Ceiling fan simulation


#1

Hi simscale users,

I am trying to learn simulation using simscale. This is my first project. Its a ceiling fan and I am trying to understand the flow of it. But I am not sure if I have done all the steps right. It would be of great help if someone can look into it and point out what and if there is anything wrong with the simulation and the results.
Here is the link Https://www.simscale.com/projects/pushpaldey/fannn_1/

Thanks,
Pushpal


#2

Hi @pushpaldey!

The setup looks good so far. Just make sure that the MRF zone is not too much in the proximity of the fan to avoid artifacts not representing real physics.

Just give it a try and let me know when you’re done - will have a look at the results then.

Best,

Jousef


#3

Thanks @jousefm for taking a look at the problem. By proximity you mean, I should increase the length and radius of the rotating zone so that the boundary is not in proximity of the fan? The fan has to be inside the rotating zone right? And should the material (air) be attributed to the whole mesh including rotating zone and bounding box and fan mesh?


#4

Hi @pushpaldey!

Exactly, leave some distance between the MRF zone and your fan. My colleague Darren (@1318980) had issues with this in the past simulating a wind turbine (if I can find the post I will share it with you).
Regarding the material you are right that you just need to assign every domain to air and that’s it.

Feel free to ask me if you have any questions.

Best,

Jousef


#5

Hi @jousefm

I have done the simulation for different blade configurations. The results looked pretty close to the test results(named incompressible) but in case of a modified fan blade (named flap) I found that the results were not close to the actual test results. So now, I am having doubt, if the simulation is right. Or is it that MRF is not a suitable approach for a configuration where there are small geometry involved that can affect the direction of the air. Can you please look at the results?

https://www.simscale.com/workbench/?pid=6667950998516157508#tab_1-0

Thanks,

Pushpal


#6

Hi Pushpal (@pushpaldey)!

I will have a look at that and get back to you asap.

Best,

Jousef


#7

Hi again Pushpal (@pushpaldey)!

Did you try increasing the fluid domain and see if that has an effect? And can you share the test results you are referring to? Just to get a feeling in which region we are moving. As for the MRF I would say that we could vary the MRF zone size a bit and additionally start one run with AMI and see how that affects the results, what do you think?

Cheers,

Jousef


#8

Hi @jousefm

I will try increasing the domain.
And, in actual testing what I found is that the flap makes the flow of air more conical i.e air throw is wider, which, in this case is straight down in the simulation, like a normal fan. Although the characteristics behind the flap is very close to the real results.
I will vary the MRF zone too and try. For AMI will the geometry be the same as in MRF or do I have to make boolean difference and make shells, as is done in ansys. I could not find a proper example project for AMI, so if you know about any project can you please share the details.

Best,
Pushpal


#9

Hi @pushpaldey!

A project template that might help you can be found here: AMI - Centrifugal Pump. For AMI you have to make sure that the quality of the interface is not poor because the simulation will diverge relatively fast.

If you need help feel free to tag @Get_Barried, @1318980 or me.

Cheers!

Jousef


#10

Hi @jousefm

Thanks a lot.

Best,

Pushpal


#12

Any update on your case @pushpaldey because you deleted your recent post?

Cheers!

Jousef


#13

Hi @jousefm

Yeah actually I had encountered an error but later realized that it was just a small mistake in my part, so I deleted the post.
The results seem quite accurate now. You can have a look into the results and let me know what you think? One thing I am not so sure is if the mesh refinement level difference between, lets say Mesh1 and sixblades, would have drastic difference on the results.

Best,
Pushpal


#14

No worries @pushpaldey!

That’s why we would need to perform a mesh convergence study and see if successively increasing the mesh fineness will actually result in more accurate results. What you can do is to set up a table in Excel and properly track your results. You can also work with the convergence of forces and see if they differ significantly with different setups.

Let me know how things go!

Jousef