Using the FEM Solver for car crash / destruction

[Abomination]

I'm trying to use the FEM solver for a car crash destruction and I need some pointers or tips.  The effect is looking great for about 5 frames after the crash, but after about 15 frames the car returns back to it's original shape, which is not what I want.  Which leads me to Question #1:  How can I keep that crashed shape without simply animating the whole object's Damping Ratio?  If I do animate the damping ratio to be an absurdly high value after the crash, the effect works great for the crashed area but the car loses all of its kinetic motion and the car just stops.

I've tried to use the SOP Solver to set attributes on the FEM geo but this doesn't seem to be working, so Question #2: Are the FEM and SOP solvers incompatible?  It's very possible that I'm not wiring these correctly, but it seems like the SOP solver is not correctly reading the FEM data, after about 8 frames the SOP solver gives me a warning saying that it fetched old data and it doesn't seem to be capable of viewing the entire animation.  Also I don't seem to be able to set primitive attributes like soliddampingratio and have it work correctly in the FEM solver.  Is this help page out of date with H16.5?

Basically what I think I want is for the Damping Ratio or the Shape Stiffness to be drastically raised for the primitives where there is crash damage and I would like the solver to see the crashed shape as the new rest shape so it no longer returns back to its original state.  Thanks in advance for any help on this.  And to anyone wondering about using other solvers (cloth or wire) I've found that the FEM solver gives me the best crash results because of the volume preservation.

[Hunter Williams]

Hey,

I think you should check out Steven Knipping's course on this- really great series on vehicle destruction, actually using rigid body system: https://www.cgcircuit.com/tutorial/applied-houdini---rigids-iii

Super flexible and fast.

As I understand it, FEM in the Houdini context is about volume conservation, so it will likely always want to return to its original shape, and you'd always be fighting that tendency if you went that route. Not really sure you want to be conserving volume with being metal, the whole point of basic vehicle crash is that it compresses and reduces "volume" -although take with a grain of salt since I'm new to Houdini

On a side note, tt's really weird because I understood the Finite Element Method generally as breaking a large system up into finite pieces to solve, so honestly not really sure why a voronoi fracture + RBD sim isn't "technically" FEM anyways.

Just saw your post hadn't been answered so wanted to give any info I had!!  

[vfxstu]

I recently had to do a car crash and used this approach as well. I'm not sure if all the attributes are the same as the ones on the link you provided, but here: http://www.sidefx.com/docs/houdini/nodes/dop/finiteelementsolver-.html about a third of the way down the page they are listed. I found that you need to apply these attributes onto the tet's as point attributes for them to work.

My big picture workflow for the crash was to simulate a low-res car frame with varied FEM attributes and some clever keyframing, then use cloth capture to deform the high-res frame. Based on that I pinned the edges of some low-res windows/windshields to it with pintoanimation and simmed them as FEM, then also cloth captured that geo's hi-res version. This gave me glass that bent and curled, which I needed for a shatterproof glass effect. To do that I timeshifted my hi-res windshield geo to $FEND and did some stuff to figure out most bent angles, then set a density based on that and used it with a Voronoi fracture back on the original rest geo. Then I made each piece of glass transform with the deforming hi-res windshield. Using that same attribute that determined the most bent angles, I released some pieces into an RBD sim at the point of impact.

Not sure if you needed/wanted any of that, but it was fresh in my mind.