Learning Forum

[abd fs]

Dear expert, 

I simulate a turbulent flow, I used a sinusoidal function that depends on time as input velocity. and its via UDF.

Regarding the time step I took in an article (even if the number of meshes is a little different), the problem is that ANSYS FLUENT exceeds the time interval that I defined.

1- Does it make sense?

2- How to force the convergence of the simulation? 

3- Is there a tool under FLUENT to estimate the optimal time step?

Best regards,

SER

[Amine Ben Hadj Ali]

1/Which time interval? What are you talking about?

2/By providing good mesh, reasonable B.C's and right models. All these 3 are requirements but no warranty. It also depends on some other numerical things like linearization, linear solver solution and stiff physics handling.

3/Switch to steady-sate, pseudo-transient increase verbosity to one and you get an estimate. You can check the docu for that. It also depends on the physics.

[abd fs]

Respected Sir,

Thank you very much for providing some of your valuable time to answer my question.

1- time interval : [0, 0.2] , deta t=5e-5 , time step size 4000

2-good mesh= see attached.

-reasonable B.C's= yes 

-  right models k-w sst  

schema of solving

Best regards,

SER

 

[Amine Ben Hadj Ali]

Ortho quality is  not really good: I hope only few cells are bad.

Regarding the time interval: I assume you stopped the run and then you started again. That might explain why it runs over the 0.2 s

 

[abd fs]

Dear Amine,

yes, I stopped the calculation several times, so I do not have to stop it until the program turns off the end of the time interval.

Regarding the calculation of the time step, I have two velocity inlet, the first is a speed profile and the second is a sinusoidal function that depends on the time via a UDF.

So can I base on the small cell size where is the sinusoidal function to calculate the time step?

Best regards,

SER

[Amine Ben Hadj Ali]

Yes you can do that as an estimate.