Chapter 4 Hybrid Substructure Assembly Techniques for Efficient and Robust Optimization of Additional Structures in Late Phase NVH Design: A Comparison Benjamin Kammermeier, Johannes Mayet, and Daniel J. Rixen Abstract In certain circumstances, not all desired NVH properties of a given mechanical structure, e.g. a vehicle, are satisfied at the end of a development process. In this situation, NVH properties of an existing structure must be improved while extensive changes of this structure are not practicable. Consequently, additional components such as mass dampers are included to improve the NVH properties. The arising task is to determine the optimal configuration of these additional components. If one assumes that no valid or accurate simulation model of the underlying structure exists, a hybrid substructuring approach is essential. The existing structure is measured at the required positions, the additional structures are modeled virtually, subsequently they are combined to a hybrid assembly. The optimization includes the repeated evaluation of such an hybrid assembly. In this contribution two major strategies are regarded: frequency based substructuring (FBS) and state-space substructuring (SSS). The possibly large number of evaluations imposes a greater demand on the computational efficiency compared to onetime assemblies. Furthermore, properties concerning the robustness towards measurement noise of the assembly technique play an important role. Based on a common notation for both assembly techniques, the relevant properties—efficiency and robustness—are compared on a numerical example. Keywords Hybrid substructuring · Frequency-based substructuring · State-space substructuring · System identification · Frequency response estimation 4.1 Related Work and Outline The paradigm of hybrid substructuring is straightforward: the assembled substructures are represented by models which are obtained from both—measurement and numerical modeling. In the measurement process, time series of inputs and outputs, e.g. forces and accelerations, are recorded. The process to obtain a model from measured data is called system identification. Besides system identification, numerical methods, e.g. FEM, can be used to derive a model. Therefore, the results of both processes will be referred to as identified and numerical model. In this contribution, a certain practical example for hybrid substructuring, rooted in vehicle NHV design, is anticipated: determine one (or more) (sub)optimal configuration(s) of additional substructures on a vehicle for a given objective. The vehicle can be measured and identified; the additional structures are available as numerical models. Such a hybrid substructuring task can be achieved by two strategies— Frequency-Based Substructuring (FBS) and State-Space Substructuring (SSS). The FBS technique was widely used during the last few years. One application case which is often considered in the NVH context is the Transfer Path Analysis. The framework set up in [18] is foremost based on FBS. A general framework for substructuring, including FBS is given in [8]. In case of FBS, the assembly process is achieved in the frequency domain. The B. Kammermeier () Faculty of Mechanical Engineering, Technical University of Munich, Garching, Germany Forschungs- und Innovationszentrum FIZ, BMW Group, München, Germany e-mail: benjamin.kammermeier@tum.de; Benjamin.Kammermeier@bmw.de J. Mayet Forschungs- und Innovationszentrum FIZ, BMW Group, München, Germany e-mail: Johannes.Mayet@bmw.de D. J. Rixen Faculty of Mechanical Engineering, Technical University of Munich, Garching, Germany e-mail: rixen@tum.de © Society for Experimental Mechanics, Inc. 2020 A. Linderholt et al., Dynamic Substructures, Volume 4, Conference Proceedings of the Society for Experimental Mechanics Series, https://doi.org/10.1007/978-3-030-12184-6_4 35