Release date: 30 October 2020
Episode 36: Ben and Stephen FINALLY are NOT talking about the latest GPUs… sorta. They discuss new phones and more importantly, Folding@home’s latest achievements! Ben and Steve are joined by aspecial guest, Sheriff Woo… Russell Waddell to discuss digital twin implementation. Ben segues into model based definition. Stephen closes a new superconductor.
Subscribe to the Weekly Newsletter https://www.amtnews.org/subscribe
Amateur Machinist Blog https://swarfysteve.blogspot.com/
Music provided by www.freestockmusic.com
Benjamin Moses: Hello everybody and welcome to the special Halloween edition of the AMT Tech Trends podcast. I am Benjamin Moses, the director for manufacturing technology, and I'm here with.
Stephen LaMarca: Stephen LaMarca, AMT's manufacturing technology analysts. And we have a special guest.
Benjamin Moses: Special guest indeed.
Russell Waddell: I'm sheriff Woody, recording live from Andy's room.
Benjamin Moses: Hi sheriff, how are you?
Russell Waddell: As played by Russ Waddell, the MTConnect managing director at AMT.
Stephen LaMarca: Awesome.
Benjamin Moses: Yeah, as we were prepping, [inaudible 00:00:38] suggested I be Jill Buck today. So send your faxes into Jill Buck at AMT if you like my outfit.
Stephen LaMarca: He doesn't get out of that outfit.
Benjamin Moses: Nonstop all day long. Welcome guys, I'm excited to record today. It's fall, and last episode we talked about phones and how I to stretch... Well, we to stretch our phones as much as possible, harness all the value out of that expensive purchase, and now that we've mentioned how the end of life for phones tend to be poor battery usage, all of a sudden I'm sensitive of how crappy my battery is now. So yesterday I was at 10% by 10:00 PM, was like, "Ah, what am I going to do? Let me just put it on my charger right next to my bed." So problem solved. It's tedious. For some reason the battery's dying quickly on me, but that was that one of my many minor problems I've faced this past week.
Stephen LaMarca: I get that too, because I also will put on... In fact it's probably a good idea since it's right next to me. I'll put my phone on the charger and I'll check in on it 45 minutes later and it'll say fully charged. I'll take it off the charger when it says 100%, and then I'll look back at it five minutes later and it says 91%. How?
Benjamin Moses: How does that happen?
Stephen LaMarca: How was it that quick?
Benjamin Moses: Sheriff Russell, how are you doing with your phone? Are you stretching it out? Are you itching to buy a new phone?
Russell Waddell: I'm a new phone convert. I find that it's much easier to sheriff with the latest technology.
Benjamin Moses: I agree with that. We were going to mention some Folding@home efforts.
Stephen LaMarca: Yeah. I know before this we were going to talk about even more graphics card and shenanigans and how AMD Radeon has just announced today their new lineup to compete with the Nvidia RTX 30 series, but we've talked about GPUs for way too much, and just the PC world slam dunking on gaming consoles all the time. But Folding@home, you posted a article this morning to AMT, no joke, 10 minutes before I was still reading that article in bed, and I thought that we need to share this because this is great. But the Folding@home article on Network World, the title is Folding@home exascale supercomputer finds potential targets for COVID-19 cure. And what's really cool is Ben and I have spoken on the program Folding@home before, how it's computational power that is distributed. Basically everybody's gaming rig or home computer can be used to do some computational efforts towards helping cure a disease, by monitoring, by simulating the misfolding of certain proteins. Well, we've spoken about this before, and it's been an entire summer of a pandemic later, and it's really great to hear that it's essentially... It's almost found a cure for the coronavirus, but one of the coolest highlights that I had to mention about this article is it said that Folding@home at its peak had more than 2.5 exaFLOPS of computational power. And I don't even know what that means, but it sounds like a lot, but at one point it does break it down and says, as of late October, now, the computational power, the compounded computational power of Folding@home has dropped to 247 petaFLOPS, which they were all depressed about. But that's still twice as much as, I think, America's most powerful super computer called the Summit. It still dunks on that. But it makes a lot of nerds that have spent too much money on gaming rigs feel pretty good about themselves right now.
Benjamin Moses: It does feel good, because... I'm glad I got back into it. I think I started folding earlier this year, and I think... I've had two computers, so one I use for my own personal streaming and recording and stuff, and I think that exploded, I think because I was letting it fold for weeks on end for no end, but that's just hearsay. But it's cool to see some type of potential contribution to the society from our efforts. So I thought that was really cool.
Stephen LaMarca: Right. And how often have you been checking on it?
Benjamin Moses: Checking on the results or checking on what my-
Stephen LaMarca: What your status, what your computer is doing?
Benjamin Moses: Oh, probably daily. Once a day. In terms of what the settings it's done and my ranking.
Stephen LaMarca: So I've noticed the past probably two weeks, not all the time, but every other day, I'll check in on it and the bar's not moving and the wheel's not turning at all, and I'm wondering what's going on now? What happens now? What do I need to call Sean from IT, figure this out for me? Because it just simply wasn't working on what they call a work unit. And after doing some digging this morning, which is actually how I stumbled across this article after you posted it was I was trying to find out why did my computer stop folding? So many people have had their computers hooked up and working on Folding@home that Folding@home and all of the institutes giving work to Folding@home have literally ran out of work to do so. The number of projects can't keep up with the number of computers that are trying to work on projects. It's wild. And that's why we saw that, that's why the report states that huge drop as of this month.
Benjamin Moses: As a person, I think I'm ranked around five... Low 5,000, breaking into 4,000.
Stephen LaMarca: Wow, dude. That's nuts.
Russell Waddell: We've got some work for them.
Stephen LaMarca: Tell us Russ.
Russell Waddell: I keep hearing about these AIs that are going to help me identify those rotten cattle rustlers.
Benjamin Moses: Well we talked about some MTConnect projects and test bed, Steve. We've got some projects-
Stephen LaMarca: Yeah, man. So I was actually on a webinar call with Russ yesterday at his house listening to a webinar that was basically on digital twin, and I think it was hosted by ISO, because they are looking to set forth some standards to define what digital twin is, and I'm not going to go one more into that because I'm going to let Russ take care of that. But when it comes to the test bed, there've been some things... As I've mentioned before in previous episodes of the podcast, we recently got the end effector, the gripper, and the vision system for our test bed's collaborative robot. And since we've been away from it all summer, I haven't really been thinking about... With no end in sight. I don't know when I'm going to be in the office next to actually do some projects on the test bed. But I was scrolling through Instagram yesterday and I thought to myself, as I'm looking at all of these watch pages, these mechanical watch pages, and these dudes flexing their Audemars Piguet and their Patek Philippe wristwatches, I'm thinking, "Dude, I could use AMT's seven axis collaborative robot as the most expensive and advanced technology watch winder." And I think as soon as I'm back in the office, that's probably the first project I'm going to do. I'm going to program just a short program to basically articulate the arm and hold a automatic self winding watch in place and just basically let it run for a day to see if it can keep the watch working and functional. But that's what I'm thinking about.
Benjamin Moses: There's a bunch of other tests to get us there, we still have a ways to go. One to plug it back in, make sure everything works.
Stephen LaMarca: Oh man. Yeah, we do have to diagnose it potentially.
Benjamin Moses: Yes. And also...
Stephen LaMarca: Dust it off.
Benjamin Moses: Doing some positional accuracy tests and repeatability tests. I think that the watch winding it could be a good short-term duration test.
Stephen LaMarca: But like I mentioned before, Russ and I... I had the pleasure of going to Russ's house yesterday to sit in with him on a webinar hosted by ISO and regarding digital twin. Russ, can you give us a brief description of that? What went down?
Russell Waddell: Well, you lost me a bit with all these Patek Philippes, but I can tell you that, that yes, I can describe the webinar. The International Standards Organization, ISO, is working on a framework document which will be published as ISO 23247, digital twin framework. That's intended to normalize what we refer to as a twin and sort out some of the confusion around what's a model versus a simulation versus a twin and get to a globally accepted definition for what's a digital twin and then how do you actually put one in place for manufacturing specifically, whether you want to wind a watch or other useful things that you might want to do.
Stephen LaMarca: So who all was there? It was really cool. You reported around or more than 500 attendants on the webinar, but I know our colleague Sharub with MTConnect actually had a huge hand in a lot of the projects that were on display there. But the three big companies putting everything on display, I believe, was Sandvik Coromant, Lockheed Martin, and Boeing, right? What did they show off?
Russell Waddell: All three of them were deploying basically examples of the framework in action. So ETRI, a research organization in South Korea, prepared the draft of the framework. And then ISO technical committee 184, subcommittee 5, working group 15 sorted through the draft that the Koreans put together and basically worked that through the ISO approval process. And then once it was pretty close to finished, that was when Boeing, Lockheed, and Sandvik each stepped up and said, "Hey, we want to deploy an example of this on some real equipment." So Boeing was looking at a robotic riveting or hole drilling operation on a wingspan or spar. The Lockheed use case was measurement of holes, also on a wing. And then the Sandvik Coromant use case was looking at tool selection and optimization and wear prediction as well. So three slightly different use cases, but all three of them were taking the architectures and information exchange systems that were laid out in this ISO document and putting each of those into a real world factory use case that was beyond the theoretical and into the practical realm.
Stephen LaMarca: So how does ISO's document look, and by that I mean how is ISO defining digital twin? What rules is ISO putting in place to basically say, "Okay, yes, this is a digital twin, or no, that's just a simulation and a detailed 3D model of something."
Russell Waddell: So this document particularly, the framework is really about exactly how you set things up in a manufacturing environment. So it's saying there's an intent component, so is the purpose for a real time evaluation and control, or is this for an offline evaluation and control? Is it tying together multiple pieces or is it one single component? And a lot of it turns out to be just making things that already had pretty in-depth simulation existing work better together, and also applying a whole host of different standards, including MTConnect, to make it so that these things are scalable and you can actually apply one set of rules that works for one set of equipment that's at Boeing and a different set of equipment that's at Lockheed, a different set of equipment that's at Sandvik, and you're repeating and reusing as much of both the best practices and the technological underpinnings as much as possible. So the other standards in there STEP AP 242 two 42 for models, GYF for quality, it's the quality information framework, it's basically telling you about inspection data, and MTConnect is heavily involved in each of these architectures as well for getting data back off machines. But the theory is you'd be able to use a whole bunch of other standards that basically are a stand in for each of those components, and you'd still be able to get the same thing, because that framework is broad enough to say you need to define data, you need to pass data, you need to have quality data, and as long as you have those components, you can mix and match and still get to this workable end product that solves the objectives laid out in that framework.
Stephen LaMarca: Wow. So I think it helps for me at least to think of it as... There's a lot of products out there, whether it's a five-axis CNC machine or a seven joint collaborative robot that already have, on their own, what may seem, or even apply to these standards as being a digital twin, but a true digital twin, that I think I'm picking up properly is the digital twin of all of these systems integrated together to a digital twin of all of these different units working together. Am I right?
Russell Waddell: Yeah, so for the AMT test bed, you've got CAD models of the robot, you've got CAD models of the Pocket NC, you have an integrated CAD model of the environment. You took measurements and modeled the space, but that's basically geometric models. So you have shapes, you have surfaces, essentially. If you want to add motion models, so a kinematic model that's describing the motion of the Pocket NC or the motion of the robot, or you want to add a functional model of what's the end effector do and what's the difference between the gripper versus the vision system or something like that. As you layer all these other models into your cell, the test bed cell, you get closer and closer and closer to a digital twin. The twin, it has to be a twin of something, and there is no end use case for having a robust simulation of the robot itself if that robot is only ever going to operate inside a work cell or on a part. You have to model the part, or you have to model the end effector that's not part of the arm. Digital twin is kind of an extension of digital thread. So you're tying these things together, and in the end you get kind of the fullest practically useful representation of reality as close to when it's happening as possible, and that's what pushes you over the edge from just a model to a simulation based on moving models to a full-blown twin. But I'm just a simple cowboy.
Stephen LaMarca: Well, I'm looking forward to seeing what... That is assuming I can actually understand everything it's saying, but I'm looking forward to this ISO document to that that will define the [crosstalk 00:18:04]-
Russell Waddell: You'll still get the pictures. If you read the diagram and it says this thing plugs into that thing, that thing plugs into this other thing, and from your perspective as the cell owner, it's sort of like, "Well, who's the expert in this thing? How do I call them? And then what does it look when it's working?" So you basically get each of those little connection points working one at a time, and you get the components plugged in one at a time, and chip away at it a little bit, little by little until you end up with a thing that works in the end, but your job is to figure out what the heck do I want it to do in the first place? Why do I even need a twin of my work cell?
Benjamin Moses: And I think that's one of the big takeaways is... I think you mentioned the word intent earlier in the description of it. That's a takeaway for myself as now there's value being harnessed from these different technologies. And I think for me, that's the biggest driver for continuing to see progress, and I think different organizations will start harvesting value and continue developing the technology because they are seeing value from this development. It's not just a science experiment.
Russell Waddell: I keep getting hung up in technical conversations and technical details, and it's been hard to separate, "Well, this is what we can do, and this is what we could make this thing do." I tend to lose sight of exactly what the business case for it is. So to see Sandvik and Lockheed and Boeing, one equipment provider and two end users, all concurrently saying, "We really want to do this." Particularly combined with looking at Siemens has digital twin for virtual commissioning, instead of sending a guy out to set your machine up, when you buy a CNC machine, they'll basically do it with a combination of an internet connection to that machine and some video. That's an immediate value right now. I don't have to get on an airplane, send a person, take a hotel night. It's very obvious to see that, that business case.
And then FANUC and Toyota also announced back on the IMTS Network a couple of weeks ago during the network week, they launched a digital twin that's basically looking at the controller itself on the CNC. So it's a twin basically as your GUI, so your interface with that machine isn't just G-code flying by faster than you could possibly read as a person right. Now it's here's the model of the thing I'm trying to cut, and that model of the thing I'm trying to cut is inside a model of the thing that's doing the cutting and I can manipulate all of that in virtual space, instead of having to see if this thing I'm trying to do is going to blow up the whole machine.
Benjamin Moses: That's a fascinating idea because if you're able to see it real time digitally, you can do more probing and investigation as opposed to sticking your head in the machine. So that's a fascinating approach. I didn't think of that.
Russell Waddell: Well not to mention a bunch of these functions already exist. So if you want to do crash prevention or look ahead or whatever, these are individual features and functions that have been rolled out for a long time, as soon as somebody said, "Let's have a computer control a mill." That was kind of when all this stuff started to be rolled out. It's tying it all together into a cohesive whole and putting it into some meaningful, useful operator facing or user facing package, that's the piece that seems like it's different now.
Benjamin Moses: Further integration. And ISO's clearly a worldwide organization, but it does feel like it's... With the different global entities participating, it does feel like everyone's taking a little piece out of the pie and moving it forward. It's not just a singular nation or one group.
Russell Waddell: Yeah I sometimes a skeptic about the speed at which ISO moves or at the ability to actually accomplish the mission of getting diverse points of view and diverse global players involved. But this one was definitely a case where you see people from all over the world, both on the writing side and the testing and deploying sides. There were two use cases that were in the US, one was in Sweden, and all the drafting was in Korea. Meetings were held at a variety of global sites and online. The meeting convener is an American from England. It was just very, very, very international in scope, which is what you would expect given the organization, but it's nice to see it actually realized in a cutting edge technology this. Which is why we need to put it on a test bed as soon as possible.
Benjamin Moses: Steve, hopefully you're comfortable with this.
Stephen LaMarca: Yeah, I'm looking forward to it, man.
Russell Waddell: We need to put it on the test bed as soon as possible. See this? This says management. I'm the management. Sheriff gold star means, unless it's roping up the bad guys.
Stephen LaMarca: Well hey man, you get me this ISO document LV 223, or what is it, 23247?
Russell Waddell: You joke, but you had the number. You had the number of the whole time. ISO 23247.
Stephen LaMarca: I'm just trying to make the Alien joke.
Russell Waddell: 8675301.
Benjamin Moses: For those of us who didn't put two and two together, Steve is a engineer in Aliens. Thanks for the practical effect there. Awesome. Steve, any more questions about the digital twin? I'm excited, I see next steps.
Stephen LaMarca: Yeah, I'm really excited about it. It's making a lot... It seems the most... The webinar yesterday was talking about all the work that these organizations, these big companies have been doing towards it. The exciting thing for me is, as every day goes by, a digital twin makes a little bit more sense. So that's a huge plus for me.
Benjamin Moses: For me it's every day the digital twin will solve a different problem. That's the realization, that's kind of a real thing. I'm excited. There's a couple of articles I wanted to talk about. One was a article from SME. They talked about model based definition, and I think it's kind of an extension where we talk about on digital twin. So the article is from... It was written by a guy from Verisurf, but he was talking about the value of model based definition in a small to medium-sized company, specifically a contract manufacturing facility. They're on the smaller side and the value of integrating GD&T, geometric dimension and tolerance, into the model and receiving those models from the customer and not just using 2D drawings, IGES models. And from my days, ages, ages ago of doing some design work, doing IGES models and STEP models, for me back then, this is probably 15 years ago, they're very, very similar. As in, to me they felt reduced, but they've come a long way in embedding digital data and being able to communicate that data in technical packages outside their building, outside their four walls. So it talks about STEP AP 242, which, Russ, you mentioned earlier, about embedding as much data into the digital model and communicating that to your suppliers, so when the contractor receives that they're able to use that data within their process and use that data in their first article inspection. That's the big takeaway is they're automating some of the tasks for first article inspection, which is actually very, very tedious and very laborious because you document everything in your manufacturing process and say, "Here, customer, I've done everything right for the first time, leave me alone. Let me keep building this." But you've done a tremendous amount of documentation, and that was the big takeaway in the article was that you can help automate some of your documentation to supply to your customer by stepping towards using some of the analytical tools or digital tools we have today.
Russell Waddell: That's the vision, right? As you already have a machine that knows what it's doing, why would you need to verify that process separately when the machine's already been qualified historically, and now you can just say that the machine, self-reporting out of the machine, "I did when I was qualified to do at this time. Here's your timestamp. Here's the part program I ran." Et cetera, et cetera. It's crazy that they have to do any of that manually if the information already exists somewhere else in your system.
Benjamin Moses: Yeah, the idea of self-reporting machines is definitely a big thing that I'm looking forward to. Steve you had an article on superconductors that you want to take us through?
Stephen LaMarca: Yeah, I got this really sweet article on a... It's about a team of physicists out of New York that have discovered the first superconductor at room temperature. And before we dive into superconductors and their uses or their usefulness, just to paint the picture, a regular conductor, and we're talking about electrical conductors here. A regular conductor is a material that is very good at, or very low resistant in allowing the pass of electrons or electricity through it. So it's got low resistance, high conductivity. They're inversely proportional to each other. In fact, resistance is measured in ohms. Do you know what the unit of measurement for conductance, which is the opposite of resistance?
Benjamin Moses: Conductance? My initial guess was one over ohms.
Stephen LaMarca: It's mhos. The opposite of ohms. So a good conductor is low resistance, and one of the physical properties of conductive materials is as they get colder, their resistance, their electrical resistance also reduces. But a superconductor, what makes a superconductor super is that when it gets below a certain temperature threshold, when it gets to a certain coldness level, let's say, that's totally inaccurate. But when it gets below a specific temperature, it loses all resistance totally. So it's a perfect conductor once you get it below a certain temperature. And the problem with a lot of superconductors is they become a superconductor, they get to this property that is so sought after at extremely cold temperatures. Well below zero, well below freezing. Really cold temperatures and to get these superconductors to work the way we want them to, a lot of energy rich and energy hog technologies are needed to keep these conductors cold so that this electrical system works the way you want it to. So the big deal here is this New York team of physicists has discovered a superconductor that has it superconductor properties at room temperature, at normal to human temperatures, which is really cool, literally. And I think it's a big deal because some of the technology that superconductors go into are MRI scanners, nuclear fusion, and what's another one... MRI, nuclear fusion, and I had a third one. Come on, man.
Benjamin Moses: That's all right.
Stephen LaMarca: What was it?
Benjamin Moses: So if we do go to room temperature... It sounds like things were working using liquid nitrogen or something, advanced cooling. If we get room temperature, that sounds nice, but what's the benefit? What do we...
Stephen LaMarca: Well now you're not using as much energy. Now you have a more efficient device, theoretically, because they haven't been made yet. The thing was just discovered. But now you have a more energy efficient device because now you don't need to dump a whole lot of energy into cooling this one component that makes the thing work the way it's supposed to work.
Benjamin Moses: Gotcha. So you could free it from your advanced cooling needs to maybe be more portable or not using tons of power. That makes a lot of sense.
Stephen LaMarca: Probably makes it smaller, lighter and more energy efficient.
Benjamin Moses: So maybe in a computer?
Stephen LaMarca: Exciting times.
Benjamin Moses: Folding@home with a super computer, super [crosstalk 00:30:59]-
Stephen LaMarca: Superconductor super folding.
Benjamin Moses: Awesome. This was a very exciting episode. I'm excited to be... A special Halloween episode. We've got aliens, we've got some sports, we've got Cowboys. Where can they find more info about us?
Stephen LaMarca: You can find more info about us and a lot of the exciting trending technologies in our industry at amtnews.org. Go ahead and subscribe
Benjamin Moses: Russ, where they can find more info about MTConnect?
Russell Waddell: Mtconnect.org. That's the place.
Benjamin Moses: That's the place. Awesome guys. This was a very exciting episode and thanks everyone.
Stephen LaMarca: Bye.
Russell Waddell: Happy trails. Sweaty Joe Buck and weird alien guy. Till we meet again.