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

When to stop refining mesh

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

Like Tree3Likes
  • 1 Post By FMDenaro
  • 1 Post By FMDenaro
  • 1 Post By FMDenaro

Reply
 
LinkBack Thread Tools Search this Thread Display Modes
Old   April 11, 2016, 04:26
Default When to stop refining mesh
  #1
Member
 
Join Date: Aug 2014
Location: Germany
Posts: 77
Rep Power: 11
harry123 is on a distinguished road
Hi,

What are the criterion for mesh refinements ?

Since that question sounds quite broad, if we consider a specific problem like a jet flow emerging from 1mm nozzle into atmosphere, how to determine the most apt mesh size ? For this particular problem, I would be refining the jet flow region to have finer mesh and coarser mesh near far field boundaries. But how much smaller should be fine region ? For a 1mm jet radius, how to determine the mesh size ?

Are there any specific methods for determining mesh refinement in particular regions of the domain other than by trial and error ?

Harry
harry123 is offline   Reply With Quote

Old   April 11, 2016, 04:37
Default
  #2
Senior Member
 
Filippo Maria Denaro
Join Date: Jul 2010
Posts: 6,762
Rep Power: 71
FMDenaro has a spectacular aura aboutFMDenaro has a spectacular aura aboutFMDenaro has a spectacular aura about
Quote:
Originally Posted by harry123 View Post
Hi,

What are the criterion for mesh refinements ?

Since that question sounds quite broad, if we consider a specific problem like a jet flow emerging from 1mm nozzle into atmosphere, how to determine the most apt mesh size ? For this particular problem, I would be refining the jet flow region to have finer mesh and coarser mesh near far field boundaries. But how much smaller should be fine region ? For a 1mm jet radius, how to determine the mesh size ?

Are there any specific methods for determining mesh refinement in particular regions of the domain other than by trial and error ?

Harry

I think you are considering the searching for a grid-independent solution in a practical flow case (grid refinement for accuracy order analysis is differently analysed).
If you are modelling turbulence in RANS formulation, the requirement for describing a boundary layer thickness give you the greater mesh size. Then you can refine the grid by checking for the variation in specific flow quantities, for example a velocity profile at different stations along the jet.
Note that often, the RANS model has greater magnitude than the truncation error that you want to diminuish by grid refinement. Therefore, starting from an initial good grid, you should obtain rapidly a grid-independent solution
harry123 likes this.
FMDenaro is offline   Reply With Quote

Old   April 11, 2016, 07:10
Default
  #3
Member
 
Join Date: Aug 2014
Location: Germany
Posts: 77
Rep Power: 11
harry123 is on a distinguished road
Thank you very much, I wasn't familiar with the term 'grid-independent'. I looked it up and thats what I was looking for.

Can you suggest any book or literature which discuss this issue in further detail, I'm doing a RANS simulation now and would like to try LES also. Since there are many other simulation parameters from time step size to various turbulence model coefficients I would have to choose, it would be nice if I could read about all that from somewhere.

Thanks,
Harry
harry123 is offline   Reply With Quote

Old   April 11, 2016, 08:01
Default
  #4
Senior Member
 
Filippo Maria Denaro
Join Date: Jul 2010
Posts: 6,762
Rep Power: 71
FMDenaro has a spectacular aura aboutFMDenaro has a spectacular aura aboutFMDenaro has a spectacular aura about
RANS and LES drive to different analyses!

- RANS is a steady solution, LES is unsteady
- RANS has zeoth-order statistics while LES has high order statistics
- RANS has grid independent solution while LES tends towards DNS for refined grids (at least for implicit filtering).
davidwilcox likes this.
FMDenaro is offline   Reply With Quote

Old   April 14, 2016, 05:03
Default
  #5
Member
 
Join Date: Aug 2014
Location: Germany
Posts: 77
Rep Power: 11
harry123 is on a distinguished road
Excuse me in case my questions are bit naive..

I understand why you said RANS to be steady, but what happens if a RANS simulation is run as unsteady ? I had run a jet flow simulation using steady and unsteady for RANS. The results from steady were not correct. What could have gone wrong ?
harry123 is offline   Reply With Quote

Old   April 14, 2016, 05:24
Default
  #6
Member
 
Join Date: Aug 2014
Location: Germany
Posts: 77
Rep Power: 11
harry123 is on a distinguished road
I think, I know why my steady simulation wasnt working. I was using a VOF method and guess it works only under unsteady
harry123 is offline   Reply With Quote

Old   April 14, 2016, 06:00
Default
  #7
Senior Member
 
Filippo Maria Denaro
Join Date: Jul 2010
Posts: 6,762
Rep Power: 71
FMDenaro has a spectacular aura aboutFMDenaro has a spectacular aura aboutFMDenaro has a spectacular aura about
Quote:
Originally Posted by harry123 View Post
Excuse me in case my questions are bit naive..

I understand why you said RANS to be steady, but what happens if a RANS simulation is run as unsteady ? I had run a jet flow simulation using steady and unsteady for RANS. The results from steady were not correct. What could have gone wrong ?

- RANS formulation is steady by definition.
- U(nsteady)RANS is unsteady provided that an unsteady external force exists. It has some difference in the modelling.

you should clarify why from your solution you get the conclusion that is wrong
FMDenaro is offline   Reply With Quote

Old   April 14, 2016, 07:28
Default
  #8
Member
 
Join Date: Aug 2014
Location: Germany
Posts: 77
Rep Power: 11
harry123 is on a distinguished road
I plot the variation of volume fraction of water along the centerline jet axis to check result correctness. When I used RANS with steady for time scheme, this graph does not look good,
Thats why I came to conclusion steady simulation was not working. Also, I read in the Star CCM user guide that VOF simulations are unsteady as the interface phenomenon cannot be captured properly while using a steady state simulation.
harry123 is offline   Reply With Quote

Old   April 14, 2016, 08:02
Default
  #9
Senior Member
 
Filippo Maria Denaro
Join Date: Jul 2010
Posts: 6,762
Rep Power: 71
FMDenaro has a spectacular aura aboutFMDenaro has a spectacular aura aboutFMDenaro has a spectacular aura about
do you have experimental or computational results for comparison?
what about the main variables?
FMDenaro is offline   Reply With Quote

Old   April 14, 2016, 08:33
Default
  #10
Member
 
Join Date: Aug 2014
Location: Germany
Posts: 77
Rep Power: 11
harry123 is on a distinguished road
I have both experimental and computational results from journals.

The variation of axial velocity looked fine, I didn't check other parameters as I just plot axial velocity along jet centerline and vol fraction of water along centerline.
harry123 is offline   Reply With Quote

Old   April 20, 2016, 03:59
Default
  #11
Member
 
Join Date: Aug 2014
Location: Germany
Posts: 77
Rep Power: 11
harry123 is on a distinguished road
Hi again, I find some strange (in my view) behavior with the simulations. I get result closer to experimental data when I use coarser mesh..!!

In the simulation, the nozzle dia is 1 mm. For fine mesh, the flow region was refined to base size of 0.5mm, for medium - 1 mm and for coarse 2.5 mm.

As can be seen from the attached image, the coarsest mesh gives better results. Can there be a good explanation for this ? In all cases, the time step was 0.001 seconds and RANS model with VOF Multiphase model was used.
Attached Images
File Type: png Vol_fraction.png (14.1 KB, 6 views)
harry123 is offline   Reply With Quote

Old   April 20, 2016, 05:07
Default
  #12
Senior Member
 
Filippo Maria Denaro
Join Date: Jul 2010
Posts: 6,762
Rep Power: 71
FMDenaro has a spectacular aura aboutFMDenaro has a spectacular aura aboutFMDenaro has a spectacular aura about
Unfortunately, this is a case in which the answer is not simple...
you are trying to compare with experimental data that are affected by many variables...Furhtermore, you solve for RANS therefore your solution has a statistical meaning that, depending on the experimental sampling, can match or not with the experiments.
Then, the turbulence model is relevant for RANS, when you refine the grid you diminuish only the local truncation error of the discretization and the solution remains mainly affected by the type of modelling.

I suggest to repeat the simulations on the same grids using different turbulence modelling to check the relevance of the closure model.
harry123 likes this.
FMDenaro is offline   Reply With Quote

Reply

Tags
jet flow, mesh and grid

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
[snappyHexMesh] SnappyHexMesh for internal Flow vishwa OpenFOAM Meshing & Mesh Conversion 24 June 27, 2016 08:54
[ANSYS Meshing] guidance about refining hex mesh? hamidciv ANSYS Meshing & Geometry 15 July 27, 2015 02:28
3D Hybrid Mesh Errors DarrenC ANSYS Meshing & Geometry 11 August 5, 2013 06:42
[ICEM] Generating Mesh for STL Car in Windtunnel Simulation tommymoose ANSYS Meshing & Geometry 48 April 15, 2013 04:24
Icemcfd 11: Loss of mesh from surface mesh option? Joe CFX 2 March 26, 2007 18:10


All times are GMT -4. The time now is 09:36.