Here's my situation: I work for a company that 3D prints photogrammetry scans, and as it stands right now about 10% of these prints break in some way during manufacturing. This is because the prints are done with an inkjet and powder type setup that leaves the parts fairly brittle until dipped in a resin.

My job is to generally improve on the 3D pipeline from photogrammetry to printable model, and I'm using Houdini's procedural tools to help manage the situation with more convenience and consistency compared to just doing everything manually.

So I'm working on a node flow that, among other things, identifies areas in the print that are thin and therefore presumably fragile. This part works great, but as is often the case with procedural operations on the geometric chaos of photogrammetry mesh, it's hard to tune things so that it perfectly identifies the exact thing you want every single time. So the identification of thin areas is probably a useful thing, but it's not perfect in real world conditions and requires a human to assess the riskiness of any particular thin area — the strength of the part is highly dependent on how it's connected to the rest of the model, for instance.

So I was thinking to myself today, would it be possible to use Houdini's RBD simulations as a sort of jerry rigged finite analysis software that can identify areas that are structurally unsound and likely to break?

I know that's an off label use case, but hear me out. The 3D printers we use work by squirting a little blob of glue into some powder, and doing that however many millions of times to build the 3D print that is made up of basically a bunch of little blobs that are stuck together. The blobs are of a fairly consistent shape, and the strength of adhesion between them is also pretty consistent.

So all in all this is a fairly simple and low stakes Finite Element Analysis problem to solve compared to rocket engines or skyscrapers or whatever, and it seems to me that the nature of it is very well in line with Houdini's ability to simulate destruction based on calculation a force's effect on a collection of volumes that are similarly stuck together with glue.

So here's a broad strokes description what I was thinking to do:

Since I already have all of the thin parts selected, I just break those out into their own geometries and make the borders where they connect to the rest of the geometry fixed so that they are unaffected by the simulation. And then I just break those pieces down into a collection of chunks that more or less resemble the structure and stickiness of a 3D printed solid, and then at that point it's just a matter of hitting it with some force from a variety of plausible directions to see if it breaks, ramping the force up to a particular cutoff threshold. If the thing doesn't break after all that abuse, then presto, it passes the test. If it does break, then we have a human double check the results and see if changes need to be made.

So do y'all think this is a crazy use case? Or is there some promise to it?

Any reason you can think of that it probably won't work would be greatly appreciated.



(disclaimer: I know that there are far better dedicated engineering softwares for this, but the ones that can handle arbitrarily shaped meshes are complex, computationally expensive, financially expensive, and extremely overkill for our use case. We're a small team, and we can't afford to get a dedicated structural analysis person, so these sorts of Houdini hacks might prove to be useful. I'm not looking for perfection here, and I'm not going to use this as a serious CAD software to design a long bridge over a deep crevasse full of crocodiles or some other scenario where lives are on the line. The goal is just to create something that might be useful in identifying fragile parts with more consistency and improving on the aforementioned 10% failure rate, even if only improving it by a few percentage points. And don't forget, the worst case scenario here is that it doesn't work but I get paid to tinker with and learn about something that interests me.)

Have experience with Z-Corp printing, you have million tricks how to Print if you use those 3D printer)Nice idea But But Geomagic , in Houdini you can do those analysis and make simulation based on that but but ,Sure you can build asset and algorithm it gonna be 40% +- accurate , remember that you must use Shell also in between 2.5 mm to 3 mm (thickness) Houdini can't compute that , can maybe some toys, but not complex shapes ..Or ??? Fun project HM...intrigggg..

Try why not Have Fun..