Opinions on CFL condition for implicit solvers and incompressible flows

boreal1 · December 17, 2023, 07:14

Hi

I know there have been a few discussions about the CFL condition, but they don't answer my question.
I'm running a simulation with a fine mesh in some parts. I usually set the max. Courant number to 5 and then the mean during the simulation is lower than 0.5. There are only very few points (20 or so and the mesh has 6981985 cells) in the mesh which reach higher Courant numbers. The issue is that these few points cause my simulations to progress very slowly. Now I set my max. Co number to 20 and the computation time decreased massivly. The residuals stay constant and don't start oscillating. I believe the simulation still gives valid results. Do you think these few points have a large impact on the results.
Do you have any sources that strengthen the hypothesis that it does not impact the results? I haven't been successful so far with finding some.
Just for your information: I'm using OpenFoam and the PIMPLE algorithm.

I'm excited to hear your answers!

FMDenaro · December 17, 2023, 13:03

Quote:

Originally Posted by boreal1

Hi

I know there have been a few discussions about the CFL condition, but they don't answer my question.
I'm running a simulation with a fine mesh in some parts. I usually set the max. Courant number to 5 and then the mean during the simulation is lower than 0.5. There are only very few points (20 or so and the mesh has 6981985 cells) in the mesh which reach higher Courant numbers. The issue is that these few points cause my simulations to progress very slowly. Now I set my max. Co number to 20 and the computation time decreased massivly. The residuals stay constant and don't start oscillating. I believe the simulation still gives valid results. Do you think these few points have a large impact on the results.
Do you have any sources that strengthen the hypothesis that it does not impact the results? I haven't been successful so far with finding some.
Just for your information: I'm using OpenFoam and the PIMPLE algorithm.

I'm excited to hear your answers!

If you are solving only for the steady and your residuals (that is the time-derivative) are constant but high, why the solution is acceptable?

LuckyTran · December 17, 2023, 14:47

But you can see for yourself the solution with low and high CFL because you have the solution. You know the impact. Just look at it.

December 17, 2023, 07:14	Opinions on CFL condition for implicit solvers and incompressible flows	#1
boreal1 New Member Join Date: Nov 2023 Posts: 4 Rep Power: 2	Hi I know there have been a few discussions about the CFL condition, but they don't answer my question. I'm running a simulation with a fine mesh in some parts. I usually set the max. Courant number to 5 and then the mean during the simulation is lower than 0.5. There are only very few points (20 or so and the mesh has 6981985 cells) in the mesh which reach higher Courant numbers. The issue is that these few points cause my simulations to progress very slowly. Now I set my max. Co number to 20 and the computation time decreased massivly. The residuals stay constant and don't start oscillating. I believe the simulation still gives valid results. Do you think these few points have a large impact on the results. Do you have any sources that strengthen the hypothesis that it does not impact the results? I haven't been successful so far with finding some. Just for your information: I'm using OpenFoam and the PIMPLE algorithm. I'm excited to hear your answers!

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
SU2 incompressible laminar flow	mkulcsar	SU2	3	April 30, 2023 23:59
QGDsolver - OpenFOAM computational framework for fluid flows based on regularized equ	mkraposhin	OpenFOAM Announcements from Other Sources	7	September 13, 2021 04:09
[Other] varRhoTurbVOF: a new set of VOF solvers for turbulent isothermal multiphase flows	Wenyuan	OpenFOAM Community Contributions	0	September 11, 2019 06:25
Classification and Selection of Solvers for low subsonic flows in CFD	udaych17	FLUENT	1	July 24, 2019 10:39
static and hydrostatic pressure in incompressible solvers	sinatahmooresi	OpenFOAM Running, Solving & CFD	1	February 22, 2017 13:16

December 17, 2023, 14:47		#3
LuckyTran Senior Member Lucky Join Date: Apr 2011 Location: Orlando, FL USA Posts: 5,747 Rep Power: 66	But you can see for yourself the solution with low and high CFL because you have the solution. You know the impact. Just look at it.