Just want to add some points on "why we model trunnion pipe as well"

Modeling trunnion pipe sometimes does not require any special reason.
Most of the time we just place Y restraint at support points and leave it as it is. Then we run the program, review the stress, loads etc.
When we encounter problems like large load on nozzles, high stresses at bends and high loads at support points. Then thats the time you do some tricks to your model by modeling the trunnion itself for your added flexibility and to make the system more real!

By the way, when we say more real doesn't mean its better, It's just a matter of being less conservative.

Cheers,
Say