Hearing your simulation
A customer had a request for a tutorial on acoustics. Turns out that we had one from ANSYS version 14.5. Went through it to make sure that it still worked as expected and mostly it did. So to quote the tutorial:
This tutorial demonstrates how to do the following:
- Perform axisymmetric simulation of a steady-state supersonic jet flow using realizable k-e turbulence model with model constants modified for free jet flows
- Calculate unsteady flow and acoustic near field by direct CAA using unsteady realizable k-e turbulence model
- Save acoustic data at nearfield microphone location for further spectral analysis
- Postprocess flowfield and aeroacoustic results
The microphones were situated like this:
Shortening the setup, we get:
Set up the case with k-epsilon and standard wall treatment.
Set the air density to the ideal-gas formulation, viscosity to Sutherland Three coefficient Method.
Operating pressure was set to 0 Pa
Gauge total pressure on the pressure inlet was set to 242469 Pa and the Supersonic/initial gauge pressure was set to 127360 Pa.
Gauge pressure on the pressure outlet was set to 100000 Pa.
Initialized and ran the steady state simulation.
Next we used the steady simulation to initialize the transient simulation, but before that we needed a custom field function that we’d record to file and plot at both microphone 1 and 2. The function was just a simple p – 100000.
Ran the simulation and ended up with the following results:
Here we can see shock diamonds in the jet, which was a nice surprise.
Sadly the recording frequency makes it look like the pressure waves are travelling backwards, towards the source, but they are really travelling outwards and they are getting picked up by the microphones:
Running the signals through Ffowcs-Williams & Hawkings gave this output for microphone 1:
Overall Sound Pressure Level in dB (reference pressure = 2e-05) = 137.837
Along with this plot:
The big spike at around 27 kHz will be difficult to hear, but the spike at around 7 kHz should be audible. According to the tutorial, this is the screech tone.
But with data of pressure and time, it should be possible to pop into Matlab and run soundsc() after some clever data management. Said and done, this is the sound when slowed to 1 second instead of the around 0.06 s simulation time:
That tone sure is screechy.
Blog post by Tony Eriksson, EDRMedeso 2019.
Follow us — Connect with EDRMedeso on social media