Keep in mind that by definition any unmoving anchor point will give false results, if the model is too restricting.
In reality, the floor pan forms a sagging sheet with a point load in the center (roll cage base plate in this case).
If you work the geometry the tensile loads on the floor go to infinity as the floor approaches being completely flat.
As soon as the floor sags even slightly those tensile loads drop very rapidly.
It is basically a geometry of the angle problem. If you simplify the situation and represent the sheet metal by a single infinitely rigid cable.
If you specify the "dip angle" as "D" the angle below the horizontal that a cable forms under load (measured at the anchor end), then the tension in the cable is (F/2)/(sin D).
So for a rope or cable stretched between two anchors and loaded in the middle with a force F applied in the center, the tension in the cable will be:
D=5.0 degrees, tension = 11.474 x F
D=4.0 degrees, tension = 14.336 x F
D=3.0 degrees, tension = 19.107 x F
D=2.0 degrees, tension = 28.654 x F
D=1.5 degrees, tension = 38.202 x F
D=1.2 degrees, tension = 45.750 x F
D=1.0 degrees, tension = 57.299 x F
D=0.8 degrees, tension = 71.622 x F
D=0.5 degrees, tension = 114.593 x F
D=0.3 degrees, tension = 190.987 x F
As you can see, as the floor approaches being completely flat the stress in the sheet to support a weight skyrockets until the floor stretches slightly under the load to achieve a dip angle of a few degrees. Even small loads will generate huge stresses until the floor dishes slightly to carry the load.
This is one of the hardest things to model as stresses might be substantially increased or decreased by the stiffness of structure elements far away from the point of load application. One of the reasons this simplified FEA is only a very rough approximation. There are two stresses in the floor sheet metal, a shearing stress at the perimeter of the roll bar support pad that depends only on the perimeter length of the pad, the thickness of the sheet metal and the applied load. The second is a tension stress in the sheet as it sags to carry the weight. It can fail due to either stress.
These are only theoretical values because no real material is infinitely rigid (has no stretch), so even very light loads placed in the center of a sheet of material will stretch the surface so it is never completely flat. It is physically impossible to support the load without developing some dip angle in the surface.
As a rough back of the envelope analysis:
If you apply 5000 # to a plate with a 21 inch perimeter resting on .035 thick steel the shearing stress at the edge of the base plate will be 5000/(21x.035)=6802 psi shear.
Shear strength is commonly about 75% of tensile strength, so an assumption of shear strength of about 25,000 psi - 40,000 psi would be reasonable.
That means in pure shear that 21 inch perimeter base plate would shear the sheet metal at a load of about 18,375 - 29,400 pounds applied load.
On a 5000 pound car slamming down on the roof that would be an impact of 4-5 G's would be enough to blow the base plate through the floor if the full impact was focused on a single base plate due to circumstances of the impact.
Needless to say any side loading that tended to bend the base plate over to the side turning a corner into a can opener and that number drops significantly.
Larry