0:01 Good morning everyone, Danan, thanks for having me back this year, I see I’m not the only one presenting my problems this time, so that’s a good sign, but no, actually I I really love this conference, and and I’m glad to be here just be, due to the authentic nature of the conversations that come out of it, you know, it’s a, it’s a bit less salesy, which I think as engineers we we all appreciate. Let me get into the view here. Yeah, so for those of you that don’t know me, I’m Chelsea Cummings, I work for the Barnes Global Advisers, I believe last year in my presentation I said “please automate me out of the process”, so good work there, we saw a lot of that in in some of the previous presentations from yesterday.

0:51 So working in an applications engineering capacity at at TBGA, more or less I take Andreas’ crazy gyroid designs and translate them into written requirements for parts, that’s one of the many things, the other thing is I keep a a pulse on the software offerings that all of you are working on developing, so another reason why I’m happy to be here. My core experience is in on the manufacturing, for being close to the process, designing the parts, doing the build preparation, sometimes running the machines if I had to, so that’s where a lot of my perspective comes from.

1:29 So I wanted to stress today, much in line with the theme of this morning, let’s design parts not shapes, and rather than telling you how to do it, we’re going to take a look at sort of like a zoomed out view, and the drivers behind that, from an AM economics standpoint. So rip 3D printing, long live AM. John Barnes, our founder, and cooworker Tim Simpson, they wrote a sort of cheeky article and 3dprint.com in response to Youris Peele’s “rip 3D printing” series, if you haven’t read it, it’s amusing if nothing else, but I wanted to carry forth the essence of that article with the addition of some feedback tuned for this group.

2:15 And in the article they basically describe 3D printing as a child star beyond its prime, so it it peaked kind of early, but we want to maintain that momentum, right, and when I was trying to think of anal, an analogy for this, unfortunately I I couldn’t help but think of like a sort of offensive one, I hope you guys will will stick with me here, and let the takeaway be not that I have anything against the Culen family, fan of all of their works, but that they actually presented a really good analogy. So yeah, we know which one we don’t want to be, right, the child star that we haven’t seen since then.

2:55 So this is my sort of madeup timeline of what the ind industry cared about, where it invested a lot of money since the the start of of 3D printing, right? So in the beginning we had the Makers Movement, we saw some commercialization into metal powder bed fusion, I just chose some photos at random, not not biased one way or another in terms of which machines I’m showing, but we saw some investment happening here. Then we, we scaled a bit, we got machines with more lasers, we have some software that’s actually tuned to the AM process to enable us to do these build preparations and slicing and things.

3:40 And then we, we kept going, more and more money got dumped in, we have more intelligent machines, we have, you know, production style machines where you have all your powder handling and modularity built in, we have a, we have implicit modeling come into the forefront by, you know, some software company we never heard of, just kidding, I saw this from ENTP’s website. So back to my analogy, I’m sort of likening this era to the MCO Culan era, Home Alone franchise, happens, everything’s really exciting, this is fun, we’re we’re having a good time. This era is more along the lines of of Signs, an exciting movie, good entertainment value, a little bit more serious.

4:26 By the way, when I was putting together this present, presentation, I thought this was the same Culen child in these movies, it’s not, his brother, it’s his brother Rory, and then I’ll get to the third one, Kieran. By the way, I have no idea if I’m saying any of their names properly, no response, okay good, I, doesn’t seem like you guys do either, that’s encouraging. I’m calling the later era the Succession era, we’re getting serious, there’s like real money here, the economy is the the VC era is ultimately succeeding, the machines are bigger, faster, better software with AM in mind exists, thanks to all of you lovely people in this room, but those are really just catalysts for a final product, right, they’re catalysts for making parts.

5:14 Of course I had to make it a heat exchanger, to stay in line with the grand theme of of this conference, so yeah, we’re, it’s the, what really matters at the end of the day is the parts, and I liked what Trinkle said yesterday about the the focus had really been on the potential of the hardware, and now the digital has a bit of catching up to do. So of course the the tools that we’re using still matter, but now the focus is is more on the endgame, and I’m likening this to Kieran, Cen, star of Succession, who has a very bright career in front of him.

So the language in the written specs has actually even changed. For a long time the proc, the definition was written for 3D printing as “the process of joining materials to make objects from 3D model data”, so yeah, that’s that’s 3D printing, and and that definition stood for about a decade, and in 2021 they actually changed it to “the process of joining materials to make parts from 3D model data”, so it’s like, okay, we have this exciting thing now, we have real returns to make, let’s, let’s do something with it. So it’s about parts, not the shapes, AM is finally humble enough to say what it is, it’s another method for manufacturing parts.

6:34 So we want to have a good grasp of where AM stands in the ecosystem to get the most out of it, right? So AM isn’t dead, it’s just different and more expensive than the conventional manufacturing methods. This is a study that we did internally at TBGA to find out how much it, really, what, what the dollar per kilogram of forming material with a different manufacturing method is, and this is what we came up with. And so what does this mean? AM’s goal should not be to replicate economies of scale, the goal should be finding the unique opportunities where that value proposition of AM yields a competitive advantage over traditional manufacturing, or in the cases of, if you’re trying to replace a casting, you’re just really desperate and you can’t get it any other way, right.

7:25 But if you intend to compete with con casting longterm for a high mass of parts, just don’t do it, go home, it’s not, it’s not it. So the false economy that was built for AM doesn’t justify a good model, time and money are hard to beat out in the long run, when you, how, when you can make stuff as as cheap as $79 a kilogram with sandc casting. So, and just because AM doesn’t have tooling or molds doesn’t mean that there’s no costs associated, there are costs later in the process and on the back end that we have to watch out for, and that we need tools to help us overcome.

8:02 We have to con, consider our process constraints, like I talked about last year, the modifying for additive manufacturing, so we can actually get the value that’s created by design for manufacturing, and we don’t want to qualify part, or sorry, processes and parameters just to qualify them, and, you know, materials that we think will have a lot of work, if you build it they won’t necessarily come. We need to come at it from a different angle, where we evaluate the parts value proposition and deliver on that. So we’re maturing our design propositions, we’re looking at things a little differently, before we sort of had this bottoms up approach, we would make a cool shape, we’re like, oh, this is interesting, we can print it for the first time ever, then we start thinking about, okay, maybe those shapes are similar to other repeating structures that we’ve seen.

8:57 But ultimately we want our structures to be comprised of shapes, it’s sort of a different way of thinking, we’re leaving behind the bottom up approach, and again our Northern Star, Kieran Kulen, he is our model of good, so we’re entering the top down approach. This is our TBGA AM maturity model, this is sort of a construct we came up with to figure out how do we think about the the things that we’re making with AM, what is it mean, how do we make sense of it, yeah, if you’ve actually seen this before, yeah, we love this thing.

9:34 So I’ll quickly go through it, at level zero we have pre-production, this is where we’re doing our prototyping and testing, kind of have a vague understanding of how to run our machine and our process constraints, we’re just learning. At level one we’re doing, we’re starting to influence production, so we’re making tooling and shop aids, getting a better understanding of our process constraints, how to actually build with this machine. At level two, this is sort of that level of desperation that we like to not stagnate at, because it’s taking parts that were not designed for AM and building them with AM, so you have to have a really good understanding of your process constraints to get anything out of this level, it’s useful at times, but we don’t want to stay here.

10:18 At level three we see more functional designs, we’re stying to get some of the benefits out of out of DfAM, right, so we have part unitization, we’re eliminating joints, we really understand our process constraints, and we’re, we’re starting to do DfAM. At level four, you, you know everything, you, you know how to run your machine, you know your process, and you’re actually designing for additive, you’re getting enhanced performance, you know, this is where Andreas lives, and your optimized structures doing doing great things, so we’re up there, yep, there he is.

Okay, so I tried to narrow down what exactly it is that I’m I’m I’m really aiming to get at, at level four, and that’s, I’m calling those yield and field enablers. So we want to improve our yield and production, make sure that we actually get all good parts off of the machine in batches, and we want to improve our quality in the field, right, so I took a look at that from a couple different lenses, and these are just ideas, I’m hopefully, if I’m hoping if nothing else, they inspire you guys to go forth and make more tools that make my life easier.

11:28 So looking at part consolidation, reduced assembly, obviously reducing our our bill of materials is advantageous from a labor standpoint, from touch point or touch time standpoint, this is really really focused on yield, and a way that we might do this is if, if I could import an assembly and auto assign a centrality score, perhaps based on mass or something, that would be really interesting. By the way, I had never heard the term centrality score, score until last week, it was, I read a paper by Dr Ja Krishnan of the Indian Institute of Information Technology, Designed, and Manufacturing, so I love that term, I thought I’d bring it up today, but yeah, this is really driving at yield from a production standpoint.

12:18 Another idea, fewer failure nodes, so if we could have a way to automatically assess the connection point types that exist in an assembly, this would reduce hardware of course, but I want to know what’s actually feasible and reasonable to attempt to to join and unitize, that that would be hopefully an improvement and yield because we don’t have as much assembly, but also an improvement on the field, because hopefully those joints are failing less. Another idea I had was, if we could pinpoint failure locations in the field with sensors and then use 5G to communicate that instantaneously back to manufacturing, now I’m not a 5G expert, I’m just just starting to learn about it, but it’s it’s had me interest, had me thinking about some interesting things, about as far as, you know, information doesn’t necessarily have to live in the cloud in the future, it it will be on the spot, so that’s driving a field enabler.

13:26 From a pure performance standpoint, in advance vantage of AM is that we can substitute a novel material for better performance, right, it’s easier said than done, because then you have to go build out a robust parameter set and and make sure that you can actually build it, and actually Way King on our team is working on a machine learning application that that can actually do this and and tell you where to start with your parameters, we’re calling it Big Island, if you’re interested in in that, and those ideas please come chat about, chat about it with me after, but that’s really driving at yield.

14:02 The design of physical properties, so yeah, this line item is really everything we’ve seen this morning, which, and and a bit yesterday, which makes me really excited, how do we get those complex simulation results actually integrated into a parametric workable model, and how do we retain those for for rev control, and things, can we translate them into architected materials, it sounds like we can these days, so that’s great. I jokingly put AI, and more AI, I thought we’d hear a lot of AI thrown around this conference, so again, another thing I like about the nature of it, it’s not not all buzzwordy.

14:43 But in all seriousness it, when I saw Sim scales presentation yesterday, I thought that was a really clever use of machine learning to like, and back analyses, for sort of like a quick and dirty assessment, but then I was thinking maybe we could take that a step further, where not only do we have the analysis, but maybe we have some sort of waiting factor that’s been trained on data of AM parts that are actually successful in that application. So we see a lot of success with heat exchanges, for example, that might have a high weighting factor of success, but maybe brackets have more failure rates in the field, and that would get a lower factor of success.

15:26 Thoughts that I’m having that hopefully inspire you guys to make my job easier, like I’m always talking about. So finally I’ll end with a company that I feel like is doing a good job of embracing what I’m talking about here, the new AM economy, and and that’s SpaceX. Unofficial report suggests that in development their their Raptor engine contains as much as 40% AM fabricated components, the engine featured here on the right is a Super Draco engine, a smaller one, but I believe their goal is to get all that tubing off of their larger materials, the Raptor, and ultimately go to homogeneous engine parts built with AM.

16:14 So people are doing it, and some are struggling with the with the old paradigm, but I just wanted to encourage us all to keep trying new things, and that really we’re the only ones holding ourselves back. So, oh, yep, there he is, thank you.