Roll Cage Base Plate FEA?

[johnneilson]

I did some work last night on it and ran into a huge issue. This is the problem with computer generated models.
The issue I ran into was how to acurately anchor the floor plate to show the deformation in real world terms, meaning that the attached structure has to deform also. You would not ever think about just placing a flat foot on the floor without some sort of doubler/structure under it.

I will run a couple with different anchors and you can make up your mind as to the results.
I will stick with my earlier comment, cage design and integration/attachment is far more important than anything else.

John

[Saltfever]

Edit . . . Sometimes stronger is actually achieved by allowing some planned flex in the design structure, so no single point overloads and fails completely. You also want it to fail gracefully which is what skip welds and plug welds would give you.

Agree, yield or elongation is far better than a failure. That is the essence of my concern and how to model it in a manner that promotes helpful discussions with tech. Stitch welding is preferred when welding outside gussets on the cage (pg.25). I have seen the base plate either continuously welded or stitch welded on cars tech-approved for >200mph. However, that would be interesting to iterate in FEA.

[Saltfever]

The issue I ran into was how to accurately anchor the floor plate to show the deformation in real world terms, meaning that the attached structure has to deform also. You would not ever think about just placing a flat foot on the floor without some sort of doubler/structure under it.

John, sorry you are running into trouble. Are you saying your FEA can’t transfer the load path through a weld to the sheet metal? Do we need to give you a solid model from something like Solidworks or Inventor?

On the few cars I have looked at none . . . repeat none . . . use a doubler between the ¼” base-plate and the thin sheet metal! You can see my concerns and why I need clever people to look at this. 

[Interested Observer]

The issue that John Neilson is referring to in “how to accurately anchor the floor plate” is one that affects any FEA model.  The model requires that the “floor plate” be restrained in some manner to keep it from flying off in space when a load is applied to it.  Should he just extend the floor plate an inch or so beyond the cage baseplate, and fix the edge there, or should it be 3” or  10” or what?  No matter what you do, you will get results for that particular model and those results would not necessarily be indicative of the result obtained for a different model.  You eventually come around to the conclusion that to really know the answer one needs to model a substantial part of the belly pan, at least to the extent that it reaches some a more structural portion of the frame.

And when he talks about having a doubler “under it”, he clearly means under the floor pan, not the cage baseplate.

[Interested Observer]

Last summer I again talked to both Steve and Lee individually trying to get a better idea of the technical thinking for the increase in thickness. Both of them are supportive but as you know neither of them could share any SCTA analysis or the causative factor leading to the change.

I would venture to say that SCTA doesn’t divulge their “technical thinking” because there essentially is none.  To summarily go from 1/8” to 1/4” baseplates without the sort of evaluation that this thread is attempting to accomplish shows that there has been no real engineering done by them on the problem.  As will be seen, it is not a trivial problem, and the 1/4” solution may well have simply moved the perceived failure point and/or changed the failure mode.  If they knew it was an improvement, why would they not disclose the rationale by which the decision was made?

What probably happened is somebody looked at the result of an “incident” and noticed that an 1/8” baseplate was distorted, said “Holy crap, we better make those thicker so they don’t bend.”  End of engineering analysis.

[johnneilson]

OK, here we go. both of these show a 1 5/8"x.120 wall tube on a 1/4" plate 6"x6".
Both are 1020 cold rolled stl, and the load is a force of 5000# applied to the end of the tube.

What you will notice is the plate retained on 3 side edges shows quite a bit of stress on the edges fixed in space, this is because the plate is deforming but not allowed to physically move and follow the deformation.

On the other sheet structure the edges of the sheet are fixed and the plate attached is allowed to deofrm and follow the structure defomation.

The last criteria not addressed is the acceleration/time for the force application, here it is constant.
Also, it is possible to run many different angles of force applied, not tonight.

John

[hotrod]

I am not used to looking at this sort of display, but if I am interpreting it correctly, the upper model with the plate restrained on 3 sides is a closer analog to a base plate set in a corner with one side along a door sill and the other along a fender kick up as that would be strongly restrained against movement on two sides and free to deform on the open sides of the plate.

The other would be for a base plate in an open floor area with some minor support under the floor.

The peak shear stress is at the base of the tube where it wants to cut a biscuit out of the base plate and or floor pan if it is stiff enough to resist buckling under the applied load.

Load spreading gussets at the foot of the tube to spread the load across the base plate would seem the solution to that issue.

Larry

[Peter Jack]

To follow on Larry's point, I would think gussets oriented toward the corners of the plates but only extending one half to two thirds the distance to the corners. That would give extra support to the punching effect of the tube while not keeping the base plate rigid enough that it becomes the punch. My seat of the pants engineering on such a design would probably use 3/16" plate for the base.

I like the way this discussion is going I'm seeing a whole bunch of well thought out ideas.

Pete

[Saltfever]

On the other sheet structure the edges of the sheet are fixed and the plate attached is allowed to deofrm and follow the structure defomation.

Very nice, John. What I don't understand is there is no highly stressed sheet metal around the perimiter of the plate applying the force to the sheet metal. If the load was high enough to deform the 1/4" plate the thinner sheet metal should have yielded long before the plate deformed?

To keep this simple the plate is pressing on a piece of .035 sheet metal bounded on all sides. It would be interesting to see the sheet metal stress around the perimeter of the plate just before the plate yields or deforms. Then what force is necessary to start the sheet metal tearing. Once we see that picture the goal would be to use a thinner base plate (eg. 0.187") to see how it behaves and its influence on the sheet metal integrity.

OK, that is a lot on your plate but your help is greatly appreciated. I'll buy you breakfast at the Red Flame 

[hotrod]

To keep this simple the plate is pressing on a piece of .035 sheet metal bounded on all sides. It would be interesting to see the sheet metal stress around the perimeter of the plate just before the plate yields or deforms.

If I understand the issue with FEA is how to set up the model so it is representative of the real situation.
Perhaps you need to think about a 0.035 sheet of metal about 2 ft square bounded rigidly at its edges, with the 1/4 inch base plate positioned in the middle or toward one side, and its edge fastened to the sheet steel. Under load that sheet steel should look like a trampoline with a heavy plate sitting in the middle of it. Maximum stress would likely be at the corners of the base plate where it focused the load just before the 1/4 inch base plate started to bend.

It would probably start to cookie cutter the sheet metal at those corners and then punch through the sheet metal around the permeter of the base plate rather than the tube punching through the base plate center.

Which if true would imply that the safest base plate shape is not a square but a shape with rounded or truncated corners like a hex or a disk that would be less likely to concentrate stress at the corners.

I also much appreciate the modeling effort to facilitate the discussion.

Larry

[johnneilson]

The basic idea is to model a real world scenaro. You would never simply mount a plate flat on a unsupported piece of sheet.
The sheet model I show is .030, and it is 2 pcs bent 90° with verticals directly under the plate/tube location. While not a true model, it represents something possible. What is not shown is that the tube displacement on the sheet supported test moved about double the distance of the 3 side supported plate. This does represent the structure moving or absorbing some of the force. The pictures show very exaggerated deformation than actual, to make it easier to see small details.

You are correct about the sharp corners on the plate possibly being the stress riser and tearing the sheet.

There are many, many factors that play into the construction. I have a friend who was almost killed when the cage broke away from the chassis during a crash. The failure was not the footing plates tearing away from the sheet, instead it was the cage tubing breaking the welds from the plates. It seems that the welds were marginal and rusted. I don't remember the foot thickness, but the car was from a time when 1/4" was thought to be standard. I just confrimed the SCCA min spec is .080 thk ('09 GCR '07 spec).

John

[Peter Jack]

Unfortunately it's somewhat hard to detect if a weld has penetrated properly but in most cases where a weld is not washed out into the main metal at the edge of the weld there is going to be a lack of penetration problem. I see this rather often in pictures and it makes me shudder when I know the vehicle has made it through tech with some organization. One of the problems with the ubiquitous mig welder is that it can make a really nice looking weld that instead of penetrating the plate tends to be cast up against the plate. Three things to help get effective penetration are:
1. turn up the heat
2. be sure to clean all the mill scale off the base plate. Mill scale is a pretty effective heat barrier and encourages the casting effect.
3. use some form of manipulation, usually a slight weave which helps to stir up the weld puddle and encourages penetration.
4. again, don't be afraid to turn up the heat. If you can't handle that without blowing holes find someone who can. Cold welds can kill! 

Pete

[WOODY@DDLLC]

To keep this simple the plate is pressing on a piece of .035 sheet metal bounded on all sides. It would be interesting to see the sheet metal stress around the perimeter of the plate just before the plate yields or deforms.

If I understand the issue with FEA is how to set up the model so it is representative of the real situation.
Perhaps you need to think about a 0.035 sheet of metal about 2 ft square bounded rigidly at its edges, with the 1/4 inch base plate positioned in the middle or toward one side, and its edge fastened to the sheet steel. Under load that sheet steel should look like a trampoline with a heavy plate sitting in the middle of it. Maximum stress would likely be at the corners of the base plate where it focused the load just before the 1/4 inch base plate started to bend.

It would probably start to cookie cutter the sheet metal at those corners and then punch through the sheet metal around the permeter of the base plate rather than the tube punching through the base plate center.

Which if true would imply that the safest base plate shape is not a square but a shape with rounded or truncated corners like a hex or a disk that would be less likely to concentrate stress at the corners.

I also much appreciate the modeling effort to facilitate the discussion.

Larry

Great discussion - here's my quick look based on comments. Lots of assumptions in the setup!

This is not a simple problem when you are dealing with structures. Which is why FEA is so useful! FEA lets you see how the various ideas respond then an optimization study can be run on the better ones!

Ultimately the cage and structure would need to be modeled and then drop tests performed at various angles to see the effects of the peak loads from the potential g-forces. But you would still start with simple models like these to get your head around it all!

Bridge builders were early hot-rodders of a sort. If the bridge did not collapse you were labeled a successful bridge builder. Then you hit the cost/benefits ratios, public liabilities and you have to start engineering for success. And even after we have "figured" it out we still have bridges collapsing!

Any engineering study would have to reviewed and certified by a professional engineer before you would go into the public marketplace. After a disaster the normal response is to over-engineer it! CYA!!!

[WOODY@DDLLC]

Two more!

[panic]

The square/rectangular shape of the plate is only useful if the axes are normal to the force received (viz., presenting a long side with the edge to the vector).
A quartering impact (left front fender hits tree) converts the harmless left front corner of the plate into a shearing wedge.

Trying to estimate the effect of a formed 3-dimensional floor is pretty tough. The stiffness under the plate is asymmetrical left to right, with the outboard side almost always stiffer than the inboard side due to the proximity of a rocker, tube or other shape, where the inboard side has almost flat floor (subject to both bending and racking) unless there's a driveshaft tunnel.