Often times it is not practical to model the entire structure with one model in great enough detail to capture all of the data required. Generally, the more detail added to a model, the longer it will take to run. With the system model approach you create a coarse model of the entire system to capture the overall effects followed by one or more substructure models to capture local details. In the system model, smaller substructures are usually represented as lumped masses at their CG locations and may be attached to the main system model with rigid links or beams and springs that represent their stiffness. Reactions or the response of the system model at the substructure location are used as input loads or boundary conditions for the substructure model.
The premise of this method is that the dynamics of the substructures do not effect the overall global dynamics of the main structure as long as their mass is accounted for. In a pure sense this is not true but this method may provide a practical solution that is ‘close enough’ for some otherwise very complex problems.
The Rule-of-Thumb for making this determination is to find a boundary to cut the substructure model off the system model such that there is an order of magnitude or more (10x) change in mass or stiffness at the proposed boundary. In other words, if you have some small components attached to a much larger structure, the local dynamics of the small components will not effect the overall dynamics of the larger structure. Also, if you have a component attached with some soft mount, it may be isolated or decoupled from the main structure.
You may want to create measures of acceleration at the CG of the substructure components on the system model. The output of these measures may be used as base motion for the more detailed substructure models. Since this method ignores any coupling between the two models, you should envelope the measure output data before you use it as input to the substructure model. This should yield a more realistic and conservative result.
The following example is of a small assembly modeled as a lumped mass at its CG location on the larger structure system model. The dynamic analysis was performed on the system model with a measure of acceleration at the smaller assembly location. An enveloping curve was made from this measure data and applied as base motion input into a much more detailed system model.
