Convergence issues with heat transfer from tube wall

Gadders · September 30, 2015, 10:42

Hi,

I am attempting to run a 2D axisym simulation of heat transfer through a stainless capillary tube (270um ID x 150mm). The tube is heated via direct resistance heating and the fluid is kerosene. Kerosene enters the tube at ambient temperature, is heated by the tube wall and exits at 290oC. The system pressure is 20MPa.

My approach thus far has been to compute the developed exit velocity profile from uniform inlet velocity, apply this at the inlet and converge, before switching on the energy equation and running the simulation with temperature dependent material properties. The inlet Re is around 2000 and the exit Re should be around 10000 at 290oC.

I've been using the pressure based solver and have tried k-e w/ enhanced wall fn and kw SST w/ low Re correction. Ive tried meshes of varying densities - ranging from 200,000 to 900,000 nodes, with y+ values from 1 to 0.1 (designing the mesh for the expected exit Re means y+ is well below 1 in most cases). Solution methods have been left unchanged (although with 2nd order discretisation).

BCs:
Velocity inlet: 10m/s, I=5%
outlet: Outflow
External wall: natural convection & radiation
Solid domain: energy source (should be 8e8 W/m^3 for 75A)

I'm having major issues with continuity as I apply heat. I have managed to get a sensible result from a run with k-e and a relatively coarse mesh, but that was after only a few hundred iterations and not converged.

Does anyone have any suggestions as to why the solution may not be converging or what I can do to aid convergence? I can't believe the mesh is the issue with such low y+ (or can you have too low a y+?), that number of cells and a maximum aspect ratio of 7.

Should i try different under relaxation values or changing the solution methods? Or the density based solver even? Could it be the change of fluid density and viscosity with T that is the issue?

Hopefully thats enough to go by for some useful advice!

Thanks,
Phil.

Gadders · October 8, 2015, 13:37

Any suggestions please? I hope I have provided enough information without actually uploading the case and mesh.

My material property definitions perhaps? I have just entered the polynomials in the GUI for both materials. Is there any benefit to defining properties with a UDF instead? Should I simplify the case with constant material properties where applicable?

Maybe the pressure? The system pressure is 20MPa and I have applied this as the operating pressure rather than the gauge pressure at the inlet. Does this have any effect?

I have been poring over every detail of the model and I am at a loss as to why the solution will not converge.
Thanks

devesh.baghel · October 9, 2015, 01:43

Hi Gadder,

Your model (stainless capillary tube (270um ID x 150mm)) is highly loaded BC & Operating conditions. For such a small model, rapid change in physical parameter could be difficult to control. Try to put all BC's i.e. input velocity, Temp and Q in gradually increasing order. This may have control over solution till reach actual BC's.

I hope it should work....

Regards

Gadders · October 12, 2015, 09:03

Thanks for the reply Devesh.

I have been slowly ramping up the heat as you suggested. I can get to around 3e8 with convergence but past that it struggles.

I am beginning to think perhaps there is a fault with my material properties, or that i simply need even more cells!

September 30, 2015, 10:42	Convergence issues with heat transfer from tube wall	#1
Gadders New Member Phil Gadsby Join Date: Sep 2015 Posts: 8 Rep Power: 10	Hi, I am attempting to run a 2D axisym simulation of heat transfer through a stainless capillary tube (270um ID x 150mm). The tube is heated via direct resistance heating and the fluid is kerosene. Kerosene enters the tube at ambient temperature, is heated by the tube wall and exits at 290oC. The system pressure is 20MPa. My approach thus far has been to compute the developed exit velocity profile from uniform inlet velocity, apply this at the inlet and converge, before switching on the energy equation and running the simulation with temperature dependent material properties. The inlet Re is around 2000 and the exit Re should be around 10000 at 290oC. I've been using the pressure based solver and have tried k-e w/ enhanced wall fn and kw SST w/ low Re correction. Ive tried meshes of varying densities - ranging from 200,000 to 900,000 nodes, with y+ values from 1 to 0.1 (designing the mesh for the expected exit Re means y+ is well below 1 in most cases). Solution methods have been left unchanged (although with 2nd order discretisation). BCs: Velocity inlet: 10m/s, I=5% outlet: Outflow External wall: natural convection & radiation Solid domain: energy source (should be 8e8 W/m^3 for 75A) I'm having major issues with continuity as I apply heat. I have managed to get a sensible result from a run with k-e and a relatively coarse mesh, but that was after only a few hundred iterations and not converged. Does anyone have any suggestions as to why the solution may not be converging or what I can do to aid convergence? I can't believe the mesh is the issue with such low y+ (or can you have too low a y+?), that number of cells and a maximum aspect ratio of 7. Should i try different under relaxation values or changing the solution methods? Or the density based solver even? Could it be the change of fluid density and viscosity with T that is the issue? Hopefully thats enough to go by for some useful advice! Thanks, Phil.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Conjugate Heat Transfer: Wall Heat Flux at Coupled Walls?	MaxHeat	FLUENT	4	September 14, 2017 10:44
Wrong flow in ratating domain problem	Sanyo	CFX	17	August 15, 2015 06:20
Compression stoke is giving higher pressure than calculated	nickjuana	CFX	62	May 19, 2015 13:32
Difficulty In Setting Boundary Conditions	Moinul Haque	CFX	4	November 25, 2014 17:30
Searching for a test with heat transfer through the wall	anya86	Main CFD Forum	0	November 18, 2012 19:53

October 8, 2015, 13:37		#2
Gadders New Member Phil Gadsby Join Date: Sep 2015 Posts: 8 Rep Power: 10	Any suggestions please? I hope I have provided enough information without actually uploading the case and mesh. My material property definitions perhaps? I have just entered the polynomials in the GUI for both materials. Is there any benefit to defining properties with a UDF instead? Should I simplify the case with constant material properties where applicable? Maybe the pressure? The system pressure is 20MPa and I have applied this as the operating pressure rather than the gauge pressure at the inlet. Does this have any effect? I have been poring over every detail of the model and I am at a loss as to why the solution will not converge. Thanks

October 9, 2015, 01:43		#3
devesh.baghel Member Devesh Baghel Join Date: Mar 2009 Posts: 84 Rep Power: 17	Hi Gadder, Your model (stainless capillary tube (270um ID x 150mm)) is highly loaded BC & Operating conditions. For such a small model, rapid change in physical parameter could be difficult to control. Try to put all BC's i.e. input velocity, Temp and Q in gradually increasing order. This may have control over solution till reach actual BC's. I hope it should work.... Regards

October 12, 2015, 09:03		#4
Gadders New Member Phil Gadsby Join Date: Sep 2015 Posts: 8 Rep Power: 10	Thanks for the reply Devesh. I have been slowly ramping up the heat as you suggested. I can get to around 3e8 with convergence but past that it struggles. I am beginning to think perhaps there is a fault with my material properties, or that i simply need even more cells!