Success with Generative Design in 2023

MIKE SMELL: Hi everyone, and welcome to AU 2023. This session is IM600969, Success with Generative Design in 2023. I'm Mike Smell, Senior Product Manager, Fusion 360. Before we get started, as always we need to cover our safe harbor statement. There may be forward-looking statements regarding planned or future development efforts throughout my presentation that should not be used to make purchasing or investment decisions.

So as I said, my name is Mike Smell and I'm the Senior Product Manager in the Design and Manufacturing group at Autodesk. I manage Fusion 360 Generative Design technologies as well as our efforts around client resiliency and stability in Fusion 360. I've been at Autodesk for over 14 years, and I'm based out of Pittsburgh, Pennsylvania. I'm a mechanical engineer by schooling with degrees from Penn State University and the University of Pittsburgh, and I've been presenting at Autodesk University since 2009.

As we get started with our time together today, we're going to look at the following learning objectives and agenda. We'll understand the range of generative design technologies that exist in Fusion 360 and their applications. We'll look at all of the generative design features that we've released in Fusion 360 in 2023.

We'll then look at how some of these features specifically can enhance your generative design workflows, and then we'll start to help you understand is generative design right for you, and show you how some of our customers are drawing a lot of success out of using generative design in Fusion 360.

Now I'm going to transition into this slide as we get started, and if you look at this image I think it exhibits what many of us have faced in our product design careers. You're sitting with your design or engineer partner and you're talking about things like, do you think we can make this design differently? Could it be better? Could it be lighter? Could it be with a different material?

These are all very common questions as we're going through the product development process, and these are all areas where generative design can play a major role in supporting answering those questions. So let's think about generative design at the very high level. What is generative design?

It is a design approach where the software is becoming a creative partner in the design process, ultimately creating designs for us. This enables a future where automation and intelligence software will play a much greater role in design creation and how we develop products, and it is a powerful engineering tool which can provide enormous value today.

Now more specifically, generative design in Fusion 360 is a market-leading design exploration technology. Generative design inside of Fusion 360 has been in the commercial market since 2018, and since then we've been generating multiple validated solutions that are performance-, manufacturing-, and cost-aware. And these solutions have been validated and valued by our customers as you'll see in some of the customer stories we'll talk about later in this presentation.

Now generative design is just one piece of Fusion 360, and I would be remiss if I didn't spend just a few seconds talking about the broader fusion in Fusion 360. Fusion 360 is not just a generative design solution, and in fact that generative design is just a small piece of what Fusion 360 actually is.

It's not just a CAD system, it's not just a CAM system, it's not just a tool to do your simulations, create documentation, rendering, or electronics. It's not focused purely on subtractive manufacturing or additive manufacturing. It's all of these things combined on a single cloud data platform that's helping bring together the next generation of how we develop products.

Let's dive back into talking about the generative design technologies that we have built in Fusion 360. Now as I had mentioned, generative design in Fusion 360 has been present in the commercial market since 2018, and over these past five years we've been working with you, our customers, and also internally working on our technology stack to figure out how can we bring the right tools for different objectives to market?

And what we have now in the current product is three unique verticals where we're serving up different solutions for how generative design technologies can be an enabler in your design process. Starting early in the design process thinking about form and function exploration, we can do this in the design workspace with automated modeling.

As we get a little bit further along into our design process and we understand the performance and physics requirements that our designs might have, we start to talk about things like generative design for structural components. This is our flagship solution that's been, again, commercially available for quite some time.

We've started extending that into generative design for fluid paths where we've added this new physics domain to start bringing the same types of design exploration capabilities to market for fluid flow applications. Now we'll look at these in a little bit more detail as we move forward.

Now this image is a good representation of where we think automated modeling fits into the design process. We're in the early stages, trying to work our way through concepts, what the size should be, what the shape should be, maybe how connections are formed. Automated modeling is exactly that.

It's about bringing a new level of accessibility to generative design technology much, much earlier in the design process where we're not burdened with having to understand all of the physics that we have to be aware of all, of the manufacturing requirements that we may have to consider-- what materials do we have available to make a given design?

We're really just focused at this stage on rapidly exploring potential design alternatives for how we need the geometry to come together. We've built this in a specific way to feel exactly like a modeling tool in the design workspace so it's part of your design process. It's not a different step in the process.

We have very minimal setup definition required to use automated modeling. You simply tell it what geometry you might like to connect and what geometry you might like to avoid. And then from that definition, automated modeling will send out to our cloud service and start generating multiple editable shape alternatives.

Now this makes a lot of sense if we're thinking about concept creation, if we're thinking about design exploration. Perhaps, what if we change these requirements about the design? How do we fit a shape around different obstacles or different parts of the assembly that we may be thinking about?

Lastly, complex connections. So rather than spending potentially hours of modeling a really, really sophisticated shape to connect many things together, it might be wise to use automated modeling to see what path you could potentially go down.

Now once we get beyond these early phases of the design process and we start to think a little bit more about the physics requirements that we have-- how will these designs have to perform? What materials will we be using? What sort of manufacturing methods will we be using? This is where our generative design solutions come into play.

We'll start off with talking about generative design for structural components, and here this is about design exploration where performance matters. And performance matters is a key statement there. One of the reasons that we just talked about automated modeling and the reason we brought that to market is because we had a lot of folks who were really interested in what generative design for structural components was offering through really unique and organic shapes, being able to explore a number of different forms.

But they weren't at the stage where they understood the physics, and without that they were really using the wrong tool for the job. But if performance matters and we're looking to build structural components, generative design is the solution to pursue. We have design objectives for mass, stiffness, stress, frequency and buckling.

We're considering multiple materials and manufacturing methods during our design exploration process. So if we want to compare additive processes, milling processes, two-axis cutting if we're doing water laser or plasma, or die casting, we can set up the same problem definition, tell it to go off explore building a shape with any and all of those manufacturing methods in mind.

And then we can observe how do the shapes differ, how do they look aesthetically, how do they perform, and what is the cost associated with that? Generative design is also providing cost insights based on the selected manufacturing process, as well as the production volume powered by our partnership with aPriori.

And then lastly, much like automated modeling, we're producing additive Fusion 360 geometry that's fully ready for use in downstream design, simulation, or manufacturing workflows. So we expect that there's probably going to be some influence that you would like to further add into the design when it comes out of generative design.

This makes a lot of sense if we're pursuing lightweight designs, part consolidation, perhaps we're looking to simplify the assembly process. Maybe we have the flexibility to move to an additive manufacturing process. And another application that makes a lot of sense for generative design for structural components is around manufacturing tooling.

This is an area where tooling-- there are a lot of benefits to having lighter weight tooling. We can increase cycle times, we can increase stiffness and precision of the machines. So an area that makes a lot of sense and is often overlooked because a lot of times we think about the products that we are selling and not the things that we use to make the products that we sell.

Let's move ahead to generative design for fluid paths. And as I said, this is still preview technology that we're providing to help users start to explore design ideas and alternatives for fluid flow applications, and this has allowed us to expand the generative design solution space into fluid domains. We've utilized our existing generative design framework for all of our workflows and our concepts.

So this means if you've already used generative design for structural components and you understand what it means to have a preserve and what it means to have an obstacle, and the impacts that starting shapes can have, you'll be well on your way to understanding how you may be able to leverage generative design for a fluid flow application.

The current state of the preview is focused on efficient fluid path design where we're minimizing the pressure drop and balancing the flow. Much like the other two solutions that we've just talked about, we're producing editable fluid path geometry for use in downstream workflows.

Now one of the things that's interesting about this geometry is that it's actually the fluid domain itself and not the wrapper. So this allows you to, for example, if we're looking at a flow control device and we want to put a more efficient flow path in there, we would be able to just fill in the flow control device and then take the new fluid domain and cut that out of it directly.

Now common applications where generative design for fluid path make sense are things like complex ducting. So think about optimizing the air flow inside of a computer chassis where we've got a lot of ways that we want to distribute the flow and ensure that we get high distributions of air flow to areas that need significant cooling.

The ductwork inside of the air conditioning units inside of your vehicle is another common application where generative design for fluid paths where we're trying to snake these flow domains through a complex ecosystem but have efficient flow-- is a great application for this solution.

Any place where we're looking at manifold distribution, ensuring that we've got equal flow across the set of inlets or outlets is another great opportunity to take advantage of generative design for fluid paths. And then lastly, any type of flow control device where we're looking to drive efficiency and remove areas of cavitation or recirculation, generative design for fluid paths is absolutely going to help smooth out those flow paths and make them much more efficient.

Now that we understand the solutions that are available inside of Fusion 360, I'd like to take some time and review all of the development achievements around generative design technologies that we made as a team in 2023. So to start this discussion, I would like to talk about the development themes that we laid forward at the beginning of the year.

The first thing that we wanted to do was improve the time to solution for our users, and this was coming through looking at how our sovereign compute stack was built and what optimizations we could deliver in that environment.

The next thing that we wanted to do was unify our technology stack. And if you've been with us for the past couple of years-- and we'll talk about some of these in the upcoming slides-- you know that we've been evolving our solver technology, and along the way having we had multiple solver stacks running, providing different results for different manufacturing methods and different load types. And we're in a state now where we've made a lot of progress and we're starting to consolidate on that and get down to a more consistent user experience.

And the last thing that we wanted to do is increase the overall robustness and reliability in the solution to help you achieve better results and higher quality the first time. And I think we've made some pretty great progress on all three of these fronts. Let's dive into those now.

Let's start in January at the beginning of the year. And one of the first things that we did was, we were at the point in time where we could graduate automated modeling from preview. So we spent a little bit of time already talking about how automated modeling helps you create multiple editable design alternatives in just a matter of minutes in the design workspace.

And we're also with automated modeling automatically detecting symmetry in your setup and ensuring that you have symmetric alternatives created, and that those alternatives are also editable in a symmetric fashion. So if we've got a T-spline model and it's symmetric, we start pushing and pulling and manipulating one side of the geometry, the same thing would happen on the other symmetric side just, as you would expect.

The other thing that we've done to help users get additional variation in their design exploration is we put a volume slider on each alternative to give you additional variation. So as you can see here in the GIF that's playing, if we want to say, like, let's make the shape a little bit thicker, a little bit thinner, rather than having to do all that editing manually we built the slider in so that we can automatically produce a shape of that smaller or larger thickness.

Now the other thing I'd like to point out here is this is included in the Fusion 360 subscription. So there's nothing stopping you from giving this a try, and I would encourage everyone who's watching today to take a few minutes and see what automated modeling has to offer for you.

Now I'd like to switch gears into our generative design solutions, and we'll start off by talking about some of the advanced features in the structural component study. In January 2023, we graduated a number of these features into the full solution outside of the preview.

Now remote forces, remote moments, remote constraints, the ability to apply off-axis moments on cylindrical faces, global and local displacement limits, and manual point masses are all available inside of the commercially available solution without having to be enabling those through preview. Because they are now part of the full solution, they can be used together with all of our commercial manufacturing methods. And we'll talk about some other features that they can be combined with later in the presentation.

Now there is one note to make here that the die casting manufacturing method is still on a different technology stack at the time of this recording and back in January of 2023 when this was released. So these features are not supported with die casting. So if we have a setup that includes any of the things that we see listed here, we will not be able to generate a die casting solution for those.

Now I'd like to talk just for a second or two about why these advanced features are important, and ultimately this comes down to how we define our problems and the level of accuracy we can obtain in our results. So let's start with an example that likely everybody can relate to.

I'm sure most folks watching this have ridden a bicycle or understands basically how a bicycle works. So here we have a crank arm, and we know that our feet push on the pedals. They turn the crank arm, which turns the chain ring, which ultimately drives the bicycle. Now you might think, oh, that's very simple. We just push down and that's how we design the loads on our crank arm.

But it's not quite that simple, because what we have typically is that where the pedal anchors to the crank arm and where the crank arm attaches to the spindle for the chain ring, isn't a consistent line. But our foot is on the pedal, which is some distance removed away from the center line of the crank arm. Now what that means is in addition to a bending moment, we will also create a torsional load around the center line of that crank arm.

So remote forces are a great way to very easily accurately represent that load where we've got the load magnitude pushing down at the pedal, But accounting for that distance away from the center line so that we can represent both the bending and torsional loads that occur in the crank arm.

Another example here is a common connection type, a ball joint or other types of universal joints or flexible connections. And oftentimes, these types of connections exist at or near where we might put constraints in the model. Well, you can imagine this ball joint example-- there is some flexibility that occurs in that union of two different components. That flexibility will ultimately reduce the stress in that area. Whereas if we were comparing that to a rigid connection, there would be a higher stress.

So remote constraints are a great option for how we can better represent some of these flexibilities in our constrained locations. So here you can see in this example, we've got a simple block. We've cut out half of a sphere that we want to represent a ball joint type of condition, and through remote constraints we have access to both translational and rotational degrees of freedom.

So specifically to get to a ball joint configuration, we would apply fixed translations, but we would leave all of the rotations free. And then about that center point that you see there, we will have all constraint translations constrained but we will have full flexibility for rotation about that center point.

Jumping back into achievements for January, now we're going to talk about symmetry, and we're going to talk about this in the context of 2- and 2.5- axis symmetry. And this falls in line with the unification of our tech stack where now we're able to create symmetric designs when non-symmetric loading is present for 2-axis cutting and 2.5-axis milling manufacturing constraints.

If you've used these before, you'll notice that they are sketch-based outcomes rather than our T-spline based outcomes. And what we're doing with symmetry here when we start to-- we've generated our symmetric outcomes, we start to talk about editability, and for the symmetric definition that you've created we will create only that section of the sketch.

As you saw there, that sketch is then editable, and in the timeline following that sketch we will have mirror operations that will then build out the fullness of the geometry, which is a very common best practice for symmetric part modeling anyways. So we wanted to ensure that our generative design outcomes and the way that we ended them falls very closely in line for how you might be modeling these types of designs manually or in a more traditional sense.

It's important to note that symmetry does require a symmetric preserve definition, and it's highly recommended that our obstacles are symmetric as well. The solutions will run with obstacles that are not symmetric, but you're going to need to pay very close attention to how those obstacles are laid out and when they're carried through the symmetric definition and geometry creation-- that they won't create areas where you actually need to have geometry.

Now we'll jump ahead a few months to April and with the work that we did back January with 2- and 2.5-axis symmetry and how we've watched our symmetry functionality mature. For our T-spline outcomes, we were able to graduate this from preview as well.

So now again, in the mainline solution, without having to activate any preview technologies, we're able to create these symmetric design outcomes for all of our manufacturing methods, or most all of our manufacturing methods, fully symmetric editability for both our T-spline-based outcomes that would be present if we're using unrestricted additive 3- or 5-axis milling manufacturing constraints. And then the sketch-based outcomes that we just talked about with 2-axis cutting and 2.5-axis milling.

So same requirements still exist with symmetric preserves and highly recommending symmetric obstacles. But again, as a note, at the time of this recording and where we were at in April 2023, symmetry was not yet supporting die-casting manufacturing methods.

We had a few other minor updates-- quality of life things that we did in April of 2023 around our Study Settings UI. We made some improvements with just the simple text that we provide to you to help understand, what does it mean to use a lower resolution? Or what does it mean to use a higher resolution?

We also added increments into the slider to make it much easier to repeat the settings that you've used from one setup to the next. It also makes it easier to collaborate with others. So if someone else is reviewing your model or asking questions about, hey, how might I pursue a problem like this, if you've got a setup that has very, very thin preserves, you might say, hey, I'm one tick mark away from the highest setting.

Or if you're just getting started and you want to get a really, really fast solution, you can say, hey, I'm two tick marks to the right of the lowest setting. And you might find that it's much easier to communicate on the grounds of how you've set up the model.

The other thing that we've done in April was we had retired the alternative outcomes option from our Experimental Generative Solvers and Features preview. And again, when I talked about our goals for the year, I acknowledged that for a while we have had the Alternative Outcomes option present and this would enable us to prove out various solver technologies as we brought them to market.

Now that we're starting to standardize on a single foundation for how we generate all of our outcomes, we no longer need to do this. And I think this is really important from a user experience point of view because oftentimes we would have users who were using the Preview, giving us feedback about, hey, I see all these outcomes and they all look the same. How do I know which one's better or which one's worse? Or, which one do I pick?

And the good news is, the fact that they all looked the same meant we were doing our job with proving out the technology and getting it working correctly. But at this point in time we don't need that extra noise in the solution set, so that is now retired.

Now we're going to turn our attention to a number of solver improvements and solver consolidation efforts that we've been making throughout the better part of the summer. Starting in May 2023, we started to add-- or we added frequency limit support to the standard solver stack. So now we have frequency limit able to support all of the commercial load and constraint types, symmetry, and manufacturing methods.

Now it's important to note that while we've brought this over to the standard stack, frequency limit is still in the preview and it does have the following limitation that we've talked about a couple of times now, that frequency limits do not support the die-casting manufacturing method.

Now I want to also take a minute while we're talking about frequency limits to explain why these are important. So if we're designing something with generative design that is either attached to or very nearby rotating or vibrating machinery, we want to ensure that whatever we design does not encounter a resonance condition with those nearby vibrating or rotating pieces of equipment.

So here you can see three example geometries where at the top we've optimized this for a given set of load cases with no frequency limit. And then we start to say, well, what if we had a minimum first modal frequency of 25 hertz? What if I had a minimum first modal frequency of 40 hertz? How does that change the design?

And here you can see in both of those cases we've beefed up the design a little bit we've also added a number of diagonal sections in there to add stiffness to the design. So because we want to drive that frequency up, the design needs to be stiffer and stiffer, and that's what the frequency limit is going to help you do.

Moving ahead to July, another solver improvement here is we've brought our rigid body mode option in line with our standard solver stack. So now much like we just talked about with frequency limits, rigid body mode support is available for all of our commercial load and constraint types, for symmetry, and all of our manufacturing methods that are not in preview.

And this is a commonly used option, or an option that you should consider using when you have a setup that requires no constraints at all so it's completely free to move around, or it has limited constraints that may also allow some amount of rigid body motion. Because this is still in preview, there are a couple limitations. One, this cannot be combined with the buckling limit option, which we will be talking about shortly. And it also does not support the die-casting manufacturing method.

Jumping ahead to September, this slide is going to look very familiar. Again, more of that solver standardization work where we're now bringing the buckling limit over to the standard stack, again supporting all commercial load and constraint types, symmetry, and manufacturing methods. Still in preview.

A couple of limitations, slightly different than what we've talked about so far. There's an additional limitation for buckling that we only support a single load case at this time. This is something that we have some ideas about how we might be able to relax in the future, but at the current state buckling is only supporting a single load case. So that will require a little bit of thought on your part as the user to think about which load case is going to have the highest potential for buckling, and we can evaluate that load case only.

Again in the images here, why do we use this? Buckling is a scenario that often happens when we have long slender members that are under compressive loads. So here you can see again, two images left and right with no buckling safety factor. You can see we've got some very long slender members. We think, hey, there might be a concern for buckling in this model. We've added in a bucket safety factor limit of 5.

And you can see what's happened. We've thickened up those members, we've added some diagonals in to support that with bracing, and ultimately what this has done in the simplest explanation is it's made all of the members shorter. The shorter the member, the less likelihood that there is for buckling.

One other major improvement coming out in September 2023 is around our overall solver robustness. I talked about the importance of us being able to deliver high-quality results the first time, and as we've watched how our solvers have performed over the past few years, one of the most common errors that users encountered is that the solutions would stop in a completed state early. They might look a little heavy, they might look a little bit blobby.

And this was because they were getting to a place where they thought, hey, it's going to disconnect. And the solver does not allow for the preserves to disconnect. So we looked at, well, how could we do a better job of handling this and preserving some of those connections and keeping all the preserves intact, but allowing the model to optimize itself further so we get a better solution?

And we came up with a strategy to do that that's going to help significantly in scenarios where you have a setup that has unloaded or under-constrained preserves, or setups that include loads where the load magnitude between the preserves on one side of the model and the preserves on the other side of the model may have significant orders of magnitude difference.

So an example of that might be having preserves in your model that are relevant for sensors, cameras, or lighting, but the rest of the preserves ultimately represent where the design space is constrained and where the main structural loads are occurring. That's a really common scenario where these disconnection errors would occur.

So with the technology that we've built in, we've seen these types of issues-- and just a few short weeks here since this has been released-- at the time of this recording have dropped by about 15%. So we're definitely moving in the right direction, giving you better results the first time through.

And if you look at the images here, you can see the old result and the new result. And I think it will be pretty clear as to what was happening. You can see that preserve that sort of stands alone on its left has a really thin section going over to it, where the rest of the geometry that's being created is still pretty bulky-looking.

So we've stopped early with this really bulky geometry, but with this thin connection. And we would say, hey, we need to stop because it's going to separate. You can see on the right the new result we've done a much better job of maintaining that connectivity to that lightly-loaded or maybe unloaded preserve, but we've also allowed the rest of the geometry to continue to evolve and optimize, so you get a much more balanced design.

The last piece of functionality that we released with the September update is around configurations. If you were watching the September What's New, you know that configurations is new functionality that was released in the design workspace to allow users to easily start to build out configurable designs with various options for how they should be developed.

And we wanted to ensure that in our generative design workflows and in our simulation workflows that we could play nicely with all of these different design configurations, so what we did in the generative design workspace is we made it very easily when you're in the study definition to switch between configurations.

We also create a unique generative model for each configuration. So now we're able to isolate or keep the results associated with the configuration in which they're related, and what this enables is now a much more streamlined way to pursue a bigger design exploration. So if we've got multiple configurations of preserves and obstacles or designs that we will build obstacles and preserves for, we can streamline a lot of this work, keep the results linked to each configuration.

And then you'll see when we move over to the Explore environment, just like today where we use multiple generative models or multiple studies, all of the results are aggregated in one view. Well now that we have an additional level of aggregation happening, we need to be able to filter and compare by those configurations as well. So again, standout feature for Fusion 360 as a whole-- generative design, making sure that it can work in the most streamlined way possible with all of those new design configurations.

With that, I'd like to give an overall summary of how we see all of the things that we've talked about and how the themes that I outlined at the beginning of the What's New come together as an overall summary of performance improvements for generative design and for structural components.

All of the improvements that we've made throughout the year have resulted in an average solve time that's about 1.7 times faster than it was previously. We now have six fully supported manufacturing methods for all the different load types we talked about, all the different constraint types we talked about, and symmetry. Things like how we're on our solver robustness and handling separations.

We've seen an overall increase in converged studies of 22%, and we've also seen a 50% reduction in the number of failed studies. So major, major progress for us as a team this year in the technology that we're delivering, which should ultimately result in a much, much better experience and rate of success for you our users with generative design.

Now I'd like to transition the conversation to helping you understand if generative design is right for you. We've talked a lot about the technologies that we offer inside of Fusion 360. We've talked a lot about a bunch of advancements that we've made in 2023 and how those are going to help drive your success with the solution.

But you have to use the solution to be able to experience all of those potential successes, so let's talk about why do we use generative design. And this is a story that we've been telling since the release of generative design back in 2018. It's really about helping designers and engineers work through this problem of balancing cost and performance versus the amount of time and energy that they can spend.

I'm sure everyone watching this has experience with going through that decision-making process to say, we need to improve the performance of this design, or we need to make it lighter, or we need to make it more cost-effective to manufacture.

But then they only have so much time and energy to invest in exploring, how do we make it lighter, how do we make it cheaper, how do we make it stronger? And if you are faced with this conundrum, I think this is an opportunity where generative design can bring a lot to the table to help you expand the range of solutions that you can consider for how you solve those problems.

Now that we understand the scenario of why and what we're trying to balance and how it can help you balance that decision-making process, I'd like to help you understand some of the areas where we think generative design makes a lot of sense for use.

When the solution came out back in 2018, a lot of our users were coming from R&D departments. They were looking for how can they bring this cutting edge technology into their design process, or maybe they were just saying, hey, we're making the move to additive manufacturing. And I would say, as the technology has matured, as folks have found out what generative design is capable of, it's not just for folks working in R&D and folks pursuing additive manufacturing.

We've seen folks have a lot of success with new product proposals where generative design is helping them uncover new creative ideas or evaluate more concepts in less time that are higher performance or better at material utilization. This is also really interesting if we're thinking about new product proposals in the context of being engineered to order or low production volumes or maybe really high volume components.

The next area where we see a good fit for generative design is in continuous product improvement, and we're going to talk about a customer success story in this area in just a bit. But this is helping our users find new ways to solve a design challenge. Potentially you want to optimize the performance of an existing design. Maybe we want to ensure that we can deliver a new level of performance to our existing customer base without having to potentially force them to overhaul an entire assembly or an entire system.

We want to take an existing design and simply make it better. This makes a lot of sense if we have high volume production type of products where shaving a little bit of material that's either used for manufacturing like in the case of casting, shave a little bit of weight if we're shipping hundreds of thousands of units, anything like that where we can have relatively low effort to change the design and make it incrementally better and drive additional value.

And then the last area where we see a lot of opportunity is around manufacturing support. I mentioned this earlier in the presentation, but the things that help you make the products you sell are often overlooked. We may be able to pursue alternative methods for manufacturing. Maybe we can make it in-house with additive manufacturing versus sending it out to be machined.

Potentially we can pursue different, more cost- effective materials, or maybe we can just simplify the way that these components are being built through a new manufacturing method like additive manufacturing, where instead of having tens of parts that we have to assemble to make tooling, we now have a single design. So these are three great areas where you can think about generative design playing a role in your product development process.

Let's take a look at this in a little bit more of a precise kind of question and answer style of conversation. So first off, if those three areas weren't clear enough, you can ask yourself, do we have organizational goals around improving product performance, increasing productivity and innovation capacity in engineering, winning more business, or reducing our product cost or weight? There's a good chance that generative design is going to be an interesting tool to help you support those endeavors.

Now if we go a bit more precise into the technical details of what you're designing, you could ask yourself, does this design have load carrying requirements? We said at the beginning of the presentation, the generative design for structural components or generative design for fluid paths are really focused on performance. If you have specific performance requirements around structural physics or fluid physics, there's a good chance the generative design can support you there.

Again, if we have performance requirements that you're already validating with a traditional simulation tool for stress or vibration or buckling, or in the case of physics-- or in the case of fluids, a CFD tool, again there's a high likelihood that generative design can help you extract more performance from those designs.

And then lastly, if you're sitting there thinking, well, you know what? We probably could make this design with a different material, or we want to pursue doing this with a different manufacturing process, again generative design is going to help you weigh those potentials and see if you can achieve equally as good designs with new processes or new materials and decide which one is best.

So the last section that we will dive into during our time together here is around customer stories. And we can talk all we want about how great the features are and when and why of generative design, but what ultimately it comes down to is real world success and how generative design can enable that. And we're going to look at two examples.

First we're going to talk about Toshulin, and they're a customer that we worked with this year to have generative design support their efforts around transforming CNC machine design through continuous product improvement. And I really, really like this story for a couple of reasons, and we'll dig into that in just a second.

It's important to understand who Toshulin is. They're a world-leading designer and manufacturer of vertical lathes and multifunction CNC centers, and they looked to generative design to help them explore how they could reduce the weight of a tool exchanger. And they wanted to do this while keeping all of their existing operating requirements intact but reduce the weight to help drive higher speeds, lower energy consumption, and improve machine precision.

So again, I talked about these continuous improvement processes. This is a great example where we know what works, but we think we can make it better. Help us make it better. And generative design did just that. It helped them develop a lightweighting strategy that would work in all of the existing form factor requirements and performance requirements and maintain the overall stiffness. We know that stiffness in this application is really critical to the overall performance.

Generative design was able to uncover or help them achieve a 25% lighter design, again while meeting all those existing performance requirements, and maybe most importantly here, is with almost no change to the manufacturing costs. So we've taken an existing design, we figured out how to make it lighter and work in all of the same constraints that we have with no cost to manufacturing.

So for Toshulin's customers, they now have a product that can integrate into their existing ecosystems where they're already using their version 1 product, and now they have a product that is lighter and higher-performance that would swap right in and allow them to then extract new value.

The next story we'll talk about is Newton Engineering and Product Development, and they are using generative design to enable space exploration through new product development. Newton engineering is a really, really cool customer of ours, a firm that's working on engineering and product development with a primary goal of providing innovative solutions to the aerospace industry.

And you've probably seen over the past year some of the work that we've been doing with NASA. Newton engineering has partnered up with NASA to build a rapid closing light and strong capture lid mechanism for the Mars sample return mission. So if you haven't seen this story, in my handout I will have a link to the full story.

But this lid is on a containment device that will be capturing Mars surface samples and bringing them back to Earth. And the way that the process works is that the capture device will open, the mechanism that we're designing here will open the lid, effectively another machine on the Mars surface will be tossing samples into this containment device, and that lid needs to rapidly close to ensure that those are captured correctly and can be returned to Earth.

And in this scenario, generative design was used in a textbook manner. Newton was looking to evaluate multiple materials to find the lightest, stiffest possible design with the lowest cost to manufacture for this bracket. Going to space-- expensive endeavor. Every kilogram counts when we're trying to go to space, so getting that lightest, stiffest design is absolute key.