Unstable results in steady state solver

Alilo · June 28, 2017, 15:33

Dear friends,

I'm using SimpleFoam solver. As you probably know, simpleFoam is steadystate solver, but when I simulate my model I saw unstable response and I have oscillation.

I've checked all boundary conditions and grid.

My question is that, what should I do to see stable results?

Appreciate.

fritznieb · December 4, 2020, 04:00

If your geometry is creating a turbulent wake for example, it is quite possible there is no "one steady solution".
For a rough guess general geometry (sharp edges), courantnumber, Reynoldsnumber might be a good idea to look at.
Also the iteration time step(assuming that you are using openfoam) might be a good place to vary.

Sorry for the vague answer but that's what I in general would look at based on the question.
Good luck

Tobermory · December 4, 2020, 07:08

Just to expand on fritznieb's response - if the flow solution is not steady, then there is no steady solution. Think of vortex shedding behind a cylinder for example - steady oncoming flow, but an unsteady wake flow as fritznieb describes. With a very coarse mesh, you might smooth out the unsteadiness through numerical diffusion; but the more accurate you make the modelling, the less artificial stability is introduced into the model and any unsteadiness in the solution will start to express itself through a limit-cycle type of behaviour.

To get the "steady" behaviour, or more accurately a time-averaged set of results, you'd need to run as transient and then field average over a sufficient number of periods of cycle.

What is the flow that you're trying to simulate, out of interest?

June 28, 2017, 15:33	Unstable results in steady state solver	#1
Alilo New Member Ali Join Date: Jun 2017 Posts: 2 Rep Power: 0	Dear friends, I'm using SimpleFoam solver. As you probably know, simpleFoam is steadystate solver, but when I simulate my model I saw unstable response and I have oscillation. I've checked all boundary conditions and grid. My question is that, what should I do to see stable results? Appreciate.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Steady State solver for High Mach No flows	mecbe2002	OpenFOAM	10	July 6, 2021 04:56
mass flow in is not equal to mass flow out	saii	CFX	12	March 19, 2018 06:21
Is steady state solver grid dependent?	mmkr825	OpenFOAM	4	March 22, 2013 13:46
convergence issues on steady state solver	icemaniac178	CFX	1	March 30, 2011 20:11
Physical time step for steady state solver	Marcin	OpenFOAM	0	July 2, 2009 03:20

December 4, 2020, 04:00		#2
fritznieb New Member Fritz Nieborowski Join Date: Nov 2020 Posts: 2 Rep Power: 0	If your geometry is creating a turbulent wake for example, it is quite possible there is no "one steady solution". For a rough guess general geometry (sharp edges), courantnumber, Reynoldsnumber might be a good idea to look at. Also the iteration time step(assuming that you are using openfoam) might be a good place to vary. Sorry for the vague answer but that's what I in general would look at based on the question. Good luck

December 4, 2020, 07:08		#3
Tobermory Senior Member Join Date: Apr 2020 Location: UK Posts: 746 Rep Power: 14	Just to expand on fritznieb's response - if the flow solution is not steady, then there is no steady solution. Think of vortex shedding behind a cylinder for example - steady oncoming flow, but an unsteady wake flow as fritznieb describes. With a very coarse mesh, you might smooth out the unsteadiness through numerical diffusion; but the more accurate you make the modelling, the less artificial stability is introduced into the model and any unsteadiness in the solution will start to express itself through a limit-cycle type of behaviour. To get the "steady" behaviour, or more accurately a time-averaged set of results, you'd need to run as transient and then field average over a sufficient number of periods of cycle. What is the flow that you're trying to simulate, out of interest?