CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   Transient flow (

omondi November 3, 2001 05:50

Transient flow
Hi CFX Users,

I use CFX4.4 to model flow in a stirred tank. I have chosen fixed time step selected such that there is one time step per cell. However, when i run the solver, it keeps loosing information at every time step. Convergence occur only within one time step then shoots up again.

it is a problem similar to that of Keny(Posted as Unmatched boundary).

Any body with an idea?


Ribeiro November 5, 2001 14:59

Re: Transient flow

What you mean "one time step per cell"?


omondi November 6, 2001 03:12

Re: Transient flow

this is the time one node of the inner bolck (solid 1) takes to move from one node of the outer block (solid2) to the next node of the same solid2.

hope this answers your question, if not, please just shout again.


Herve November 6, 2001 07:15

Re: Transient flow
How many ietrations per time step are you using? If you use multiple iterations per time step, then you should see a series of "spikes" coming down showing that convergence is achieved within each timestep. This is a typical convergence plot for transient cases.

omondi November 6, 2001 09:15

Re: Transient flow
Hi Herve,

Thanks for taking your time to respond to this.

I use 20-100 iterations per time step. Yes, convergence takes place per time step, it then swings back to the original values, then converges a little and back again. And the oscillation continues.



Herve November 6, 2001 12:12

Re: Transient flow
Okay so what you need to worry about is the convergence within each time step. This is the important feature. So check the convergence within each time step to ensure that after n iterations your residuals have decreased by x order of magnitude and that the solution at the selected nodes does not change. If this is not satisfactory carry out more iterations. I think the oscillatory feature you observe is fairly normal for a transient run.

omondi November 7, 2001 05:39

Re: Transient flow

it oscilates between 1.0E+03 and 1.0E+01 residual values. Yesterday I used 2000 iterations but there was no improvement.

In summary again my problems are: 1.No significant convergence 2.The solver tells me, in the output file, that I need pressure boundary [details of thes are in my 1st posting].

why?? Back to you Herve!


matej forman November 7, 2001 08:12

Re: Transient flow
Hi, could you, please be more specific and write down what physical models do you use.

you should not need any boundary. at least I am doing combustion calculations in geometry with just walls. and it is working for 1d case. I didn't work out how to make it converging in 2d case yet.

Hi, Matej

omondi November 8, 2001 02:47

Re: Transient flow
Hi matej, How would you use a stirrer in a 1-D case?

If you read my first posting, you will notice that I specified some of the details I am now being asked towards the end of the string. However, I will always give the information again. -It is a stirred tank(conventionaly undestood as a cylindrical vessel, unless specified otherwise) -the vessel geometry conforms to the classical configurations (T=3D, T=H, Baffle, B=0.1T, 4 baffles) -agitator is Rushton turbine 6-bladed.

Incase this is still not clear feel free and ask for specific details of the geometry.

All times are GMT -4. The time now is 22:07.