wind farm e ciency assessed by wrf with a statistical
TRANSCRIPT
Wind farm efficiency assessed by WRFwith a statistical-dynamical approach
P.J.H. Volker, A.N. Hahmann, J. Badger,
and H. Sørensen
DTU Wind Energy (Risø Campus)
1
Motivation
Adams and Keith, Environ Res Lett, 2013
“The results suggest that the maximum energy thatcan be extracted by turbine arrays at these scales isabout 1 W m-2”
Miller et al., Proc Natl Acad Sci, 2015
“. . . expanding wind farms to large scales willlimit generation rates, thereby constraining meanlarge-scale generation rates to about 1 W m-2 evenin windy regions”
2
Method of Adams and Keith 2013
They use the WRF model to simulate:• Actual Power Density (APD) (wake effects with wind farm parametrisation)• Reference Power Density (RPD) (no wake effects)
Simulations over the Great plains in winter/summer 2006
The Power Density (PD) in function of
• Wind farm size 103 – 105 km2
• Turbine density 0.25 – 16 km−2
3
Result of Adams and Keith 2013
Actual (wakes) versus Reference or expected (no wakes) Power Density (PD)
APD/RPD is the degree to which the turbine drag reduces the wind speed
It seemed that the APD converges to around 1 W m-2
4
Consequence
Current: the 20 km2 offshore wind farm Horns Rev I(8 MWi km-2
)has a annual power density of up-to 3.98 W m-2
Future: very large(104- 105 km2 ∼ Dogger Bank
)wind farms would
have a power production per area of 25% compared to Horns Rev I
5
Experimental set-up of WRF
4 wind farm sizes
• Small (Horns Rev I)
• Medium (London Array)
• Large (Dogger Bank)
• Very large (Iowa)
3 turbine spacings
• 5.25 D0
• 7 D0
• 10.5 D0
2 WF schemes
• WRF-WFFitsch et al.2012
• EWPVolker et al.2015
Number of 2 MW turbines
Small Medium Large Very Large
Wide (10.5 D0) 6 × 6 22 × 22 202 × 202 402 × 402
Intermediate (7 D0) 9 × 9 33 × 33 303 × 303 603 × 603
Narrow (5.25 D0) 12 × 12 44 × 44 404 × 404 804 × 804
Volker et al.: Prospects for generating electricity by large onshore andoffshore wind farms Environ. Res. Lett. 2017
6
Wind Conditions
For each wind farm we simulated a range of idealised case experimentsbetween the turbine cut-in and cut-out wind speed.
From the set of simulations we define 3 wind conditions
Region A (land)Moderate windsGreat Plains
Region B (sea)Strong windsNorth Sea
Region C (sea)Very strong windsStrait of Magellan
Region A
0.00
0.05
0.10
0.15
0.20
0 10 20 30U
(ms−1
)
Freq
uenc
y
Region B
0.00
0.05
0.10
0.15
0.20
0 10 20 30U
(ms−1
)
Freq
uenc
y
Region C
0.00
0.05
0.10
0.15
0.20
0 10 20 30U
(ms−1
)
Freq
uenc
y
a b c
7
Wind speed reduction in very large wind farms
At equilibrium wind speed a balance between the dragforce f(Ct, U) and turbulent influx of momentum
EWPWRF-WF
Region ARegion BRegion C
4
6
8
10
12
0 50 100 150Distance (km)
Uh(
ms−
1 )
• Offshore there is less mixing and equilibrium is reached much later
• Equilibrium wind speed remains higher with better wind conditions
8
Actual vs Reference PD for very large wind farms
1Wm−2
Adams and Keith (Great Plains)
Parametrisatrion ApproachEWPWRF-WF
Region A
0123456789
101112
0 1 2 3 4 5 6 7 8 9 10 11 12RPD
(Wm−2
)
APD
( Wm
−2)
• In the Great Plains also 1 W m-2 (differences are due to parametrisation)
9
Actual vs Reference PD for very large wind farms
1Wm−2
2Wm−2
Adams and Keith (Great Plains)
Parametrisatrion ApproachEWPWRF-WF
Region ARegion B
0123456789
101112
0 1 2 3 4 5 6 7 8 9 10 11 12RPD
(Wm−2
)
APD
( Wm
−2)
• In the Great Plains also 1 W m-2 (differences are due to parametrisation)
10
Actual vs Reference PD for very large wind farms
1Wm−2
2Wm−2
3.5Wm−2
Adams and Keith (Great Plains)
Parametrisatrion ApproachEWPWRF-WF
Region ARegion BRegion C
0123456789
101112
0 1 2 3 4 5 6 7 8 9 10 11 12RPD
(Wm−2
)
APD
( Wm
−2)
• In the Great Plains also 1 W m-2 (differences are due to parametrisation)• However: In regions with very strong winds the APD is around 3.5 W m-2
⇒ The APD is not limited, but depends strongly on wind (and roughness) conditions
11
Wind farm efficiency (APD/RPD)
0
25
50
75
100
102 103 104 105
Wind farm area
Effi
cien
cy(%
)
Region A
– Typeset by FoilTEX – 1
Region A: A Very large wind farm (160.000 turbines) produces 700 TWh
Region B: A cluster of nine medium wind farms (total 9.801 turbines) 77 TWh
Region C: A small wind farm 1 TWh (50% more than Horns Rev I). A verylarge wind farm would produce 1.7 PWh
12
Wind farm efficiency (APD/RPD)
0
25
50
75
100
102 103 104 105
Wind farm area
Effi
cien
cy(%
)
0
25
50
75
100
102 103 104 105
Wind farm area
Effi
cien
cy(%
)
Region A Region B
– Typeset by FoilTEX – 1
Region A: A Very large wind farm (160.000 turbines) produces 700 TWh
Region B: A cluster of nine medium wind farms (total 9.801 turbines) 77 TWh
Region C: A small wind farm 1 TWh (50% more than Horns Rev I). A verylarge wind farm would produce 1.7 PWh
13
Wind farm efficiency (APD/RPD)
0
25
50
75
100
102 103 104 105
Wind farm area
Effi
cien
cy(%
)
0
25
50
75
100
102 103 104 105
Wind farm area
Effi
cien
cy(%
)
0
25
50
75
100
102 103 104 105
Wind farm area
Effi
cien
cy(%
)
Region A Region B Region C
– Typeset by FoilTEX – 1
Region A: A Very large wind farm (160.000 turbines) produces 700 TWh
Region B: A cluster of nine medium wind farms (total 9.801 turbines) 77 TWh
Region C: A small wind farm 1 TWh (50% more than Horns Rev I). A verylarge wind farm would produce 1.7 PWh
14
Conclusion
Power Density
• The power density is not limited to 1 W m−2 as previously assumed
• Instead it depends also for very large wind farms on the local
up-stream wind and surface conditions
Wind farm efficiency/production
• In onshore regions with moderate wind conditions very large wind
farms can significantly contribute to the electricity production
• Offshore, clusters of smaller wind farms are more efficient
• However, in regions with very strong winds very large wind farms
become also efficient
15
(II) Efficiency of a wind farm cluster in 2 regions
Can the overall cluster efficiency be improved by separating the same number ofturbines on fixed area (3658 km2) in 2 wind farms?
Separation in km:
S00 S10 S20 S30
WF1 WF2 WF1 WF2 . . .
Density 1 with 9145 turbines:
- S00 5.0 W m-2
- S10 6.0 W m-2
- S20 7.5 W m-2
- S30 10 W m-2
Density 2 with 12802 turbines:
- S00 7.0 W m-2
- S10 8.4 W m-2
- S20 10.5 W m-2
- S30 14.0 W m-2
16
Wind speed reduction for different WF spacings
S00
6.5
7.0
7.5
8.0
8.5
9.0
0 25 50 75Distance (km)
Uh(
ms−
1 )
Hub-height wind speed
Highest efficiency is a balance between:
• wind speed reduction in the wind farms f (turbine density)
• wind speed recovery between the wind farms
17
Wind speed reduction for different WF spacings
S00S10
6.5
7.0
7.5
8.0
8.5
9.0
0 25 50 75Distance (km)
Uh(
ms−
1 )
Hub-height wind speed
Highest efficiency is a balance between:
• wind speed reduction in the wind farms f (turbine density)
• wind speed recovery between the wind farms
18
Wind speed reduction for different WF spacings
S00S10S20
6.5
7.0
7.5
8.0
8.5
9.0
0 25 50 75Distance (km)
Uh(
ms−
1 )
Hub-height wind speed
Highest efficiency is a balance between:
• wind speed reduction in the wind farms f (turbine density)
• wind speed recovery between the wind farms
19
Wind speed reduction for different WF spacings
S00S10S20S306.5
7.0
7.5
8.0
8.5
9.0
0 25 50 75Distance (km)
Uh(
ms−
1 )
Hub-height wind speed
Question:
• Can the overall wind farm cluster efficiency be higher by separating wind farms?
20
Efficiency of WF1 and WF2
Region B (blue) and Region C (green)
Density 1Density 2
0.6
0.7
0.8
0 10 20 30Wind Farm separation (km)
Effi
cien
cy(%
)
Efficiency of up-stream WF1
Density 1Density 2
0.85
0.90
0.95
1.00
0 10 20 30Wind Farm separation (km)
APD
WF2
/WF1
(%)
Power reduction of down-stream WF2
• The efficiency decreases withincreasing turbine density
• In region B the efficiency is alwayslower than 70%, because the windfarm size is too large
• The power reduction for the 2attached wind farms is up-to 20%
• The power reduction does notconverge to 1!
21
Overall efficiency for the 1st case
Single WFTot. cluster
0.45
0.50
0.55
0.60
0.65
0.70
0 10 20 30Wind Farm separation (km)
Eff
cien
cy(%
)
Region BSingle WFTot. cluster
0.65
0.70
0.75
0.80
0 10 20 30Wind Farm separation (km)
Eff
cien
cy(%
)
Region C
• The chosen wind farm is to large for the Region B wind conditions,since the efficiency is relatively low
• In the experiments with the lower installed capacity a wind farmseparation could improve the efficiency
22