I'm trying to determine the stresses on a firearm barrel during firing by running a static simulation on an assembly consisting of the barrel and a brass casing in the chamber.

This is what it looks like:

There is a small clearance between the casing and the chamber. However, when I first ran this simulation I simply made the outside diameter of the casing match the inside diameter of the chamber and used Bonded contacts. I figured that this would be okay because the thin case quickly expands after firing and is pressed firmly against the barrel. This resulted in a fairly high, but not unreasonable amount of stress on the barrel, given what I know about the materials properties of the barrel.

Afterwards, I decided to change the dimensions of the brass case to accurately reflect the size of unfired ammunition, and thus preserve the small gap that would be present between the case and chamber walls. I changed the contacts to No Penetration and ran the simulation again. This time, the stress on the barrel was much lower; only about 2/3s that of the result given by the previous method.

So my question is: which one of these results is more accurate?