Roll Cage Base Plate FEA?

[Rex Schimmer]

I certainly appreciate the FEA work done by John and Woody, and I am sure that everyone that has posted on this subject can look at these analysis and see that there are very inherent and dangerous consequences from this design. I also think that all of the people that have posted recognize, without any FEA analysis, that the picture posted by Saltfever of the down tube welded to a 1/4 inch plate in the middle of an expanse of thing sheet metal is wrong and a disaster looking to happen but what is frightening is that this cage was built to SCTA rules and was in a car that had passed SCTA inspection at Bonneville. Whether SCTA will or can change the rule criteria for this type of cage connection is a question I don't think we can answer but I do think at all of us who have or will build cages for uni-body cars would and will not make this mistake simply because we can recognize the danger of this type of connection.

As John said "cage design and integration/attachment is far more important than anything else." These are word that we all need to keep in mind the next time we fire up the welder and tube bender. You have to get the cage to distribute the load throughout the car structure..

Rex

[Glen]

There is a big difference in vehicle types. Coupes and sedans and open cockpit  type. Coupes and sedans have the advantage of the body absorbing much of the impact acting as a crush zone where a open cockpit type vehicle the roll cage takes the impact. I have never seen a tube to plate torn away from the body with the welded and in some cases bolted as well. In open cars I have seen roll bars bent and in one case the roll bar torn off. This was not because of a plate type attachment but thin wall tubing. Maybe SCTA will address this further or clairfy
it as the reason for the thickness change.

[hotrod]

I think it is also important for folks to remember that many of these cars weigh 4000# - 5500# which is far heavier than most other motor sports environments. The only racing organization that I am aware of that has comparable vehicle weights is the desert truck racing series, and their roll cage regulations specify cage tube size and wall thickness based on GVW of the vehicle in race condition.

Larry

[Saltfever]

I don’t know what to say but a big THANK YOU to all of you that have recently joined in on this thread. All of your postings, on just about any subject, add value and are carefully thought out. There are always certain particular names I look for on this forum and you guys are coming on-board this thread. Please hang around. My apologies for the length of this post . . . I promise shorter ones to come.

John, re your post #25 I agree a real world scenario would be best but the complexity could easily overwhelm people’s generosity and sharing at this point. FEA still takes CPU horsepower and I don’t want to get so complicated that enthusiasm is killed. I don’t know the computer ET now days but I can remember starting calcs that ran for 8 hours. If CPU power if not an issue I’m still after simplicity if only to get a basic picture or understanding, to help filter through the complexity.

There is essentially no typical real world scenario. As Glenn pointed out, in the more than 500 entries last year, every one of those were different! The number of attach points on any one cage can be 6,8,10,12, or 14. All would be approved by SCTA.  The geometry of cage designs is about as individual as fingerprints! Everyone is different and none will be the same. Even a design from someone like Chris Alston who replicates roll bar kits from a NC mandrel bender will be “personalized” by the installer.

Here is the only common scenario. The rule book forces us to weld (or bolt) a ¼” plate to a piece of sheet metal approximately 0.035” thick. That would be a Vega, Monza, Camaro, etc. I have not measured Ford products. That base plate used to be only 1/8” thick but for some unknown reason was changed to ¼” thick. The purpose of this thread is to try and see how the increased thickness changed the failure mode and if some other thickness is more beneficial. We do need to see loading that produces failure. Again, for simplicity, let’s keep the load orthogonal.

To further complicate the issue there is no data on a Bonneville crash. If you watched the Danny Thompson crash video, a good engineer may be able to infer approximate G forces by measuring time and distance but that would be only one data point. Regarding the landing attitude of the car . . . who knows? In other words; on a 10 point cage did one corner hit first, 4 corners, or more than that at the same time, etc? I have seen anything you can imagine and Glenn has seen a ton more. There is no typical accident at Bonneville. That being said, I “think” G-loading could be 10, 20, or 30Gs . . .  on single or multiple members! So our model loads, most likely should be greater than 5,000lbs.

A real world model would be at least 500 plus individual scenarios and 500 solid model cage designs. We simply can’t go there. Let us focus on a generic sheet metal patch, bounded on all sides, a perpendicular force, and vary the base-plate thickness. Let’s get that picture first. Complexity can be added according to thread interest.

[Saltfever]

The simplistic model.
1.)  1 square foot of 0.035” sheet metal bounded on all 4 sides
2.)  The rule book base plate welded in the center.
3.)  A 1-5/8 x .120 wall mild steel tube welded perpendicular to the base plate
4.)  Load applied perpendicular to base plate
5.) Stress & Strain at failure.

Change No. 1.
Base plate changed to 0.187: thickness

Change No. 2.
Base plate changed to 0.125” thickness

Change No. 3
1/2”radius corners on the 0.250” base plate.

Change No. 4
1/2”radius corners on the 0.187” base plate.

Change No. 5
1/2”radius corners on the 0.125” base plate.

Change No. 6
60 degree edge bevel on half the thickness with radius corners on 0.250” plate

Change No. 7
60 degree edge bevel on half the thickness with radius corners on 0.187” plate

Change No. 8
60 degree edge bevel on half the thickness with radius corners on 0.125” plate.

[Interested Observer]

Ref. Reply #28, figure 1

Woody,
I am not convinced that the model is as representative as it should be--there seem to be some obvious peculiarities in the results.  Granted, it was a rush job.

1)  Why are the stresses at the ends of the 1/4” baseplate higher than those in the thin sheet metal that they are immediately attached to?
2) Why are the stresses at the ends of the 1/4” baseplate higher than those in the thin wall tubing that is welded to the middle of the plate (being thinner than the plate and much smaller than the length of the plate--i.e. having a much smaller cross-sectional moment of inertia) ?
3) Why are the stresses at the ends of the 1/4” baseplate higher than those in the plate itself at the area where the tube is welded to the plate?
4)  Why is the stress distribution in the thin sheet asymmetrical?
5)  Why does the stress distribution in the sheet have that odd banded pattern?
6)  Seems very peculiar that the maximum stress is at an unrestrained external upper corner of the baseplate.

7)  How about a plot of the FEA elements?
  What FEA are you using?

[WOODY@DDLLC]

IO this was just a quick & dirty in SolidWorks Simulation Express to emulate "pushing" on the pipe and wiggling the sheet metal!

As I stated there are a lot of assumptions in even a simple model like this. The .035 sheet is 24" square because it is restrained around the outside and we don't want to see that applied stiffness in our wiggle test. I applied a .035 "weld" bead around the outside of the .25" base plate so the only constraint was at the outer edges and the .035 sheet can distort under the .25 plate. If it were bolted then the contact conditions and reactions would change. Or if the plates were brazed together.

As you have deduced most of the visual anomalies aka errors are related to the coarseness of the mesh. Normally you start simple and then refine the mesh until the answer does not change much (converges) and that's your result.

It looks like Saltfever wants a load test that is straight into the sheet metal under more controlled conditions. This sounds more like what is the punch press force to poke a hole in the sheet metal.

I am stacked right now but I will try to model this in between jobs. I will try to post something simple later for feedback and then refine it.

[johnneilson]

Salt,

What you are asking for is not applicable to any real application. It is like asking which screw driver point will penetrate a paint can lid easiest.

Just to test this I modeled 2 scenarios, 1/4" plate and .080 plate.

Take a look, both fail, like the Mustang poking the cage through the floor of the car.

Best regards, John

[Saltfever]

Well John I’m not sure what to do. If you count just sedans there are probably 300-400 unique designs. The only thing they share in common is the attachment interface required by the rule book and a very thin sheet metal floor pan. Even at that there could be 8-14 attachment points.

When SCTA summarily increased the thickness of the base plate 100% I wanted to see the effect it had on only one point to keep the FEA effort manageable. The force and load path possibilities are almost infinite. However, with the knowledge you and others bring to bear an individual might use it to evaluate their unique case. What do you suggest?

What load caused the failures in your examples?

[hotrod]

More than one way to model a cage

These were models I built out of 1/4 inch polyethylene tubing and hot melt glue.
I got a lot of lessons about how a roll cage structure shares loads with these tests.

Bought the tube in rolls, cut it in 36 inch lengths threaded them on welding rods and heat soaked them in the oven on warm then let them cool still on the welding rods to straighten the tubing. Then cut and notched with an exacto knife and assembled my prototype cage structure with hot melt glue (sets fast, is flexible and tough).

Allowed me to apply force to a real object and see what tubes gave under the load. Sometimes the load sharing was surprising. The rigidity of the side hoops is strongly effected by the stiffness of the fire wall bulk head mounts. If that is not stiff a downward force on the top of the side hoop pushes the bend at the front of the door near the firewall forward. Stiffen up that front fire wall area (why I inserted the cardboard bulk head) and the structure becomes much stiffer.

side hoops were heated then formed on a cardboard template and allowed to cool to freeze their shape, same for the main hoops.

Larry

[hotrod]

More tests with stiffened bulkhead at firewall.

[hotrod]

Buckle behavior without restraint at the firewall to prevent the side hoop from kicking forward at the firewall bend.

[SteveM]

Awesome idea HotRod.  Now I have some ideas that my 10 year old son and I can work on to model the cage in my Rampage. 

It looks like you have some good scale modeling skills.  Is the polyethylene cage made to a particular scale (1/10th or so)?

Steve.

[johnneilson]

Salt,

Like I said earlier, this is not a real world application.

The force applied to the previous models was #5000.

I did another at #50 load, it lived.
This shows a #100 force and failed.

John

[hotrod]

Is the polyethylene cage made to a particular scale (1/10th or so)?

It is scaled so that the 1/4 inch tube is a scale 1 5/8 or 1 3/4 or what ever size tube I was planning on (simply can't remember now)
If you look closely at the cardboard floor plan on a couple of those pictures you will see scale dimensions.

I "think" I was using 1 mm = 1 inch, which would scale the tubes to 1.5875 diameter. It is not exact but a reasonable approximation of a scale size.
Metric scale in mm at 1 mm=1 inch works nicely because for all practical purposes it is a 1/25 scale (1 inch = 25.4 mm)
To hold some of the tubes in place long enough for the hot melt glue to set, dress maker pins worked nicely to "pin" the end of the tube in place long enough
for the hot melt to cool.

It was fun to build up the cage incrementally, and play with it then add a tube to try to limit a certain weakness and repeat. It is surprising how stiff it becomes with the addition of a couple critical tubes, and very interesting to see where the inherent weak points of a common cage. The most critical weak points in that design were always a front diagonal impact on the drivers side top corner of the windshield, and the middle of the top bar across the front windshield.

I was never really satisfied with how much less force it took to bring that front windshield corner down compared to the main hoop behind the drivers head.
That bar that runs across the top of the door and down the A post of the windshield needs to be a double tube (stacked vertically) to really stiffen it up to that front quartering top inpact.

Larry