Parametric model order reduction without a-priori sampling for low rank changes in vibro-acoustic systems

A parametric model order reduction scheme is presented for second order systems that does not require a-priori sampling of the parameter space. The proposed scheme \revi{transfers} the parameter dependence to the throughput matrix of the dynamic system by using an auxiliary input matrix. The resulting model can thus be reduced with non-parametric model reduction techniques. Furthermore, it allows for independent low rank changes in the stiffness, damping and mass matrix of the system. It is shown that in combination with the tangential iterative rational Krylov algorithm, a high number of low rank changes can be parametrized, while keeping the reduced model accurate and of moderate size. Also, a scheme is proposed to further reduce the model size, by a frequency limited post-processing step. The methodology is illustrated with two numerical examples: A purely structural example that simulates an unknown defect by locally reducing the stiffness and damping, and a fully coupled vibro-acoustic example that demonstrates how the method can be used to simulate added mass loading, due to for instance the placement of sensors/actuators.status: publishe