Heat transfer within a porous domain

rbezerra · January 4, 2015, 10:09

Dear all,

I’m having difficulties in simulating heat transfer within a porous medium in CFX. I tried to set up a simple case of a rectangular channel (600 x 1.5 x 0.3 mm). The top boundary (y = 1.5mm) is modelled as a wall with constant temperature of 80°C, the bottom and side boundaries are symmetry.
The mesh has 1 element thickness in the Z-direction (~2D simulation), 15 elements in the Y-direction (biased, so that the region near the top wall has a better resolution) and element length of 0.5 mm in the X-direction.

The fluid (water) is entering at x=0 with a static temperature of 25°C. For the outlet boundary (x = 600 mm) I have tried different settings: fixed temperature of 80°C and adiabatic.

It seems to be a global convergence issue. I checked the residuals which are more or less spread throughout the whole domain. I reached after 1000 iterations a RMS for H-Energy of the magnitude 1E-2. I also monitored the global imbalance for H-Energy, which gave me the following terms:

Boundary: outlet -2.4882E-02
Boundary: wall 1.2982E+01
Domain Src (Neg) : Domain 1 -4.2079E+00
Domain Src (Pos) : Domain 1 4.2251E+00
Domain Imbalance : 1.2974E+01
Domain Imbalance, in %: 99.9405 %

Any ideas, as to why the simulation is not converging?
In the Ansys help I found an explanation for the term “domain src (neg)” as being the thermal energy that flows between fluid and solid, which makes sense. Why do I get a domain src (pos) term in the same order of magnitude? Why am I not getting a term for the inlet?

Might this be related to a limitation of CFX and/or thin slice domain?

Thanks for your help and happy new year!

Renato

ghorrocks · January 4, 2015, 18:04

Are you sure it is converged? 1E-2 is very loose convergence and rarely gives useable results. I think you just need to run longer.

rbezerra · January 7, 2015, 08:25

Thanks for the reply.
I'm assuming the simulation has not converged.
But after about 300 iterations the residuals are not getting any lower, and I ran 1000 iterations, so I don't expect it to be an issue of slow convergence.

Opaque · January 7, 2015, 10:38

what flow boundary conditions are you using for the inlet ? You only described the heat transfer boundary conditions.

Have you tried the case without heat transfer first ? Does it converge ?

What timescale are you using for the fluid ? How about the solid timescale ?

ghorrocks · January 7, 2015, 18:53

Quote:

But after about 300 iterations the residuals are not getting any lower

Sounds like an FAQ: http://www.cfd-online.com/Wiki/Ansys...gence_criteria

rbezerra · January 9, 2015, 06:06

Increasing the timescale helped in my case. Thanks for the replies!

January 4, 2015, 10:09	Heat transfer within a porous domain	#1
rbezerra New Member Join Date: Jan 2015 Posts: 6 Rep Power: 11	Dear all, I’m having difficulties in simulating heat transfer within a porous medium in CFX. I tried to set up a simple case of a rectangular channel (600 x 1.5 x 0.3 mm). The top boundary (y = 1.5mm) is modelled as a wall with constant temperature of 80°C, the bottom and side boundaries are symmetry. The mesh has 1 element thickness in the Z-direction (~2D simulation), 15 elements in the Y-direction (biased, so that the region near the top wall has a better resolution) and element length of 0.5 mm in the X-direction. The fluid (water) is entering at x=0 with a static temperature of 25°C. For the outlet boundary (x = 600 mm) I have tried different settings: fixed temperature of 80°C and adiabatic. It seems to be a global convergence issue. I checked the residuals which are more or less spread throughout the whole domain. I reached after 1000 iterations a RMS for H-Energy of the magnitude 1E-2. I also monitored the global imbalance for H-Energy, which gave me the following terms: Boundary: outlet -2.4882E-02 Boundary: wall 1.2982E+01 Domain Src (Neg) : Domain 1 -4.2079E+00 Domain Src (Pos) : Domain 1 4.2251E+00 Domain Imbalance : 1.2974E+01 Domain Imbalance, in %: 99.9405 % Any ideas, as to why the simulation is not converging? In the Ansys help I found an explanation for the term “domain src (neg)” as being the thermal energy that flows between fluid and solid, which makes sense. Why do I get a domain src (pos) term in the same order of magnitude? Why am I not getting a term for the inlet? Might this be related to a limitation of CFX and/or thin slice domain? Thanks for your help and happy new year! Renato

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Water subcooled boiling	Attesz	CFX	7	January 5, 2013 04:32
Heat transfer thorugh porous	kmgraju	CFX	3	December 15, 2012 00:22
Convective / Conductive Heat Transfer in Hypersonic flows	enigma	Main CFD Forum	2	November 1, 2009 23:53
heat transfer in porous domain	Tiago	CFX	3	October 19, 2008 02:20
REG POROUS MEDIA HEAT TRANSFER	Rashmi Venkat	FLUENT	6	June 1, 2006 06:13

January 4, 2015, 18:04		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,871 Rep Power: 144	Are you sure it is converged? 1E-2 is very loose convergence and rarely gives useable results. I think you just need to run longer.

January 7, 2015, 08:25		#3
rbezerra New Member Join Date: Jan 2015 Posts: 6 Rep Power: 11	Thanks for the reply. I'm assuming the simulation has not converged. But after about 300 iterations the residuals are not getting any lower, and I ran 1000 iterations, so I don't expect it to be an issue of slow convergence.

January 7, 2015, 10:38		#4
Opaque Senior Member Join Date: Jun 2009 Posts: 1,880 Rep Power: 33	what flow boundary conditions are you using for the inlet ? You only described the heat transfer boundary conditions. Have you tried the case without heat transfer first ? Does it converge ? What timescale are you using for the fluid ? How about the solid timescale ?

January 9, 2015, 06:06		#6
rbezerra New Member Join Date: Jan 2015 Posts: 6 Rep Power: 11	Increasing the timescale helped in my case. Thanks for the replies!