Structural dynamics analysis is an essential step in the design of high-tech mechanical systems. Complex products such as home appliances are designed in an increasingly modular fashion, combining in-house-developed products together with outsourced components. This generally requires the creation of virtual dynamics models for each component in the system, which can be assembled or “substructured” to evaluate the dynamic properties of the complete product. Efficient substructuring techniques accelerate the finite element analysis and enable the vibro-acoustic optimization of complex systems like home appliances.
Numerical substructuring methods have been well accepted over the last decades, however with increasing product complexity the question arises as to how accurately, using only numerical models, the actual behavior of the individual components of the whole system can be represented. In recent years, the structural dynamics community showed a renewed interest in the structure coupling techniques, especially in the context of experimental applications. This leads to an increase in the experimental modeling of relatively complex structures. Yet stand-alone experimental models are strongly influenced by essentially independent and often imperfect measurements. Therefore, its application to complex real-life engineering structures is often hindered by the method’s notorious sensitivity to experimental errors. The choice between numerical and experimental modeling is made for individual components based on strengths and weaknesses of the respective modeling approach. This choice presents a compromise based on which a selection between the numerical and experimental modeling is made individually for each component of the system. Hence, the research activities in last years are orientated towards the incorporation of strong suits from different modeling techniques into a single model of given component. This is referred to as a mixing of multiple equivalent models (numerical and experimental) of the selected component into a hybrid model using substructure coupling techniques. This represents a hybrid and therefore very powerful modeling methodology that would implement an “as is” description of the experimental model and the consistency associated with a numerical model.
