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

NLH Cells requirement

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

Reply
 
LinkBack Thread Tools Search this Thread Display Modes
Old   February 14, 2018, 03:22
Default NLH Cells requirement
  #1
New Member
 
Vishwas Verma
Join Date: Mar 2016
Location: Mumbai India
Posts: 25
Rep Power: 10
vishwasvemra is on a distinguished road
Dear all,

I am trying to run an NLH solution in fine turbo. As the paper by Vilmin "UNSTEADY FLOW MODELING ACROSS THE ROTOR/STATOR INTERFACE USING THE NONLINEAR HARMONIC METHOD" says for capturing one harmonic one require at least 30 cells in circumferential direction.
Going by the paper by Zoe Burton "The Influence of Inlet Asymmetry on Steam Turbine Exhaust Hood Flows", she says cells requirement is 30*N*(nu/nd) for downstream component (where N is number of harmonics, nu = number of blade in upstream blade row and nd = number of blade in downstream blade row).
By using this formula, say if i want to capture 3 harmonics and for me nd/nu = 2.7, i am getting approximately 119 cells upstream and 70 cells downstream.
Here is my doubt, is this number represent number of points in b2b plane normal to the flow direction in grid topology ? Further if i follow this my node count is going very high approximately 2.5 million nodes for rotor domain. As a result for one stage calculation i am ending up getting nearly 12 million cells in domain, which seems very high to me.
Could anyone please comment on this ?

Thanks

Kind Regards
Vishwas Verma
vishwasvemra is offline   Reply With Quote

Old   February 14, 2018, 18:26
Default
  #2
Senior Member
 
DarylMusashi's Avatar
 
Holger Dietrich
Join Date: Apr 2011
Location: Germany
Posts: 174
Rep Power: 14
DarylMusashi is on a distinguished road
Dear Vishwas,

at first please let me answer your first question, I have marked the related grid nodes in the B2B control panel, which you want to adjust for a proper resolution of the harmonics.

Secondly, I would suggest to use the User Manual and its recommendations as a primary source. The most impact has the first harmonic, it should be resolved with 30 grid points, as you stated correctly. As the influence of the other harmonics decrease quickly you still get a good result when using only 15 for each other harmonic. For 3 harmonics the grid nodes at the interface in circumferential direction are 30+15+15 = 60. (FINE™/Turbo > User Guide > Dedicated Turbomachinery Models > Rotor/Stator Interaction > How to Set-up a Simulation with Rotor/Stator Interfaces? > Harmonic Method).

2.5 million nodes for a rotor mesh is not that unusual in my opinion, but that depends on your available hardware, of course. But I wonder why you need 10 million nodes for the stator (and what else?) to get a resulting mesh size of 12 million nodes? Do you model cavities (ZR-effects)?

Kind regards
Holger

PS: I did several NLH computations with very good agreement to experimental and unsteady results. From my experience the generated meshes in AutoGrid fulfill the desired grid resolution if you use the Row-Wizard (Refinement Level: 0).
Attached Images
File Type: jpg NLH_grid.JPG (78.9 KB, 20 views)
DarylMusashi is offline   Reply With Quote

Old   February 15, 2018, 00:41
Default
  #3
New Member
 
Vishwas Verma
Join Date: Mar 2016
Location: Mumbai India
Posts: 25
Rep Power: 10
vishwasvemra is on a distinguished road
Dear Holger,

Many thanks.
I wrote 12 million cells or elements(not nodes, sorry to mixed up both terminology).
From the attached figure i would like to ask one thing, will it produce matching mesh, if i don't match node count at inlet and outlet side of domain in cross-stream direction. Wouldn't it be fair to go with tip conformal mesh? Will there be the loss in accuracy if i keep non-matching mesh.

Thankyou

Kind Regards
Vishwas
vishwasvemra is offline   Reply With Quote

Old   February 15, 2018, 04:39
Default
  #4
Senior Member
 
DarylMusashi's Avatar
 
Holger Dietrich
Join Date: Apr 2011
Location: Germany
Posts: 174
Rep Power: 14
DarylMusashi is on a distinguished road
Dear Vishwas,

would you like to attach the figure as mentioned? It doesn't really matter in this context if we are talking about nodes or cells. I just would like to know what are the remaining parts of your mesh (besides the stator, of course). As 10 million nodes/cells are a lot.
You are aware of the fact, that you just need 1 rotor passage and 1 stator passage for NLH?

Kind regards,
Holger
DarylMusashi is offline   Reply With Quote

Old   February 15, 2018, 05:20
Default
  #5
New Member
 
Vishwas Verma
Join Date: Mar 2016
Location: Mumbai India
Posts: 25
Rep Power: 10
vishwasvemra is on a distinguished road
Dear Holger,

In the attached figure(a sample case), the red box shows number of node points and the green box in statistics shows number of cells. By nodes i was referring to number of points(red box) and by cells i refer to green box. Aren't node counts different from cell counts.
Earlier i used icem cfd which tells cell counts, i meant in that reference.
Could you please clarify ?
Yes, i am aware of the fact for NLH, only one passage is required.

Thanks

Kind Regards
Vishwas
Attached Images
File Type: jpg final.jpg (97.5 KB, 9 views)
vishwasvemra is offline   Reply With Quote

Old   February 15, 2018, 07:19
Default
  #6
Senior Member
 
DarylMusashi's Avatar
 
Holger Dietrich
Join Date: Apr 2011
Location: Germany
Posts: 174
Rep Power: 14
DarylMusashi is on a distinguished road
Dear Vishwas,

you are right, of course cell counts are different from node (points) counts. What the point is: They are in the same range. I am just wondering about the following: Your rotor passage has 2.5 million elements and your complete mesh 12 million. Which means that about 10 million elements must be in the stator mesh (or what else).
Thats the only point I wanted to discuss, as here might be further potential for reducing the number of cells.

Furthermore, can you clarify what you meant with this:
"From the attached figure i would like to ask one thing, will it produce matching mesh, if i don't match node count at inlet and outlet side of domain in cross-stream direction. Wouldn't it be fair to go with tip conformal mesh? Will there be the loss in accuracy if i keep non-matching mesh."

I think you are referring to my attached B2B topology figure? No, you never get a matching mesh between the rows, as it is an rotor-stator interface. It can't be matching since one row rotates, the other not.

I would recommend to use the same wall spacing for both rows and approximately the same number of flow pathes. As a rule of thumb the ratio of 3:1 shouldn't exceed between the rows. Please have a look at the attached figure (B2B view, left row 1, right row 2). I hope you get the point and this answers your question?

Kind regards,
Holger
Attached Images
File Type: jpg rows.jpg (197.3 KB, 10 views)
DarylMusashi is offline   Reply With Quote

Old   February 15, 2018, 07:53
Default
  #7
New Member
 
Vishwas Verma
Join Date: Mar 2016
Location: Mumbai India
Posts: 25
Rep Power: 10
vishwasvemra is on a distinguished road
Dear Holger,

Thanks alot.
As i mentioned, i have 2.5 million nodes in rotor and 1.5 million nodes(red box in my above figure) in stator (which according to above figure will mean 12million cells as shown in green box all together).

Further, by saying non-matching i was referring to mesh in tip gap region, as in the attached figure red boxes shows non-matching mesh in tip gap, due to unsymmetrical grid distribution(as you told in your figure).

Thankyou

Kind Regards
Vishwas
Attached Images
File Type: jpg 2.jpg (181.7 KB, 8 views)
vishwasvemra is offline   Reply With Quote

Old   February 15, 2018, 08:13
Default
  #8
Senior Member
 
DarylMusashi's Avatar
 
Holger Dietrich
Join Date: Apr 2011
Location: Germany
Posts: 174
Rep Power: 14
DarylMusashi is on a distinguished road
Dear Vishwas,

okay, sorry for the confusion. 2.5 and 1.5 million nodes for each stage sounds reasonable. But if you need less nodes you may use again the RowWizard and use a refinement level of -2 or -3. The quality and overall result should be in the same range, but with less nodes.

Non-matching periodicity in the gap:
I would recommend to enforce a matching connection in the flow channel instead of in the gap. From my experience there is no loss of accuracy in the tip gap region when using a non-matching connection. If you want to make further investigations you can switch the matching regions in the B2B Mesh Topology Control -> Topology dialog (top: "Matching Periodicity"). Please click "Initialize Topology and Grid Points Distribution" in the top right corner. I attached two pictures with the two different matching and non-matching locations.

Kind regards,
Holger
Attached Images
File Type: png matching_periodicity.PNG (104.2 KB, 16 views)
File Type: png non-matching_periodicity.PNG (95.9 KB, 17 views)

Last edited by DarylMusashi; February 16, 2018 at 07:49.
DarylMusashi is offline   Reply With Quote

Old   February 15, 2018, 14:04
Default
  #9
New Member
 
Vishwas Verma
Join Date: Mar 2016
Location: Mumbai India
Posts: 25
Rep Power: 10
vishwasvemra is on a distinguished road
Dear Holger,

Thanks alot.

Kind regards
Vishwas
vishwasvemra 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
Foam::error::PrintStack almir OpenFOAM Running, Solving & CFD 91 December 21, 2022 05:50
[snappyHexMesh] snappyHexMesh does not remove internal cells Binglian Hu OpenFOAM Meshing & Mesh Conversion 16 June 3, 2021 10:38
[snappyHexMesh] sHM layer process keeps getting killed MBttR OpenFOAM Meshing & Mesh Conversion 4 August 15, 2016 04:21
[snappyHexMesh] No layers in a small gap bobburnquist OpenFOAM Meshing & Mesh Conversion 6 August 26, 2015 10:38
[snappyHexMesh] snappyHexMesh won't work - zeros everywhere! sc298 OpenFOAM Meshing & Mesh Conversion 2 March 27, 2011 22:11


All times are GMT -4. The time now is 04:18.