Discussion of Model Reduction and Reservation

A large structural model has to be reduced to a manageable scale to facilitate its feasible use for damage identification and structural health monitoring. The issues aroused from traditional methods for preserving force transfer paths and model connectivity are discussed along with the requirement of eigen pair accuracy, namely, four model reduction methods, Guyan(static), Kuhar/Paz(dynamic), Iterated improved reduced system, and system equivalent reduction expansion process are extensively discussed in the paper. The balance between model reduction and fidelity is analyzed with constrained system matrices updating. A five degrees-of-freedom system is employed for demonstrating associated issues and the new approaches for model reduction. It is demonstrated that certain kind of model connectivity has to be sacrificed for achieving model fidelity.