unsteady convergence problem for bluff body

Byukuya · October 19, 2010, 17:11

Hi people,

I have been investigating flow around a bluff body (building-3D simulations). I firstly made steady calculations. I started iterating with 1st order and 2nd order. After some thousand iterations my residuals started oscillating. Then i changed to make unsteady calculations. For the time step choosing i obtained it from the CFL number. When my calculations finished in the time i wanted, i checked the residuals, they were oscilating beween 10-6 and 10-12. When it comes for the results, for example when i check the contours there isnt any change with the time, also any values didnt changed when i plot them. My question is while solving unsteady shouldnt we observe any differences between time periods? My problem has to be unsteady. Am i missing something? Can anyone help me please,

Regards,

Byukuya

sreerao · October 20, 2010, 06:41

Hi

Well, Unsteady should show some fluctuations as you expected. However, it may mainly depends on what turbulence models you choose for the problem. For instacne, std k-e model may not be suitable for complex problems which has some wake and seperation. Even the order does not show you much for for the chosen problem.

Try to use steady to get the flow features correct. I mean boundary layers or pressure or flow around you building first. You may try differnt turbulence models (check some literature for suitable turbulece models). Once you established then try unsteady simulations. You may see some differnce.

You mentioned CFL number in you question. How did you get this value? I presume you are using pressure based solver. did you mean cell courant number?

Regards
Sree

Byukuya · October 28, 2010, 11:59

Hello,

Thank you for answering my problem.

Firstly i have chosen many different turbulence models and they show similar things and i also checked other people works but i couldnt see differences between my conditions except they see the unsteadiness. For the cfl you ask i meant the courant number which calculated below,

CFL = (sound speed + inlet speed)* time size / mesh size

I choose CFL =1 and calculate the time

October 19, 2010, 17:11	unsteady convergence problem for bluff body	#1
Byukuya New Member Join Date: Mar 2010 Posts: 8 Rep Power: 16	Hi people, I have been investigating flow around a bluff body (building-3D simulations). I firstly made steady calculations. I started iterating with 1st order and 2nd order. After some thousand iterations my residuals started oscillating. Then i changed to make unsteady calculations. For the time step choosing i obtained it from the CFL number. When my calculations finished in the time i wanted, i checked the residuals, they were oscilating beween 10-6 and 10-12. When it comes for the results, for example when i check the contours there isnt any change with the time, also any values didnt changed when i plot them. My question is while solving unsteady shouldnt we observe any differences between time periods? My problem has to be unsteady. Am i missing something? Can anyone help me please, Regards, Byukuya

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Any body help me!, Impeller simulation problem.	Desperado	FLUENT	2	June 9, 2009 14:04
Continuity eq convergence problem	carno	FLUENT	4	February 8, 2008 05:04
Convergence problem	Roland	CFX	12	June 20, 2006 22:48
convergence problem	Trushar	Phoenics	5	August 28, 2002 00:40
convergence problem with SIMPLER	NURAY KAYAKOL	Main CFD Forum	1	February 24, 1999 14:43

October 20, 2010, 06:41		#2
sreerao New Member sreenivasa rao gubba Join Date: Sep 2010 Location: Leeds University, UK Posts: 28 Rep Power: 16	Hi Well, Unsteady should show some fluctuations as you expected. However, it may mainly depends on what turbulence models you choose for the problem. For instacne, std k-e model may not be suitable for complex problems which has some wake and seperation. Even the order does not show you much for for the chosen problem. Try to use steady to get the flow features correct. I mean boundary layers or pressure or flow around you building first. You may try differnt turbulence models (check some literature for suitable turbulece models). Once you established then try unsteady simulations. You may see some differnce. You mentioned CFL number in you question. How did you get this value? I presume you are using pressure based solver. did you mean cell courant number? Regards Sree

October 28, 2010, 11:59		#3
Byukuya New Member Join Date: Mar 2010 Posts: 8 Rep Power: 16	Hello, Thank you for answering my problem. Firstly i have chosen many different turbulence models and they show similar things and i also checked other people works but i couldnt see differences between my conditions except they see the unsteadiness. For the cfl you ask i meant the courant number which calculated below, CFL = (sound speed + inlet speed)* time size / mesh size I choose CFL =1 and calculate the time