 |
| Titel |
Large Eddy Simulations to determine the role of dispersive stresses in the urban canopy layer |
| VerfasserIn |
Andreas Christen, Marco Giometto, Marc Parlange |
| Konferenz |
EGU General Assembly 2013
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 15 (2013) |
| Datensatznummer |
250082115
|
|
|
|
|
|
| Zusammenfassung |
| Urban-scale weather and air pollution forecasting models need to realistically predict
conditions in the urban canopy layer (UCL) - the atmosphere in-between buildings where
people live and most activities take place. Nevertheless, for performance reasons,
forecasting models cannot resolve every detail of the flow field around individual
buildings and obstacles in a city. In common urban canopy parameterizations (UCPs),
exchange processes between the UCL and the overlying atmosphere - including
momentum transfer - are simplified to one-dimensional bulk flow representations,
where the time-averaged flow field is also horizontally averaged over a larger spatial
subset of the urban canopy. In the spatial averaging process of RANS equations,
additional covariance terms arise in the time-averaged momentum balance, called
‘dispersive stresses’. Physically, a dispersive stress can be explained as spatial correlation
between the mean horizontal flow and mean vertical flow around buildings at a
given height layer. Due to lack of knowledge on the role of dispersive fluxes, they
are neglected in all current UCPs and transfer formulations. Only limited CFD
studies for idealized cubical arrays show that dispersive fluxes are relevant and
important to properly describe the overall momentum transfer in those specific rigid
canopies.
The current contribution determines the role of dispersive stresses to the overall
momentum transfer for a more realistic urban canopy by means of large eddy simulation
(LES). LES takes into account the unsteadiness that characterizes canopy layer flows,
offering indisputably superior performances in predicting momentum exchange with respect
to traditional methods, in particular when the effects of canopy elements play a major role.
LES also showed to be able to properly represent the flow in areas of strong separation and in
wakes, features that are strongly present in urban canopies, where most RANS
and URANS models fail due to their under-prediction of turbulent kinetic energy
and turbulent fluxes. In the current simulation, we adopt a Cartesian grid approach
with cyclic boundary conditions, the subgrid terms are parameterized adopting a
Lagrangian scale-dependent dynamic model and the urban morphometry is taken into
account through a discrete, direct-forcing, immersed boundary approach which has the
numerical advantage, over other techniques, of not introducing stiff forcing terms in the
equations.
Simulations are run for neutral flow conditions with different morphometries,
informed by light-detection and ranging data of the morphometry of a real suburban
surface in Vancouver, Canada for which long-term tower-based turbulence and flux
measurements exist as a benchmark dataset. Different resolutions are tested in order to
validate the LES subgrid model and results are averaged over 800T where T is
the turnover time for the characteristic eddy shed by the given morphometry. The
vertical profiles of dispersive stress and horizontally averaged Reynolds stress are
computed and compared. The statistical analysis of the time averaged flow field
is done using spatial quadrant analysis, which allows determination of scale and
space fraction of regions effectively contributing to the overall dispersive stress. |
|
|
|
|
|
|