CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   OpenFOAM Running, Solving & CFD (https://www.cfd-online.com/Forums/openfoam-solving/)
-   -   Can we restart the simulation from a different time in OpenFoam (https://www.cfd-online.com/Forums/openfoam-solving/80119-can-we-restart-simulation-different-time-openfoam.html)

jf_vt May 20, 2016 17:08

how iteration number is calculated in steady state calculation?
 
Hi Foamers
this is quite a silly question but I am at loss:

If I run a simulation with a simple algorithm with:

startFrom startTime;

startTime 0;

stopAt endTime;

endTime 1000;

deltaT 1;

the log file state as expected that it will run for 1000 iterations ( no convergence criteria)

When I continue the simulation ( after remapping on a finer mesh)

startFrom startTime;

startTime 1000;

stopAt endTime;

endTime 2000;

deltaT 1;

The log file then report that it will run for 2000 iterations while 1000 will be expected.

what do i miss?
Thanks
JF

aeroMatze May 20, 2016 18:06

sorry, my reply was nonsense

alexeym May 21, 2016 19:09

@jf_vt

Since your message lacks exact OpenFOAM message, I suppose it is something like:

Code:

SIMPLE: no convergence criteria found. Calculations will run for 2000  steps.
And it is a result of not quite precise wording in simpleControl class constructor:

Code:

if (residualControl_.empty())
{
    Info<< algorithmName_ << ": no convergence criteria found. "
        << "Calculations will run for " << mesh_.time().endTime().value()
        << " steps." << nl << endl;
}
...

Since your endTime is 2000, message says that it will run for 2000 steps.

Maybe it is worth reporting bug (http://bugs.openfoam.org), since for SIMPLE mesh_.time().endTime().value() can be changed to mesh_.time().endTime().value() - mesh_.time().startTime().value().

Yet simpleFoam simulations without convergence criterion is something ridiculous.

jf_vt May 21, 2016 19:21

Thanks
Sorry for the lack of accuracy in my message (As you can easily guess I am pretty new to the forum) but you get it right of course..

As for interrest running simple without convergence algorithm, I agree with you in general but at the moment I am investigating convergence speed while inserting remeshing and field mapping in the process ( this is research not sure where it leads) but in a simplified form:

I run a cas for 1000 step remesh it finer, map fields and run it for another 1000 step ... and so forth... and compare solution quality vs cputime if i run on the finer mesh all the time.

I dont need convergence test for that and this is why I stumbled on this and found it annoying in the context of the work described above.

I reported it as a (small) bug

Thanks for your reply
JF


All times are GMT -4. The time now is 22:05.