Modelling Of Wind Speed and Turbulence Intensity For A Forested Site In Complex Terrain

1
Modelling Of Wind Speed and Turbulence Intensity
For A Forested Site In Complex Terrain

• C. Montavon1, I. Jones1, D. Malins2,
• C. Strachan2, R. Spence3, R. Boddington3

1ANSYS UK Ltd, 2Scottish Power Renewables,
3SgurrEnergy Ltd
2
Motivation

• More and more onshore sites developed in complex
  terrain and complex forestry environment.
• Associated risks
• Separation, negative shear exponent factors
• Increased turbulence
• ? implications for turbine longevity and energy
  output
• On such sites, standard industry tools
  (linearised models) used outside of the envelope
  where they are meant to operate
• CFD models increasingly advocated to be better
  suited for difficult sites.
• Question is what can CFD deliver in terms of
  accuracy?
• ? need for validation

3
Contents

• Part 1
• Harestanes site description
• Modelling assumptions
• Mast to mast cross predictions
• Wind speeds
• Turbulence intensity
• Part 2
• Comparisons with Galion LIDAR campaign
• Conclusions

• Acknowledgements
• Carbon Trust for providing funding to the POWFARM
  project
• Scottish Power Renewables for providing us with
  their mast data, and allowing the deployment of
  the Galion LIDAR on their site.

4
Complex terrain

• Harestanes is situated in the Scottish Borders
• Developed by Scottish Power Renewables
• Site elevation between 50 and 700m
• Wind regimes predominantly from the south-west
  and east/north east
• Complex terrain but not very complex as only few
  regions where slopes exceed critical slope for
  separation

Wind rose at Holehouse Hill
5
but mostly complex forestry
Forestry cover in the proximity of the deployment
location for the Galion LIDAR ? Lots of forestry
coupes, each with its own tree height
6
Modelling assumptions

• Simulations done with WindModeller (ANSYS CFD),
  solving for
• Transient RANS, SST turbulence model
• Atmospheric stability accounted for via equation
  for potential temperature, buoyancy effect in
  turbulence model
• Lopes da Costa forestry model (variable tree
  height, constant loss coefficient), background
  roughness of 0.05m
• details in paper

7
Masts on site

• 4 masts on site, 3 near hill tops, 1 in a
  location with difficult flow conditions (valley,
  proximity to forestry)
• Cross prediction done for 3 masts with long
  concurrent time series

All masts with data at 70m, 60m, 40m, 30m Hill
top masts Holehouse Hill Hareshaw Rig Valley
mast Bran Rig
8
Mast to Mast Cross Prediction (wind speed)

• Maximum relative errors in wind speed cross
  predictions for three model configurations
• Purely neutral
• Conventionally neutral i.e. stable conditions in
  free stream, with potential temperature gradient
  of US standard atmosphere, and neutral conditions
  at ground (adiabatic).

3 masts (70m only) 3 masts (70m only) 3 masts (70m only) 3 masts (all heights) 3 masts (all heights) 3 masts (all heights)
Model BRAN-HOL BRAN-HAR HOL-HAR BRAN-HOL BRAN-HAR HOL-HAR
neutral 11.8 13.3 1.8 24.0 24.0 3.2
stable 0.4 4.1 5.4 2.9 6.5 6.2
9
Shear exponent factor (40 to 60m) predictions
Purely neutral
Including freestream stability
Holehouse Hill
Bran Rig
Data red symbols and error bars (average
stdev) Model black symbols
10
Relative wind speed between masts, normalised by
HOL 70m
Purely neutral
Including freestream stability
Holehouse Hill
Bran Rig
Data coloured symbols and error bars (average
stdev) Model black symbols
11
Relative TI between masts at 70m, normalised by
HOL 70m
Purely neutral
Including freestream stability
Hareshaw Rig
Bran Rig
Data red symbols and error bars (average
stdev) Model black symbols
12
Separation prediction
Massive separation bubble crossing RHI planes
s270
s310

• 2 RHI scans (azimuth 275 and 290) are analysed
  for 2 distinct wind directions (270 and 310)

Not really explained by slopes exceeding
critical angles
Still not explained, even when making critical
angles a function of forestry height. ?
Separation off forestry edge?
13
CFD vs LIDAR sector 270 - LOS Velocity
RHI 1
RHI 2
GALION
CFD
14
CFD vs LIDAR sector 310 - LOS Velocity
RHI 1
RHI 2
GALION
CFD
Vref 10 m/s _at_ 30m
15
CFD vs LIDAR sector 310 - LOS Velocity
RHI 1
RHI 2
GALION
CFD
Vref 18 m/s _at_ 30m
16
Conclusions

• Simulations with stability showed significant
  improvement in average wind speed cross
  predictions, particularly so at difficult mast
  location
• Decent prediction of TI at 70m from purely
  neutral simulation, but TI ratio by direction
  between masts better reproduced when free stream
  stability is included.
• Massively separated flow region seen in CFD for
  specific wind directions.
• Not necessarily explained by slopes alone.
• Even taking effect of forestry into account on
  reduction of critical slope for separation does
  not seem to explain why the flow separates so
  much more for sector 310 than 270.
• ? likely that separation is induced by well
  defined forest edge
• Massive flow separation for sector 300-310
  confirmed by Galion LIDAR data