fvSchemes with tetramesh

bye bye my blue · February 2, 2017, 21:47

div(phi,U) bounded Gauss linearUpwind grad(U);

div(phi,U) gauss linear;

the difference of these schemes is order (1st or 2nd). because of this, solutions (result of 2 schemes) can be different??

bounded's velocity profile is more accurate than gauss linear's . Is it acceptable??

matejfor · February 6, 2017, 07:16

Hello your blue,
(really? how old are you to use such a silly nick?)

both schemes are second order (http://www.openfoam.com/documentatio...ar-upwind.html)

still the results will be different.
Linear should be more precise in the mesh is reasonably fine and you do not break the stability. You may have a look here: http://www.openfoam.com/documentatio...e-example.html for classical comparison of the basic schemes.

happy foaming,

your Great Hyperlobic Omni-Cognate Neutron Wrangler

bye bye my blue · February 6, 2017, 09:06

umm.. then, can 'bounded' makes more accurate solution than gauss linear?

thanks

p.s) bye bye my blue is my favorite song's title ^^.

matejfor · February 6, 2017, 09:27

To answer the question, let me rephrase it:

Q:Can 'bounded' make the linear solution more correct?

A: Yes it can.

And why and how is described here:
http://www.openfoam.com/documentatio...ivergence.html

hope this helps

bye bye my blue · February 6, 2017, 20:01

umm,.. sorry to say that, I have a question..

in (http://www.openfoam.com/documentatio...ivergence.html), for steady state cases the bounded form can be applied:

it says that only for Steady state? If not, in unsteady state, bounded can make no or small oscillating solution as the linear solution + bounded?

(http://www.openfoam.com/documentatio...e-example.html)

Thanks.

matejfor · February 7, 2017, 09:26

Hi,

In the steady state (more importantly with SIMPLE solver) the solution of momentum equation is important to set the initial field for pressure equation. But basically the velocity coming from the momentum equation is garbage. It is based on the pressure field which does not conform the continuity equation. This may lead to disturbances in the pressure field from which the corrected velocities are computed. When the velocity fields contains unphysical values it: (1) takes long to get out of them from the solution, (2) typically leads to steep gradients resulting in excessive source terms to epsilon equation. This leads to jumps in the turbulent viscosity and finally unstable solution at the start of my run. So it is good idea to make sure your steady state solution is bounded and the values are reasonable.

Why at steady state? That is typically the case where your initial field is miles away from a physical situation. So this bounding does not increase your precision (as the term introduced by it should diminish as the solution is progressing) but the stability.

You can do a simple easy test. Let's take the pitzDaily tutorial from $FOAM_TUTORIALS/incompressible/simpleFoam and run the case with bounded Gauss and without. Then run foamLog and compare continuity errors, velocity residuals. I am using (OpenFOAM+1612 which is essentially enhanced OpenFOAM v 4.1) and I do see the difference only when I plot kMax_0 or epsilon. where the bounding cuts the values down. but the final solutions are equal. as they should be.

To sum it up. Using bounded schemes for steady state is a good idea, but not necessary.

Hope this helps.

February 2, 2017, 21:47	fvSchemes with tetramesh	#1
bye bye my blue Member bye bye my blue Join Date: Sep 2016 Posts: 37 Rep Power: 8	div(phi,U) bounded Gauss linearUpwind grad(U); div(phi,U) gauss linear; the difference of these schemes is order (1st or 2nd). because of this, solutions (result of 2 schemes) can be different?? bounded's velocity profile is more accurate than gauss linear's . Is it acceptable?? Last edited by bye bye my blue; February 2, 2017 at 21:48. Reason: just modifying

February 6, 2017, 09:06	thx for replying to me	#3
bye bye my blue Member bye bye my blue Join Date: Sep 2016 Posts: 37 Rep Power: 8	umm.. then, can 'bounded' makes more accurate solution than gauss linear? thanks p.s) bye bye my blue is my favorite song's title ^^.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
using fvc::interpolate without fvSchemes / manually prescribe interpolation schemes	linch	OpenFOAM Programming & Development	5	March 20, 2018 05:49
about fvSchemes with tetramesh	bye bye my blue	OpenFOAM	0	February 2, 2017 20:29
First order in fvSchemes does not seem to work	gerritgroot	OpenFOAM Running, Solving & CFD	0	September 30, 2015 20:06
Unrealistic values in cells at edge (high pressure & velocity) - fvSchemes?!	matzbanni	OpenFOAM Running, Solving & CFD	8	August 24, 2015 16:46
fvschemes and fvsolutions in MRFSimpleFoam	renyun0511	OpenFOAM Running, Solving & CFD	23	August 3, 2011 04:07

February 6, 2017, 07:16		#2
matejfor Senior Member matej forman Join Date: Mar 2009 Location: Brno, Czech Republic Posts: 182 Rep Power: 17	Hello your blue, (really? how old are you to use such a silly nick?) both schemes are second order (http://www.openfoam.com/documentatio...ar-upwind.html) still the results will be different. Linear should be more precise in the mesh is reasonably fine and you do not break the stability. You may have a look here: http://www.openfoam.com/documentatio...e-example.html for classical comparison of the basic schemes. happy foaming, your Great Hyperlobic Omni-Cognate Neutron Wrangler

February 6, 2017, 09:27		#4
matejfor Senior Member matej forman Join Date: Mar 2009 Location: Brno, Czech Republic Posts: 182 Rep Power: 17	To answer the question, let me rephrase it: Q:Can 'bounded' make the linear solution more correct? A: Yes it can. And why and how is described here: http://www.openfoam.com/documentatio...ivergence.html hope this helps

February 6, 2017, 20:01		#5
bye bye my blue Member bye bye my blue Join Date: Sep 2016 Posts: 37 Rep Power: 8	umm,.. sorry to say that, I have a question.. in (http://www.openfoam.com/documentatio...ivergence.html), for steady state cases the bounded form can be applied: it says that only for Steady state? If not, in unsteady state, bounded can make no or small oscillating solution as the linear solution + bounded? (http://www.openfoam.com/documentatio...e-example.html) Thanks.

February 7, 2017, 09:26		#6
matejfor Senior Member matej forman Join Date: Mar 2009 Location: Brno, Czech Republic Posts: 182 Rep Power: 17	Hi, In the steady state (more importantly with SIMPLE solver) the solution of momentum equation is important to set the initial field for pressure equation. But basically the velocity coming from the momentum equation is garbage. It is based on the pressure field which does not conform the continuity equation. This may lead to disturbances in the pressure field from which the corrected velocities are computed. When the velocity fields contains unphysical values it: (1) takes long to get out of them from the solution, (2) typically leads to steep gradients resulting in excessive source terms to epsilon equation. This leads to jumps in the turbulent viscosity and finally unstable solution at the start of my run. So it is good idea to make sure your steady state solution is bounded and the values are reasonable. Why at steady state? That is typically the case where your initial field is miles away from a physical situation. So this bounding does not increase your precision (as the term introduced by it should diminish as the solution is progressing) but the stability. You can do a simple easy test. Let's take the pitzDaily tutorial from $FOAM_TUTORIALS/incompressible/simpleFoam and run the case with bounded Gauss and without. Then run foamLog and compare continuity errors, velocity residuals. I am using (OpenFOAM+1612 which is essentially enhanced OpenFOAM v 4.1) and I do see the difference only when I plot kMax_0 or epsilon. where the bounding cuts the values down. but the final solutions are equal. as they should be. To sum it up. Using bounded schemes for steady state is a good idea, but not necessary. Hope this helps.