[Issa]

Hi,

Please be informed that I'm using CFX to simulate airflow in a chamber. The chamber is ventilated by displacement ventilation with air velocity of 0.1 m/s and there are some heat sources in the chamber. When i used steady state simulation, i didn't get a converged solution because the flow is mainly buoyancy driven flow. So, I decided to go for transient simulation. But my question is what is the appropriate time step and total time should I use for transient simulation? Thank you

Regards,
Issa

[Issa]

Hi,

Still waiting for a reply. Kindly, I need your help

Regards,
Issa

[ghorrocks]

Time step size selection is discussed in the documentation. In summary you want to have 3-5 coefficient loops per time step. You can use an adaptive time step technique to find this quickly if you like.

Have you tried the tips here to get convergence in steady state?
http://www.cfd-online.com/Wiki/Ansys...gence_criteria

But I suspect you are right in that the result is transient which means you only real approach is a transient model. HVAC flows, especially ones with heat sources tend to have large scale transient behaviour.

[Issa]

Hi ghorrocks,

Thank you for your reply. Regarding the documentation, where i can find it? Thank you.

Regards,
Issa

[Issa]

Hi ghorrocks,

Thank you for your reply. Kindly, where i can find the documentation? I would like to know what is the required time step and the total time for transient simulation. Thank you.

Regards,
Issa

[Attesz]

In CFX help!

[kingjewel1]

Originally Posted by Issa

Hi ghorrocks,

Thank you for your reply. Kindly, where i can find the documentation? I would like to know what is the required time step and the total time for transient simulation. Thank you.

Regards,
Issa

You can set your timestep based on the air residence time within your room. Typically your Courant number should not be greater than 20.
Personally I run the simulations for a few thousand iteration on a Timescale factor of 5 then switch to physical.

[Issa]

Dear kingjewel1,

Thank you for your reply. Kindly be informed that I'm using transient simulation not steady state. For steady state i tried physical timescale but the solution didn't converged. Actually for my case the flow is buoyancy driven flow. So, it is too difficult to get converged solution using steady state simulation. For transient simulation I would like to know what is the appropriate time step and the total time for the simulation. Actually I don't know what is the residence time? So what to do?

Regards,
Issa

[ghorrocks]

I have already told you how to set time step in a transient simulation.

Time step size selection is discussed in the documentation. In summary you want to have 3-5 coefficient loops per time step. You can use an adaptive time step technique to find this quickly if you like.

The total time for the simulation is problem dependent. It probably has a large scale transient structure so you will want a few oscillations of this structure so you can get a reasonable overall picture.

Personally I run the simulations for a few thousand iteration on a Timescale factor of 5 then switch to physical.

Kingjewel - most CFX steady state simulations converge with 100-200 iterations. So why are you running thousands? Is your physics complicated?

[kingjewel1]

Originally Posted by ghorrocks

I have already told you how to set time step in a transient simulation.



The total time for the simulation is problem dependent. It probably has a large scale transient structure so you will want a few oscillations of this structure so you can get a reasonable overall picture.



Kingjewel - most CFX steady state simulations converge with 100-200 iterations. So why are you running thousands? Is your physics complicated?

Sorry I did actually mean hundreds. Thanks for the correction.