Problem using Central differencing scheme with LES WALE

selvam2487 · June 10, 2013, 06:28

Hello Friends,

I am trying to run a Large Eddy Simulation with WALE model for analysing thermal mixing of fluids in a T junction piping system. Currently I am performing simulations in fluid domain only.

(i) Initially I did a steady state simulation using SST model and gave its result file as input for transient simulations involving LES. I chose LES WALE model and gave Central differencing scheme for Advection scheme and Second order backward Euler scheme for transient scheme.

(ii) I chose IAPWS library for material properties (Water in this case). I selected Table Generation checkbox and gave Min. and Max. Temperature as 273.15 K and 1000 K and Min. and Max. Absolute Pressure as 1000 Pa and 1e07 Pa.

(iii) Inlet and boundary conditions of my simulation are:
Hot water: T = 393 K and mass flow rate = 600g/s
Cold water: T = 298 K and mass flow rate = 100g/s
Adiabatic wall boundary.
At outlet 'Relative Pressure' was set to zero.

No. of Nodes: 5 Million

When I run the simulation and check the trn files after few hundred iterations, the temperatures are very unphysical (218 K and 425 K in this case). Therefore, I stopped the simulation and chose Bounded Central Difference scheme this time and the same thing happens. I ran the simulations in two different meshes with aspect ratios 90 and 8 respectively and nothing changes. But when I performed the same simulatons with upwind scheme for just the energy equation, then the simulation is ok with temperature in the limits of 298 K to 393 K.

If anybody has experienced this problem before or if you know the reason why this happens and what's the mistake which I need to correct, please guide me in dealing with this problem.

Thanks in advance for your help.

ghorrocks · June 10, 2013, 23:19

You are certainly making it hard for yourself by running a LES model with IAPWS. When you run IAPWS you will have a much harder time converging. Do you really need IAPWS? If you can use a constant properties fluid your simulation will be much better behaved and run much faster.

selvam2487 · June 11, 2013, 04:03

Can you please explain me why using IAPWS will not be good for the convergence of LES? My simulation involves mixing of fluids at two different temperatures (298 K and 393 K). I thought that using IAPWS library, one can know the variation of temperature dependent properties like density, viscosity etc. If using it makes the solution difficult to converge or produce unphysical results (in my case now), then can I incorporate expressions involving variation of water properties with temperature and run the simulation?

ghorrocks · June 11, 2013, 09:22

When material properties go non-linear then the equations get much harder to solve, even when the non-linearity is small. If you don't believe me then simply switch to a constant properties fluid and I bet it converges just fine.

You are absolutely correct in the IAPWS is the best way to make sure your water properties are correct for all conditions. But you pay a big price for that in difficulty of convergence, and to get around that you normally have to use much smaller timesteps = much longer simulation.

So the cost of the additional accuracy of IAPWS is high. So only use it when you really need it.

You can simplify it by defining simple functions for fluid properties in the conditions yo uare interested in. This is usually a lot more stable than IAPWS in my experience at least for small deviations from constant properties.

selvam2487 · June 12, 2013, 04:22

Dear Glenn,

I tried giving constant properties or expressions for water properties and it seems to be ok. The question which is bugging me is, "Inspite of the usage of IAPWS library being very difficult for convergence, why should it produce unphysical results when using Bounded Central Differencing Scheme for LES WALE model? But, when I switched to upwind scheme for only energy equations, then the temperature variation is within the defined limits of my simulation." Is it a problem related to my mesh or is it something else?

ghorrocks · June 12, 2013, 07:07

The unphysical results from IAPWS is due to it not converging, or having other numerical instabilities which cause it to result in unphysical results. To get IAPWS to work you are going to need to improve numerical stability -which means better mesh quality, smaller time steps, a better initial condition and/or double precision numerics.

selvam2487 · June 20, 2013, 08:28

Hello Glenn,

Thank you for your reply. I am working on the numerical grid and will see how the solution behaves when I try to implement the changes you mentioned. Once again I wanted to thank you very much for your help. I will let you know how the solution behaves after my simulations are over. Take care

raunakjung · June 4, 2017, 11:07

Quote:

Originally Posted by selvam2487

Dear Glenn,

I tried giving constant properties or expressions for water properties and it seems to be ok. The question which is bugging me is, "Inspite of the usage of IAPWS library being very difficult for convergence, why should it produce unphysical results when using Bounded Central Differencing Scheme for LES WALE model? But, when I switched to upwind scheme for only energy equations, then the temperature variation is within the defined limits of my simulation." Is it a problem related to my mesh or is it something else?

Hello can you share your experience in using LES with constant properties? Also, did you face unphysical results using constant properties with BCD and CD?

Thank You

selvam2487 · July 23, 2017, 16:33

Dear Friend,

Firstly, my sincere apologies for the delayed reply.

Moving on to your question, I have not used constant properties since there is a significant temperature difference between fluids in my calculations. So I finally used IAPWS library for defining water properties and performed all my LES calculations (with WALE sub-grid model) with IAPWS library for water properties.

In terms of numerical scheme, I ended up using High resolution scheme in CFX since CD or BCD ended up producing unphysical oscillations in terms of temperature. May be it has something to do with the CFX software code but I don't know. When I tried the exact thing in FLUENT using BCD, it was working. Hope it answers your question.

June 10, 2013, 06:28	Problem using Central differencing scheme with LES WALE	#1
selvam2487 Member Karthik Join Date: Oct 2012 Location: Germany Posts: 53 Rep Power: 14	Hello Friends, I am trying to run a Large Eddy Simulation with WALE model for analysing thermal mixing of fluids in a T junction piping system. Currently I am performing simulations in fluid domain only. (i) Initially I did a steady state simulation using SST model and gave its result file as input for transient simulations involving LES. I chose LES WALE model and gave Central differencing scheme for Advection scheme and Second order backward Euler scheme for transient scheme. (ii) I chose IAPWS library for material properties (Water in this case). I selected Table Generation checkbox and gave Min. and Max. Temperature as 273.15 K and 1000 K and Min. and Max. Absolute Pressure as 1000 Pa and 1e07 Pa. (iii) Inlet and boundary conditions of my simulation are: Hot water: T = 393 K and mass flow rate = 600g/s Cold water: T = 298 K and mass flow rate = 100g/s Adiabatic wall boundary. At outlet 'Relative Pressure' was set to zero. No. of Nodes: 5 Million When I run the simulation and check the trn files after few hundred iterations, the temperatures are very unphysical (218 K and 425 K in this case). Therefore, I stopped the simulation and chose Bounded Central Difference scheme this time and the same thing happens. I ran the simulations in two different meshes with aspect ratios 90 and 8 respectively and nothing changes. But when I performed the same simulatons with upwind scheme for just the energy equation, then the simulation is ok with temperature in the limits of 298 K to 393 K. If anybody has experienced this problem before or if you know the reason why this happens and what's the mistake which I need to correct, please guide me in dealing with this problem. Thanks in advance for your help.

June 11, 2013, 04:03	Les wale, bcds	#3
selvam2487 Member Karthik Join Date: Oct 2012 Location: Germany Posts: 53 Rep Power: 14	Can you please explain me why using IAPWS will not be good for the convergence of LES? My simulation involves mixing of fluids at two different temperatures (298 K and 393 K). I thought that using IAPWS library, one can know the variation of temperature dependent properties like density, viscosity etc. If using it makes the solution difficult to converge or produce unphysical results (in my case now), then can I incorporate expressions involving variation of water properties with temperature and run the simulation?

June 20, 2013, 08:28	Les wale	#7
selvam2487 Member Karthik Join Date: Oct 2012 Location: Germany Posts: 53 Rep Power: 14	Hello Glenn, Thank you for your reply. I am working on the numerical grid and will see how the solution behaves when I try to implement the changes you mentioned. Once again I wanted to thank you very much for your help. I will let you know how the solution behaves after my simulations are over. Take care

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Problem with LES	Roy	AVL FIRE	1	March 28, 2013 09:51
problem about numerical scheme in LES.	libin	Main CFD Forum	4	July 1, 2004 05:32
the numerical scheme for LES.	John S	Main CFD Forum	2	March 14, 2004 08:52
Jameson's scheme for LES ?	Li Yang	Main CFD Forum	8	February 25, 2004 14:02
Differencing scheme for k-eps model	Ron Barer	Main CFD Forum	14	November 2, 2000 01:02

June 10, 2013, 23:19		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,871 Rep Power: 144	You are certainly making it hard for yourself by running a LES model with IAPWS. When you run IAPWS you will have a much harder time converging. Do you really need IAPWS? If you can use a constant properties fluid your simulation will be much better behaved and run much faster.

June 11, 2013, 09:22		#4
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,871 Rep Power: 144	When material properties go non-linear then the equations get much harder to solve, even when the non-linearity is small. If you don't believe me then simply switch to a constant properties fluid and I bet it converges just fine. You are absolutely correct in the IAPWS is the best way to make sure your water properties are correct for all conditions. But you pay a big price for that in difficulty of convergence, and to get around that you normally have to use much smaller timesteps = much longer simulation. So the cost of the additional accuracy of IAPWS is high. So only use it when you really need it. You can simplify it by defining simple functions for fluid properties in the conditions yo uare interested in. This is usually a lot more stable than IAPWS in my experience at least for small deviations from constant properties.

June 12, 2013, 04:22		#5
selvam2487 Member Karthik Join Date: Oct 2012 Location: Germany Posts: 53 Rep Power: 14	Dear Glenn, I tried giving constant properties or expressions for water properties and it seems to be ok. The question which is bugging me is, "Inspite of the usage of IAPWS library being very difficult for convergence, why should it produce unphysical results when using Bounded Central Differencing Scheme for LES WALE model? But, when I switched to upwind scheme for only energy equations, then the temperature variation is within the defined limits of my simulation." Is it a problem related to my mesh or is it something else?

June 12, 2013, 07:07		#6
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,871 Rep Power: 144	The unphysical results from IAPWS is due to it not converging, or having other numerical instabilities which cause it to result in unphysical results. To get IAPWS to work you are going to need to improve numerical stability -which means better mesh quality, smaller time steps, a better initial condition and/or double precision numerics.

July 23, 2017, 16:33		#9
selvam2487 Member Karthik Join Date: Oct 2012 Location: Germany Posts: 53 Rep Power: 14	Dear Friend, Firstly, my sincere apologies for the delayed reply. Moving on to your question, I have not used constant properties since there is a significant temperature difference between fluids in my calculations. So I finally used IAPWS library for defining water properties and performed all my LES calculations (with WALE sub-grid model) with IAPWS library for water properties. In terms of numerical scheme, I ended up using High resolution scheme in CFX since CD or BCD ended up producing unphysical oscillations in terms of temperature. May be it has something to do with the CFX software code but I don't know. When I tried the exact thing in FLUENT using BCD, it was working. Hope it answers your question.