Dynamic analyses really don’t support submodeling methodology. This is in part due to the fact that the cut boundaries are what are used in submodeling, but the cut boundaries do not contain all necessary information, and the acceleration/velocity terms inside of the submodeling region will be missed/omitted. Thus, we typically don’t use submodeling for dynamic analyses. For a bulky region, one could try using submodeling in a transient analysis at a specific time if inertial effects can be neglected and if the variation in displacements within the region can be assumed to be independent of velocity/acceleration. Since stresses are only determined by strains (displacements), we could use submodeling in such a case. However, this typically won’t work well for ‘flexible’ (non-bulky) regions used in submodeling. Other dynamic analyses are much harder to address, however, as we are dealing with combination of modes, so displacements at the cut boundary won’t reflect the actual shape (again, with a very bulky section, it may be similar to the static shape, but that is a big assumption and a specialized case). Avoid using submodeling in dynamic analyses because of these considerations.