3D Stress Simulation Service from STL (Free)

We are looking for feedback on an easy-to-use online stress simulation service we are working on thanks to a grant from the (US) National Science Foundation. The service operates on STL models uploaded to the cloud and requires a WebGL-capable browser such as Chrome or Internet Explorer 11.

This short video demonstrates its operation. You can also just jump right in. (Be patient, it may take a few seconds for the dynos to wind up.)

Please try it out and let us know what you would like, don't like, or would like to see added.

Thank you,
Michael
Intact Solutions, Inc.

This is wonderful!

I used to simulate turbine blades and vanes in aerospace, I did mostly heat transfer and CFD but structural folks would love this too. This is really cool!

I havent signed up yet but I definitely will.

Can you give me a quick run down on how you plan on meshing the STL geometry, as well as inspecting the mesh, modifying it (if at all) and then applying boundary conditions and post process viewing? Sorry for all the questions, I did not sign up yet so maybe all of them can be answered, but can you answer them for me first before I get into it?

Thank you,

- D

Whoa....

This will help a lot of my designs. I've been making prototypes and testing to failure. Now I can eliminate at least a few iterations.

One question though, the files as described as "In our inventory". To me that implies you guys acquire some rights to the files. Do you do anything with the files or do we retain all rights to them?

One immediate comment, It'd be really nice if the pan/tilt mouse control was not the same as the face select mouse control. In attempting to spin the model, I end up selecting faces I don't want.

Suggestion: Leave mouse wheel as Zoom, but change pan/tilt to mouse wheel click.


And another suggestion, some sort of pop-up that notifies you that the simulation is completed. I waited 10 minutes before hitting refresh, and it was done.

Also, the ability to edit previous simulations. As in, instead of having to select various faces again, I can just alter the amount of force.

Quote:

---

Originally Posted by Marm

...
And another suggestion, some sort of pop-up that notifies you that the simulation is completed. I waited 10 minutes before hitting refresh, and it was done.

Also, the ability to edit previous simulations. As in, instead of having to select various faces again, I can just alter the amount of force.

---

+1
and i was struggling a bit to select faces to apply loads.

Great tool !

Quote:

---

Originally Posted by DBFIU

This is wonderful!

I used to simulate turbine blades and vanes in aerospace, I did mostly heat transfer and CFD but structural folks would love this too. This is really cool!

I havent signed up yet but I definitely will.

Can you give me a quick run down on how you plan on meshing the STL geometry, as well as inspecting the mesh, modifying it (if at all) and then applying boundary conditions and post process viewing? Sorry for all the questions, I did not sign up yet so maybe all of them can be answered, but can you answer them for me first before I get into it?

Thank you,

- D

---

We don't mesh. Instead the geometry is embedded in a grid of basis functions that are used to represent the solution field while all geometric computations (integration and visualization) take place over the STL geometry. The resolution of the basis functions is controlled on the simulation setup page. Boundary conditions are applied to facets of the STL model.

An overview of the technology can be found on this page where there is also a link to a white paper with more details.

Thanks for your interest.

~Michael

Thank you for all the suggestions. We are compiling a list of items to improve, among them mouse interaction on the face selection page - that one is particularly annoying. Oh, and the ability to edit previous simulations... would you want this operation to make a duplicate on which you could operate without overwriting the original?
~Michael

Quote:

---

Originally Posted by Marm

Whoa....

This will help a lot of my designs. I've been making prototypes and testing to failure. Now I can eliminate at least a few iterations.

One question though, the files as described as "In our inventory". To me that implies you guys acquire some rights to the files. Do you do anything with the files or do we retain all rights to them?

---

Hmmm... I hope it doesn't say "our inventory" anywhere. It is YOUR model inventory. The models and associated solutions just reside on our instance of Amazon Web Services S3.

Is retention of the original geometry a big concern? To be clear, I'm not arguing here... I ask because we are considering techniques where we never actually hold onto users' models.
~Michael

Any chance we are going to be able to upload file's larger than 3MB? :P

Watch the video you linked. Pretty sure it says "Our Inventory".

Quote:

---

Originally Posted by Marm

Watch the video you linked. Pretty sure it says "Our Inventory".

---

Ahh... I see that now. The gist of the commentary in the video is intended to be read as: "As users we see this... or we do that... to our models... to our simulations." We'll try to make the commentary more clear in future videos.

As far as I know I can't repair the video without breaking the link, but rest assured have no desire to claim ownership of any user's models.
Thanks,
Michael

Quote:

---

Originally Posted by Bassna

Any chance we are going to be able to upload file's larger than 3MB? :P

---

How big of an STL file would you find useful?

that does look like a very useful tool.

Not sure I understand a lot of the terminology as i never studied engineering.
But I can test it to see if your program says a piece willl break when i destruction test it and it actually does break :-)

As CA said, I'm not familiar with a lot of the terminology, nor theory. Are there any good guides or tutorials for understanding the basics of what you have here? I mean, I think I know what I'm looking at, but I'm not totally sure what it's trying to tell me.

Well, I stood up and started throwing things around the room.

Ran 1 simulation on a design I had already deployed into use, and had noticed a delamination failure, I then had cobbled up a Hubblesque fix for it, and it's doing fine, but the amount of material I had to add to the design, post deployment, negated any weight and space advantages I had originally gained. The failure was mainly due to me designing to an absolute minimum, and not building in redundancy. If I had (and later versions do), I still would have had a weight advantage, but a much stronger piece.

Well, after running the simulation, it predicted a failure EXACTLY where the delamination had occurred, which is the dark red line on the near vertical post. I'm going to use this service again to pre-test any of my load bearing designs. If I had this service 6 months ago, Would have saved me quite a bit of headache.

Attachment 6477

Interesting stuff here, at least in my mind, lol.

After Noticing the fail point originally months ago, I extended (in design only, haven't produced yet) one of the support structures on the sled. Showed a noticeable reduction in stress points, aside from one particular location.

Attachment 6481

Extended that same structure even more, thinking it would eliminate that stress point. But even with doubling+ the support, it showed no noticeable improvement.

Attachment 6482


Interesting stuff here....

Thanks for the site guys, it's already helped a lot!

Are you obliged to have a 90 degrees sharp angle at this point ? If not, round off your corner, it will help.

Also note that "red" does not necessarily mean "bad". The legend simply maps the minimum stress value to blue and the maximum stress value to red. Compare the numeric value that corresponds to "red" in each of your designs to judge the amount of improvement/not.

We are happy to hear you've found it useful.

Quote:

---

Originally Posted by Hugues

Are you obliged to have a 90 degrees sharp angle at this point ? If not, round off your corner, it will help.

---

Good point. Issue is a servo motor has to sit in that opening, so expanding the corner to round it without losing material isn't an option. But I'll play with it and see what happens.

Quote:

---

Simulation "Simulation 1" was successfully destroyed.

---

Cool way to phrase that lol.

Quote:

---

Originally Posted by Hugues

Are you obliged to have a 90 degrees sharp angle at this point ? If not, round off your corner, it will help.

---

Interesting results. Took the model from the second picture, and applied a rounded corner as you suggested.

Attachment 6492

Took a big step backwards. LOL. Interesting.

Edit:

Ok, jsut remembered what The OP jsut mentioned, Red does not mean bad.

In the original design red = 8.03 somethings (The unit is not specified).

In design 2, red = 5.4 somethings.

In the latest design, red = 2.6 somethings.

So yes, it looks like it took the stress, spread it out, and reduced it by a quarter.


It'd be really nice if I could mouse over the color scale bar at the bottom and have it say what the value is, as the scale is probably not linear.

Cool !
Sharp angles are to be avoided, they concentrate the stress, basics of mechanical engineering, which I studied :-)

Quote:

---

Originally Posted by Marm

Interesting results. Took the model from the second picture, and applied a rounded corner as you suggested.

Attachment 6492

Took a big step backwards. LOL. Interesting.

Edit:

Ok, jsut remembered what The OP jsut mentioned, Red does not mean bad.

In the original design red = 8.03 somethings (The unit is not specified).

In design 2, red = 5.4 somethings.

In the latest design, red = 2.6 somethings.

So yes, it looks like it took the stress, spread it out, and reduced it by a quarter.


It'd be really nice if I could mouse over the color scale bar at the bottom and have it say what the value is, as the scale is probably not linear.

---

Sorry, units are on the list of things to add. They are consistent with the geometry units you are using. So for stress:
mm --> MPa (mega Pascals)
cm --> Ba (barye)
m --> Pa (Pascals)
in --> psi (Pounds per sq. Inch)
ft --> lb/ft^2 (Pounds per sq. Foot)

All the deflection quantities in the results display are in the units of length used for the geometry.

The legend scale IS linear.

This is NOT a sales pitch, but our plug-in for Rhino has A LOT more capabilities... building the cloud solution to the same level of sophistication will take time.

Very cool! Thanks for the answer.