Put it this way:

The simple method of putting API-520 loads on the PSV is extremely conservative, but only for one dimension, and completely dismisses 1 dimension per elbow. Call this perhaps a 90% solution that overdesigns supports and misses 10% of the time.

Analyzing all the imbalanced forces all at once via Peng typically diminishes the over-conservatism of the API-520 method and brings the other dimensions into the fold. Call this a 99% solution that requires only 10 minutes of extra time to set up if you have yourself a spreadsheet handy, with the side-benefit of reducing needlessly monstrous relief loads.

Analyzing the imbalanced forces individually is also a 99% solution, but probably comes, but if combined with analyzing everything all at once, it could be a 99.9% solution. Individually is perhaps another 15 minutes to set up, but can also easily balloon into solver convergence problems.

The question becomes, what is adequate for your analysis? The more you familiarize yourself with the method and the loads that will result via the calculation, the better you can streamline, simplify, yet still meet requirements.