Professional Documents
Culture Documents
net/publication/261348660
CITATIONS READS
3 1,409
5 authors, including:
All content following this page was uploaded by Asher Yahalom on 13 January 2017.
Abstract—In this paper, two approaches to calculating outcome is given by a numerical value without any insight into
Capacity Factor of fixed speed wind turbines are reviewed and the CF formation.
then compared using a case study. The quasi-exact approach In this note, the quasi-exact and approximate approaches to
utilizes discrete wind raw data (in the histogram form) and calculating the CF are reviewed and then compared using a
manufacturer provided turbine power curve (also in discrete
case study. It is shown that the obtained results are similar,
form) to numerically calculate the capacity factor. On the other
hand, the approximate approach employs a continuous justifying the use of analytically derived relations in spite the
probability distribution function, fitted to the wind data as well fact that they describe an approximate solution only. Hence,
as continuous turbine power curve, resulting from double- analytical derivations of additional merits of wind turbine
polynomial fitting of manufacturer-provided power curve data. performances which rely on capacity factor (such as Turbine
The latter approach, while being an approximation, can be solved Performance Index [5], Turbine-Site Pairing Performance [6],
analytically thus providing a valuable insight into the factors, etc.) are justified and the results obtained using these relations
affecting the Capacity Factor. Moreover, several other merits of are reliable.
wind turbine performance are based on the analytically derived
equations and hence are an approximation as well. The note II. THE WIND DATA
shows that the results obtained by employing both approaches
are very close, thus the analytically derived approximations are The wind speed data is usually provided by meteorological
valid and may be used for wind turbine performance evaluation. stations as raw matrix of wind speed versus time at 10m height,
while sample times vary from 10 minutes to 1 day. In reality,
Index Terms— Wind turbine, Capacity Factor, Data fitting, the sample time is much higher than stated and the available
Probability Distribution Function data sample is actually an average of tens to thousands of faster
samples. An example of monthly wind speed raw data
I. INTRODUCTION represented by 10 minutes samples is shown in Fig. 1.
Since the wind blows inconsistently, wind turbines barely
18
operate at their rated power. Therefore the concept of capacity
factor (CF) is usually engaged to assess the expected wind 16
usually formulated as 8
E P
P(v) f (v)dv ,
1 6
CF (1)
PR PR 4
0
where E[∙] is the mean value operator, P(v) and PR are the 2
is a discrete series rather than a continuous function, and the Fig. 1. A typical monthly wind speed raw data
wind PDF in (1) is the result of fitting the discrete raw wind
data to an a priori assumed PDF. Hence, (1) is actually an The raw vector can be either transformed into a histogram
approximation. Nevertheless, several analytic solutions of (1) (discrete PDF) or fitted to a known continuous PDF, typically
have been presented in the literature [2], [3], allowing thorough of Weibull type, as shown in Fig. 2. When creating a
understanding of the factors, affecting the CF. The more histogram, the bins are typically chosen to be 1m∙s-1 wide to
accurate (referred to as quasi-exact thereafter) solution, taking match the resolution of the manufacturer provided turbine
into account the original rather than processed wind speed and power curve data (explained in Section 3), resulting in the
power curve data, exists in a spreadsheet form only [4] and its following discrete PDF,
f HST (v) f (vi ), vi 0.5 v vi 0.5 , (2) turbine is stall regulated; therefore the output power in the
rated region reduces with the increase of wind speed, as shown
where f (vi ) is the magnitude of the histogram bin, centered in Fig. 3.
at vi .
TABLE I. NEG MICON 1000/60 FIXED SPEED TURBINE POWER CURVE
As to Weibull fitting, several methods of deriving Weibull DATA
parameters from the raw data were compared in [7]. The
maximum likelihood estimates of the Weibull distribution
parameters are typically employed in dedicated software
packages, e.g. MATLAB.
0.16
1000 Manufacturer Data
Polynomial Fitting
0.14 900 ZOH Fitting
0.12
Histogram 800
Weibull fit
700
0.1
power, [kW]
600
probability
0.08 500
400
0.06
300
0.04
200
0.02 100
0
nonrated region rated region
0
0 2 4 6 8 10 12 14 16 v vR vF
0 C
5 10 15 20 27
wind speed, [m/s]
wind speed, [m/s]
Fig. 2. Histogram and Weilbull PDF fit of wind speed raw data of Fig. 1. Fig. 3. Power curve of NEG Micon 1000/60 fixed speed wind turbine
Weibull PDF is defined as The power curve data, provided by the manufacturer may be
k v ( v )k extrapolated using either Zero Order Hold (ZOH) method [4]
fWBL (v) ( )k 1 e c (3)
or polynomial fitting [3], as shown in Fig. 3. In the former
c c
with parameters c and k are related to the site wind speed mean case, each data point P(vi) is replaced by a 1m∙s-1 wide discrete
and standard deviation as bin, having the same magnitude, thus creating the following
staircase approximation of the power curve,
1
v c 1 (4) PZOH (v) P(vi ), vi 0.5 v vi 0.5 , (7)
k
for i = 1…N.
and
In the latter case, the data is divided into nonrated and rated
2 1 subsets and each subset is fitted to a distinct polynomial,
v c 1 2 1 (5)
k k creating the following polynomial approximation of the power
curve,
respectively, where
0, v vC
( x) t x 1et dt
0
(6) n
a1i v i , vC v vR
is the complete Gamma function. In case the wind raw data of i 0
a site is absent, but the mean and standard deviation of the
PPOLY (v) PR n . (8)
a vi ,
wind speed are known, Weibull PDF is usually assumed and its 2i vR v vF
parameters are calculated using (4) and (5). i 0
III. TURBINE POWER CURVE 0, v vF
The power production of a wind turbine is associated with one IV. CAPACITY FACTOR CALCULATION
of the two nonzero regions of the power curve: the non-rated
region for wind speeds between the cut-in speed vC and the In case the raw wind data of the site at hub height is available,
rated speed vR or the rated region for wind speeds between the a discrete PDF in the histogram form, described by (2), may be
rated speed and the furling (or cut-out) speed vF. The turbine constructed. Hence, combining (2) and (7) as
power curve is usually supplied by the manufacturer as N-point PZOH (v) f HST (v) P(vi ) f (vi ), vi 0.5 v vi 0.5 (9)
discrete series {vi, P(vi)}, i = 1…N, with vi – vi-1 = 1, i.e. the and substituting into (1), the quasi-exact solution for capacity
data resolution is 1m∙s-1. For example, consider a NEG Micon factor is given by
1000/60 fixed speed turbine power curve, given in Table I. The
1 N TABLE IV. 2008 MONTHLY QUASI-EXACT AND APPROXIMATE CAPACITY
CFQE
PR
P(vi ) f (vi ) .
i 1
(10) FACTORS
i i
n vF k n
( i v i v i v
CFAP a v e ci
)
i
2i F
c
( ) a1i (( C )k , ) (a1i a2i ) (( R ) k , ) a2i (( F ) k , ) ,
i 0 i 1 k k c k c k c k
where
y 30
probability
0.1 Weibull fit
Ariel, Israel. The turbine parameters are summarized in Table
II and its power curve in given in Table I. Coefficients of 0.05
TABLE II. NEG MICON 1000/60 FIXED SPEED WIND TURBINE DATA
30
wind speed [m/s]
February
20
10
TABLE III. TH
6 ORDER FITTING COEFFICIENTS OF NEG MICON 1000/60 0
POWER CURVE 1 5 10 15 20 25 29
time [days]
0.1 Histogram
probability
Weibull fit
0.05
0
0 5 10 15 20 25
The 2008 monthly wind raw data in Ariel (10 min resolution) wind speed [m/s]
at 70m height is shown in Fig. 4 along with the appropriate
histograms and fitted Weibull PDFs. The 70m height wind
wind speed [m/s]
speed data was extrapolated from the 10m height data, 20 March
provided by the Israeli Meteorological Service, using the
following relation [4], 10
0.3
v70 70
0
(13) 1 5 10 15 20 25 30
v10 10 time [days]
calculated using (10) and (12), accordingly, are plotted in Fig. Weibull fit
5 along with the percentage difference between the results. 0.05
According to Table IV and Fig. 5, the results are very alike; the
0
worst case error is 2.5% in October, reducing to less than 0 5 10 15 20
wind speed [m/s]
0.25% in January. Two interesting point arise from the case
study. First, the quasi-exact capacity factor is higher than the Fig. 4. Monthly 2008 wind raw data in Ariel and appropriate PDFs
approximate throughout the year. Second, the higher the
capacity factor is, the lower is the percentage difference
between the results, i.e. the October capacity factor is the year
lowest while the January capacity factor is the year highest.
30 20
wind speed [m/s]
0 0
1 5 10 15 20 25 30 1 5 10 15 20 25 30
time [days] time [days]
0.1 0.15
Histogram Histogram
probability
probability
Weibull fit 0.1 Weibull fit
0.05
0.05
0 0
0 5 10 15 20 0 5 10 15
wind speed [m/s] wind speed [m/s]
20 30
May September
20
10
10
0 0
1 5 10 15 20 25 30 1 5 10 15 20 25 30
time [days] time [days]
probability
Histogram Histogram
probability
0.1 0.1
Weibull fit Weibull fit
0.05 0.05
0 0
0 2 4 6 8 10 12 14 16 0 5 10 15
wind speed [m/s] wind speed [m/s]
30
wind speed [m/s]
June 20
wind speed [m/s]
20 October
10 10
0
1 5 10 15 20 25 30
time [days] 0
1 5 10 15 20 25 30
0.15 time [days]
0.15
probability
0.1 Histogram
probability
15 July 30
wind speed [m/s]
10 November
20
5
10
0
1 5 10 15 20 25 30
time [days] 0
1 5 10 15 20 25 30
time [days]
0.15 Histogram
0.15
probability
Weibull fit
probability
0.1 Histogram
0.1
Weibull fit
0.05
0.05
0
2 4 6 8 10 12 14
wind speed [m/s] 0
0 2 4 6 8 10 12 14 16
wind speed [m/s]
Fig. 4. Monthly 2008 wind raw data in Ariel and appropriate PDFs (cont'd) Fig. 4. Monthly 2008 wind raw data in Ariel and appropriate PDFs (cont'd)
30
VI. CONCLUSION
wind speed [m/s]
Histogram
Weibull fit data), results in an analytical solution and hence provides a
0.05 deeper understanding of the factors, affecting the Capacity
Factor and several other merits of wind turbine performance,
0
0 5 10 15 20 25 based on the analytically derived equations. The note has
wind speed [m/s] revealed the close similarity between the results obtained by
Fig. 4. Monthly 2008 wind raw data in Ariel and appropriate PDFs (cont'd)
employing both approaches, enforcing the validity of the
analytically derived approximations.
REFERENCES
0.8
[1] K. Abed and A. El-Mallah, "Capacity factor of wind turbines,"
Quasi-Exact
Energy, vol. 22(5), pp. 487 – 491, 1997.
Capacity Factor
0.6
Approximate
0.4 [2] M. Albadi and E. El-Saadany, "New method for estimating CF
0.2
of pitch-regulated wind turbines," Electr. Power. Syst. Res., vol.
80, pp. 1182 – 1188, 2010.
0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec [3] Y. Ditkovich, A. Kuperman, A. Yahalom and M. Byalsky, "A
generalized approach to estimating capacity factor of fixed
2.5 speed wind turbines," IEEE Trans. Sust. Energy, vol. 3(3), pp.
2 607 – 608, 2012.
% Difference