Drain The Data Lake - Model And Contextualize Your OT Data at the Edge

Speakers

Travis Cox

Chief Technology Evangelist

Inductive Automation

Arlen Nipper

President & CTO

Cirrus Link Solutions

Benson Hougland

Vice President, Marketing

Opto 22

Dan Stauft

Director of Operational Technology

SugarCreek

Michael Manzi

Practice Lead, Digital Manufacturing

Feyen Zylstra

Join a panel of Ignition community experts who helped the State of Indiana launch a Digital Transformation program for manufacturers quickly and simply. Energy data, manufacturing output, and other OT data can be collected and modeled in-plant, and efficiently published into cloud infrastructure and unsupervised AI for actionable insights with a pre-built “I4.0 in a Box” solution.

Transcript:

0:00
Travis Cox: All right. Good afternoon, everybody. How y'all doing today? Yeah, having a good time?

00:04 
Audience: Yeah.

00:05 
Travis Cox: All right. Welcome to our panel discussion. This is going to be good. We have a great panel up here today on “Drain The Data Lake - Model And Contextualize Your OT Data At The Edge.” So you guys know me. I'm Travis Cox, and I'll be introducing the panel here in just a moment. But I want to kind of set up sort of the state of where things are if you look at a Digital Transformation journey and why it's really important to look at contextualizing and modeling your data at the edge of the network. So most industrial organizations today that we talk to, they understand why they should implement IIoT solutions and why they should start their Digital Transformation journey, but they really struggle with the “How do I get started and what kind of solutions do I use?” There's a lot of noise out there, if you will, in the IoT world, especially with a lot of tools that are really primarily just IT-focused. So there's still a lot of question, a lot of doubt out there. And really, our message here is to kind of look at it from an OT level first, right? Looking at what we can do from that, how we can bring that up in a real meaningful way.

01:15 
Travis Cox: So the problem really is simple. There's a lot of people trying to solve these issues. They're typically not cost-effective. A lot of it, you see a lot of custom code being written out there. It's not scalable. It's very proprietary. A lot of nightmares in that and data inconsistencies that are there. A lot of stranded data, lack of data context, and we aim to really fix these kind of problems. So there is a new approach, though, of course, and0 we've been talking about this quite a bit, in that when we're getting data from our operations, we want to bring it up and model that data once and bring it up to every solution, every place. We want to look at that data going all the way up to the cloud. So we really want to democratize our data, and we're going to talk a lot about that here today with our panelists. We want to leverage platforms and look at basically using tools rather than coding on operating systems. It's really important to have tools that people can maintain going forward and to really kind of facilitate the cooperation between different departments, getting OT and IT to really work together within organizations and to kind of change that company culture that drive their company culture to cultivating change, to want to really accomplish these solutions in a meaningful way.

02:34 
Travis Cox: So we've got a really great panel here to kind of talk about some of the tools, some of the things that we're doing to make that happen. We have Benson Hougland here, who is Vice President of Opto 22, and really excited to have him here. Got Arlen Nipper, a man who needs no introduction. The king of MQTT, and he's CTO of Cirrus Link Solutions. We have Michael Manzi, the Practice Lead, Digital Manufacturing at FZ Industrial Tech. And we have Dan Stauft from SugarCreek. He is the Director of Operational Technology. So super happy to have you guys here. And Benson, we're going to start with you to talk about Opto 22 and also what you guys are doing to help with this modeling your data at the edge.

03:14
Benson Hougland: Terrific. So thank you for having me on the panel and with these esteemed guests as well. Yeah, my name is Benson from Opto 22, the obligatory Opto 22 slide. We've been around for a long time, nearly five decades. I've been with Opto for about 28 of those years. I'm responsible for product strategy at Opto. One of the things that really makes us kind of unique, particularly when you're talking about hardware manufacturers, is we all do it right here in California. Down in Temecula, California, about an hour north of San Diego, is a picture of the other factory there. Everything we make is designed, built, supported right from that factory in Temecula, which means that we're still shipping product. We are essentially a PLC to some degree, but we're much more than that, as those of you may be familiar with some of our products that we've been showing at this event. One of the things that has always kind of set us apart from other traditional PLC manufacturers, if you will, is we've always had a very distinctive design engineering philosophy of taking the notion of rugged OT-type systems like you would expect for something that might be in a plant floor or out on a remote asset, but combine it with technologies that are prevalent in the IT space. Now, that's not just for IT people or for OT. The idea is to try to pull these two technologies to better solve problems that exist, particularly in applications that are driven by this notion of Digital Transformation.

04:49 
Benson Hougland: Those two worlds must work together, and so we try to provide the tools that allow that to happen, and we primarily do that with our flagship products that we have today. The first is something called a groov EPIC. EPIC is an Edge Programmable Industrial Controller. Much more in a PLC, a lot of gateway functions. More on that later. Also, our groov RIO, which is a remote I/O unit. And something else called the groov RIO EMU. EMU is an energy monitoring unit. All of these things have come together in a very symbiotic relationship that we have with Cirrus Link. Arlen, friends for a couple of decades, tried to come up with an idea of how we could take some of these products from an OT perspective, combined with some technology on the IT side to really get people started with Digital Transformation, but do it right, right at the beginning. So part of that came into this evolving system that we put together. And part of what we're going to be talking about today is how we used modeling at the edge to solve a Digital Transformation problem for the manufacturers of the state of Indiana. More on that from Arlen.

06:00 
Benson Hougland: But indeed, what we did was we put together essentially a launch pack, a combination of hardware and software tools that can be placed inside a facility and immediately start to knock off some wins. Let's start with something simple. Let's start capturing energy data and start moving that information wherever it could be useful. We're gonna talk about where some of those places are coming up. So how are these deployed in a plant floor? This is kind of how it looks here. You can see we'll instrument a motor. We'll instrument a furnace, a pump, a press, whatever it might be, and also the whole building energy that is coming into a given plant to start bringing some visibility about energy use in manufacturing facilities. We take all that data, and we start to move it through these platforms coupled with the Ignition platform running on this CPU.

06:48
Benson Hougland: So indeed, in this case, it's not even acting as a PLC. It's acting as more of an IPC. So we're running full Ignition on this industrial device, quad-core, four gigs of RAM, lots of SSD on there, and we put it all in a nice little bundle and made it available for the state of Indiana manufacturers. There's a lot of technical information on this slide relative to some of the specifications of the box, but indeed it's a small Linux box. It's capable of running full Ignition or Ignition Edge. And we've been Onboard partners since the very beginning for running Ignition on our platforms, and indeed, it's all built-in and really nice to use.

07:29 
Benson Hougland: One of the things that makes the unit specific to this case and a lot of DT cases is it does have dual NICs, dual network interfaces. And the idea is to be able to separate the OT network from another network, maybe an IT network or some valid gateway network, so we can get the data where it needs to be quickly and do all this in a very cyber-secure way. But in doing so, you have to have the right tools. We're gonna talk about Ignition tools, but even at this level, the tools are meant to be used by folks in OT and IT. So, this isn't a foreign device on the network. IT knows how to manage this with the tools that they know how to use. DHCP, DNS, LDAP, I can run off a bunch of acronyms that all those tools are built in. So it's a quick deployment, gets you up and running very quickly. And that management interface is all web-based. So it's easy to use, it's easy to manage, it's easy to update, and so on.

08:24
Benson Hougland: And then, we have the interface right there for managing your Ignition instance that's running on the device. But again, the key thing here is that this is not only OT approved, and they understand this technology because it kind of looks like a PLC or an IPC, but it's also IT-approved again because the tools that are used to allow it to exist on a network are tools that IT is familiar with. So, in short, what we've really done here is give people an opportunity to get started quickly, get a quick win, get the data where it needs to be. We're gonna talk more about that in a minute. But then, once this platform is in place and it's all set up now, you can really start expanding. You can start utilizing all the power of Ignition running on this platform to start to connecting other systems within the plant floor, other PLCs, other devices, perhaps even other I/O signals.

09:15
Benson Hougland: Whatever it is, we now have a way to pull it all together and start using those powerful Ignition tools to model the data appropriately for the applications that will be consuming them. So, that's really a key aspect of this program, is to help people get started and be able to expand. So, very quickly, my last slide, this is just kind of an architecture view of how it might look in a given plant, where we have the OT side, where we're talking to PLCs that are largely unsecure out there. We'll put them behind the EPIC and make them secure and then start to model all that data, including I/O, put it all into Ignition, and start sending it up to a broker for it to be ingested by these applications you're going to learn more about. So there you go. Perfect.

09:58 
Arlen Nipper: Thanks, Benson.

10:00
Travis Cox: Thanks, Benson. All right. So, over to Arlen. Talk about more of the modeling aspect and how we can do that.

10:06 
Arlen Nipper: Everybody knows who I am, I think. So I'm going to skip over this slide. We are gonna leverage MQTT. So what this program is about, let's talk about draining the data lake, is that we had been doing a lot of different projects with Amazon. The state of Indiana had a lot of federal money that they needed to spend, right? And so, at the time, it was Pugal at AWS got with the state and said, “We can help you put together the program.” So the state of Indiana has over 8000 manufacturers. The state wanted to put together a program that could solve an initial problem. i.e. how is the state going to reduce their energy overhead? As a state, they've got to stay competitive. But also provide these manufacturers with a starting point for their adoption of Industry 4.0. So it wasn't this, "Oh, we're going to come in, we're going to put in a box that's going to monitor your energy, and then we're going to go away. In a year from now, somebody's going to walk around and say, why is that box in our office? And what does it do?" So, we wanted to give them a technology starter kit as well.

11:15
Arlen Nipper: As we know, no two manufacturing facilities are the same. So how can we put together a technology pack that could be installed in a day but flexible enough to grow across the factory? So I spent about two years … Where we started is we had four early adopter manufacturing companies. Not naively, but I said, "Okay, state is gonna buy Ignition for you. We're going to put it in your VPC or your onboard system.” So the mistake there was the customers were installing Ignition. They didn't really know how to do it. There wasn't somebody tasked to do that. So we kind of had to back up one step there. We got those four customers up and running, and then we started, "Well, how are we going to measure the incoming power?" Well, we knew we had to have that meter. And that's where I heard Benson. He had been doing a presentation. He said, "Oh, we've got the ability to measure KYZ, and we've got these new EMUs that measure three-phase power, and we've got an EPIC.”

12:26
Arlen Nipper: So, the next logical step was, what if we put Ignition on the Opto 22? We've got the energy monitoring units, we've got the KYZ. Now we can go in and target a customer very, very quickly. Not the whole plant, but at least the original goal, which was to get something in there and start measuring some energy. Now, the next part of the problem was we still had this notion of the data lake, right? While this was going on, Amazon was still looking at, well, we're going to send, pump all this stuff up into S3 buckets, and then from there, we'll try to feed that into some unsupervised AI, and maybe we can get some insights out of that. But that was going to become a manufacturer-by-manufacturer exercise.

13:16 
Arlen Nipper: And we wanted to be able, really, to scale this. We need to be able to go in and install the entire system in one day. That's the only way we're going to scale to 100, 400, 600 manufacturers in the state of Indiana. So from the notion of, “we're not going to go IT down, we're gonna build the models at the edge and not go into a data lake. We're gonna go into the models that SiteWise can give us through Amazon.” So now we can pre-configure a KYZ, two EMUs, have those UDTs ready, plug the box in, power it up, and point it to the MQTT broker in Amazon, and we're up and running. And we've got three power sources that we can start monitoring, and then we can scale from there. Because again, we wanna make sure that the manufacturer, the customer, understands Ignition. So what we're doing is going in with Discovery Days and saying, "Okay, here's what the state gave you. They gave you all of this. They bought it. But now, these are the tools that you've got in your factory, and these are how you can start expanding.” So Benson already mentioned this. We put together the launch pack. Now, although we're using this for the state of Indiana, this is available to everybody. This isn't relegated to just this program. This launch pack is put together, and Amazon has told me to tell all the integrators that if you find customers, then Amazon has funding to be able to put in this launch pack. And let's go ahead and take it forward.

14:54
Arlen Nipper: So, real quick, it was, “Keep It Simple, Sweetie,” in that, “How do you get started with Industry 4.0?” So, let the state start putting this journey together for you. You got a factory, and you got some machines, you got power. Okay, well, with the kit, you can monitor two of your larger machines, and you can monitor the KYZ input. So now we've at least got started, but you've got all the other protocols. So if I've got an Allen-Bradley PLC or a Modbus PLC, the customer can start bringing that in and start his journey for Digital Transformation. So this literally is the simplified drawing of what we're doing in the state. Is that where the goal is to be able to install the launch pack in a day, connect it into Amazon, and from there, we can go into SiteWise and into all the machine learning. So from there, I'm gonna turn it over to Mike.

15:49
Michael Manzi: Thank you, Arlen. I'm Mike Manzi with Feyen Zylstra. It's a mouthful. A lot of us just go by FZ. So, Feyen Zylstra was founded in 1980. We're a systems integrator. I like to think of as more of a solutions integrator. We have about 600 people, 100 million dollar company, 500 pole wire, 100 program PLCs. And I got about 10 or 15 in the Nerdery that do the reporting and the data collection and figure everything out in the analytics. I run that group. We have offices in Grand Rapids. That's where the home corporation is. I'm out of Cleveland myself with a couple of people also in Nashville and the Carolinas. My background a little bit. I was in the US Navy for six years. 90% of the boats I was on sank. They came back up. I was a nuclear mechanic on a sub for six years.

16:42
Michael Manzi: Stick around with me long enough, you'll hear that joke again.

16:46 
Michael Manzi: But then I pulled wires for a few years after that, and then I became an engineer. So I like to really think I bring a practical view to what the plant floor is doing.

16:55 
Michael Manzi: So even before this all happened, to some point last year, where I heard about this meeting outside of Detroit for Amazon. And that's where I got introduced to Arlen and Benson. And this box. But I was even thinking a year and a half ago, “There has to be this magical black box that I can just put on a machine, grab its data, and shoot it out to the cloud and start doing analytics on it”. It's just not that complicated. Why don't we have this yet? And so, I went to a meeting, met these guys, and here is this magical black box. I like this slide right here. “Your opinion is not my reality.”

17:29 
Michael Manzi: And another way I've heard this put is facts are not feelings. So, when you're running your corporation, you want to actually have the correct information to make the right decisions. And you start by getting that by bringing in your real data. Not manual data, not gut feelings. Your real data tells a story, and especially if it's verified and validated data. So the key to that is just how to get started because how do we even start? The big guys have been doing this for years. But the first thing, I was the global OT manager for both PPG Automotive and Kennametal for two large modernization projects, multimillion-dollar projects. You do not start off by buying a product. I'm saying that here. But we're telling you to buy this product. Right? But you don't. You have to start off with, what are we trying to solve? And from there, you create the business case, and then you bring in the right people for that question. 'Cause I don't know how many of you seen Hitchhiker's Guide to the Galaxy. They put in all the information, and they asked the computer, what is the answer? And it said it was 42.

18:36
Michael Manzi: And they're like, "42? What do you mean?" You failed to ask the right question. I had to make another computer to ask the right question. So sometimes, you see an IT system come down to the OT layer, and it fails because it had no OT input. I go so far as to, “I want an operator in the room, I want the plant manager, I want quality, I want maintenance, I want engineering in the room.” And we're gonna sit down, and we're gonna talk about what are the things that we're trying to solve within this plant 'cause then we'll have a success story, everybody has buy-in, everybody is a part of it, and they feel like they want to move forward.

19:11
Michael Manzi: The last line on here is very important as well, and this is where I think the AWS box comes in handy. The hardest part is just get started, and the IoT, the promise of IoT is that it's really lowered the cost and time to entry. Doing this for decades, especially in the days of MES, like the early 2000s, maybe even the late '90s, you looked at Procter & Gamble, you looked at GE, you looked at all these major players that were going with MES systems. What happened to the small and medium manufacturers? They were still operating off of Excel spreadsheets, Access databases. And today, it really hasn't changed.

19:56
Michael Manzi: So 90% of business in the United States, or manufacturing in business, is small and medium manufacturers. A lot of companies target the larger ones 'cause those are the bigger paychecks. But you're missing the biggest opportunity, which is the 90% that are small and medium. But the small and medium can't afford infrastructure for these data systems. You're not gonna see them put in a $300,000 process control network with their own domain controllers and switches. You're lucky if they VLAN off a few things.

20:30 
Michael Manzi: So when we look at the AWS solution, this is a great example of public-private cooperation and collaboration. There is money by Amazon and money from the state of Indiana that essentially pays for this box to go in. This box requires very little infrastructure, and I believe it can also do cellular.

20:48 
Arlen Nipper: Yeah.

20:49
Michael Manzi: Yeah. So you don't even need a network. You can put this right on a machine, and it can send data right out. We like energy 'cause energy is a known data model. You know, voltage, current, power. I think we kinda know what we're looking at. There's other organizations out there right now that are developing data models. You look at CESMII, you look at an Omadi, the OPC Institute. So there's a body of work going in there that are looking at known processes and assets. You have a press. Here are the inputs and outputs you should be monitoring.

21:17
Michael Manzi: When I was at PPG and Kennametal, we would write playbooks. Here are the inputs, the outputs, here's the protocols, here's what... And you bring in all the experts to find out what is the system doing and what are the true KPIs we want out of it? So that we can get the correct analytics and knowledge from it. And that is not a trivial body of work. That would take anywhere from three months to two years to do, to go through all your assets and processes. So there's actually bodies out there right now doing that work, so I'm not gonna trivialize that work, but to be able to jump in with energy. And energy is real savings, too. When you look at facilities management, you could probably see a 20% savings in facilities management. Am I running my compressors too much? Do I need all those boilers on? Those are all questions that could be answered by monitoring your facilities management.

22:05
Michael Manzi: So what is the role of the SI?

22:05
Arlen Nipper: They can read it.

22:09 
Michael Manzi: They can?

22:09
Arlen Nipper: It's behind you.

22:09 
Michael Manzi: Good 'cause I can't read this. They say in your 40s, “You're either wearing glasses, or you're lying.” So the role of the SI is going to change. We are not gonna be the ones programming. You're seeing a lot of solutions come out now that are low code for a reason. We had all these great solutions. Great analytics package 15 years ago, GE Predix, fantastic. Take it to a customer, nobody knew how to use it. You had to have a Master's or a Ph.D. in analytics to be able to understand this thing. So you're looking at this new generation of software that is low code and easy to use. So what is the role of the SI? 'Cause we're not gonna be coding?

22:47
Michael Manzi: We serve now as mentors and educators, and we come in to try to you lead people through the process. So if a company is going through a modernization journey, it's probably their first one. It might be their second, and they'll really listen to you if it's their second. But as an SI, you've done this tens, dozens, maybe even 100 times for some of the older guys. Dan, you might have done it like what 200?

23:13
Dan Stauft: What are you trying to say?

23:17
Michael Manzi: I think it's pretty obvious.

23:19
Michael Manzi: So it's just the old adage is, “You wouldn't represent yourself in court.” So you bring Rockwell in to talk to you about a modernization journey ... Guess what you're gonna hear? The Rockwell modernization story. You're gonna hear GE, Emerson, any of them, and even Ignition. But Ignition is playing more of an IoT role. A lot of SIs, and particularly FZ, is product agnostic. We are a Premier Integrator with Ignition, but we're only gonna put Ignition into places that makes sense. We're not gonna set it up to fail. So we don't lead with the product. We figure out what the problem is, what the solution needs to be, and then pick the product. Quite often, that solution is Ignition. Sometimes it's not.

23:58
Michael Manzi: So that's gonna be the role of the SI. It's gonna be more consultative, and there is an aspect at the machine level, there is a data cleansing aspect, and I'm not gonna minimize it. So there's a lot of... If everything was greenfield, it would be easy, but not everything's greenfield, so there's a lot of going in there and figuring out, "How do I get the KPIs out of the machines?" So there's actually bringing the data to the box in a contextualized way, and you think about standard tag naming conventions, asset models, basically turning an asset into an object. Once you can do that, it's all templatized, parameterized. You're actually seeing the real dream of what was S95 and S88. It's all modeled and parameterized.

24:43
Dan Stauft: That's me.

24:44 
Michael Manzi: That's you now.

24:46
Dan Stauft: See, just look for the food.

24:47 
Michael Manzi: That is true. You wanna do the last slide, too, though?

24:51 
Dan Stauft: Great. So, SugarCreek Packing. We are a large, privately held food manufacturer with manufacturing locations in three states. We call ourselves “the protein experts,” you see some examples of some of our products there. We were founded in bacon. We've expanded to a very large sous vide presence in the United States. We do a lot of formed patties, meatballs, some whole meat traditional cook systems. We are a co-developer and co-manufacturer, so you'll never see our brand name on anything, but we may have developed something like a venison sous vide that might have gone to Arby's, that you might see on a sandwich in Arby's. So we develop things and sell them to customers, and then we also do the traditional co-man where they come up and say, "Match this product," and we go.

25:52 
Dan Stauft: Oh, we've got five primary manufacturing sites, and we've got around 3000 employees. The big thing is we've quadrupled in size in the last 10 years. I joined the company nine years ago. We were doing about 300 million in sales. This year we're at 1.2 billion. We have been using Ignition for eight years, so we are not new to the program. We are based in Cincinnati, Ohio. So now the question you're gonna ask is, "What the hell is a company that's been using Ignition for eight years and based in Cincinnati got to do with an Indiana initiative for new manufacturers?"

26:31
Dan Stauft: Well, we have a large facility in Indiana, and I like free ****.

26:41
Dan Stauft: No, in reality, I was approached by Arlen. Arlen did a pitch on this at the ARC Conference in June. And I heard free in Indiana and raised my hand in the back. And part of the thing, as Arlen alluded, to is Discovery Day. So, whole idea of this is to get manufacturers introduced to Ignition, introduced to advanced analytics, and then on a regular basis. I think the cadence is supposed to be every six months or something like that.

27:09 
Dan Stauft: We're gonna get the companies that are involved together in Indianapolis, and they're gonna share their successes and their failures so they can learn from each other. So in talking to Arlen, again, I like free stuff, so I tried to convince him that, "Hey, if we got involved, we could kind of step ahead of the curve and show the other member companies what's possible." We just happen to have a very large facility in Cambridge City, in Indiana.

27:36 
Dan Stauft: This facility, if it was cut off from the rest of the company, is a $400 million facility by itself in annual sales. It was built in 2015. It has got Ignition everywhere. It's 420,000 square feet. It's a very large cook capacity. We've got six traditional cooking lines. We've got a 10-tank sous vide system. We do everything from meatballs to deli logs to American Airlines filet mignons. We do a lot of stuff. We've got a very large refrigeration system in it, and we've got a very large wastewater system. Again, everything's already connected on Ignition. So why do I need Benson's box? Well, we haven't used MQTT yet, and we haven't really delved into the cloud, and they're giving it to us for free. And they're giving us a box with an Ignition license on it that we can throw in another gateway.

28:29
Dan Stauft: So, what are we gonna do with it? Well, right out of the gate to kind of mirror what the intent of the Energy INsights program is, is we're going to take all of our already connected devices plus a couple new ones, and we're gonna use MQTT and shove them up into Mega AI and AWS and start getting a lot of data really fast. So at the next Discovery Day, hopefully, we'll be able to go up and go, "This is what we found. This is the improvement that we have." I don't think Arlen mentioned it, but the objective is an 8-10% reduction in energy usage just based on this.

29:11
Dan Stauft: So what are we gonna put out there? Well, we're gonna put 70 plus water meters, and they range from IO-Link smart devices to the old-school pulse devices say, "Why the hell do we need 70 water meters?" Well, this plant uses 500,000 gallons of water a day. So a little improvement on the water is a big improvement to the bottom line. And, oh, by the way, we're only allowed to use 500,000 gallons per day. So we're bouncing up against the regulation, environmental thing. So, that's where our biggest focus is gonna be water right out of the gate.

29:47
Dan Stauft: And then steam and gas that's already in place. The electrical's gonna be a little different because our power company doesn't want us to analyze the power we're using because we use a lot of power and they make a lot of money. So, they've got a 120-day lead time on getting us new meters that, and we've got two main feed meters to the plant. Those are getting upgraded to the latest and greatest technology in, I guess, four months. We have 12 or 10 switch gear locations that already have controllers. We're gonna put Modbus adapters on those.

30:28 
Dan Stauft: Again those, I think we got two of the 10, the other ones they're all back-ordered. And then we're gonna use Benson's box for the EMUs for the high-load equipment like refrigeration compressors. So what's the architecture gonna look like? It's gonna look kind of like this, which is a little bit more complex than I think the typical Energy INsights user. We're gonna take all of our existing meters that are going through data concentrators and feed into an Edge Tag Provider.

30:57
Dan Stauft: We're gonna combine them with the new Modbus addresses for the electric meters that are gonna go directly into the Edge provider. The EPIC. We're gonna use the EMUs, and we're gonna load up once we figure out which of the 10 switchgear is our problem child. We're gonna figure out what high-load devices are on that switchgear, and we're gonna daisy chain all of that using the groov EPIC. All of that's gonna feed up to an existing Ignition Gateway that has MQTT Transmitter, and we are going to set up on Monday the connection and hopefully, by Tuesday, I will have all of this, minus the stuff we don't have 100% on the cloud.

31:43 
Arlen Nipper: And also, Benson. Oh sorry, Dan, didn't you mention that you have the honor of having the highest penalty for power?

31:50
Dan Stauft: Yeah, well, there's that. Yeah, I guess for Hauser Energy, we set the record for power factor fines of over a million dollars. So that's another reason that they really don't want to upgrade their meter so we can figure out how to balance the load. But yeah.

32:06 
Arlen Nipper: That's quite literally a lot of bacon.

32:08 
Dan Stauft: It's a lot of bacon. Yeah.

32:11
Audience Member 1: Need little samples man...

32:12
Dan Stauft: Excuse me?

32:14 
Audience Member 1: Need some samples, man.

32:15 
Dan Stauft: Yeah, I can ship them to you.

32:17
Audience Member 1: Okay.

32:18
Travis Cox: All right. Well, thank you, guys, for going over the program here and all the details. We're gonna open up to questions now. If you guys have questions, just come over here to the microphones in the front, and don't be shy with all of those. While you guys are thinking about questions, I have a couple of questions I'll start with here. Arlen, you mentioned the speed at which these things can be put into place, and we've had some examples of real quick successes so far with the box.

32:45
Arlen Nipper: Yep.

32:46 
Travis Cox: Can you provide just some sort of idea of what this could look like if they had a brand new customer where they want to put this box in place?

32:52
Arlen Nipper: Well, typically, right now, the state has ... The factor of the future have some engineers that typically go out, do a site visit, just see what the setup of the facility is, then they'll come back, they'll configure up an Opto 22 box, and then either the state or FZ will schedule time with the customer, they go out, and I think FZ already has done a couple, it's one day.

33:18 
Michael Manzi: Right.

33:19 
Arlen Nipper: And they're even... I like the fact that FZ are giving us feedback on, "Hey, this wasn't quite set up right. But we'd like to get it, of course, to a half a day, and you've got data going in." Now, again, it's not the whole plant. It's just the incoming power, it's a couple of power meters, but we wanna make sure that we can go back to that customer and say, "Hey, we can pick up ... We noticed when we did the walk-through that you've got some Siemens power meters, you've got an Allen-Bradley PLC, you've got a Modbus PLC, you've got a Haas CNC machine that talks MTConnect." Well, we can start picking all that up. But that's where either the customer, we want them to self-learn that Ignition as much as possible or lean on some of the integrators but, hopefully, this will foster these small-to-medium manufacturers to be able to, like everybody has said, at least get started and grow from there.

34:19
Dan Stauft: The cool thing about the box as it comes from EMC2 is they've already done the site visit, they know what you're gonna get, so they've got the hardware connected, they've got the I/O cards installed, and they've also got all the UDTs set up, they've got the Ignition project pre-loaded, pre-configured, ready to go, and it's literally plug in, make a connection to the cloud and you're in business.

34:45
Benson Hougland: That's a good point. We did a lot of heavy lifting early on to try to identify what that data model would look like for energy use, and as we've already discussed, that tends to be relatively simple. You're working with current, you're working with voltages, and then, of course, we're deriving kilowatt hours and kilowatts, usage, and consumption. Model that data up, and so when you start thinking about, "Okay, are we gonna throw a bunch of watts and voltages up into the cloud?" No, we're gonna model it right at the device.

35:15
Benson Hougland: So now we take that data, and it's sent over through a UDT for an EMU or a KYZ meter, and now they can assign an asset to that particular UDT, and when it pops up, up into the AWS cloud for the Mega AI folks to start doing unsupervised AI, they're doing it with a real asset that was derived from the model that started right at the edge. And so having those templates done, and we put them on the [Ignition] Exchange so that anybody can download these templates that work with EMUs or RIOs connected to KYZ meters and get up and running very quick.

35:51
Arlen Nipper: Well, and that was the whole... This whole drain the swamp, drain the data lake notion, right? If we wouldn't have taken this tact for this solution, we would have been engaging them going up and doing all the custom stuff in the cloud that you would have to do to build the data model to get it into AI. Now that literally is completely gone. It went from ... The first project we did was 18 man-months of configuration and consulting fees and stuff like that. It literally has collapsed from 18 man-months to zero using this technique.

36:33
Michael Manzi: From our end, we're actually looking at our electricians being able to install this. We're not talking about a high level of ... I'm not gonna use this derogatively, but a high level of education or competency to put this in. We've lowered the technical capability to be able to install this. We're talking about having them, our electricians, driving around with a couple of them in the back of their truck as we're going to various factories and saying, "Hey, guess what? We got here. Do you want one?" So, yeah, it's much easier to use. Yep.

37:02
Dan Stauft: All pre-configured, right?

37:04
Michael Manzi: Yeah. And the closer you can get that UDT down to the actual PLC ... I mean, if you can actually program that UDT in the PLC, those data models will hold true throughout, and there's actually ... The body of work I talked about earlier is talking about having the same data model throughout the entire data stream.

37:23 
Arlen Nipper: Single source of truth.

37:24 
Michael Manzi: Single source of truth, everything, contextualized the same throughout the data flow.

37:28
Dan Stauft: So that's the part I'm most excited about. I've known Arlen for years, and he's been beating me about the face and neck about MQTT. And we never really found a use case for it. But this is an outstanding use case because, quite frankly, we're gonna blatantly use the hell out of the free AWS and Mega AI stuff we have because we're gonna be dumping production UDTs up that have OEE data, and have production rates, and have performance rates. We've got so many UDTs that are already set up. All of that's going up. And oh, by the way, since the same gateway that talks to Indiana talks to the other five plants, we're gonna shove all of them up too. So...

38:08
Michael Manzi: In the same way.

38:09
Dan Stauft: In the same way. Yeah. So within three weeks, I would expect every single UDT that means anything to us will be in the cloud.

38:20
Travis Cox: All right. We've got a couple of questions?

38:22
Audience Member 2: Yeah, how do I, as an integrator in Indiana, get some of my customers the information about this, or how would I get involved with them?

38:31
Arlen Nipper: There is a website, but please get a hold of me after this, and I will get you in touch with the ... Paul Mitchell runs the economic development for the state of Indiana, and he is the head guy in Indiana that's running this program.

38:46 
Audience Member 2: Okay, thank you.

38:47
Arlen Nipper: Yeah, contact me after this and I will get you that.

38:49
Audience Member 2: Okay.

38:50 
Travis Cox: It is really important. We'll iterate one more time that going forward, once this is in place, this is a starter kit. It's a foundational piece that allows to continue on and find other additional use cases.

39:01
Arlen Nipper: They get to keep it.

39:02
Travis Cox: You've got an energy use case right now, which is great, start with that, but then there's many others, as like Dan's looking and thinking about his systems, what he can get value out of going forward as well.

39:12
Audience Member 3: Okay, so ... It's great, you guys ...

39:16
Dan Stauft: Why you looking at me?

39:18
Audience Member 3: 'Cause the question is gonna come to you, Dan. So we've talked about the configuration, getting the data, getting into AWS, but as an end user, how are you going to get that data back into your system to take value from it? And as I'm assuming that you guys have already created the mobile application so that once the data gets to AWS, you have a Perspective screen that's coming back to?

39:45
Arlen Nipper: No, no.

39:46
Dan Stauft: No, it's all web-based. I'll defer to Arlen because we're not in it yet, but I know that the unsupervised AI takes place outside of Ignition completely. Once it goes up there, it's really … The intent is for the AI portion of it.

40:03 
Audience Member 3: So you're just gonna get an output. You're not... So if you wanna get that data back into your system...

40:07
Dan Stauft: Back into Ignition?

40:09
Audience Member 3: No, back ... Well, back into your on-prem Ignition.

40:14
Arlen Nipper: Well, right now, we are working with Mega AI to take the anomaly detections and get those back into ... So that they can build a model that we can use engines that subscribe to just that model and then be able to take that back in. But I think right now it's the notion of finding those anomalies and figuring out what data we're actually gonna feed back. So I look at it two ways. I think the unsupervised AI is gonna be cool at some point. But they say you can't fix it if you can't measure it, and the first thing I think these customers are gonna see is that "Oh my gosh, look at how much energy we're using," and that could be local display on the EPIC.

40:58
Audience Member 3: Yeah, I guess that was my question. Is there going to be ... The data that you're collecting, I get the idea of getting it to the anomaly detection, but how is Dan gonna take advantage of that data on-prem for his systems?

41:15
Arlen Nipper: We will get that. Like I said, when we figure out what that model looks like, so we can feed it back to Dan when we'll be able to do that. Now, to your point, Perspective or Vision or anything like that, that's what we're hoping either the integrator or the customer can go in there and say, "Okay, now I'm running Perspective in trial mode. I really like this. Okay, now I'm gonna enable this part of it." Because, again, trying to get this envelope of 50K, it's only Ignition base with [MQTT] Engine, Distributor, and Transmission, and then they can go in, and everything else is already on the groov, you can run it in trial mode, but then the customer would have to go in and say, "Okay, I wanna purchase Perspective," or whatever.

42:00
Dan Stauft: Yeah, so my thought to your question is we could ... 'Cause obviously, we've got Perspective, we've got everything. So the first case, because it's all web-based analytics that you can pull up in a web browser, so that means we can get to the web page and either scrape it or just show their web page in an Ignition client. Using the provided ...

42:24 
Benson Hougland: Everything's provided through Grafana dashboards that come right out of the AI system.

42:26
Travis Cox: But another important part about this is SiteWise. Thinking of that as an API, if you will, the data that's getting put in there, it's being ingested, is contextualized, if their models are there. It's stored long-term in time sharer’s database. It's easy to query. Mega AI happens to be doing that, doing the INsights, unsupervised AI. The results of that, going back to what they're working on, getting that back in the SiteWise means automatic, then back into us.

42:52|
Michael Manzi: And Mega AI is also working on ...

42:54 
Travis Cox: You wanna see a lot of that happen.

42:55
Michael Manzi: Yep. Mega AI is actually working on reporting dashboards. We've been working with them on what those need to look like. They're building to your typical OEE. If you think of ... To bring up another bad word like PTC ThingWorx, their MES capability is what you'll probably see within the next few months, or definitely not any further out than a year within Mega AI. Also, within the Opto box is full Ignition. So ...

43:21 
Audience Member 3: I guess to ask my question, so here's ... I tried to throw you a softball. You missed it. Well, here's the question. 

43:26
Benson Hougland: We can't get it back right now, so the answer, Arlen.

43:30
Benson Hougland: We're dancing.

43:30
Audience Member 3: You guys have the opportunity to, yes, send this to AWS. You also have the opportunity to send this to Hive, to send it to Chariot, to send the same data to another broker at the exact same time that the company could take advantage of. Correct?

43:44
Arlen Nipper: Yes.

43:44
Benson Hougland: Yep.

43:46
Audience Member 3: You could do it from Opto, you could do it from Ignition ...

43:47
Benson Hougland: We're doing it in our own facility where we like to eat our own dog food.

43:51
Audience Member 3: Yeah, that's all I was trying to get you...

43:52
Benson Hougland: We would put this exact same system into Opto, connect to our compressors, started to really get the data that we needed to understand how to stage those compressors properly to avoid demand charges. All that data is right there locally, and now it's up on being analyzed by AI at the same time.

44:08
Audience Member 3: Yeah, at the end of the day, this data belongs to Dan in your factory. You now have access to get that data on top of sending it to AWS.

44:18
Dan Stauft: Oh yeah, no doubt. No, we're gonna dashboard everything in Ignition, also, yeah.

44:21
Arlen Nipper: Absolutely.

44:22
Audience Member 3: Yeah, I could have sat down five minutes ago ...

44:25 
Dan Stauft: I thought ... I didn't understand your question. That ...

44:31 
Audience Member 4: That's too bad.

44:31
Dan Stauft: But for ... So to the point, for the Energy INsights consumer that is starting with the kit, you don't get Vision, you don't get Perspective, it's not included. So for that customer, all they get is the web-based trends and analytics, so I was speaking based on what the program capabilities were.

44:54
Audience Member 4: Also, as they said, the same data could go to Azure the same way it could go to any other consumer applications.

45:01
Benson Hougland: Right, exactly.

45:02
Arlen Nipper: There's money from AWS. That's the ...

45:04
Dan Stauft: Yeah, this is free. Did they mention that?

45:06 
Michael Manzi: You won't get the money from AWS to send it to Azure.

45:11
Travis Cox: Arlen, we have another question here.

45:13
Arlen Nipper: Yes, go ahead.

45:13
Audience Member 5: I have a quick question. I think you answered most of it, but you're doing energy, but what we want to do is to look at the sensors to have ... Read the vibration of the motors, whether it is compressors or filters or the water quality, those ... We want to read the values, although we want to have the analytics, but we don't want to ... We have a private cloud network, so we don't want to use any AWS. Is it possible to have, use your unit to read the vibration and then predict the conditional base?

45:47
Benson Hougland: You bet. Yeah, any sensor we can tie into the EPIC system through the RIO.

45:53
Audience Member 5: Do you recommend any sensors, or we can go with VIE, or any of the sensors that we like?

45:57
Benson Hougland: Yeah ... Yeah, more traditional sensors. I'd like to know precisely what kind of sensor you might be using to confirm that it would connect to the EPIC, but in most cases, that's what we do. We're an I/O system and a control system, so.

46:08
Dan Stauft: Yeah, I'm gonna upset the inductive guys right now. There's a lot for vibration. Specifically, there's a lot of off-the-shelf systems that are dedicated 100% to predictive analytics for maintenance, and that might be easier than trying to create that from scratch in Ignition.

46:27
Travis Cox: With that being said, it would be incredibly easy though …

46:35
Travis Cox: To get that data into an Ignition in their private cloud. So they can do a lot of amazing things with that.

46:41 
Michael Manzi: We could have a whole session on the groov EPIC and its capabilities. It could be a remote access gateway. We can have a two-hour session just talking about the capabilities, but this is just one thing that it's doing.

46:51 
Dan Stauft: We could have an eight-hour session on when is it the right tool for the job and when is it not.

46:57 
Michael Manzi: I would just say it always is, right?

46:58
Audience Member 5: That's good. I'll connect with you later to discuss more. Thank you.

47:02
Arlen Nipper: Thank you.

47:04 
Travis Cox: Alright, well, we are out of time. I was going over every other session I've done, so I'll wrap this up, and I thank our panel very much for being here today.

47:13
Travis Cox: Okay, good job. Have a good rest of the conference. Don't miss the Build-a-Thon.

47:19 
Arlen Nipper: Don't miss the Build-a-Thon.

47:19 
Benson Hougland: If you wanna see EPIC in action.