Unphysical horizontal velocity pattern in vicinity of vegetation

[r-broek]

Dear FDS developers,

The horizontal velocity profile of an MO wind profile blowing towards a tree gives the following results:

t = 0:

t > 0:

The general behaviour is, I believe, correct. However, in the vicinity of the treetop, we observe an unphysical pattern. This is not due to Smokeview, as the CELL_CENTERED values also show such patterns.

Based on older discussions, it seems that this may be due to discretization errors. However, I find the magnitude difference of these patterns at the treetop unusually high. Is there any way to reduce this?

The Lagrangian particles that model the tree are given a constant drag coefficient of 2.8. The finest meshes that contain the tree are $1.2 \times 1.2 \times 2.4 \ \mathrm{m}^3$ and have grid cells of $7.5 \times 7.5 \times 7.5 \ \mathrm{cm}^3$. The tree has $D = 1.19 \ \mathrm{m}$, $H = 3.96 \ \mathrm{m}$.

[rmcdermo]

This is simply numerical (dispersion error) from the central differencing scheme in the U-component momentum equation. This is inevitable with low-order central schemes at practical grid resolutions. Look at either the velocity magnitude or kinetic energy, these should look better.

[r-broek]

Thank you!