0:00 Hi everyone, thanks one for being here, thank you all for being here. My name is Christian Waldvogel, and I represent a small startup from Switzerland that goes by the name of Spherene. I will tell you a few sentences on how spine came to be, and then go into the details, more or less, of what spine the material, or spine the geometry, is, and I’m really thrilled to be trying to demonstrate our software for the first time, as it is, will be released in the coming weeks.

0:44 So it actually all started just 2 kilometers from here, when I was in an artistic residence here in Berlin for half a year. I have a background in architecture, but I spent 15 years as an artist, and spine actually is founded in the jos of three projects that started here. One was a small one about Archangel Gabriel being expelled from heaven because he couldn’t tile the plane using the holy heptagon. Another one was about a machine that feeds a steady drop of wax into a random motion on the, the bottom, which creates the most regular shape, the sphere, it’s one of my favorite projects.

1:37 And the third one is about this quote, it’s 2 and a half thousand years old, geometry was used to try to describe the, the shape of the universe, and already then it was understood that maybe the universe is something like an inverted sphere, a sphere that has its center everywhere and its boundary nowhere. A couple of years and hundreds of paper models later, we found that spine in 2018, and realized that this whole undertaking was about minimal surfaces.

2:16 Now, for those of you who don’t know what the minimal surface is, if you imagine how you describe something that’s curved, you would look at it in one direction, and in the other perpendicular to that. If those curvature are equal, you get a sphere, if those two curvatures are zero, or one of them is zero, you’ll get a plane or a kind of cylinder type shape, and if those curvatures are equal to the amount but with the opposite sign, like one like this and the other one in the other direction, you’ll get a minimal surface, the minimal surface is in a way the opposite of a sphere.

2:57 Now we are, we all know the TPMS, triply periodic minimal surfaces, for example the gyroid, and we have something that we call ADMS, adaptive density minimal surface, except from being minimal surfaces and having two labrants, they’re vastly distinct from the TPMS in that, for one, they are surface conformal, meaning they always end on the outside in a, you could say, correct way, as can be seen here, where you have a perpendicular entry point to whatever geometry you, you, you fill with, and you have somewhat equal distances between where the minimal surface joins its bounding shape, we call this the envelope.

3:52 For example, you can do the, you can use these shapes to do organic grading, simulate or emulate shapes you might know from nature, you could use it to, let’s say, lumin, emulate the mass of aluminum in a part that’s printed in steel, or we could use it to protect something very valuable from tremendous loads. So the characteristics of spines are that they are controlled, whatever shape you put them in, whatever, ever, thickness you want to apply, whatever position in the ideal range you want it to have, if you want it to end correctly or incorrectly, so to speak, they are inverted spheres, as I said.

4:46 You can see this here by these graphs that show the, the distribution of face normals on a sample, and you can see that the gray spikes, which is the distribution of the gyroid, is all but regular, and that’s very logic, because the, the gyroid, as any TPMS, has to end correctly onto, on, on a cube, or it has to end where it started, right, so that it kind of, that it, that it fits together, so it cannot be isotropic. Whereas the blue print is a sphere, a perfect sphere, and the orange print is a spher, high hus, fine, the small deviations you see is basically because of, we use voxels to compute them, so it’s, it’s kind of a our own induced error.

5:44 Spines are predictable, you can see that they are extremely regular, that they behave in a way in a compression test which can be somewhat described by the, the, the formula on the, on the, bottom left, stress is related to the strain, and well, strain to the power of one divided by strain, and it will break at the equivalent of e, oil’s number, and will yield at the equivalent of strain equals one. So these curves, the dotted ones, are theoretical, and the, the ranges are measured using titanium samples that were done four years ago, so this is kind of an experimental result, and what you can also see is that those different samples have yield and break points that are again related in exponential curves, the gray dotted lines.

6:46 Springes are stochastic, they’re not regular, but, yes, they are regular, and you, this can be seen, for example, in the fact that a meaningful grading happen, happens, in the same row of number, in the same sequence of number, that we all know from the, the aperture in a photo camera, right, so if you want to, if you start with, say, 4% volume fraction, or density, as we call it, and you want to increase it by one step, you would go to 5.6, and not to 4.1, or five.

7:23 All parameters, or all statistics of these shapes, they always follow this same rule, so the pore size estimate, or the size of the outer features, or even the genus, or the triangle count, they are all related by this square root of two series. As every minimal surface, they, they are double domain, they have two labyrinths that don’t ever intersect, they are surface conformal, throw it into whatever shape you want, it’s always going to correct, and they allow for an intuitive control, which uses basically the three parameters: density, which can be seen in the sample series in the Y or X direction, vertical, the thickness, and left and right the surface bias, if the surface is at the ideal center or offset to the left or to the right, which has obviously quite interesting effects on the stiff and, and damping behavior of a structure.

8:33 As one could imagine from what I said until now, the stress strain curves from these different samples are closely related, so changing the density by, let’s say, one step, and increasing the thickness by one step, would lead to a predictable difference in performance, and this goes for the thickness, the surface bias, and also for the equal volume. We can always, of course, find the, the, the ideals, or the, the optima, for, for parameter combinations, but we don’t need AI, so there’s no AI involved here, this is, let’s say, insight, not AI.

9:20 So we implement this technology as an API on the cloud, it’s cloud native, and we currently provide access through a plugin in Rosor 3D, where we have a very simple workflow: you load an envelope, which can come from any CAD system, whatever format, you set the requirements, and then you have the API compute the meta material. In the workflow, it’s basically CAD, simulation if you want, API, print, the output is printable without supports.

10:05 From this, as I said, we see that sine is a meta material, it is a geome, geometric shape that behaves like a solid, like a bulk material, it has locally adjustable density, it has foreseeable characteristics, and you can tweak it very simply and very easily, and I’ll try to show you how this works. So we’re going to try to, to do a simple fixture, right, we have a design space here, in white, remove that, and we have an envelope inside of that space that we designed using engineering intuition, but it could also be designed with a topology optimization, for example, or let’s say pure design will.

11:03 Behind this we also have some interfaces that need to be met, 3D shapes that need to be present at the interfaces, we have some loads that we’re going to observe, and we have this envelope. Now what’s needed is, this has to be printed in aluminum, which has a yield of 255 MPA, more or less, we’re doing a bit bar park engineering here, if we add the safety factor we have 180 MPA, that’s the max stress that we can have in the structure, and we need to print it at 6 mm. We did a simulation on the, on the solid body of the envelope, and found that we have these stresses present in all these different kind of vibrations and loads in XYZ.

12:04 Now we can do very simple, very simple computation, so what’s light green, in 30 MPA, has to, can become 180 MPA, right, so how many times do we fit 30 into 180, it’s six times, so that means that we can have a density of 16% at that place, which gives us a density stop of 16, and for the other ones 11 and 3.3. Now what I can say already is, with 3.3% density there’s not going to be enough material to keep the thing in shape, so we’re going to put a little more.

12:51 Now I’m selecting this envelope, it’s a nice mesh, I go onto this, can you see the point, on this button here, creates a new project, we’re going to call it CD F demo, we get a new layer down here, and we can remove that one, so now we have the envelope layer that contains this object, and we’ve gotten a density point of 5.6% standard. Now I’m going to manually add more density points, this button here allows me to select the field type, we’ll go to density, start with the 16, let’s put in a few points here and there that improve them, real quick.

13:54 Then we add more density points, going to do the 11%, which is around here, I’m being a bit quick and dirty here, but I could always invest a little more time, of course, that, and then we’re going to add, now we have, we’ve gone from 16 to 11, one left, would be question to the audience, if you think about photography, let’s do eight, right, so we can have a few points of eight around here, down here. Okay, now because this is a demonstration, and we’ve, we a little bit limited in time, we used the same simulation for the front s and the back side, side, it’s a 2D simulation basically, but this could obviously be be 3D.

15:01 Now in order to have a, have a, an acceptable result, we’re just going to copy this couple of times. Okay, so now we are ready to go, hit this, the compute button here, we have here set density reference thickness, that’s the thickness upon which the system computes the density that we’ll get, I set this to the value that is the same as the printing thickness, and I’m going to do this, give it a comment here. Okay, I’m, I’m computing on a bit on our internal server, which is a bit faster than the one that’s on the cloud.

15:52 Now in the meantime, while we’re doing this, while we’re doing this, this 2D version of the surface, we are looking at the interfaces feld, so we’ve prepared these, and I added also these small ones up. We got the result already, let’s put that away for a second, so these 3D shapes here we want them sol, solid, right, so we select them, we go to this button here, add boundary, it already did it, we can check up here the parameters, we leave it at standard, look, actually we got to put this to zero, and this one to four, and we’re going to add these, well, they make sure that the, that the part remains printable, they’re kind of like, they’re bending, bendy, in this direction but not in the other one, and make sure that the part remains printable, as I said.

17:10 So select these, going to give them a thickness here of8 mm, standard rest is standard, and now we’re going to compute this again, and we ask for the 3D version. Now looking at the 2D result quickly, we see this 2D surface, meaning it has no thickness, it cannot be printed, and we mapped the, the density in color on it, so that it, it helps you to kind of navigate what you’re doing if you’re doing some iterations, we can also map surface bias, and mean curvature, and thickness, whatever can be mapped in, in, in color schemes, on those.

18:03 Now while we wait, another couple minutes, we’re at 3 of six, running out of stuff to say, sorry, that, yeah, well, five of six, it’s called, thank you, thanks for, maybe you want to say something, pardon me, both. So, so now we have the final result here, so we have this, this is a solid now, a solid mesh, error free largely or mainly, with all the boundaries smooth, smooth joint, directly printable, following the densities that we just specified before. This one here is a, is a 2D surface, can be used for sheet simulation, but, or, and the good thing is, it’s not actually just a toy, it’s, it’s a real part we just did, it’s, this, it’s a bracket that we did for, for the European Space Agency, it survived all the tests, this is the actual part that you saw on the shaker before.

19:36 And naturally the very brief and course engineering and dimensioning I just did is a first step, so in order to really then do this, we had to do a couple of iterations, we resimulated the result, we corrected the, the wall thickness, in order to kind of contain all the residual stress concentrations, but the part passed with flying figures, the simulation on the ion frequencies was matching the ones in the test, and the part wasn’t damaged. Now speaking of simulation, we used ANSES to simulate the first part, and you might know, it’s a bit time consuming, so if there’s anyone in the audience here who’s willing to take on the challenge to work with us to great, integrate simulation in order to accelerate the somewhat still necessary process of iterations, yes, I saw you guys, thank you.

So, o sorry, I have to change the program, so if I would, if you look at, let’s put away the interfaces here, so this would obviously be not ideal for printing, right, so there’s different strategies, one is that we have an algorithm that kind of can bend the surface, you can add in, you can close the whole thing, can al also be closed, you can use something that we call de featured, or density field envelope, that changes the outer behavior, there’s plenty of things you can do to ascertain printability, but you have these problems only on the outside, internally it’s always print, printable without supports, so what you get is complexity for free.

21:52 So this part took the original customer 3 hours of engineering, we had an intern that did it in 20 hours, so it can be really complex, it can be used to customize autonomously, and if you would like, please join us in one of our upcoming webinars, where we’ll be talking about simulations, API, and integrations, not only in terms of us talking to you, but hopefully also you talking to us. And to end, we imagined that this could be used as a, a locally printed moon habitat, where the actual surface will be filled with regolith in order to keep way radiation, printed from regolith. We are always very happy to hear and see what you guys are going to do with it, and I am glad that I was able to leave my artist life behind, going from author to tool maker. Thank you very much.