CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   Torque Calculation of 2D rotating disk (

playmaster July 2, 2012 01:57

Torque Calculation of 2D rotating disk
Dear all

I have simulated a rotating disk floating in water by a 2D model in CFX using symmetry boundary condition. after solution, I want to determine torque on this disk. would anyone help me how to calculate it?

Thank you in advance,

ghorrocks July 2, 2012 08:44

Use the function calculator in CFD-Post, with the force_x/y/z or torque_x/y/z functions.

playmaster July 3, 2012 14:34

thank you ghorrocks,

actually I did, but I expected that it would be about 11000 N.m, when I saw the calculated value (about 450 N.m). It seems that torque of 2D antisymmetric flow with swirl competent need another calculations by writing an expression.

ghorrocks July 3, 2012 19:06

Have you taken into account the lack of a z dimension in 2D models? So your expected value will actually be of units Nm/m (where the /m comes from per metre of length in the z dimension), and your CFX model will have a torque over whatever extrusion length you did in the z direction, so you need to convert that to Nm/m also.

playmaster July 6, 2012 02:44

Mesh between rotor and stator was generated in X-Y plane. Then , in CFX, the domain has been considered as a rotating domain around X-axis. In CFD post, torque was claculated around x-axis. it is also interesting that I expect that torque value around Y and Z axis be zero, but its value around Z- axis is greater than X-axis. Confusing!!!

ghorrocks July 6, 2012 06:27

Confusing - yes definitely, I did not understand a word you said.

But my point (which I suspect I did not make clearly enough last post) is that if you make the z extrusion 1mm it will give you half the torque of if you made it 2mm. The simulation is not normalised for 2D, you have to do that yourself.

hadikhayyamian July 26, 2012 23:55

Hey Glenn;
I had the same problem. you are telling the output value of the solver is actually in Nm/m. is it right?
the problem is that there is no extrusion in z direction. actually it is a disk rotating around x axis? So I dont understand that how output value should be converted in real 3D case?
thanks in advance for your reply.

ghorrocks July 27, 2012 06:25

No. The output of the solver in simply Nm on the body as modelled. You have to convert that to a 2D result yourself if that is of relevance.

I do not understand your second sentence, can you post an image?

hadikhayyamian July 27, 2012 13:23

Hey Glenn;
thanks for reply.

I would like to model a disk like in this figure:

for simplicity I used 2d axisymmetric swirl. so the computational domain and geomtry in 2d is like:

the output torque is much lower than expected torque! Do you mean that 450 Nm torque which I obtained from solver has to be changed?

Please note that I expect final torque to be about 10K Nm.


ghorrocks July 28, 2012 08:03

In your case, from the torque returned on the disk you will have to double it (for symmetry) then scale it up to 360 degrees - so it you model a 2 degree slice you will need to multiply it by 180.

Also this type of flow generates 3D features when the rotational velocity is high enough, and if you are in this regime your 2D model will be inaccurate. Your model is only applicable in the 2D flow regime.

hadikhayyamian July 29, 2012 14:45

I think I could not explain my model well enough. So I will ask in another way:

If you see Tutorial Guide: Using a Single Rotating Reference Frame ( 2D, axisymmetric, co-rotating disk cavity system)

there are two rotating disks. in this model there is no 2 degree slice of the disk. it is just a section of the disk. and so it is a line (2D).
Can you tell me, for example in that example, how you calculate fluid viscouse torque on the right rotating wall? will you just use the value returned by going to Report>Force>Moment around x-axis?

Thanks for your reply!

ghorrocks July 29, 2012 19:42

Which example are you referring to?

hadikhayyamian July 31, 2012 16:50

See below, please:

ghorrocks July 31, 2012 19:08

That is a fluent example. CFX is a bit different. Fluent has a true 2D solver, but CFX does not.

All times are GMT -4. The time now is 22:41.