Topology optimization of 3D compliant mechanism flexures
Вставка
- Опубліковано 5 жов 2024
- Music by Kamui Yata - 404 Hack
/ 4o4hack
3D model files (.stl) can be found on:
github.com/art... under GPL-3.0 License
AM in TPU by Materialise
• TPU 92A-1, the first f...
Stijn Koppen
scholar.google...
Delft University of Technology
Faculty: Mechanical, Maritime and Materials Engineering (3ME)
Department: Precision and Microsystems Engineering (PME)
Specialization: Structural Optimization and Mechanics (SOM)
Project: Stable and Adjustable Mechanisms and Optical Instruments (SAMOII)
FAQ
What do (TX, TY, TZ) and (RX, RY, RZ) stand for?
These are the degrees of freedom of the mechanism, three Translations (TX, TY, TZ), and three Rotations (RX, RY, RZ) between the top and bottom.
Do you share the 3D model files?
Yes, 3D model files (.stl) can be found on GitHub:
"artofscience/flexures" under GPL-3.0 License.
( github.com/artofscience/flexures )
Link in the description.
Wow, what is this music?
This is music by an old friend of mine: Kamui Yata - 404 Hack ( soundcloud.com/kamuiyata/4o4hack )
Is this the future?
Yes. Compliant mechanisms, and thus monolithic flexures, do not suffer from "play" or "backlash" and hence reach extreme repeatability.
This is mainly useful in high-tech applications (nano/micro technologies, (aero) space, robotics, medical devices, etc.)
What tool are you using to design these flexures?
Together with a colleague, I wrote my own code in MATLAB, PYTHON, and C++.
The flexures featured in this video are generated running the C++ code in parallel using PETSC ( www.mcs.anl.gov/petsc/ ) routines to handle the parallel Message Passing Interface (MPI). The design method is very versatile and can be implemented with minimal effort in any state-of-the-art Finite Element Analysis (FEA) code (e.g. COMSOL, ANSYS).
How did you manufacture these?
I did not manufacture those myself.
Those flexures are additively manufactured by Materialise ( ua-cam.com/video/H2kLEG_RBRw/v-deo.html ) using a Selective Laser Sintering (SLS) process.
What material did you use?
The material is TPU 92A-1 ( ua-cam.com/video/H2kLEG_RBRw/v-deo.html )
Is this work published already?
No, unfortunately, it is not.
I will present this work at the World Congress of Structural and Multi-disciplinary Optimization (WCSMO 14) (@t in June.
The conference is fully virtual and I expect that authors have to make a video explaining their work.
Will probably make a separate one, to share with you on this channel.
Who are you?
I am Stijn, a Dutch PhD researcher at the University of Technology in Delft.
My expertise is in structural optimization and mechanics. More specifically I focus on the topology optimization of compliant mechanisms.
This work falls under the project "Stable and Adjustable Mechanisms for Optical Instruments and Implants (SAMOII)".
Nice work!! congrats
What a time to be alive! Hold on to your papers!
Hi I'm doctor Károly Zsolnai-Fehér's
Will share with you asap :)
www.researchgate.net/publication/356025963_A_simple_and_versatile_topology_optimization_formulation_for_flexure_synthesis
The video: compliant mechanism
The music: impending sexual assault
Would love to see a full explanation of what went into the optimization of these cool structures
I will present this work at the world congress of structural and multi-disciplinary optimization in June. The conference is fully virtual and I expect that authors have to make a video explaining their work. Will post that on the channel!
@@StijnKoppen congrats 👏👏👏
@@StijnKoppen Is a paper on the way? Or have you already published?
@@fetsackz2665 A paper is on the way!
THWg
I watched this while high, I dont know what the fuck that was, but I liked it, you got my sub
You sob, I'm in
@@SirLoinOfHamalot have fun
They are mechanisms that can only flex one way
Good music from Irak has done it...
@@derpdawg3492 or flex in all ways expect one (last mechanism)
Thanks for sharing, this is such an interesting field of study. Also, thanks for the cool music "kamui yata - irak"
i can't wait for bionics with little joints like this everywhere!
As someone who delves into this I appreciate at least somewhat the difficulty in achieving what I have just seen Bravo good sir
Hard to imagine what led you to believe that was a good audio choice.
just about to comment how great it was
this was an amazing audio choice!!
I enjoyed it!
I enjoy the variety of opinions on the audio choice :)
Good job on the optimisation, but that track is what made this video awesome.
fully agree here!
You have earned my Sub in about 2.69 seconds. Great work!
Welcome aboard!
I love how weirdly organic those springs are
These are absolutely brilliant man! By far the most impressive compliant mechanisms (with USEFUL characteristics) I've ever seen!!
Good to hear!
very good insight into the future, thanks for sharing
My pleasure!
why is this video so scary?
Vaguely organic shapes, and slightly melancholic music
@@leavoa The color and texture of the structures? The bizarre movements?
not an Aphex Twin fan?
I had no clue you could sls tpu, that's some pretty neat lookin' stuff
This looks like it was made by some alien
I’m curious - what material did you use to make these awesome geometries?
TPU!
@@StijnKoppen thats a dope tpu texture
@@jason_man prob spray painted for visual analysis tools
u can track the dots with software
but thats just my guess
@@officer_baitlyn check ua-cam.com/video/H2kLEG_RBRw/v-deo.html
Very interesting. Actually practical too. Any number of ways this is useful.
This is just some examples of couplings. Any type of coupling can be designed :)
"Your Ancestors Called it Magic, but You Call it Science. I Come From a Land Where They Are One and the Same" Thor
WOW! this probably is the future!
Also it explains some sci-fi things, like those weird flying bikes at the end of The Avengers (2012)
Do you have a link to the "flying bikes" you refer to?
@@StijnKoppen ok, I had to look it up. I meant "chitauri chariot". But looking at it now it isn't as similar to your work as I thought it was.
@@StijnKoppen But still your work is really cool. It showed me a new world and a new word: topology optimization.
I also read about generative design - could this method also be used to create these? (or even a better version of them)
I'm really new to this, I don't mean to disrespect.
@@SamoScopom In my opinion, Generative Design is simply a "commercialized" word for topology optimization. You can compare it to people calling their tablet from the brand apple their iPad, whereas, it is a tablet.
Very cool, great work! And the music was as interesting as the video 😉👍
I will let the authors of the music know :)
Preprint of the paper is available: www.researchgate.net/publication/356025963_A_simple_and_versatile_topology_optimization_formulation_for_flexure_synthesis
UA-cam recommendations found yet another gem
*casually rips off the platform with a gentle tug*
Nice work! Does the optimization account for nonlinear deformations (Both geometric as material nonlinearity), or are these results from a classic linear topo optimization?
Again, nice work!
Thanks for your comment! Unfortunately, it does not account for nonlinearities. That would be a good next step. However, including geometric nonlinearities in topology optimization is a tough challenge.
Thank you for sharing this!
No problem!
that is interesting. the rx, Ry would only really work with foams though.
can you explain ? :)
@@StijnKoppen sorry i should clarify to "work well".. you need that center post to have a lot of bending flexibility. that equation is modulus*(.25*pi*r^4)/length^3. you have a tiny length which has a cubed term and a ^4 for bending.. large deflection cause high stresses fast. and all of the axial load has to go through that tiny joint. essentially, you will fail fast in stress for a large deflection in rotation or fast in any reasonable axial load.
@@captainvlog good thoughts! Indeed, in theory, it only works for infinite small rotations and translations. Stress levels are not considered in these designs (however I can include them quite easily). One of my other videos includes a compliant mechanism design with manufacturing tolerance and stress constraints. There you can see longer flexures, as you explain.
TZ would be a step towards a printable speaker driver.
Hi! which univeristy are u affiliated with? im a msc. student in structural optimization at tudelft
Hi! I am also affiliated with TU Delft, 3ME, PME.
Beautiful
This is awesome!
super cool vid and stuff!!
Hey, this is amazing! Could you tell me what software you used for topology optimization?
thank you for your positive comment. these results are part of my phd project, for which my colleague and I write our own code in C++. however, you should be able to achieve the same results using any commercial package that can handle basic topology optimization requirements (compliance). this work will be published soon. I will share a link in the comments.
wow i was going to write this same comment! but ya, great job, this is awesome.
@@StijnKoppen any update on publishing?
@@williamhinrichs6558 working on it! Expect it to be publicly available in about one to two months.
@@StijnKoppen cant wait!
This is so cool
Pretty cool, would like to know how to come up with designs.
"topology optimization" in fusion 360 just go to the generative design tab and optimize for material and you get the process for a different feature in a few clicks. In these cases they're optimized for flexture but it's no different, just different parameterization
@@GRAYgauss
I am studying physics + math, learn to program an iOS App, learn Chinese.
- I didn’t have time to get in 3D modeling yet.
I thought you might calculate those shapes by hand based on a set of rules.
I have heard about topology optimization:
Somebody on UA-cam had a video about strong supports for a shelve made from the least 3D filament.
He said something about topology optimization.
The mechanism in this video weren’t as „stringy“, they looked cleaner, so I couldn’t correlate both topics.
@@GRAYgauss Fusion 360 indeed can do topology optimization for stiffness design. Unfortunately, most software are limited to stiffness design (as is fusion) and thus cannot design mechanisms yet.
R stands for Rotating, but what does T stand for? Torque? Transform?
Probably a language thing, since I learned this in german long ago and can't remember it too well
Translation
Very cool
great design!!
Music : When you can enter Poisson's negative
The compliant RZ one was interesting, was that what the software spit out? Other ones looked like something generated while the RZ one was something a human might come up with.
Indeed, the compliant RZ is close to what an engineer would come up with. A common design is simply 2 sheets in a cross. In this RZ design (if you look closely), the flexible lines are half-circles as opposed to straight lines. In general I observed that easy design problems (single objective, single constraint) results in recognizable design, whereas harder problems with multiple stiff/flex modes provide more innovative topologies.
wild compliant mechanisms distracting music LOL ! Wouldnt it be better if these mechanisms were not printed of rubbery material from get go ?
Flexing is better than joints.
My phd dissertation on this topic can be found here: doi.org/10.4233/uuid:21994a92-e365-4679-b6ac-11a2b70572b7
Hey, thats very very nice. I wanted to know is it the variation of thickness along the extrusion curvature that can yield for any material's compliant mechanism tendency. And if so, how do you determine the optimal thickness and it's location ? is the location through the springs in your PRBMs(pseudo rigid body models)? Do you also think polar moment of inertia can help predict curvatures thickness ?
Thank you for your time. I would also want to know can I be given the chance to email you related to this topic ?
Rajkaushik, I do not use a prbm. These are designed using topology optimisation. My email is in the description ;)
@@StijnKoppen thank you so much for your reply, Sir. I'll shoot you an email.
Hello! I've read in the comments below that these mechanisms are printed using slm. Are those printers available at TU Delft for students?
Thanks for linking introduction to topology by Jun Wu down below. Can you recommend more videos / other sources about topology optimization / compliant mechanisms?
Michael, unfortunately, they are not. At least not at 3ME. I would recommend following the TOP Webinars, which are monthly presentations by topology optimization experts. The discussions can be followed on youtube!
@@StijnKoppen I see, thank you
that's cool
Awesome research! I'm wondering will FDM way of printing cause on directional properties of these structures. Did you research the differences between FDM vs SLS printed structures? I'm an active 98A TPU user on FDM.
It is indeed expected that FDM will cause anisotropic behavior, although I did not experiment with this. My research is on the computational design :)
Please some one tell if there are some basics to read or watch for designing such structures .
You might want to check the video of Veritasium: ua-cam.com/video/97t7Xj_iBv0/v-deo.html&ab_channel=Veritasium
You can also check the intro to TO by my colleague Jun Wu : ua-cam.com/video/5ocnVS_HvdY/v-deo.html&ab_channel=JunWu
Very interesting thank you❤️
Wow! Nice!
Awesome work!!
I have studied a little about TO for CMs. Can't wait to read the paper on this.
How were the required DOFs specified? Because generally for Rx Ry, the hinge is near the bottom and symmetry has to be enforced right?
This is amazing work. Is there a paper about this research. We'd love to read and cite.
Currently under review!
Really cool and inspiring!
Would be cool if those shapes could be sand casted in some metal, and have those properties. Like if you could sand cast a spring, instead of making a spring from a metal thread which has been shaped into a helix.
@@bosna737 AM using metals is becoming achievable! If I may ask, why would you want to sand cast?
@@StijnKoppen what I'm aware of at this moment there aren't any international standards that ensure that a 3D printed part meets the requirements stated on a drawing and there aren't any standardized methods to ensure the quality. But if you use the 3D print as a core in a mold, then it should be possible to "piggy back" of the already well defined standards for casting 😄
@@bosna737 that would be a smart "workaround"!
I still don't know if it's possible but would be interesting to explore
What tool are you using to do this topology optimization, and what degree of simulation of material properties and dynamics does it support?
I use my own code (C++). For those simulations, I use linear isotropic material properties and did not include any dynamic considerations. One of my other videos includes dynamic considerations ;)
@@StijnKoppen nice, do you have a github where you've posted your published works reference code? I'd like to dabble in this stuff
Wow amazing work! I am doing similar research looking at optimizing lattices for energy dissipation but am currently killing my poor laptop with heavily nonlinear Sims. Have you found significant differences between linear and nonlinear simulations of your mechanisms? I would be very interested to see your linear material models for SLS tpu as I am sadly unable to do any printing or testing. Do you plan to have any research available before May?
Good question. For mechanisms the geometrical nonlinearities can be very dominant, this can also be seen in the video (e.g. stiffening of RZ-compliant structure). In general, I am surprised how good they work for a finite range, but this is fully case-dependent. I am working hard on the paper, however, I do not think it will be online before May.
www.researchgate.net/publication/356025963_A_simple_and_versatile_topology_optimization_formulation_for_flexure_synthesis
Muy interesante y educativo!
Muchisimas gracias amigos!
amazing cover :D
Incredible work here! Can this work be replicated on Fusion 360 or Ansys? Also, is your work publicized yet Stijn?
George, thanks for your comment. This work can easily be replicated, a publication will follow soon. Will post the link in the description.
@@StijnKoppen time for link yet?
@@donaldviszneki8251 working on it!
www.researchgate.net/publication/356025963_A_simple_and_versatile_topology_optimization_formulation_for_flexure_synthesis
that music is fkng dope
Kamui Yata - 404 Hack
Please show scripts or whatever programms used!
I wrote my own code in C++. I will present this work in June on an optimization conference. A simplified code (Matlab) will be available on my github (artofscience) from that moment on :)
@@StijnKoppen great, thanks, is that conference going to be public? thank you and great work!
Is there a model that covers all rotations and translations?
WOW
Had a problem at work that called for a coupling that was compliant in x,y,Rx,Ry while also being rigid in z and Rz.
Couldn't crack the problem with any sort of flexure, and had to resort to a more complicated and less elegant assembly.
Would you be willing to throw that problem at your algorithm and see what it comes up with?
Yes, will try it out!
Sounds like a though one..!
@@StijnKoppen Remembered this today, Have you maybe had a chance to give it a try? really curious!
What material was used here? Can this be achieved with PLA on FDM printers?
The material used is tpu, achieved using slm. Due to overhang limitations this can not be achieved using fdm techniques. I hope this clarifies things. Feel free to follow up with questions
@@StijnKoppen Thanks for the quick response. I'd love to recreate this on an FDM with multi extrusion, either using water solubles or plain PLA as support that could be broken out when using TPU as main material. Will you open source the STL files?
@@Bananananamann That sounds possible (not sure though about the connection of TPU with other materials). I will open-source the STL files soon. Will share a link in the description. Looking forward to see your results.
@@Bananananamann The .stl files can be found on github.com/artofscience/flexures under GPL-3.0 License
@@StijnKoppen you're the best! Thanks so much. I'll report back
Dann that’s cool
How difficult would it be to rerun the simulation with a Z dimension at 0.25 the original size?
Not much effort; all parameters are relative; dimensions do not really matter (as long as constraints are properly scaled along). This also means you will simply get the same result ;)
@@StijnKoppen I understand that you would get the same result if you scale down x, y, and z to 0.25 scale but if you left x and y the same and just reduced z it seems like some of the designs it made would have to be adjusted to account for the reduced space.
Which song is this? Sounds sampled from Broken by Lund.
Kamui yata - 404 hack
how did you convert your C++ to a mesh file format Materialize could use?
Using my C++ code i generated a structured grid. The output of my code is simply an nxn density cloud with corresponding displacements on the nodes. The density cloud I exported to paraview. Using some postprocessing in paraview I created STL files to be used by Materialise.
can you send me the part file of compliant RX RY
❗👍👍👍👍👍
Great work ❤
Thanks fauzi!
@@StijnKoppen Im curious about your design and manufacturing process?
@@StijnKoppen I see that you already published your work so I’ll check that out. But what about the manufacturing process that you used to make these?
Quick update: paper submitted :)
Where can I get it?
' Topology optimization of flexures: a simple and versatile formulation ' right? Can't wait to read it.
@@vaishakhcuntoor7349 indeed, you joined my presentation? :)
@@StijnKoppen I wanted to watch your presentation but couldn't make myself available.
@@thiwillilo www.researchgate.net/publication/356025963_A_simple_and_versatile_topology_optimization_formulation_for_flexure_synthesis
Do you have a version for RX only?
Hi, I love your work. I'm currently doing a Master's degree in Additive manufacturing and I would really like to apply something like this to an upcoming project. Do you have an email I could contact you on?
Cool to hear about that! I do agree this set of flexures might be a nice benchmark set for AM. You can contact me on: s.koppen@tudelft.nl
First I read Addictive manufacturing
Was that a left-handed coordinate system?
haha i was expecting this, you are the first though ;) couldn't correct it after making it public
@@StijnKoppen Lmao that's funny. I'm glad that you said "correct" it, though. Because left-handed coordinate systems, while mathematically valid, are wrong.
no you didnt search for this video
How many movements is this good for/what's the fault tolerance like?
Good question. I did not do any experiments (yet). These designs do not take into account stress levels nor fault tolerance. Thereto, one cannot expect to much from it at this point. However, those can be easily included. One of my other videos does take those into account.
Could this be used as a car CV joint?
In theory it could, but in practice it will not mostly due to fatigue issues. It is also not designed to retain its constant velocity property in the deformed configuration. This one does: D. Farhadi, N. Tolou, J. L. Herder, “A Fully Compliant Homokinetic Couplings”, Journal of Mechanical Design 140 (1), 012301 (2018).
how the hell do you even manufacture them?
Could be 3D printed, just a guess though.
Sls 3D printing
I don't understand and it's giving me headache
Each displayed structure is flexible in some directions, and stiff in others. It is useful for different applications, e.g. in mechanical tools. "Compliant RX RY" means that it rotates (R) around the X and Y axes, but is stiff in the Z axis. "Compliant TZ" means that it translates (T) in Z, but does not move in the other directions, neither rotates.
@@henriksundt7148 Well explained!
What 3d printer is it?
Sls
in which software did you do it?
I write my codes in C++ (incl. openmpi), Python or Matlab, these results are obviously from C++ considering the huge amount of dofs
How did You make That? What did you studied?
I’m a phd candidate my work is on structural optimisation, specifically compliant mechanisms for high-tech applications
Wat material is tis
tpu
weird flex but ok
music is attrocious...
How to print this thing?
SLA or FDM?
Sls:)
Ah yes
The general population for some ungodly reason
The horrible music makes my teeth ache.