The displacement and rotation computed by the API-650 routine can be included in the model. To incorporate these values, I suggest adding them to the "nozzle CNODE", since this will put them on the other side of the computed stiffnesses. Applying these values to the nozzle node defeats the benefit of computing and applying the nozzle stiffnesses.

As to the "primary versus secondary" issue, my opinion is that it depends on how the tank cycles, and how often.