[Andrea159357]

Does anyone know why i'm getting this kind of trajectory in a particle tracking simulation of a pipe?schifo CL_10.jpg

[FMDenaro]

First of all, what about the velocity solutions you are using?

[Andrea159357]

Velocity has been calculated with Realizable k-eps with scalable wall function.
The model is 15 m pipe with 170 bar pressure and 350 k inlet temperature. Velocity at inlet is 22.6 m/s.
y plus is varing between[15-70].

[FMDenaro]

Originally Posted by Andrea159357

Velocity has been calculated with Realizable k-eps with scalable wall function.
The model is 15 m pipe with 170 bar pressure and 350 k inlet temperature. Velocity at inlet is 22.6 m/s.
y plus is varing between[15-70].

Again, what about your obtained velocity solution? Show the field

[Andrea159357]

Mi scusi, vuole uno screenshot del campo di moto?
Le va bene un contour ed un vector field nel piano?

[FMDenaro]

Originally Posted by Andrea159357

Mi scusi, vuole uno screenshot del campo di moto?
Le va bene un contour ed un vector field nel piano?

Si se il campo è stazionario

[Andrea159357]

Si, il campo di moto è stazionario

[FMDenaro]

I see some angle deviations in the vector plot that are strange. Check better if the residuals are enough small.
Then, plot a single lagangrian particle trajectory superimposed to the velocity plot you showed here. Let the particle going from inlet to outlet. If the flow is steady, the trajectory must coincide with the streamline.

[Andrea159357]

These ara the residuals:
continuity x-velocity y-velocity z-velocity k epsilon
9.5933e-09 4.5807e-11 1.9709e-11 3.2791e-11 2.1988e-09 5.0063e-10

[Andrea159357]

The main problem is that not all the particles escaped. Even if ai increse max number of time step, those particles remain insides because the keep oscillating in the boundary layer.

[Andrea159357]

these are the screenshot of the particles that seems like stuck in the boundary layerJ_1_y 30_trajectory1.jpg

J_1_y 30_trajectory2.jpg

J_1_y 30_trajectory3.jpg

[Andrea159357]

I tryed i different mesh with y+ = 300 so the aspect ratio and min orthogonal quality are far better.
The first mesh is y+=30 with AR = 1300 an MinOQ = 0.05
The second mesh is y+=300 with AR =121 and MinOQ = 0.5

Could be that particle tracking blow up with bad cell quality?

[FMDenaro]

The velocity vector shows oscillations but I don't see in a clear way that you have recirculation region.
However, the particle cannot deviate so abruptly.
I strongly suggest to check first all you are doing in the case of a straight channel. Find the solution for the velocity and then integrate the particle.
In any case, when you post the results show the whole domain and the particle locations using markers

[Andrea159357]

The case is ok, it was tested on a small model of 90° bend pipeline. The ricirculation is not present but plotting the streamline it can be seen the swirl after the curve.
I know that trajectory with 90° change in sirections are unphysical, i'm trying to understand the source of the problem. Coursing the mesh erase the problem but erosion value are littel bit to high; may due because the poor quality of field velocity (y+=200/430).
Someone have told me that this particles behavior could be done by the ratio of particle diameter and wall first cell thickness. In my opinion this can not be true because the lagrangian particle tracking consider a single particle as a point.
Are my consideration so wrong?

[FMDenaro]

Originally Posted by Andrea159357

The case is ok, it was tested on a small model of 90° bend pipeline. The ricirculation is not present but plotting the streamline it can be seen the swirl after the curve.
I know that trajectory with 90° change in sirections are unphysical, i'm trying to understand the source of the problem. Coursing the mesh erase the problem but erosion value are littel bit to high; may due because the poor quality of field velocity (y+=200/430).
Someone have told me that this particles behavior could be done by the ratio of particle diameter and wall first cell thickness. In my opinion this can not be true because the lagrangian particle tracking consider a single particle as a point.
Are my consideration so wrong?

If you are integrating the equation dx/dt = v there is no dimension and mass in the particle model. If you compute the trajectory of a single particle from the inlet to the outlet it must be equivalent in tracking the streamline from the same point at the inlet. Try to superimpose those two and check if coincide.


What about the software you are using?