You say "The stresses dropped dramatically at those points". I'd rather you said the CALCULATED stresses dropped. It's up to the engineer/designer/technician/user to confirm that the model reflects the real-world system and the piping code intent. I'm not wagging my finger at you; I just want to remind everyone that there's more to it than simply saying this number is less than that number.

Regarding the layout, I will not say more than this - I would be concerned about the added dynamic response if you allow those valves to open while the header is not solidly supported. Perhaps, too, the thrust loads may run down your riser and do some disturbing things below. Remember that you have the piping code issues ( usually stress) and you have your mechanical or structural issues that are independent of Code recommendations.

Don't just play the numbers game...

Regarding John's point on lift-off. I know of some (older, wiser?) engineers who automatically say lift-off is poor design. Most, however, while they appreciate that software can analyze these situations, they cannot offer a quantitative limit on what is OK and what is too much.