27.02.2013 Views

Wind Energy

Wind Energy

Wind Energy

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

18 W. Bierbooms and D. Veldkamp<br />

3.4 Load Calculations Based on Normal and Constrained<br />

<strong>Wind</strong> Field Simulations<br />

The (un)constrained wind fields have been used as input for two standard<br />

wind turbine design tools: Flex5 and Bladed. Both packages feature among<br />

other things dynamic inflow, stall hysteresis and tip losses. Logarithmic wind<br />

shear is assumed and tower shadow is taken into account. In Flex5 the NEG<br />

Micon NM80/2750–60 is considered and in Bladed a generic 2 MW turbine.<br />

Both are pitch-regulated variable speed and have a diameter of 80 m, hub<br />

height of 60 m; the rated power is 2,750 kW and 2,000 kW respectively.<br />

As example the time series of the calculated blade root flapping moment<br />

is shown in Fig. 3.3 based on the wind fields of Fig. 3.2. The response based<br />

on the “measured” wind is assigned “measured” load. The 1P (Ω=1.9 rad/s)<br />

response is clearly visible. The mean response of the constrained simulations<br />

gets more detailed, with increasing number of anemomenters / constraints,<br />

and the standard deviation decreases since the wind fields gets more fixed.<br />

Comparison with the filtered “measured” load shows that the constrained<br />

response captures the low frequency phenomena well.<br />

In order to consider fatigue, the damage equivalent load ranges are considered<br />

of the blade root flapping moment (S–N slope 12). The load ranges<br />

and the frequencies of occurrence are found with the usual rain flow counting<br />

procedure.<br />

It is anticipated that constrained simulation results into a smaller scatter<br />

of the obtained equivalent damage. As measure for the scatter the COV<br />

(coefficient of variation) is taken. The convergence of the COV has been<br />

checked by doing 100 simulations for a particular situation. It turned out<br />

that about 20 simulations are enough; the uncertainty in the COV is about<br />

0.1 to 0.2.<br />

Blade root flapping<br />

moment [Nm]<br />

1.5 2<br />

2.5 x 10 6<br />

0.5<br />

0<br />

0 10 20 30 40 50 60<br />

1<br />

x 10 6<br />

1.5 2<br />

2.5<br />

0.5<br />

0<br />

0 10 20 30 40 50 60<br />

1<br />

Time (s)<br />

1.5 2<br />

6<br />

x 10<br />

2.5<br />

0.5<br />

0<br />

0 10 20 30 40 50 60<br />

1<br />

x 10 6<br />

1.5 2<br />

2.5<br />

x 10 6<br />

0.5<br />

0<br />

0 10 20 30 40 50 60<br />

1<br />

1.5 2<br />

2.5<br />

0.5<br />

0<br />

0 10 20 30 40 50 60<br />

1<br />

Time (s)<br />

Fig. 3.3. Generic turbine: Left, top: “measured” flap moment; bottom: filtered<br />

flap moment. Right, from top to bottom: mean flap moment (plus/minus standard<br />

deviation) for 0, 1 and 7 anemometers/constraints. Note low frequency similarity

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!