Strange behavior of velocity residuals

skullCFD · April 28, 2011, 18:47

I have a converged solution for a conjugate heat transfer problem. I changed a property (thermal conductivity) in the solid and it started to solve the velocity field in the fluid again i.e., the velocity residuals rose back to 1e-2 from less than 1e-6! When all the properties of the fluid that I have are independent of temperature, should it not simply solve the energy equation in the entire domain and let the velocity residuals in the fluid to continuously go down starting from 1e-6 in this new simulation. Could anyone please help me know what is happening? Thanks in advance.

meangreen · May 5, 2011, 15:24

I have noticed fluent does this with a lot of things. Even if you change the URFs, you will see them spike. I think it is because fluent clears all unnecessary information while solving so that your computer memory doesn't get eaten up. I think it just has to recalibrate everything once something is changed.

Have you checked to make sure the actual value of your velocities hasn't changed? You could also try implementing the command to keep fluent from clearing the memory and see what happens? I think in the end it isn't something to worry about.

skullCFD · June 3, 2011, 12:53

Thanks, meangreen. I am waiting for Fluent customer support to respond and they are still working on understanding it. I checked the actual velocity values and they changed by about 1%. As of now, what you mentioned about the memory clean up makes the best sense to me. It takes about a month for a single simulation to converge in my case (high velocity separation flow in the transitional regime; ~6M cells) and I was depressed to see it trying to converge the velocity field all over again just because I changed a property of the solid. Could you please mention how to prevent Fluent from clearing the memory? Thank you.

April 28, 2011, 18:47	Strange behavior of velocity residuals	#1
skullCFD Member Join Date: Apr 2011 Posts: 30 Rep Power: 15	I have a converged solution for a conjugate heat transfer problem. I changed a property (thermal conductivity) in the solid and it started to solve the velocity field in the fluid again i.e., the velocity residuals rose back to 1e-2 from less than 1e-6! When all the properties of the fluid that I have are independent of temperature, should it not simply solve the energy equation in the entire domain and let the velocity residuals in the fluid to continuously go down starting from 1e-6 in this new simulation. Could anyone please help me know what is happening? Thanks in advance.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
UDF error - parabolic velocity profile - 3D turbine	Zaqie	Fluent UDF and Scheme Programming	9	June 25, 2016 20:08
Velocity profiles problem behind the elbow (3D problem)	kabat73	FLUENT	8	May 9, 2010 05:26
Velocity in Porous medium : HELP! HELP! HELP!	Kali Sanjay	Phoenics	0	November 6, 2006 07:10
How to calculate Velocity?	Kali Sanjay	Phoenics	1	May 1, 2006 14:57
Strange oscillating velocity	zonexo	Main CFD Forum	2	April 6, 2006 12:38

May 5, 2011, 15:24		#2
meangreen Member Casey Join Date: Jun 2009 Posts: 98 Rep Power: 17	I have noticed fluent does this with a lot of things. Even if you change the URFs, you will see them spike. I think it is because fluent clears all unnecessary information while solving so that your computer memory doesn't get eaten up. I think it just has to recalibrate everything once something is changed. Have you checked to make sure the actual value of your velocities hasn't changed? You could also try implementing the command to keep fluent from clearing the memory and see what happens? I think in the end it isn't something to worry about.

June 3, 2011, 12:53		#3
skullCFD Member Join Date: Apr 2011 Posts: 30 Rep Power: 15	Thanks, meangreen. I am waiting for Fluent customer support to respond and they are still working on understanding it. I checked the actual velocity values and they changed by about 1%. As of now, what you mentioned about the memory clean up makes the best sense to me. It takes about a month for a single simulation to converge in my case (high velocity separation flow in the transitional regime; ~6M cells) and I was depressed to see it trying to converge the velocity field all over again just because I changed a property of the solid. Could you please mention how to prevent Fluent from clearing the memory? Thank you.