Frame motion in Fluent with translational and rotational velocities input

[GiulioRemo.Cupilari]

Hello,
I am simulating the propellant in a tank of a space launcher during its flight. In input, I pass via MATLAB the linear and rotational velocities of the launcher to Fluent, defined in the body reference frame of the launcher which rotates with the launcher. These inputs go into Frame Motion in Fluent, which has the option “Relative Specification - Relative to Cell Zone” set to Absolute as default, since only one zone is moving. The walls are set to stationary, meaning that they should rotate with the reference frame. The velocity specification is set to Relative in the General tab.
However, the results I get do not match the ones that are expected, with results showing possibly an erroneous decomposition of the velocity vector in the wrong axes, causing the formation of an angle on the free surface. I know that the simulation with all the inputs except the rotations works and was validated, however when I try adding only the rotations, the propellant seems to be sloshing as if it was angled to one side.
Must translational and rotational velocities be defined in the body frame, ie. the relative reference frame which moves and rotates with the launcher and cell zone?
Does the option “Relative Specification - Relative to Cell Zone” set to Absolute in Frame motion mean that the velocities and rotations in the Frame Motion tab must be defined in the Absolute Reference Frame of Fluent, or do they have to be defined in the body reference frame, ie. the relative frame which rotates in Fluent according to the specified rotations?

My walls are set to stationary with respect to the rotating frame

[GiulioRemo.Cupilari]

I would also like to add, does the centre of rotation and rotation axis directions must be defined in the absolute frame, ie. not rotating, or do they have to change to account for the rotation of the body axis through frame motion? For example, if I have rotations around the axial and one of the lateral axes, respectively x and z, with the rotation axis origin that stays on the x axis during this manouver, does the origin point have to account for the rotation of the x axis due to the rotational speed around the z axis? Or are they defined relative to the rotating and moving reference frame, and thus its definition in Fluent would remain of the type (1,0,0)? The same question with the rotational axis directions

[Rob]

Walls would move at zero relative velocity on the moving reference frame, they're not stationary.  If you're using reference frame (motion) the geometry isn't changing so I'm not sure why you need to worry about changing locations. 

Pictures might help. 

[GiulioRemo.Cupilari]

Thanks Rob for replying.
In the first image you can see I have set a changing value for the centre of the rotation axis, which stays on the axial (x) axis. The direction of the axis was calculated using vectors calculated as the angular velocity in each direction normalised by their modulus. Finally, Angular and Translational velocities were inputted in the body reference frame, which rotates and moves with the object. In the second image you can see an example showing for simplicity only a rotation of angle alpha around the y axis (into the page), which causes the x and z axes to rotate.
First, is the definition of the rotation axis centre and direction correct, ie. being the centre set initially on the x axis does it remain on the x axis even after this rotates due to the angle alpha? Or does this rotation have to be taken into account in this definition (ie. do they have to be defined in the body (relative) frame which moves and rotates and thus the above definition is correct)?

Similarly for the velocities, these were defined in the relative frame which rotates, ie. as Vz and Vx in the second image, or do they have to be defined as Vz' and Vx' in the non rotating frame by appllying to the input a rotation matrix R?

 

[Rob]

You're setting a frame motion, so not actually moving anything: the coordinates are still aligned with whatever the mesh was built at. If you want to actually rotate the zone you may want Mesh Motion, but if there are cells "outside" the moving part you may need overset or moving mesh. 

[GiulioRemo.Cupilari]

I see. Then would you have any suggestions on why the rotations cause the results to become different than the validated model (pendulum) considering that without such rotations, the simulation does match it and nothing was changed aside from the rotation input? In the simulation with data shown below, only a rotation around x was added and translational velocities only in x and z, but not in y. Despite this, there is a large error on the y axis, as shown below (red and black curves should match).The inputs were double checked for the rotations.

[Rob]

What curves? 

[GiulioRemo.Cupilari]

The site is not allowing me to add images. It gives me an error saying "Image upload faile due to a XHR error"

[GiulioRemo.Cupilari]

I don't want to rotate the zone, i want to simulate a rotation with a translation also. I am defining the velocities, both rotational and translational, referring to a frame of reference which rotates and moves with the body (body frame) and I am wondering if this is what frame motion wants, or if I have to change these inputs and provide them to FLuent in a frame that does not rotate or move. In the latter case, the translation velocity for example would become v = R*v where R is the rotation matrix to remove the angles of rotation

[Rob]

Translating and rotating? What exactly are you trying to replicate? The values are relative to zone or absolute, but as the zone isn't actually moving you need to set interacting motions carefully.