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

Propellor case diverges for K-epsilon model while stable for Laminar

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

Like Tree2Likes
  • 1 Post By Krao
  • 1 Post By Krao

Reply
 
LinkBack Thread Tools Search this Thread Display Modes
Old   October 10, 2019, 04:14
Default Propellor case diverges for K-epsilon model while stable for Laminar
  #1
Member
 
chandra shekhar pant
Join Date: Oct 2010
Posts: 62
Rep Power: 11
chandra shekhar pant is on a distinguished road
Dear All,


I am trying to run the propeller case for the changed geometry. I had put my CAD model (comprising of the 3 blade and a hub), innerCylinder_small, innerCylinder and outerCylinder, exactly as been given in the tutorial. I tried to run the exact case (same RANS model, dt, etc.) as been given in the tutorial but found that after a time of 0.28 the simulations stop since it becomes unstable, CFL value, K, \epsilon value increases by an order of 100. However if I run the same case as laminar then it stays stable. The result of checkMesh -allGeometry -allTopology, the snapshot of grid, control panel, boundary is attached herewith. I have been struggling this problem for about a month. Any help/suggestion is highly appreciated.
Attached Images
File Type: png xz_mesh.png (125.5 KB, 5 views)
Attached Files
File Type: c boundary.c (2.0 KB, 1 views)
File Type: c controlDict.c (1.6 KB, 1 views)
File Type: c CheckMesh_updated.c (56.9 KB, 2 views)
chandra shekhar pant is offline   Reply With Quote

Old   October 10, 2019, 05:08
Default
  #2
Member
 
Kmeti Rao
Join Date: May 2019
Posts: 91
Rep Power: 2
Krao is on a distinguished road
Hi chandra shekhar pant,

***Max skewness = 8.45626, 10 highly skew faces detected which may impair the quality of the results
<<Writing 10 skew faces to set skewFaces

As far my experience with OpenFoam is concerned, you need to improve your mesh, with this mesh it is difficult to get a transient simulation running. Also, by copying everything from tutorials will not solve the problem. You need to calculate the various model parameters, which are valid for the case you are studying. But for the initial setup it should not be a problem. Also try to start with lower order fvSchemes and with lower speed and time steps, as the simulation gets stable change everything according to your requirement. Also try to run your case in laminar first, and then use these laminar flow results as an initial point for your turbulent flow study. You can use lower relaxation factor in the beginning, and change it accordingly once your simulation gets stable. Check out the following link for some useful tips regarding mesh generation and simulation setup http://www.wolfdynamics.com/images/O...sandtricks.pdf. All the best

Krao
Krao is offline   Reply With Quote

Old   October 10, 2019, 05:41
Default
  #3
Member
 
chandra shekhar pant
Join Date: Oct 2010
Posts: 62
Rep Power: 11
chandra shekhar pant is on a distinguished road
Hello Kmeti Rao,


Sincerely thanks to you for looking the files and commenting them. I am a beginner in the OpenFoam. I am using snappyHexMesh, so if you say the mesh is not of good quality what I think is that, you mean that I should increase the maxLocalCells and maxGlobalCells ?

My RANS simulations are stable for 0.28 seconds and before that close to 0.26 everything looks perfect, the CFL, rotation of mesh/rotation of propellor etc.

But after that everything goes bad. As of now the simulations for the laminar looks stable (simulated around 1 sec).

Could you please elaborate what are the next steps to be taken?


Many thanks for helping me.
chandra shekhar pant is offline   Reply With Quote

Old   October 10, 2019, 06:31
Default
  #4
Member
 
Kmeti Rao
Join Date: May 2019
Posts: 91
Rep Power: 2
Krao is on a distinguished road
These skew cells are not great to have in simulation. If you have too many sharp edges try to make those edges blunt or rounded. Add sufficient boundary layers. Also making the mesh finer than what you require would simply increase the simulation time. Therefore try to calculate the required cell size in advance, for example you have stated that you are using k-epsilon model, for this model calculate in advance the probable y+(dimensionless wall distance needed) and make mesh accordingly. Also after simulation check your y+ value and rebuild the mesh, if you think you require some refinement in some regions, such as trailing edge, tip or wake etc. I know as a beginner it takes a lot to learn, but you have sufficient materials here in cfd online, go through, have patience you'll get going.

By coming to the second part
Quote:
My RANS simulations are stable for 0.28 seconds and before that close to 0.26 everything looks perfect, the CFL, rotation of mesh/rotation of propellor etc.

But after that everything goes bad. As of now the simulations for the laminar looks stable (simulated around 1 sec).
If you have turned off the turbulence, now turn it on and observe what happens. Or if you are using steady solvers and change it to transient and getting the errors mentioned above, only good mesh can help you.

But once again I am reiterating, what you have already gone through in the above links, the one who owns the mesh owns the result!!!

Krao
Krao is offline   Reply With Quote

Old   October 11, 2019, 08:54
Default
  #5
Member
 
chandra shekhar pant
Join Date: Oct 2010
Posts: 62
Rep Power: 11
chandra shekhar pant is on a distinguished road
Hi Kmeti Rao,


Going through your post I realized the importance of y+, I just check this value after the sims (using pimpleFoam -postProcess -func yPlus) and found that initially (at time=0), its giving:
# y+ ()
# Time patch min max average
0 outerCylinder 1.118179e+03 7.877846e+03 5.058236e+03
0 propellerTip 4.638374e+02 3.690934e+04 5.864245e+03


and at the end of some time (0.09) its giving me:
# y+ ()
# Time patch min max average
0.09 outerCylinder 1.058541e+03 7.848940e+03 4.898665e+03
0.09 propellerTip 3.146227e+02 7.994214e+04 1.134515e+04


So I don't understand what this huge numbers are indicating? Now lets assume that I have these values, how should I proceed?
Again thanks for your valuable comments.


Thanks and regards
chandra shekhar pant is offline   Reply With Quote

Old   October 11, 2019, 09:26
Default
  #6
Member
 
chandra shekhar pant
Join Date: Oct 2010
Posts: 62
Rep Power: 11
chandra shekhar pant is on a distinguished road
After you comments, I was going through the subject of y+ and found that if I am using wall functions, I need not to worry about the y+ values . Please correct me if I am wrong, this is what I understood through some of the internet stuffs. Also just curious will anything change if I move my RAS model from k-\epsilon to k-\omega SST ?
chandra shekhar pant 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 On
Pingbacks are On
Refbacks are On


Similar Threads
Thread Thread Starter Forum Replies Last Post
Question about adaptive timestepping Guille1811 CFX 25 November 12, 2017 17:38
bounding of k and epsilon diverges solution Dave110 OpenFOAM 3 October 13, 2017 01:19
At high Y+ values does the K Omega SST model just behave like the K Epsilon model? JuPa CFX 0 December 22, 2015 06:44
Overflow Error in Multiphase Modelling with Two Continuous Fluids ashtonJ CFX 6 August 11, 2014 14:32
Jump in epsilon values near the wall : low re k-epsilon model malaboss OpenFOAM Verification & Validation 1 February 1, 2013 16:36


All times are GMT -4. The time now is 09:24.