Another XJ cage design thread, with an ME twist.

[vetteboy]

I've got some downtime while I'm on vacation in South Carolina, so I dusted off my copy of ANSYS and remembered how to use it again.

Basically, what I did was to create a program in ANSYS (a structural modeling software package, for those non-nerds out there) that allowed me to put in the coordinates of the A, B, C, and D pillars in the vehicle, generate a cage from these points, and analyze it from a variety of loadings. I did this because I don't have an XJ here to measure, so I created variables for each node in the cage (A-pillar base, A-pillar height, vehicle width, etc) that I plugged in arbitrarily for now and can later go back and edit to make them accurate.

Once the cage geometry is generated, there are 6 points that can be easily loaded (theoretically EVERY point can be loaded, but I only made variables for the top 6 nodes: a, b, c pillar on both sides). Each node can be loaded in any combination of X, Y, and Z direction, to whatever value I want.

The connections of the cage to the vehicle are fixed in all degrees of freedom, which means that I'm assuming the floor of the vehicle is rigid and won't deform, and also that the connections themselves are rigid. This is not completely accurate, but for the purposes of this program (and design exercise), it'll be close enough I think.

All tube connections are also assumed to be rigid, and all connected nodes share the same DOF constraints. This way I can be sure that all members in the cage are 'interacting' the way they would when welded.

I've also left variables for tube size and modulus of elasticity; the program then takes care of calculating moment of inertia (XX and YY) and cross-sectional area. I can also determine useful things like overall length of tubing used, and overall cage weight.

If anyone is really interested (you gotta be REALLY nerdy to appreciate this), here's a link to the program batch file.
http://www.phatserver.net/users/chris/cage/rollcage.txt

So enough chit chat, here's some pics of the first-generation of my XJ cage, scaled to an XJ from measurements that I approximated from memory. Yes, the tubing appears square, but that's because the program just generates a generic cross-section with the equivalent section modulus of whatever round tube I'm using.

Side view:

Front view:

Top view:

Isometric:

Keep in mind that I pretty much only have to input 10 or so coordinates, and it'll automatically scale and generate all the triangulation and stuff. Pretty neat.

Part II coming up: first-generation testing

 

[vetteboy]

The first loading simulation I wanted to do was a pretty bad one: landing squarely on the driver's side A pillar top node, with loading in all 3 directions. I used 4000 pounds from each direction, which works out to about a 7000-pound vector loading at a 45* angle to each axis.

Here's the resulting Von Mises stress plot, which shows that as expected, the highest stress is at the node experiencing impact.

You can really see which members are really helping, and which members aren't really doing much. It's through this kind of process that I want to 'optimize' the cage so as to minimize weight and materials. It's also good to know that with a 7000-pound impact (the XJ only weighs ~4000), I'm still well within the structural limits of the material.

The next thing to consider is displacement of the members of the cage. Here's a vector sum displacement plot to show how much it moves under impact, which is exaggerated to highlight movement:

According to these results, the most deflection experienced by any point in the cage is right around 1/32". Good deal. It'll also generate nifty videos showing the 'impact' to the cage:
http://www.phatserver.net/users/chris/cage/rollcage.avi

The next loading scenario I tested was a side-roll, landing on the top of the driver's side A, B, and C pillars.

Each arrow there is 4000 pounds, so there's a vector at 45* to the pillars (equivalent to landing ONLY on the edge of the roof) of about 5500 pounds.

Stress result:

The stress is slightly higher in this situation (so is the total loading), and you can see how more members in the cage are helping to distribute it.

Displacement result:

Here, even with the greater and different loading, the most deflection is still only less than 1/16", and again at the least supported node in the cage, the A-pillar top.

The next step then becomes optimizing the cage - are these loadings sufficient? Is the factor of safety sufficient? Are there members that can be removed, or members that should be added? Right now, it's using about 103.5 feet of tube, which I'd like to reduce if possible.

*A note about the constraints here: The cage has the shape it does, because that's the profile the XJ will have after this project is done. I'm transplanting an MJ tailgate into the rear, which is why I've omitted the bottom brace across the D-pillars - the tailgate will help transmit this load, and while not as effective as a rigid tube member, I think it's enough that I can afford to leave that member out for the sake of convenience in using the back area.

 

[AzRick]

Huh?


























































































Seriously.....get rid of the X in the roof.....you'll just hit your head on it.

Then, step away from the computer and pick up your welder and start fabbing.


Rick

 

[ckh550]

you forgot the flux capacitor for your space ship.




wow, at first i thought i didn't know much about fabbing but now i just feel retarded.

Chris

 

[tkm]

I stoped at "on vacation"

 

[vetteboy]

Seriously.....get rid of the X in the roof.....you'll just hit your head on it.

Then, step away from the computer and pick up your welder and start fabbing.

I'm short.








I'm also 850 miles away from my XJ, welder, and shop...and a month away from even being able to start this. The best I can do right now is Virtual Fab. :rattle:

FWIW, from start to finish I probably spent about an hour writing that thing and making the pretty pictures. At least I know how much tube I'll need.

 

[Thayer]

That is really cool. I've never used ANSYS, but some ABAQUS for a class, so I can appreciate your simulations. I see static loading, but what about impact loads? I'm not sure the best way to do that, but maybe you can use the strain energy theory. You know the mass of the XJ and can estimate some impact velocities. In any case, keep posting simulation pics of your optimization process.

 

[CRASH]

I highly doubt you'll be able to get the "X" bracing on the b-pillars without seriously compromising rear door entry. May not matter to you, though.

 

[Matt S.]

I highly doubt you'll be able to get the "X" bracing on the b-pillars without seriously compromising rear door entry. May not matter to you, though.

I concur.. unless you are 4'11", it aint happenin. For my tall ass and a comfy seat, check where the b pillar came to with a straight X.

i could have moved it maybe 3 inches forward, but I figured I was butchering the back seat anyways, might as well give myself "cab" room.


I think you have to many Xs as well. its just not reasonable to have them where you do.

 

[CRASH]

Also, I'd like to know what your assumptions are for mounting the cage to the existing chassis. I think I have a very stiff mounting system, but I know for a fact that it deflects a bit under load. Not a bad thing, considering the relative stiffness of the OEM chassis.

Basically, with a full body XJ, I work with the assumption that a single hard roll will destroy the chassis. I like a cage that does more for chassis stiffness, with roll protection as a secondary design consideration. Remember, we are dealing with a full chassis vehicle with a roof structure designed to withstand NHTSA rollover standards. The cage is just icing on the cake. Your cage design is one that would work for a cut-up body or buggy that will see a lot of time leaning tube-on-rock, but is way overkill for a full body rig that has inherent rollover protection.


So, for to improve chassis stiffness, you really need to tie into and through the firewall to the front spring mounts and then down to the bumper mounts. Take some of the tube you are using for triangulation in the rear and move it to the front of the rig.

 

[vetteboy]

Also, I'd like to know what your assumptions are for mounting the cage to the existing chassis. I think I have a very stiff mounting system, but I know for a fact that it deflects a bit under load. Not a bad thing, considering the relative stiffness of the OEM chassis.

Basically, with a full body XJ, I work with the assumption that a single hard roll will destroy the chassis. I like a cage that does more for chassis stiffness, with roll protection as a secondary design consideration. Remember, we are dealing with a full chassis vehicle with a roof structure designed to withstand NHTSA rollover standards. The cage is just icing on the cake. Your cage design is one that would work for a cut-up body or buggy that will see a lot of time leaning tube-on-rock, but is way overkill for a full body rig that has inherent rollover protection.


So, for to improve chassis stiffness, you really need to tie into and through the firewall to the front spring mounts and then down to the bumper mounts. Take some of the tube you are using for triangulation in the rear and move it to the front of the rig.

The assumptions for mounting are first-order in that I'm assuming it's rigid. I realize the floor will deflect under load, but I've got no data or numbers to figure that into the simulation unfortunately. Like, I don't know how much it'll move (besides "a little" or "not too much"), so for now I'm content with establishing a good factor of safety to make up for it. If anyone has any numbers ("I rolled off a 5-foot ledge and landed on the passenger a-pillar and the floor moved up half a foot") that'd be good though.

It's not gonna be a full-body XJ anymore, that's also part of the problem. Chopping is gonna be similar to Kaczman's rig. I'm not going to have a sealed tailgate on the back for rigidity in that way, so the cage has to take up both flexure and torsion in the chassis that I'm losing from eliminating the tailgate. That's the reason for the X on the back, and also because I'll most likely be mounting the spare tire there.

As far as chassis stiffness, I've currently got 3x3x1/4" angle running the length of the 'frame rails' from the LCA mount to the rear spring hanger. I recently cut the LCA mount off because I went to a radius arm setup, so I'm extending the 3x3 as far forward as possible, and ideally tieing it all in near the steering box and panhard mount area up front. The rear is getting 4-linked so the 3x3 will also continue back to where the rear coil bucket will be, and possibly tie into the bumper and hitch area. With that much reinforcing and also two large crossmembers and rigid bumpers front and rear, there's a lot of help for the chassis. Once I get to chopping, I'm going to further examine running more tube through the firewall and forward.

As far as the B-pillar hoop - I'm not planning on keeping it planar as shown in the diagram. The hoop will stay planar all the way around, but the harness mount behind the front seats will be bent around the seats, and any cross-bracing will also be bent to match. Of course, like I said, some of this may change when I actually get down to it, but it's nice to dream.

 

[Weasel]

for the loading figure anywhere from 5-30g's and a violent roll will easily get above 30 g's. If your apply the forces as point loads over a round structure you anlaysis isn't going to be accurate. Not familair with your program but with ABAQUS it's recommend to model it as a beam or truss. Then you can use a force in a direction otherwise you need to partion a section or area and apply the equivliant force per unit area.

 

[BrettM]

I concur.. unless you are 4'11", it aint happenin. For my tall ass and a comfy seat, check where the b pillar came to with a straight X.

I've sat in your seats, you've got the gangsta lean goin on

 

[Kaczman]

I think adding tubing to connect the bases of the A to B, and B to C pillars can help minimize unibody flex, especially when tied to a subframe. No calculations or formulas involved, just my experience.

-Jon

 

[Matt S.]

I've sat in your seats, you've got the gangsta lean goin on

haha... thats what everyone else says... but with those 90* seats, got no choice. PLus i am going for the playa status. Since Jes wont give me "gettin it" status, i go for what i can.

 

[BrettM]

"Jes wont give me" ?!?!

but you said he's your bitch... or were you wrong?

 

[Matt S.]

"Jes wont give me" ?!?!

but you said he's your bitch... or were you wrong?

oh... well... um... you see...


he is still my bitch

 

[Bent]

I think you are way over rating the tail gate's ability to provide a sheer panel. The pivot cups allow for quite a bit of up and down movement and the top latch does only a marginal job of preventing the sides of the bed from flapping in the wind. I think it was ~ the mid '70s Datsun was running a cam latch at both ends of the gate; these would keep the tail in place a bit more.

That said, cool software!

 

[vetteboy]

I think you are way over rating the tail gate's ability to provide a sheer panel. The pivot cups allow for quite a bit of up and down movement and the top latch does only a marginal job of preventing the sides of the bed from flapping in the wind. I think it was ~ the mid '70s Datsun was running a cam latch at both ends of the gate; these would keep the tail in place a bit more.

That said, cool software!

Yeah, it's certainly not gonna be as good as a full structure back there. That's why I wanted to keep all the X braces - to make up for that flex somehow. I'm still working out the details on how to actually mount the damn thing in there anyway.

It appears that the MJ tailgate is pretty much a GM piece anyway, except for the jeep embossing - is this correct? The latch and pivot cups all look GM sourced from a Sonoma or S-10 pickup of that era.

 

[Captain Ron]

I think you are way over rating the tail gate's ability to provide a sheer panel.
...

Maybe it should be a queer panel.

--ron