LES for wind environments and dispersion at Bristol City
Matthew Coburn, Zheng-Tong Xie
Introduction & Background
Figure 1: Topography and building plan of the city of Bristol and surrounding areas. Source release location is shown in red, with numbered measurement locations. (Coordinate unit is 2.5m)
- A 5kmx5km area of interest was selected to incorporate key terrain elements assumed to have an impact on the local wind properties.
- 3D Terrain elevation data was gathered from Ordnance Survey.
- A 4kmx4km area using LIDAR data at 1m resolution is surrounded by DTM data at a spatial resolution of 5m, the 2 datasets were then merged.
- Interpolation was done to smooth the geometry to the inlet plane.
- Wind data collected within Bristol at location 8 in figure 1.
- The wind angle is 281 degree.
- This wind angle is represented in the U and V components for the prescribed inlet profiles in figure 2.
Acknowledgements
- The field data were provided by James Matthews and Dudley Shallcross @ the University of Bristol
- The computations were carried out on the Archer2 and Iridis5 computer systems
- This project is funded by NERC
Inflow profiles for mean wind velocity and synthetic turbulence generation. (Surrey wind tunnel data)
Figure 2: Prescribed vertical profiles at inlet.
Figure 3: 3D view of Bristol domain, with lidar and DTM terrain combined.
| |
| Syn. Inflow. Turb, Dirichlet |
| |
| |
| |
| |
| |
Figure 4: Boundary Conditions.
- Palm4U running on Archer 2.
- 1200 CPUs x 24 hours
- 2400x2400x900 grid
- Resolution 2.5m
- 1 hour initialisation
- Release for 15 mins
- Sample for 30 mins (from start of release)
Boundary conditions used for the case
- Figure 5 shows the mean concentration contour on the vertical plane across the source along the axial direction, source release location approximately 2700m.
- The plume shows a second peak downstream likely due to the large buildings and the upslope from approximately 3800m-4400m.
Figure 5: Source concentration along the source centre line.
- Figure 6 shows over prediction of the time average dimensionless concentration measured at each of the locations compared to field measurement data.
- Location 2, Brendan Hill, seems to give the closest approximation, overall the difference between CFD and field measurements seem similar.
- Figure 7 shoes the time series of dimensionless concentration at locations 2-5.
- The convection speed and distance from the centreline of the plume affects time taken for the scalar cloud to reach the station.
- Simulation data provides strong evidence that the sampling length used in the field measurements is not sufficient. The plume, especially in the region of the university is still close to it’s peak value.
- Future work focuses on using different wind angles (+/-10 degrees) to assess the uncertainty due to changes in wind direction.
Figure 6: 30 min average normalised concetration
Figure 7: Instantaneous normalised concertation
- In order to gain a better understanding of O(100m) terrain effects within a real urban area.
- Part of the ASSURE project (Bristol, Imperial, Reading, Southampton, Surrey.)
- Figure 1 shows the location of the source release and measuring station locations used in field measurements.
- The release location is given as: (s)
- The locations are as follows:
- 1) We the Curious,
- 2) Brandon Hill,
- 3) Wills Roof,
- 4) Med Teaching Lab,
- 5) Mobile lab,
- 6) Site 6,
- 7) Wills Ground (same location 3).
- Location 8 is the location of a 3-D sonic anemometer for local wind measurements.