Art of Possible: Generative Design and Lattice Structures

00:00 Good afternoon. My name is Andreas and Jose and I will try to share some of our experiences about lattice structures. What we will try to cover 4 or 5 topics. The time is very short, so we'll go a little bit fast, but after that you can ask any questions you want. We'll talk a little bit about taxonomy of lattice structures. What are they and how we can combine them, how we can organize them so we know what they are. And then what happened?

00:33 I don't know. About 20 years ago we could build things we couldn't manufacture. About five years ago. We could manufacture things like lattices, complex lattices we couldn't build in the computer. Now we can do both. We can design and manufacture very complex lattices. Let's talk a little bit about taxonomy and then we'll see the need.

01:06 What what can we learn from an experienced designer? Nature is a very experienced designer. It has amazing designs. They know about gyroids, about lattices, about complex geometry and etcetera. So what can we learn from them? We can see their arms hitting each other and they don't get brain damage with severe force. So what can they do?

01:36 You know, if we analyze a cross-section, we can see some intricate lattice inside and say, okay, so what? Well, that inspires to do our own lattices for custom lattices which can absorb energy. Woodpeckers do the same thing. They have incredible energy absorbing lattices. There is another creature which is the strongest creature on earth, on earth, the mantis shrimp. Check on YouTube to see videos of mantis shrimp. Incredible.

02:10 Up to 170 miles an hour. Speed of this claws so it can crush the crabs and have a nice crab meal. In seashells. And the question is, how can you crush a seashell? And it doesn't break its own club because it uses some complex lattice inside like this. If you build it in CAD, you will see that type of structure. In general, if we try to organize this,

02:41 we have on surface lattices we can see like the dragonflies has very nice voronoi and we'll talk about in a minute voronoi type structure. And it's very thin and it's very controllable. Of course we know about two and a half d lattices for from the bees and we use them, let's say in helicopter blades as an infill and their paper. So they don't need to be strong to be a good infill. And then we have 3D beam lattices that those are all in nature,

03:13 three beam lattices which they used in corals and sponges. And also the designer is so good it doesn't give them uniform cross-section. It gives them a helical cross-section which looks like that inside. My favorite is the butterflies. Butterflies. He has the very long wingspan and very small thickness. So the spine to thickness ratio is amazing. How can they do that?

03:45 Because they use Gyroids as an Infills. What kind of what is a gyroid? It's triply periodic minimal surfaces and we'll see some of them. You know, later. So nature is using lattices for high performance designs, either aerospace type or structural or, you know. So today in Creo we can do all of these lattices. We can. They're well-organized.

04:15 There is a single feature and they're parametric. So in other words, we can do voronoi structures on the surface. We can do two, two and a half d lattices, the beauty in two and a half d in one direction. But we can wrap them, we can convert them, we can unfold the surfaces, fold them back and etcetera. So we can do, we can put two and a half d lattices to provide ribs in any, any complex surface. Very simply, one feature everything parametric and 3D beam lattices.

04:48 And you can see here a small heat exchanger which the thickness or the diameter here is bigger and there is smaller. So variable thickness or variable diameter lattices can be done very easily with one command and infill it. Try to imagine the shell feature we use instead of filling it up with air. You infill it with a lattice you choose. So the interface is pretty good. And here is my favorite one. The triply periodic minimal surfaces like the Gyroid.

05:20 You can see we can have good control on the thickness based on either simulation results or proximity to geometry and etcetera. And here is a heat exchanger thicker and thinner and thinner. So good we can manufacture them now and we can design them in CAD. Are we done? Well, not quite. Let's see. There are several challenges. What are the challenges?

05:51 Massive. That element size. If I try to infill to mesh this little one unit cell of beams with tetrahedrons because they're circular, I will not need a lot of tetrahedrons around it. So I will need, you know, about a couple of hundred elements to just do one cell. Try to imagine now if I have 2000 cells or if I have million cells. So it's very impractical to go ahead and say, okay, I'm going to since it's standard geometry,

06:22 I'm going to mess it with pets and I'll do it. It's very impractical because there are way too many. So that's the first one. What is the solution or the antidote to this Well, we have three options when we generate lattices. One is full geometry. Okay. If it's small, you can analyze it like with regular geometry. Now. You have the option of simplified. Simplified ? Well, yeah. Instead of having the cylinder and intersection and the fillet

06:53 and all this stuff, you can have just the beginning and end and it's like a little datum curve and say, Well, then I can see it. You don't need to see it. You can export it as three maps, so you can print it. And if you try to do simulation on that, for example, Creo simulate, understand those are beams. So you can do beam modeling of these massive structures, which is super exciting because now we can analyze them. So now we can build them, we can make them in CAD,

07:25 now we can analyze them and if it's extremely massive or it's, you know, it's random, then you can do homogenize. You can say, forget the even the beams won't cut it. I'm going to take one element, one unit cell, squeeze it in all three directions, find the equivalent stiffnesses. Use that as a solid and analyze it. Of course, you don't expect to find stresses there, but you can find natural frequencies, displacements and etcetera. The other big issue is interoperability.

07:59 So what do you mean? These TPMs or the Gyroid type lattices are not b-rep geometry, so I cannot say file save a step. Why? Because their unique kernel. So we can step it out and bring it to Ansys or Nostrand or whatever is your favorite code. So how do we solve them? Well, the good news is the real time simulation tools like Creo Simulation Live

08:29 can understand TPMs structures. So I can do fluids, I can do structural, I can do whatever I want using real time simulation. In other words, the the the Creo simulation live understands gyroids. The third one I won't focus on that is integration with generative design. We can use homogenized property to generate a complex generative design

08:59 and then infill it with a unit cell. We decided to do the homogenization. This is very powerful for combining generative design and lattice structures. Now. You say, okay, good. We can. Analyze it. We can run on. So what? How can I optimize my design? So the first was make it in CAD, be able to make it in, you know, in real world with 3D printing. The third was analyze it.

09:33 The fourth step is let me optimize it since I have it there. What I can do, we can optimize it. Here is a very simple example of hexagonal mirror you have to support. Here is the simple trick you can use if you go to Creo simulation live and you see a result, let's say the deflection of the plate, you can go and say, Can I have a probe? What's a probe? Something which can tell me the deflection. So where do you want the deflection? Well, in that point, how about the surface?

10:06 So on that surface, then in the surface, you need to tell me what you want. The maximum? Yes, the maximum. So you can take the maximum deflection on the bottom surface and say capture it as a probe and then save it. If you save it, the probe becomes a feature of the model. So every time you regenerate the model, the deflection will update. You say, Aha, So I don't have to redo the analysis. Okay, big deal.

10:37 No, you can do even better because you can go to behavioral modeling and say, can you please plot the lattice size versus the deflection? And you can generate automatically because it's live. It's nearly live. You can get very fast. Like this will take about, I don't know, 60 to 90s to generate these curves. When I was teaching at the university, it will take us a master thesis to do this, to plot the, you know, design variables like, you know, in this case,

11:10 the thickness of the gyroid versus, you know, the deflection. And then at the same time, you can plot the weight and you can come and say, oh, my constraint was 10th of a millimeter, so I can come here, hit the curve, find out what is the size, go to the size, go up and find the weight. And if you are greedy, I don't know, some of you may not. But if you are greedy, that optimization can become a feature. So every time you regenerate the model,

11:42 it runs the optimization to minimize the weight and give you the four in displacement. Now you say, okay, that's too complex and our designs are a little bit more complex. I want something simpler. Well, here's a simpler solution. Let's say I have a drone wing and I infill it with lattice and I say in the probe, now you can say, Can you please keep track of my iterative solutions and my iterations? So you put the load, you fill it out with gyroid boom, and it gives you the result.

12:13 You say, What do you mean? Gyroid? Can I look at it inside? Yes, you can. That's how it looks like inside. Okay. Of course, in the real one, you can really modify the thickness across based on either simulation results or proximity to surface. And you make it super efficient. Okay. And you say, okay, can I try another gyro? Because this is discrete variable. I say Gyroid diamond or whatever you want. It's not easy to do a plot, but you can do a bar chart.

12:44 You can say, okay, now I can tie the primitive lattice and you get another number. And then if you keep doing that, now let me try Diamond and you get the bar chart automatically. This is not Excel. This happens within you say keep track of the deflection for all my imperative directions you need still an envelope and a piece of paper to write down what you try. So you remember the order, but you know you can do that. In, you can do even better.

13:15 You can say, okay, in this direction I can really change the width or the length of the gyroids. So can I do a deflection on the vertical axis in the gyroid length? On the other you say Wow. So I can go to gyroids. And do we call this sensitivity analysis? So this is great, but it can't get any better. Yes it can because now you can come and set up an optimization and say minimize

13:45 the weight, make sure the deflection is less than that, and you can play with four design variables. All the four numbers, they define the gyroid. In in the training and we don't have time. I would love to show you this example. You can generate these three cantilever beam. This is from the training class. But you know, we can put Star Honeycomb and Gyroid and get automatically the deflections. So you can say, Oh man, Gyroid wins.

14:15 So you can find which design wins very fast. It doesn't need to be a master thesis. This is this literally takes less than a minute. Okay. So you can give you an idea. You say, okay, all the structure stuff, linear, elastic, small deflections. It's easy. Can we do something better than that? Yes. Here's another example. A mold cavity here is a mold. It's very complex underneath. In the past we didn't cool the molds. Then we start doing conformal cooling.

14:46 We drill holes, holes, holes, plug this, plug this so we can create a U-shape and do, you know, liquid cooling. Then we improve and say, can we go very close to the to the to the part so we can cool it fast? Because if we improve 1 or 2 seconds on the injection molding, it's very big deal. And they did that. You know, the last decade, a lot of people do, you know, since we can 3D printing, we can put a little snake type hole inside to happen.

15:18 But the pressure drop was too high. So the idea was can we use lattices and big cavities within so we don't have pressure drop? And we filled them with gyroids and they pump fluid here and we get it on the other side. So, yes, can we control the pressure drop on that? Yes. You can find out. For example, here is the inlet. It goes like that. It goes around and goes to the outlet. This is Creo simulation life.

15:48 So you can do that literally in seconds and say, okay, what is this red spot? Well, there is a narrowing there. So it gives you the idea to go take that narrow part and increase it a little bit because you can have more uniform. So you can say, Oh, the gyroid hits the corner here, I need to take this out of the infill volume so you can improve it. First. It gives you immediate feedback. It design spellchecker. Then you can say, How about if I see the profile,

16:23 the distribution of the velocity and you can see that and say, Yes, but you told me we're going to optimize the pressure drop. Okay, let's see the pressure drop. So here is the pressure drop. What you do is simply that the whole idea this is very simple. You go select that surface and say, create a probe of the pressure, drop the maximum pressure drop in this phase and save it. Now, that became a feature in the Model three. Okay. They became a feature of the part.

16:54 If it's a feature and updates every time, I can very easily go and say. Here. I can plot now the dimension of the gyroid and the pressure drop on the horizontal. Okay, this will take a minute. This will take four minutes. But still you can live with it in four minutes. You can find a sensitivity of the gyroid length versus pressure drop. It's pretty powerful. And also it gives you other insights. You can see that, hey, this is a better point than this.

17:26 So the bigger is not always the better and depends on the geometry. Sometimes the flow has a hard time making the U-turn so you can find the optimum size of the gyroid to do that. Super exciting stuff. If you guys don't use them now, you will be fallen behind by your competitors. So or you'll be acquired and you know it will be the same thing, but super exciting stuff. I'm, you know,

17:56 I use them and I'm always amazed on what we can do today. Okay. Then my colleague here, Jose, will talk about simulation driven lattices in Creo and if you have questions afterwards, I'll be around or in the reception. I will be glad to share this information. So let's review a couple of topics on on simulation driven lattices in Creo. First of all, as a foundational information for you, maybe you are aware, maybe not.

18:30 I just want to have a common ground here. You can create lattices, as Andreas presented in his slides with different cell types, but also you can induce variability in the lattice feature using geometric references. So for instance, you can select the curve that is in red in those diagrams. This curve is my reference in this case, and I'm inducing some changes,

19:00 inducing variability. So this is what I'm calling variability based on geometric references. So you can define how big the impact will be. Like in a distance, you can define the diameter of that reinforcement, I will say is a reinforcement of my lattice structure. So the beam diameter is bigger in, in, in the proximity or that particular reference.

19:30 Okay. This is something possible having around for several releases. Now you can do it any time you want. Now the topic discussed here as well was about simulation or optimization. You in the latest modeling context? Well, in the case that you are using, for instance, Creo six, Creo, eight, etcetera. I wanted to give you this progression chart. I will say.

20:02 So if you are using Creo five, for instance, you can have there the simulation and optimization of lattices using. Idealizations. So this concept presented by Andreas that there is one concept called. A simplified representation of a lattice. What it means is that when you send this model. To simulate all those beams instead of meshing individual beams,

20:35 you are considering inside Creo simulate as an idealization element. A beam in the modeling side is a beam in the simulation side. What is what is the meaning of this? That your your simulation will be 2050 times faster. That's the impact of using that. So going in this progression Creo six, we introduce material homogenization,

21:05 another concept to speed up further the simulation and optimization of lattices. So in the in the homogenization side, you can further speed the computation of of your problem. Also, in Creo six, you can find the real time simulation using Creo simulation live. Going further Creo seven. We including more functionalities like simulation additive

21:36 process simulation. So using a third party. But now you have a the capability to simulate the printing process, especially for metal. Creo eight. Automatic lattice variability based on simulation results, not only based on geometric references, which is very powerful, but now using the results of the finite element analysis problem. So the results, let's say a stress the stress map,

22:08 and then let the lattice feature to adjust the variability based on the stress maps. Okay. Creo eight Increate for beam based lattices. Creo nine. You can do that. Also, you can do real time simulation for fluids and in Creo 10 this variability extended to the gyroid types so deeply periodic minimal surfaces. So we are expanding and extending.

22:40 So my my advice to you is please use Creo 10 now because it's the most complete set we have for you. Please consider it. So let's go deep down a little bit on the lattice variability piece. The lattice variability based on simulation works like this. You simulate your part, you get your stress paths, you export the simulation results into a file. Okay? You export into a file.

23:14 This file will contain any number of sample points that that will be random. In the geometry, it could be 200, 100, etcetera. Then you come back to your lattice and use that file in order to induce the variability. You will see here that if we can say that the blue areas in the stress path is the lowest stresses in that particular case,

23:45 then is where the beam diameter is a minimum and where the stresses are bigger then you have thicker beams. Okay, so that's the effect. Bigger stress, bigger reinforcement. I will not explain this map, but this is in the a document that you can see in the in the in the application. But here is a map of the progression.

24:15 So, for instance, if you want to do simulation of lattice beam based lattice and you are using full geometry, you can simulate it using Creo. Simulate Creo. Simulate live Ansys Simulation Creo flow analysis. There are several use cases where you have the eggs are not still covered. This is information for you. We are working on that. Eventually we will have full coverage. But I think this is important for you to reduce any any potential frustration that why this is not happening.

24:49 Anyway, so we are working on that. You will have it handy. Let's see now as a final topic, some special cases, and this is special cases I, I included because this is to respond to some common questions. For instance, the first special. Topic here is. Structural ribbing. So this notion of having a very thin in this case. Yeah, design, very thin wall design.

25:25 And then I want to reinforce it with ribs, structural ribbing. So the question can be in many ways, but for instance, is, hey, where is the structural ribbing command? Well, there is not there is not a structural command. It's a procedure. You start with two and a half D lattice that, by the way, I will present it in the tips and tricks session later in the week. If you attend that, you can see it live, how it works.

25:57 But basically the basic concept is you create a two and a half D and then you wrap it around. So then you have your structural ribs and there are many ways to do it. Okay. This is one special use case that I think is is important to know that is possible. This is another case where we were using the new Auxetic lattice that, by the way, this helmet will be on display in Xtropolis you please visit the Creo booth

26:30 and Xtropolis here. What we did is to create an auxetic lattice which is new in Creo 10 and rapid normal to the surface. Also, there is not a command to do that. It's a procedure just using regular Creo functionality. You can you can flatten the quilt, you create the lattices, they're normal to the flat surface and then wrap it around again.

27:02 And then you have the the shape that you want. Another very interesting topic is custom cells. So there are only so many cells that we can provide to you. Meaning we started with a few cells and we added more and more and more and more, and we are adding more cells at every Creo release. But the reality is that you will find that another way to do a lot a cell, right? So that's why we open up this to,

27:34 to you be at the custom cells so you can create many different custom cells in Creo using any Creo tool that you want as long as you put it inside a small box. This is an example of creating a model just using curves. Straight curves. When you use this part because this is a PRD file in the lattice feature,

28:04 it will end up being something like this. So this individual curves lines, it will became beams, solid beams. Okay. This is the first notion that I would like to communicate here. So just create datum curves and you will have beams and then all this lattice is another lattice for you to be used and simulate and so on. This is another very interesting example.

28:34 If if the cell that you are designing is based on quilts. Check. What is the effect? So you create a quilt. Could be as complex as you want as long as it's in a small. Is square a cubical type of shape in order that this side connect to the opposite and the other the top connect to the bottom, etcetera. This is what we are looking for as long as it's possible.

29:05 Here we are in this example, we are combining getting a custom cell based on quilts. Use the methodology that I told you to create a two and a half the lattice in flat state and then wrap it around. And here you can see we just put it like that. Let's say zero degrees and here we are rotating it 45 degrees. So in order to make it more printable. Okay.

29:36 But this is the effect and this is the kind of flexibility that you have today. The chiral type of cells. So this is a cell. When you press a load here in the top is not only the form in the direction, but also rotates. Okay. So then the structure in itself, we have another properties not only structurally but in, let's say, hold loads,

30:06 but also rotate in some way depending on the dimensions. So you can optimize this, you can you can create a the different features in order to measure the displacement in rotation and then create the behavior that you want in the structure. This is from a Creo solid model into a lattice using the user defined or custom cell type. This one is also interesting. This is the resulting structure.

30:40 Okay. When it's propagated inside this cube. But check the way how the model of the cell is being done. They are not even connected. So you design it in a way that will connect with the adjacent cell. So just I think it was interesting for you to see the possibilities and variations that can be possible. It another example of a solid model.

31:12 So this is another solid model. It could be this shape, could be this shape, and then it's propagated. It can it will just fill up the a the space here is an exercise of creating this sorry, creating this cell. And then I have an stochastic surface lattice inside and I can make it as complex as I want. For sure. More complex, more time. But if it makes sense for you,

31:43 what what we are here giving you is the flexibility to to do that. Well. To to finish this very short presentation that I just presented the cases, I will be happy to explain it if you have interest, just to say that there are many possibilities in Creo today. I think we are reaching a maturity state. Very interesting that we have many tools in order to design

32:16 and optimize different metamaterials which metamaterials in in in essence is create a material engineered to have properties that does not exist in nature, which is the concept of auxetic. For instance, even in Creo 10, if you attend to to my session in in in Wednesday and Wednesday, I think is Wednesday, we will talk more about Auxetic being cells that behave

32:51 differently and maybe. Behave in a way that there is no other material in nature that can behave in that way. Okay. So this is the this is the line for the future of the lattice modeling in Creo more and more creating meta materials for you to push the boundaries of design. Thank you so much. This is the this is the last time.