Brad Rothenberg: nTop Removes CAD's Limits

We sit down with Brad Rothenberg of nTop to explore how implicit modeling and signed distance fields make computer models that are robust, physics-aware and ready for fast iteration. From aircraft wings to heat exchangers and turbine cooling, we show how fields, splines, and optimization unlock design spaces that B-reps can’t handle.

• Why B‑rep models fail under topology changes 
• How signed distance fields encode geometry and space 
• Spline-driven aircraft surfaces and robust lofts 
• Custom blocks for reusable parametric assemblies 
• Duct and inlet optimization tied to flow targets 
• Integrated CFD and meshless solver connections 
• Heat exchangers for 3D printing and AI surrogates 
• Turbine blade cooling strategies and manufacturability limits 
• FEA, topology optimization, and nTop Connect SDK 
• Design sprints that compress vehicle-level development

Transcript

Roopinder 0:20

Hello, and welcome to FoDES, the future of design and engineering software podcast. My name is Roopinder Tara. On the show, we will have guests that will discuss tools and technology that engineers will find interesting and useful. It's my pleasure to welcome today Brad Rothenberg as a guest. Brad is a CEO and founder of nTop I've nTop since the time it was called nTopology. My first article was after meeting Brad was in 2017. In an ensuing conversation, me being frustrated by what CAD could not do and what nature could do quite easily. I've been most impressed by Brad's approach, an alternative to conventional CAD modeling that in some ways can duplicate the real world, not the straight and perfect world of CAD. Brad is not an engineer by education. He has a bachelor's of architecture from the Pratt Institute. But engineering is to where he sets his mind, and end top solutions may well be the future of engineering and design software. Today I have Brad Rothenberg. Brad, you go by Brad?

Brad 1:33

I write Bradley, but everybody calls me Brad, and I'm okay with that. So you can call me Brad.

Roopinder 1:38

I've always called you Brad, so it's good that you accepted. I always call you Roopinder. That's good. We've known each other for very long. I remember the first time we met. You actually came and sat next to me at a breakfast. I think I was with Brad, another Brad, Brad from COFES. And that was my I think my first introduction to nTop and topology was called back then. I looked at it as, oh, this is another generative design software. But the stuff that nTop could do was somehow more sophisticated. I think you've taken the flag, as it were, on generative design, but you do a lot more than that. I want to talk about that later after you show your software, but use this thing called sign distance fields and uh as your core mathematics. And that's unique. I know a couple other applications that use it, but in design and engineering software, it's very uncommon. I want to discuss that with you a little bit later after you give the demo. But I've been very impressed. Small cohort of people that are changing the way design is being done, or let's say try to change. I know engineers are very difficult to change, right? Once they get something, it's very difficult to talk them out of it. You're changing the way people design in a more traditional way, using CAD, making CAD easy to use, but you are changing everything. You are changing the way geometry is used. So tell me a little bit about that.

From B-Rep Pain To Implicit Promise

Brad 2:55

Yeah, that was my thought early on. Back in the day when we met, I was still obsessed with two things airplanes and CAD, which I've been obsessed with since I was a kid. My goal has always been like, how do you actually represent something as complex as a full aircraft in a computational model that's really close to what gets built, right? The high fidelity model. I grew up hacking into CAD systems and I was very familiar with the core data model, like using analytical surfaces, conics, NURB surfaces, constructing a B rep model out of these surface attaches that have to fit together with edges and topology. At the time, 3D printing was just becoming really popular. And I knew that as a foundational data model we use to represent 3D shapes on the computer, it wasn't really holding up. Like when you make changes to a B rep, the B rep is fragile, it breaks the design intent. When you construct a B rep, the design intent is not direct. It's more like you're capturing geometry, not necessarily the design intent in that geometry, but it's like drawing operations. And that would lead to some issues downstream. If you reference an edge and add a fillet to it and that edge disappears, where does that fillet go? Or if that edge gets broken into two edges, what happens? The name of the company, N topology, in the early days, started with, hey, I want to make software that allows you to represent components regardless of the topology. Because when you have these topological changes, like when you punch bowl and things, the models would tend to break. And with 3D printing, you have these manufacturing technology that can do this. Now, what's amazing is 3D printing pushed us to develop and take to market literally the world's most powerful geometry engine using sign distance. I didn't when I started nTop, I didn't know that we were going to use signed distance fields for geometry. So we're exploring everything at the time. There was one night in 2017 or 2018 when I wrote an SDF generator and render directly in GLSL code on the GPU and saw like how you can still basically you capture the parametric that's in the code that produces the geometry. And so there's this direct connection between the parametric model you build and the requirements that come into it. And it's very robust. You don't have the issues, the topological issues. We've been spending a huge amount of time making the interface between a signed distance field-based model and a BREF smooth so that customers could build parts and manufacture from nTop models. Only in the last year did we figure out how to produce clean lofts at a signed distance field and to drive the SDF model from splines, which are something engineers work with all the time with defining shapes.

Roopinder 5:26

Dwell on that a little because that was very interesting to me when I went to the design computational design summit you had a couple of months ago. You introduced that whole symposium about aircraft design, and you said your first love was airplanes.

Brad 5:40

My mom has a drawing I made of an SR-71, actually a YF-12 variant from when I was eight. It's an ugly drawing, but I was obsessed with it when I was from the age of five to even now, I'm still obsessed with aircraft.

Roopinder 5:52

You have a lot in common. SR-71, the Blackbird. That's a beautiful plane. To this day, it has not been eclipsed in terms of what it can do. That's made before CAD. I'll recommend a book to you, by the way, about the making of the Skunk Works. I gotta give a little story here to the audience. We share a lot of the same books. We also share a basic philosophy that CAD imposes a tyranny of design upon the designer. Like we can only make things that CAD can make. It's gotta be an architect. Okay, the only buildings are gonna be made out of four by eight sheets of plywood and two by fours if you're making residential buildings, right? CAD's the same way. You can only design things that are made of Boolean operations and B reps, right? In the world around us, that's just not the case. That's not the way nature makes things. I always loved his book about fluid flow. All it did was show smoke trails. It's just photograph after photograph of smoke trails. And to me, that was like, wow, if I could visualize the way things should be designed, and I wasn't held to the tyranny of CAD, right? What a different world this would be. If I could look at the way flow is moving, I wouldn't have chunky looking trucks on the road, for example. It was liberating to see that. And so anyhow, I thought I'd go at that, but I just wanted to share that.

Brad 7:06

I think that was one of the best moments we had when we made that connection. You were like, I have a book for you. It's my favorite thing because I still have that.

Roopinder 7:12

I was so happy that it landed in the hands of somebody that could actually use it.

Brad 7:15

Reference it all the time. It's one of the best books. As somebody who makes engineering software, I think one of our roles was to make the invisible things visible. And that's really stuck with me because when you're dealing with engineering, when you're building a model, there's all these forces driving the shape of that model, right? When you have an aircraft, you have the forces of physics that are making that shape the way it is. You're trying to optimize parameters so that you can meet some mission requirement. And a lot of times you need to visualize what those things are so you understand the impact of the design on those aspects. I always think about that. When you're working with sign distance fields, the math represents the entire space around your part as well as the part itself. And so when you're visualizing these models, you can really understand the impact of physics on the model because everything is controlled through what we call sign distance fields. The fact that a field is just a bunch of numbers in space that are changing. And those numbers could be vectors of fluid flow direction. And maybe you want to orient things based on that. Those numbers could be temperature, et cetera. And so everything in nTop, when you're modeling, you're using these fields to control and generate geometry, which makes it really robust, really easy to tie to physics, great for the new world of AI engineering that's happening because all the requirements are captured in these models. We've been seeing enormous success in the deployment of our software for building aircraft in the last couple of months. And that was really spurred by this introduction of spline.

Roopinder 8:40

Distance fields are the way nature works. If you look at a shape that forms in nature, whether it's a river or bone growth after a fracture, it's influenced by the field around it to take its shape. Not springing out of God's mind, right? This is actually forming because of its surroundings.

SDF Breakthroughs And Robust Parametrics

Brad 8:57

It's usually influenced by what the CAD tool is capable of doing. When you look at the shapes of cars, pre-CAD, you had all these loopy curvy cars in the 30s, like when we invented when messy curves were invented and stuff like that. And then when CAD came along, you had all these boxy cars. I think the F-117 is an even better example of this. When I was a kid, I thought stealth means sharp angles. And it turns out that's not the case. The F-117 only has sharp angles, but the CAD tools at the time and the analysis tools could only represent a certain number of mesh elements. So they were limited to modeling based on the tools accessible to that. I've now seen it successfully deployed by our customers in aerospace and effect. My hope is always okay, we're bringing a new capability to the market that will enable new designs to get to the market much faster. And so we're trying to not just incrementally improve a B reps and make there be less clicks and SOLIDWORKS to make something happen. But what we're trying to do is deliver a technology that can give an order of magnitude and capabilities so you can create these really robust, high-fidelity models. The tools that we're using in the traditional engineering stack being pushed to the limit in capabilities. Like we've basically the improvements coming in terms of a traditional built-on of B rep model have only incrementally improved over the last years. And we haven't had a new technology that's allowed us to make an order of magnitude leap and improvement. Airplanes and CAD have always been what I've been obsessed with since I was a kid. My goal has always been to create higher fidelity digital models for aircraft. Why aircraft? It's because I think it's an aircraft design where ambition is pushed completely to the limits. This is an industry where systems and parts have to work, driven by physics, the shape, it's really this function and form overlap. It's a highly regulated, highly conservative industry. There's airworthiness requirements, et cetera. And we've seen a lot of uptick in the adjacent industry. We have a lot of customers in high performance automotive and F1, with a lot of customers in turbomachinery. What's happening throughout the industry is they're being pushed to deliver much faster. So the design cycles, they're shrinking, and that's amplifying the lock-in risk early on in a program. As you try and hone in on the right design, if you get locked into the wrong design, that could have massive budget implications and massive cost implications later on. The decisions you make early in a program could lock you into certain decisions later on. The problem that we're set out to solve is how do you go from requirements to design as fast as physically possible. To do that, we've introduced a new modeling technology, implicit modeling, into the industry. And so I'll show a couple examples of some models. An example of a really simple parametric wing model in nTop, if we look at this model for a second. So this model is built, it's a smooth. You can see if we look at the curvature here. So this is a wing that's parameterized by the span of the wing. It's just a wing. You have a leading edge sweep, a trailing edge sweep, and a root cord going in. And so the taper ratio is a fallout from these two inputs here. And so I could make this much smaller, I can make it much bigger, I could change the leading edge sweep. You'll notice here the wing actually inverted and flipped. The model didn't break when it did that. If I make the trailing edge sweep 58 degrees, also it'll come back to a Hershey bar here. This wing itself, like everything that's generated here, is what's called a signed distance field. The math underlying this, if we look at it in section, let's actually just put zero degree sweep on this for a second.

Roopinder 12:20

Right. Am I correct in saying that you're not actually modeling the geometry itself, you're modeling the field around it?

Making Invisible Physics Visible

Brad 12:27

Yeah, you're modeling calculating and creating deep. Now, when you put a model like this, it's actually a fairly simple model for the root profile. I have an outboard profile, which I just made up with a spline. If I wanted to, I can tweak the points for this initial spline. Let me just close this. I have these airfoils defined in normal coordinates. If I looked at these profiles, I can tweak these points and I say let's move that point down slightly, the whole model directly with that new spline. What's happening here is basically there's still parametric, the parametric definition of the here we have a curve for the leading edge, curve for the trailing edge, and two different profiles at the root and the tip. And then there's a blending function between them. And when you're building this model at nTop, you're essentially creating a workflow in what we call the notebook on the left-hand side. That notebook is where you define the parametric relationships, which allow you to create a model where you have a wing span of whatever you want, believing edge sweep, let's do 25 degrees here and zero degrees for there. We could increase our root cord to eight feet. This would always update. When you're working in nTop, you're defining a parametric workflow. Now, this essentially is like a computer program to generate this wing. And so you'll notice I have an output box and input functions. I could add a box. I could also import this wing. Let me just see where that is. I think it's just on my desktop. I could basically import the actually that's not this right wing. What I import. I'm gonna save this one on my desktop. I'll just call this wing Roopinder's wing. Oh yeah. And so now if I just go find, I could bring Roopinder's wing in and let's see if that's oh, I need some of these to make these inputs up here. Maybe I'll make the blending function an input as well, but we'll just keep this for now. Okay so now I could do file import, and now I have what's called a custom block. I have this wing testing 1.01. This now is just a function for this wing. And I have the span here, so I can make that 50 or whatever, 120. I can tweak my leading edge sweep 60 degrees. Obviously, that's gonna create the bow tie, which is this would just destroy any other model here, it just updates. And it knows it's still bow tie, so you can check for that, but the model's not gonna break or fail. So you can do big design sweeps. Now, what's nice is by constructing these, we call these custom blocks. And so you could have libraries of these functions that allow you to basically assemble an aircraft like this. Here we have something like a CCA type vehicle, and again, everything here is controlled by spline. So if I wanted to go tweak the lofted fuselage, maybe this nose is a little too droopy in the front. I can make that to like negative five, and the whole model is just going to update based on that. And so everything stays very rigidly parametric and robust, and the sign distance field-based model is what enables that in the background. A lot of these will update in real time, also. So you can see.

Roopinder 15:21

Is this what you used in the airplane design seminars symposium?

Brad 15:25

It's a version of that. We've been constantly improving those models over time. Okay. This is just like a clean OML. There's an inlet duct in here as well, which you can see that's flowing through. And maybe there's not enough capture area there, which it looks like. So basically, I have a capture area point in here. If we look at the duct, this start point, and maybe I just move that up a little bit. And so any number in nTop, you can tie into an optimizer or connect into a workflow that allows you to basically figure out what those right, what that right parameters are. And so what becomes really interesting is when you actually do calculations on these models for performance, you can do things like set up optimization routines on an inlet duct. Here we have a variable inlet duct, and this duct has a cross-sectional area going through it. The red is the cross-sectional area through the geometry on the screen. The green might be the cross-sectional area that the propulsion people want because they need the air to slow down and then speed up as it enters. And you can hook this up to an optimization routine, which allows you to punch in the new model or a new curve. And now it just changed the offset function in order to meet those requirements.

Roopinder 16:28

Oh, so you're actually synthesizing the design then.

Brad 16:31

Yeah, it's really here. We can see now there's more capture area. We can take measurements of that, tweak that, etc. Looks like if you're a ramjet type aircraft.

Roopinder 16:40

It could be.

Brad 16:41

There's one we were doing with a company called Spectre, which I don't think I have the model to show, but maybe they'll be able to show that at some point in the future. This becomes really interesting when you start to tie in the physics as well. We have our own fluid solver. So if you can hook these models up to our lattice Boltzmann solver as well to run fluid flow, you can also call out to the luminary cloud platform. It's the combination of the geometry and physics that makes this work. So the again, we have our own CFD tool inside of nTop so we can visualize some of those fluid flows.

Roopinder 17:12

But what's is that something you acquired or did you develop?

Brad 17:16

We acquired a company in Germany called Cloud Fluid that had been developing this that we were working really closely with. And what we found was we were testing the different CFD tech out there, and we were looking for a really fast solver that didn't require us to mesh our model. Because you'll notice with these models, if I turn on the kind of curvature view, these are very smooth models. You can control the curvature, you have a very fine control over the curvature and the blend functions that are happening here. Really important when you're working. Now, that's because these models are mathematically precise. When you're working with a mesh or even a B rep that's going to be meshed, the B rep itself is usually mathematically precise, but the CFD solve requires a mesh to be made, so you have to create that mesh. We're like, okay, is there a method that doesn't require a normal mesh to be made and building it?

Roopinder 18:04

I have to point out that something like this would be very difficult for ordinary CAD modelers to create and it would take something like Alias or CATIA to generate this, and it would not be easy. If I need to sheet for CG, do you how's the audio coming in on your side? You were cutting in and out on mine. Sounds fine. Is mine not good? Yours is not good. Is it no? Yeah, now it's good.

Brad 18:28

I guess Zoom is using the GPU for sound because if I switch to adaptive rendering, is my sound ban it now? Interesting. I wonder if that's it. I wonder, yeah. That's what I was thinking, but Okay, okay.

Roopinder 18:40

Oh so this is it's very impressive in the aircraft shape. nTop has found a particular niche though. I was impressed how heat flow exchangers had that whole industry, if you can call it that, that's been a specialty. It seems perfect for that, and it's been adopted for heat exchange, right?

Brad 18:58

Yeah, I mean, we have been adopted in kind of two totally different parts of the engineering workflow. Uh-huh. Historically, we were always in detail part component design for 3D printed parts and heat exchangers. 3D printing was almost made for these heat exchangers because you have these heat exchangers that were traditionally made by braised fins that have to be assembled together with thousands of pieces, and they're really hard to put together. You can 3D print them for cheaper and have even more effective heat exchangers. We have a number of customers across aerospace for cold plates and automotive for cold plates and heat exchangers. There's a heat, there's uh commercial aircraft heat exchangers where they're building these computational models in nTop that basically allow you to have a very fine-grained control over the fluid flow and to control the pressure gradient of the heat exchanger. Carlos Mendez at Lockheed gave a presentation at our event. I think it was right after mine. There's some really good stuff that we're releasing on the blog around that work specifically. What's really interesting with the heat exchangers is that you can actually train up these AI surf physics prediction engines. So you can basically run high fidelity CFD in parallel via Luminary Star or via Fluent and basically train up a surrogate model that allows you to then perform inverse design or do design optimization using the physics AI to give you performance prediction. I think we're going through a transition as an industry, an AI-first engineering process, and everyone's trying to figure out what that is. There's the agentic generative side of the AI and the physics AI side. Physics AI is much more mature in the market. We have customers using the physics AI today. What's interesting compared to CAD is that like the end top model using sign distance fields, it's almost like it's an AI native model compared to a lot of times with training these AI models, you don't have the data that you need to build the training data set in order to train them. When you build an end-top model, it's not just one data point, it's the whole design space. Having this robust parametric model allows you to push a button and train an AI model. It's about using these synthetic data sets that you generate to build real-time performance predictions. And then the other side of the AI.

Roopinder 21:10

You're learning on the flies, so to speak, almost literally with airplanes, right? You're learning the airplane is shaping itself as it's plowing through exactly. And I think you don't need data, just governing principles, the first principles.

Aircraft Obsession And Spline-Driven Models

Brad 21:21

What's amazing is you can validate these models through physic and reinforce them with actual physical tests as well. And so we actually had this like crazy process over the we had this crazy like we had three of the best interns this summer, and they built an airplane at nTop over 12 weeks and flew it. And it was a group one drone. And so what's amazing is they took the it was a sub-scale model, wasn't it? Or it's a so it's small, it's 40-inch wingspan. Okay. You probably saw the first version of this model since then. This was the model you saw probably. That's with George Irving. He's working with us, he's one of the configurators who drew up the X-35 and the YF-22. He's an amazing person to work with. So this model was the fit check version. Over 12 weeks, the team built seven new aircraft with modding it each time. And they crashed four of them since we met at the summit. Initially, the aircraft was sized to be thrown. Of course, like the engines, they're when we measured the thrust of the engines, it was half of what the spec sheet showed. So I can probably send you these videos. You can see the crash one from the throw. The team installed landing gear, but no steering. So they also did ground loops for a while. We even put the car on the runway. I actually hung out in the sunroof to launch the aircraft and got the car up to 40 miles an hour. Plane flew perfectly straight for about two seconds, about a thousand times longer than any previous flight, but then it hit the vortex coming off of the car and just immediately crashed into the ground. So the team put steering on the nose gear, and then finally, two days before their internship ended, 12 weeks into their internship, had multiple successful flights where the plane took off. It's a little intense to take off. Now, what's doing the flight control? It's a remote flight control. It's a remote control, and we were it was in manual mode. The pilot is amazing. It's this 80-year-old guy, Bruce, who's just an incredible pilot. And once he turned it out, it was perfect. And this thing is fast and it landed fast. Penetrator.

Roopinder 23:18

Wow. So it actually works, flies. Oh, yeah, and it flies pretty well. Really tell me again what was the reason why it only had half the thrust, or what did you say? Half the thrust or half the the engines had half the thrust.

Brad 23:30

The spec sheet said the thrust to weight was supposed to be like three and a half to one. The spec sheet said the airplane is like 1.8 kilograms, and each motor was supposed to put out almost like double that. But when we measured the thrust, the motors themselves, it was more like one to one thrust ratio, like 0.8 to 1, which would apply. But throwing it to take off was tricky. So you crashed a lot of throwing them. These were 3D printed. The longest build takes about nine hours. We have one FDM machine at the office. Next summer, maybe we'll get four of them and can print a whole airplane in a night. Wow. We're looking for more interns, by the way. If anybody's listening and wants to apply, we're hiring an aircraft team for next summer.

Roopinder 24:08

Certainly would be very interesting work for sure. And I I gotta I'm gonna suggest that you market these videos because this could be like that Boston robotics company that makes the robots and the dog. They get more views than anything, and maybe even some revenue for the videos for all the fails that they have, which people find very interesting.

Brad 24:28

I think it's important because the lesson is that valid exploring the design space, failing fast and learning faster. As toolmakers, we should provide the tools that make it easy to fail fast and easy for you to learn. You should not be like you have to, it takes so long to draw and you get one shot. It should be they should be driving these like iterative processes.

Roopinder 24:47

Yeah, the design exploration, that's really the key here. You're not just designing, you're exploring the design space. That's something we as engineers don't do. Ideas, the shape forms come out of our head in its final shape, and we're pretty much stuck with it. We don't explore all the possibilities. We have this conceit, maybe it's based on experience, that we have the right shape and we're not willing to give that up and explore things. Who knows? We might discover something. This is very good for design space. We mentioned heat exchange, but one of the most interesting applications of heat exchanger I've had, I've heard of, was the cooling channels and turbine blades, because they can only be 3D printed. And there's no other way to do that, but that allows the turbine blades to, because they can be cooled by this micro channels in these intricate shapes, like they operate in higher environments and faster rotational speeds. Is that something that you've Oh yeah, spot on?

Brad 25:39

I think it's probably one of the most interesting applications in our software right now. We've been working really closely with the team at Siemens Energy on some of the engines they have at RTX, at Pratt Whitney, and at GE. There's been phenomenal work that's been done on the turbine blade side, which it's really important because that first turbine after the combustion really hot. And the hotter that you can do the combustion, the more efficient the engine is going to be. It's almost like you have every physics trade going into that one blade design. You have to manufacture at the highest quality material. So they do single crystal casting. The metal is literally a single crystal that's grown as it's solidifies the metal, solidify as it's in cast, investment cast. But you also have all of the thermal drivers, so you need to keep it cool because any metal that you put in there is gonna melt. You need to actively cool it. Then you need to make sure that it's capturing as much energy as possible.

Roopinder 26:31

So it needs to be efficient from an aerodynamic standpoint. It has to cover all areas of the blade that are gonna get superheated. Maybe I did see it at some it, but it you have to reach into the area where it's not being cooled and where it is getting really hot. You have to reroute the circus. I had a question after that. How do you account for that being a viable channel? And how can it be produced? How does it actually get produced in 3D? Do you have to shake that material out? The hot section turbine blades are cool.

Brad 26:59

Yeah, as far as I know, the ones I've seen are all investment casts. I don't know if I've seen any that are 3D printed directly. They might have some that are 3D printed, but for the cooling, I've seen very complex heat exchangers with tubes that are like crazy shapes. And the way they get the powder out is they just vibrate. They put it in a vibrating machine and then they run fluid through it and they make sure that the fluid doesn't have particles in it. I don't know of any production hot section blades that are 3D printed directly. Just because the material properties you need on those blades, they really need to be a single crystal cast. Usually the crystals you get from a 3D printed structure, like you have these kilometer column grains, that's usually not good enough for the process.

Roopinder 27:40

So the optimum cooling channel has not yet been created for these turbine blades.

Brad 27:45

If you were to each somebody with their own really big unique proprietary cooling the blades because it's such an important problem that they've been working on for so long, I think the the question where I've seen nTop add a lot of value is in how quickly you can iterate and explore and understand the correct cooling strategy. There's all sorts of details in these things. How do you texture the wall to provide turbulent zones? How do you get laminar flow where you want laminar flow and turbulent flow where you want turbulent flow? And so there's all sorts of like geometric solutions. If you know how the fluid flow is flowing, you could place things in that bacteria field and control it in nTop, which are some of the coolest workflows I've seen in the software. I might have a chill plate that looks that's a nice actually Pick our demo file on the home screen of nTop. Does that okay, okay?

Roopinder 28:30

I've seen it.

Live Wing Demo And Custom Blocks

Brad 28:31

You see it's that anomaly. But it's looking at the basically placing structure to flow along that curve, and then we're measuring the flow. But you can use this flow to do all sorts of stuff, like where it's lower velocity, you want to make it smaller so that the velocity goes up, or you can add turbulators where if you want it to be a certain way.

Roopinder 28:51

Oh, turbulators meeting things that'll make it. Turbulence. Yeah, exactly. Okay. That's the best way to heat exchange, to have turbulent flow where the molecules can wrap around and not go into laminar flow.

Brad 29:02

Yep, exactly. It's very easy to tweak like all these parameters you can tweak and change and model.

Roopinder 29:08

Correct me if I'm wrong, but the randomness that does it could nature, for example, the way bones heal, they attain a final shape. That seems random to me. Is that something that's not been captured yet by your software? Or you can use the noise and do stuff with noise.

Brad 29:22

So if I wanted to make a texture, I could do that. There's algorithms to do like if you school with the motherful device industry does this with their software where they make the bone implants and simulate what bone looks like. I can see if I can find a picture of it.

Roopinder 29:35

I've been fascinated by bone growth, I think, because I've broken so many bones. It just seems to me like that's nature at its best, the way that can heal a bone.

Brad 29:45

And we've not got any control over. We've seen applications like that in our software as well. I would love to see that type of system integrated into aircraft structure, also.

Roopinder 29:56

Self healing? If we could make more robust structures, like structures. That takes on its shape. So much of an aircraft structure is based on how the stress is during landing. The shape could change depending on what kind of stresses it's feeling. And maybe you don't need all that weight equally. The weight you distribute somehow to be in a different place when it's for level flight. So I gotta ask though, you spent a lot of time and had a lot of results on fluid flow and NTAP acquired that fluid flow program. Are you doing something similar? Have you acquired any kind of FBA or structural analysis to help with yours, or are you relying on the shop?

Brad 30:31

Structural analysis inside of nTop. It's just a traditional meat and potato spina element solver. And we have a lot of the normal boundary conditions in that setup. And you can do like inertial relief, you can do on the in the solver. We have topology optimization that we've built on top of that as well that a lot of people know us for. On the analysis side, I think what's most interesting recently is we released what's called nTop Connect. And it's an SDK for basically taking nTop models, calling out to other solvers, and bringing the data back into nTop. There's a company we've been working closely with called Intact Solutions. You might know Vadim Shapiro. He was at our event. They have a mesh-free structural solver, thermal solver, and vibrations, linear static using their mesh-free method. It doesn't require a traditional file method mesh. And so we're really excited about that method. And that's connected into nTop via nTop Connect. So it feels like it's part in nTop, but it's actually calling a partner software. But it's just like I made that building block for the wing. You have a structural analysis set up by NTAC Solutions block and nTop. And you can prepare your model for that and ship it out to InTAC. InTAC solves it and sends the data back. And now you can use that for optimization. So it's really awesome. We've seen other tools connect, like Sphere Polyfhem is another nonlinear solver with contact. We actually have an intern this summer from Vinelli, the Italian company, and they're doing energy absorbing structures for their shotguns. They're using Polyfem for the analysis, connected intend, because it's a very nonlinear contact column, and the results are pretty interesting so far.

Roopinder 31:58

How did they find out about you guys? They've been a customer of Intel for a long time.

Brad 32:02

Okay. And there's there's the when you do PhD in Europe, you can do like a work program where you go for a couple months at a site and do that. And so we've all we've hosted a number of PhD students. Like we give them access to train them in the software and they do some really awesome stuff.

Roopinder 32:15

I'm thinking of other cool applications, like F1 racing cars. Have you done any era work for automotive companies, a racing company?

Brad 32:23

We're in 3D printed components and things like teens. F1 customers. One that we've been really public with is Sauber, who's been speaking about the stuff they've done in it. We're in a lot of other F1 companies as well. And uh it's mainly more for 3D printed structures. They loosen the regulations in terms of integrating 3D printed parts throughout the car. And so we're seeing a lot of interesting work there on the part design area. But I see a lot of these areas, the work we're doing in aircraft design was probably the most innovative work I've seen in nTop since starting the company. And I imagine a lot of now we're gonna see flow down into like on the vehicle level design, on the aero design. I think that'll flow into F1 pretty soon as well. Every single vehicle has one or multiple heat exchangers in it. They're just like hidden. There's a lot of and a lot of them get made every year, and they're really important. That was one of the first big application hits for us. But now the vehicle level design. What we've been doing is we've been doing these design sprint workshops with our customers where we'll send in a few of our best engineers, a couple of the team that works for us, that like a couple of our kind of experts in aircraft design, and we'll go do a three or four-day design sprint with the customer where we'll take the building blocks that we've built already and then improve them, enhance them, update them together with the customer on a real system that they're building. In the last six weeks, we've done five of these week-long workshops, and the results have been phenomenal to see. Hopefully, we can release some of those. Going for the awesome to share those with you.

Roopinder 33:56

Brad, let me thank you for joining me. It was wonderfully informative and quite an eye-opening. We plan on doing everything we can to support us. I think this is very valuable for engineers to explore design. So again, thanks very much. Great having you. I hope we see each other again soon.

Brad 34:11

We're bender, it's always awesome to talk. I'm gonna shoot you a text. We'll catch up soon. Happy to get you more deep dive demos and stuff like that, too. Thanks. Have a good rest of the winter. Bye bye. Thank you. Sure.

Roopinder 34:22

Thank you for listening to Faux Des, the Future of Design and Engineering Software Show, brought to you by Ench Technica. I hope you have learned of a new application or technology that will help you with your job. If you have an application you think would be of interest to other engineers, please let me know by emailing me at Repinder at Enge Technica dot com or message me on LinkedIn.