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

Residuals querry

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

Reply
 
LinkBack Thread Tools Search this Thread Display Modes
Old   March 24, 2010, 09:56
Default Residuals querry
  #1
Senior Member
 
Mohsin Mukhtar
Join Date: Mar 2010
Location: South Korea
Posts: 249
Rep Power: 17
Mohsin is on a distinguished road
Hello

My residuals are set to 10^-4 for continuity, x,y,z, and turbulent dissipation rate. while doing iterations the residuals goes down just near 10^-4 after 1000 iterations and again moves upwards towards 10^-2 and then doesn't come back.

Is it ok If i stop after 1000 iterations or I should stop the iterations when I m getting constant results (like straight line and unchanged residuals).

Pelase guide me
Mohsin is offline   Reply With Quote

Old   March 25, 2010, 03:01
Default
  #2
Super Moderator
 
-mAx-'s Avatar
 
Maxime Perelli
Join Date: Mar 2009
Location: Switzerland
Posts: 3,297
Rep Power: 41
-mAx- will become famous soon enough
No you cannot stop iterating after 1000 iterations.
If it drops towards, then there is a problem. You have to find why...
__________________
In memory of my friend Hervé: CFD engineer & freerider
-mAx- is offline   Reply With Quote

Old   March 25, 2010, 03:27
Default
  #3
Senior Member
 
Mohsin Mukhtar
Join Date: Mar 2010
Location: South Korea
Posts: 249
Rep Power: 17
Mohsin is on a distinguished road
You mean to say that I have to find out why the residuals move from 10^-5 to 10^-3 and goes up and up until 10^-2.

How to find it...any suggestions?

For Velocity, Turbulence What parameters are neeeded to be seen?
Modifying Geometry or mesh? what can be done?

Thanks
Mohsin is offline   Reply With Quote

Old   March 25, 2010, 04:12
Default
  #4
Super Moderator
 
-mAx-'s Avatar
 
Maxime Perelli
Join Date: Mar 2009
Location: Switzerland
Posts: 3,297
Rep Power: 41
-mAx- will become famous soon enough
*Is your mesh ok?
grid/check
*check if the mass continuity is conserved (massflow in = massflow out)
...
It is typical debugging, and it's a part of CFD.
Everybody already had and will have to deal with it
__________________
In memory of my friend Hervé: CFD engineer & freerider
-mAx- is offline   Reply With Quote

Old   March 25, 2010, 04:31
Default
  #5
Senior Member
 
Mohsin Mukhtar
Join Date: Mar 2010
Location: South Korea
Posts: 249
Rep Power: 17
Mohsin is on a distinguished road
Flux report indicate the mass flow rate of all the boundaries= -6.5*10^-7 kg/s. So the continuity seems to conserve.

What else I can see?
Mohsin is offline   Reply With Quote

Old   March 25, 2010, 04:55
Default
  #6
Super Moderator
 
-mAx-'s Avatar
 
Maxime Perelli
Join Date: Mar 2009
Location: Switzerland
Posts: 3,297
Rep Power: 41
-mAx- will become famous soon enough
-what is the max skewness of your mesh?
-what kind of BC did you used
-is your massflow or pressure at inlet in agreement with your expectations?
etc...
__________________
In memory of my friend Hervé: CFD engineer & freerider
-mAx- is offline   Reply With Quote

Old   March 25, 2010, 05:13
Default
  #7
Senior Member
 
Mohsin Mukhtar
Join Date: Mar 2010
Location: South Korea
Posts: 249
Rep Power: 17
Mohsin is on a distinguished road
Max skewness = 0.89
At inlet = velocity Inlet
At outlets = Pressure Outlet
Others are wall and fluid.

Ya they are in accordance with my expectations.

Really exhasted
Mohsin is offline   Reply With Quote

Old   March 25, 2010, 06:07
Default
  #8
Super Moderator
 
-mAx-'s Avatar
 
Maxime Perelli
Join Date: Mar 2009
Location: Switzerland
Posts: 3,297
Rep Power: 41
-mAx- will become famous soon enough
*did you compared the results for iterations = 10000, and then for iterations where the residuals drop?
*compressible/uncompressible?
*what kinds of solver/schema/turbulence model
__________________
In memory of my friend Hervé: CFD engineer & freerider
-mAx- is offline   Reply With Quote

Old   March 26, 2010, 03:35
Default
  #9
Senior Member
 
Mohsin Mukhtar
Join Date: Mar 2010
Location: South Korea
Posts: 249
Rep Power: 17
Mohsin is on a distinguished road
Sorry fro late reply as i was comparing the results of iterations.

At 2500 iterations I got better results as compared with doing 10,000 iterations (at 10,000 iteration the residual plot was constant and was not fluctuating).

I think we should look at the point of minimum in residual plot. After seeing at the residual plots of 10,000 iterations I initialized the calculations and gave the number of iterations until the minumum value (2500 iterations) which i saw from 10,000 iteration plot and got the better result as compared to doing 10,000 iterations.

I m using the following:

Pressure based solver
Incompressible
K-E 2 equation Realizable model
DPM model

Whats your point of view?
Mohsin is offline   Reply With Quote

Old   March 26, 2010, 04:28
Default
  #10
Super Moderator
 
-mAx-'s Avatar
 
Maxime Perelli
Join Date: Mar 2009
Location: Switzerland
Posts: 3,297
Rep Power: 41
-mAx- will become famous soon enough
ah ok you are running DPM model.
I have 0-experience with it, so I won't be helpful from this side.
Sorry.
I would check tutorial about DPM, if one exists...
__________________
In memory of my friend Hervé: CFD engineer & freerider
-mAx- is offline   Reply With Quote

Old   March 26, 2010, 05:00
Default
  #11
Senior Member
 
Mohsin Mukhtar
Join Date: Mar 2010
Location: South Korea
Posts: 249
Rep Power: 17
Mohsin is on a distinguished road
http://my.fit.edu/itresources/manual...pdf/ug/pdf.htm

This is Fluent's User guide Chapter 22 is about DPM modeling.

But i was asking generally about the residuals behaviour. once it starts to move up can we stop it there or we should wait for it until it becomes constant.

In my case it became constant until 7000 iterations. But i was getting better results at 2500 iterations. Does it allways supposed to go towards convergence(means the residual plot contnually goes on decreasing), If not then whether something is wrong or its normal.

Please guide
Mohsin is offline   Reply With Quote

Old   March 26, 2010, 05:07
Default
  #12
Super Moderator
 
-mAx-'s Avatar
 
Maxime Perelli
Join Date: Mar 2009
Location: Switzerland
Posts: 3,297
Rep Power: 41
-mAx- will become famous soon enough
No the residuals should decrease and then stay constant. But they are not suppose to drop towards.
Monitor the pressure at your inlet, and check its convergence
__________________
In memory of my friend Hervé: CFD engineer & freerider
-mAx- is offline   Reply With Quote

Old   March 26, 2010, 05:13
Default
  #13
Senior Member
 
Mohsin Mukhtar
Join Date: Mar 2010
Location: South Korea
Posts: 249
Rep Power: 17
Mohsin is on a distinguished road
it means that it should always decrease then remain constant. In my case it first decreased to 10^-5 until 2500 iterations then started to increase until 10^-3 and then remained constant until 10,000 iterations.

So it means that there is some problem somwhere that originates at 2500th iteration which casues my solution to diverge at that point.

I did not specify pressure at the inlet (velcity inlet) . its 0. I just specified the operating pressure.

my Conditions

Max skewness = 0.89
At inlet = velocity Inlet
At outlets = Pressure Outlet
Others are wall and fluid.

??
Mohsin is offline   Reply With Quote

Old   March 26, 2010, 05:44
Default
  #14
Super Moderator
 
-mAx-'s Avatar
 
Maxime Perelli
Join Date: Mar 2009
Location: Switzerland
Posts: 3,297
Rep Power: 41
-mAx- will become famous soon enough
if you set a velocity at your pressure, the solver will compute the pressure inlet.
Monitor it, to see if the computed pressure converges or not
__________________
In memory of my friend Hervé: CFD engineer & freerider
-mAx- is offline   Reply With Quote

Old   March 26, 2010, 10:01
Default
  #15
Senior Member
 
Mohsin Mukhtar
Join Date: Mar 2010
Location: South Korea
Posts: 249
Rep Power: 17
Mohsin is on a distinguished road
Welocity is different as pressure. I have just specified the operating pressure of 1 atmosphere. My inlet is velocity inlet whose velocity is 6m/s. However i used pressure outlet at the outlet and there i specified a guage pressure of 0 Pa. As in my case there is no reaction so Inlet and outlet pressures are same. So I dont think i need to specify it.

What do u thinK?

I bothered you so much. Thanks for ur help.
Mohsin is offline   Reply With Quote

Old   March 26, 2010, 10:28
Default
  #16
Super Moderator
 
-mAx-'s Avatar
 
Maxime Perelli
Join Date: Mar 2009
Location: Switzerland
Posts: 3,297
Rep Power: 41
-mAx- will become famous soon enough
without talking about DPM, if you have a massflow, then you have a drop pressure.
As your outlet-pressure is fixed (your BC), then the drop pressure will be dp= pressure_inlet - pressure_outlet --> that means fluent will adjust the pressure at your inlet.
Maybe with DPM it is different...
__________________
In memory of my friend Hervé: CFD engineer & freerider
-mAx- is offline   Reply With Quote

Old   March 26, 2010, 10:34
Default
  #17
Senior Member
 
Mohsin Mukhtar
Join Date: Mar 2010
Location: South Korea
Posts: 249
Rep Power: 17
Mohsin is on a distinguished road
I dont know the pressure at outlet. I just know the pressure at inlet which i included in operating pressure. 1 atm + 7 atm-g. I couldnt specify the pressure at inlet coz of velocity inlet boundary condition and on the outlet i dont know the pressure so i just let it be 0 guage at the outlet (the outlet pressure is with relative to operating pressure so as it is 0 it means that outlet pressure at the outlet is 1 atm +7 atm.g.

Thats what i know about the pressure. If u can give me some additional information i will be thankful.
Mohsin is offline   Reply With Quote

Old   March 26, 2010, 10:36
Default
  #18
Senior Member
 
Mohsin Mukhtar
Join Date: Mar 2010
Location: South Korea
Posts: 249
Rep Power: 17
Mohsin is on a distinguished road
Also please tell me. I am using parallal processing on windows 7 with 1TB hard disk and 4 GB ram Intel core quad 4 cores. So i m using fluent 3d -t4 in CMD but still it is slow. At first it was fast but suddenly it became very slow. I restarted my computer but still the same result. What may have caused this to happen?
Mohsin is offline   Reply With Quote

Old   March 26, 2010, 14:30
Default
  #19
Super Moderator
 
-mAx-'s Avatar
 
Maxime Perelli
Join Date: Mar 2009
Location: Switzerland
Posts: 3,297
Rep Power: 41
-mAx- will become famous soon enough
Quote:
Originally Posted by Mohsin View Post
I dont know the pressure at outlet. I just know the pressure at inlet which i included in operating pressure. 1 atm + 7 atm-g. I couldnt specify the pressure at inlet coz of velocity inlet boundary condition and on the outlet i dont know the pressure so i just let it be 0 guage at the outlet (the outlet pressure is with relative to operating pressure so as it is 0 it means that outlet pressure at the outlet is 1 atm +7 atm.g.
If you don't know the pressure at your outlet, then you set at it at 0. You give the massflow at outlet, and fluent computes the pressure at inlet (eg: your drop pressure). Finally your "real" pressure at outlet will be <the pressure inlet you know (1 atm + 7 atm-g)> - <the pressure inlet computed from fluent>

Quote:
Originally Posted by Mohsin View Post
Also please tell me. I am using parallal processing on windows 7 with 1TB hard disk and 4 GB ram Intel core quad 4 cores. So i m using fluent 3d -t4 in CMD but still it is slow. At first it was fast but suddenly it became very slow. I restarted my computer but still the same result. What may have caused this to happen?
Do a test with another case (a basic one), maybe it comes from your model which has some convergence issue
__________________
In memory of my friend Hervé: CFD engineer & freerider
-mAx- is offline   Reply With Quote

Old   March 26, 2010, 14:41
Default
  #20
Senior Member
 
Mohsin Mukhtar
Join Date: Mar 2010
Location: South Korea
Posts: 249
Rep Power: 17
Mohsin is on a distinguished road
I changed my opearting system to 32 bit windows 7. It is still slow so it was not the operating systmes problem. I m using 0.4 million cells with parallel processing of 4 cores fluent 3d -t4 but still it is slow. earlier it was fast but i donty know...Its the basic case which i m doing but its ok.

Also The mass flow rate i cannot give at the outlet coz this is what i have to find. Basically i have to know the concentration of particles at the outlet so that they are uniformly distributed among the 3 outlets of the cylinder. I will give non uniform flow at the inlet and check whether the particles can be uniform at the outlet. So i cant give mass flow rate.

My result with a simple case (when the flow at the inlet is also uniform) goes down on the residual plot until 10^-5 and then goes up (as u said it shouldn't go up) but it goes up and then becomes constant. Ssshhh
Mohsin is offline   Reply With Quote

Reply

Thread Tools Search this Thread
Search this Thread:

Advanced Search
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 Off
Pingbacks are On
Refbacks are On


Similar Threads
Thread Thread Starter Forum Replies Last Post
[Tutorials] Tutorial of how to plot residuals ! wolle1982 OpenFOAM Community Contributions 171 February 20, 2024 03:55
what to monitor besides residuals? franzdrs FLUENT 5 March 21, 2013 04:59
Mesh size and solver residuals... Scott CFX 5 December 15, 2008 18:10
Convergence - scaled vs unscaled residuals HS FLUENT 1 November 7, 2005 06:45
residuals Karl CFX 2 June 25, 2003 00:13


All times are GMT -4. The time now is 03:02.