bruinbro

What is the radius uses on the rollhoop bends? 3 inches. It's somewhat cosmetic as I really intend on doing a simple beam model first.

What are you using for material weight/composition? (assuming 1.5x.120 DOM mild steel?) 0.29 lb/cu in

What are you assuming for Weld strength and weight? See the answer to the first question.

Also, what are you using for constraints on the end of each mounting point? When I build the model, I will first clamp them in 6 DOF, then after seeing how that behaves ditch the clamps and tie them together with beam elements and play with stiffnesses and end releases.

Thanks.

95m3racer

[/QUOTE]

What are the details for the Mesh you're using? I guess i could also ask how powerful your computer is?

Also, in the cage model you last poted you might want to put some of the normal extra tube truss/gussets that are usually put in. For example from the B pillar up to the top of the door hoop, and from the front roof crossbrace to each part of the A pillar bars.

The only real issue with trying to use these as comparitive tests is that the actual results will vary so much, that I don't really see any data being conclusive to apply to any structural component of the cage...would not be much better than a edumacated engineering guesstimate of what will happen.

I just can't see the direct relationship between what people will actually be welding to what you are modelling.

You can model tons of different structures, but if the structure is only as strong as the weakest link, and you are not modelling any of the potential weak links, it doesn't prove anything to me, other than what the stronger internal structure is, which is not very useful to building a cage at the caliber we're disussing.

bruinbro

It's a step by step process. First you get a crude model to tell you where the issues are and at the same time play with sensitivities to things like joint stiffness and restraint conditions. From there you develop a more refined model and check it against the crude one to see that the refinements are doing what you think they should be doing. Then you can build a comprehensive model that has all the nitty gritty details that should give definitive answers to the problems at hand. The fidelity of the modelling and boundary conditions is somewhat dependant upon what you are trying to find out. I disagree that there won't be useful info coming out of this. What George and others must understand is that the results will reflect what the design will do with the assumptions as to the boundary conditions. I can model a very accurate cage with various manufacturing variables taken into account. It is the underlying restraints and loads that make this less than a comprehensive excersize.

BTW, I have a dual processor Dell 530 workstation with twin 1.8 GHz processors, 4 GB ram and a Quadro 900 graphics card. Would like to upgrade to 3.2 GHz processors some time.

Bro

ETA: Haven't meshed anything yet. Wanted to see if the solid is what George had in mind.

Geo

Indeed. What Steve sometimes forgets is that not everyone w/o an engineering degree is a dummy. I, and I think at least most of us playing at home, understand the limitations of what we are doing. I don't give a woop about absolutes. I care about relative performance which is what we have been talking about all along. I don't care about small differences, but I care about orders of magnitude or near orders of magnitude which is what we saw between the X and the pyramid.

It sure is. And let me say, this has been probably the most educational thread I've participated in on any on-line forum. I thank you for your efforts. I hope others have not gotten bored.

Geo

"Normal" depends upon the class/category a cage is designed for. Since this is my cage, in an SCCA ITS car, they are purely illegal. It would however be interesting to look at the results after we see them w/o.

bruinbro

Hey George, I wasn't calling you a dummy. I was trying to say was that I'm not the one that is going to define the boundary conditions. I've read your posts here and elswhere. You ain't no dummy.

Bro

Geo

Steve, thanks! I meant the other Steve (95m3racer), not you.. Too many Steves. Sorry for the confusion.

And once again, thanks for all your effort. This has been and is fantastic.

gbaker

Ditto. All tools are valuable, even if imperfect.

That's just not fair. You have all the toys.

bruinbro

Yeah, but they are getting a little long in the tooth.

Bro

fatbillybob

I think I got lost. The summary chart did not show orders of magnitude better results of the pyramid over th flat plane but less deformation than expected similar to the flat -x. I think the way to summarize this is the flat x is lighter but more driver confining. The pyramid is heavier but gives more ingress and egress space and crash depth.

Does any of this hold up in reality? if I understand what has happened so far the load was only 12,000lbs which is 3g's from a 3000lb car. That is a bumper tap! I would think the metals and their configuation would act quite differently at more g's. Wasn't earnhardt's crash delta V inot the wall 45mph with 80g's resultant?

bruinbro

I think your summary is correct. BTW it's 4 g's , not 3 g's. 3000 b x 4 g's = 12,000 lb. The reason I arbitrarily picked 4 g's is because the material was still in the elastic reagion for the analysis. Much higher than that the stresses would have gone into the inelastic region and the results would have no comparative value. Actual g loading from a crash is very hard to simulate, especially since the restraint conditions ( the connection to the flexible car) are somewhat unknown to me.

Bro

Geo

The differences in deformation (max) was 6-8x. That is approaching an order of magnitude as I see it.

It would be interesting to see how much deformation there would be with say a series of 20g impacts. I realize the rest of the car is difficult to simulate, but again, the relative results would be interesting.

fatbillybob

Well 6-8x a very small number with totals all still under 1" of total deformation. If my body is say 8" from the bars if one bar moves .125" and another .85" I know that is 7x difference but still that really is not anything and if I understand BRO correctly with 12000lb impact that's "elastic" deformation (does that mean springs back?) too boot. If this thread proves anything it says that if you want light weight do the X. If you need more room do the nascar or pyramid and you will be just as safe. However, I still did not quite get what Bro is talking about not going over the 4g's. See how dumb I am? 12/3 = 4!! not 3. Anyway, I think we need to analyze at over 4g's because that where injury will occur and that is what we are trying to build for. Instead of being under the elastic deformation point at 4G's or 12000lbs lets be really rediculous and hit all three bar systems with a baseball. lets assume that MV squared is 50lbs. In my very simple mind that means there will be zero measurable deformation elastic or permenant. I think everyone will agree that all three bar types are equal when I throw a baseball at them. So why 12000lbs instead of 50lbs? I think you have to test more than 12000lbs before you get any useful data and see what the permenant deformation is beyond the elastic limit because it is that permenant deformation that is going to hit me when I crash and violate the integrity of the cage where I can get hurt. So....I love all the FEA and thank Bro for doing it but I still don't understand the benefit of this analysis. I know Geo is changing his mind about X vs pyramid but intuitively I really think he was still right, X is the way to go but I have no engineering ability to prove it. GEO..."come on brother" stick to your guns!

bruinbro

Fatbillybob,
If we (I) had the capabiltiy to define all of the variables that go into a realistic crash scenario including the high loading that would permanently deform or break the cage elements, we would need a much more capable computer than I have at my disposal and we would be in business doing this, not doing it as an internet excersize. The deformation results, although small, are a snap shot of the range of loads I ran to determine the elastic (springback capable) properties of the cage door designs. The relative deformations from one design to another didn't change appreciably with decreasing or increasing loads. Once the loading exceeds the yeild strength of the material, the analysis ceases to be linear (time invariant) and becomes non-linear, which then means we would have to know how the load is dissapated and what it's magnitude and direction would be as a function of time. This is waaayyy beyond what I am capable of doing, even if we knew what was happening with the load. As I ponted out in a previous post, these analyses show only comparative performance with a limited set of boundary conditions. They are not meant to be conclusive evidence of how a design will behave in a real crash, just showing trends that serve to somewhat validate or invalidate people's intuitive sense of how one design should behave as compared to another design. It's kind of like saying I know a steel rod is stiffer than an aluminum one and I don't have to analyze them both to beyond their breaking point to prove it.

Hope this helps,

Bro

Geo

I do think that running the scenarios at a somewhat higher force would at least tell us how consistent the results are. I think we all realize that this modeling cannot predict absolute performance in a crash. All of that is speculation. But at least with modeling of additional force we can make a more educated guess.

For my part, looking again at my cage and where the main hoop is placed, I am not too likely to replace the X, although I may scrap the gussets and reinforce the center with additional tubes.