[maddy11]

I'm simulating transonic flow around RAE2822 aerofoil in rhocentralfoam and the simulation has been running for the last 44 hours. But the time has decreased to just 0.0077 s. I have given adjustTimestep as Yes. Without this flag, errors are produced.


The deltaT is 0.004 and maxCo is 0.2. But the deltaT shown in the simulation is in e-09.

[hyFoam]

Since you set adjustTimeStep to True, the maxCo entry is used instead of the fixed deltaT to calculate the simulation time step. This latter value depends on the grid size and velocity field.


When fixed to 1e-4 sec, your time step is too large by several orders of magnitude and your simulation will crash.

[maddy11]

thank you for your reply. i tried changing the deltaT and commenting the adjustTimestep but it is only running with deltaT as small as 1*e-09 sec.
as i told earlier thread that deltaT is of the same order when adjusttimestep was on.

for this simulation i am using SA turbulence model.

my grid have y+=1 and first grid spaceing off the wall around 4*e-06m.

can you please suggest what is the reason for this very small deltaT taken by the code.Is it because of the fine grid?
how can I use larger deltaT because my simulation is taking long time.


this is my controlDict file
------------------------------------------------------------------------------------------------------

application rhoCentralFoam;

startFrom latestTime;

//startTime 0.0223313653;

stopAt endTime;

endTime 1;

deltaT 0.001;

writeControl adjustableRunTime;

writeInterval 1e-03;

purgeWrite 0;

writeFormat ascii;

writePrecision 6;

writeCompression off;

timeFormat general;

timePrecision 6;

runTimeModifiable true;

adjustTimeStep yes;

maxCo 0.85;

maxDeltaT 1;
-------------------------------------------------------------------------------

[hyFoam]

Hi Maddy,


The equations you are solving are not unconditionally stable, the simulation will diverge if the time step is too large. In rhoCentralFoam, you must ensure that the maximum Courant number do not exceed 0.5. Thus, your initial maxCo of 0.2 with adjustTimeStep activated was the correct setting.



Please consult a good textbook or lecture notes on numerical methods for high-speed (compressible) flows for more details.


Cheers,
Vincent

[maddy11]

can you please tell me how can i use larger deltaT for these simulations. it is only allowing deltaT as low as 1e-09 and it is taking long time to reach at steady state

[hyFoam]

To speed up your simulation, you may

1) Use a coarser grid*
2) Use Local Time Stepping (LTS)*
3) Use more CPUs / submit your job on a HPC



but keep your maxCo to 0.2.


*might not be a satisfactory solution in your case.

[maddy11]

Hi Vince,

Thank you for ur reply. By changing the grid, I was able to get lesser time.

Maddy

[hyFoam]

Hi Maddy,


That's good to know. You could do a grid sensitivity analysis now, i.e., refining the grid by a given factor where it matters (near the wall here) and see how meaningful wall quantities change.



Cheers,
Vince

[maddy11]

Yes Vince, I can begin that now.

Maddy