tomorrow and make the system produce that in a real-time basis.
But the real key to success is to get the operators in what I call a pacing information, where you're taking bits and parts of that ROI and pushing it down to the operational level and having rewards, as simple as a trophy or a lunch pizza for people who are trying to improve that pacing data which, in fact, drives your ROI.
That's where people are most successful.It fits an ROI that everybody in the back room is calculating and seeing if you won or lost.I'll save you the trouble, you're probably losing.
VAVRA:Bob, I guess as you're all looking at and implementing these in various stages, do you find out that what either the enterprise folks or the plant folks think is the reality is really not even close, that it's actually worse than it seems going in?
ZEIGENFUSE:One of you said it earlier: that the first result of a system is the shock associated with the real numbers.
FREDE:If we don't get thrown out, then we can show them how to improve.
ZEIGENFUSE:I have a real life story to tell you. Earlier in my career as an enthusiastic engineer, we had a printing press, and I realized all the analog values had never been calibrated.So they were running this press based on fictitious engineering numbers.I mean, they were so far out of calibration.
So I decided to be a heady engineer, and I came in with a team, and we recalibrated the entire press right down to the minutest detail on every engineering unit, on every parameter, maybe a hundred parameters on that press. Well, can anybody guess how Monday's production went?
FROM THE FLOOR:It didn't?
ZEIGENFUSE:It didn't, right, because they all went to those numbers.And I said, "You can't run at that tension.The paper would explode."They go, "We have been running that tension for three years, Bob.That's what we run at."I go, "No. We calibrated it.You can run it at the real number now."Really, we don't compute.
So I had to take those hundred parameters and put them all back, decalibrate them back to their old numbers so we could run the press.
Well, that analyzes a microscopic view of your enterprise data.The numbers are just -- they're just numbers.I mean, you put engineering units by them for a feel-good factor, but unless you can engineer that from a sensor, all the way up, and prove that that's a value, it's just a number that you've agreed because, by a lack of any other good numbers, you're going to run your facilities by.So it's no more complex than that.
FROM THE FLOOR:Is there movement on the enterprise level to actually understand the concept of real-time?
FREDE:I think SAP is definitely working on it with xMII and their acquisition of Lighthammer several years ago, so, I mean, they're definitely pushing to get to real-time on the plant floor. I think they've partnered with all of the companies we have named so far to extend their solution down to the plant floor.
GAY:I think it really comes down to, we can get that Lighthammer connection down to the plant level. But what is it a connection to?Does the information make any sense to the people that actually need to use it?A guy sitting up at the ivory tower in the financial room that's looking at a dashboard is one thing, but he's not making anything happen.He's not creating, he's not forcing change in the plant floor. He's just looking at the numbers, going, "Holy cow, we're off by 5% in our yield numbers for this week.Why is that?" He's got no idea.
It's the guy on the floor that's going to take that information and really use it.That's why I think that, in a way, from an automation provider perspective, it's like SAP coming in and buying these, a company like Lighthammer, you get a little nervous.Then you think to yourself, well, when it really comes down to it, these guys don't understand the plant floor.They don't understand how to use that information to make really effective change. That's why I think MES is still there, and they still supply the value that's coming out of the plant.