Simulation of Sound Propagation under Background Flow Field (Impact of Airflow on Sound Propagation in High-Speed Train Tunnel)

Please login with a confirmed email address before reporting spam

Through StarCCM+, I obtained the velocity distribution data at a cross-section perpendicular to the direction of travel when the high-speed train runs at a speed of 350 km/h in the tunnel, as well as the temporal variation data of pressure (sound pressure) at two measuring points in that cross-section. This cross-section was imported into COMSOL by defining the functions and interpolating the data to generate the corresponding interpolation functions. The COMSOL model uses linear Euler and transient settings, assigning the velocity interpolation function to the model's background mean flow speed, while the sound pressure was applied to two boundaries through pressure (isentropic). After adding materials and meshing, the output time step for the study was set to range(0, 1/f/10, 0.1) (where f = 445 Hz, as the Fourier transform of the sound pressure data showed significant energy at this frequency), and the solver's time step was set to 1/60/f. Upon running the calculations, we found that after a certain time, the values across the entire surface began to increase abnormally, rising from 10^2 to 10^4 then 10^15, and so on. What could be the possible reasons for such results? How can I improve the simulation to successfully generate the sound field from a sound source under a velocity field? When the background mean flow field is set to 0, this issue does not occur; it only happens when it is not 0. I'm confused and seeking help.


2 Replies Last Post 13 nov. 2024, 06:24 UTC−5
Edgar J. Kaiser Certified Consultant

Please login with a confirmed email address before reporting spam

Posted: 2 months ago 27 sept. 2024, 07:09 UTC−4
Updated: 2 months ago 27 sept. 2024, 07:10 UTC−4

Hi,

I experienced similar issues with the time explicit method some time ago. I played a lot with the meshes and finally achieved a pretty long runtime with plausible results. However, I didn't find a specific pattern in this behavior. It was pretty much trial and error, a certain mesh worked and the next mesh didn't. It was not the finest mesh that worked best. The mapping between flow and acoustic mesh seem to play a role as well. With zero flow I didn't see issues as well. My advice is, make a small model that runs fast and play with it to eliminate the issue. Once you have something apply it to the full scale model. In 2D this issue didn't show up.

Good luck Edgar

-------------------
Edgar J. Kaiser
emPhys Physical Technology
www.emphys.com
Hi, I experienced similar issues with the time explicit method some time ago. I played a lot with the meshes and finally achieved a pretty long runtime with plausible results. However, I didn't find a specific pattern in this behavior. It was pretty much trial and error, a certain mesh worked and the next mesh didn't. It was not the finest mesh that worked best. The mapping between flow and acoustic mesh seem to play a role as well. With zero flow I didn't see issues as well. My advice is, make a small model that runs fast and play with it to eliminate the issue. Once you have something apply it to the full scale model. In 2D this issue didn't show up. Good luck Edgar

Mads Herring Jensen COMSOL Employee

Please login with a confirmed email address before reporting spam

Posted: 2 days ago 13 nov. 2024, 06:24 UTC−5
Updated: 2 days ago 13 nov. 2024, 08:40 UTC−5

Hi,

You cannot use imported data (the interpolation function) directly in the input fields for the background mean flow in the linearized convective acoustic interfaces. First of all, this dramatically slows down the solution time (a function call is made at every time step, the function is not optimized for this type of use); secondly, you will not get smooth gradients of the background flow values. The non-smooth (noisy) gradients is probably the cause of your instabilities as the gradients end up in the so called reactive terms.

To do this correctly you need a separate study where you map the interpolated data to COMSOL shape functions adding some stabilization. I suggest you contact support, they can help you.

Best regards

Mads

Hi, You cannot use imported data (the interpolation function) directly in the input fields for the background mean flow in the linearized convective acoustic interfaces. First of all, this dramatically slows down the solution time (a function call is made at every time step, the function is not optimized for this type of use); secondly, you will not get smooth gradients of the background flow values. The non-smooth (noisy) gradients is probably the cause of your instabilities as the gradients end up in the so called reactive terms. To do this correctly you need a separate study where you map the interpolated data to COMSOL shape functions adding some stabilization. I suggest you contact support, they can help you. Best regards Mads

Reply

Please read the discussion forum rules before posting.

Please log in to post a reply.

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.