CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CFX (http://www.cfd-online.com/Forums/cfx/)
-   -   Rotating propeller in a finite freestream velocity (http://www.cfd-online.com/Forums/cfx/77057-rotating-propeller-finite-freestream-velocity.html)

asma June 11, 2010 07:47

Rotating propeller in a finite freestream velocity
 
Hello,

I am trying to model a simple two-blade propeller in cfx. The propeller is rotating and there's also a free stream horizontal velocity. I tried modelling it using a rotating immersed solid (propeller) in a stationary rectangular block fluid domain with a finite inlet velocity. Outlet was defined as static pressure outlet.

One, I need to know if this approach is correct. My solution converges but the prop force in direction of thrust is coming out negative. How else can I model this problem?

Attesz June 11, 2010 15:16

Hi,

the general way of these simulations is defining a separate domain which includes the propeller, and setting it to rotating domain. In CFX help, there is a compressor simulation, the basics is the same for you. Your BCs seems to be Ok. Keep in mind, if you calculate the force of the propeller by the post calculator, you calculate the reaction force, so the opposite of the thrust vector. The cordinate system is also important.

asma June 11, 2010 15:26

Thanks. I am going through this "Mixing Vessel" tutorial in CFX help and it works the same way by defining a rotating domain in the region of the prop. Since CFX tutorials don't have any information on geometry and mesh sequence, I still dont know wat the approximate size of the rotating domain should be. As in wat fraction of the prop radius. And prop blade should be cut out from the domain as in done usually with vehicles when simulating flow over them?

Attesz June 11, 2010 15:33

Usually the best is to minimize the rotating domain dimensions, because you rotate the whole flow field, which is correct near the blades, but its not physically valid far from the blades. So go to the blades as close as possible.I dont know your geometry, but a simple cylinder near the blades is a good start. The blades and the cylinder walls may not have any connections, keep a little bit distance to get a good quality boundary layer mesh around it. And of course, you should cut out the blades in the first approach (you should use it if you want to do an FSI simulation).

asma June 12, 2010 02:03

Ook I think I get that now. There are however still a few ambiguities. I am not sure wat sign convention does Ansys CFX use for rotating frames. Is it the right hand rule or clockwise/anti clockwise criteria? Secondly post processing would give reaction force only when the prop is made to rotate, as with the immersed solid technique. If it's the other way round, with a rotating domain of free air defined with a stationary prop, the post processing results for prop force will be the actual thrust and not the reaction, and should be positive in thrust direction right?

Attesz June 12, 2010 04:49

Hi,
the rotating frame uses right hand rule. And yes, you will measure the force on the blades, which is thrust.

read these posts: http://www.cfd-online.com/Forums/sea...earchid=488731

Best

asma June 22, 2010 11:41

Simulation results
 
3 Attachment(s)
Hi,

Thanks for all the help in setting up my simulation. I ran my simulation, with an rpm of 8000 and freestream velocity of 75fps. As you suggested, I defined a rotating domain for the propeller and since the prop rotates in an anticlockwise direction about the y axis, as may be seen from the image "prop2", and which is positive rotation according to right hand rule, domain's rpm was set to negative. When I check for force on the propeller blade, it still comes out negative. From what I understood from the discussion, now that I have kept the propeller stationary and made the freestream around it to rotate, blade force should be the actual force on prop blade and hence positive in direction of thrust.

The image "prop result" shows the velcoity contour in the propeller plane. I wanted to know if they look ok. The image "propset" shows the geometric setting of the problem, with a rotating domain encapsulated in a stationary domain with a finite inlet velocity.

sodjaj June 28, 2010 05:04

Hello,

I also work on propeller simulations however I use only one large rotating domain that encompasses the prop. blade. I disagree that the physics changes because the domain is rotating. One just needs to monitor or transform certain variables in stationary frame of reference.

I also tried to use this approach with stationary and rotating domain. However even thodugh the two meshes were confomally connected I experienced abrupt changes in vorticy acros the domains boundaries.

Kind regards,
Jurij

SupunRandeni August 17, 2012 01:46

Hi!
I am doing a similar analysis. That is a self-propulsion test of a submarine. I need to find the thrust produced by the propeller. Therefore I measured the force along the thrust direction at the propeller. But i got a negative value. So I rotated it otherway round but still I get a negative value. Is there any perticular way to find the thrust of a prop? How did you overcome your problem?
With Best Regards,
Supun.

shiyun January 7, 2014 07:14

Quote:

Originally Posted by Attesz (Post 262675)
Usually the best is to minimize the rotating domain dimensions, because you rotate the whole flow field, which is correct near the blades, but its not physically valid far from the blades. So go to the blades as close as possible.I dont know your geometry, but a simple cylinder near the blades is a good start. The blades and the cylinder walls may not have any connections, keep a little bit distance to get a good quality boundary layer mesh around it. And of course, you should cut out the blades in the first approach (you should use it if you want to do an FSI simulation).

Hi Attesz

I also found a issue, as you said, actually the rotating domain rotates the fluid in this domain rather than the real solid body (eg, blades), so is that means if the blade rotate in +ve direction, the rotating domain velocity need give -ve direction, however if do in this way in transient simulation, in CFD-POST you will see the blades rotate in opposite direction compare with the physical condition, do you know how to solve this issue?


All times are GMT -4. The time now is 15:19.