I think the worst case that could be seen is a set of 20" rims on a rig. so if this is GROSSLY oversimplified, we could say that on an 8 lug 6.5" bolt pattern, each bolt is supported by a 1/8th pie slice of the wheel disc. the capacity of the pie would increase as it widens linearly with the diameter of the wheel.

so say 20" rim, forget about all the rim crap along the outside, and that's (20"-6.5")/2 = 6.75" between any bolt and the outer rim.

the section at edge is 20*Pi/8 wide = 7.85"

at 3/8" thick, the section modulus of the pie slice is 0.184-ish.

on 60 ksi steel, that should yield at about 11 k-in.

say we toss 8k lbs on the wheel for easy math, each lug would take 1k. so 1k at 6.75" is 6.75 k-in. that's about 38% safe.

I'd use a saftey factor of 3, personally, but this is a really conservative way to run the numbers and doesn't account for any internal plate action.

if you re-run the numbers for 1/4" plate, the plate fails by about 12% on a 15" rim and about 28% on a 20" rim.

somebody please check my numbers. I've been in a slump lately doing "duh" work.