The non-linear NLMSFD model of atmospheric boundary-layer flow over topography
has been carefully analysed and in particular the relaxation parameter in the iterative scheme has been tuned to
provide convergent solutions for steeper topography than had previously been possible. So far this has been done
with a 2-D surface layer version of the code but a 3-D planetary boundary layer version has also been developed.
Work on flow over multiple roughness changes has continued and a 3-D version has been tested. Both models were
used to generate predictions of the modifications to flow over the Bolund hill in a "blind comparison" organised
by the Riso laboratories in Denmark with a workshop held there in December 2009. We have also used the 3D model
for test studies of flow over Lake Erie offshore from Cleveland and will use it for other Great Lakes studies.
Plans for 2010-2011
We now have a 3D roughness variation model running and once it is well tested we will collaborate with Zephyr North
personnel in order to incorporate it in their MS-Micro and WindFarm models for practical applications in wind farm
design. These wind flow
calculations are critical for evaluating the commercial viability of wind farm proposals. Applications of our 2D
NLMSFD model have now included revisiting a comparison with some wind tunnel studies reported by Gong et al (1996).
With a 3D time dependent model we hope to investigate the development of steady 3D longitudinal vortices in the
flow over 2D waves. This could prove to be an interesting and significant study and we are in communication with
Dr Peter Sullivan at the National Center for Atmospheric Research who is planning a parallel study using his
Large Eddy Simulation (LES) model. Comparing his LES results with those from a RANS (Reynolds Averaged Navier
Stokes) model like ours for this, and other flows, will be a valuable exercise.
So far we have treated roughness and topographic flow perturbations separately, and for real terrain in models
like MS-Micro, we combine the perturbations linearly. This is computationally efficient, and the NLMSFD model is
also highly efficient in comparison with other approached to solving RANS equations. We are however interested
in a fully non-linear model and exploring that option is now a significant goal for the second year.
Reference: Gong, W., Taylor, P.A. and Dornbrack, A., 1996, Turbulent boundary-layer flow over fixed, aerodynamically rough, 2D sinusoidal waves, J. Fluid Mech., 312, 1-37.
|
