CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CFX (http://www.cfd-online.com/Forums/cfx/)
-   -   Boundary condition setting regarding turbine simulation using CFX (http://www.cfd-online.com/Forums/cfx/98183-boundary-condition-setting-regarding-turbine-simulation-using-cfx.html)

 Lacerlacer March 5, 2012 11:15

Boundary condition setting regarding turbine simulation using CFX

To all CFD-ers,

Hi there, my name is LOH, a student from Malaysia. Currently I am researching on project regarding tidal turbine. I am using ANSYS CFX and Fluent for the simulation. There are some questions encountered in the simulation which I hope to seek some opinion and advices from all the experts here.

Firstly, tidal turbine is quite similar to that of wind turbine. Since wind energy already been research for more than 35 years, and with the availability of NREL phase 4 full scale wind turbine experiment, there are quite some numbers of numerical simulations work done on the performance prediction of particular turbine. The wind turbine in NREL is a 2 bladed machine, and the two blades are same in all aspects. Therefore, most of the authors of published works are doing their simulation with only a single blade domain analyzed (periodicity of 180deg). By referring to CFX workshop and documentation, compressor turbines are analyzed in similar fashion. Torque of one blade with respect to the rotating axis is get from simulation and multiplies by respective number of blades and speed of rotation for power calculation. So, I guess those authors are doing the same thing (please correct me if I am wrong).

In tidal turbine, the simulation is done in similar ways by all others researcher. A benchmark experiment from University of Southampton is done on 2005 and is used for numerical codes verification. Most of the verification published work analyses single blade domain too. The turbine is a three blades machine.

Since I am researching on some design of blade for power improvement, I carry out the same simulation as what all others do. I model the respecting blade, and run a single blade domain simulation by referring to some of the papers. Somehow, problems exist. Firstly, the results I get is far away from what it suppose to be. The expecting torque supposed to be in 20+ Nm range and I get it for more than 100 Nm. I thought it was some meshing issue, and try to solve the issue by improve the mesh quality by decrease the Y+ to meet the turbulence model requirement. The mesh I used is fully hexahedral blocked by ICEMCFD, the pre-mesh quality should be ok (determinant 2x2x2 is above 0.5). Somehow, the result doesn’t improve much, still in the range of 100 Nm. The torque value is monitored throughout the simulation and only be taken when the value doesn’t fluctuating much.

I am running out of idea and suspect that the error come from boundary condition setting. I would like invite all forum-er and expert to comment in the boundary condition setting. Here is the picture of my domain and boundary conditions setting:

http://www.cfd-online.com/Forums/mem...c-settings.png

Inlet : 1.73 m/s Normal speed
Outlet : Pressure outlet 0 gauge pressure
Farfield: No slip wall
Bottom wall : No slip wall
Periodic faces : rotational periodic interface

Latest Update 5 March 2012
Been try to run simulation using 180deg periodicity and 120deg periodicity, changing the boundary condition of farfield to free slip wall:
120deg : No slip wall – 112.14Nm
120deg : Free slip wall – 68.589Nm
180deg: No slip wall – 112.14Nm
180deg: Free slip wall – 50.065Nm

120 deg shall reflected 3 bladed machine and 180deg shall represent 2 bladed machine after multiply their respective number of blade….

Any input is welcome and appreciated…

 tauqirnawaz March 5, 2012 11:59

One problem immediately visible from the picture you have provided is the far field boundary, it seems that you have set it to pressure opening. You need to change it to wall boundary. Secondly it seems you have got only one domain?

 ghorrocks March 5, 2012 20:16

You do not need the far field to be a wall. It can be done using pressure boundaries but may have convergence issues.

Have you read the FAQ on accuracy?:
http://www.cfd-online.com/Wiki/Ansys..._inaccurate.3F

 Lacerlacer March 5, 2012 20:40

Quote:
 Originally Posted by tauqirnawaz (Post 347731) One problem immediately visible from the picture you have provided is the far field boundary, it seems that you have set it to pressure opening. You need to change it to wall boundary. Secondly it seems you have got only one domain?
I already change it to wall boundary, the simulation was done using pressure outlet before and results is not acceptable also. The results that i show is using wall boundary,been try for free slip and no slip wall as well.

 Lacerlacer March 7, 2012 10:16

Quote:
 Originally Posted by tauqirnawaz (Post 347731) One problem immediately visible from the picture you have provided is the far field boundary, it seems that you have set it to pressure opening. You need to change it to wall boundary. Secondly it seems you have got only one domain?
Regarding the domain, u are trying to say that if i set one domain rotating, another one stationary? i got try on that, somehow, with single blade domain, it prompt alot of errors, mainly from the interface of rotating domain and stationary domain. May u share on how u set on that? the domain interface

 Lacerlacer March 7, 2012 10:19

Quote:
 Originally Posted by ghorrocks (Post 347810) You do not need the far field to be a wall. It can be done using pressure boundaries but may have convergence issues. Have you read the FAQ on accuracy?: http://www.cfd-online.com/Wiki/Ansys..._inaccurate.3F
I tried both pressure boundaries and wall boundary, results is quite similar. Convergence is achievable with tolerance of less than 1 percent fluctuation of torque value. the torque is get by torque_x()@blade , rotating axis is on X

 tauqirnawaz March 7, 2012 14:59

Quote:
 Originally Posted by Lacerlacer (Post 348146) Regarding the domain, u are trying to say that if i set one domain rotating, another one stationary? i got try on that, somehow, with single blade domain, it prompt alot of errors, mainly from the interface of rotating domain and stationary domain. May u share on how u set on that? the domain interface
It is a bit tricky to set up domain interfaces, it depends on your mesh and analysis type. If you can please take a screen print of the interface settings and a close view of the mesh at the interface then I would probably be able to comment.

http://hikwww2.fzk.de/hik/orga/hlr/A...interfaces.pdf

 ghorrocks March 7, 2012 18:04

This model can be done using either pressure or wall far field boundaries. Your problem is likely to be elsewhere.

 Lacerlacer March 11, 2012 01:20

Quote:
 Originally Posted by tauqirnawaz (Post 348214) It is a bit tricky to set up domain interfaces, it depends on your mesh and analysis type. If you can please take a screen print of the interface settings and a close view of the mesh at the interface then I would probably be able to comment. http://hikwww2.fzk.de/hik/orga/hlr/A...interfaces.pdf
I been simulate the two domain successfully, will post the results in short. Really thanks for the comments.

 Lacerlacer March 11, 2012 11:21

Quote:
 Originally Posted by tauqirnawaz (Post 348214) It is a bit tricky to set up domain interfaces, it depends on your mesh and analysis type. If you can please take a screen print of the interface settings and a close view of the mesh at the interface then I would probably be able to comment. http://hikwww2.fzk.de/hik/orga/hlr/A...interfaces.pdf
http://www.cfd-online.com/Forums/mem...e-global-x.png
http://www.cfd-online.com/Forums/mem...-rotate-p2.png
http://www.cfd-online.com/Forums/mem...terface-p1.png
http://www.cfd-online.com/Forums/mem...al-cgi-180.png

Above is the pictures of interfaces for 180deg periodic domain(two blades). The setting of interface also shown.

Below is the pictures of interfaces for 120deg periodic domain(three blades):

http://www.cfd-online.com/Forums/mem...120deg-bcs.png
http://www.cfd-online.com/Forums/mem...-interface.png
http://www.cfd-online.com/Forums/mem...-interface.png
http://www.cfd-online.com/Forums/mem...eriodic-p2.png
http://www.cfd-online.com/Forums/mem...eriodic-p1.png

Simulation is Done for 120deg periodic interface with a rotating domain (inner one) and stationary domain (outer). The torque result got is even far from single domain interface. The mesh used is different, i was using full tetrahedral for the simulation. Torque value i got is 148.45Nm which is far from what it should be in that particular rotating speed (TSR of 7, responding on 40cm radius of blade). The referring literature for those torque value is " Power and Thrust measurement of marine current turbines under various hydrodynamic flow conditions in cavitation tunnel and a towing tank", by As Bahaj, A Molland J chaplin and W.Batten.

Any input is welcome...

 tauqirnawaz March 12, 2012 09:21

Quote:
 Originally Posted by Lacerlacer (Post 348794) Simulation is Done for 120deg periodic interface with a rotating domain (inner one) and stationary domain (outer). The torque result got is even far from single domain interface. The mesh used is different, i was using full tetrahedral for the simulation. Torque value i got is 148.45Nm which is far from what it should be in that particular rotating speed (TSR of 7, responding on 40cm radius of blade). The referring literature for those torque value is " Power and Thrust measurement of marine current turbines under various hydrodynamic flow conditions in cavitation tunnel and a towing tank", by As Bahaj, A Molland J chaplin and W.Batten. Any input is welcome...
Hi Loh

1. The resolution of the pictures that you have sent is too low, so I am not able to identify them clearly.
2. I think you have combined all the interface surfaces in one interface, which is incorrect. You would have to define a separate interface for each surface of two domains that are touching each other, like top,bottom and side surfaces of the rotating domain.
3. It is also unclear from the pictures, that which "Frame Change" and "Mesh Connection options" have you chosen.
4. Please go through "http://iceberg.shef.ac.uk/docs/cfx-pdf/tutorials/Tut15MultiphaseMixer.pdf", to see how to define interfaces between rotating and stationery domains.

 Lacerlacer March 12, 2012 10:32

Quote:
 Originally Posted by tauqirnawaz (Post 348935) Hi Loh 1. The resolution of the pictures that you have sent is too low, so I am not able to identify them clearly. 2. I think you have combined all the interface surfaces in one interface, which is incorrect. You would have to define a separate interface for each surface of two domains that are touching each other, like top,bottom and side surfaces of the rotating domain. 3. It is also unclear from the pictures, that which "Frame Change" and "Mesh Connection options" have you chosen. 4. Please go through "http://iceberg.shef.ac.uk/docs/cfx-pdf/tutorials/Tut15MultiphaseMixer.pdf", to see how to define interfaces between rotating and stationery domains.
Thanks alot for the suggestion! i will try that tutorial and set accordingly, really appreciate ur help~

 All times are GMT -4. The time now is 09:45.