RollCage FEA

[nestockcar]

I want a rear seat in my car and I want a rollcage. I consider the rear down bars the largest potential hazard a passanger could face. Forming them to the roof I still feel puts them in too close a proximity to a passangers head. So, now I'm considering a clamp-in/bolt-in version that I can leave out for highway use and put in for track use. I know it won't be SCCA/NHRA legal, but I will not need it to be for my use.
So, to figure the performance/safety diference I performed a little FEA. And as I expected there is little diference between welding the bar and clamping it to the main hoop. Attached are deflection figures for a cage applying 300lbs to the underside of one side of the chasis (simulating a chasis load). The stress numbers were very close on each, so I am publishing the deflection #'s as they have more granularity. I plan on running a rollover simulation next. We'll see what 3500lbs on the roof does?
Please provide any insights you can think of.

Thanks,
Steve

[DeepBlue68]

Cool project! This is the kinda stuff that I wanna start doing on the side when I get out of school. I had a couple of questions for you about the FEA analysis:

- What program did you use?

- Did you apply a "fixed" constraint to all other support points of the cage besides the one that you loaded?

- How did you come up with a 300-lbf load being applied to the front corner support? What real-world chassis condition would that simulate? I would think cornering would put the whole cage in torsion, not just apply a point load to one of the supports. I'm not saying you're wrong, just wondering what your thinking behind that loading condition was...I can definitely see how doing that would simplify things if all you're trying to do is analyze the difference between no rear supports, clamped rear supports, and welded rear supports.

- Lastly, for the rollover simulation, wouldn't you need to apply a lot more than 3500 lbf to the roof? I would think that if you were gonna pick the car up and set it down nice and easy on the roof, 3500 lbf would be accurate. But to simulate a real rollover, the car is gonna get hammered with a lot more than 3500 pounds isn't it? I'd think that applying a combined vertical and horizontal force (perpendicular to the longitudinal axis of the car) to one side of the cage would probably be the most realistic situation. That would basically simulate the car flipping over and the first thing to hit the ground is right where the roof and window meet on one side of the car. Your thoughts on this?

If you have any other experience doing design/FEA on rollcages, I'd love to know more. I've always wanted to design my own cage, but was never sure exactly how to go about it. Modeling the cage and doing the actual FEA analysis isn't the problem; it's coming up with the right constraints and loads to simulate what would actually happen in real-world situations.

[Jewce]

I worked in the crash protection area for a while. Typical chassis radial accelerations for rollovers from hitting the ground are in the 5-10 g range on pavement. Your roof will absorb the impact of the first hit or too, but if you roll multiple times, the roll bar will eventially take it all.

[nestockcar]

I used Cosmosworks (from SolidWorks). It is fairly new to me, had it about 6 months and I've run a handful of studies. I did apply fixed constraints to all the other support points. The 300lbs was just a number I picked. All I'm after is to solve if a clamp-on roll bar will provide close support to a welded one (though it would be cool to run a full study on it). Solving for an impact load would require another add-in to the software which I do not have (lots of extra $), so I'm left with applying static loads. I think that at least gives me some info to the validity of the approach. I'm also a bit low on computing power so it's difficult to apply lots of loads in different directions. In fact, trying to run the roll-over sim, it crashed on me. I'll try again today.

Thanks for the input.

[silver69camaro]

I'm also a bit low on computing power so it's difficult to apply lots of loads in different directions. In fact, trying to run the roll-over sim, it crashed on me.

When designing spindles here, Cosmos would be calculating for a good 30 minutes during each test. I'd have to walk away and do something else while the computer was working away (and I've got a pretty high end PC, too).

If you do get a chance to use the full version of Cosmos, you'll really like it. Much more flexability than the lite version.

[DeepBlue68]

Has anyone used Pro-E for doing modeling? I'm gonna be graduating in 2 weeks, and the place I'm goin to work uses Pro-E, but all I've used in school is Inventor. I imagine they're all somewhat similar, and if you understand the concepts behind one, you can pick other programs up relatively quickly. We've done all of our FEA stuff in NE Nastran here at school, and from what I've been told, it's not one of the easier / more user friendly FEA programs to learn.

Matt - what kind of stuff do you do at AME? I'm goin to work in the oil industry designing downhole tools, mostly because the amount of mechanical design training I'll be able to get there is hard to beat, and I wanna learn as much as I can as fast as I can. But I really wanna get into automotive stuff at some point because it's always been my passion. Have you worked in any other industries to compare it to?

[nestockcar]

I tested the cage simulation with 4000lbs on the hoop and rollbar at a 30deg angle. The results are the welded is about 2% stronger than a clamped-on rear bar. Both are significantly (about 20%) stronger than no rear bar. So I will pursue a clamp-on design. I will provide some pics when it's complete.

Thanks for the comments.

[silver69camaro]

Matt - what kind of stuff do you do at AME? I'm goin to work in the oil industry designing downhole tools, mostly because the amount of mechanical design training I'll be able to get there is hard to beat, and I wanna learn as much as I can as fast as I can. But I really wanna get into automotive stuff at some point because it's always been my passion. Have you worked in any other industries to compare it to?

I'm in charge of all engineering here, which includes design, FEA, destructive/non destructive testing, etc etc. Design varies from hard core suspension geometry to spindle modeling to chassis design.

If you have a place to go to that will get you experience, that would be your wisest move. Working in other areas will allow you to be more creative when you're presented with a difficult design task. Myself, I've only been in the automotive engineering world, so I can't compare it to other industries.

[29od1]

I was wondering how you applied a clamped connection between your rear bars and upper main hoop? Did you input a frictional force between the two? It may be a better scenario to use a pinned joint between the two? Unless you actually run a bolt through the main hoop and rear bar clamshell connection, your rear bars are free to rotate about the main hoop and won't provide any out of plane stiffness that a weld would have. Just a thought to consider.

[nestockcar]

In the simulation the "clamp-on" bar is actually floating. I wanted a worst case, and it turns out that is only 2% different from welded, at least under this load condition. In reality I plan to use something like McMaster-Carr#6436K24 or make my own, so there will be a significant friction force. The other end will be connected to a reinforced trunk area with bolts.
Bottom line is it will not be as strong as welded, but should be a bunch better than no bar when doing a little track time.

[firebob]

Kind of where I was going when I installed mine last week.

http://im1.shutterfly.com/procserv/4...6108BcNW7ly0b8

I was just trying to avoid welding near my headliner so I welded some stubs onto the main hoop before I installed it in the car and pinned it to the down bars through two sleeves.

Robert

[Elusive R]

What is the difference between the two when you go to 40,000lbs? As suggested above, your load will actually be more like 10g's. Also, you might want to consider narrowing the scope of your FEA to the part you're actually worried about - the clamp. What happens if the clamp breaks open? I looked at the part number you referenced and I take it you are welding the clamp onto the end of the downtubes? I've never done a rollcage, but my personal opinion is that you'd be much better off welded or pinned at minimum. It's your decision in the end since you know the usage best, but if you feel you need a cage, why skimp?

Ryan