Running localEuler ddt on transient case

sufjanst · February 15, 2019, 09:19

Hey everyone,

I am running a LES combustion simulation with PaSR.

First, I calculated a steady solution with localEuler as ddt scheme and k-omega-SST.
Afterwards I switched to LES, changed the timestep size and let it run. I forgot to change localEuler to Euler. The results look really good.

My question is: Does localEuler affect the solution in a transient case? Can I trust the solutions or do I have to run the calculation again with Euler ddt?
Solver is reactingFoam on OF6.

Thanks
sufjanst

clapointe · February 15, 2019, 14:25

Quoting from here :https://cfd.direct/openfoam/user-guide/v6-fvschemes/

"localEuler: pseudo transient for accelerating a solution to steady-state using local-time stepping; first order implicit."

So if you want an unsteady simulation, you'll need to use a different scheme.

Caelan

schuyler · February 15, 2019, 17:59

Unfortunately, if you are using local time stepping your statistics will not be correct. If you had areas with relatively small velocity variations, they may appear nice, but the small finite changes in velocity will affect the time steps. In reality, your simulation will be giving LES statistics in some sort of pseudo-time domain.

mfrl · February 17, 2019, 21:05

I believe in localEuler the solution is advanced in time at each node by using the maximum 'local' timestep. The solution att ecah node may be even updated before computing all the other nodes. Any relationship with the physical transient is lost.
Mrfl

sufjanst · February 18, 2019, 04:27

I only need the time-averaged values so it should be fine I guess. Anyways, I will run it with Euler ddt scheme to be sure.

Thank you guys!

February 15, 2019, 09:19	Running localEuler ddt on transient case	#1
sufjanst Member Join Date: Mar 2016 Posts: 73 Rep Power: 10	Hey everyone, I am running a LES combustion simulation with PaSR. First, I calculated a steady solution with localEuler as ddt scheme and k-omega-SST. Afterwards I switched to LES, changed the timestep size and let it run. I forgot to change localEuler to Euler. The results look really good. My question is: Does localEuler affect the solution in a transient case? Can I trust the solutions or do I have to run the calculation again with Euler ddt? Solver is reactingFoam on OF6. Thanks sufjanst

February 15, 2019, 17:59		#3
schuyler Member W. Schuyler Hinman Join Date: Apr 2013 Location: Calgary, Alberta, Canada Posts: 38 Rep Power: 13	Unfortunately, if you are using local time stepping your statistics will not be correct. If you had areas with relatively small velocity variations, they may appear nice, but the small finite changes in velocity will affect the time steps. In reality, your simulation will be giving LES statistics in some sort of pseudo-time domain. sufjanst likes this. __________________ Schuyler

February 17, 2019, 21:05		#4
mfrl New Member MF Join Date: Jan 2019 Posts: 6 Rep Power: 7	I believe in localEuler the solution is advanced in time at each node by using the maximum 'local' timestep. The solution att ecah node may be even updated before computing all the other nodes. Any relationship with the physical transient is lost. Mrfl sufjanst and hogsonik like this.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Transient species transport case	Hong M.H	Main CFD Forum	0	February 25, 2018 04:44
Time-averaging of variables over different time periods for transient case	m.uzair	FLUENT	4	December 20, 2017 06:40
Transient case with no inlet region	Ancient_Nik	CFX	11	March 21, 2011 02:46
Free surface boudary conditions with SOLA-VOF	Fan	Main CFD Forum	10	September 9, 2006 13:24
Fluidisation transient case	Vikas Kumar	Main CFD Forum	0	April 24, 2003 04:29

February 15, 2019, 14:25		#2
clapointe Senior Member Join Date: Aug 2015 Posts: 494 Rep Power: 15	Quoting from here :https://cfd.direct/openfoam/user-guide/v6-fvschemes/ "localEuler: pseudo transient for accelerating a solution to steady-state using local-time stepping; first order implicit." So if you want an unsteady simulation, you'll need to use a different scheme. Caelan

February 18, 2019, 04:27		#5
sufjanst Member Join Date: Mar 2016 Posts: 73 Rep Power: 10	I only need the time-averaged values so it should be fine I guess. Anyways, I will run it with Euler ddt scheme to be sure. Thank you guys!