MESOSCALE AND CFD COUPLING: AN IMPROVED TECHNIQUE FOR PREDICTING MICROSCALE WIND

1
MESOSCALE AND CFD COUPLING AN IMPROVED TECHNIQUE
FOR PREDICTING MICROSCALE WIND

• ALICE ELY, PETER STUART, MIN ZHU
• WEDNESDAY 18th APRIL 2012

JOSÉ PALMA, CARLOS RODRIGUES, ROMAN
CHERTOVSKIH CENTRE FOR WIND ENERGY AND
ATMOSPHERIC FLOWS UNIVERSITY OF PORTO
2
Aim and Contents

• Can microscale wind climate predictions be
  improved by using a model chain?

• Overview of the example wind farm site
• The new coupled model
• Comparisons with measured data and a linear flow
  model

3
Example Wind Farm Site
No steep slopes No trees Linear flow model
expected to perform well
Original Mast
Thermally-driven flow
Additional Masts
4
Directionally Weighted Average Wind Speed
MS3DJH over predicted the wind speed at the
additional mast locations.
WRFMS3DJH downscaling showed improvement over
MS3DJH alone.
Stationary VENTOS did not substantially improve
upon MS3DJH results.
5
New Coupled Model VENTOS/M

• Developed by the Centre for Wind Energy and
  Atmospheric Flows, University of Porto under a
  joint venture with RES and Natural Power.
• Provides coupling of WRF mesoscale simulations
  with VENTOS CFD software.

• OUTPUT FROM
• VENTOS/M
• Velocity
• Wind direction (azimuth)
• Turbulence
• Shear
• Inflow angle
• Potential temperature

• INPUT FROM WRF
• Velocity
• Friction velocity
• Pressure
• Potential temperature
• Upward heat flux at the surface
• Planetary boundary layer height

Initial Conditions
Transient Boundary Conditions
6
The Model Chain

• Global model
• NCEP/NCAR T62
• Horizontal resolution 2.52.5
• 6 hourly output
• WRF model for each day
• Domain 400km x 400km x 20km
• Horizontal resolution 2000m
• Transient boundary conditions interval
• 1 hour
• VENTOS/M model for each day
• Domain 23km x 19km x 7km
• Horizontal resolution 200m at centre, expanding
  to 400m at domain edges

www.nws.noaa.gov
7
VENTOS/M for the Example Wind Farm Site

• All masts were operational between March and
  October 2010.
• 100 single days randomly selected as a
  representative sample from 214 available days.
• 4 days total simulation time for all days on
  RESs 128 core HPC.
• VENTOS/M results stored as 10 minute averages.

8
Wind Conditions across the Site for a Selected Day
Sunrise 0626 Sunset 1944
Height above ground level 71.5m
9
Time Series of Wind Speed for a Selected Day
4th September 2010
WRFVENTOS/M resolves smaller scales than WRF
alone. Accuracy is important and all stages of
the model chain.
10
Processing the VENTOS/M Results

• All 10 minute averaged results records over the
  100 selected days are binned into 30 directional
  sectors.
• Mean speed-ups are calculated from Mast A to
  Masts B, C and D for each directional sector.
• Speed-ups are used to predict the wind speed at
  Masts B, C and D given the measured wind speed at
  Mast A.

11
Prediction of Wind Speed at Mast B from Mast A
Mast B is 2.3km to the east of Mast A - this is
relatively nearby.
Both MS3DJH and WRFVENTOS/M closely match the
measured data.
12
Prediction of Wind Speed at Mast C from Mast A
Mast C is 6.4km south-east of Mast A.
MS3DJH remains close to the measured wind speeds
at Mast A, rather than Mast C. WRFVENTOS/M
more closely matches the measured wind speeds at
Mast C.
13
Prediction of Wind Speed at Mast D from Mast A
Mast D is 10.4km south-east of Mast A.
MS3DJH remains close to the measured wind speeds
at Mast A, rather than Mast D. WRFVENTOS/M
more closely matches the measured wind speeds at
Mast D.
14
Directionally Weighted Average Wind Speed

• WRFVENTOS/M results greatly improve the MS3DJH
  results at Mast D.
• MS3DJH results cancel out errors predominant
  wind directions are 30 and 60.
• WRFVENTOS/M results appear slightly worse than
  WRFMS3DJH, but
• WRFMS3DJH directional results not considered.
• WRFMS3DJH would not be available for sites
  with complex terrain.

15
Conclusions

• MS3DJH is unable to predict the full variation in
  wind speeds observed across the site.
• WRFVENTOS/M greatly improves upon MS3DJH
  predictions more closely match the measured wind
  speed at mast locations across the site.
• WRFVENTOS/M takes into account the surrounding
  area and the effects of varying atmospheric
  stability.
• WRFVENTOS/M can be used to visualise real flow
  and atmospheric conditions over the whole site.

Can microscale wind climate predictions be
improved by using a model chain?
Yes!
16
(No Transcript)