2D simulation of Cross Flow Fan - Rotating region

[Pawel_M]

Hi, my name is Pawel and I'm glad to be part of this community since now:)

I'm having a difficulties with a simulating a CROSS-FLOW FAN in 2D. The results are strange. After some iterations there is high air velocity comming from the outside of the model which is kind of weird.

My initial approach was:

mesh
regions

- model ash shown in the picture, divided into sections in order to have better control during meshing
- separated inlet and outlet with boundary conditions set to 0 Pa pressure
- SST k-omega
- ADV5
- manual mesh (pic)
- uniform mesh around Rotating Region (pic)
- uniform mesh at RR
- 6 mesh enhancement layers
- blades tips have 2x more edge points than the rest of the blade ( picture)
- 50 iterations ramp-up speed  picture)
- time-step size chosen so that 1 mesh element pass each other over each time step

RR speed

Results:

results

Note the crazy velocity magnitude. Real measurements show that air velocity at the outlet of the fan is around 7-8 m/s.
Y+ value is max 2.3

Other approaches were also a failure. Here are the other setups:
1. 6 mesh enhancement layers (initial setup)
2. 10 mesh enhancement layers
3. 6 mesh enhancement layers with time-step 2x smaller
4. 10 mesh enhancement layers with time-step 2x smaller
5. 6 mesh enhancement layers with 2x thicker mesh around the RR and time-step accordingly set up .
6. 6 mesh enhancement layers with 2x thicker mesh around the RR with time-step 2x smaller than in point 5.

Which ended with the same result as in the initial setup.

Could you tell me if my approach is correct?  What can cause such behavior of air at the inlet?

Greetings Pawel

[David.Short.]

Hi Pawel,

Setup looks great except for maybe the inlet and outlet geometries. Is there any way you could extend the inlet and outlet and use straight rather than curved surfaces? What does it look like in reality?

All the best,

David

[Pawel_M]

Hi David

Thank you for your fast reply! In experimental study the fan was lying on the table. I have divided inlet and outlet just to be sure that there is no flow re-circulation that could cause any other problems.

As you suggested I have extended inlet and used straight line. I have also joined inlet and outlet and doubled the volume of the air. Results remain the same as in the previous scenarios. I have also investigated other designs -with no change in results.

Results and other designs:

With extended air volume, high speed air velocity comes in the model respectively later in time.  Having greater air volume could be a "solution. However what if I would like to compute greater amount of time steps.

High air velocity comes in the model exactly when the air from the outlet escapes out of the model.
It's because of the mass conservation law but why is it happening so aggressive, am I right?

Best regards
Pawel

[David.Short.]

Hi Pawel,

I was thinking more along the lines of this:

However the length of the inlet boundary may also cause an issue due to pressure variation along it causing recirculation so maybe also try something like:

Let me know how you get on.

All the best,

David

[Pawel_M]

Hello David

Actually the fan blows the air in other direction so I have created the extended inlet on the proper side. Unfortunately the results are the same as in previous runs.

ad

On the second picture rotating region is all blue because I had to recover data.

I am thinking if I should do more mesh-dependent and time-step designs. On the other hand the list of analysis I mentioned in my first post should be enough.

What is worth mentioning that conception with half-circle air volume was present in scientific articles (however another software was used).

All the best

Pawel

[David.Short. · Accepted solution]

Hi Pawel,

Can you try enabling the adv5_no_dtime flag (provided you are using ADV5, which is recommended!!) this can help avoid suspect behavior near the inlet.

Let me know how you get on.

David

[Pawel_M]

Thank you very much David!

Works like a charm! I had to use smaller timestep but overall its working very good!

Let me check some other designs because I might have some further questions, but I will accept this as solution as soon as possible 🙂 

Thank you again

Best regards

Pawel