Moonhawk roof

So for over a year now I've been trying to get the Moonhawk roof to deform even remotely like the Grand Marshal in a T-bone "poleshot" type wreck.
This type of crash is my favorite and of course involves major roof deformation. I have tried every conceivable j-beam code tweak imaginable with zero
success. The closest I've gotten to a satisfying "upbend" on the Moonhawk was smashing it with the T-75 while parked against a wall, and I don't consider
that viable. The Moonhawk roof tends to deform laterally versus horizontally and when getting close to bending upwards it starts doing a "sharktooth" thing
where it looks like the j-beams are confused which way to go. I think this has to do with more than just the roof and is probably a systemic whole car beam
thing.



Poleshots on the same pillar at the same speeds and angles.

Has anyone gotten the Moonhawk roof to bend upwards realistically?....post shots!....I don't care how you do it.

 

jesus, I thought I was the only person tweaking the roof deformation of lots of vehicles. The covet is another oddball I'm trying to fix. Good luck with it man

 

heh..heh...we are a special breed you and I. I have a Grand Marshall simply called "Grand Marshal 2" that I use to test with. I wasn't happy with the pillar deform on it and found
a happy medium to get pillar "push up" in a severe head on crash.

I've had so many "Moonhawk 2's" I lost track.

 

Okay, that makes sense and describes it perfectly. I figured the Moonhawk had more nodes/beams in it's roof, (being a larger roof/car), and mere beam value changes
were probably not going to work. Question answered, thank you.

 

2 rules G, A pillars do not normally bend and the B pillar tries to move inward in a head on crash. Save me and aborath some agony



This is the extreme case of what happens. The floor bends in a textbook manner, the A pillar doesn't get bent in half, but levers the roof up and the B pillar collapses inward. The only car this happens on is the Sunburst and only kinda. They drive well and crash well already, but they can crash even better. You may even notice we get some of the same kind of damage we see in a pole crash, the movement of the B pillar pulls the roof into a "tent" shape. I think some tweaks are needed still.

 

That vid perfectly illustrates the issue I have with Beam vehicle pillars.They bend and warp a bit too much. They tend to follow what the firewall is doing and in some
instances will actually drag the roof down in a head on. Part of my rather amateurish fix was to make the A-pillar deforms extremely high and the
roof corner beams extremely weak. This is acceptable for the time being. I need to dig further but another thing affecting the pillar/roof bend is the firewall
and how it tends to bend backwards from the floor pan join. Is it possible to assign a selected group of nodes/beams within a larger group of nodes/beams
and assign a separate set of value's?? Below is frontal offset with modified values and is about as good as I can get it. Having both A's at the same values need to be
a little different for realism otherwise the roof will crinkle in a straight "line".

 

pillar deformation depends a lot on the vehicle. newer cars treat the passenger compartment as a 'safety cell' and the pillars are stiffened to protect them. older cars the pillars are just there to hold up the roof. looking at crash tests, one of the key indicators is the structural integrity of the safety cell.

 

Good points. I need to watch a shitload of crash tests and get a better perspective on the issue while at the same time continue to tinker. Some how the "downward" roof
collapse in head on collisions is due to two things, the firewall is interacting with the frame and will fall backwards dragging the pillars and roof with it and at the same time
the roof is reacting to the rear end pushing forward. Coincidentally, the same rear push force is what also provides upward bending force to the roof/A pillar intersection in
a collision not involving the frame so much.

 

Three pics illustrating the firewall bending backwards on a GM with various modified deform/spring/damp values. Merely just to show the firewall action
with driver side door removed for clarity.

At the current state I can get fairly good roof buckle for now. No matter what I do with the A-pillar values they always seem to "bend" in a round fashion
versus a clear cut crinkle. Is this because the beam/node count isn't sufficient enough to replicate a real world A-pillar? I would have the same question
about the firewall, beam/node count doesn't support highly detailed deformation?

Regardless, Lowering the roof/A-pillar corner beam deforms to 10-13K and raising the A-pillar to 900K gets some pretty good roof crinkle. Stock roof width/length
values appear to be fine as lowering them too much yields unwanted deformation in the rear.

 

Apologies for the late bump, but it takes time to do these things and well... One of these suffered a 40mph offset test, the other was T boned at 30mph.



and here's a modified jbeam

View attachment hatch.jbeam

Is this better or worse? this is the covet and not the moonhawk. I probably overweakened the B pillar, you get a hell of a lot of deformation out of a side impact now. 30mph side impacts resembles what I see in those russian dash cam videos. A pillars act properly and the B pillars collapse slightly during a frontal impact. What do you think?

 

Don't have time to look at the beam file right now but the A-pillar/roof deform looks straight and not "round", Something that annoys me about the Covet. No problem with
other vehicles, this subject applies to ALL vehicles. Can't wait to see what those values look like.

 

Two questions: Is your covet born of the experimental build and did you add new deform/strength values to select node clusters? I used your values with miserable results
without adding the new value lines. Those lines are the only thing new in your hatch jbeam file different from my old stable covet jbeam file.

 

Stock unmodified Covet T-bone at 30mph. It'd be nice to be able to remove damaged parts in the parts selector and take a look at that B pillar.

 

Yeah, I did a lot to the covet, and I've done even more which I'm going to attach in this post with a brief on how I did it. Yes it's the experimental build version.

Gabester has structured everything the same for all his vehicles. With the covet I managed to get a solution with the following: There's a beam for both A pillars that goes from the top to the bottom. Single them out and set the beamdeform to 10-15k. There's a set of beams that go from the top of the B pillar to the middle of the A pillar. You will find four of them. In a crash these tend to be what pushes the A pillar outward. Single them out and set their beam deform to 25 so they still hold the windscreen stable but they apply next to no real force in a crash. You now want to make four new beams to give the A pillar more support from the bottom of the firewall to the top of the A pillar, these serve to push the top of the pillar upwards in a head on impact. Then from the middle of the firewall to the middle of the A pillar again to stop the windscreen wobbling.

In a head on crash the B pillar is usually forced inward. There's a set of beams that hold the roof up from the B pillar. Weaken them significantly. You will find also a beam going from the middle of the B pillar to the roof. Delete it, for some reason it limits the movement of the B pillar far too much which gave you that weird shape in your screenshot. Notice my B pillar has now deformed more "smoothly". Don't worry, that node is being held in place also by a similar beam to the floor. To which locate the set of "corner" beams that join the B pillar to the floor, weaken them but less so than you did for the roof. And finally locate all the beams that run the entire length of the roof, again weaken those. All this should allow the roof to bend more freely, do a better job of keeping the A pillar straight and allow the vehicle's B pillar to move more freely with the bending of the floor. Just like you see in the chinease car crash videos.

When I weakened everything for the covet I set the beam deform values down to about 2 to 4K.

If you impact the B pillar further up like in that video, it does obliterate the car. The crash there was between the two covets, so the impact was closer to the floor which means deformation isn't as bad. When it's between the D15 and the covet the results are much much worse.

Moar tuning.
View attachment hatch.jbeam

A better solution would likely be to delete the entire front portion of the roof and redo it with one "segment" for the A pillar and two roof segments. but I decided to get used to tuning beams by working with what gabester set up. Because now there's a problem with when you land a jump the top of the A pillar flexes way too much but it looks better in a crash. The other problem is the A pillar can still bulge, the results are mixed but they are in the right direction. ITs just hard because the A pillar isn't rigid. I'm considering adding a node behind A pillar to turn it into an actual... pillar. I've used "//edit" and "//added" in the file to highlight major changes in the jbeam.

Ayway, because gabester's made all vehciles essentially the same, whatever works out for this could work out for the rest. It's not just value changes, I've done quite a bit of rearranging, adding and deleting.

 

I found the B pillar effect on the GM last night. Makes a big difference in the roof up bend as it provides another axis for the A pillar to release
energy against via the roof beams. It bends the B pillar in slightly and gives a nicer whole body deform look in super violent head ons. The info
on why the Covet A pillars are the way they are is priceless and I'd love to move that shit around like you say.

So I currently have no idea how to add/remove/adjust jbeams. I've been modifying values in the main jbeam files via notepad and that's it.
Blender?...modify those in Torque?....whada' I do.

 

Literally just adding beams and adjusting things in notepad, like you are. Get notepad++. It lays things out better. To add a beam all you do is write out this ["",""] and then fill in the nodes like this ["q1r","rf3"], and new beam on car between those two nodes. To assign a new set of values for strength, copy this line {"beamDeform": ,"beamStrength": }, and put it above the group of beams you want to change the strenght of

If you want to temporarily disable the beam, put a double slash // infront of it to temporarily turn it into a comment. if you determine you will never want it again, delete it.

{"beamDeform":100,"beamStrength":"FLT_MAX"},

["p4r","q2r"],
["p4l","q2l"],
["p3r","rf1r"],
["p3l","rf1l"],

{"beamDeform":10000,"beamStrength":"FLT_MAX"},
["p3r","rf2"],

That group of four beams are weak, with a strength of 100. while that single beam below is 100 times stronger. Whoever wrote up the parser for jbeam files is pretty ingenious, it's so simple and flows intuitively enough.

Anyway the new nodes have done wonders, it works! ... I also made the floor more rigid but much more weak, and I think it's mirrored the crash test videos of those early 90's civics now.




https://www.youtube.com/watch?v=5Y4yIjT83kA

Here's the file.

View attachment hatch.jbeam

Basically I've made it behave how you would want it to behave, and it's nearly as rigid as gabe's original. It's very messily edited tho, but maybe it'll help you modify the moonhawk and GM to behave like this.