CFD Online Logo CFD Online URL
www.cfd-online.com
[Sponsors]
Home > Forums > FLUENT

QUICK scheme oscillation

Register Blogs Members List Search Today's Posts Mark Forums Read

Reply
 
LinkBack Thread Tools Display Modes
Old   March 30, 2012, 14:39
Default QUICK scheme oscillation
  #1
New Member
 
kong
Join Date: Sep 2011
Posts: 21
Rep Power: 5
kongl1986 is on a distinguished road
i am running gas-solid flow in e-e.
all the discretization are selected to second order upwind, except volume fraction.
When i use first order upwind scheme of volume fraction, the result got converged.
However, when i change the volume fraction to QUICK, it can not get converged.
I plotted the volume fraction, there is a big oscillation along the height.
Can any one give me some suggestions? what i should improve to solve this oscillation under QUICK scheme

thank you
kongl1986 is offline   Reply With Quote

Old   March 31, 2012, 16:02
Default
  #2
Senior Member
 
Lucky Tran
Join Date: Apr 2011
Location: Orlando, FL USA
Posts: 594
Rep Power: 11
LuckyTran is on a distinguished road
Quote:
Originally Posted by kongl1986 View Post
i am running gas-solid flow in e-e.
all the discretization are selected to second order upwind, except volume fraction.
When i use first order upwind scheme of volume fraction, the result got converged.
However, when i change the volume fraction to QUICK, it can not get converged.
I plotted the volume fraction, there is a big oscillation along the height.
Can any one give me some suggestions? what i should improve to solve this oscillation under QUICK scheme

thank you
kong, this is a typical and expected behavior of higher order schemes (2nd order and up). You trade stability for improved accuracy. Not much you can do except to investigate the reason for the oscillation.

Have you tried lowering our under-relaxation factors? Doing so may allow the solution to change more gradually. I do not recommend this method myself, but I am throwing it up as an option to try. Fluent 14 and (possibly 13 also) has a higher-order term relaxation option that is essentially an under-relaxation term on your higher order discretization. I recommend trying this first if you have it available. I do not recommend lowering the base under-relaxation factors, especially if you have HOD available.

There are few things you can do for an easy solution. The oscillation may be enhanced by a inherent instability in the flow or one of the models used. Even a very fine mesh can produce these oscillations, it all depends on the flow. The NS equations inherently lead to instabilities (turbulence!) so it is really not too surprising.

A brute force method is to run an unsteady simulation and start collecting statistics, but that is just avoiding having to figure out what is the meaning of the oscillation (in the steady sense). Doing the unsteady simulation might help you find the reason for the oscillation.
LuckyTran is offline   Reply With Quote

Old   April 4, 2012, 15:36
Default
  #3
New Member
 
kong
Join Date: Sep 2011
Posts: 21
Rep Power: 5
kongl1986 is on a distinguished road
Quote:
Originally Posted by LuckyTran View Post
kong, this is a typical and expected behavior of higher order schemes (2nd order and up). You trade stability for improved accuracy. Not much you can do except to investigate the reason for the oscillation.

Have you tried lowering our under-relaxation factors? Doing so may allow the solution to change more gradually. I do not recommend this method myself, but I am throwing it up as an option to try. Fluent 14 and (possibly 13 also) has a higher-order term relaxation option that is essentially an under-relaxation term on your higher order discretization. I recommend trying this first if you have it available. I do not recommend lowering the base under-relaxation factors, especially if you have HOD available.

There are few things you can do for an easy solution. The oscillation may be enhanced by a inherent instability in the flow or one of the models used. Even a very fine mesh can produce these oscillations, it all depends on the flow. The NS equations inherently lead to instabilities (turbulence!) so it is really not too surprising.

A brute force method is to run an unsteady simulation and start collecting statistics, but that is just avoiding having to figure out what is the meaning of the oscillation (in the steady sense). Doing the unsteady simulation might help you find the reason for the oscillation.
thank you for ur reply LuckyTran.
Actually, i am now using unsteady simulation.
For under relaxation number , do you have any suggestion how to choose or lower how much would be ok? the iteration can got converged each time step, only the solution is oscillating in a large range.
thank you so much for your reply again
kongl1986 is offline   Reply With Quote

Old   April 4, 2012, 16:05
Default
  #4
Senior Member
 
Lucky Tran
Join Date: Apr 2011
Location: Orlando, FL USA
Posts: 594
Rep Power: 11
LuckyTran is on a distinguished road
Quote:
Originally Posted by kongl1986 View Post
thank you for ur reply LuckyTran.
Actually, i am now using unsteady simulation.
For under relaxation number , do you have any suggestion how to choose or lower how much would be ok? the iteration can got converged each time step, only the solution is oscillating in a large range.
thank you so much for your reply again
lowering under relaxation factors by a factor of 1/2 or 1/3 is common when running into issues such as you are facing. If possible, it is better to not change URF at all.

For unsteady simulation, rather than lowering URF, it is easier & better to just make the time-step smaller without touching URF.

The unsteady simulation will not make the oscillation (in time) go away obviously, but per time-step you should be able to get converged results.

Last edited by LuckyTran; April 4, 2012 at 16:27.
LuckyTran is offline   Reply With Quote

Old   April 4, 2012, 16:22
Default
  #5
New Member
 
kong
Join Date: Sep 2011
Posts: 21
Rep Power: 5
kongl1986 is on a distinguished road
Quote:
Originally Posted by LuckyTran View Post
lowering under relaxation factors by a factor of 1/2 or 1/3 is common when running into issues such as you are facing. If possible, it is better to not change URF at all.

For unsteady simulation, rather than lowering URF, it is easier & better to just make the time-step smaller.

The unsteady simulation will not make the oscillation (in time) go away obviously, but per time-step you should be able to get converged results.
thank you so much LuckyTran.
I will try smaller time step next. and will let you know whether it works.
kongl1986 is offline   Reply With Quote

Reply

Tags
oscillation, scheme

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are On


Similar Threads
Thread Thread Starter Forum Replies Last Post
Upwind, central, QUICK scheme for Re~10000 quarkz Main CFD Forum 4 June 23, 2011 15:04
Implementation of QUICK scheme Romuald Skoda Main CFD Forum 10 August 3, 2010 12:23
QUICK Scheme mmahdinia OpenFOAM Running, Solving & CFD 3 December 9, 2009 17:30
Oscillation of a upwind scheme Mehdi Main CFD Forum 4 June 26, 2003 10:47
QUICK scheme kim FLUENT 1 August 29, 2002 10:16


All times are GMT -4. The time now is 00:21.