convection from rotating cylinder without crossflow

prabulchandran · June 6, 2016, 05:41

i am modelling convection from rotating cylinder in CFX at a reynolds number of 22000 which comes within the mixed convection regime. bouyancy is activated . i observe that my residual values go down upto 10-4 and the solution converges. but as i set my convergence criteria to 10-5 the residues rise unacceptably and solution looks bizzare.why could this be happening

ghorrocks · June 6, 2016, 06:23

This is related to FAQ: http://www.cfd-online.com/Wiki/Ansys...gence_criteria

prabulchandran · June 8, 2016, 04:50

Does making the smallest element too small distort the results. While doing grid independence the solution doesnt seem to stop, goes on varying on and on. iam doing a natural convection problem

ghorrocks · June 8, 2016, 05:58

Quote:

Does making the smallest element too small distort the results.

As the mesh gets finer you will eventually reach the point where numerical round-off errors stop convergence.

Quote:

While doing grid independence the solution doesnt seem to stop, goes on varying on and on.

This is saying:
1) your mesh is miles too coarse, you need it much finer. That is why lots of the worlds supercomputers are doing CFD. OR

2) Your modelling approach is wrong. A better approach would allow convergence with a coarser mesh.

prabulchandran · June 10, 2016, 05:40

The cylinder is located in free space.i have used pressure inlet and outlet boundary conditions for the simulation. i also tried with wall boundary conditions with the walls located very far away from cylinder and set to ambient temperature. I find that using the wall boundary condition gives results which are more close to the experimental ones . But i am worried about the air recirculating within the domain. is it appropriate to use wall boundary conditions when simulating such a case ??

ghorrocks · June 10, 2016, 06:22

If you have done a sensitivity check to ensure that the proximity of the walls does not affect the results then yes, you can use them. But if you are trying to model a far field (ie: free air) then the normal way of doing this is with an inlet/outlet pair so if you cannot get convergence with this then it is very unlikely a wall boundary will have converged. A wall boundary should converge at a larger distance away than an inlet/outlet pair as it is less like a free field condition.

June 6, 2016, 05:41	convection from rotating cylinder without crossflow	#1
prabulchandran New Member Prabul Chandran Join Date: Nov 2015 Posts: 12 Rep Power: 10	i am modelling convection from rotating cylinder in CFX at a reynolds number of 22000 which comes within the mixed convection regime. bouyancy is activated . i observe that my residual values go down upto 10-4 and the solution converges. but as i set my convergence criteria to 10-5 the residues rise unacceptably and solution looks bizzare.why could this be happening

June 8, 2016, 04:50	Size of the smallest element	#3
prabulchandran New Member Prabul Chandran Join Date: Nov 2015 Posts: 12 Rep Power: 10	Does making the smallest element too small distort the results. While doing grid independence the solution doesnt seem to stop, goes on varying on and on. iam doing a natural convection problem

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Mesh type in rotating blade inside cylinder	Mah.tavana	Mesh Generation & Pre-Processing	0	July 26, 2014 06:53
Turbulent natural convection in a veritcal cylinder	m.vegad	Main CFD Forum	0	March 28, 2014 03:54
Air natural convection inside a vertical cylinder	Pavolo	FLUENT	3	December 12, 2012 06:00
forced convection on a cylinder	mountaineer	OpenFOAM	3	September 27, 2011 10:30
Fluid in a rotating cylinder	Fluid Novice	Main CFD Forum	1	December 17, 1998 17:25

June 6, 2016, 06:23		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,703 Rep Power: 143	This is related to FAQ: http://www.cfd-online.com/Wiki/Ansys...gence_criteria

June 10, 2016, 05:40		#5
prabulchandran New Member Prabul Chandran Join Date: Nov 2015 Posts: 12 Rep Power: 10	The cylinder is located in free space.i have used pressure inlet and outlet boundary conditions for the simulation. i also tried with wall boundary conditions with the walls located very far away from cylinder and set to ambient temperature. I find that using the wall boundary condition gives results which are more close to the experimental ones . But i am worried about the air recirculating within the domain. is it appropriate to use wall boundary conditions when simulating such a case ??

June 10, 2016, 06:22		#6
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,703 Rep Power: 143	If you have done a sensitivity check to ensure that the proximity of the walls does not affect the results then yes, you can use them. But if you are trying to model a far field (ie: free air) then the normal way of doing this is with an inlet/outlet pair so if you cannot get convergence with this then it is very unlikely a wall boundary will have converged. A wall boundary should converge at a larger distance away than an inlet/outlet pair as it is less like a free field condition.