Partitioned Co-Simulation for CAD-integrated Vibroacoustic Problems in Unbounded Domains
This paper presents a modular, partitioned co-simulation framework that couples Isogeometric structural and acoustic solvers directly on CAD geometry for exterior vibroacoustic analysis, featuring novel complex-valued convergence accelerators that ensure robust and accurate solutions for both weakly and strongly coupled problems.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Modern engineering relies heavily on the ability to predict how structures behave before a single piece of metal is cut or a prototype is built. In industries ranging from automotive manufacturing to aerospace, designers must ensure that vehicles and machines do not vibrate excessively or generate unwanted noise. This challenge is known as vibroacoustics, the study of how solid structures vibrate and how those vibrations create sound waves in the surrounding air or water. To solve these problems, engineers use powerful computer simulations. However, a significant hurdle exists when trying to model the interaction between a solid object and the vast, open space around it. While computers are excellent at analyzing the vibrations of a solid part, they struggle with the infinite expanse of the surrounding fluid, such as the atmosphere or the ocean, because the sound waves travel outward forever without hitting a wall to bounce back. Traditionally, solving these problems required combining the mathematics for the solid and the fluid into one massive, rigid computer program. This approach is difficult to manage, especially when engineers want to use specialized, pre-existing software tools that were not designed to talk to each other.
A team of researchers from the Technical University of Munich and KU Leuven has developed a new way to tackle this problem that allows different software programs to work together seamlessly. Instead of forcing all the calculations into a single, monolithic block of code, they created a system where a structural solver and an acoustic solver run separately but communicate with each other in a coordinated dance of data exchange. The researchers focused on a specific type of advanced simulation that works directly with the original computer-aided design files used by engineers, preserving the exact shape of the object without needing to simplify it. Their method allows these two independent programs to pass information back and forth: the structural program tells the acoustic program how the object is moving, and the acoustic program tells the structural program how the surrounding fluid is pushing back. This back-and-forth continues until the two programs agree on a stable solution.
The core of this new approach lies in how the two programs handle the boundary where the solid meets the fluid. In many real-world designs, the computer models for the structure and the surrounding fluid are built differently, meaning their internal grids do not line up perfectly. The researchers tested several methods to translate data across these mismatched grids and found that a technique based on projecting points from one grid onto the other worked best. This ensured that the forces and movements were transferred accurately, even when the two models had different levels of detail. A critical breakthrough in their work involved how they handled the mathematical nature of sound waves. Because sound waves have both a height (amplitude) and a timing (phase), the data exchanged between the programs is complex. The team discovered that treating the real and imaginary parts of this data separately, as many older systems do, caused the simulation to become unstable and fail to converge. By developing new mathematical tools that treated the data as a unified complex quantity, they were able to stabilize the process, allowing the two programs to reach an agreement quickly and reliably.
The researchers validated their method using a series of rigorous tests. They simulated a simple metal plate vibrating in the air and found that their separate, communicating programs produced results that matched a traditional, all-in-one simulation almost perfectly. They then moved to a more difficult scenario: a plate submerged in water. Because water is much denser than air, the fluid pushes back on the structure with much greater force, creating a strong interaction that is notoriously difficult to simulate. In these tests, the new method successfully handled the intense feedback loop between the water and the plate, whereas older techniques that treated the data separately failed to find a solution. Finally, they tested a spherical shell separating an enclosed fluid from the open ocean, a scenario that mimics real-world underwater structures. The results showed that their approach could accurately predict both the loudness and the timing of the sound waves generated by the vibrating shell, matching both analytical formulas and commercial software solutions.
The findings demonstrate that it is possible to achieve high-precision results in complex engineering problems without sacrificing the flexibility of using specialized, independent software tools. By allowing different solvers to communicate effectively while preserving the exact geometry of the design, this method offers a robust alternative to the rigid, monolithic approaches of the past. The study confirms that handling the complex nature of sound waves as a single, unified entity is essential for stability, particularly when the interaction between the structure and the fluid is strong. This work provides a practical pathway for engineers to simulate the noise and vibration of future designs with greater accuracy and less computational friction, ensuring that the virtual models used in the design phase are as reliable as the physical prototypes they replace.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.