Background and Aims
Large wind farms are to be expected in the coming decades. We are talking
about farms of for example 20 times 20 turbines or larger. Of course
it is of utmost importance to be able to predict the production of such
farms accurately. However, little is known on cumulative effects of
wind turbine wakes. Therefore this project studies the wind turbine
interference and especially wake losses for larger offshore farms.
Model
The decrease of the wind speed in the wake of wind turbines can be compared
to the frictional drag exerted to the planetary boundary layer by surface
roughness. We calculated the effect to the boundary layer when the wind
encounters a wind farm. The wind farm was implemented as a sudden change
of the surface roughness. Our calculations showed that after an increase
of the surface roughness, the wind speed was decreasing up to large
distances downstream. Several wind farm models assume that equilibrium
is reached after 5 rows of wind turbines, our model showed that it will
take at least 10 rows before equilibrium is reached to some extent.
This means that our classical models predict a too high power output
for really large wind farms.
Experiment
In order to validate the results of the model we set up an experiment
in the boundary layer tunnel of TNO, Apeldoorn. We designed and manufactured
30 wind turbines on a scale of 1:400. The turbines are 25 cm diameter
and have a hub height of 25 cm. Our design philosophy has delivered
rotors which are very similar to full scale rotors regarding the axial
force and wake properties, however regarding the efficiency their performance
is less. The power coefficient is about 0.30, while is can be
between 0.45 and 0.50 for commercial rotors. In the wind tunnel the
surface roughness (excluding that due to the wind turbines) was set
to sea conditions: after scaling this was approximately 0.2 mm / 400.
Sometimes the roughness was adapted to onshore circumstances as well.
We installed several wind farms layouts and measured the wake losses
and the effects on the boundary layer.
Results and Discussion
The validation suggests that the wind velocity has not reached its equilibrium
value after 5 rows, which indicates that the numerical model can be
used for global studies of atmospheric flows above (and behind) large
wind farms. The words ?suggests? and ?indicates? were used since the
situation in the tunnel was different from the real situation in several
aspects. For several reasons the interpretation of the results was difficult.
Especially the accurate measurement of the wind speed was a difficulty.
A way out for this accuracy problem was the application of differential
measurements: we finally set up two farms next to each other in the
tunnel.
Conclusion
We can not rely on wind farm models that assume that the wind flow in
a farm is stabilizedafter five rows of turbines. Many models will
over predict production.
Recommendation
We recommend that much attention is to be paid to improvement of the
accuracy in the wind tunnel and to the organization of good experiments
with actual farms in the field. Both suggestions have been implemented
in successive projects, which already have been started.