block coupled solver in 2.3.x

tjliang · February 3, 2015, 07:57

hello foamers,

i am now struggling with the conjunct heat transfer problem with chtmultiregionfoam. As in the fluid region a lot of equations need to be solved and at the moment the simulation is very unstable. And i was once advised by a nice foamer in our forum that to use a impilcit coupled solver in version 1.6.x, which will solve many equations simultaneously. But i have already worked with version 2.3.x for a long time. So i wonder if anyone could tell me if it's possible to have the same block coupled solver in 2.3.x. Otherwise i can only use very small time step with segregated solver to solve it which costs a lot of time.

Bests,

Peng

Henning86 · February 4, 2015, 10:00

compiling a blocked-coupled solver from extended 3.1 in OF 2.3.x isn't straight forward at all. Its easier to start with extended 3.1.

you mentioned you had stabilty issues:

do you use fixedFluxPressure for the pressure boundary condition?

tjliang · February 4, 2015, 13:59

Quote:

Originally Posted by Henning86

compiling a blocked-coupled solver from extended 3.1 in OF 2.3.x isn't straight forward at all. Its easier to start with extended 3.1.

you mentioned you had stabilty issues:

do you use fixedFluxPressure for the pressure boundary condition?

Hi, thank you for the reply. actually yes except for the outlet boundary. I mentioned this coupled solver is because i have a lot of equations to solve including two temperature equations and about 7 equations which describe the transport properties which are in turn also dependent from the two temperatures. So this set of many equations is actually coupled with each other. When i use segragated solver, namely chtmultiregionfoam, it's simply very hard to converge unless i use a time step of 1e-12s, it very time-demanding so i gave it up. For example when i have a temperature of 12000K in the last time step and i will get suddenly a temperature of ,say -2000000K, it's just very very unstable. I hope some one could offer me good suggestions, thanks.

Henning86 · February 5, 2015, 05:48

block coupled solver are more effective with these equations.

in extended 3.1 there are some solvers which use the block coupled matrices.

elvis · February 5, 2015, 10:31

Hi,

it makes sense to have multiple OF installations (ESI´s and OF-extend Project)side by side and just switch to that OF install by type "OF31ext" "OF231" or "OF23x" you can handle this by creating various/multiple "alias-scripts" in your $HOME/.bashrc-file.

"block coupled" can speed up things at the price of "bigger memory footprint" needed to handle bigger matrices.
comparison of segregated vs OF extend vs company optimised solver http://sourceforge.net/projects/open...W09_P_0037.pdf

February 3, 2015, 07:57	block coupled solver in 2.3.x	#1
tjliang Member Peng Liang Join Date: Mar 2014 Posts: 60 Rep Power: 12	hello foamers, i am now struggling with the conjunct heat transfer problem with chtmultiregionfoam. As in the fluid region a lot of equations need to be solved and at the moment the simulation is very unstable. And i was once advised by a nice foamer in our forum that to use a impilcit coupled solver in version 1.6.x, which will solve many equations simultaneously. But i have already worked with version 2.3.x for a long time. So i wonder if anyone could tell me if it's possible to have the same block coupled solver in 2.3.x. Otherwise i can only use very small time step with segregated solver to solve it which costs a lot of time. Bests, Peng

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
NASA Rotor 37 & Pressure-Based Coupled Solver	Razvan	FLUENT	6	June 11, 2015 04:58
Runge Kutta for (point-)implicit coupled compressible solver?	Shenren_CN	Main CFD Forum	0	October 15, 2012 04:12
segregated solver vrs coupled solver	sm	FLUENT	0	November 6, 2007 02:24
Can I use coupled solver for this problem	Frank	FLUENT	0	April 11, 2006 07:28
Coupled or Seggregated solver	Ejaz Ahmed	FLUENT	4	July 15, 2005 13:35

February 4, 2015, 10:00		#2
Henning86 Member Join Date: Nov 2012 Posts: 83 Rep Power: 13	compiling a blocked-coupled solver from extended 3.1 in OF 2.3.x isn't straight forward at all. Its easier to start with extended 3.1. you mentioned you had stabilty issues: do you use fixedFluxPressure for the pressure boundary condition?

February 5, 2015, 05:48		#4
Henning86 Member Join Date: Nov 2012 Posts: 83 Rep Power: 13	block coupled solver are more effective with these equations. in extended 3.1 there are some solvers which use the block coupled matrices.

February 5, 2015, 10:31		#5
elvis Senior Member Elvis Join Date: Mar 2009 Location: Sindelfingen, Germany Posts: 620 Blog Entries: 6 Rep Power: 24	Hi, it makes sense to have multiple OF installations (ESI´s and OF-extend Project)side by side and just switch to that OF install by type "OF31ext" "OF231" or "OF23x" you can handle this by creating various/multiple "alias-scripts" in your $HOME/.bashrc-file. "block coupled" can speed up things at the price of "bigger memory footprint" needed to handle bigger matrices. comparison of segregated vs OF extend vs company optimised solver http://sourceforge.net/projects/open...W09_P_0037.pdf