ANSYS Fluent Aeroacoustics

Important to measure and even predict it. Hello, this is Ertan Taskin from Ozen Engineering, and in this video, I will be talking about Ansys Fluent Aeroacoustics application, starting with some introduction and providing some basics. Sound is a pressure disturbance that the human ear can hear.

Identifying and reducing noise sources would be an engineering goal for design modifications to meet emission standards. One of the benefits of this is that sound simulations give designers a better understanding of how and where the sound is generated and thus can be reduced.

Needless to say, simulation does not need prototypes and installation conditions. That's why it is mostly utilized for these kinds of applications. And furthermore, Ansys Fluent has an Aeroacoustics solver. Let's answer this question. Is sound pressure the same as static pressure? Definitely not.

Static pressure, as it is shown on the left side of the screen, is a vital element in sound production. The force exerted by the fluid on the surface or within a volume.

Whereas sound pressure refers specifically to the local pressure variation from the ambient atmospheric pressure caused by the sound wave. When a sound wave travels through a medium like air, it causes compressions and rarefactions, leading to fluctuations in pressure.

Sound pressure levels for some common sounds are tabulated here. If I whisper to somebody, it's about 30 decibels. If it is a quiet office such as what I have right now, it's about 50 decibels.

Whereas if you go to a rock concert, it's going to exceed 100 decibels, which could be definitely harmful for the hearing. Well, if you're on the rocket lounge side, you better watch it. I mean literally, just watch it, do not listen.

A typical sound pressure level versus frequency plot is shown on the screen. You will definitely notice the tonal noise as the first spike on the chart. The following spikes are the harmonics of the tonal noise.

Following that, a group of signals with reduced order spikes can be observed, which is called broadband noise. Since sound is defined as a disturbance, modeling of turbulence could be very important for acoustics.

There are multiple types of turbulence modeling methods that can be utilized in literature. In this slide, we have shown a few of them. On the left-hand side, you will see the Reynolds Average Navier-Stokes models. They are typically called RANS models.

In this case, the eddies, the turbulence eddies, are not modeled. Whereas if you go to the far side of it, you will see the direct numerical simulation. In fact, in this case, we are not modeling the turbulence, but all the eddies are explicitly resolved.

So, as you may imagine, this is a computationally very expensive approach. So, in between, there is a scale-resolving simulation, where the large eddies are actually resolved, however, the small eddies are modeled.

This is typically important because the scale-resolving simulations can capture both broadband noise and the tones, which we have shown in the previous slide. Whereas the RANS models, or unsteady RANS, will only capture the tones into large scale structures.

NCC-FT offers different aeroacoustics methods. In terms of the computational effort, in terms of the compressibility situation, as well as the accuracy, there are multiple methods available for the user to choose.

In the following application, we will use the low-compressibility, and the low-computational effort one, just for demonstration purposes. The sample application will be modeling the aerodynamics, as well as the aeroacoustics, for a propeller utilizing Ansys Fluent.

The major propeller noise components will be the thickness noise, as due to the volume displacement of the blades, or steady loading noise due to the steady forces, or the non-uniform loading due to the circumferentially non-uniform loading, non-linear noise, or quadruple noise, as well as the broadband noise.

We will start with the already developed geometry and mesh and perform the setup for the steady-state CFD solution. After following the steady-state solution, we will set up for the acoustic solution. After obtaining the acoustic solution, we will post-process it.

Here is the Fluent job of the sample application. As you can see here, the blade body, hub, and nose are already on the screen. Now I'm going to turn on the pieces one by one, so that we can see the details of the entire geometry. Let's start with the inlet and outlet.

Here is the flow domain, pretty much described with the inlets and outlets. And, the side surface is set as symmetry, and it's actually covering the entire numerical domain. And, the interfaces have two pieces. One is called the MRF, Multiple Reference Frame Interfaces.

As you can definitely see, the blade is pretty much encapsulated in this particular region. As well as the interface with the outers. They are, of course, on the other side of the same, corresponding location. So, there are two fluid regions. One is the propeller region, as shown here.

The other one is the outer region. And, the inlet boundary condition is set as the velocity inlet setting with a velocity magnitude of 2 m per second. The default turbulence intensity and the turbulent viscosity ratio values were utilized here.

So, the outlet is outlet pressure with zero pressure condition. As shown here. So just in the final note, let me show you the mesh particularly with the there you go, let me put the faces so it's going to look better. As you can see, the mesh is exactly fine and is finer on the tip of the blades.

For the turbulence k-epsilon turbulence model with a realizable option, we have utilized standard wall functions.

The selection of this is because the relatively coarse mesh on the boundary layers and the wall functions are used due to the larger cell sizes adjacent to the walls for improved accuracy.

Though boundary layer meshing could be utilized in the region in the wall region, the shear stress transport k-omega turbulence model would be a better choice in that case. For this particular example, we will consider air and with an incompressible flow for simplicity.

So, the density is a constant in that case. Note that in actual aircraft propellers, due to the high blade tip speeds, there could be compressibility efforts to be considered. But for the sake of this application, we have considered a constant density.

For the rotating domain, we will consider the rotational speed of 7770 revolution per minute on the rotational axis of the positive z direction. The rotation is pretty much defined accordingly.

Since this portion of the blade is considered in the rotating region, the rest of the blade structure is defined in the walls as the rotating walls.

For example, the wall hop is the adjacent cell zone, which is actually the MRF rotating domain, and that's why the moving wall selection is highlighted and selected for this.

Similar to the blade, when the steady-state simulation is performed, the residuals were stabilized at about a thousand iterations. And looking at the force, the thrust and the torque applied to the blade, they are also stabilized.

Looking at the calculated thrust force on this blade denoted with the red line, and the high-speed motion of the hooks in hertz for the blade, the method of calculating that is actually shown on the screen, depending on the number of blades as well as the rotational speed.

And for the speed, which I provided already, the BPF is going to be 125.7 hertz. And the higher harmonics of this tonal noise will be the integer multipliers, such as second and third and the fourth harmonics. So, let's see how the acoustics module is set. Let's go to the acoustics.

And the model we will select is the FWH. And we will disable the convective effects in this particular application. The far-field density is the same as the air density. And of course, the far-field sound speed is the speed of sound.

The number of time steps per revolution is for every rotational angle, for every angle of rotation, we're going to have the time steps for the acoustic calculation.

So that's why for one full rotation, it's going to be 307. So we're going to have 60. But we're going to perform this analysis for 16 number of revolutions. And if you wonder how we came up with this number, it's actually demonstrated here.

Utilizing the speed, as well as the time step, as I described to you before, the number of revolutions to achieve the frequency resolution of four hertz is about 60. So we're going to have 16 revolutions.

To calculate the acoustic solution, if you come to the run calculation task base, you're going to click the acoustic signals and then the compute. Once the simulation performs, we will post-process the acoustic signal.

In order to do that, we're going to go to the results and fast forward your transport. We will perform the acoustic analysis. For that purpose, we will select the sound pressure level and frequency with the x-axis. Click on the receiver. Here is the receiver. And we're going to have a signal.

And here is the plot that we should get after the acoustic analysis. As I mentioned to you, the tonal noise is actually the first spike, and the two harmonics afterwards. And then the broadband noise is pretty much indicated in this plot.

One thing that I forgot to mention before, defining the sources and the receivers. The sources are pretty much defined as the wall, blade, and the hub. And the receiver is defined with this coordinate.

So we have one receiver, which is in the x-coordinate 2.3 meters away, y-coordinate 2.69 meters, particularly for the z-coordinate, one meter away from the impeller blade region. And according to that receiver, we have actually obtained this spectral analysis.

So with that application, we have seen an example on the aerospace and defense industry. But you may imagine that the acoustic analysis can be performed in many different areas, including energy and power, automotive, electronics, as well as the medical. So we have a very good example here.

And this concludes this video. Thank you for watching.