Wind energyProject Using Upstream LIDAR Measurements
This Lessons Learnt Report outlines the key learnings Windlab Limited has learnt in the six month period of April 2019 – October 2019 in regard to their involvement in the ARENA Short Term Forecasting trial.
These points follow from the supporting information supplied below:
- Theory and practice show that randomness essentially dilutes the performance of all-time series prediction algorithms. The objective of our early work is to understand the origins of the randomness and, where possible, define a strategy for minimising its effects. Doing this gives the prediction algorithms the best chance of being able to predict the time series into the future. This applies to both the wind turbine SCADA measurements and the upstream LIDAR measurements.
- The placement of the LIDARs is key to optimising the correlation between upstream measurements and measured wind speed (or generation) at any turbine location. The placement of measurement locations gives a pattern of influence that is quite variable across a geographic area the size of even a small wind farm. As such, careful choice of measurement locations is key to coverage of allturbines.
- The measured hub height wind speed and the resultant generation contain a high level of randomness. This randomness varies throughout the day. This turbulence/randomness is strongest during the daytime when the sun’s heating provides buoyancy for the creation of turbulent eddies. This is seen very consistently in the diurnally changing variance in both the horizontal and vertical wind speeds and is expected to continue and become stronger as we move into summer when the heating is strongest. Any forecast algorithm needs to accommodate this diurnal change in variance.
- The LIDARs used in the project measure horizontal and vertical wind speeds every minute at several levels up to 133m. This capability has been key to building a complete picture of the atmospheric processes that are in play across the wind farm as we are trying to make generation predictions.