CDFAM NYC 2023 · New York · 14–15 June 2023
Hybrid DfAM: From Topology to Toolpath & Production to Performance
Abstract
This recording is part of the CDFAM Computational Design (+DfAM) Symposium and features Kirill Volchek, Chief Technology Officer at OQTON.
Presentation Title: Hybrid DfAM: From Topology to Toolpath & Production to Performance
Kirill Volchek, the Chief Technology Officer at OQTON, presents on ‘Hybrid DfAM: From Topology to Toolpath & Production to Performance’. Volchek explores the multifaceted landscape of Design for Additive Manufacturing (DfAM), encompassing topics from topological design to toolpath generation, and from production methodologies to performance evaluation. Drawing from his rich experience at OQTON, Volchek illustrates how these elements interconnect and influence each other, providing unique insights into the development and implementation of hybrid DfAM strategies.
This presentation is intended for professionals, researchers, and academics in the fields of computational design, additive manufacturing, and production engineering. Individuals interested in the synergies between topological design, toolpath generation, and production performance in the realm of additive manufacturing will find this session exceptionally insightful.
Transcript
From YouTube’s automatic captions, lightly cleaned; expect some errors. Each timestamp opens the video at that moment.
Read the full transcript · 7,936 words
Foreign. Good, my name is Kirill Volchek, I’m CTO of Oqton. I worked on the software company, and today I’m going to talk about 3D Expert, which is one of the products worked on, developed, and I’m really proud to show you this picture here. I’m holding tent and rocket that was printed seven years ago, after two months, when we decided to do our editing click to go into edit of manufacturing domain. The roots of 3D Expert, it’s actually 38 years, I would say out of those 38 years I was with a company for 25, as a chief software architect of Symmetron, and then after large 3D printing company acquired us, we started to think about additive. So on this picture, on this rocket, after two months we decided to go into editive: we have B rep, we have lattices, beams like lightest beams, variable sickness, fully designed within first prototype of 3D Expert, sliced within slicers that also be developed in those two months, and printed on metal printing printer. So it’s just happened, because we have, like, strong team of people, like, working, and, like, doing our KFCM solutions, and all the algorithms around this for years.
0:16 And today I’m going to talk about some general information about 3D Expert, and then jump into use cases, and show you a couple of very nice videos, and just showing the processes: vertical applications, and carbon capture applications, heat exchangers, gyroids like stuff that you really really like, right? You like jayroids, heat exchangers, complicated geometries that’s, that’s going to be here today.
2:00 So our design, it’s our just the beginning, but we are, we need to, like, bring this functional property to parts, and that’s why we have also the simulation, please, please place, an optimization in place. After that’s done, we need to to go to manufacturing assessments, and here on the screen you see our latest analytics tool, when they can, like, beam lattices our tool analyze all different parameters of those lattices, and this specific case we are looking into overhand candles, and then you can just see, like, in my light is, those red beams are under specific angle that probably is not going to be really well manufactured in my manufacturing technology. So it’s our manufacturing assessment.
2:43 When you’re done, you need to prepare it for printing, and 3D Expert, it’s all about, like, a process. So you can do also design, ulcer, simulation, assessment, printing, print preparation, and in the print preparation stage you’d like to design your supports. In this specific latches you have overhand nodes, and you’d like to support them, and you can support them with support which is also, like, this, right? In a bottom picture it’s implicit geometry, and implicit geometry, in some cases, when you put it into the bounding volume, you may get of floating or overhang dangling geometries that you don’t want freely to to print, and you don’t want to support them also. So we have, like, this digital debarring tool that takes those geometries, and, okay, like, you can just remove them out of your implicit field.
3:38 Our brain simulation you know, parts become more complex, you have more geometries, so simulating this process of printing within the metal printer is very important, because the print is, like, print processes costly, you don’t want to fail your print. So print simulation here, and last but not least, scanpas generation. So scanpas generations are very important for many many different reasons, it’s a really domain that being developed right now, this supportless printing ideas, and ability to optimize your prints, and we are going to talk about it today more.
4:14 So our motto, and of 3D Expert, is hybrid modeling. So hybrid modeling is nothing more than just take different paradigms of design and bring them together into one part, as different objects, without, are, without a comment, eliminating those geometries into something like triangles or implicits. Or so, we treat each geometry, each geometry type, and native manure. So if we create B rep— and which is very good for power design, original part, standard part design, because in each part you have, like, interfaces probably, or you are getting it from assembly, or you’d like to design them through brap tools so it’s very convenient way to do this. Our meshes for guiding design, topology optimized geometries, parametric beams for largest complexities, implicit for, like, we’ve seen today many many examples of this, and voxel algorithms, and voxel generative algorithms, read-based generation of algorithms, also very useful for some specific applications. And again, for each geometry modify, like, everything is done natively, are with these types of different geometrical paradigms and models.
5:31 Definitely, I need to have a slide about logistics, right? A 3D Expert, as I showed them on a tendon rocket before, like, after two months we had already lattices. So we spent those seven years making this license portfolio reach, and we have beam types of lattices, implicit types of lattices all those lattices could be created in a uniformal way, in partisan coordinate space, or radio, or spherical, or also it could be done within conformal spaces, right? It could be also surface lattices. Texturing analysis for largest is also very important like, on the bottom picture you can see porosity analysis, where for some medical applications, like, porosityopholetos, is very critical features, so you can get, using actual voxel algorithms coupled with beam lattices, you can get, are, those porosity histograms.
6:27 So, and definitely, heat exchangers, right? We need to talk about heat exchangers here, because those are places where we are replacing original designs this TPMS with gyroids.
6:39 So I’m going to talk about this case study, are, with this case study together with UDRI, University of Dayton, and the idea here was to to take existing, like, Cuban shell heat exchanger, like, oil fuel oil cooler, and a case study was about, like, how to keep outer, outer shape, like, exactly this, and just make this insert within this heat exchanger to be replaceable. So, or, like, it could be like maintenance could be easier, and then definitely one of the other side effects of this, like, to make this heat exchanger way more efficient, right? So, and to be efficient, it needs to have, like, extra scene walls, and the print, and aluminum, it’s quite quite challenging.
7:29 So the game, there’s this design so in this design we took, like, what we was before in this heat exchanger, he’ll place it with gyroid, but this gyroid is designed in conformals, by conformal space. So, for you see, like, inlets and outlets are on top of a part, so flow should go, start on top and then go down, make a U-turn and climb up, right? So we created this conformal space and designed gyroids within it.
7:54 So I’m going to show you not just the process, but also what we call dedicated application. You know, our, we see, like, many companies, and, like, we can, like, you can design all those types of geometries, but in our solution is not just the creative geometries, but the gift to engineers ways to automate the process and give, a, bring engineering logic into into devices.
8:27 So, so first we took the design and just took out the tubes, and then we had to design within this u-shape. We use this u-shape as a conformal shape to to create UVW parameterization by by cross sections, so you can see here the Divine defining cross sections, and then we have this conformal parameter parameterization. So what’s important here is you can see that in the bottom you may have, like, singularity point, but we created, like, those cross-sections in a way that, like, no singularity point in in a place in the center of this, like, native rotation, because in this point, if we create design without taking care of this, you’ll get extremely thin walls, because it’s actually singularity point, everything becomes slick, you know, very very small.
36:48 So our, then in this volume we created jayroid structure, so you can see that we are changing parameters here, everything is updated, because computation is done on GPU, it’s blessingly fast performance. So, and you can go to unlimited complexity, and everything is evaluated very fastly.
37:10 So let’s see what we got this is the geometry of our insert. Let’s look in the cross section of this geometry, you can see, like, on the bottom, very nice, like, flow pass that makes this U-turn. So let’s zoom in and on the screen, upgrade here and we are, we have logical representation of our heat exchanger, right? So it’s not just the geometry, but it’s, like, right, currently, right now I’m selecting outer shell, so explaining to the system here, this is my outer body. Next, selecting the inner body, in the middle, system understands, like, my hot and cold domain, and you can, like, visualize it, see, like, where is my hot and cold, I mean.
38:45 And then you, you pick your B rep, and now you start to define and select your inlets and outlets. So here you picked on the B rapper phase that actually represent an inlet, and then automatically system created the the domain for the fluid, so reddish object here is actually a fluid, a hot fluid, that’s going to be in this specific pipe, right? That was understood automatically by a system, so we know how liquid is going to enter the internal structure of of the heat exchanger. Or if you do the same on our outlet, again just selecting the geometry, and then system automatically recognize domains on a fuel side, also called inlet and outlet, and then you can see that on top and on the sides, on those inlets and outlets, geometry actually those domains are intersecting.
38:55 So we need to solve this problem, because the system knows where each domain is, at, what, what’s, what is inlet and outlet it capable automatically to create those stapling separation, it’s just implicit functionality, nothing complicated here, but, but user can do it in one click, because system knows what specifically is where. So you are not just performing design iteration, you’re just saying, okay, this is my heat exchanger, is that what I want to do I want to just separate, like, inlet and outlet from from a mixing of liquids, and it’s done.
Our next, what’s important, is to see what type of information you can get out of the geometry. So, on, you see on the screen here, volume, surface area of each domain, surface to volume ratio, it’s important parameters that you’d like to know, and you can also see a distribution of of a cross section, because you need to have, like, constant hydraulic diameter over cross sections, in order to have, like, correct weightless numbers, because viscosity is different for oil and fuel. So that’s why we created actually a separate sickness of our domains, which it didn’t say, and now system analyze water tightness here, and you can see that leak analysis said, like, no leaks. But this water tightness and let me stop here for a second this water tank, it’s possible to analyze, because system knows, like, whole design, like, it’s all those are my outer bodies here, all the liquid domains that I have. So if any of our liquid domain intersects with another liquid domain, which is actually a leak, or spills out, which is actually leak out. So everything is, its logical representation, everything could be analyzed in the system automatically.
And last but not least, it’s a scampus generator. So if you slice everything again natively so if you don’t have here any problem of taking this geometry and converting it, let’s say, to triangles, right, because this geometry is pretty complicated so if you slice it natively, but this is very thin heat exchanger, like, wall thickness here is designed, this could come up to just 100 microns, because, you know, for gyroids, and some, click, in some areas, 2D sickness on on a cross section could be really small, but we don’t want really in aluminum to go to this wall thickness, it’s actually mostly impossible to print. So in order to print symbols, you need to have very dedicated slicing functionality. So you can see here the scan pass doesn’t look like normal kind of, right, because you can see all those islands of HS, and then, like, blocked passes, and then any, like, there is no place on this scanning where you have, like, one baskets for one pass of a laser, is not good for water, water tightness, or gas tightness. So we are always ensuring, like, water title and gust type print. It’s a lot of know-how here, we can talk about it later, I don’t want to spend all the time on the slicing, because, yeah.
42:06 So, yeah, we have immune carbon emissions from industrial sources, we don’t like it, you didn’t, we want to capture it, and we want to store it or use it, right? So, in order to do this, we need those carbon capture parts, and our intent actually to to not to put it on the ground, which is one of the possibilities, but really to use it in consumer products, in carbonation, and then price of this carbon needs to hit, like, 30 dollars per ton price target. So the parts that we are designing here needs to be very efficient, and very economically, like, we need to, like, cost per part needs to be low.
So first part that I want to show you here is a printed sealer. So this is actually a device that Gus goes from a bottom, and it cooling jacket around this part cools it down, and, like, captured carbon leaks down, like, guys, that sir becomes a liquid, and, like, leaks down through the whole system, and being absorbed, like, captured on the bottom. We’d like to make efficient design here, if you’d like to make it as small as possible, as efficient as possible, a long Bart life, very important, and those parts are stacked one on top of another, and you can put them into a pipe, like, in a big large pipe, right, so they connect into each other.
43:38 So in order to achieve it, we create a design that, and, and cooling efficiency should be very very high, and gradients, thermal gradients in a layer and between layers, needs to be very very low, because, if it’s, like, we need to reduce terminal, a German load here, right, excuse me. So in order to achieve it, we have here a b-wrap designed fins, those fins are hollow, and visit those fins may have conformal lighters to to optimize the durable properties. So here we have those dual, dual chilling circuits, like, our calling jack jacket, and then those lightest systems on films, fins, and then intensification structure which needs to be like, kind of, uniform mode, but with some randomization properties, in order to make efficient precipitation of the CO2. So that’s first use case, and this is part being designed inside 3D Expert, right.
44:56 So, second use case is even more interesting part, it’s also a carbon capture type of a part, it’s called helical conductor. It’s more complicated part, with more dense geometry here, with more functional functional components, and you can, you can purchase such parts of the shell, but they’re extremely inefficient, right? So we’d like to make a very efficient part for it. Part is ginormous, it’s half meter and half meter and half meter part, so amount of geometry and fidelity of geometry is very high, and one of the very important parameters here was to reduce the cost of a part, and cost of this part is a printing time, because this huge part, if sliced and printed, like, the standard parameters on some printers, could be printed in just one part in three weeks like, you can imagine how this is, how what’s going to be cost of this part. So we had to reduce it x3, so we’d like, we had to print it in a week, and majority of a print time is actually a print of a lattice, like, internal structures, so we have special treatment for this, I’m going to show you that.
46:07 So again, we need to minimize, like, to make it single part, and make it more as efficient as possible, monolithic design. It has three different subsystems, is one of the gas conducting, second is solvent shower head system, so there is a solvent that goes, like, steam of solvent goes into the part, and then it washes out carbon which is captured by lattices, and cool a jacket around. So that’s again, let’s, let’s see how it works.
46:41 So inside 3D Expert we have this B rap, and this B rap geometry defines, like, it climbs up, and on top we have outlets for for gas, and also for coolant. Let’s see how cooling jacket is created, so it’s a self-supported design, it goes on the side of our, of our contactor, so that’s a cooling jacket. And for for solvent, we also have here solventeen, and sold without, like, inlets and outlets there in the bottom, because solvent, like, steam of solvent goes from a bottom, from a middle of a part, distributed through those three shower heads, and then they are going down and washes it out, and solvent, like, enriched this carbon goes out from the outlet here.
47:29 So, and this is shower heads, right, so this far, this is very complicated beer up geometry, full with holes and channels, so you can, like, let’s zoom in and see how it looks like from close up complicated geometry design, and this B rep standard approaches, right. So this is a shower hand, and let’s see, like, what we did with the lightest.
47:55 So here we see lights is inside, so this light just was designed within conformal space, so it’s a spiral lattice. So we use this body to create this conformal UW spacing by pointing on six sides of this, like, it’s kind of topological cube here, right, like we have six pairs of faces, so it gave us the UW parameterize space. And yeah, so you see this parameter space, and we use it to to put here this type of the lightest cell. So this lightest cell is actually pretty simple, so that 2x, two x’s, like the shift, so gas goes in, heats those, like, structures, and then being absorbed, because it’s also coated lattices, not just simple lattice, and they need to be one millimeter thick according to design requirements, are.
But, iffy, like I said, like, it was three weeks of printing, because initially, if you designed that this, like, volumetrically, and then if you do it through implicits, or you do it through, like, beams, volumetric beams, you slice it as a volumetric object, right, and then when you slice it as a volumetric object, you have this long long slice, because you have, like, contours, you have hatches, and some layers those hashes are very short, so it, it, your play this print time. So instead of designing it this way, we just created zero thickness beams here, which are actually membranes, because the cross profile of this beam is an open line, right, so we created, like, zero sickness structure, and we put the zero sickness structure within this conformal volume.
49:38 So now you see that each beam, or each membrane, has its own direction, and it, because of on HP, and we have, like, two rotation angles on both sides, so you have, like, almost, you have, or kind of, blend surface, so each beam is not a plane, it’s kind of blend surface here, and those beams are nicely following this conformal, not beams of cells, so following conformal space that we designed. So you see it’s bigger, larger, smaller, when you close at the center, there’s a little bit of overlap, which is also very important, because we want it to be, like, stiff, and, or, because of the distance on top, on those, on those, are, like, you can see here, distance pretty large, in some places could come up, come up to, up to 15 millimeters between those peaks, our upper surface will be non-supported if you’ll print it this way.
50:33 So in order to take care of this, we had to design on top of this structure topologically connected another layer of latches, which is self-supported. You see, it’s absolute topological connected, which is not so trivial to do it in UVW coordinate spaces, because you need to adjust, like, everything, and now this is really nicely printed part.
50:57 And not, let’s talk a little bit about slicing. So in order to print it with the sigma so zero, it’s zero sigma spins, right, but in order to print it with the sickness, we assigned to this zero sigma spins thickness value, and this thickness value is taken in a slicer. So slicer knows, when it sees the the cross section of this, it knows the instead of one cross section, it builds, it creates several hitches, but, like, we applied plane here a little bit with the quality of the material, but quality of material here is not so important, and aluminum melt pool is about 300 microns, so it’s actually it’s pretty pretty close to one millimeter. But printing time here is extremely extremely efficient, because, you know, you just do three hatches, and you have, like, your full membrane actually printed.
51:51 It’s not so, also, trivial, because when those are you see here in the middle, when those membranes meet and start, like, going up, you have a place where you have a lot of geometry, very dense, on almost on the same line, like, just one layer above this joint. If all those H’s, like, from each theme, three, each film, three, all of them going to be in one place, so then it’s going to be overheat and failure of the print, absolutely. So, instead of, in order to take care of this, we have also specific treatments, so we are taking care of this, so never, never, no place where we have, like, all those, or. So it’s a lot of smart innocent slices, that’s what I want to say, and this smart innocent slicer, only we can do it, only because we have, like, all the geometry natively presented and understood by our tools. Just trying to take it to different system and explain intent will ruin the whole functionality of a system. So that’s why ability to design, simulate, our support, assess manufacturability and slice, is so important, because there is no data conversion, and the whole intent has been explained to the system, so system understand what trying to do, what you’re trying to achieve.
53:15 So it was almost my, now, my last slide. So 3D Expert is a hybrid designed for the manufacturing tool that allows you to tackle highest complexity with finance details and total manufacturability, and that’s actually that, yeah, this is last slide. So we are here, would be happy to answer your questions, we have few parts on the table, so we can look into them. So thank you very much.
Foreign. It was a long day full of amazing presentations and information, right, but it’s only me, keynote, and drinks, and, like, we are done. Okay.
So my name is Kirill Volchek, I’m CTO of Oqton, worked on the software company. Today I’m going to talk about 3D Expert, which is one of the products worked on, developed, and I’m really proud to show you this picture here. I’m holding tent and rocket that was printed seven years ago, after two months, when we decided to do our edited like, to go into edit of manufacturing domain. Roots of 3D Experts, it’s actually 38 years, I would say out of those 38 years I was with a company for 25, as a chief software architect of Symmetron, and then after large 3D printing company acquired us, we started to think about additive. So on this picture, on this rocket, after two months we decided to go into editive: we have B rep, we have lattices, beams like lightest beams, variable sigmas, fully designed within first prototype of 3D Expert, sliced within slicers that also be developed in those two months, and printed on metal printing printer. So it’s just happened, because we have, like, strong team of people, like, working, and, like, doing our KFCM solutions, and all the algorithms around this for years.
And today I’m going to talk about some general information about 3D Expert, and then jump into use cases, and show you a couple of very nice videos, and just showing the processes: vertical applications, and carbon, carbon capture applications, heat exchangers, gyroids like stuff that you really really like, right? You like jayroids, heat exchangers, complicated geometries that’s, that’s going to be here today.
So our design, it’s our just the beginning, but we are, we need to, like, bring this functional property to parts, and that’s why we have all the simulation, please, please place, an optimization in place. After that’s done, we need to to go to manufacturing assessments, and here on the screen you see our latest analytics tool, when they can, like, beam lattices our tool analyze all different parameters of those lattices, and this specific case we are looking into overhand candles, and then you can just see, like, in my light is, those red beams are under specific angle that probably is not going to be really well manufactured in my manufacturing technology. So it’s our manufacturing assessment.
When you’re done, you need to prepare it for printing, and 3D Expert, it’s all about, like, a process. So you can do also design, ulcer, simulation, assessment, printing, print preparation, and in the print preparation stage you’d like to design your supports. In this specific latches you have overhand nodes, and you’d like to support them, and you can support them with support which is also latches, right? In the bottom picture it’s implicit geometry, and implicit geometry, in some cases, when you put it into the bounding volume, you may get, or, floating or overhang dangling geometries that you don’t want freely to to print, and you don’t want to support them also. So we have, like, this digital debarring tool that takes those geometries, and, like, you can just remove them out of your implicit field.
Our print simulation you know, parts become more complex, you have more geometries, so simulating this process of printing within the metal printer is very important, because the print is, like, print processes costly, you don’t want to fail your print. So print simulation here, and last but not least, scanpas generation. So scanpas generations are very important for many many different reasons, it’s a really domain that being developed right now, this supportless printing ideas, and ability to optimize your prints, and we are going to talk about it today more.
So our motto, and of 3D Expert, is hybrid modeling. So hybrid modeling is nothing more than just take different paradigms of design and bring them together into one part, as different objects, without, are, without a common denominating those geometries into something like triangles or implicits. Or so, we treat each geometry, each geometry type, and native manure. So if we create B rep— and which is very good for power design, original part, standard part design, because in each part you have, like, interfaces probably, or you are getting it from assembly, or you’d like to design them through brap tools so it’s very convenient way to do this. Or meshes for guiding design, topology optimized geometries, parametric beams for largest complexities, implicit for, like, we’ve seen today many many examples of this, and voxel algorithms, and voxel generative algorithms, read-based generation of algorithms, also very useful for some specific applications. And again, for each geometry modify, like, everything is done natively, are with these types of different geometrical paradigms and models.
Definitely, I need to have a slide about lattices, right? A 3D Expert, as I showed in on a tendon rocket before, like, after two months we had already lattices. So we spends on seven years making this latest portfolio reach, and we have beam types of lattices, implicit types of lattices all those lattices could be created in a uniformal way, in partisan coordinate space, or radio, or spherical, or also it could be done within conformal spaces, right? It could be also surface lattices. Texturing analysis for lightest is also very important like, on the bottom picture you can see porosity analysis, where for some medical applications, like, porosityophyllitus, is very critical features, so you can get, using actual voxel algorithms coupled with beam lattices, you can get, are, those porosity histograms.
So, and definitely, heat exchangers, right? We need to talk about heat exchangers here, because those are places where we are replacing original designs this TPMS with gyroids.
So I’m going to talk about this case study, are, with this case study together, versus UDRI, University of Dayton, and the idea here was to to take existing, like, Cuban shell heat exchanger, like, oil fuel oil cooler, and a case study was about, like, how to keep outer, outer shape, like, exactly this, and just make this insert within this heat exchanger to be replaceable. So, or, like, it could be like maintenance could be easier, and then definitely one of the other side effects of this, like, to make this heat exchanger way more efficient, right? So, and to be efficient, it needs to have, like, extra scene walls, and the print, and aluminum, it’s quite quite challenging.
So the game, there’s this design so in this design we took, like, what it was before, in this heat exchanger, Hill Place this gyroid, but this gyroid is designed in conformance, by conformal space. So, for you see, like, inlets and outlets are on top of a part, so flow should go, start on top and then go down, make a huge turn and climb up, right? So we created this conformal space and designed gyroids within it.
So I’m going to show you not just the process, but also what we call dedicated application. You know, are, we see, like, many companies, and, like, we can, like, you can design all those types of geometries, but in our solution is not just the creator of geometries, but the gift to engineers ways to automate the process and give, a, bring engineering logic into into devices.
So, so first we took the design and just took out the tubes, and then we had to design within this u-shape. We use this u-shape as a conformal shape to to create UVW parameterization by by cross sections, so you can see here the Divine defining cross sections, and then we have this conformal parameter parameterization. So what’s important here is you can see that in the bottom you may have, like, singularity point, but we created, like, those cross-sections in a way that, like, no singularity point in in a place in the center of this, or this, like, native rotation, because in this point, if you create design without taking care of this, you’ll get extremely thin walls, because it’s actually a singularity point, everything becomes slick, you know, very very small.
So our, then in this volume we created a gyroid structure, so you can see that we are changing parameters here, everything is updated, because computation is done on GPU, it’s blessing with fast performance. So, and you can go to unlimited complexity, and everything is related very fastly.
So let’s see what we got this is the geometry of our insert. Let’s look at the cross section of this geometry, you can see, like, on the bottom, very nice, like, flow pass that makes this U-turn. So let’s zoom in, and it’s on the screen up right here, and we are, we have logical representation of our heat exchanger, right? So it’s not just the geometry, but it’s, like, right, currently, right now I’m selecting outer shell, so explain it to the system here, this is my outer body. Our next, selecting the inner body, and immediately system understands, like, my hot and cold domain, and you can, like, visualize it, see, like, where is my hot and cold, I mean.
And then you, you pick your B rep, and now you start to define and select your inlets and outlets. So here you picked on the B rapper phase that actually represent an inlet, and then automatically system created the the domain for the fluid, so reddish object here is actually a fluid, a hot fluid, that’s going to be in this specific pipe, right? That was understood automatically by a system, so we know how liquid is going to enter the internal structure of of the heat exchanger. Or if you do the same on our outlet, again just selecting the geometry, and then system automatically recognize domains on a fuel side, also called inlet and outlet, and then you can see that on top and on the sides, on those inlet cell and outlets, geometry actually those domains are intersecting.
So we need to solve this problem, because the system knows where each domain is, at, what, what’s, what is inlet and outlet it capable automatically to create those stapling separation, it’s just implicit functionality, nothing complicated here, but, but user can do it in one click, because system knows what specifically is where. So you are not just performing design operation, you’re just saying, okay, this is my heat exchanger, is that what I want to do I want to just separate, like, inlet and outlet from from mixing of liquids, and it’s done.
Next, what’s important, is to see what type of information you can get out of the geometry. So, on, you see on the screen here, volume, surface area of each domain, surface to volume ratio, it’s important parameters that you’d like to know, and you can also see our distribution of of a cross section, because you need to have, like, constant hydraulic diameter over cross sections, in order to have, like, correct rayless numbers, because viscosity is different for oil and fuel. So that’s why we created actually separate sickness of our domains, which it didn’t say, and now system analyze water tightness here, and you can see that leak analysis said, like, no leaks. But this water tightness and let me stop here for a second this water tank, that’s possible to analyze, because system knows, like, whole design, like, it’s all those, those are my outer bodies here, all the liquid domains that I have. So if any of our liquid domain intersects with another liquid domain, which is actually a leak, or spills out, which is actually leaked out. So everything, because it’s logical representation, everything could be analyzed in the system automatically.
And last but not least, it’s a scampus generation. So if you slice everything again natively so if you don’t have here any problem of taking this geometry and converting it, let’s say, to triangles, right, because this geometry is pretty complicated so if you slice it natively, but this is very thin heat exchanger, like, wall thickness here is designed, this could come up to just 100 microns, because, you know, for gyroids, and some, in some areas, 2D sickness on on a cross section could be really small, but we don’t want really in aluminum to go to this wall thickness, it’s actually mostly impossible to print. So in order to print symbols, you need to have very dedicated slicing functionality. So you can see here the scan pass doesn’t look like normal kind of, right, because you can see all those islands of hatches, and then, like, blocked passes, and then, and, like, there is no place on this scanning where you have, like, one baskets for one pass of a laser, is not good for water, water tightness, or gas tightness. So we are always ensuring, like, water title and gust type print. It’s a lot of know-how here, we can talk about it later, I don’t want to spend all the time on the slicing, because, yeah.
So, yeah, we have immune carbon emissions from industrial sources, we don’t like it, even if you want to capture it, and we want to store it or use it, right? So, in order to do this, we need those carbon capture parts, and our intent actually to to not to put it on the ground, which is one of the possibilities, but really to use it in consumer products, in carbonation, and then price of this carbon needs to hit, like, 30 dollars per ton price target. So the parts that we are designing here needs to be very efficient, and very economically, like, we need to, like, cost per part needs to be low.
So first part that I want to show you here is a printed sealer. So this is actually a device that Gus goes from a bottom, and it cooling jacket around this part cools it down, and, like, captured carbon leaks down, like, guys, that sir becomes a liquid, and, like, leaks down through the whole system, and being absorbed, like, captured on the bottom. We’d like to make efficient design here, if you’d like to make it as small as possible, as efficient as possible, a long Bart life, very important, and those parts are stacked one on top of another, and you can put them into a pipe, like, in a big large pipe, right, so they connect into each other.
So in order to achieve it, we create a design that, and, and cooling efficiency should be very very high, and gradients, thermal gradients in a layer and between layers, needs to be a very very low, because, if it’s, like, we need to reduce terminal, a German load here, right, excuse me. So in order to achieve it, we have here a b-wrap designed films, those fins are hollow, and within those fins we have conformal like this to to optimize the durable properties. So here we have those dual, dual Chilean circuits, like, calling jack jacket, and then those lightest systems on fields, fins, and then intensification structure which needs to be like, kind of, uniformal, but but with some randomization properties, in order to make efficient precipitation of the CO2. So that’s first use case, and this is part being designed inside 3D Expert, right.
So, second use case is even more interesting part, it’s also a carbon capture type of a part, it’s called helical conductor. It’s more complicated part, with more dense geometry here, with more functional functional components, and you can, you can purchase such parts of the shell, but they’re extremely inefficient, right? So we’d like to make a very efficient part for it. Part is ginormous, it’s half meter of half meter and half meter apart, so amount of geometry and fidelity of geometry is very high, and one of the very important parameters here was to reduce the cost of a part, and cost of this part is a printing time, because this huge part, if sliced and printed, like, the standard parameters on some printers, could be printed in just one part in three weeks like, you can imagine how this is, how what’s going to be cost of this part. So we had to reduce it x3, so we, like, they had to print it in a week, and majority of a print time is actually print of a lattice, like, internal structures, so we have special treatment for this, and I’m going to show you that.
So again, we need to minimize, as, like, to make it single part, and making some more as efficient as possible, monolithic design. It has three different subsystems, is one of the gas contacting, second is solvent shower head system, so there is a solvent that goes, like, steam of solvent goes into the part, and then it washes out carbon which is captured by lattices, and cool a jacket around. So that’s again, let’s, let’s see how it works.
So inside 3D Expert we have this B rap, and this B reap geometry defines, like, it climbs up, and on top we have outlets for for gas, and also for coolant. Let’s see how cool injection is created, so it’s a self-supported design, it goes on the side of our, of our conductor, so that’s a cooling jacket. And for for solvent, we also have here solventeen, and sold it out, like, inlets and outlets there in the bottom, because solvent, like, steam of solvent goes from a bottom, from a middle of a part, distributed through those three shower heads, and then they’re going down and washes it out, and solvent, like, enriched this carbon goes out from the outlet here.
So, and this is shower heads, right, so this for, this is very complicated beer up geometry, full with holes and channels, so you can, like, let’s zoom in and see how it looks like from close up complicated geometry design, and this B rep standard approaches, right. So this is a shower hand, and let’s see, like, what we did with the lightest.
So here we see light is inside, so this latches was designed within conformal space, so it’s a spiral lattice. So if you use this body to create this conformal UW spacing by pointing on six sides of this, like, it’s kind of topological cube here, right, like we have six pairs of faces, so it gave us the UVW perimeterize space. And yeah, so you see this parameter, a space, and we use it to to put here this type of the lightest cell. So this lightest cell is actually pretty simple, so that took x, two x’s, like the shift, so gas goes in, heats those, like, structures, and then being absorbed, because it’s also coated latches, it’s not just simple latches, and they need to be one millimeter sick according to design requirements, are.
But if we, like I said, like, it was three weeks of printing, because initially we designed it with, like, volumetrically, and then if you do it through implicits, or you do it through, like, beams, volumetric beams, you slice it as a volumetric object, right, and then when you slice it as a volumetric object, you have this long long slice, because you have, like, contours, you have hatches, and some layers those hatches are very short, so it, it, you play this print time. So instead of designing it this way, we just created zero thickness beams here, which are actually membranes, because the cross profile of this beam is an open line, right, so we created, like, zero sequence structure, and we put the zero sequence structure within this conformal volume.
So now you see that each beam, or each membrane, has its own direction, and it, because of on H beam, we have, like, two rotation angles on both sides, so you have, like, almost, you have, or kind of, blend surface, so each beam is not a plane, it’s kind of blend surface here, and those beams are nicely following this conformal, not beams are cells, a following conformal space that we designed. So you see it’s bigger, larger, smaller, when you close at the center, this is a little bit of overlap, which is also very important, because we want it to be, like, stiff, and, or, because of the distance on top, on those, on those, are, like, you can see here, distance pretty large, in some places could come up, come up to, up to 15 millimeters between those peaks, our upper surface will be non-supported if you’ll print it this way.
So in order to take care of this, we had to design on top of this structure topologically connected another layer of latches, which is self-supported. You see, it’s absolute topological connected, which is not so trivial to do it in UVW coordinate spaces, because you need to adjust, like, everything, and now this is really nicely printed part.
And not, let’s talk a little bit about slicing. So in order to print it with the sigma so zero, it’s a zero sigma spins, right, but in order to print it with the sickness, we assigned to this zero sigma spins a sickness value, and this thickness value is taken in a slicer. So slicer knows, when it sees the the cross section of this, it knows the instead of one cross section, it builds, it creates several hitches, but, like, we applied the plane here a little bit with the quality of the material, but quality of material here is not so important, and aluminum melt pool is above 300 microns, so it’s actually it’s pretty pretty close to one millimeter. But printing time here is extremely extremely efficient, because, you know, you just do three hatches, and you have, like, your full membrane actually printed.
It’s not so, also, trivial, because when those are you see here in the middle, when those membranes meet and start, like, going up, you have a place where you have a lot of geometry, very dense, on almost on the same line, like, just one layer above this joint. If all those H’s, like, from each theme, three, each film, three, all of them going to be one place, so then it’s going to be overheat and failure of the print, absolutely. So, instead of, in order to take care of this, we have also specific treatments, so we are taking care of this, so never, never, know place where we have, like, all those. So it’s a lot of smart and listen slices, that’s what I want to say, and this smart innocent slicer, only we can do it, only because we have, like, all the geometry natively presented and understood by our tools. Just trying to take it to different system and explain intent will ruin the whole functionality of a system. So that’s why ability to design, simulate, our support, assess manufacturability and slice, is so important, because there is no data conversion, and the whole intent has been explained to the system, so system understand what trying to do, what you’re trying to achieve.
53:15 So it was almost my, now, my last slide. So 3D Expert is a hybrid designed for the manufacturing tool that allows you to tackle highest complexity, these finance details, and total manufacturability, and that’s actually that, yeah, this is last slide. So we are here, would be happy to answer questions, we have a few parts on the table, so we can look into them. So thank you very much. Thank you.
Listen, search and cite
Listen to this talk as a podcast
More from CDFAM NYC 2023

Generative Design & Digital Manufacturing at NASA Goddard
Ryan McClelland · NASA Goddard Space Flight Center

Generate-to-Validate: Optimal Designs and the Need to Deliver Under Uncertainty
André Wilmes · Rafinex

Defining Performance not Just Geometry with Software Process + Material Science
Hardik Kabaria · Carbon

Reinventing the Wheel – Representing & Manufacturing Geometry with a Program
Mark Burhop · Siemens Technology

Interspecies Forms: Robotic Feedback Systems Mycelia Growth & Computational Form
Natalie Alima · Adidas / Biolab Studio

Designing Material Properties with Advanced Toolpath Generation
Alex Roschli · Oak Ridge National Laboratory

Design Automation Using Machine Learning & Computational Design
Rhushik Matroja · Cognitive Design Systems






















