LocostUSA.com

Hi Lonnie,
This isn't a direct reply to your question, but in Victoria each new chassis gets tested for torsional stiffness.
We have to meet:
1) At least 4,000 Nm/degree (~2,950 lb-ft/degree in your money)
&
2) A "somewhat" linear graph of stiffness vs distance down the chassis.
This is so that fatigue between different sections with different stiffness is avoided.
From memory, this applies to a 4 cylinder engine, and the figure becomes 6,000Nm/degree for 6s & 8s.
Cheers - Gavin.

Krepus, thanks, I'm flattered. It is a big help if you can download the files and run sketchup, that was my idea anyway. It might help you make changes etc. Also I'm hoping to get people to help with the models of parts so that this gets easier and easier.

Then it occurred to me that the files that are used for Grape give the location of every tube junction and a description of every tube. It's very easy to read. So you just need to encourage me to keep the FEA files up to date. That seems like a good idea anyway.

I've started looking at the loads our suspensions put into the frame. Starting with cornering loads on the front lower wishbone. I am using wishbones with one leg perpendicular to the chassis. So I expect all my lateral force to basically go thru that leg.

I am planning on using 1"x2" rectangular tube some places and this is one place I am considering this. So I chose to compare numbers for the rectangular tube in both orientations and also 1" square and 1.25" square because they are the commonly used choices. My frame design does not feed these loads directly into a tube junction for practical considerations of ride height and typical frame construction. My early designs did not have a flat floor, but now I am opting to start out simple.

For this test I simply chose to put the load into the tube 1/4 of the way up from the floor to the top tube. This is higher then the natural height for a Mustang II type spindle, but probably lower then a Miata front spindle. The load chosen is 1000 lbs. because that's an easy number to work with.

Deflection Stress Element
0.011" 23,000 1"x1
0.008" 15,213 1.25"x1.25"
0.009" 13,377 1"x2" ( flat side towards load )
0.004" 10,977 1"x2" ( 1" side towards load )

The first result concerns me a bit. that would be close to the yield for hot rolled tubing.

Marcus, I just returned from some "offline" holiday travel and am impressed with the progress you've made. Looks like I have a lot to catch up on.
Two items caught my eye from your latest post: Given your previously mentioned thought to bolt suspension mounting brackets thru sleeves welded into the verticle chassis tubes, wouldn't the stronger "1" side towards the load" with 2" to mount the brackets be a natural choice?

Thanks, Gavin.

Lonnie

Hello,

I've just read the entire thread and It was very intersting to see there are many other people ot there are dealing with FEA and chassis rigidity.

I'm new to this topic, so I would like to know if anybody can validate my math. From my calculation I should have about 1126 Nm/degree of torsional rigidity.

Here the brief results of my FEA on the suspension mounts on the front bulkhead:

Your math looks right to me.

if it so, my design is pretty weak...

Seems like an unrealistically high load though.

The size of the load will not affect the stiffness though....



I was somewhat appalled after my first run too. If you are using the same software we are, you can download Andrew's Locost model and that will likely make you feel better.

Then just keep at it. You can post a screenshot of your frame for comments and also study the frame during animation to find weak spots...

Good luck.

I have no idea what his chassis or model looks like so take this for what it is worth. I'm speaking from the suspension mount design standpoint and not whether that load could actually occur in a bad situation on the road because it could occur. Will the area around the input deflect significantly? Real world testing could show his deflection is way off the FEA model simply because it is a very large local load.

Essentially he is modeling for a catastrophic situation, not what his chassis will react to in a road racing/autoxing situation when you actually care about torsional rigidity.

It all goes back to knowing what you are testing. "Garbage in, garbage out" as they say. Not to say analyzing all chassis loads doesn't have some merit but checking torsional stiffness is a pretty simple test.

If you are measuring torsional stiffness, any reasonable load will give you the stiffness. It is the same theory as a linear spring - whether you apply 100 lbs or 10,000 lbs, it will have the same spring rate (lbs/in, N/m, etc).

I won't disagree that 5,000N is high. If anything I'd lower it to 250N or 500N and see if the torsional stiffness is the same - it should be close

There are at least 10-20 more loads that would have to be placed on the chassis to accurately analyze some sort of catastrophic situation. I'd be nervous if only a reasonable torque applied the shock mounts did a chassis in.

I think 5000 N is very realistic in fact..

consider the following.. Let's say upper frame width is about half track width... for sake of argument... that means at that point you see vertical corner load x 2....

with a 500 KG locost, in a corner under braking you could see 250 KG at front outside corner... now multiply that with a normal bump load of 3G, and x2 for the leverage effect I mentioned at the start, and we get... 15000 N....

Ooops...I messed up the conversion before.

I don't know what I was looking at since 5,000 N = 1,125 lbs. Why can't the rest of the world use a normal method of measuring? I've heard inch-pound-second is good.

funnily enough.. I went to high school in the US, and I do not remember having to deal with inches/pounds at all... Gosh that was too long ago..