CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM Programming & Development (http://www.cfd-online.com/Forums/openfoam-programming-development/)
-   -   icoFoam - high number of iterations for pressure field computation (http://www.cfd-online.com/Forums/openfoam-programming-development/116699-icofoam-high-number-iterations-pressure-field-computation.html)

aylalisa April 23, 2013 09:48

icoFoam - high number of iterations for pressure field computation
 
Dear All,

I ran an icoFoam job.
Geometry: flow channel (L=3100mm, B=250mm, H=20mm).
Like in the cavity tutorial case I've defined a moving top.

icoFoam case 1:
v_{top} = 0,6m/s
kinematic viscosity \nu=0,001m^2/s
newtonian fluid

icoFoam case2:
v_{top} = 0,009m/s
kinematic viscosity \nu=0,000001m^2/s
newtonian fluid

\Delta t=0,002s
total simulation time = 2s

Unfortunately I didn't log the simulation progress.

According to the couple of last iteration steps print on the screen there is a tremendously high number of iterations for the pressure field computation in case 2 whereas in case 1 the number of iterations for pressure (just as for velocity) is in the beginning of the computation small and finally zero.

I read following thread:
http://www.cfd-online.com/Forums/ope...-pressure.html

I've tried to understand chapter seven (Ferziger, Peric) about pressure correction but am still too much of a beginner to be in the know.

I would like to understand the high number of iteration steps for pressure field in case 2.

Can anybody help?

Aylalisa

fredo490 April 23, 2013 11:02

Just for curiosity, what is the pressure gradient in your second case ? How much is the inlet and outlet in your final solution ?

I don't know how is icoFoam but some other solver have troubles if the pressure gradient is too small. Try to change your scheme maybe.

aylalisa April 24, 2013 07:14

Hi Fredo,

Quote:

Originally Posted by fredo490 (Post 422636)
Just for curiosity, what is the pressure gradient in your second case ? How much is the inlet and outlet in your final solution ?

I don't know how is icoFoam but some other solver have troubles if the pressure gradient is too small. Try to change your scheme maybe.

I've only a moving top that should produce the flow. That is at least what I wanted to try out to get a feeling for icoFoam (and its weakness, if existing).
I've defined Inlet and outlet as walls. It's not for example like in an extruder where you get a flow produced by a given \Deltap and a moving wall (extruder screw).

Did I maybe misinterpret your question?

Aylalisa

fredo490 April 24, 2013 08:21

Try to plot the resituals to check what happen during your computation:
http://www.cfd-online.com/Forums/ope...residuals.html

aylalisa May 15, 2013 12:29

Plotting Residuals
 
Hi,

I've followed instructions of the 'how to plot residual' thread:
http://www.cfd-online.com/Forums/ope...residuals.html

The icoFoam computation runs. The residuals of Ux, Uy, Uz and p have been plot.
simulation time: 3s, delta t : 0,002s

Plot of first run (simulation time: 2s, delta t: 0.002s):
http://i.imgur.com/iQ2tSl1.png

Question 1:
Why stops the residual plot of Ux and Uz after about 150 iterations?
The simulation momentarily runs and the script 'Residual.txt' supplies, each time it is called, following message on the screen:

'warming: Skipping data file with no valid points'

Does that message have something to do with the fact the curves of Ux and Uz have already stopped beeing plot???

Have Ux and Uz already been converged?

Question 2:
Another fact, according to the entries in controlDict, confuses me:

endTime 3s
deltaT 0.002s
writeControl timeStep
writeInterval 2

Since I've set writeInterval to 2 I've expected that entries in the log file will be written after 2 timeSteps. But there is an entry for each timeStep in the log file. What does
'writeControl timeStep' actually mean?

Question 3:
Has the fact, that pressure computations needs about 1000 iterations per time step, something to do with the pressure correction method used in icoFoam?

Thanks for each reply! :o

Aylalisa

mgg July 9, 2014 08:59

Hi aylalisa,

did u find the answer of high iteration number? I am doing DNS of turbulent pipe flow. I am facing also the same problem.


All times are GMT -4. The time now is 01:46.