Using Advanced Features in Generative Design

GRZEGORZ BOROWSKI: Hello I would like to welcome you in our presentation about using advanced features in Generative Design. My name is Grzegorz Borowski. I am Senior Product Owner in Autodesk. I've been working in Autodesk since six years, mainly involved in the Generative Design product and automated modeling. And together with me, we don't have Tomasz Kasperek, that we build this presentation for you, but I'm sure you will be able to meet him at Autodesk University.

So let's just jump right through it. But first, let's go to-- let's have a quick introduction what Generative Design is.

So Generative Design is a design exploration technology available in Fusion 360. It simultaneously generates multi-CAD ready solutions based on real world manufacturing constraints, cost evaluation, and product performance requirements. So in simple words, what that means is that if we look at traditional design, when you want to design a part at first, you need an idea, and then you have to start putting it into life.

So you build your CAD model with CAD software, with computer aided design software. Then you try to validate it. You validate it through computer aided engineering software, simulations, et cetera. You check for stresses, trying to simulate it with real life conditions, loads, constraints, et cetera.

If for some reason that doesn't work out, you have to make modifications. So you go back to computer aided design. You make modifications, and you iterate through this process. And later, on you go to computer aided manufacturing. And you do the same.

If that doesn't work, and for some reason you didn't consider the right tools for the job, a model is too big in some area, you have to go to computer aided design again, and you iterate through each of those parts of the work. This can be time consuming, and also it needs a lot of cooperation between the designers, simulation, manufacturing colleagues, different departments. Only then, after some time, you are ready to manufacture.

So Generative Design contains all of it in one package. So it provides you a design that is validated, simulated with computer aided engineering technology. So we take into account stresses, frequency buckling. We then, of course, have manufacturing constraints that ensure that your design can be actually made.

So how do you do it? Generative Design is pretty straightforward. And what it provides is this bar on the top of the Fusion, and it is placed here. And this bar contains all the features you need to set up your model. So let's go through some of those options.

To generate your studies, you need to first create a study. And you do it through in this section. And then you have to define preserve geometry, optionally obstacle and starting shape. You do it from here.

Assign body and boundary conditions and loads and constraints. This is taken care of here, design conditions. You need to select material, define manufacturing methods, set objectives, and, if needed, adjust the resolution, which is under this tab.

After you do that, you are ready to go to check your setup by our precheck technology. And what it does, it checks your model for any potential issues that you might face. And you can do that through this option right here. If there will be something wrong with your model, you'll get not a green checkbox, but an extension mark, et cetera, for example. It is highly recommended to go inside this option and check your design. Once you are done and everything seems to be correct, you can generate your study through Generate Option and explore your results after a while.

So Generative Design comes with some standard features, and they are-- I divided them in three parts. First is Design Control, which is more related to shape itself. So you do have preserve geometry. You have obstacle geometry, starting shape, obstacle offset. Manufacturing methods, you can use unrestricted additive 2 axis cutting, 2.5, 3, and 5 axis milling.

You can also set up your model 2 to mimic real life conditions by applying loads like forces, pressure, moment bearing loads, constraints like fixed constraints and frictionless. And of course, we cannot forget about objectives limits. You can set a mass target to your design, the mass that you don't want to go below, and also safety factor for your stresses.

Generative Design provides also advanced features, and it does it through the Generative Design review options. So the preview content, we have several previews, first of all. And we have Experimental Generative Solvers and features that enables some advanced functionality that I will cover in this presentation. We do have also dive casting preview, which enables you additional manufacturing method. We have fluid path option that enables generation of a flow path optimized for pressure drop based on your geometric and performance requirements. In this presentation, we will concentrate on those first two options.

And first of all, before we do, I would like to cover what they actually contain. So if we look at Design Control, we do have symmetry constraint, and we do have manufacturing method, die casting as an option. There is boundary condition, additional boundary conditions. We have remote loads, remote constraints, point masks, remove rigid body modes. We have additional limits, objective limits, like frequency buckling and displacement limits. This will be all covered in today's presentation.

But first, let's see how to enable it so you can actually play around with it by your own. So the preview features are available for all Generative Design users, and they are available to you by going to Preferences on the top right corner of Fusion. You just select Preferences, marked with red box here.

What will happen in the dialog with Preferences will be open, and on the left side, you will see Preview Features. Once you click on it, the right side will show you all Preview Features that are available in Fusion. We are trying to select a Generative Design section, where you do see this in those options, like die casting, fluid path, experimental generative solvers, and features. You just enable whichever you wish to use and apply your changes.

Design Control. As I mentioned to you, the first an advanced feature that I would like to show you is symmetry constraint. And application is we use it to drive shape generation by enforcing symmetrical design. We also apply symmetry constraints in final design. So when you already calculated, generated your design, you can export it to design section, where you can edit it through this plane modeling.

How to use it? Go to Design Space, Section. There, after you turn on the precheck, you will see-- after you turn on the preview, sorry, you will find additional options, like symmetry plane. You click on it and the dialog asks you to select planes that you want to use for symmetrical constraint. You choose construction planes in your model and you click Accept.

There are some requirements to this. So now, we can currently, we support up to three symmetrical planes. And those symmetrical planes have to be orthogonal towards each other. The setup. So the preserve and obstacles need to be symmetrical. And this needs to be fulfilled in order to be able to get proper results. If your setup is not symmetrical based on preserves and obstacles, you will get a warning about it in the precheck section that I discussed at the beginning of this video.

The important part of it is that loads and/or constraints do not have to be symmetrical. They can be totally asymmetrical and still will be able to provide you the symmetrical shape. Some manufacturing settings can be contradictory to symmetry settings. So imagine let's look at the model on the right. We do have a model containing of two preserves. One is fixed at the bottom. Second has a load, [INAUDIBLE] load and you do have-- we do have two symmetry planes here, one in xy and second in yz plane.

So if you choose a manufacturing method, let's say 3 axis milling, or that enable material reduction just from z plus, then you won't be able to get symmetrical outcome, and because there is no way to remove material from minus z. So in the end, your result will be most probably a box of material without any material reduction.

Symmetry is supported for unrestricted objective 3 axis milling and 5 axis milling manufacturing methods. It's not available in die casting that I will cover in next slides.

The good practice is to use symmetrical set up, total symmetrical setup so you do-- I mentioned about preserve and obstacles that is required. But starting shape can be unsymmetrical. However, it's a very good approach to have it symmetrical, as well. It will benefit your result. And as I mentioned, always verify symmetrical setup for precheck. It will assure that your results will be accurate.

So if you look up on the result on the right side of the slide, and you see our model, it has two symmetry planes. If we don't run it with symmetrical constraint, you will get the result in the middle. You can see that it's totally unsymmetrical. And while it is accurate from a simulation standpoint, it will survive the loads that you applied, It might not be pleasing, pleasing to an eye. Maybe for visual aspects, you like to have your results symmetrical. So after you apply symmetrical constraints, you get symmetrical outcome that you see on the right.

And now I would like to show you another benefit of symmetry constraints, which is visible in Edit mode. So let's look at the model here. So as I mentioned, we have two preserves and we have two symmetry planes.

We can run our model. In this case, it's already calculated, so we can choose and export the design we want. And what we wish to do is to export, the export the design outcome we selected. In this case, it's also already done. So we just pick the outcome and go to design section.

Once we are here, we see that the construction planes were exported to design. We can Edit Model in this plane option, and we can use various tools to edit, do the edits. So we can select few elements, but for the purpose of this video, let's check the larger area.

You can already see that other faces that were marked by yellow are the places where you will be doing the symmetry modification. So you just drag your elements in the way you-- the version you wish to modify, and you finish your form. This way, you can see that dragging just one, let's call it leg of this structure, because it was symmetrical, it infers your modification to other parts, as well, because it's according to the symmetrical planes that you chose.

So other advanced functions that I would like to show you is die casting. Die casting is an additional manufacturing method that is applicable, as the name suggests, for die castable models. You can find it under Design Criteria, Manufacturing Option. In the dialog of manufacturing methods, you will find additional method that you can check, die casting. You can select the injection direction, xyz. And those are global coordinates. And if you select one, you will get one additional outcome. If you select two, you will get two outcomes. And those additional directions, of course, this is plus and minus y. And this is the injection of your

You can specify minimum draft angle, minimum thickness of your model, and maximal thickness. All of those options will be considered while making your design. There are some requirements, also, to follow with this manufacturing method. The most critical is that setup must be die casting friendly.

So what we mean by that, there might be some complex setups with a lots of preserves with holes with preserves flying around, with holes that cannot be actually-- that are not die castable. We might have a-- it might be tricky for Generative Design to create a design on the setup that is not die castable and die casting friendly. So it is the most important factor.

It's also worth to mention that some advanced features are not supported together with die casting. So symmetry, remote loads, remote constraint, point mass, remove rigid body molds, are not available while using die casting manufacturing.

So on the right, you do see the set up model, which contains five preserve geometries. And they are, four of them are constrained. The last one is being pulled upwards with a purse, and you can see the result. This is die casting result for direction plus x, minus x. So the injection goes in along the x-axis. And you do have draft angles for purpose, also, of injection. Such a design should be die castable when you wish to manufacture it.

Boundary conditions. So there are a few boundary conditions that I would like to cover. First is remote loads. This is an additional boundary condition that allows you to define a point load that there are no direct-- that are not directly on the preserves. This is commonly used for simulation of assemblies.

And how to use it? It actually is placed as any other load. So you go to Design Conditions. You go to structural loads. And from the dropdown menu, you search remote force or remote moment, and you select the target surface that you wish to apply your load in.

Requirements are similar to any other loads. So however, in this case, the first and moment loads here, the remote force and remote moments are supported here. However, they have to be placed on the face. And the target has to be a face. It cannot be a body, edge, or point. And the good approach is to have remote loads realistic, like as any other loads. Some small loads and relatively small loads to your design can lead to disconnection.

And the same could happen with loads that are self-cancelling. So you can imagine that you have one preserve, where you put two loads with opposite direction, same magnitude. In this case, they will be self-canceling each other, resulting with zero force on this preserve, which may lead also to disconnection.

On the right, you do see a dialogue for remote force and remote moments you can use any kind of way to apply it like any other load. What is important, you have one additional option here. You specify the-- let's create an anchor point where the load is applied. So you can see from those examples on the left, you have force. On the right, you have moment. And the load, even though we selected the faces of those cylinders, the point that is dragged through the force, where the force is applied, is not directly on the surfaces. This way, you can mimic some other parts that are having the effect on your structure.

Remote constraints. They are used to define constraints that are not directly on the preserves. We can define joint and jointlike connection. They are also commonly used for assemblies. How to use it, similar to loads, you can find them in structural constraints along with all other structure constraints. In this case, you go to dropdown menu and try to find remote constraints.

Once you've done it, you apply the target, so the surface you want to have your constraints connected, and you have to apply degrees of freedom, and that are visible here. You have six translations in global coordinate system in x, y, and z. You have also rotations in x, y, and z. And you have to select the anchor location to place your remote constraint.

A few requirements, also. Only faces can be chosen as a target point. Edges are not supported. A good practice, selection of degrees of freedom should be done with caution. And what I mean by that, if we look at the model on the right, you do see-- you can see two cases, one at the top and one on the right. In this case, we have chosen our remote constraint to be attached to this particular body on the left. And it stays directly on the face of the body.

In another option, we moved a bit to our anchor point to the left, but they all have the same degrees of freedom. In theory, if you leave rotations around z and z-axis, free, this will mean that this model will be able to rotate. Right it will rotate differently in those two cases, because it will rotate around the anchor point that you selected. That's why the position of it, it's important.

However, this is just to show you how it works. But it will not-- such a setup will not be able to run in Generative Design, as it is not fully constrained. If you wish to run something similar, of course, it depends from your purpose you are trying to solve, challenge you like to solve. But you can, for example, fix another body with translations, and that will prevent your model to rotate to infinity around anchor point. So you will have a fully constrained setup that is ready to be run.

But it depends on your application. You can use any other constraints. What is most important, be sure to fully constrain your model.

I would like to show you a few examples what you could do with remote constraints. So with 6 degrees of freedom that you can use in remote constraints, you are given a lot of flexibility to simulate your case. So first of all, you could do fixed constraint. This needs all six degrees of freedom to be checked here. You can do a set up without rotations. We leave translations free, and we mark only the rotations. We can do ball joint, which means we want to have only translations-- we don't want to have any, sorry. We don't want to have any translations. We allow rotations.

We can have cylindrical joint, which means that these surveys have free translations and rotations along around the z-axis, in this case. Regular joint, we allow only rotations and frictionless, as described here. We fix movement in y and rotation in x and z. But there are many, plenty options that you can do here. Those are just a few examples.

The next advanced feature that I would like to show you is point mass. So point mass represents the effect of external component attached to a generative model. And it can be found as a separate option under Design Conditions, visible here.

So you go to point mas option, and the dialog allows you to select define position, choose preserve to which point mass is connected through those options, and define mass.

The requirements point mass has only an effect in load cases that contain gravity load, or in load cases where the frequency limit is set. The point mass will not have an effect on the mass target or mass minimization values. It will not be also added to the mass shown in Explorer.

So in Explorer, you will see just the mass of your design, not design plus mass, plus point that you've chosen. The good practice is that point mass should be realistic and point mass should not have significant impact on stiffness. So avoid point masses that are way, way heavier than your design. And in this case, on the right, you can see a model that aspired to be a table. So there is a pressure load on it, and there is also a gravity. And what I did, I placed a point mass in the middle of this table attached to the top surface.

So on the top, you do see our results from that. You can see that the design is relatively symmetrical. The material is distributed symmetrically across the design. However, if you decide to do a point mass somewhere on the edge, somewhere closer to the edge, you definitely see the impact. If we run it again, we do see that the distribution of material is different than what we saw on the top picture. We do see a lot more material around the point mass, which is here, trying to support the loading.

There are many various cases where you would like to use point mass. I can imagine trying to design a car or truck with some loading. But you can use it for anything you wish, where you want to simulate the effect of some additional mass on your model.

Remove rigid body modes are used to stabilize generative study when model is not entirely constrained. It is commonly used for simulation of assemblies, as well, and it is available to you under Study Settings. Once you click on the Study Settings, you do see a dialog with resolution, alternative outcomes. But there is also Intertial Relief, which is another way to remove rigid body modes. In this case, you just check this box and you should be good to go.

There are a few requirements that would allow you to create accurate results. First of all, it is recommended to use this feature without constraints. And if you use constraints, the sum of loads must be equal to or close to 0. Basically, loads must be balanced.

What it means, when you look on the right, on the one on the right, you do see two cases. Hers is not constrained at all. Usually in a simulation, this would not be. You would not be able to run it, because it's not constrained, it's not stabilized. But here, we have two forces balancing each other, and you will be able to get results from such model. If you don't use constraints, be sure that they will not affect the loading that you have. So in this case, you have loading in y direction, and you do not constrain the model in direction of loading.

Best practice, instead of constraints, please use all known forces, contact forces, reaction forces to stabilize your model. And make sure that the loads are balanced in the direction where the constraints are applied.

Now we head to Objective Limits, and there are a few.

First of all, I would like to talk about frequency constraint. This constraint allows you to prevent resonance phenomena in design part by including limit of first natural frequency. And where you can find it is through design criteria, where you have all your objectives and limits. After you turn on the preview, you will get additional option, such as modal frequency. Here, you just check it and you put your first mode frequency in Hertz. And you apply your choosing.

So a few requirements. Only first natural mode is supported. Frequency must be applied in Hertz, and this might cause some longer runtimes to your design. But the good thing is that having this first frequency is very beneficial for your design.

The good practice is to select minimal first mode frequency to be higher than excitation frequencies. This is because when you-- basically, every model have some frequency modes, and the first one is the smallest one in volume. When you have your excitation, when it is equal, the theory says that when you have excitation equal to modal frequency mode, any frequency mode, you are in danger of going into resonance, which can destroy your structure.

So if the excitation frequency is lower than your first frequency mode, you should be on the safe side with your structure. That's why we set a minimum frequency mode, first frequency mode as a limit to save your design.

On the right, you do see a setup where we apply some of this limit. So you have two preserves without, and they are fixed. And when you do a frequency-- when you do not do frequency limit, the first mode is actually 340 Hertz. And you can see the result here.

What happens if we select a higher limit for our structure, let's say 500 Hertz? What it will do, it will provide us a design with very small, equal to 508 Hertz. And you can immediately see that the design is more bulkier and stiffer. That is an effect of a frequency limit.

Buckling option, prevent loss of stability of a design under compression loading, it is also essential during simulations. You can find it under Design Criteria in objective section, along other limits. Previously, we looked at the modal frequency. Now, we check buckling, and we put a safety factor that we want to fulfill.

Requirements is that only one load case is supported with such a setup. And it will also extend your runtime for this kind of-- if it contains such limit. You do a single load case per generation. It is advised to choose the most critical load case for your design. But once you do that, you can prevent buckling situation.

So if you look on the right, we have a model containing of two preserves. One is fixed. The second one has a compression load.

If we do not do any buckling limits, you will have very slender design. That is like the buckle. When we have such a design, of course, we can export it, go to a simulation, and do buckling analysis. But why do so if we have such a limit, also, in Generative Design that can prevent such a situation automatically?

So when you set your buckling factor of safety 2, you can see that your design is way different. It separated into two supporting geometries that are safer in terms of buckling. If you put factor of safety 4, they get even wider, supporting your design, and you have those supports that add additional stiffness.

Another advanced feature that I would like to cover here is displacement limit. So it's being used to set maximum displacement limit for your Generative Design outcome. You could find it in Design Criteria, as well, in objective section. You can check displacement option under limit section. And then you can use global elements or local. If you choose global, it will affect the whole design. If you choose local, you can select parts of your model surfaces, edges, points that you want to prevent from some extensive displacement.

And what you need to do is to set the direction in which you want to prevent it and value that is allowed. Once you set it up correctly to your needs, just press OK and it will be applied.

And displacement limits, as I mentioned, they could be global and local. We already covered that. There are good practices to have displacement limit realistic. If it will be-- if those limits will be too strict, too small, it will prevent material reduction, because we will be asking for a very stiff structure that cannot displace, that does not have some large displacements.

In the case where many displacement limits are needed, please consider using global limit. Those limits also have an effect on your runtime. So if you have many of them, why not just use a global one?

On the right, you do see a model of a setup. For this purpose, I used the previous one that we already know. It's from symmetrical constraints. But this time, we have run it and we looked at displacements. When we don't have any displacement limit, we have a mass of 0.37 kilograms and max global displacement of 0.113 millimeters.

When we add displacement limit, let's say we'll be very strict for this case of 0.025 millimeters in each global direction. We'll get much better shape than the previous one. We already see that they have an effect on structure. The mass almost doubles, and the maximum displacement is 0.036.

You might ask why it's not 0.025. And it's because this limit corresponds to each global direction. So it will be covered in x, in y, and in z. So it will not allow the displacement could be higher, but only in specific, in each global coordinated direction.

There are some practical examples I would like to show you. In this case, let's focus on the example we can all find on YouTube. It is titled, Optimized Space Telescope Structure, Penn State Capstone Video. The authors are the Penn State students mentioned here, and I would like to play it for you. It's three minutes, but it shows you the application of advanced-- some of the advanced features.

[VIDEO PLAYBACK]

- Supernovas, quasars, neutron stars, and black holes are all magnificent phenomena that happen within our universe. But how can we get imaging of these fast forming objects? NASA research engineer Ryan McLellan and his team at the Goddard Space Flight Center are developing a new telescope to capture these items in space. Its name, Star X. Mission Star X is a combination of a UV telescope, an X-ray telescope, and a rapidly responding spacecraft that captures images of high energy objects in space.

Our Penn State team, composed of four student engineers, are working with NASA and Autodesk to design, build, and test an optimized structure for the Star X telescope using Generative Design to advance research on how these technologies can be used in spaceflight. We used Autodesk's program Generative Design within Fusion 360 to create our alpha prototype of the detector bench, which was made with titanium and had asymmetric geometry. In the final prototype of our detector bench, we reduced the number of holes and changed the material to aluminum 6061, since it is a material that is commonly used in space applications.

With the final design of the detector bench complete, three FEA simulations were conducted in Fusion 360. First, the static load test. According to the NASA requirement, the detector bench must support approximately 956 pounds from the launch force and have a minimum factor of safety of 2 or higher. The simulation showed that our prototype met both of these requirements.

A static load test was conducted in Penn State's Materials Characterization Lab and confirm that the park could support over 1,000 pounds, indicating the simulation was correct.

Second, the modal frequency test. According to the NASA requirement, the detector bench must have a first mode frequency higher than 100 Hertz, with the loads of constraints included. Our simulations show that the structure passed the requirement by having a first mode frequency of 174 Hertz. We reran the simulation without the loads of constraints and found the first mode frequency that we could compare this to in the lab.

A physical modal frequency test we conducted on our prototype at Penn State's Laboratory of Sound and Vibration Research. The detector bench was suspended freely from one hole so that it would be symmetric along the diagonal. Next, an accelerometer was attached to the part, and a hammer was used to tap each hole in the structure, which recorded the data into a computer.

This graph displays the postprocess data that was collected in the lab. As you can see, the first mode was found to be 127 Hertz. This is very close to the 129 Hertz found in the simulation. We can conclude that the simulation is correct in comparison to the real world results. Displayed here is the mode shape found from the laboratory data. As we can see, it is the same shape that was generated by the FEA simulation on Fusion 360.

Lastly, the thermal conductance test. The detector bench must have a total thermal conductance, which is neighboring components less than 0.1 watts per kelvin. We were able to meet this requirement by adding G10 washers made of fiberglass to our prototype. In this capstone project, we were able to meet all the NASA requirements as well as complete physical testing in various labs, to confirm that the FEA simulations were accurate. The Star X mission is one step closer to becoming a reality.

[END PLAYBACK]

GRZEGORZ BOROWSKI: So I would like to thank these students for doing this. You can find many similar applications on YouTube, on many articles, where you can see how very satisfying, at least for me, I think, that Generative Design is being used widely to produce often very futuristic and very important parts for across all the industries.

So you can see in this video that what they did is actually they tested parts created by Generative Design for frequencies. We could see point masses that were applied there. So those additional advanced features are being used, and pretty successfully, as the validation showed. And I would like to-- with this, I would like to encourage you to really use it, because it's there. It's available for all Generative Design users.

Before I'll let you go, I would like to share a few good general practices. And that applied to all Generative Design. So let's go through some of it and some of those.

And so the first point is to investigate your model category for any precheck warnings. Any warnings can have a direct effect on your results. I already mentioned to you where to find it, and it's true, there are many warnings that could benefit your set up, such as symmetry, such as interference between preserves, obstacles, et cetera. There are many of them. Please use them. It will speed up your setup process.

Preserve an obstacle and starting shapes. It is recommended to avoid relatively small preserves, as it may lead to disconnections. Ensure realistic line of sight between all preserve geometry. This basically means that you can imagine where you have two preserves, and in the middle you have very huge obstacles. This means there is no line of sight and their material will not be able to connect through those.

Even if the line of sight is very small, it might be tricky to connect preserves throughout some small hole, so be sure that this line of sight is realistic.

Avoid interference between preserves and obstacles. Use starting shape to drive your design. This provides quite a few benefits. You can really affect your design with using a starting shape.

You could also in some cases push your optimization farther. There are cases where the available iterations are [INAUDIBLE] for ourselves, because of the large material reduction that is needed. And sometimes when you want to use a thinner starting shape you allow a larger material reduction in your design.

But also, please remember to avoid very thin structure in starting shapes, as it may also lead to disconnections. Those starting shapes need to be connected to preserves, to all preserves. And please avoid unnecessary details, like part numbers, complex geometry, et cetera.

When you use material, be sure that material is realistic. When you look at boundary conditions, the loads and constraints should be realistic, as well. As I mentioned, I believe I mentioned this in remote loads, we have to be realistic and assess. Please avoid low loaded setups, because this may lead also to disconnections, and the resultant load at given preserves should be a nonzero value. So avoid those self-canceling loads that you might get.

Avoid loads on a small area. Manufacturing methods, not every setup is valid for each manufacturing method. So for example, for 2 axis cutting, the setup will be totally different than 4 or 5 axis, 5 axis milling. Be sure that you choose the right manufacturing methods for your design. The tool dimension is also very important. It should be reasonable compared to user setup.

And please ensure that the obstacles are not preventing material removal. This could happen when you, for example, have a preserve and you surround it with obstacles and the tool dimension prevents the tool to actually go inside the preserve to remove material.

If it comes to a resolution, change your resolution based on your setup. So in general, complex setups might need higher resolution. And preserves might also need higher resolution. But this doesn't mean that accurate resolution does a better job, it can provide better results always. First of all, it comes with a price of performance. Second, there might be cases where default or even coarse resolution is better for your design.

I would like to thank you all for going with this presentation with me. I would like to encourage you all to use advanced features in Generative Design, because they are there for every user and everyone of you who use Generative Design, and they bring a lot of control to your design.

Thanks again, and I hope we'll meet in person at AU. Thanks.