I have to agree with MoverZ - it's not that CAESAR II doesn't match the "real world", it's probably that your MODEL doesn't match the real world (or there is a disconnect between program operation and your expectations).
Take node 265 for example. This is a node where a hanger is to be selected. The program picks a constant effort support so it has the "exact" load to balance the deadweight and that load (H) is included in your load case "W+P1+H". For this load case in CAESAR II, the Y deflection of node 265 is zero. That's what you and I should expect. But your reported, from the field, installed position of node 265 is -80 mm.
So, is the hanger out there in field, properly connected? What load is displayed on that hanger? Is it the value predicted by CAESAR II?

Your solution is model specific.
Please report back when you have discovered your correction.