If you want to skip ahead: Design space generation: 1:42 Boundary conditions: 7:22 Non-design areas: 9:50 Meshing: 11:10 Optimization results: 14:20 Creating design from optimization: 15:30 Printing the model: 22:20
Holes can be done in an other way that seems more logicial to me : - add points at the right locations, exit sketch - hit "H", select point(s), "Simple" type, "Simple" tap type, distance "all", enter diameter. Done. You could also fiddle with "clearance" type and choose the screw type that fits in. :)
The main difference between your channel and other 3D-oriented channels is that you tackle many technical issues, such as this, the one about the infill strength, etc. Thanks for that!
What a great video. As a very novice Fusion 360 user I never leave the modelling environment - I guess it's time to! Also, the visible highlighting of the shortcut keys makes it much easier to follow along. I wish every youtuber doing Fusion 360 videos would add that
this Video is absolutely great, I love it. I'm a mechanical engineer using Catia V5 (sparsely using the integrated FEM Analysis), and it's incredibly interesting to see how FEM Analysis and Shape Optimization is integrated into Fusion!
Thank you really much for making such extended tutorials, the topics you choose are really, *really* helpful and the amount of detail you put into them make it super easy to pick up; even if topics are as abstract as this one.
Seriously one of the best videos 3-D printing related topic videos I have ever seen, also one of the best advanced fusion 360 videos I have ever seen. Thanks!
Yet another AWESOME tutorial from you. I'm new to the 3D modeling and printing world and your videos are very rich in information. Learned so much from you. Keep it up.
i work on a students race car and we used Topology Optimization for nearly every part and these parts look so beautiful. purly functional, nothing there whats not needed
Hey man, i have a suggestion that in such build you should set infill at 100% to achieve actual results. Reason for this is that Topology Optimization do not calculate infill. It calculate topology for full-body solid object, and infill will broke equation. What i want to say that with 15% infill there will be totally different optimized shape that without infill 100% and program do not calculate it.
wow, I didn't even know that! A few weeks ago I was designing a filament holder myself in Fusion 360 and this would've been really handy... Nice video!
Is there a reason why the simulation would put a support asymmetrically when the constraints and forces are all applied along the centerline of the part? It works out nicely for printing but seemed strange.
Tim Mortensen he didn't go into it, but there are settings that tell the software how the part will be made - 3d printing, cnc, extrusion, etc. Depending on these selections, the software will change placements of buttress features like this. I'm not sure if fusion has it yet, but other software packages also allow you to specify whether or not you want the part to be symmetric about a given plane as well. Hope this helps.
I haven't been able to find anything in Fusion 360 to set that it's a 3d printed object for that sort of thing. Do you know where it may be if there is such a setting?
The software generally deploys a generative algorithm, so the initial source of the structure is random. And the algorithm tries to make it better with each iteration.
FEA simulation is going to be off more because of the weak z axis and layer bonds rather than the infill. Infill does not really weaken parts that much, not as much as they are weakened by poor layer bonding. Really nice video, thank you!
I'm not sure if Fusuon allows for changing element size in areas but you only need the smaller ones near the stress concentration areas. Could get by with way less elements, especially if Fusion has plate elements
Damn, looks like I won't get to try this as simulations will be scrapped for Fusion 360's hobby version. :'( You'll have to do this tutorial under Solidworks!
I looooove these segments! It's such a good basis for further experimentation. Thanks! Nice to see you're using FormFutura recycled petg, I don't see many creators do that (recycling that is). I'm testing it right now but it does seem a bit brittle at any temperature, did you experience that?
Why did the FEA show the cross support to be on only one side, the 'bottom' of the model (in fusion)? You made that support only half width as well when you made your design, but in the axis the forces and supports were fully symmetric.
Thank you for the tutorial. TO is the essential tool for AM. We are looking forward to see the next one. A couple of comments: - Following your design strategy: in order to keep the original dimmensions of the design you could Project the initial sketch to draw the arc of the holder. - Alternative way: Why you haven't convert the STL to solid and then correct/soften the desire areas?
My personal opinion is that "softening" optimized designs is a really bad practice in a lot of cases and especially in the, the redesign is pretty easy.
You may be able to shell and then add a "cap" - top layers, maybe join body, hopefully 360 leaves the inner cavity open. Then specify your material and try an FEA?
As 3D parts’ strength comes mainly from the perimeters, would you be able to approximate the analysis a little more if you made a hollow version of the part? Excuse my grammar 😅 Ps: Your content is awesome
The hard edge on the upper mounting bracket is a weak point and should also be modeled organically. Because what's the point in maximizing the rigidity of the body if the attachment is weak..? It's like having a steel part screwed to a cardboard part ._.
You usually can't model those parts organically, because the math used to simulate the structure doesn't produce reasonable results around where the loads and the attachment points are. Also, if you ask an optimization program to minimize the stress in your part and give it the ability to modify material around where the stress is applied, the "best" solution is to just not support the load at all, so it will remove 100% of the material from that area.
Regardless of whether the static stress simulation is accurate for the printed material, I still think it would be interesting to run both the original and the optimized shapes and see how much weaker the optimized part is.
Depending on what stress the part has to withstand, the optimized part can be stronger. E.g. a H or T steel bar bends less than a massive piece of steel.
@@Mobin92 I found your comment interesting and decided to check it. Draw 5 horizontal bars 100mm*100mm*1000mm in FreeCAD. Profiles are: "standing H", "laying H", "solid square", "T", "upsidedown T". Wall thickness is 10mm. I've fixed in one end of each bar and let it bend on its own mass. Results are quite interesting. "Laying H" is the best. "Standing H" is the worst. "Solid" is the second worst. So does that mean that solid bar is not strong? No. It is just heavy and uses own mass in not optimal way, but when you give it an external load, it will be the strongest.
@9:11 you show the load coming in from a single direction. I would think the load would also be coming in from the sides and front-to-back as a smaller force. In this case, with no side-to-side load, your project will be very floppy, as in a piece of paper. It will be strong in 2 dimensions, but not in 3 dimensions. Possibly I am missing something..
You've got good videos and descriptions, well done. The one thing that drives me nuts is how slowly your intros wrap-up, the "I'm Stefen and welcome to CNC Kitchen!" part seems to take F-O-R-E-V-E-R. It was amusing the first video, but made me want to jump out the window (or just click away) by the third time I heard it. 2-3 seconds max, not 5. I know it seems trivial, but when you have quality editing and planning (which clearly you do) such rough edges stand out.
And there we have it... at least 1 semester of college in 20 minutes... seriously now, those places are only neccesary to get that piece of caedboard with your picture on it
Is there a way to force Fusion 360 to do the analysis locally? I'm not a fan of the 'cloud credits' thing Autodesk has going on and I'm perfectly fine with waiting longer if it's all local to my machine.
I'm totally with you. For normal static analyses it's possible to perform the solution locally, topology optimization unfortunately doesn't support that.
It looks like if you have a Startup/Education license you get unlimited credits at least but I bet they give you lower priority on their simulation servers.
Ese es el perfil optimizado para un punto en la horizontal que recibe la fuerza. La fuerza la recibe la barra, y la reparte entre dos elementos. Tal vez la forma sea un tetraedro.
great video! thank you so much. I have one question/observation: although the load and mounting holes are symetrically placed on the part (mid point), the optimization results in an asymetrcal design with the middle truss member being offset to the right. can anyone explain why this is?
The last time i worked with finite elements was helping somebody optimising some old code for calculating chemical penetration depth. 3 mm grain size - that is ginormous XD I think we were working on something like 20 nm. of course that was not running on a normal home computer.
Servus Stefan - Merci, great video as always! A learning question from the engineer who installed Fusion360, but always uses the software he has been using for 20 years when something needs to be constructed. It's a shame, I know. But it is what it is. However, when I see something like this, it motivates me to make the switch, however: At what point in this FEM process do the material properties such as weights or stress tensors actually come into play? Are they hidden somewhere in the settings or is it all based on the freely chosen percentage of material savings, which would not be so tingling after all. I assume that the answer here will be more concise than the question, or could give rise to a separate video, because you have mentioned the imponderabilities of FDM printing. I pull the crippled strut inside in this form (quite constructively) a bit in doubt, but it is also very funny - so why not follow the software. :-) Translated with www.DeepL.com/Translator
So if you do these simple topology optimizations, with only one part, then the material is negligible because you maximize stiffness (here the norm of the strain energy is used) and are only looking for a mass percentage. You could actually replace mass by volume. Actually even the magnitude of load doesn't play a role if you only have one load case. You could set materials but if you work with isotropic materials (with no crazy Poisson's ratio) the result wouldn't change a bit. If you go more advanced (assemblies, orthotropic materials, numerous load cases) then these things come into play, but then the problem gets really complicated. There is a reason why I said, that this takes years or even decades to master.
I think it doesn't matter. The simulation assumes homogeneous tensile strength throughout the part and goes from there. This is also why these simulations can only get a reasonable approximation for 3D-printed parts, as they are shells with a small percentage of infill (not at all homogeneous).
Thanks for the great Tutorial. @ 15:25 you say that you wouldn't recommend to print it as it is, but you don't say the reason why. Could you provide a reason to why one shouldn't print the optimized topology as it is. Thank you.
1. The result is rough, so the print wouldn't look nice, but smoothing could be done. 2. The solution doesn't consider any manufacturing constraints. Even though you can theoretically print everything, it is not advisable to print everything. 3. Most importantly: The solution can show un-plausible results, especially at the locations where you added constraints. These result from simplifications you've done with the model (e.g. not simulating pretension or contacts...). So it is the job of the engineer to spot them and design them properly.
The mesh from the optimization is also very coarse - depending on the mesh you specify in the FE settings. Normally, 3D printing uses a much finer mesh; unless you're ok with a rough looking part, in which case the mesh from optimization is perfectly useable.
At which point the simulation takes into account the material? Maybe I missed it in the video. But I guess the reaction to 100 N it's not the same on a 10mm steel beam or a 10mm crystal bar.
Can someone please explain to me why that truss was offset to one side, instead of spanning the entire thickness of the structure? Shouldn't the solution be symmetric in that axis since the load was given to be straight downward?
why is there no straightforward way to convert from optimized mesh to sketch?? Ive looked at solidworks optimization workflows too, and it requires manual sketching.
That was a pretty awesome tutorial, I sure would like to see a lot more tutorials in the future,... I can't even design a 30mm bed levelling knob in Fusion 360, yet, because I don't know where everything is, tools-wise. anyway, I came from the 3D Printing Nerds, Channel and loved what I saw and had to subscribe. Simple tutorials with-in fusion 360 when you have time, thank you.
Hallo Stefan, Vielen Dank für das perfekt geklärte Video. Ist das möglich die Topologieoptimierungs-Ergebnisse im offline-Modus zu öffnen. Wie und in welchem Format sollte das gespeichert werden? Danke im Voraus. Amer
Awesome job, but I'm wondering why you don't need to set a material before doing the optimization? Because wouldn't different materials require different structures to support the 100N force?
Can you give me advice about how I should do? I made a form of a planter from thin cardboard. I want to make it more rigid to avoid shape-changing when I will pour concrete into it. I made a 3d model of my planter in 3ds max before. How I should do constraints and loads in Fusion 360?
Thanks for this informativ video. My Question is, is there a way to get the Model in a sliser generate an infill generat an .slt, put it back to fusion and run a simulation
Interesting thought, there might be softwares out there. But I would say that the strength of a 3d print practically only come from 3 key elements, thickness/number of perimeters, orientation of the layers and the quality of the bond between layers. Even with an accurate STL of the sliced model only the thickness of the walls could be accounted for in a simulation on fusion. Simulations are all theoretical and used mostly to speed up human analysis. Meaning let the computer do calculations and then use humans to look and decide if results are appropriate or if a revision is needed. Humans still out think computers, machines are just quicker not better.
