As a 21st-century concept, the Digital Twin captures the idea that we are moving work from the physical world into the virtual one. It moves toward using bits, which are getting cheaper all the time, in place of atoms, which are getting more expensive. We are replacing expensive physical time, energy, and material with much cheaper digital information.
The Digital Twin is the concept driver behind Manufacturing 4.0. It moves us from a functional-centric approach to a product-centric one. It integrates manufacturing as a critical aspect of product creation so that how to manufacture products is not an after-thought once a product is designed and engineered. It will improve both productivity and quality.
Hi, I’m Dr. Michael Grieves. I’m the Chief Scientist for Advanced Manufacturing at the Florida Institute of Technology. And my discussion today is on the digital twin, enabling 21st-century Manufacturing 4.0. What I’m going to talk about is industry, or 4.0, or Manufacturing 4.0, or Smart Manufacturing, or Industrial Internet of Things, or IoT, or however you want to call it. But it’s this idea that we’re in the fourth wave of the Manufacturing era. And just to give a brief reminder of this is that the first stage or era was machines, steam power, central power shafts. And so this sort of took place after the steam engine was developed and, and occurred in the 1800s. When we moved to the beginning of the 20th century, we started to have electricity, we have individual motors. And then clearly the idea of mass production introduced by Ford Motor Company of my hometown, Detroit was a particular value in terms of being able to manufacture the third era, it gets a little fuzzy, but we can consider that as computer-based PLCs, and robots. And so this took place in the second half of the 20th century, let’s call it around the 70s 80s, maybe even a little bit into the 60s, but it was this idea that we were going to have robots and computer-controlled machines.
And so the claim is now that we’re in the fourth era, which is smart, connected machines, and cyber-physical systems, and as I will sort of content that is digital and physical twins, and so this whole idea of where the digital twin comes into play, but I’m gonna also talk about the physical twins because I think it gets a little short shrift with respect to how we have to look at smart connected products to basically drive our digital twins. And so this is kind of the presentation piece where I’m going to talk about what do we mean when we have the sort of digital and physical twins. And I’m first going to start off with the idea of the digital twin model. And this is the underlying premise, I introduced this in about 2002, it hasn’t changed much, except for the fact that the technology has made it a reality, as we have had exponential increases in computing technology to enable us to have the kind of robust digital representations that we would need for digital twin. From my perspective, and I’m not big on definitions, because I, I think that, like complex concepts, have a tough time being captured in a few word definitions. And if you believe the old adage of a picture’s worth 1000 words, then I think an idea of a model is, is much more relevant in terms of being able to describe this kind of capability. And so on the on the right-hand, left-hand side, we have the physical space that we’ve always had. And this includes the, the physical products, and in point of fact, as soon as we sort of had an idea about a physical product, we actually had to, to put it into physical form, we either drew it or created blueprints or or even created physical prototypes, where this basically changed as the ability to having this digital information in a computer based system, and that’s on the right side where you can see that we have the digital twin itself. So these are the are the the digital representations of what was on the left hand side. And our focus here was basically to create a robust digital versions of the driven by use cases, because information is granular and we can have as little or as much information about the product on the left side where we have to have the entire product or no product at all. The third aspect is basically the connection between the two spaces. And so we want to collect data from the physical space to populate the virtual space. And then we can use that information in the physical space to to basically be more effective and productive with with with our products. And we can develop them quicker we can manufacture them quicker and better and we can support them. And my focus is going to be more on Manufacturing but but Basically entails the entire product lifecycle. And the what we’re doing is is, is moving work, which we’ve always done in the physical space, from the 20th century and before into the 21st century. Why because bits are much cheaper than atoms, they continue to get cheaper, or atoms continue to get more expensive. And so so we can actually do much, much more work in terms of, of working with our digital twin products. And the ideal is we like to basically create the product to test the product, and back to the product support the product, and then we got it all right, in this digital environment, we would go out and actually move around expensive atoms. The the middle part with the connections is often called the digital thread. Frankly, I’m not a fan of the term I’ve hung on by a thread too often in my career, to feel comfortable, I’d like to be called the digital pipe or the digital eyebeam. But we’re sort of stuck with the with the digital thread nomenclature. So when you hear digital thread, it doesn’t contain any information. It just connects the physical space and the virtual space, my digital twin and my physical twin. I’m going to talk about the digital twin types, because I think that that’s kind of important and understanding that the digital twin encompasses the entire life cycle. It is not simply about connecting a digital twin to a specific physical model. But it’s about being able to use a digital twin throughout the entire life cycle from the beginning of the lifecycle until the product is disposed of. And even then the digital twin can continue on. And so that’s sort of represented in the lower left corner, where I
divide the lifecycle into sort of four areas, I create the product, I build the product, I operate and sustain the product, and then I dispose of the product. And what the digital twin has allowed us to do is to move from a functional centric version, where every area had its own information to this product centric version, where we could basically have the information populated and consumed no matter where we are in the product lifecycle. The first kind of product that we digital twin type we have is what I call the prototypical digital twin, or the digital twin prototype. And this is not the same as the sort of the the prototypes that we make for experimentation. But but is basically all the products that can be made. And so as you can kind of see from the from the visual, it starts off as this fuzzy idea of what we want, we start to build requirements up. And as we move through the the, through the process here, in terms of creating the first physical product, we basically could start to develop and to collect all the information that we need to build that only that product, but the variants of that product. The next kind of digital twin type is the digital twin instance. And now we start to make individual products that we want to track throughout its entire life. And so so as we manufacture physical products, we want to in essence, create the digital twin of it that we’re going to connect it through a slide, we obviously need to collect it at this point in time because this is the as builds and this will provide the base for our digital twin moving forward. And and we obviously want digital twins for for complex products, pacemakers, jet airplanes, nuclear aircraft carriers, industrial machinery, we certainly don’t need them for things like paper clips. So it’s all about, you know, the complexity of information that we want to caption. So these are all the products that are made. And so the focus here is basically that we want it to, to start to manufacture these, these complex physical products. And then we want to connect them through their digital twin, and connect them throughout his life so that we can have information about that product no matter where it is. And we can get that information instantaneously and simultaneously with others. The third type of digital twin that I propose is the digital twin aggregate. And these are all all the products that have been made. And why this is important, is what we really want to do here is to basically be able to collect all this information about the products as they’ve been formed. And the more products that are out there, the more information I’m going to get. And then we can do such things as be able to predict performance. So if I see this sensor reading and that sensor reading and another sensor reading, I know that I’m going to have failure of that particular product within a period of time. And if I’m good enough, I can even put probabilities on it to say, you know, within four weeks, there’s an 80% chance of failure. And within six weeks, there’s a 90% chance of failure. So I can also use it for machine learning. So So Earlier versions of the product can help later versions of the product to, you know, maintain the performance, because they now have gone through the learning process. And they can share that with with, with newer products that are coming down the line. So these are the digital twin types, and these, again, come out throughout the entire lifecycle. And, you know, there’s sort of a misunderstanding that I only can have a digital twin, once I have something that rolls out on the factory, that can’t be farther from the truth, I have my digital twin first, because I can do a whole lot of work in the virtual environment that will save me time, energy and material. And so speaking of that particular issue is why is information important. And, and I think this is not as well understood as it should be. And that the idea that information is a trade off for time, energy and material. So if I take any particular task, and let’s take Manufacturing, since we’re sort of talking about that, the idea of Manufacturing a product, I could have it in my factory, and I can divide that, that task of Manufacturing that product into do two distinct areas. One is the the most efficient, effective way, using the minimal amount of, of physical material, and, and physical resources as possible. And that’s the green bar down at the bottom. And I can cost that because I can cost time energy material, the red part is all the information and efficiencies that that are over and above that green area. If I look at what the use of information,
information replaces, not the physical resources, I need to complete that task. So if I was omniscient and omnipotent, I’d only want to do the stuff in the green and to minimize the use of resources. But but the blue information doesn’t replace that it basically can replace the information in efficiencies. Now I show it replacing all the information and efficiencies. And that clearly is not the case. This is the ideal. But it’s the idea that that I can can replace information in efficiencies with with the information. And the issue is, is that I don’t have a unit of measurement for information. So so but I can cost it with a proxy of the cost of my hardware, software, people’s time and organizing information. So I can get to that. Why does this work. And when and this works, if I have sort of repeatable tasks, which I usually do in Manufacturing, and if the cost of the information over all the time for the task is less than the cost of wasting resources, physical resources, I’m basically going to save money and in point of fact, make go far, far less mistakes in the physical world. And for those of us that may or living in the information system business, this pretty much holds true in terms of using information to replace wasted physical resources. And this is the underlying premise behind not only the digital twin, but pretty much all other information system. As I said, I don’t think that that is particularly well understood. Lee Manufacturing, which is this idea of the minimal use of resources, you know, the digital twin, I will contend probably gets us closer to lean Manufacturing, than moving things out around on the actual Manufacturing floor and expanding energy because we can do this in the virtual world and basically be able to model and simulate this information and be able to to come up with the leanest way of Manufacturing products without having to expend that time and effort on a factory floor. I’m going to talk a little bit about sort of the digital twin model of the factory because that’s sort of one of the more extensive uses of the digital twin because not only requires the digital twin of my products, which I have on the far right and left hand sides, but it also requires the digital twins of the means of production. So don’t know Yeah, good to basically create this wonderful virtual product, if I can’t actually make the physical version of it. Otherwise, all I have is a pretty picture. And so so what I want to do is to basically use the information and and have my digital twin of my machines be able to then send that information to the physical machines to create that the physical product. And the ending result here is is while I use my digital twin prototype, the prototypical digital twin on the right hand side. What I want to do is as I’m making physical products, be able to create the digital twin instances I talked about that match up with the individual products that exist and so so so this whole idea of the digital twin model for the factory is important. In fact one of the first papers that I wrote on the topic, but it has garnered great attention because it is a great use of digital information to to Minimize resource use in a factory. And quite frankly, the ideal
situation in a factory is it’s Groundhog Day, assuming that, that the day was the minimal use of resources, the the factory manager wants every other day to look exactly like that. And so so this helps that in terms of being able to capture that information, and to be able to use that information to to produce products in an efficient and effective fashion. This is industry or Manufacturing 4.0, as I say, it’s called Smart Manufacturing. And it uses the industrial internet of things because I have to send to these machines, in order to be able to feed that data, that solid arrow into my virtual world, because I’ve got to collect that information, and be able to turn it into, I gotta collect that data and turn it into information that I can use in order to replace wasted resources. So in terms of the factory itself, again, the first sort of use of it, and the first one I propose is what I call a factory replication. And that’s the idea of, if I can capture this information, if I can have that data flowing into my digital twins of my Manufacturing process, I can have the information about the factory, no matter where I am, and it could be instantaneously and simultaneously across doesn’t matter where I’m at. And and and if you’ve been on the factory floors, you’ll notice that the the the factory management is up above, so they can see this actually gives them the visualization. And in point of fact, they can collect data information off the factory floor and merge it with their simulations, and be able to understand exactly what’s occurring on on the factory floor. And so so this idea of replication, I think, is an interesting one, and was sort of the first ones I proposed for, for factories using digital twins. Where I really would like to go with this is the idea of obviously replication. So quite simply, I talked about the idea that we could replicate this information, and have the the information that I would have to gain from the factory floor by being in close proximity to it, but I also gonna have information that I couldn’t even get so so. So how is the machines producing? You know, is there any sensor readings that are causing a problem. And so what I would like to do is, in essence, I like to engage what I call a front running simulation. This is the idea that says that, that if I have this information, I can merge my my collection of real time data with my modeling and simulation to predict the future. And so that every time is zero, I want to run a new simulation to see what’s happening. So So is a bottleneck occurring, because there was a glitch in one of the robots is, are we going to run out of material at a particular station based on production. So so I would ideally like to predict what’s going to happen in the future with respect to, to my digital twin of my factory, and, and production, where I sort of differ from industry. 4.0 is the idea of industry. 4.0, as you can see from this, from this chart, is about the idea of decreasing the amount of time from a some sort of failure to remediation. And my position is I think that’s good. I mean, you really want to do that if you can, but I would like to eliminate the problem altogether, if I could predict that there was going to be a failure. And and I could basically fix that failure before it even occurred. So I could bring material to a station that was going to have a stock out, I could readjust my assembly line if I knew that there had been a glitch on a on a on a robot that was going to cause a bottleneck later on. So so it’s sort of the difference between the the digital twin concept and industry 4.0 there’s a lot of similarities there. But the digital twin first of all allows me to implement industry 4.0 but secondly is I really would like to get out and predict that particular problem. rather than wait till the problem occurs and an attempt to remediate it even quickly. Even a quick remediation means that there is going to be increased cost of not having production up to my Groundhog Day of being perfect. So this is a quick idea of the digital twin and its use in in the in Manufacturing 4.0 industry 4.0 and I’m going to leave you with a couple resources that you can can use. I have a couple book chapters. The first one is was probably the seminal chapter on digital twins in it. It’s in a book called transdisciplinary complex systems. And the second is one that came out in 2019. And it’s in a book called complex systems engineering. theory and practice, you can find that chapter on researchgate under my name, but if you if you would like, you can simply send me an email, and I’ll be happy to send those chapters. So with that, I’d like to, to thank you for your attention. And I hope I’ve at least given you a pretty good view of Manufacturing 4.0 you know, the use of the digital twin which I think is sort of critical in order to be able to implement that and and our ability of an essence being able to decrease the, the the waste of physical resources by using a digital twin. So thank you very much.
Get full Q/N Access
Sign up to Q/N with a few details to watch this presentation.