Just take note that FEA also has some mathematical limitation which is not a closed form solution unlike direct stiffness method being employed by most of pipe stress programs. In FEA, you must check if you have enough mesh density & check if your solution converged to a reasonable value. But take note of the sharp corners, which causes singularity & stress going to infinity. Normally, you can do a stress convergence study at the toe of the weld which is mostly where fatigue failure occurs. This is useful where there is a potential for fatigue. Some consultants have extensive fatigue testing to validate FEA software.

Any other opinion is highly appreciated.

Cheers!!