Remotely measuring the wind using turbine-mounted lidars: Application to power performance testing

Forward-looking wind lidars mounted on the nacelle of a wind turbines allow to remotely measure the flow upwind. The newest generation of nacelle lidar systems can sense the wind at multiple distances and multiple heights, and consequently has profiling capabilities. Wind lidars are cost-efficient and provide measurements more representative of the wind flow field than conventional meteorology mast.

For the purpose of power curve measurement, it is essential that lidars provide traceable measurements and to assess their measurement uncertainty.  A generic calibration methodology was developed, using the so-called whitebox approach. It consists mainly in calibrating the lidar primary measurementsof line-of-sight velocities.

The line-of-sight velocity is the projection of the wind vector onto the laser beam propagation path. The calibration is performed in situ, by comparing the lidar velocity measurements to a reference quantity itself traceable to the international standards of units. The uncertainty of the line-ofsight velocity measurements was assessed using a normative methodology (GUM) which is based on the law of propagation of uncertainties.

The generic calibration procedure was applied to two commercially developed nacelle lidars systems, the Avent 5-beam Demonstrator and the ZephIR Dual Mode lidars. Further, the lineof-sight positioning quantities such as inclination angles or beam trajectory werealso calibrated and their uncertainties assessed.

Calibration results were of high quality, with line-of-sight velocity measurements within 0.9% of the reference. In the lidar measurement process, line-of-sight velocities taken in multiple locations (different heights, distances, and directions) are used to reconstruct useful wind characteristics such as wind speed, direction, shear, etc.

Wind field reconstruction methods based on model-fitting techniques were developed. The model-fitting wind field reconstruction technique allows to clearly define the wind model – and state its inherent assumptions. Different wind models can be used without changing the general principles of the wind field reconstruction methods.

Two wind models were developed in this thesis. The first one employs lidar measurement at a single distance – but several heights –, accounts for shear through a power law profile, and estimates hub height wind speed, direction and the shear exponent. The second model combines the wind model with a simple one-dimensional induction model.

The lidar inputs were line-of-sight velocity measurements taken at multiple distances close to the rotor, from 0.5 to 1.25 rotor diameters. Using the combined wind-induction model, hub height free stream wind characteristics are estimated (speed, direction, shear, induction factor). With the help of a seven-month full-scale measurement campaign at the Nørrekær Enge wind farm, the model-fitting wind field reconstruction technique and models were demonstrated.

The same methods were applied to both the Avent 5-beam Demonstrator and ZephIR Dual-Mode nacelle lidars. Nacelle lidar estimatesof wind characteristics were compared to those measured by instruments mounted on a mast located 2.5 rotor diameters from the turbine on which the lidars were mounted. For wind directions in the ‘IEC free sector’, the wind speed comparison results showed that lidar-estimates where within 0.7% from the top-mounted cup anemometer measurements.

The secondary wind characteristics (direction, shear, induction factor) were also compared to reference quantities and proved to provide valuable information on the upstream flow field. The uncertainties of wind field characteristics estimated by the model-fitting reconstruction method were quantified using numerical error propagation techniques called Monte Carlo methods.

These numerical methods are particularly relevant to propagate errors trough complex non-linear models, since such models are outside the scope of the GUM methodology. The procedures used to apply Monte Carlo methods to wind field reconstruction codes were detailed. The uncertainty results are provided for a wide-range of wind field characteristics values, and for all the estimated wind characteristics.

In particular, the model wind speed uncertaintieswere shown to be equivalent to the cup anemometer uncertainty that was used to calibrate the lidar line-of-sight velocity. Finally, the methods were applied to power performance testing, using the experimental data of the Nørrekær Enge campaign. The IEC 61400-12-1 (ed. 2, 2017) standards for ‘Power performance measurements of electricity producing wind turbines’ provided the basis to develop procedures applying to nacelle-mounted lidars.

The measured power curves using wind speed measurements from the two profiling nacelle lidars and from the mast top-mounted cup anemometer werecompared. The power curve uncertainties were also quantified. Further, the annual energy production (AEP) was computed for a range of annual mean wind speeds.

At 8ms−1, the lidar-estimated AEP was within 1% to the one obtained with the cup anemometer. The combined wind-induction reconstruction technique represents a paradigm shift in power performance testing: it is no longer required to measure far upstream the rotor – between two and four rotor diameters – to approximate the free stream wind speed.

Instead, measurements taken close to the turbine rotor by nacellemountedprofiling lidars can be used to accurately estimate the free stream windspeed. In the future, nacelle lidars are likely to replace meteorological masts for turbine power performance testing.