Dirichlet–Neumann Partitioning BCs

Giancarlo_IngChimico · April 13, 2013, 09:05

Hi all,
I have to implement the Dirichlet–Neumann Partitioning BCs for my solver that treats two different regions (fluid and solid).
Is there any preconfigured BC in the framework of OpenFoam?

Thanks

Regards

Giancarlo

ngj · April 14, 2013, 06:14

Hi Giancario,

I am not familiar with the term "Dirichlet-Neumann Partitioning", however, maybe the mixedFvPatch class is what you need.

In this class you can switch (smoothly) from a dirichlet to a neumann condition depending on the instantaneous dynamics of the system. I use it in the binary fashion to switch between fixed value and zero gradient, where the choice depends on the internal value of some concentration field.

Kind regards

Niels

Cyp · April 14, 2013, 16:45

Hi Giancarlo,

I fear there is no Direchlet-Neumann BC coded by default in OF and that you have to develop your owns. You can find a lot of informations in http://www.personal.psu.edu/dab143/O...ven_slides.pdf and its associated tutorial http://www.personal.psu.edu/dab143/O...erTraining.tgz.

The relaxation factors are quite important in this example.

@++
Cyp

Giancarlo_IngChimico · April 14, 2013, 18:20

Thanks a lot Cyp. This materials is fantastic. It's just what I was looking for.

Regards

Giancarlo

chriss85 · May 30, 2014, 09:48

In case you still read here, have you got this working?

I'm currently implementing something similar for electric potential between different regions.

By now OpenFOAM has a boundary condition for the temperature that uses an approach like this, called turbulentTemperatureCoupledBaffleMixed.

I adopted it to my needs, but I don't understand the weighting factor in the mixed formulation. Why is it set to this value? Does it help convergence if this value is used instead of some arbitrary value between 0 and 1 (like 0.5)?

Furthermore, how does relaxation help here? My current solver doesn't use relaxation (it's transient), and it requires quite some iterations initially until a consistent solution is reached (up to 30 in a test case I made).

The link mentioned using Aitken's delta² method (see here), which is basically a quadratic extrapolation of the convergence series of the field. In its current implementation, OpenFOAM doesn't use this. Have you implemented this? Is it worth it for transient cases? I believe that the convergence should be better in transient cases because the solution changes only a bit between steps.

chriss85 · June 2, 2014, 10:46

After some more investigation I figured out that relaxation didn't refer to the field relaxation, but rather to the coupling between the patches of different regions.

I also compiled the MRConjugateHeatFoam solver with OF2.3, but it doesn't calculate any sensible results and the solution of the example looks very incorrect. Has anyone tried this solver with success?

ahmmedshakil · June 9, 2014, 11:24

Hi Chirss85,
I worked with the MRConjugateHeatFoam. In my case, I validated the solver with some experimental cases, and the results were really good. BTW, my problem was heat conduction between two solids (i.e. silicon on a glass substrates) due to laser annealing.

Cheers
shakil

Quote:

Originally Posted by chriss85

After some more investigation I figured out that relaxation didn't refer to the field relaxation, but rather to the coupling between the patches of different regions.

I also compiled the MRConjugateHeatFoam solver with OF2.3, but it doesn't calculate any sensible results and the solution of the example looks very incorrect. Has anyone tried this solver with success?

April 13, 2013, 09:05	Dirichlet–Neumann Partitioning BCs	#1
Giancarlo_IngChimico New Member Giancarlo Join Date: Apr 2013 Location: Milan Posts: 21 Rep Power: 13	Hi all, I have to implement the Dirichlet–Neumann Partitioning BCs for my solver that treats two different regions (fluid and solid). Is there any preconfigured BC in the framework of OpenFoam? Thanks Regards Giancarlo

April 14, 2013, 16:45		#3
Cyp Senior Member Cyprien Join Date: Feb 2010 Location: Stanford University Posts: 299 Rep Power: 18	Hi Giancarlo, I fear there is no Direchlet-Neumann BC coded by default in OF and that you have to develop your owns. You can find a lot of informations in http://www.personal.psu.edu/dab143/O...ven_slides.pdf and its associated tutorial http://www.personal.psu.edu/dab143/O...erTraining.tgz. The relaxation factors are quite important in this example. @++ Cyp fumiya and wayne14 like this.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
ERROR #001100279 has occurred in subroutine ErrAction.	smnaryal	CFX	11	December 20, 2017 17:32
Dealing with BC's in OF 1.6	vkrastev	OpenFOAM Running, Solving & CFD	5	September 4, 2012 12:58
Understanding Code behind BCs	Linse	OpenFOAM Programming & Development	8	January 9, 2012 09:58
BCs for Pressure Correction Equation (SIMPLE)	Bharath Somayaji	Main CFD Forum	1	March 1, 2006 07:12
Need Assistance: BCs for k-e model implementation	Dean Schrage	Main CFD Forum	3	August 11, 2001 02:02

April 14, 2013, 06:14		#2
ngj Senior Member Niels Gjoel Jacobsen Join Date: Mar 2009 Location: Copenhagen, Denmark Posts: 1,903 Rep Power: 37	Hi Giancario, I am not familiar with the term "Dirichlet-Neumann Partitioning", however, maybe the mixedFvPatch class is what you need. In this class you can switch (smoothly) from a dirichlet to a neumann condition depending on the instantaneous dynamics of the system. I use it in the binary fashion to switch between fixed value and zero gradient, where the choice depends on the internal value of some concentration field. Kind regards Niels

April 14, 2013, 18:20		#4
Giancarlo_IngChimico New Member Giancarlo Join Date: Apr 2013 Location: Milan Posts: 21 Rep Power: 13	Thanks a lot Cyp. This materials is fantastic. It's just what I was looking for. Regards Giancarlo

May 30, 2014, 09:48		#5
chriss85 Senior Member Join Date: Oct 2013 Posts: 397 Rep Power: 19	In case you still read here, have you got this working? I'm currently implementing something similar for electric potential between different regions. By now OpenFOAM has a boundary condition for the temperature that uses an approach like this, called turbulentTemperatureCoupledBaffleMixed. I adopted it to my needs, but I don't understand the weighting factor in the mixed formulation. Why is it set to this value? Does it help convergence if this value is used instead of some arbitrary value between 0 and 1 (like 0.5)? Furthermore, how does relaxation help here? My current solver doesn't use relaxation (it's transient), and it requires quite some iterations initially until a consistent solution is reached (up to 30 in a test case I made). The link mentioned using Aitken's delta² method (see here), which is basically a quadratic extrapolation of the convergence series of the field. In its current implementation, OpenFOAM doesn't use this. Have you implemented this? Is it worth it for transient cases? I believe that the convergence should be better in transient cases because the solution changes only a bit between steps.

June 2, 2014, 10:46		#6
chriss85 Senior Member Join Date: Oct 2013 Posts: 397 Rep Power: 19	After some more investigation I figured out that relaxation didn't refer to the field relaxation, but rather to the coupling between the patches of different regions. I also compiled the MRConjugateHeatFoam solver with OF2.3, but it doesn't calculate any sensible results and the solution of the example looks very incorrect. Has anyone tried this solver with success?