Aaron: Hello and welcome to another episode of Being an Engineer podcast. Today, we’re privileged to speak with Bradley Rothenberg, founder and CEO of nTopology, a company pioneering next-generation engineering software for advanced manufacturing. Launched in 2015, nTopology enables engineers to create complex, optimized geometries, particularly for sectors like aerospace, automotive and medical devices, optimized for additive manufacturing. Bradley brings a unique perspective to computational design, bridging the gap between form and functionality with his background in architecture from Pratt Institute. Bradley, thank you so much for joining us today.
Bradley: Thanks so much for having me.
Aaron: Do you refer to the company as “nTopology” or just “nTop”?
Bradley: When I started the company, we called it “nTopology,” with “N” standing for “any” and “topology,” meaning the mathematical structure of shapes. I wanted to create software that allowed you to make any shape without limitations. About a year and a half ago, after several years of negotiations, we acquired the domain “ntop.com,” and we officially shortened the name to “nTop.” I like to say we lightweighted our name. That said, some of our social media handles still use “nTopology,” so either is fine.
Aaron: You have a background in architecture. How did you move from architecture to creating nTop?
Bradley: I think the origins of nTop go back even before architecture school, to when I was in high school. I was obsessed with our CAD program. My CAD teacher taught me from freshman year to senior year and we were fortunate—this was in Fairfield County, Connecticut, where GE was headquartered and they sponsored the local schools. Even though I was in public school, we had this amazing CAD program. We were learning computer-aided design (CAD) on a 2D pen plotter and I used to finish my assignments early so I could teach myself how to program add-ons to AutoCAD. I even hacked the pen plotter to do some wild things.
When I got to college, I thought the architects were using more interesting software than the engineers, so I decided to study architecture. In 2005, I was working in a digital manufacturing lab and one of the professors said, “We just got this new machine—it’s called a 3D printer. It can consume a 3D file and print it in 3D.” I thought it sounded like the coolest thing ever, so I volunteered to manage it over the summer. That was my first real connection between programming and manufacturing. The 3D printer was a game-changer because, for the first time, you could take something modeled digitally and produce it in the physical world.
I quickly realized that the current tools for representing 3D shapes were a bottleneck. In architecture school, we used the term “computational design,” which is about capturing engineering requirements into computer algorithms. I became obsessed with how we could use computers to help create new shapes and products. The existing CAD tools were based on drawing and drafting principles from the ’70s and ’80s, so they were limited by the sequential operations of drawing. Instead of capturing the requirements as algorithms, you’d draw a sketch, extrude it, add bodies and manipulate edges. But I wanted to take advantage of modern computational power—GPUs, parallel computing—to create shapes in new ways. That’s why I created nTop, to remove the limitations of traditional CAD tools and use computers to help us design.
Aaron: Give us a tactical example of how nTop differs from, say, SolidWorks or Creo, which are more traditional CAD programs.
Bradley: A great example is Siemens Energy. They design enormous turbine engines, the biggest jet engines on the planet—way bigger than the ones on airplanes. These engines power entire cities, like the one providing electricity to my apartment in New York. These engines currently run on fossil fuels but Siemens is working on converting them to hydrogen power, which burns much hotter. So, they need to redesign components like burner chambers and heat exchangers, which are extremely complex. The goal is to transfer as much heat from one material to another with minimal pressure drop, which involves thousands of tubes weaving in and out of each other.
In a traditional CAD tool like SolidWorks, you’d have to draw all those tubes manually or you might create a basic array of tubes and adjust it. But each design iteration takes time and you’d need to redraw everything to optimize it. With nTop, you capture the requirements in a set of parametric building blocks. You can define parameters like the number of tubes, the periodicity of sine waves or the placement of inlets and outlets for fluids. These building blocks allow you to instantly update the model by adjusting parameters, like changing from 10,000 tubes to 20,000. The model is directly connected to fluid simulations, so you can quickly evaluate how well each design performs and make adjustments in real time.
To give you a simpler example, think of an airfoil or wing. In traditional CAD, you’d draw one specific version of that airfoil. But in nTop, you capture the requirements—the thickness, camber and other variables. This allows you to generate countless variations of the airfoil and run a design study to determine the best-performing option, all without manually redrawing it each time.
Aaron: What kind of computer do you need to run nTop effectively? Do you need a really high-end machine or is some of the computing done in the cloud?
Bradley: I run nTop on my laptop, which is a couple of years old and it works well. I also have a desktop with a fast GPU—a Threadripper with 256GB of RAM and an NVIDIA A6000 GPU—which runs everything much faster, of course. But nTop performs well on both setups. What’s interesting is that our models are extremely lightweight. A fully parametric model of a jet engine that might take up gigabytes in a traditional CAD system could be just a megabyte in nTop. That’s because our core technology is based on signed distance fields. Instead of storing surface data, we store mathematical equations that define the shapes, which makes the models compact and allows for faster computations.
Aaron: That’s impressive! Have you ever considered licensing your modeling technology to other CAD platforms? I imagine that would be a huge advantage for companies like SolidWorks.
Bradley: Absolutely! We’ve actually created a product called nTop Core, which is a library that partners can use to read and write nTop data. Companies like Materialise have integrated nTop Core into their software, allowing them to consume nTop models and slice them for 3D printing. We also partnered with Autodesk to release a plugin for Fusion 360 that reads nTop Implicits for build simulation and processing. We’re considering opening up more of our modeling API so that partners can build more complex applications on top of nTop Core.
Aaron: Interesting. How does the modeling in nTop compare to traditional CAD programs? Is it similar or completely different?
Bradley: It’s definitely different. In traditional CAD, you’re defining geometry by drawing edges and surfaces. In nTop, you’re creating relationships that produce geometry. We use the term “implicit modeling,” meaning the geometry is defined through a set of primitive shapes and the intersections create the edges. You can add fillets and rounds at those intersections but the core modeling approach is fundamentally different. If you’ve been using traditional CAD for 25 years, nTop might feel hard to grasp at first. But we had a high school intern with us over the summer who picked up nTop in just two days. He used it to design and 3D-print a tricopter drone in six weeks!
Aaron: That’s really impressive. So, it’s not like traditional modeling where you’re sketching out everything—you’re defining constraints and the software builds the model based on those?
Bradley: Exactly. You’re setting up relationships and constraints and nTop generates the geometry based on those. It’s more about defining what you want to achieve and the software handles the specifics of how the shape is created. This makes it very powerful for exploring different design variations quickly.
Aaron: What types of physics does nTop handle? Does it work with things like loads, vibration, heat and fluid dynamics?
Bradley: Yes, we have our own solvers for linear static stress, vibration and thermal analysis, which use traditional finite element analysis (FEA) methods. We’ve also integrated meshless simulations, where the implicit model itself is used in the solver without needing to generate a mesh. We’ve partnered with Intact Solutions for stress analysis and a company in Germany called CloudFluid for computational fluid dynamics (CFD). nTop is also integrated into mainstream engineering workflows, so you can use it alongside tools like STAR-CCM+, Fluent and LS-DYNA for advanced simulations.
Aaron: Let’s talk about the handoff between nTop and traditional CAD programs. For example, if I’m working in SolidWorks and want to optimize a part using nTop, how do the two systems exchange files?
Bradley: If you’re working on an assembly in SolidWorks and you want to optimize a part, you can export the part as a STEP file and bring it into nTop. In nTop, you can set up your problem—apply loads, and boundary conditions, run simulations and generate the optimized geometry. Once you’ve optimized the part, you can export it as a STEP file and bring it back into SolidWorks for assembly.
We also have customers using nTop within a PLM environment like Windchill or Teamcenter alongside NX or CATIA. In these cases, the nTop file is managed directly in the PLM system and any changes that happen upstream automatically get fed into nTop to update the design process. In addition to detailed part design, we’re seeing customers use nTop earlier in the process for low-fidelity concept design. For instance, in applications like drones or small robots, engineers can rapidly iterate through thousands of variations of a system-level model in nTop to figure out what works best. Once they settle on a concept, they can export it as a STEP file and continue refining it in SolidWorks or CATIA.
Aaron: So nTop is primarily used for additive manufacturing and 3D-printed parts. I’ve seen some parts made with nTop and they often have these really organic shapes. It’s almost anatomical, with structures that look like veins and surfaces weaving in and out. Is 3D printing the best way to realize those shapes, given that traditional methods like injection molding wouldn’t be able to create such complex geometries?
Bradley: Yes, that’s right. For problems involving highly complex geometries—like heat exchangers or parts with intricate lattice structures—nTop is the only tool that can produce those shapes robustly, quickly and efficiently. Traditional CAD tools would drive you nuts trying to model something like a heat exchanger with hundreds of thousands of tubes. But with nTop, you can handle those kinds of geometries with ease.
Additive manufacturing (3D printing) is particularly suited for these complex shapes because the limitations of subtractive processes like machining or molding do not constrain you. You can print structures with internal channels, organic lattices and other complex features that would be impossible to machine or mold. That’s why we’ve seen so much success with nTop in the additive manufacturing space.
Aaron: If we consider a structural part in nTop, where you’re applying loads and boundary conditions, does the software strip away material where it’s not needed, leaving only the load-bearing structure? Is that how you end up with those organic shapes that look almost like bone structures?
Bradley: Yes, that’s exactly what happens when you run a density-based topology optimization. You start with a block of material, apply loads and constraints and the algorithm removes material where it’s not needed. What’s left is a representation of the load paths, the areas that are crucial for structural integrity. That’s a common use case for nTop and our implicit modeling technology is very good at handling those types of problems.
However, nTop goes beyond just topology optimization. You can also build parametric models that take into account manufacturing constraints, like making sure a part is machinable or suitable for injection molding. For instance, if you’re designing an injection-molded part, you can set parameters that ensure all walls maintain a uniform thickness and that certain regions remain accessible for tooling. So, it’s not just about producing organic shapes; nTop allows you to build models that are optimized for different manufacturing methods, whether it’s additive, subtractive or injection molding.
Aaron: That’s fascinating. I didn’t realize it had those capabilities. Does that mean you can set up your models so that the exported geometry—if you need to bring it into another platform like SolidWorks—has flat faces or other features that are easier to work with in traditional CAD?
Bradley: Yes, absolutely. Our mesh-to-STEP file conversion is getting really good at identifying and preserving flat faces. But my question would be: why bring it back into CAD? You can do a lot of those operations, like adding draft or machining features, directly in nTop. Draft, for example, is a block in nTop—you can set it up automatically, apply it to your model and then run your analysis with the drafted part. You don’t need to think of it as a post-processing step like you would in a CAD system.
Aaron: Does nTop also allow you to add discrete features, like holes with specific dimensions?
Bradley: Yes. Everything in nTop is defined by parameters, so you can create a feature like a hole with precise dimensions and tolerances. If you later find that the load at a bolted connection is too high, you can adjust the size of the hole, and the entire system will update instantly. That’s one of the strengths of nTop—it’s parametric like a traditional CAD system but much faster and more robust.
Aaron: It sounds like FEA (finite element analysis) is typically considered a separate process from CAD modeling. You create your model in CAD, then export it for simulation in ANSYS or another tool. But in nTop, it seems like the simulation is integrated into the design process. Is that the case?
Bradley: Yes, that’s exactly right. In many organizations, simulation and modeling are separate processes, often handled by different teams—one team focuses on drafting and another handles analysis. But I see simulation as just one step in solving an engineering problem. You should have a tight connection between the simulation and the geometry so that when you make a change to the model, you can immediately see how it impacts performance.
Ideally, I’d like to make a change—like tweaking some ribs or adjusting the draft angle—and instantly see the effects on the stress distribution or other performance metrics. We’re not quite there yet because physics simulations still take time to run, but I think we’ll get there in the next five years or so. Right now, simulation in nTop is integrated into the workflow. For example, I have a model running on my desktop right now, where I’m using nTop Automate to run a parameter sweep on the wall thickness of an injection-molded part. It’s automatically running through 100 different design options to find the optimal combination of weight and stress distribution. The meshing and solving take time, but nTop handles it all in the background.
Aaron: In traditional modeling, you’re often creating a part based on engineering judgment and only in critical cases do you run an FEA. But it sounds like nTop is different, where simulation is directly linked to the final geometry. Is that a good way to think about it?
Bradley: Yes, that’s a great way to think about it. In nTop, simulation isn’t something you do just to validate a design; it’s a tool to guide the design process itself. Each design iteration is automatically evaluated based on performance metrics, like stress distribution or weight. So, rather than relying solely on engineering intuition, you’re using real data to optimize your design in real time.
Aaron: Are there any tasks or operations that you wouldn’t recommend doing in nTop?
Bradley: There are still some things nTop isn’t optimized for. For example, setting up assemblies is more difficult in nTop because we don’t have all the constraints you’d find in traditional CAD systems. We also don’t have a robust sketching tool yet, so if you’re designing something simple, like a bracket or a sheet metal part, traditional CAD is still the better option.
However, for more complex problems—like designing bulkheads for aircraft or optimizing intricate geometries—nTop excels. If your design requires frequent changes and those changes are hard to manage in a traditional CAD system or if you’re dealing with thousands of surfaces that are similar but slightly different, nTop is the right tool for the job.
Aaron: What are some key questions engineers should ask themselves when deciding whether to use nTop for their design problem?
Bradley: The main question is: will spending more time upfront defining relationships and constraints lead to a better design in the long run? If you’re working on a problem that requires a lot of iterations or where changes in the design are manual and time-consuming, that’s probably a good fit for nTop. Another question is: are you dealing with hundreds of thousands of surfaces that are all slightly different but share common characteristics? If so, implicit modeling in nTop would likely make the design process much faster and more efficient.
Aaron: Are there any other topics or questions we haven’t touched on yet?
Bradley: We could talk all day about computational design and the types of problems nTop solves. One thing I’d like to mention is that we recently launched nTop 5, which represents a huge leap forward. We re-architected the core model to be an order of magnitude more precise and faster, which opens up new possibilities for solving much larger and more complex problems. There’s a video on YouTube called The Powers of nTop, where we demonstrate some of these capabilities, like concept design for drones and engines.
We’re also active on social media, so you can follow us there to keep up with what we’re doing. And for students, we offer nTop for free. You can go to our website, sign up under the EDU tab and get access to the software for learning purposes. That’s really important to us because we believe that’s where the next generation of engineers will come from.
Aaron: How can people get in touch with you or learn more about nTop if they have further questions?
Bradley: You can reach me on LinkedIn or email me directly at brad@ntop.com. I’m also on Instagram and respond to DMs there.
Aaron: Thank you so much for your time, Bradley. Congratulations on all the success you and your team have had with nTop and thanks for being on the podcast.
Bradley: Thanks so much for having me. Let’s talk again soon.
Recommended Comments
There are no comments to display.