Time consumer of simulation time

nmc · June 15, 2023, 11:54

I am running a sediment simulation using sedFOAM from the OpenFOAM software. I did a 10s simulation and it took me approximately 24h. Then, keeping everything the same except for the application of an equation in the inlet of the Boudary condition, I performed a simulation of 0.6s and it took me about 24h.
Could someone tell me if it is really normal that applying an equation in the inlet of the boundary condition increases the simulation time so much? If it is not normal, could someone provide me with some tool to make the simulation more efficient?

Thank you very much in advance!

Tobermory · June 16, 2023, 03:20

Okay - let's break this down a little. The main question you are raising is "why is the simulation taking so much longer"?

I don't know sedFoam, but in any typical low Mach# flow solver there are three parts to each time step - calculations of initial velocity estimate; pressure calculation and velocity correction; calculation of turbulence and other transport equations. The first and last are generally very quick - from a numerical perspective they are typically trivial. The pressure solution is almost always the most expensive part of the simulation.

So - take a look at your log file and look at how many linear solver iterations are being done on the pressure field for each solver iteration (i.e. time step) - has this gone up significantly? Is it hitting the maxIter bounds? If so, then this will be why it is taking so much longer.

Assuming that this is the case, then the next question is of course "why is it doing that?" Simply put, you are asking the solver to do too much work on each time step - the pressure field and velocity fields are probably changing too much, and indeed you may not have got a sensible pressure solution by the time your solver hits the maxIter limit on the linear solver iterations. Try reducing your time step ...

I hope the above gets you thinking in the right direction. Good luck!

June 15, 2023, 11:54	Time consumer of simulation time	#1
nmc New Member Naomi Mestre Join Date: Jun 2023 Posts: 1 Rep Power: 0	I am running a sediment simulation using sedFOAM from the OpenFOAM software. I did a 10s simulation and it took me approximately 24h. Then, keeping everything the same except for the application of an equation in the inlet of the Boudary condition, I performed a simulation of 0.6s and it took me about 24h. Could someone tell me if it is really normal that applying an equation in the inlet of the boundary condition increases the simulation time so much? If it is not normal, could someone provide me with some tool to make the simulation more efficient? Thank you very much in advance!

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
bash script for pseudo-parallel usage of reconstructPar	kwardle	OpenFOAM Post-Processing	41	August 23, 2023 02:48
courant number increases to rather large values	6863523	OpenFOAM Running, Solving & CFD	22	July 5, 2023 23:48
[Other] Contribution a new utility: refine wall layer mesh based on yPlus field	lakeat	OpenFOAM Community Contributions	58	December 23, 2021 02:36
pressure in incompressible solvers e.g. simpleFoam	chrizzl	OpenFOAM Running, Solving & CFD	13	March 28, 2017 05:49
plot over time	fferroni	OpenFOAM Post-Processing	7	June 8, 2012 07:56

June 16, 2023, 03:20		#2
Tobermory Senior Member Join Date: Apr 2020 Location: UK Posts: 668 Rep Power: 14	Okay - let's break this down a little. The main question you are raising is "why is the simulation taking so much longer"? I don't know sedFoam, but in any typical low Mach# flow solver there are three parts to each time step - calculations of initial velocity estimate; pressure calculation and velocity correction; calculation of turbulence and other transport equations. The first and last are generally very quick - from a numerical perspective they are typically trivial. The pressure solution is almost always the most expensive part of the simulation. So - take a look at your log file and look at how many linear solver iterations are being done on the pressure field for each solver iteration (i.e. time step) - has this gone up significantly? Is it hitting the maxIter bounds? If so, then this will be why it is taking so much longer. Assuming that this is the case, then the next question is of course "why is it doing that?" Simply put, you are asking the solver to do too much work on each time step - the pressure field and velocity fields are probably changing too much, and indeed you may not have got a sensible pressure solution by the time your solver hits the maxIter limit on the linear solver iterations. Try reducing your time step ... I hope the above gets you thinking in the right direction. Good luck!