[Course] VAMOR 1st Public Technical Course on Model Order Reduction Within Vibro-Acoustics Held at TUM
The VAMOR Doctoral Network organized its 1st Public Technical Course on Model Order Reduction (MOR) within Vibro-Acoustics on 16 March 2026 at the Technical University of Munich (TUM).
The course featured lectures by leading researchers, covering key advances in MOR for vibro-acoustic systems, including:
- Reduced-Order Modeling of Poroelastic Materials: Olivier Dazel
- ROMs for highly damped vibro-acoustic problems: from modal to Padé-based methods: Romain Rumpler
- MOR in the Boundary Element Method: The Automatic Krylov subspace Recycling (AKR) algorithm and its extensions: Dionysios Panagiotopoulos
- Consistent Parametric MOR via Matrix Interpolation for Structural Dynamics: Sebastian Resch-Schopper
- Efficient Frequency Sweeps for (Vibro-)Acoustic Problems: Steffen Marburg
The event forms part of the training and dissemination activities of the VAMOR network under the Marie Skłodowska-Curie Actions (MSCA) programme.
Lecture Videos
Reduced-Order Modeling of Poroelastic Materials
Speaker
Olivier Dazel (LAUM, UMR CNRS 6613)
Overview
Poroelastic materials (PEMs) are widely used for noise and vibration control but lead to large, frequency-dependent finite element models with high computational cost. This seminar presents reduced-order modeling strategies for such systems, starting with classical Component Mode Synthesis (CMS), whose application to dissipative biphasic media is limited by modal definition issues. Alternative approaches based on complex modes and coupled formulations are discussed, followed by a hybrid strategy that separates solid and fluid phases into substructures. This approach enables efficient model reduction using decoupled modal bases with correction terms, reducing computational cost while preserving key dynamic effects.
ROMs for highly damped vibro-acoustic problems: from modal to Padé-based methods
Speaker
Romain Rumpler (KTH Royal Institute of Technology)
Overview
Reduced-order modeling of vibro-acoustic systems with highly damped materials, such as poroelastic media, remains challenging for classical modal approaches. This talk reviews reduction strategies from modal decomposition to more general approximation frameworks, including mode selection techniques to retain the most relevant contributions. It then introduces Padé-based methods for direct approximation over frequency domains, extended to parametric settings. Finally, moment-matching methods are presented, linking Padé approximation with projection-based MOR and enabling efficient reduced models for damped vibro-acoustic problems.
MOR in the Boundary Element Method: The Automatic Krylov subspace Recycling (AKR) algorithm and its extensions
Speaker
Dionysios Panagiotopoulos (KU Leuven)
Overview
Frequency sweep analyses using the Boundary Element Method (BEM) are computationally expensive due to dense, frequency-dependent system matrices. This seminar focuses on Model Order Reduction (MOR) techniques, particularly Automatic Krylov Subspace Recycling (AKR), and its application to BEM systems. The method is extended to parametric problems, indirect Galerkin formulations, and iterative solvers. Adaptive subspace selection, non-intrusive implementations using Chebyshev approximations, and further compression via Reduced Basis methods are presented. The framework is also extended to multi-parameter settings and used to construct efficient preconditioners, enabling scalable and accurate solutions for large-scale acoustic simulations.
Consistent Parametric MOR via Matrix Interpolation for Structural Dynamics
Speaker
Sebastian Resch-Schopper (Technical University of Munich)
Overview
Finite Element Method (FEM) simulations of complex dynamical systems are computationally expensive, especially for repeated evaluations over time, frequency, or parameter variations. This talk introduces projection-based model order reduction (MOR) and its parametric extension (pMOR) for efficient simulations. A focus is placed on matrix interpolation-based pMOR, which avoids affine parameter assumptions by constructing and aligning local reduced models across the parameter space. The challenges of subspace inconsistency due to dynamic changes, mode switching, and truncation are discussed, followed by strategies such as adaptive sampling and localized approaches to improve robustness within a unified framework.
Efficient Frequency Sweeps for (Vibro-)Acoustic Problems
Speaker
Steffen Marburg (Technical University of Munich)
Overview
This talk addresses passive noise control for vibroacoustic radiation problems analyzed using finite and boundary element methods. It highlights the challenges of frequency sweep analyses, where repeated solutions at many frequencies lead to high computational cost. Classical approaches such as modal superposition and acoustic transfer vectors are limited in practical applications. Alternative strategies based on frequency interpolation and Krylov-subspace model order reduction are introduced, along with recent developments that enable more efficient evaluation of radiated sound power.