[santiagomarquezd]

Hi there, a few days ago Dr. Axel Larreteguy told me about some concerns about Figure 3.7 in Hrv. thesis (see attachment). Here we have the vectors and , which have units of area and then vector , the correction, which seems to take in account the displacement of from the face centroid , so that it has units of distance. The concerns are:

1. Why the differences in units?
2. Do the vector decomposition make sense as is drawn?

Below the figure it is stated that vector is not used

3. The correction vector is not used because simplification or because it is zero (I guess so)?

Following the work of Mathur and Murthy

http://www.tandfonline.com/doi/abs/1...915105#preview

it leads to a zero correction. If it is the wrong interpretation

4. How do we interpret the figure?

Thanks in advance,

[bigphil]

Originally Posted by santiagomarquezd

Hi there, a few days ago Dr. Axel Larreteguy told me about some concerns about Figure 3.7 in Hrv. thesis (see attachment). Here we have the vectors and , which have units of area and then vector , the correction, which seems to take in account the displacement of from the face centroid , so that it has units of distance. The concerns are:

1. Why the differences in units?
2. Do the vector decomposition make sense as is drawn?

Below the figure it is stated that vector is not used

3. The correction vector is not used because simplification or because it is zero (I guess so)?

Following the work of Mathur and Murthy

http://www.tandfonline.com/doi/abs/1...915105#preview

it leads to a zero correction. If it is the wrong interpretation

4. How do we interpret the figure?

Thanks in advance,

Hi,

Not sure if I can answer your questions, but

In Hrv's thesis (and in OpenFOAM), non-orthogonal correction is not employed on the boundary. This assumption is fine if we assume that phi does not vary along the boundary.
Although, if we assume that phi varies linearly along the boundary then non-orthgonal correction is required to maintain accuracy, as outlined in
E. De Villiers thesis (Fig. 4.7).

I believe that this correction is not performed in OpenFOAM because it may lead to mass flux across walls.

But boundary non-orthogonal correction can be employed through use of custom boundary conditions, such as fixedDisplacement (corrected version of fixedValue) in the solidMechanics branch of OpenFOAM-1.6-ext.

Hope it helped,
Philip

[santiagomarquezd]

Thanks Philip, it helped! I'm going to elaborate some ideas to post.

Regards.