topic Re: Inlet recirculation in CFD Forum

Inlet recirculation

Anonymous — Tue, 04 Jun 2019 18:03:05 GMT

Hi All,

I am simulating smoke removal from an area. I attached pressure=0 BC to inlets, and volume flow rate BC to outlets.

Fire is modeled as resistance with transient total heat generation, and scalar = 1 attached to its surface. I also used film coefficients on all walls to simulate heat exchange.

I am running transient analysis with 3 inner iterations, and 0,5s time step.

I believe that I modeled inlets long enough but I keep getting recirculation on them, and after that divergence. Is there any type of solution for this kind of problem?

I attached several screenshots and CFD share file.

Best regards, Branimir

Re: Inlet recirculation

frederic.gaillard.7 — Tue, 04 Jun 2019 19:33:24 GMT

Turn on the flag adv5_no_dtime.

it will solve your problem

Re: Inlet recirculation

Anonymous — Tue, 04 Jun 2019 19:45:11 GMT

What does it do?

Re: Inlet recirculation

frederic.gaillard.7 — Tue, 04 Jun 2019 19:46:40 GMT

http://help.autodesk.com/view/SCDSE/2019/ENU/?guid=GUID-D64BDA7D-7753-44A4-BC80-45BEB68F7AC5

Re: Inlet recirculation

Anonymous — Sun, 09 Jun 2019 19:00:28 GMT

I tried with that flag turned on but the problem is still there...Again it diverges, and I believe because of recirculation...

Any other ideas?

Re: Inlet recirculation

frederic.gaillard.7 — Sun, 09 Jun 2019 20:01:02 GMT

Can you add picture of your convergence plot?

Your velocity field has realistic values?

What is your mass balance of your system?

Thx

Re: Inlet recirculation

Anonymous — Mon, 10 Jun 2019 09:12:58 GMT

Hi,

1. Convergence plot is attached to my first post.

2. Velocity results are realistic until divergence happens.

3. What do you mean by this? I have two p=0 inlet BC, and two 22.500 m3/h outlet BC.

Best regards, Branimir

Re: Inlet recirculation

frederic.gaillard.7 — Mon, 10 Jun 2019 13:28:37 GMT

Hello Branimir,

Normally i can simulate a quick run of your scenario and have a good appreciation of your model but right now i am in the middle of a simulation as well. You will be my eye here.

Few questions first :

Did you really turn on the flag adv5_no_dtime ? The value must be displayed at 1, in the flag manager.
I don't see divergence in your convergence plot. Only odd behavior in the early stage of the analysis (where the huge pressure drop occurs) After that Pressure & TED converge and Temperature TKE & Scalar is about to converge as well.
Can you draw a isosurface of the unrealistic velocity in your model. This will help to spot where the divergence happen.
You can find the mass balance in the ''summary file'' (check the attached picture) The mass flow out need to be equal to the mass flow in. 5% off is generally okay.
Refining the region around the volumetric heat load might be a good idea...

THX
Fred

Re: Inlet recirculation

Anonymous — Thu, 13 Jun 2019 09:18:09 GMT

Hi,

I attached screenshots according to your questions.

It is strange that mass flow in is 0 😕 (see picture 4.jpg)

Best regards, Branimir

Re: Inlet recirculation

frederic.gaillard.7 — Thu, 13 Jun 2019 13:21:24 GMT

Hello Branimir,

Thank for sharing your picture, very helpful.

Flag adv5_no_dtime
1. Unfortunately the flag is not operating, you should see the flag below the scenario 1_45000
2. Check the empty box to make it work on your analysis
Convergence Plot
1. Indeed, Something is very wrong here.
Unrealistic Velocities
1. Check the geometry carefully around this area
2. refine the mesh in this region
Mass Balance
1. Right Now you have a mass imbalance of 58% which is too much to produce a good CFD analysis. This needs to be corrected in the first place. Furthermore, I slightly change the way you set you transient BC
3. What are the purpose of the shut off conditon for the first minute ? Why does not simply start your simulation at t=60s ?? I think that cause recirculation thought your pressure BC.
Heat load Meshing Region
1. It might be a good idea but for the moment i will focus on the point 4.
Realize a state steady simulation before might be a good idea.

FRED