Making STRONG shelves with Topology Optimization

A project, that's fun, educational, cool-looking, and actually useful! What more could you want?
We're making topology-optimized floating shelf brackets today 👍
Grab the design www.prusaprint...
Why the topology optimized brackets weren't screwed to studs (and other questions answered)!
• Why the topology optim...
Materials used/shown:
MK3 go.toms3d.org/M...
Proto-pasta TOM filament go.toms3d.org/P...
DAS Filament Multicolor Galaxy www.dasfilamen...
Elegoo Mars go.toms3d.org/E...
Read the article to this video here: toms3d.org/201...
💙 Enjoying the videos? Support my work on Patreon! / toms3dp
Product links are affiliate links - I may earn a commission on qualifying purchases (at no extra cost to you)
🎥 All my video gear toms3d.org/my-...
I use Epidemic Sound, sign up for a 30-day free trial here share.epidemics...
🎧 Check out the Meltzone Podcast (with CNC Kitchen)! / @themeltzone
👐 Enjoying the videos? Support my work on Patreon! / toms3dp

[@Mattakattak]

How about a topology-optimized BIKE ? I wanna see that alien bike.

[@1248erik]

It would break

[@Markfps]

@@1248erik no

[@aaronbell5994]

Yes!

[@memoryerror]

That was my first thought as well, I had been dreaming of it since the '90s when doing solid modeling and mountain biking.

[@ivoandricic1088]

@@Markfps yes. topology optimisation doesn't just remove unnecessary material, it also removes strength from it. its called optimisation because you're saving more on not spending that extra material because you don't need all the strength. in bikes, you need the strength. even then, bikes can break (if you dont go into dumb detail about what its made of). so no - you're wrong.

[@DarthBlazer.]

Fusion no longer supports this in the free version, boo

[@jonstenSE]

Would be nice if the algorithm were aware of fdm printing and that you could configure layer orientation, then it could account for layer to layer weakness...

[@StephanWahlen]

Stefan and Thomas talked in one of their meltzone podcast episodes i think :)

[@dibblethwaite]

People have been asking for anisotropic materials for a good while. I don't know if the'll ever do it. The usual response is use autodesk inventor if you need that.

[@dejayrezme8617]

@@dibblethwaite I've looked into anisotropic materials for fiberglass composites but haven't found much for FEA software. It's (much) more complex and I think you need to know more about structural engineering.

[@WilkoVehreke]

Generative us doing that.byou can define a bunch of materials and different manufacturing processes (including fdm printing) and the generated design takes this into account.

[@neur303]

Same for the infill! But that sounds like a really hard problem. I would assume it would be easier with generative design, because it should be easier to predict properties of synthesized parts than parts where a wall moves when you remove something. Hmmm well if it is a linear property maybe you could just scale one dimension before the optimization and scale it back afterwards.

[@ashwalk85]

For future etymologists: This may be the first use of the term "analog documents". 2019, folks.

[@questionablecommands9423]

I was recently discussing how the term, "paperwork" now often refers to online forms (specifically forms required for tax, insurance, rental applications, and other formal purposes) but often requires no actual paper.

[@smorris12]

Wing mirror. Dash board. Trunk. English (and no doubt, all other languages) is full of words divorced from their original meaning. This is how it happens.

[@cryonim]

@@questionablecommands9423 funny, because we have been talking soft copies and actual paper-work as long as i can remember. Even though things are digital here in India there's just too much paper . Well we will someday look back at it too.

[@hrgwea]

Analog is still electronic, though. For example, old TVs, radios and video tape players & recorders are electronic devices but they are analog, not digital. A better term would be "physical documents".

[@jonc8074]

@@hrgwea i still have lots of electronic analog stuff in my house

[@betabenja]

they all break at the unsimulated part , where it attaches to the wall

[@JT-ir6vw]

yea the connection seems to be the weak point on this. Seems like it might fail from tear out due to shearing if overloaded, or possibly bearing if plastic is somewhat malleable.

[@scout2369]

Thing is, did he use a stud finder to more securely place the shelves?

[@greenthizzle4]

Scout2 it does not look like it lol, he just put it into the drywall.. he's clearly not informed on house work much, hes put his brain towards 3-D printing

[@danilooliveira6580]

I wonder if the software take into account fixed points, or only cares about load and shape.

[@cyan933]

@@greenthizzle4 I think most houses in Germany are built different from America. they don't use wood... it seems to be mostly masonry. If you notice Germans always mock American house construction methods

[@kriskeersmaekers233]

But youre using topology optimalisation and then a hollow print.....

[@TonyRios]

But did you see him stand on the shelve...

[@operator8014]

@@TonyRios But you could have done that with a straight beam bracket, that weighed 1/6 as much. This is poorly designed.

[@henrychan720]

@@operator8014 But it looks 1/6 as cool

[@operator8014]

@@EngineeringTechnikcom Not wrong. Actually completely correct. You may now leave this conversation that you are not qualified to speak on. Thank you.

[@Psyrecx]

@@operator8014 Says the person thinking their opinion is a factual statement.

[@CNCKitchen]

Those are some beautiful shelf baskets! Awesome work, Tom.

[@revertfpv2928]

They are beautiful but they gonna collect a LOT of dust

[@revertfpv2928]

@Zygy __ compressed air will just throw dust elsewere

[@DrRiq]

They're pretty hideous but the science is beautiful

[@MaxC_1]

@@revertfpv2928 All you need is a tissue and some water. Done. Something like Denture tablets in water are not corrosive at all and clean stuff very well. Always use them for my keyboards over Iso or anything else.

[@HeyMaruniko]

me: "youtube, why you recommending this? this is not my contents" youtube: "umm, it's 3 am?" me: "alright i take it"

[@CafeSarcasmo]

totally.

[@Leafyfpv]

its 3am again, youtubes doing its thing- clicked on this ~3:10am

[@R4V3-0N]

It's literally just about to hit 3 am here in the land of Kiwi. Your comment hits too close to home.

[@MushookieMan]

This is what it would look like if the Covenant had shelves.

[@BayLeafff]

You just nailed what my subconscious mind was thinking the whole time, it's definitely the purple haha

[@Ensource]

halooooooooooooooooooo

[@daos3300]

except the covenant would never ruin their shelves with white paint.

[@azoozmrz9944]

great video thank you so much

[@peksn]

2:30 Wait, they charge you for running simulations on a software you've already paid for, using store credits?? Have micro transactions found their way into even 3d engineering software? What is wrong with this world

[@TouchedAlot]

you're paying to rent compute resources to run the simulation. you could do the same thing at home if you had a supercomputer

[@danepher]

@@TouchedAlot You don't need a supercomputer to run such simulation. Engineers at my work run simulations on 16 core PCs at work with no problem on other software. They are just milking people for money.

[@baconlegend4170]

jus because your computer is a potatoe PC doesnt mean modern computers are sUpEr cOmPuTeRs

[@TouchedAlot]

most people don't have lots of computers ready and available to throw at simulations. there is no reason to buy, configure and maintain lots of computers yourself if all you want to do is run simulations because you can just rent the amount of computing power you need on demand

[@aerobyrdable]

As others have said, it's not a microtransaction. You're paying for their server time to run it on their computers. One reason given for them not allowing us to run these simulations on our own computers is that it would require them giving out the code for how to do it.

[@mattpictaggi7504]

Interesting how the supports resemble tree branches

[@marcowen1506]

interesting but not surprising: biology has been doing topology optimisation for millions of years. Trees are very, very efficient structures.

[@mastermati773]

For me those resemble bones. Bones grow where they are under stress. I don't know if that applies for trees but I suspect yes. It's all literally runtime topological optimization :P

[@skinny7010]

@@mastermati773 bingo -> smaht

[@joshuawiens1344]

I died a little bit when I saw those hardwood boards go from being beautiful to painted white 😨

[@StCreed]

Yeah. That was barbaric.

[@MaddieGraffeo]

Same! Saw the beautiful wood and thought 'he spent extra for this project' then came the paint. :(

[@Kabodanki]

Sadness level maxed

[@TheAudiopulse]

I liked the look of it.

[@TheCyclingArtist]

I would have used a filament that was wood pulp-based painted the edges on the boards and supports, and tied it all together that way.

[@madddog7]

Too bad that TO is not available in free version of F360 nowadays.

[@AndreasIndustriePro]

wasted $100 on simulation has sharp edge on highest stress area 10/10 engineering

[@boozebeard9501]

Wasted 15 minutes watching a video. Makes a comment about spending $100 when he says multiple times he's using the free option. 10/10 youtube viewer.

[@ZeUbinator]

0/10 listening comprehension

[@KyleDB150]

@@fayenotfaye wasted 15 seconds reading reply to comment Wastes 15 seconds writing reply about wasting 15 seconds replying to a comment me: wastes 5 minutes figuring out reply to reply about wasting 15 minutes watching a video me: next level youtube replier

[@Humbulla93]

@@KyleDB150 at this point my math skills aren´t sufficient enough to leave a good reply

[@Some__Guy]

You're all 10/10 people in my book. You beautiful strangers.

[@Buchsbaumschere]

In Rhinos Grasshopper there is an extension called tOpos which runs topology optimizations localy.

[@justasingledronevideo2583]

Buchsbaumschere what is rhinos grasshopper

[@MysticalDork]

@@justasingledronevideo2583 Rhino is a CAD package, and Grasshopper is an algorithmic modeling plugin for Rhino. A commercial license for Rhino 6 costs $995.

[@justasingledronevideo2583]

Ryan Willis that doesn’t sound like an optimal alternative to F360 if it’s free

[@sparkyferret550]

I also use Rhino, it's a much better option and comes with much more functionality then the limited subscriptions of fusion 360

[@Krizbonz]

@@sparkyferret550 I try to avoid Autodesk whenever possible lol

[@wtechboy18]

This would be a really interesting design to do an investment casting of and get aluminum brackets.

[@VMfilm329]

Spends countless hours designing new futuristic shelve brackets using modern technology and complicated mathematical algorithms... Doesn't drill into the studs and attaches it to just dry wall...

[@henrystoetzel6336]

Haha I was going make this comment but thought I should check the thread first

[@MadeWithLayers]

Uses brain power to write a comment.... doesn't think to scroll down and see the same superficial assumptions about the wall being made by dozens of people before him...

[@quaditz]

"Its 25 Bucks to run one Cloud Job" - Wow thats crazy! "Its 100 more Bucks to get the result to your PC" - WHAT? How is that not included in the initial price?" "Its both on top on a yearly subscription you have to pay anyway for" - "WT........... Ok I am out guys"

[@immediateentourage]

If we lower the price then people will actually use it!

[@chrisakaschulbus4903]

i'm sure that there will be some open source alternatives at some point... at least i hope so... i would never support garbage like this, no matter how "professional" it is

[@zeiphon]

I would like a strength comparison between the smaller topology optimised brackets and a similar (roughly equivalent) squarish design printed in a more optimal orientation (on its side)

[@ARepublicIfYouCanKeepIt]

I assume that Thomas was going for a more organic design. One for which generative design is better suited, but requires cloud credit expense.

[@fitrianhidayat]

I'm guessing squarish design won. the purpose of topology *optimised* is to use the least materials to achieve the necessary strength, as oppose to make it stronger than conventional shape

[@schwarzarbyter]

in the next episode Thomas will print walls for his home

[@MadeWithLayers]

Don't tempt me

[@rok1475]

Entire walls - too big. But small, topology-optimized interlocking bricks would be viable. People are building houses out of plastic bottles and cans - why not out of 3-D printed bricks?

[@greenthizzle4]

Thomas Sanladerer you need to print some studs for inside the drywall so your not screwing into plain drywall

[@joshuawiens1344]

Jake Mitch drywall anchors are perfectly suitable for light to medium duty shelving. Most of the crappy ones are rated for 50lbs per anchor.

[@NigelTolley]

@@greenthizzle4 It isn't drywall. It is actual material worth building a house from, a type of aerated concrete block which is the internal insulation, with a brick or solid concrete outer skin. Waterproof and light, and cheap, and no timber to rot or get eaten. The only downside is you need the correct fixings if putting up shelves!

[@hieronymusbosch8817]

How is maintenance on them? I'm imagining they trap dust in their phalanges.

[@ark8705]

You could add a solid layer (maybe glass) to the sides.

[@jonc8074]

try to imagine trying to clean a xenomorph hive or polish the furniture in Giger Bar

[@rafbuelens4908]

I hate it when I get dust in my phalanges.

[@SuperMakeSomething]

Very good video Thomas! One note (as Stefan from CNC Kitchen also points out) -- a big issue with this approach is that FDM 3D printed parts are anisotropic, so results should be taken with a grain of salt (or large factor of safety). I would be very interested in seeing comparison results in this regard between FDM and resin prints and their predicted vs. measured failure loads.

[@timbarrett4580]

Also the way he set up his simulation; he only constrained where the countersunk screws were and didn't add a planar constraint where the bracket touches the wall. That's probably why the program almost immediately lightweighted that section and didn't connect any ribs to it. And then he printed it with infill, which is definitely different than what the program was calculating for. It's a cool way to get very organic looking parts or to visualize your load paths, but I wouldn't trust it without a big safety factor.

[@Jaze2022]

@@timbarrett4580 it's a gimmick. I can't imagine this optimization being useful for any manufacturing process

[@_..-.._..-.._]

@@Jaze2022wrong

[@bene20080]

"teach you about topology optimization, while you're at it". Not really, it only teaches you how to use Fusion 360 to use their topology optimization tool. (Which is also good, of course)

[@KyleDB150]

it teaches you the essentials, go do an engineering degree if you want to learn it properly

[@Cyborgzerozero9]

That bottom part of that piece hes holding at 11:22 aint looking so hot . But then it gets glossed over like nothing happened so im like :V

[@david102994]

and I thought it was funny that he was like "what is this...foam concrete" and if he means drywall then it's amazing that someone with such specialized knowledge doesn't know about drywall. It's great to see examples of how experts are just normal people in areas other than their field. Like how Ben Carson is a brain surgeon but that doesn't mean his politics are good.

[@83hjf]

@@david102994 hey you numbnuts, do you understand this person is german? he probably speaks english better than you.

[@fhuber7507]

Most likely he printed another after the anchor failed to hold. Not blaming the design of the 3D printed part for the failure of the anchor.

[@r3dluff]

david102994 in europe houses are made of bricks and concrete, but sometimes if youre unlucky you can hit a hollow or foamy part that just doesnt hold. We dont have drywalls. He just hit a bad spot. Happens to the best of us

[@Joel-co3xl]

@@r3dluff idk where you're from, but drywall is pretty common in sweden.

[@akura2]

Dang!... I JUST found a use for topology optimization and now they've removed it from the free version

[@mhelvens]

Pretty neat! I don't think I'd have painted the models. At least not any of the curved sections. The colour gradient seems to work well with the organic shape.

[@chris15325]

Video should be called "making overly complicated shelf brackets with industrial design software nerds"

[@harrison298]

Damn, what a bunch of hypercritical blowhards in the comment section. I think your shelves look awesome man. Looking forward to the next topology optimized thing you come up with!

[@MikaelLAOhman]

Strength is not rigidity. This type of topology optimization of this sort can only ever optimize rigidity, typically at the *cost* of strength. It specifically computes forces * displacement because the math allows you do to this very easily, and inexpensively. Choose to optimize anything else, and the problem becomes thousands of times more expensive, requiring completely different approaches.

[@elyasrasti8943]

It works though....

[@JamesSeedorf]

@@thealienrobotanthropologist agreed, a simple right angle triangle would have been the strongest with the least material. All of the support for the shelf is adding zero strength here because the shelf is self supporting. It needs braced on the outside and at the wall, that is it.

[@Jehty_]

@@thealienrobotanthropologist Please upload a video that shows how you would design such a beautiful piece. For someone who isn't artistic this method is fast and simple and apparently it works.

[@Jehty_]

@@JamesSeedorf and a right angle triangle looks just exactly the same like the one your neighbor has. If you want your apartment look like an Ikea catalog, just buy the brackets. If you want something unique the video shows how its done.

[@carbon1255]

@@JamesSeedorf This is incorrect, the topology optimization also specifies a right angle brace. In this video he left in dynamic curves as limiters on the outside so a more "natural" shape would form. Also rigidity is the objective for this application. Strength is based on the slicer methods.

[@suplified]

youtube: making something spectacularly complicated me: using a triangle and doing just fine

[@aserta]

Neat, but i would never pay a farm-bait spawned product or wait hours on a farm-bait product to do it's designed and intended speed to get you to pay. It's scammy and baity and we shouldn't encourage such bad behavior. We have enough of it already.

[@DnBastard]

agreed. seeing those prices and the extra 100 it takes JUST TO GET THE THING THATS ALREADY DONE is terrible PR. at this point there's no reason I'd give them money, even though I would find the product useful and could afford to.

[@skaltura]

Sadly yes - Fusion360 is excellent, but all the cloudyness of it sucks. I would probably be happy to pay something like 300-350€ for a license i own and everything works locally - but i will never go for this cloud model never ending cost BS. Especially when enforced like this.

[@aserta]

Same, i have no problem with the price, i have a HUGE problem with the way it's done and i have an even bigger bone to pick with the online only aspect of it. This is a scam as far as i'm concerned, because i pay (i have the full blown program for work) a sum of money, but essentially get only the UI (or fully get to that point in the future, because that's what they want to do). Like buying a car, but you can only drive it when they let you drive it, and at some point, if they want to, they can take it back from me. Ross over at AccursedFarms (creator of Freeman's mind) has a video on a similar subject ("Games as a service" is fraud.), but for the gaming world. It's a stupid trend that's been getting pace lately. I refuse to use such features. And i would never buy this kind of a product if i wasn't forced by circumstance.

[@MarkWheadon]

I'm pretty sure that PLA under load creeps over time, which would not be good. It looks //great// though.

[@AlgoDar]

That's true what you said in the last part, the process does not care how much load you have if you are only specifying (1) Strain Energy and (2) Volume Constraint. You need to put additional constraints such as deflection so that the software optimizes your structure given a certain load. I use ABAQUS for Optimization by the way. By the way, your video is awesome! so much appreciated!!!

[@duicenasty1078]

Good to know that 3D prints are stronger than modern houses

[@williamprimeee]

Amazing video you are great as allways its autodesk that is the problem now Perhaps an upgraded version that uses freecad could be nice 😀 So much for it being free, for a while now its felt like fusion has been on a downwards spiral and the older versions were better :( oh how i wish i could downgrade to 2019 fusion or 2020 fusion

[@AlohaMilton]

Question: Does the simulation understand the model is full of hollow cavities? Or did you print them at 100% infill? How does this work with the material properties related to FDM?

[@spagamoto]

It does not. You can shell your part before doing an optimization. Maybe a subject for a follow up video?

[@MadeWithLayers]

Simulating FDM 3D printed parts is nearly impossible - between the intricate detail that is infill and the anisotropy that layer adhesion causes, it would be a massive simulation project unless you over-simplify the model to the point where it's only a very rough approximation.

[@AlohaMilton]

@@MadeWithLayers hmmm, I wonder if simulating a vertical slice of layers would be enough, the layers simulated in 2d across the center for adhesion and compression applied to the rest as a value of its tensile and compression strength might be enough to be 'close enough'. Making parts function under compression through layers and extension only through horizontally printed loops is tricky... its tricky, its tricky tricky tricky. Uh.

[@benschofield1361]

Simulation is not available on the free version

[@BryceSchroeder]

Wow, Autodesk is such a phenomenal racket. Remember when people were just upset over how expensive perpetual licenses of CAD software were?

[@solarEbike]

This video: ...topology optimization in the free version of Fusion 360... Me: oooh, I've been dying to try this out [fires up Fusion 360] AutoDesk: $60/month to play Me: F**k!

[@lawsonbarney9556]

should have just anchored the shelves in studs.

[@leukocyte3145]

Or use wallmates.

[@fakename287]

Drywall on stud construction is considered an absolute joke outside of America, what this guy has is aircrete panels (ergo, no studs to anchor into)

[@Randywm01]

@@fakename287 An absolute joke is not being able to securely mount things to your walls lol

[@keldencowan]

@@fakename287 if there are no studs, what are the laths nailed to? Or are there no structure components in a European wall? Just solid "concrete foam"?

[@MadeWithLayers]

It's an aircrete wall, with plaster as the visible layer. You'll often find a polymer mesh as a reinforcement in the plaster as well. More commonly, walls are hollow brick instead of aircrete here.

[@kal5765]

I really dislike how much of fusion360 requires the cloud. It even makes it laggy, the moment I turn on offline mode the app stops being laggy

[@dertraurigerest]

The installation is so botched the whole fancy structure optimization is utter bs

[@Locane256]

I thought the same thing, but I still think the video is great and useful. Some of us have normal walls that anchor properly 😂

[@TurboNitroMonkey]

Did EA buy Adobe and Autodesk? Remember back in the day when you would purchase a copy of software, install it on your computer, then do what you'd like with it without microtransactions? Oh... Those were the days.

[@MrGuyCali]

No longer free. :(

[@medyk3D]

That Multicolor Galaxy PLA is really nice.

[@pteppig]

Looks similar to the "waste" filament i got, that has a slow merging discoloration, which happens during colour changes in filament production. Great to see an other version with more significant, "forced" colour changes

[@Vicros]

Now I wish I wasn't colourblind

[@Dzeno2010]

Have you considered creep (material aging under load)? PLA is notorious for it. Also I still don't understand why people just don't design the part hollow where you don't need the material instead of using infill. You're infill also makes the simulation nonsense (besides anisotropy).

[@Ruthavecflute]

"Also I still don't understand why people just don't design the part hollow where you don't need the material instead of using infill." I think a part of it may be that it reduces the distance that the printer needs to bridge or create over hangs for when creating the top layers. As printers can be quiet restricted in how far they can bridge and overhang the fill makes it a lot easier to design printable structures.

[@joshuawiens1344]

Designing hollow parts for 3d printing requires special design considerations (I.e. limitations) and printing hollow will *never* produce a stronger part than printing with until. Computer simulations don’t have to be absolutely accurate, relative accuracy is good enough in the world of 3d printed parts.

[@sawilliams]

The more complex our algorithms and processes get the more and more they mimic nature.

[@tombratfred3102]

I would rather say that nature and simulation converge on the same optimum.

[@forloop7713]

@@tombratfred3102 I would argue that evolution is a good optimisation algorithm which ran for a long time and found almost optimal solution for some problems, and machines are finding good solution to these problems now

[@tombratfred3102]

@@forloop7713 This is exactly what I implied ;)

[@speedbumphu]

I neither have ideas about mechanical engineering nor about Autodesk, but I have found this video very understandable, interesting and reasonable. Keep up the good work!

[@dadjake]

Welp... Fusion removed the simulation tab from the free version (because autodesk is greedy)... there goes that idea.

[@NoLan_YT]

Same, was just working on a project that needed the weight saving and strength, but Autodesk has to be the most Greedy company ever.

[@dadjake]

@@NoLan_YT I'm pretty sure the most greedy title still goes to adobe, but Autodesk is getting closer. Why use cloud computing for this? Why not try to run it locally? It's their only justification for making people pay for that service. Sure, it'd probably be slower locally, but I got time.

[@Shadowmech88]

I'd like to see a stress test comparing a topology-optimized shelf bracket with more conventional 3D-printed brackets, to see exactly how much more weight it would allow it to hold before failing/sagging.

[@almarc]

I don't think this allows *more* weight then the full brick if that's what you're asking. It's obviously weaker, but it takes out a huge percentage of the material for an insignifant strength loss.

[@ggeorgo555]

almarc the whole point of topology optimization is to reduce used material while maintaing the same properties as the original structure.

[@almarc]

@@ggeorgo555 I'm not a certified professional of course, but it can't remain "the same". It can only be 99% or something as strong. Unless you remove some bump on the side that literally carried no weight on it whatsoever.

[@Shadowmech88]

@@almarc I didn't mean testing whether it was stronger than the full brick. By "conventional" brackets, I meant like, 3D-print the kinds of brackets one might find in a store. One that's literally just an L-shape, one that's a triangle shape (L with a single crossbeam), etc. The topology optimized bracket, judging by its appearance, would actually have slightly more material than these. Sure, topology optimization is built for retaining mostly the same strength while massively reducing material used, but another way to look at it is that it can massively increase strength with only a small increase in material used, depending on what is used as the input shape. A bracket with 90% of the strength of a solid square block is likely going to be stronger than a bracket with 100% of the strength of a hollow triangle.

[@user-lf9gu3ij8h]

Is there anyway to do this for free? It looks like using simulation is no longer free. Please correct me if I'm wrong

[@holykoolala]

Could you compare the strength of these to one that you optimize yourself by intuition with the same amount of material?

[@regmigrant]

Am I alone in wondering why an accomplished 'tech needs so many paper files?

[@MadeWithLayers]

Accounting. Not everyone moves as fast as us when it comes to getting rid of paper.

[@regmigrant]

@@MadeWithLayers I hear you, nobody amasses paper like a bean counter except maybe a lawyer where they can charge by the linear metre :)

[@douglasmeadwell1417]

Simulation is not available in the free trial😭

[@easaspace]

First off all: great video! Awesome shelves! Then to my question: why gyroid infill? While cool to watch and fast to print, it's not strong at all. No two lines really rest on top of each other, so no load bearing compared to triangular or grid. I know your shelves were more than strong enough, but how about some experiments with this? Or should I poke Steffan with it? 😜

[@MadeWithLayers]

Definitely a topic worth investigating! Gyroid is supposed to be the ultimate (isotopic) infill pattern, so I use it whenever print time isn't super crucial.

[@michaelkowalski3745]

Did you figure out what settings make more fingers show up? I was inspired to give this a shot, and made a 80% similar shape to test shape optimization. Sometimes I get a finger or 2, but...mostly it solves without much interesting geometries. Are there any secrets or tips to force more interesting solve geometry? Currently experimenting, but it takes a while for each solve : /

[@evanalmighty9444]

had you found out how?

[@MrMauseri]

I have to say that this has to be one of the prettiest functional print I've seen (prolly one of the prettiest prints I've seen in total). Thank you for the STL file. Cant wait to get into Fusion 360 and topology studies.

[@WilkoVehreke]

Great video and I like that you've shown your iterative process. That you have not used the MMU but simply painted the parts is great, too. And the result is awesome.

[@5ilverstreak854]

TFW; When you 3-D print something stronger than your wall

[@the9file]

interested in industrial design. mounts brackets onto drywall with molly bolts and glue.

[@MsHojat]

LOL yeah.

[@almarc]

Here where I live all the walls are always thick and strong, so when I tried mounting some things with walls like those, it was very confusing for me at the beginning :D

[@kurtwagner6574]

It's not drywall lol. Here in Germany walls are typically made out of concrete or concrete bricks with plaster/render added on top, also inside of houses. So mounting a shelf with dowels and screws is the way to go. You can see him using a hammer drill in the video too.

[@the9file]

@@kurtwagner6574 thx 4 elucidating

[@TBFSJjunior]

@@kurtwagner6574 Actually it is foam concret bricks (ytong), so definitely not dry wall, but also not solid concrete.

[@spambot7110]

i would assume the point of setting a weight in the topology optimizer is for if you have multiple loads and want to configure their relative forces

[@mick0matic]

But the real question is, how did it print so well without supports!

[@lordtyroxx]

I really wish this was still possible in the free version of fusion 360

[@19mitch54]

You should use a studfinder to locate the screws in material (wood) that will support some weight. Drywall won't hold much. When I was doing finite element stress analysis with NASTRAN (a while ago), I had to accurately model the loads and constraints and the strength of materials. There was some voodoo involved. It was easier in that they were steel structures; I wasn't working with the complicated material of a 3D print.

[@antalz]

European buildings are almost never timberframed. They are nearly always built around brick or concrete walls. That's why he got out such a heavy drill, you pretty much need a percussion drill to hang anything in Europe in most cases. A studfinder won't do you any good here.

[@MadeWithLayers]

This is neither a stud wall not covered with drywall.

[@dieselphiend]

I am so excited for this tech to see wider adoption. The algorithms used studied things like bone structure during their creation. There's not many options as of yet, but I expect that to change. This kind of optimization almost spells the end for what we think are straight lines. Our world is going to get far more curvy!

[@ionpunk341]

I've done simulations like this in Fusion before, but I'm always confused what material to select. I wonder which material would most closely simulate PLA, and if there's maybe something that can be done relating to force vectors to compensate for the weakness between layers.

[@michbushi]

...try wood, maybe - as it has a similar directional strength profile? Cross-grain should approximate layers shearing pretty well, for the purpose of calculations?

[@jbhann]

No...the brackets did not hold up, as clearly seen at 11:18 with it being cracked.

[@jamesbra4410]

Looks like an intake manifold.

[@imbw267]

Tom: Says the guy who lives in an IKEA catalog IKEA: 👍

[@radarmusen]

Reminds me about fight club.

[@christophershafer5615]

I guess the simulation assumed solid material cross-section and not infill patterns?

[@casinoroyal93]

If that was the case this coul've been a research paper

[@burn1none]

the shelves are super strong, too bad they are secured to the wall with something far less durable

[@retoruest]

If you could place the top screw lower, you could make a flat surface and print them upside down. Maybe a bit harder to install them.

[@peregrinespurway]

Hi Thomas, great video, unfortunately Fusion has removed the functionality to solve locally. :(

[@VincentFischer]

Are there papers describing the math behind this? Would love to do this in blender as I think autodesk are criminals, even more so in the gfx market

[@loukask.9111]

due to the price?

[@shenlong21]

I know industry used 'finite elements' to calculate loads, temperature, pressure, deformation and other physical properties on structures. Maybe it can help you finding what was done in Autodesk's software. :)

[@Seecalator]

There's a nice paper called 'topology optimization in 99 lines of matlab code' or something like that and I think there's also a python implementation of this floating around online. If you happen to own Rhino 3D you can also look at the grasshopper and millipede plugins for the same effect. That being said, both are a lot more fiddly to get up and running than the tools inside Fusion.

[@DnBastard]

@@Seecalator python eh? nice that's what blender's programmed in.

[@CNCKitchen]

Search for "soft kill option" which is an easy way to implement such a bionic optimization.

[@512TheWolf512]

fucking hell, i'd rather not pay autodesk anything and not optimize my topology, than do it! how fucknig greedy, they're inexcusable

[@jpjokela1]

I think topology optimization is mostly good for solid components (as in, not FDM printed). When it comes to strength and lower material use, at least. I've noticed, that you can sometimes actually save material cost / weight by removing holes, as they will basically replace some (low) infill with perimeters. On the other hand, they can increase part strength. And look cool. The most important thing.

[@AnthonyHigham]

The simulation is made for a solid part. It looks like you printed it at 25% infill. Try 100% infill and you should get better results.

[@NigelTolley]

Better than holding 65kg+? What's the point of making it more expensive and slower to print?

[@AnthonyHigham]

​@@NigelTolley My point is that after testing the bracket in the video, he could even further reduce the material using the topology optimization using an infill at 100%. By doing this, the overall design will be reduced with less material. But like you said, even a 25% grid (or whatever infill type and percentage he was using) is surprisingly strong.

[@Mavrik9000]

I'd like to see it redone with the top wider than the bottom where it connects to the wall. Since that was the failure point, and it doesn't have to be straight and square to save material or make it easier to produce.

[@bru57000]

When I see your broken bracket without 100% infill, I feel like the topology optimization is a bad good idea because the finite element method will not take into account the "almost empty infill".

[@kevfquinn]

Can the simulations take account of infill patterns/density?

[@attitudeadjuster793]

No.

[@StuermischeTage]

Does this still work for free? Can't seem to solve the simulation without paying with cloud credits.

[@YetAnotherAaron]

you need to mount those to studs in the wall, anchoring to the drywall is usually a bad idea

[@MarkHanford]

It is not drywall. He literally says they are construction blocks. Not many countries build houses like the US does. They're still the wrong fixings though ;)

[@T0BBi94]

@@MarkHanford He literally has no idea what kind of wall it was... It was most likely a plaster wall. The can withstand quite some force when plugged correctly but it is always best to find a stud.

[@joshuawiens1344]

Drywall anchors are rated for like 50lbs + per anchor, and that’s for the crappiest ones.

[@NigelTolley]

@@joshuawiens1344 good luck putting a drywall anchor into masonry block. That's even more worryingly wrong than the slightly wrong fixings Thomas used.

[@marcusvoss]

lol concrete not drywall...

[@hvip4]

Shame the simulation tab is now locked in the version for individuals

[@johnschindler9641]

And the price went up, way up, for paying subscribers.

[@miperroesmuyguapo]

Great video and just as professionally made as always.

[@FMHikari]

Person: optimized shelves Me: you mean triangles..? Person: no, *unusual term here* Me: i'll stick to triangles, they never disappoint me.

[@kevinm3751]

Would love to find something like this for Blender.

[@operator8014]

Use Solidworks or Inventor. If you're a student, you can buy the non-commercial use license for $90 and it can do actual simulation work, not this nonsense.

[@kevinm3751]

@@operator8014 I have Solidworks but that was not the point! I favor having as many all in one solutions as possible. Sick of bouncing from one program to another for a single project.

[@BodywiseMustard]

You didn't mind that one of your braces snapped at the bottom when the rawl plug came out?

[@versatile3d64]

Great Videos Thomas!! Your my go-to guy for everything 3D.

[@JosephAlanMeador]

Wow this is really cool, topology optimization looks so organic (human or alien) and strong. Definitely liked and subbed, thanks for sharing this process!

[@clementhathaway6778]

Aaaaand F360 is getting nerfed in 2 weeks :(

[@manarallego3745]

Love this function. I am finally getting into actually working with 3D, making my own functional parts, so this is absolutely fascinating :D P.S.: Anybody wondering about the music in this video: Farrel Wooten - Lead Power Druid - Forever Falling Absolutely lovely

[@EnuffsEnuff318]

Toggle bolts would be better for that application....at least on the top set of fasteners. Those anchor screws/glue on drywall dramatically reduced the load bearing that you feverishly tested.

[@MadeWithLayers]

(not drywall)

[@ajlenze8]

This seems like a great (and extremely cool) design. Did Tom mention his print settings like the layer height and the type and percentage of infill? Also, I wish he had spent a bit more time describing the modified mesh he used to increase infill around the top mount. Finally, am I crazy, or does the design fit across the width of a Prusa MK3S?

[@theanonomous9465]

Those look sick. Great job

[@MadeWithLayers]

Thanks!

[@DanBader]

"...my 16 cores are bored to death while...." Lol, spot on!

[@RokkitGrrl]

Topology optimization examples reminds me of bones.

[@SciAntGaming]

That's because the algo is based exactly on human bone microstructures. :)

[@Fochit8611]

cloud credits are fusion 360's version of micro transactions...

[@OwenBuckingham11]

I used this same process to make a stronger version of my RC car chassis, worked amazing! Love Fusion 360

[@ricky107_]

That sounds cooool

[@yodaiam1000]

It would be interesting to see the optimization result if a stress limitation was used which proportioned a lower stress in the cross grain direction. It would be good to know if the optimizations was for strain energy or a limited stress/strain.