Escolar Documentos
Profissional Documentos
Cultura Documentos
1,2 2 3 4 5 6
R.J. Barthelmie , S.T. Frandsen , K. Hansen , J.G. Schepers , K. Rados , W. Schlez , A.
Neubert6, L.E. Jensen7, S. Neckelmann8,
1
Atmospheric Science Program and Center for Environmental Research, Indiana University,
USA (rbarthel@indiana.edu)
2
Risoe DTU Laboratory for Sustainable Energy, Denmark
3
DTU, Denmark
4
ECN, The Netherlands
5
National Technical University of Athens, Greece
6
Garrad Hassan and Partners, Germany/UK
7
DONG Energy, Denmark
8
Vattenfall, Denmark
Summary Introduction
The aim of this research as part of the As wind farms offshore increase in size,
UpWind project is to improve wind farm one of the research challenges is to
modelling and address the issue of evaluate how to model interactions between
providing more accurate power output the individual turbines, the atmosphere and
predictions for wind. Specifically the focus neighbouring turbines so as to accurately
is on wake modelling in large offshore wind predict power output before wind farm
farms. Detailed case studies of power construction. Here we present research
losses due to wakes at the large wind farms from the UpWind project where specific
at Nysted and Horns Rev have been case studies at Nysted and Horns Rev have
analysed. Despite relatively high stochastic been simulated with a range of wind farm
variability in the observational data, and CFD models to evaluate wind farm
similarities in the depth and width of the modelling. The UpWind project presents a
wake are presented. The case studies are unique platform for model evaluation
simulated with a range of wind farm and because the co-operation of a number of
computational fluid dynamics (CFD) groups means that more models can be
models. Results shown indicate power evaluated on standardised cases. It is also
losses due to wakes can be modelled, fortunate that DONG Energy and Vattenfall
provided that the standard models are have allowed data from a number of cases
subject to some modifications. We also studies to be used in this project.
present some of the first simulations of Preliminary evaluation of the models
large offshore wind farms using CFD. presented in [1] suggested that standard
Despite this progress, wake modelling of wind farm models were under-predicting
large wind farms is still subject to an wakes (i.e. over-predicting power output)
unacceptably high degree of uncertainty. while computational fluid dynamics models
(CFD) were over-predicting wake losses.
Re-evaluation of model parameterisations
and constants improved model
performance in some cases [2]. However, a
wide range of model predictions remains,
together with high variability of the
observations even for specific and narrow
wind speed and direction bins. Some of the
difficulty in evaluating models arises from
what might be termed natural variability
deviations of wind speed and direction from
stationary conditions such as when a front 6051000
A1
passes or from eddies on varying scales. 6050000
A2
B1
C1
D1
E1
B2 F1
C2 G1
D2 H1
Part of the difficulty also derives from 6049000
A3
B3
C3
E2
F2
G2
ER-15o
UTM Coordinates N
D3 H2
A4 E3
B4 F3 ER-10o
C4 G3
issues pertaining to measurement error and 6048000
A5
B5
C5
D4
E4
F4
G4
H3
A6
D5
E5
H4 ER-5o
B6 F5
G5
model limitations [3]. Here we examine a A7
B7
C6
D6
E6
F6
H5 ER
6047000 C7 G6
D7 H6
A8 E7
B8 F7 ER+5o
new set of simulations from the two large A9
B9
C8
D8
E8
F8
G7
H7
6046000 C9
D9
G8
H8 ER+10o
E9
Danish wind farms. F9
G9
H9 ER+15o
6045000
672000 674000 676000 678000
UTM Coordinates E (M)
6152000
Observations
1 11 21 31 41 51 61 71 81 91 ER-15o
The data are from the SCADA systems 6151000 2 12 22 32 42 52 62 72 82 92 ER-10o
UTM Coordinates N (m)
3 13 23 33 43 53 63 73 83 93 ER-5o
from the large offshore wind farms at 6150000
4 14 24 34 44 54 64 74 84 94 ER
6 16 26 36 46 56 66 76 86 96 ER+10o
Figure 1 shows the locations and layouts of 7 17 27 37 47 57 67 77 87 97 ER+15o
6148000
6147000
Wind farm Nysted Horns Rev predict wake losses, a number of case
0.6 0.6
0.4 0.4
2552.5o 2602.5o
0.2 0.2
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
1.2 1
Normalised power
1 0.8
0.8
0.6
0.6
0.4 0.4
2652.5o 2702.5o
0.2 0.2
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
1 1
Normalised power
0.8 0.8
0.6 0.6
0.4 0.4
2752.5o 2802.5o
0.2 0.2
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
1
Normalised power
0.8 Observations
WAsP
0.6 GH Windfarmer
ECN
0.4
NTUA
2852.5o
0.2
1 2 3 4 5 6 7 8 9 10
Turbine
Figure 3. Case studies of normalised power at Horns Rev: comparison of models with
observations for U=8.00.5 ms-1. Error bars shown are one standard deviation of the
observations. Note that observations, NTUA, GH WindFarmer and ECN simulations are 2.5
the centre angle but WAsP simulations are 5.
Normalised power
1 1
0.8 0.8
0.6 0.6
0.4 0.4
2632.5o 2682.5o
0.2 0.2
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
Normalised power
1 1
0.8 0.8
0.6 0.6
0.4 0.4
2732.5o 2782.5o
0.2 0.2
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
Normalised power
1 1
0.8 0.8
0.6 0.6
0.4 0.4
2832.5o 2882.5o
0.2 0.2
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
Normalised power
0.8
0.6 Observations
WAsP
0.4 GH Windfarmer
2932.5o ECN
0.2
1 2 3 4 5 6 7 8
Turbine
Figure 4. Case studies of normalised power at Nysted: comparison of models with observations.
Error bars shown are one standard deviation of the observations. Note that observations, GH
WindFarmer and ECN simulations are 2.5 the centre angle but WAsP simulations are 5.
Figure 5. Average efficiency by direction (2.5). Left: Nysted. Right: Horns Rev.