CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   Flow in a rectangular duct (

saisanthoshm88 July 27, 2013 05:44

Flow in a rectangular duct
1 Attachment(s)
I have some questions on the simulation of flow in a rectangular duct.

Could some one please look into a short presentation attached here to better explain my questions and provide clarifications to the questions there.

ghorrocks July 27, 2013 07:07

You will get better comments if you post your questions directly in the forum rather than as attachments.

1) Yes, extend it as you describe. But put a monitor point at the real measurement start location to get your inlet pressure value.

2) I usually do this by sensitivity analysis. But there is stuff in textbooks and the literature about development length as a function of all sorts of things.

3) Yes, you can use the velocity profile funciton to do this. This would be an approach I would certainly consider for this case.

saisanthoshm88 July 27, 2013 08:35

1 Attachment(s)
Hi Glenn,

Thank you very much for your clarifications , regarding the extraction of velocity profile from the simulation of an isolated extension, please suggest if I have to use a velocity inlet (for the extension) and a pressure outlet at atmospheric pressure (for the extension) (or) should I use a vice versa approach.

Please have a look into the attached image if my question is not clear.

ghorrocks July 28, 2013 07:29

You should arrange the boundary conditions to match what you know about the flow. If the exit is known pressure, then put the pressure bounary there. If the inlet pressure is known put it there.

There is little difference in numerical stability in both cases, providing the inlet and out are far enough away from the action to not have separations in them.

DaveyBaby October 14, 2013 06:20

Fully developed inlet conditions

It is possible to run a simulation separately with a periodic boundary condition, as opposed to an isolated extension. You could then not concern yourself with the length needed for full development as your development length could be effectively infinite.

As you mentioned, you could then extract all three velocity and all three vorticity components within a cross-sectional plane and use these to define their respective components at the inlet, while still using the atmospheric pressure boundary condition at the outlet. This would mean that any secondary flows (such as the stream-wise vortices found in the corners) would already be present.

You could run the fully developed solution just once, and use it to set the inlet velocity profile for multiple runs.

I would be interested to know if you try this and it works, or if you foresee any problems with it that I haven't!

Kind Regards

ghorrocks October 14, 2013 06:39

This works and it is a very good approach to use if it is applicable.

oj.bulmer October 14, 2013 06:52

Also, you may want to make sure you capture the secondary flows due to vortices at the four corners. In rectangular duct, typically, the general 2 equation models fail to capture these, because of their isotropic turbulence assumption. I have seen nonlinear EARSM k-eps and RSM work well with this, although the getting a convergence with both is difficult.


QIAN06 December 11, 2015 07:42

Hi saisanthoshm88,
This is interesting, did you work out it ?

All times are GMT -4. The time now is 23:56.