[amelia]

Hi,
I have many experimental results obtained using the LDA technique, inside a small annular flume (see file attached). The fluid movement is provided by a lid, at the top of the mini flume. Although the gap between the bottom wall and the lid is 49 mm, measurements were made only up to 26 mm above the bottom.
Now, I am using Open FOAM, trying to find the best model for my case. The experimental tests were made at both laminar and turbulent Re; the fluid was an aqueous solution of glycerin (dynamic viscosity=3e-3 Pa.s and density=1076kg/m3). I send attached the results obtained experimentally (uz_exp.) and numerically (uz_num.), for the angular velocity of 41,89 rad/s. The numerical results were obtained using simpleFOAM and the k-e turbulence model considering the default values. Comparing the experimental and numerical results, it is possible to observe that profiles (Ux, Uy, Uz) are similar in form, however the magnitudes are quite different.
I am trying to justify that difference, thus I would like to know if I am using the correct model and if there are important factors inside the flume that are not being considered in the equations, such as centrifugal forces, since the experimental results show that the flow is pushed to the outer wall.

[amelia]

[amelia]

[amelia]

[amelia]

Circular miniFlume:

[billy]

Hello,

I modeled this geometry in 3D and obtained the above results.

Now, I modeled this geometry as an axi-symmetric geometry using wedge patch type. Actually, since the geometry has an inner diameter the block is not a wedge but a hexahedron. Then I defined two patches with wedge type. Is this the correct thing to do?

[amelia]

Hello,

I modeled this geometry in 3D and obtained the above results.

Now, I modeled this geometry as an axi-symmetric geometry using wedge patch type. Actually, since the geometry has an inner diameter the block is not a wedge but a hexahedron. Then I defined two patches with wedge type. Is this the correct thing to do?

[henry]

It's still a wedge desite it not reaching the axis so yes wedge is the correct patch type.

[amelia]

Thanks Henry.
I tried a wedge with 5º and it seems to work fine for lower velocities (RE=2000). when I increase the angular velocity of the lid, the results present very high oscilations using both turbulent and laminar models. Do you have any suggestion?

[amelia]

Hello,

I am having problems reaching a stable convergence at higher Re, even using upwind schemes for all divSchemes except two. I noticed that it is not possible to select upwind scheme for div(R) term in simpleFoam. Is there a reason for this?

Also, is this scheme implemented in OpenFoam:

Khosla and Rubin scheme "A diagonaly dominated second-order accurate implicit scheme"?

[henry]

1) What flux do you the upwind should be evaluated on for the term div(R)?

2) No.

[amelia]

I am using upwind on the convection term and I am having many oscillations. Do you have any idea why?

In my simulations using a 3D mesh of the flume I had less oscillations, although the magnitude of the velocity was much lower than the experimental results.
When using the 2D mesh at lower Re, the numerical results were very good, much better than other commercial software I used before.