## Fluids

#### meaning of particle x, y, z velocity for rotating reference frame

• xin.yang
Subscriber

The details of my questions are as follows:

(1) The fluid cell zone of my case has enabled rotational frame motion since my case is in rotational reference frame.

(2) I am using discrete phase model (DPM) to track a single particle.

(3) In order to setup the injection of discrete phase of 'single type', I need to give the values of particle initial x, y, z positions and velocities. Like below figure from ansys fluent:

(4) I need to set zero-slip velocity at the inlet. I have already got the fluid solution of my case in ansys fluent and so I can easily get gas x, y, z velocities and relative gas x,y,z velocities at the injection location of the inlet.

(5) According to 'ansys fluent theory guide', when rotational (moving) reference frames are enabled, the additional forces are added in particle tracking due to the rotation of reference frame.

(6) Question 1: My understanding is that 'when rotational (moving) reference frames are enabled, DPM is solving and reporting particle velocities relative to the rotational reference frame. Am I right?

The theory guide is not giving details of this. It's confusing me. It doesn't tell whether the velocities are stationary or rotational. See figure below from ansys fluent theory guide:

(7) Question 2: If Fluent is reporting the relative particle x, y, and z velocities, then I think I need to use the relative gas x, y, z velocities at the inlet to set up the initial particle x, y, z velocities at the injection location. Am I right?

(8) I am using ansys fluent 19.2 and 2020R2.

• DrAmine
Ansys Employee

1/Please calculate the particle trajectories with relative velocity when solving MRF. You need to select that under Discrete Phase Numerics. (default in recent versions).

2/Relative to the zone motion they are injected from.

• xin.yang
Subscriber

Thanks, Dr Amine,

1) I think the seting under Discrete phase numerics you mentioned is disabling 'Track in Absolute Frame'. It is disabled by default in my Ansys Fluent.

2) Therefore, for my case, the particle velocties are in relative as the whole domain is in the rotating reference frame.

I just did a double check whether my 'zero-slip' velocity at the inlet is correctely setup as follows:

(1) I use a udf to get the gas absolute velocities at the injection point using 'C_U(cell, thread)', 'C_V(cell, thread)', 'C_W(cell, thread)'.

(2) As there is no data access macros for gas relative velocties, then I transfered these velocities into the relative velocities using 'Vector_velocity_absolute=Vector_velocity_relative+ Vector_angular_velocty * Vector_position'. The transformation should be correct as I double checked the formulate by comparing the absulte velocity and relative velocity in Ansys fluent fluid solution. I can get almost same relative velocities to the values provided in the fluid solution.

(3) Then I use these gas relative velocties as the particle initial velocties at the injection location.

(4) Since I cannot directly get particle slip velocty from ansys fluent, But I can get particle Reynolds number. Then I output the data of particle velocties, positions and particle Reynolds number from 'Results->Particle Tracks->Report type: step by step->Report to: File'.

(5) I was hoping the particle Reynolds number is zero or a very small value as the particle velocities are set to be equal to gas velocities. However, the value I got from Ansys Fluent is 8.58. This is quite big for a 'zero-slip' velocity condition.

(6) As this result is surprising me. Then I tried a case without rotating reference frame. For a normal flow case, if I set particle inlet velocity=gas inlet velocity, The particle Reynolds number I get from Fluent is very small, at a level of '1 e-3'. This is what I expect.

(7) I think there should be something wrong for the rotating reference frame case. A reason I guess may be related to how Fluent calculate particle slip velocity. If the gas velocity a particle sees is not determined by 'C_U(cell, thread)', 'C_V(cell, thread)', 'C_W(cell, thread)', but by multiple nodes/cells near the particle, then it is possible that the slip velocity is not equal to zero for my rotational case. But I am not sure what's happening in Fluent.

(8) Could you give me some advice?

• DrAmine
Ansys Employee

Why are you providing the gas velocity as particle velocity? I thought that should be zero-slip? You might let particle coming in then check the Reynolds Number if it is correct.

• xin.yang
Subscriber

Thanks, Dr Amine.

1 > Why are you providing the gas velocity as particle velocity?

It is because I want to set a 'zero-slip' velocity at the injection location for particles.

2>  I thought that should be zero-slip?

Ideally, yes. But as I cannot directly output slip velocity from Fluent. If there is a slip-velocity macros in Fluent UDF, please let me know.

According to the particle Reynolds number (Rep) from Fluent, the slip velocity inside Fluent is not zero as Rep=8.58 for this condition.

3> Comparison between Rep by Fluent and Rep by UDF.

I tried two cases to compare the Rep calculated by UDF and Rep output by Fluent.

'Rep output by Fluent' means: Rep is obtained from here:

Case 1: no rotational reference frame. See figure below:

Case 2: rotational reference frame (not a 'zero-slip' velocity at the injection location in order to remove 'incomplete' particle tracking).

As you can see in both figures, the Rep output by Fluent is close to the Rep calculated in the UDF, but they are not equal to each other. I was expecting they should be equal or the difference within 1%.

4> This is the udf I wrote to calculate relative gas velocity and particle Reynolds number:

• DrAmine
Ansys Employee

What do you mean with zero slip?

• DrAmine
Ansys Employee

If zero slip relative the velocity at inlet: why are you providing any velocity for the particles then?

• xin.yang
Subscriber

1> What do you mean with zero slip?

slip velocity = gas velocity - particle velocity.  So zero-slip velocity means particle velocity equals to gas velocity here.

2> If zero slip relative the velocity at inlet: why are you providing any velocity for the particles then?

Rep Figure of case 1 is at a condition of 'zero slip'.

Rep Figure of case 2 (roational reference frame) is not at a condition of 'zero slip'.  This is because if I set zero-slip at the injection location for this case, the particle tracked is 'incomplete' so I changed the velocity.  It is aimed to understand whether I can get same Rep in UDF to Rep reported by Fluent.

Please see figure below Case3. Case 3 is 'zero-slip' and all other settings are same to case 2.

• DrAmine
Ansys Employee

I know the meaning of zero-slip :)

How does your case look? You know that MRF is an approximation and it might look not really intuitive when it comes to particle tracks. The need to run with relative frame is because espcially for small particles they should not penetrate any walls. However trust-worthy analysis can only be achieved if you run with moving mesh. Let's keep that first aside.

The point properties are relative to the zone motion. You can verify that even without MRF by just injecting from a rotating wall.

The particle tracks (and positions...) that Fluent will calculate and visualise in the rotating reference frame zone will be what an observer sees who is rotating with the shaft / rotor / .... (Think of what a camera "sees" that is attached to the shaft and rotates with it.)

Just create a rotating cell (ring) zone inside a box. The box has 4 sides as pressure outlet. The inner side of the ring is wall and from there particles are injected. The whole ring rotates. Just look there after the particle plots with relative velocity and abolsute velocity to get a better idea what is going on.

• xin.yang
Subscriber

Thanks a lot, Dr Amine.