|
[Sponsors] |
March 13, 2013, 10:35 |
Can't get the mesh deformation to work.
|
#1 |
Member
Sanyo
Join Date: Apr 2009
Location: India
Posts: 62
Rep Power: 17 |
Hello All,
I am trying to simulate flapping of the flat plate in water. Total angular movement is 74 Deg in one second. I need to calculate the pressure & viscous moments on plate w.r.t. time. I had been through various posts about this and tried to collect as much information I can. My problem is that whatever I do, I cant get the plate to be rotated by desired angle. I had tried different ways as follows. 1. First I tried with immersed solids - problem is I cant get moments on plate walls. Also when plate has been moved, cavity in fluid volume is visible. 2. Then I tried by Rigid body with mesh deformation option. Mass moment of inertia was zero. No external force or torque. Rotational degree of freedom- Y axis. Mesh motion for plate walls was kept as rigid body solution. Angular velocity of 1.3 rad/s was set for plate walls in wall velocity. Wall velocity was kept relative to mesh motion. Solver exited with an error at 2nd timestep due to negative volume. When I checked the solution, plate was moved by almost 70 Deg just in one timestep. 3. Next, I created 2cylindrical domains around plate. The cylinder containing plate was also specified as subdomain & mesh motion was kept as rigid body solution. Rest of the BC kept same as of case 2. Mesh motion was kept stationary for Interface side in inner cylinder. This simulation worked, though it gave warning that rotational degrees of freedom not converged. When checked the solution, the plate movement was only around 2-5 Deg in one second. 4. I think, I shouldn't define rotation in wall velocity. But I couldn't figure out the way to define rotations in rigid body options. As I want to find out the forces acting on plate, I don't want to define any external force or torque. 5. I also tried with rotating domain and transient rotor stator. Motion is correct in that case, however the swirling in flow is very high & definitely unrealistic. Swirls can be seen in whole rotating domain which is obvious because fluid itself is rotated. I don't get where am I making mistake. I am quite frustrated. Can anybody please help me? Thanks in advance. Regards, -Sanyo |
|
March 13, 2013, 16:23 |
|
#2 |
Super Moderator
Glenn Horrocks
Join Date: Mar 2009
Location: Sydney, Australia
Posts: 17,703
Rep Power: 143 |
How is the motion defined? Do you know the motion and just want to apply it, or is the motion coupled to the fluid flow somehow (if so, then how)?
|
|
March 14, 2013, 01:33 |
|
#3 |
Member
Sanyo
Join Date: Apr 2009
Location: India
Posts: 62
Rep Power: 17 |
Thank you very much Glenn for reply.
I know the motion of the plate and want to apply it directly so that I can evaluate the transient moments. Motion is not coupled with flow field. Angular velocity of plate is 1.3 rad/s. I am defining the motion in Wall Boundary condition as rotating wall as follows. I am not sure though that it is the correct way to specify the motion. But I didn't find any other way either. Any suggestion please? BOUNDARY: Plate Boundary Type = WALL Location = Primitive 2D G BOUNDARY CONDITIONS: MASS AND MOMENTUM: Option = No Slip Wall Wall Velocity Relative To = Mesh Motion WALL VELOCITY: Angular Velocity = 1.3 [radian s^-1] Option = Rotating Wall AXIS DEFINITION: Option = Coordinate Axis Rotation Axis = Plate cordinates.2 END END END MESH MOTION: Option = Rigid Body Solution Rigid Body = Rigid Plate END WALL ROUGHNESS: Option = Smooth Wall END END END Other Parameters are as follows: MESH DEFORMATION: Option = Regions of Motion Specified MESH MOTION MODEL: Option = Displacement Diffusion MESH STIFFNESS: Option = Increase near Small Volumes Stiffness Model Exponent = 10 END RIGID BODY: Rigid Plate Location = Primitive 2D G Mass = 1 [kg] Rigid Body Coord Frame = Plate cordinates DYNAMICS: DEGREES OF FREEDOM: ROTATIONAL DEGREES OF FREEDOM: Option = Y axis END END END MASS MOMENT OF INERTIA: xxValue = 0 [kg m^2] xyValue = 0 [kg m^2] xzValue = 0 [kg m^2] yyValue = 0 [kg m^2] yzValue = 0 [kg m^2] zzValue = 0 [kg m^2] END END |
|
March 14, 2013, 05:27 |
|
#4 |
Super Moderator
Glenn Horrocks
Join Date: Mar 2009
Location: Sydney, Australia
Posts: 17,703
Rep Power: 143 |
If the plate just rotates (you mention a constant angular velocity) then put it is a rotating frame of reference - easy.
If the plate oscillates and it is just sweeping an arc then put it in a rotating frame of reference and put in a angular velocity which is a function of time. |
|
March 14, 2013, 08:09 |
|
#5 |
Member
Sanyo
Join Date: Apr 2009
Location: India
Posts: 62
Rep Power: 17 |
Thank you Glenn for the reply and extremely sorry for forgetting to state that I need to see actual movement of plate. I need to create an animation for the plate movement and vectors around it. I guess, its not possible with MFR approach.
I didn't get why can't I see the plate movement with Rigid body approach. Regards, -Sanyo |
|
March 14, 2013, 17:03 |
|
#6 |
Super Moderator
Glenn Horrocks
Join Date: Mar 2009
Location: Sydney, Australia
Posts: 17,703
Rep Power: 143 |
? The plate "actually" moves in a rotating frame of reference simulation. I would not have suggested it if it didn't. You will need to use a GGI to connect it to the stationary frame around it, and put it in transient rotor-stator mode to get the motion. If you used frozen rotor or something like that you might be confused into thinking the plate does not move.
If the motion is only angular then RFR is by far the easiest and most accurate way of doing it. |
|
March 14, 2013, 17:26 |
|
#7 |
Member
Sanyo
Join Date: Apr 2009
Location: India
Posts: 62
Rep Power: 17 |
I had tried it. As I stated in my first post (point#5), motion of plate is correct however, it sets rotating motion to all of the fluid in the rotating domain. It could be correct for pump or turbines. But in my case, plate will not affect whole domain. The motion of the plate is kind of flapping. In 1 second, it travels from 0 Deg to 74 Deg. Then it is stationary for next 0.5 second. And again travels back to 0 Deg in 1 second. Initially, I am just trying to model motion of 1 second. Later, I will model back movement.
I had also read that fluent has very good capabilities for dynamic mesh. Does CFX has similar capabilities? Or CFX can not handle dynamic mesh problems? |
|
March 14, 2013, 17:32 |
|
#8 |
Member
Sanyo
Join Date: Apr 2009
Location: India
Posts: 62
Rep Power: 17 |
I also don't understand, why we can't define angular velocity to rigid body?
|
|
March 16, 2013, 05:15 |
|
#9 | |
Super Moderator
Glenn Horrocks
Join Date: Mar 2009
Location: Sydney, Australia
Posts: 17,703
Rep Power: 143 |
Quote:
What you are seeing is the velocity in the rotating field of reference. As the flow is moving in pretty much a straight line in the absolute frame of reference, when viewed in the rotating frame of reference the velocity shows a counter-rotating velocity. To see the velocity in the stationary frame of reference view velocity in the staionary frame of reference. It is all pretty obvious when you understand it. So try rotating frame of reference again, and this time look at the velocity in the stationary frame. It will behave as you expect. Yes, CFX has full moving mesh capabilities. But don't bother with it in this case, rotating frame of reference can do what you want just fine, and will be MUCH faster and easier to set up. |
||
March 16, 2013, 06:53 |
|
#10 |
Member
Sanyo
Join Date: Apr 2009
Location: India
Posts: 62
Rep Power: 17 |
It's very embarrassing. How simple it was and how little informed I am! Thank you very much Glenn for the support. I find your posts always invaluable. Thank you very much.
Anyway, if I had to solve the same problem by moving mesh approach, how could I proceed? Regards, -Sanyo |
|
March 17, 2013, 04:26 |
|
#11 |
Super Moderator
Glenn Horrocks
Join Date: Mar 2009
Location: Sydney, Australia
Posts: 17,703
Rep Power: 143 |
To do this by moving mesh I would cut out a cylinder around the plate (just as for the rotation frame of reference approach). Link the cylinder to the outer stationary bit with a GGI. Then you move the cylinder with moving mesh using a simple rotational transform. The moving mesh approach is almost identical to the rotating frame of reference approach but more complex and will run much slower.
I would not put the plate in a single domain and use mesh deformation to handle the motion. You will get folding mesh and horrible mesh quality. |
|
March 18, 2013, 08:58 |
|
#12 |
Member
Sanyo
Join Date: Apr 2009
Location: India
Posts: 62
Rep Power: 17 |
Thanks Glenn. Exactly that was happening when I tried moving mesh. I used only one domain & I was getting negative volumes.
Thank you very much. |
|
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
how to set periodic boundary conditions | Ganesh | FLUENT | 15 | November 18, 2020 06:09 |
Mesh motion with Translation & Rotation | Doginal | CFX | 2 | January 12, 2014 06:21 |
Mesh Deformation issues | Jon P | CFX | 0 | November 27, 2007 18:20 |
Mesh deformation | Elian81 | CFX | 5 | September 6, 2006 06:57 |
Problem with rotational mesh deformation | NymphadoraTonks | CFX | 1 | October 17, 2004 18:43 |