Вы находитесь на странице: 1из 9

(This is a sample cover image for this issue. The actual cover is not yet available at this time.

This article appeared in a journal published by Elsevier. The attached


copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elseviers archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright

Author's personal copy


J. Wind Eng. Ind. Aerodyn. 109 (2012) 18

Contents lists available at SciVerse ScienceDirect

Journal of Wind Engineering


and Industrial Aerodynamics
journal homepage: www.elsevier.com/locate/jweia

Modeling of atmospheric wind speed sequence using a lognormal


continuous stochastic equation
R. Calif a,n, F.G. Schmitt b
a
b

-P, Guadeloupe
EA 4098, LARGE laboratoire en Geosciences et energetique, Universite des Antilles et de la Guyane, 97170 P-a
CNRS, UMR 8187 LOG Laboratoire dOceanologie et de Geosciences, Universite de Lille 1, 28 avenue Foch, 62930 Wimereux, France

a r t i c l e i n f o

abstract

Article history:
Received 14 October 2011
Received in revised form
10 May 2012
Accepted 1 June 2012

In this paper, we have presented a spectral and a multifractal analysis performed on 412 time series of
wind speed data each of duration of 350 s and sampled at 20 Hz. The average spectrum for the wind
speed displays a scaling behavior, in the inertial range, over two decades, with b 1:68 close to the
Kolmogorov value  5/3. A multifractal analysis has been motivated by the presence of scaling
invariance in data set. Then we have considered their scaling properties in the framework of fully
developed turbulence and multifractal cascades. The results obtained for wind speed conrm that the
exponent scaling function zV q is nonlinear and concave. This exponent characterizes the scaling
functions in the inertial range indicating that the wind speed is intermittent and multifractal. Moreover
the theoretical quadratic relation for lognormal multifractals is well tted. We investigate the
consequence for wind energy production: we generate stochastic simulations of a multifractal random
walk, and using a power curve derived from experimental data, we generate the associated power time
series. We show that, due to the saturation of the power curve for large speed values, when the input
time series (turbulent wind speed) is multifractal, the output can be almost monofractal.
& 2012 Elsevier Ltd. All rights reserved.

Keywords:
Turbulence
Wind speed uctuations
Intermittency
Multifractal analysis
Fractional Brownian motion
Lognormal model
Multifractal random walk

1. Introduction
Wind energy is a ubiquitous resource and a promising alternative to meet the increasing demand for energy. However, this
energy production is very sensitive to uctuations in wind speed,
and these uctuations are stronger for larger values of the wind
speed, due to turbulence. Rapid variations of wind speed due to
turbulence can lead (i) to electrical power variations in the order
of the nominal power output, (ii) to mechanical stress on the gear
boxes and result in substantial voltage swings at the terminals
(Fariley, 2003; Peinke et al., 2004). In this context, a precise
understanding and modeling of small scale turbulent wind eld is
important.
In the atmospheric boundary layer, the Reynolds Number
(ratio of inertial to viscous force) can be as large as 108 (Burton
et al., 2001). Large values of the Reynolds number lead to a huge
intermittency of wind speed uctuations at all temporal or spatial
scales ranging from large-scale variations (years) to very smallscale variations (few minutes down to seconds). Small scale
intermittency remains a challenging problem for the turbulence
community research (Frisch, 1995). Several approaches can be
used to consider the scaling intermittency of small scale

Corresponding author.
E-mail address: rcalif@univ-ag.fr (R. Calif).

0167-6105/$ - see front matter & 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.jweia.2012.06.002

turbulence, the most classical one being structure functions


analysis. In recent years several works have been dedicated to
the analysis of scaling laws and intermittency at small scales for
speed in the laboratory (She and Leveque, 1994; Anselmet et al.,
1984) and in atmospheric boundary layer (Schmitt et al., 1993;

Katul et al., 1995; Schmitt, 2007; Bottcher


et al., 2007). These
studies have shown that atmospheric turbulent speed at small
scales has multifractal scaling uctuations and exhibits long
range power correlations.
In this paper, we characterize the scaling properties of high
frequency wind speed data sampled at 20 Hz in order to quantify
and model the intermittency and statistical properties of their
uctuations using a multifractal analysis technique considering
the moment functions estimates. The paper is organized as
follows. In Section 2 the theoretical framework, traditional
spectral analysis and structure functions analysis are presented.
In Section 3 the datasets are described. In Section 4 the results of
the analysis are presented. In Section 5 the previous results are
checked and conrmed using stochastic simulations.

2. Theoretical framework: turbulence and scaling


We recall here in a few lines the basic scaling properties of
fully developed turbulence, rst concerning Kolmogorovs theory

Author's personal copy


2

R. Calif, F.G. Schmitt / J. Wind Eng. Ind. Aerodyn. 109 (2012) 18

(Kolmogorov, 1941), and then discussing scaling intermittency


descriptions and multifractal models.
2.1. Scaling properties following Kolmogorov (1941)
The scaling and intermittent properties of wind speed have
been studied in the turbulence community for several decades
(see reviews in Frisch, 1995; Tennekes and Lumley, 1972; Monin
and Yaglom, 2007; Vulpiani and Livi, 2004; Schertzer et al., 1997).
We recall here some basic properties. Following the ideas of
Richardson and Kolmogorov for fully developed (and statistically
homogeneous) turbulence, the RichardsonKolmogorov energy
cascade develops on the inertial range between the large injection
scale (of the order of several tens or hundreds of meters) and the
dissipative scales (of the order of millimeters). In the inertial
range, wind speed uctuations are scaling: they possess a powerlaw spectrum
b

Ek C k

with E the power spectral density, b 5=3 for K41 turbulence


(Kolmogorov, 1941; Obukhov, 1941) (k is the wave number). This
relation is written here for wave numbers, but can be converted
to frequencies using Taylors hypothesis (Tennekes and Lumley,
1972) to relate spatial uctuations with temporal uctuations.
Let us note that in the wind energy community, some ts of
speed power spectra are often used, based on von Karmans
2
formula of the form Ef a=b cf 5=6 (von Karman, 1948) or
on Kaimals formula Ef A=1 Bf 5=3 (Kaimal et al., 1972),
where a, b, c and A, B are constants. Von Karman wanted to
perform an interpolation between a low-frequency f4 spectrum
5=3
and a high frequency Kolmogorov spectrum f
and Kaimal
et al. proposed their formula on purely empirical grounds for the
low frequency part. In both cases the main idea is to capture in a
single expression the injections scales and the inertial range with
Kolmogorov spectrum. Indeed, these ts are based on Kolmogorov
theory (Simiu and Scanlan, 1978) which denes a general form of
the uctuating wind velocity shown as follows (Zhangb et al.,
2008):
fEz,f
Ang

v2n
1 Bna b

where f is the frequency, Ez,f is the spectrum, vn is the shear


wind speed, n fz=vz is the Monin coordinate and vz is the
mean wind speed. A, B, a, b and g are coefcients dependent on
the measured data and assumptions used in different models.
Furthermore, the uncertainties and randomness are inherent in
these power spectrum density functions (Solari and Piccardo,
2001). In the present paper, we mainly focus on the inertial range
frequencies and therefore do not need to consider the ts
proposed by these authors.
Kolmogorovs scaling expression can also be written for spatial
uctuations using the absolute increment DV 9Vx Vx9
with V the wind speed, x dening the position, l a spatial distance.
Originally Kolmogorov proposed to consider the scaling
properties of the second moment of these structure functions
(Kolmogorov, 1941):
/DV 2 S C

with the angle braces /  S indicating the ensemble average and


g 2=3 obtained from dimensional analysis.
2.2. Intermittency and multifractal properties
Intermittency in turbulence has been a subject of research for
almost 50 years now, following Kolmogorovs (1962) seminal
work. Here we may dene intermittency as the property of having

large uctuations at all scales, with a correlated structure: large


uctuations are much more frequent than what would be
obtained for Gaussian processes (Frisch, 1995; Vulpiani and Livi,
2004; Schertzer et al., 1997). This is typically studied considering
the probability density function (pdf) or the moments of the
speed uctuations, or more often, using the moment of order q of
these uctuations, called structure functions of order q
Sq t /DV t q S C tzq

where we have written the uctuations in time (t is a time scale),


since we deal with time series analysis in this paper. zq is the
scale invariant moment function, which is nonlinear and concave;
z3 1 is a xed point; z2 b1 relates the second order
moment to the power spectrum scaling exponent. Knowledge of
the full q, zq curve for integer and non-integer moments
provides a full characterization of speed uctuations at all scales
and all intensities. The parameter H z1 is the Hurst exponent.
Monofractal processes correspond to a linear function zq qH,
where Brownian motion is described by H 1=2, fractional
Brownian motion is described by 0 o H o1 (H a 1=2) (that can
be dened as a fractional integration of order b (0 o b o1) of a
Gaussian noise, and H 32b) and homogeneous turbulence is
described by H 1=3.
The values of the function zq are estimated from the slope of
Sq t versus t in a loglog diagram for all moments q. The function
zq denes the types of scaling behavior; in other words, this
exponent function is useful to characterize the statistics of the
random process. If zq is linear, the statistical behavior is
monoscaling and if zq is nonlinear and concave, the behavior
is dened as multiscaling, corresponding to a multifractal process.
The concavity of this function is a characteristic of the intermittency,
the more concave the curve is, the more intermittent the process
(Frisch, 1995; Vulpiani and Livi, 2004; Schertzer et al., 1997).

3. Experimental dataset
In this study, we consider time series of speed V measured on
the wind energy production site of Petit-Canal in Guadeloupe, an
island in the West Indies, located at 16115N latitude and 60130W
longitude. This 10 MW production site is positioned at approximately 60 m (197 ft) above sea level, at the top of a sea cliff. The
wind speed is measured with an ultrasonic anemometer (model
CSAT3) mounted on a tall mast erected 20 m (66 ft) from the cliff
edge, at 38 m (125 ft) from the ground. We analysed a database
recorded at f s 20 Hz during 40 h in July 2005. We separated this
sequence into 412 series of length 350 s each, and perform
ensemble average on the result, in order to characterize and
model the properties of the atmospheric wind speed in the
inertial zone, i.e. for the small time scales. An example of atmospheric wind speed sequence is displayed in Fig. 1. It shows a
strong intermittency, with uctuations at all scales.
Let us characterize the turbulent ow using the speed time
series data and relations which are well-known in the inertial
range of fully developed turbulence (Tennekes and Lumley, 1972).
First, the dissipation E can be estimated by the following exact
relation (Tennekes and Lumley, 1972; Frisch, 1995)
/DV t 3 S 45EV 0 t

which is written here in the time domain, and considering that


the spatial distances writes V 0 t, where V0 8.4 m/s is the
mean speed of the data collected during 40 h. The above relation
is used to obtain the dissipation: E 0:0027 m2 =s3 , which is quite
a large value. Then, the uctuating speed or the standard deviation sv is estimated and can be used to determine the kinetic
energy k 3s2v =2. We nd sv 1:95 m=s; this corresponds to a

Author's personal copy

14

105

12

104

10

103
E(f) ((m/s)2/Hz)

wind speed (m/s)

R. Calif, F.G. Schmitt / J. Wind Eng. Ind. Aerodyn. 109 (2012) 18

8
6

Ev(f)
slope 1.68
slope 5/3

102
101
100

4
101
2

102

0
0

6
8
time (hour)

10

12

14

Fig. 1. An example of wind speed data sampled at 20 Hz and recorded continuously during approximately 14 h, illustrating its intermittency.

large turbulence intensity of sv =V 0 22%. With these values we


can estimate the injection scale L s3v =E (Tennekes and Lumley,
1972), giving L2746 m. The Reynolds number can then be
estimated as Re V 0 L=n 2:57  1010 , where we have taken the
kinematic viscosity n 0:896  106 m2 =s3 at 25 1C. This very
large value of the Reynolds number conrms the fully developed
turbulence framework. We can also estimate the Kolmogorov
scale Z n3 =E1=4 0:13 mm; this corresponds to the smallest
scale of turbulence. The inertial range is from the injection scale
to the Kolmogorov scale, or considering time uctuations (and
using Taylors hypothesis to relate time to space using the mean
speed), from 326 s to 1.3 ms.
p
The Taylor microscale l is estimated as l 15n=Es 0:13 m.
Finally, this allows us to estimate the Taylor-based Reynolds
number Rl sv l=n 3  105 , which is much larger than the
largest values of Rl obtained in the laboratory (Rl 104 ).

103
103

102

101
f (Hz)

100

101

Fig. 2. The mean Fourier power spectrum EV f (solid line) of the 412 samples of
wind speed data in loglog plot, showing a power-laws with a slope of  1.68
(line) compared to slope 5=3 (dashed line).

demonstrates a scaling behavior for frequencies from about


f0.1 to f 10 Hz, with b 1:68 close to the Kolmogorov value
5=3 (Kolmogorov, 1941; Obukhov, 1941): this is consistent with
the values obtained for the inertial range in the previous studies
(Schmitt et al., 1993; Katul et al., 1995; Schmitt, 2007; Katul and
Chu, 1998; Lauren et al., 1999; Morales et al., 2012). The slight
difference with the exact 5=3 value is usually caused by intermittency effects (Frisch, 1995; Schertzer et al., 1997).
However, the power spectrum E(f) is second-order statistic
(proportional to the square of the amplitude of a given frequency
uctuation) and its slope is not sufcient to fully specify a scaling
process. Multifractal analysis is a natural generalization to study
the scaling behavior of a nonlinear phenomenon, using qth order
structure functions: this is discussed below.

4. Multifractal analysis of atmospheric wind speed


We rst conrm here the intermittent character of wind
atmospheric uctuations at small scales through multifractal
analysis. For that, the time period considered is two days sampled
at 20 Hz for wind atmospheric measurements.
4.1. Spectral analysis
In order to estimate the power spectral densities of the wind
speed, the discrete Fourier transform of the times series considered are computed. The expression of the power spectral density
for a process x(t) is recalled here. A N point-long equal interval
time sample of process is used to construct the value at frequency
domain point f, Xf (Bracewell, 1999)
Xf

N
1
X

xj e2pEf =N

4.2. The structure functions scaling exponent zV q for wind speed


data
Here, we analyzed 412 time series of atmospheric wind speed
V t t each of duration 350 s and sampled at 20 Hz (ensemble
average is performed over 412 realizations). The structure functions
analysis is realized for speed increments DV t t 9Vt tVt9 for
t belongs to the inertial range between 0.05 s and 1 s and for all
moments between q0.15 and q5 with increment 0.25. Fig. 3
gives the scaling of the structure function for a sequence V t t for
q1, 1.5, 2 and 2.5, in loglog plot. The straight lines in this gure
indicate that the scaling of the relationship (3) is well respected. We
then estimate zn q for each time sequence n. The scale invariant
function zV q is the ensemble average obtained from all these
exponents zn q

j0

zV q

with f 0; 1, . . . ,N1. The power spectral density is


2

Ef 9Xf 9

Fig. 2 shows the average spectrum EV(f) for the wind speed,
obtained as an ensemble average estimated over 412 sections
compared to 5=3 slope. The wind speed spectrum, EV(f)

412
1 X
z q
412 n 1 n

Fig. 4 presents the empirical structure function scaling exponent zV q compared to the linear model K41 (zq q=3) and the
lognormal model. The lognormal multifractal model is a classical
model rst proposed by Kolmogorov (1962) and still advocated
in recent years (Arneodo et al., 1996); it corresponds to the

Author's personal copy


4

R. Calif, F.G. Schmitt / J. Wind Eng. Ind. Aerodyn. 109 (2012) 18

100

<|v|q>

Following these results, the goal is to reproduce the scaling


law for the moments and the power-law correlations of atmospheric wind speed. Thus, a lognormal continuous stochastic
equation is considered in Section 5.

q=1
q=1.5
q=2
q=3
Fit q=1
Fit q=1.5
Fit q=2
Fit q=3

101

5. Stochastic modeling of wind speed for wind power


uctuations from one turbine

102

103
0.05

0.5

(s)
Fig. 3. The scaling of the structure functions for wind speed SV(q), for q 1, 1.5,
2 and 2.5.

5.1. Stochastic equation for monofractal processes: fractional


Brownian motion

2
v

1.8

Introduced by Kolmogorov (1940), Yaglom (1957) and fully


described in other studies by Mandelbrot and Van Ness (1968),
fBm is a fractional integration of a Gaussian white noise process. A
fBm time series X(t) is a Gaussian time series whose autocorrelation function is given by

q/3
Lognormal

1.6
1.4
1.2
v(q)

The uctuating character of the output power from a Wind


Turbine Generator (WTG) is a consequence of the intermittency of
the atmospheric wind speed. The passage from the wind eld to
the output power is usually modeled using a power curve. In
order to test the smoothing effect of a wind turbine, we model the
wind turbine behavior using a power curve. For that, we perform
in this section stochastic simulations of wind speed time series,
and analyze the scaling properties of the output power time series
using a virtual wind turbine using a specic power curve. In this
section, we model the turbulent wind speed using two scaling
stochastic models: a fractional Brownian motion (fBm) and a
multifractal non-stationary sequence.

Rt,s EXsXt

s2
2

2H

t 2H s2H 9ts9

10

where s is the standard deviation of X(t) and the Hurst parameter


H A 0; 1. The fBm is a monofractal process which has stationary
increments and has scaling structure function following zq qH.

0.8
0.6
0.4

5.2. Multifractal random walk simulation

0.2
0
0

q
Fig. 4. Empirical values of the structure function scaling exponent zv q for wind
speed (&) compared to the linear model K41 (dashed line) and the lognormal
model (solid line).

non-linear multifractal curve

zq

q
m

3qq2
3 18

where the parameter m is called intermittency parameter; the


value m 0:25 70:05 has been proposed (Arneodo et al., 1996).
The function zV q obtained from our database is nonlinear and
concave similar to the results obtained in previous studies
(Schmitt et al., 1993; Schertzer et al., 1997) has the following
properties:

 0 o qo 3, zV q 4 q=3 for small uctuations,


 q3, zV 3 1 in agreement with the Kolmogorov 4=5 law,
 q4 3, zV q o q=3 for large uctuations.
The lognormal model provides a reasonable t for the empirical
exponent function zV q. Other models have been proposed such
as log-Poisson (She and Leveque, 1994) and log-stable (Schertzer
et al., 1997), but we do not test them here.

To generate a non-stationary multifractal process mimicking


the turbulent speed, we generalize multifractal random walk
ideas (Bacry et al., 2001; Muzy and Bacry, 2002), and generate a
non-stationary multifractal time series u(t) as
Z t
ut
E1=2 x dBH x
11
0

where BH(x) is a fractional Brownian motion. Taking lognormal


statistics for E, an intermittent and long-range correlated time
series (Schmitt, 2003), the scaling exponent zq given by
q
/9Dut t9 S  tzq can be written as

zq qH q2 q
2

12

where m is the intermittency parameter (0 r m r1) characterizing


the lognormal multifractal cascade. This was done in Huang et al.
(2008) for H 1=2; we use H o 1=2 here. This multifractal
stochastic process is continuous in scale ratio and in time. It has
a simple generating equation and can be used to sequentially
generate time series of any length.
5.3. Stochastic simulations of atmospheric wind speed data and
wind power through the power curve
Here, we simulate a sequence of length 4096 data points for a
fBm process with H 1=3 using the circulant matrix method
algorithm (Wood and Chan, 1994). For lognormal multifractal
process, we provide a stochastic simulation for a sequence of

Author's personal copy


R. Calif, F.G. Schmitt / J. Wind Eng. Ind. Aerodyn. 109 (2012) 18

length 4096 with m 0:25. The sampling frequency of the simulated data is 20 Hz. Then, the spectra and the exponent structure
functions obtained in each case are compared with the empirical
spectrum E(f) and curve zv q of a measured wind speed signal
presented in Fig. 1. The simulated signals have been rescaled to
have the same mean value, 6 m/s (this value is the most probable
occurring on the site Calif et al., 2005), and the same standard
deviation 0.3 m/s as the measured signal vexp. Fig. 5 illustrates the
measured signal vexp, a multifractal random walk vmrw generated
by the lognormal continuous stochastic equation and a fractional
Brownian motion vfBm generated by the circulant matrix method
algorithm. Fig. 6 shows the normalized densities of measured
signal Dvexp , fBm process DvfBm and multifractal random walk
Dvmrw , compared to the Gaussian pdf G  N0; 1 for the temporal
increments Dv vt tvt with t 0:5 s. We observe that
large uctuations have larger probabilities in the multifractal

random walk. Gaussian statistics are compatible with fBm, contrary to the measured signal and the multifractal random walk.
In Fig. 7, we have plotted the spectra of measured and
simulated speed signals compared to a straight line of slope
b 5=3: the scaling behavior is veried. We checked also the
structure functions scaling exponents of vfBm and vmrw. Fig. 8
shows the scaling exponents of vfBm, vmrw and vexp compared to
the linear model K41 (zq q=3 and lognormal model
(zq q=3 m=183qq2 ). For the signal generated with fractional Brownian motion, the curve zfBm q is linear and tted by q/
3. The curve of zmrw q of the generated signal with lognormal
multifractal random walk is superimposed to empirical curve
zv q. zmrw q is nonlinear and concave: the eld that is generated
with this type of equation is intermittent, scale invariant, and
characterized by multiple-scaling properties. Such properties are
fully compatible with experimental results.

vexp (m/s)

7
6.5
6

vfBm (m/s)

vmrw (m/s)

5.5
0

20

40

60

80

100

120

140

160

180

200

20

40

60

80

100

120

140

160

180

200

20

40

60

80
100
120
time (seconds)

140

160

180

200

7
6.5
6
5.5

7
6.5
6
5.5

Fig. 5. (a) An example of measured signal vexp, (b) a multifractal random walk vmrw generated by the lognormal continuous stochastic equation and (c) a fractional
Brownian motion vfBm generated by the circulant matrix method algorithm.

100

100

101

101

101

102

102

102

density

100

103

0
vexp/

103

6 4 2 0 2 4 6
vmrw/

103

vfBm/

Fig. 6. The increment normalized densities of Dvexp t (left), Dvmrw t (center) and DvfBm t (right), compared to the Gaussian pdf G  N0; 1.

Author's personal copy


6

R. Calif, F.G. Schmitt / J. Wind Eng. Ind. Aerodyn. 109 (2012) 18

102

200

101
150
P (kW)

E(f) ((m/s)2/Hz)

250

Emrw
Eexp
Efbm
slope 5/3

103

100
101

100

102
50
103
104
103

0
102

101
f (Hz)

100

101

Fig. 7. The Fourier power spectrum E(f) of wind speed (Eexp), fBm (Efbm) and
lognormal multifractal random walk (Emrw) compared to slope 5=3.

(q)

1.6
1.4

Lognormal
v

1.2

vmrw

1
0.8
0.6
0.4
0.2
0
0

10

15
20
v (m/s)

25

30

35

Fig. 9. Power curve providing the power curve of a 220 kW wind turbine
generator versus the input wind speed.

In order to analyze the scaling law of the moments for the


output power from one wind turbine, we generate the output
power with the wind speed modeling results for 4o vt r15 m=s
(increasing zone of the power curve) and 15 o vt r26 m=s (stall
region of the WTG).

1.8
q/3
vfBm

q
Fig. 8. The function zv q of wind speed zvexp (&), fBm zvfBm X and lognormal
multifractal random walk zvmrw (n) compared to lognormal model zvlognormal (solid
line) and linear model K41 (dashed line).

The wind speed simulation using two stochastic model (fractional Brownian motion and multifractal random walk) are
translated into wind power using the mean power curve of a
220 kW wind turbine currently installed and operating in the
wind farm described in Section 3: this type of turbine possesses
two blades of length 13 m, with stall-regulated and xed speed.
The goal is to see how wind speed modeling affects wind power
output, a point that is particularly important for the design and
operation of a wind turbine. The turbines have a cut-in speed of
4 m/s, a cut-out speed of 26 m/s and produce their peak power
output at 15 m/s. The power curve of the wind turbine generator
is shown in Fig. 9. To simulate wind power P(t), we t the
following power curve to experimental data collected and processed according to the IEC standard 61400-12 (Burton et al.,
2001):
8
2
3
>
< 13063vt 9:1v t0:3v t
Pt 140 11vt0:36v2 t
>
:0

for 4 ovt o 15 m=s


for 15 r vt o 26 m=s
for vt Z 26 m=s

5.3.1. Modeling for wind power in the increasing zone of the power
curve
We rst consider here the increasing zone of the power curve,
for which 4 o vt r15 m=s. We study three simulated power
series: (i) P simul is generated with the experimental wind speed
data vexp; (ii) Pfbm is simulated with the wind speed modeling
using a fractional Brownian motion data vfbm; and (iii) Pmrw is
generated with the wind speed modeling using a lognormal
multifractal random walk vmrw. We simulated 20 segments of
4096 points sampled at 1 Hz for all three simulated power series.
Fig. 10a illustrates the Fourier power spectra EPsimul, EPfBm, EPmrw
for all three simulated power series. The power spectra of the
simulated power series follow a power-law spectrum for
0:001 rf r 0:5 with a slope close to 5=3.
To determine the nature of scaling for Psimul, PfBm and Pmrw, we
estimated the structure functions of their time increments and
the exponent structure functions zPsimul , zPfbm , zPmrw plotted in
Fig. 10b. For all three simulated power series, the scaling behavior
of the output power is similar to that of the wind speed modeling
results obtained in Section 5.3. Indeed, Pfbmhas the same monoscaling behavior as vfBm, whereas Psimul and Pmrw have the same
multiscaling behavior as respectively vexp and vmrw. These results
show that the spectral and statistical properties of the output
wind power are similar to those of the wind speed. In other
words, although the power curve is a cubic function, the intermittency of wind speed is unchanged by the power curve in this
region contrary to modeling of the output wind power using
Langevin equation presented in Gottschall and Peinke (2007) in
which the intermittency of the speed increment distribution
seems to be amplied.
5.3.2. Modeling for wind power in the stall region of the power curve
In order to analyze the intermittency of the output power in
the stall region of the power curve, we simulated 20 segments
of wind speed having a mean value around 16 m/s, using a
lognormal multifractal random walk. The spectral and

Author's personal copy


R. Calif, F.G. Schmitt / J. Wind Eng. Ind. Aerodyn. 109 (2012) 18

105

1.8

104

EPmrw

103

slope 5/3
EPsimul

Psimul

1.6

102
(q)

E(f) (kW2/Hz)

EPfBm

101

1.4

Lognormal
q/3
PfBm

1.2

Pmrw

1
0.8
0.6

100

0.4
101
102
104

0.2
0
103

102
f (Hz)

101

100

105

1.8
EPS
slope 1.64
EPL
slope 1.63

PS

1.6

PL
q/3

1.4

p(q)

E(f) (kW2/Hz)

1.2

100

1
0.8
0.6
0.4
0.2

105
104

0
103

102
f (Hz)

101

100

2
q

Fig. 10. (a) The Fourier power spectrum E(f) for P simul (EPsimul), Pfbm (EPfbm) and Pmrw (EPmrw), compared to slope 5=3; (b) the function zp q for simulated wind power from
speed data zPsimul (&), fBm zPfbm X and lognormal multifractal random walk zPmrw (solid line) compared to t q/3 (dotted-line); (c) the Fourier power spectrum of
simulated wind power from speed generated using lognormal multifractal random walk, for the linear zone EPL and EPS for the stall region of the power curve (saturated
zone of power curve); (d) the function zp q for 10 simulated samples of wind power from speed synthesized using lognormal multifractal random walk, for the linear zone
zPL and stall zone zPS .

statistical properties of Ps are compared with those obtained for


Pmrw in the increasing zone of the power curve. Fig. 10c
illustrates the Fourier spectrum Eps(f) for Ps and EpL(f) for Pmrw.
The two spectra have a 5=3 power spectral slope for
0:001r f r 0:5. Moreover, as shown in Fig. 10c, the variance of
Pmrw is larger than that of Ps. We estimate the structure function
for the time increment of Ps for determine the exponent
function zps q. In Fig. 10d, we have plotted zps q (&) and
zpmrw q (1) compared to linear t q/3 (solid line). zpmrw q is
nonlinear and concave whereas zps q tends to be linear. Indeed,
the gure shows a slight difference between zps q and the t q/
3. In the stall region, the large uctuations of the wind speed
are reduced for the wind power uctuations.
In conclusion the behavior of the output uctuations seems
multifractal for the increasing range and monofractal for the
stalled range. However the lack of real data for a single turbine
does not enable us to conclude in a nal way.

6. Conclusions
We considered here wind speed time series sampled at 20 Hz.
Their scaling statistics have been studied in the framework of
fully developed turbulence and Kolmogorovs theory. Their Fourier power spectra have been estimated and we considered their
scaling properties in the physical space. We showed that the wind
speed is intermittent and multifractal as previous works. In order
to check this results, we generated turbulent wind speed time
series using two stochastic models, fractional Brownian motion
(fBm) and multifractal random walk. The analyses have shown
that the multifractal random walk is intermittent, and possesses
scaling properties very similar to measured data, unlike fBm.
Using these stochastic simulations and a measured signal, we
provide a simulation for wind power data: the wind speed
simulation results and the wind speed measured are translated
into wind power using the power curve of one of the 220 kW

Author's personal copy


8

R. Calif, F.G. Schmitt / J. Wind Eng. Ind. Aerodyn. 109 (2012) 18

wind turbines we collected data from. The simulated power


presents two scaling behaviors: (i) in the increasing region of
the power curve, P is a multifractal random walk and exhibits a
long-range power correlations according to Gottschall and Peinke
(2007), and (ii) in the stall region of the power curve, P is close to
a monofractal random walk. In other words, the wind power is
intermittent and multifractal in the increasing zone, whereas the
wind power is less intermittent and close to monofractal process
in the stall zone, meaning that the large uctuations of the wind
speed are smoothed in the stall region. Although no data for an
individual turbine are available, we attempt to give different
possible explanations concerning the smoothing effect. Besides,
we could only obtain the data corresponding to the aggregate
power of an entire wind farm. The dataset has been analyzed in
the same way (spectral analysis and multifractal analysis). The
analysis results published in Calif and Schmitt (2012) have shown
that the aggregate output power possesses intermittent and
multifractal properties.
This work thus predicts that output power time series possess
intermittent and scaling statistics in relation with the multifractal
properties of the input wind speed series. Furthermore, we
numerically found that the saturation behavior of the nonlinear
transfer curve has an inuence on the output power series: for an
absolute wind speed not too large, the power series will also be
multifractal whereas the effect of the saturation has the consequence of producing non-intermittent in the power series.
Future works involve simultaneous analysis of turbulence time
series and instantaneous power data, in order to consider the
multiscale correlation between both series. With the increasing
importance of wind energy, the understanding of the non-stationary and intermittent nature of power series is an important
issue. The present study shows that this can be related to
turbulent wind speed properties in the framework of locally
homogeneous Kolmogorov turbulence.

Acknowledgment
This study is nancially supported by the Regional Council of
Guadeloupe and the European funding (No 1/1.4/-31614).
References
Anselmet, F., Gagne, Y., Hopnger, E.J., Antonia, R.A., 1984. High order velocity
structure functions in turbulent shear ows. Journal of Fluid Mechanics 140,
6389.
Arneodo, A., Baudet, C., et al., 1996. Structure functions in turbulence, in various
ow congurations, at Reynolds number between 30 and 5000, using
extended self-similarity. Europhysics Letters 34, 411416.
Bacry, E., Delour, J., Muzy, J.F., 2001. Multifractal random walks. Physical Review E
64, 026103.

Bottcher,
F., Barth, St., Peinke, J., 2007. Small and large uctuations in atmospheric
wind speeds. Stochastic Environmental Research and Risk Assessment 21,
299308.
Bracewell, R., 1999. The Fourier Transform and Its Applications, third ed. McGrawHill Science.
Burton, T., Sharpe, D., Jenkins, N., Bossanyi, E., 2001. Wind Energy Handbook. John
Wiley & Sons, Chichester. (p. 12).
Calif, R., Schmitt, F.G., 2012. Analyse des series temporelles de production
eolienne: loi de Taylor et proprietes multifractales, in Rencontre du non
lineaire 2012 ed. In: Josserand, C., Lefranc, M., Letellier, C. (Eds.), Institut Henri
Poincare, Paris, Non-lineaire Publications, 2012. ISBN 978-2-9538596-1-4,
pp. 6166.

Calif, R., Blonbou, R., Deshaies, B., January 2005. Wind speed analysis for time
scales smaller than 1 hour: application to wind energy forecasting. In: 24th
AIAA-ASME Wind Energy Symposium, Reno, 1332.
Fariley, P., 2003. Steady as she blows. IEEE Spectrum 12, 3539.
Frisch, U., 1995. Turbulence the Legacy of A.N. Kolmogorov. Cambridge University
Press, Cambridge.
Gottschall, J., Peinke, J., 2007. Stochastic modelling of a wind turbines power
output with special respect to turbulent dynamics. Journal of Physics:
Conference Series 75, 012045.
Huang, Y., Schmitt, F.G., Lu, Z., Liu, Y., 2008. An amplitudefrequency study of
turbulent scaling intermittency using Hilbert spectral analysis. Europhysics
Letters 84, 40010.
Kaimal, J.C., Wyngaard, J.C., Izumi, Y., Cote, O.R., 1972. Spectral characteristics of
surface-layer turbulence. Quarterly Journal of the Royal Meteorological Society
98, 563589.
Katul, G., Chu, C.R., 1998. A Theoretical and experimental investigation of energycontaining scales in the dynamic sublayer of boundary-layer ows. BoundaryLayer Meteorology 86, 279312.
Katul, G.G., Chu, C.R., Parlange, M.B., Albertson, J.D., Ortenburger, T.A., 1995. Lowwavenumber spectral characteristics of speed and temperature in the
atmospheric surface layer. Journal of Geophysical Research 100 (D7),
14,24314,255.
Kolmogorov, A.N., 1940. The Wiener spiral and some other interesting curves in
3218 Hilbert space. Doklady Akademii Nauk SSSR 26, 115118.
Kolmogorov, A.N., 1941. The local structure of turbulence in incompressible uid
for very large Reynolds numbers. Doklady Akademii Nauk SSSR 30, 299303.
Kolmogorov, A.N., 1962. A renement of previous hypotheses concerning the local
structure of turbulence in a viscous incompressible uid at high Reynolds
number. Journal of Fluid Mechanics 13, 8285.
Lauren, M.K., Menabe, M., Seed, A.W., Austin, G.L., 1999. Characterisation and
simulation of the multiscaling properties of the energy-containing scales of
horizontal surface layer winds. Boundary-Layer Meteorology 90, 2146.
Mandelbrot, B.B., Van Ness, J.W., 1968. Fractional Brownian motions, fractional
3253 noises and applications. SIAM Review 10, 422.
Monin, A.S., Yaglom, A.M., 2007. Statistical Fluid Mechanics, vol. II. MIT Press,
reprint Dover, p. 92.

Morales, A., Wachter,


S., Peinke, J., 2012. Characterization of wind turbulence by
higher-order statistics. Wind Energy 15 (3), 391406.
Muzy, J., Bacry, E., 2002. Multifractal stationary random measures and multifractal
random walks with log innitely divisible scaling laws. Physical Review E 66
(5), 056121.
Obukhov, A.M., 1941. Spectral energy distribution in a turbulent ow. Doklady
Akademii Nauk SSSR 32, 2224.
Peinke, J., Barth, S., Bottcher, F., Heinemann, D., Lange, B., 2004. Turbulence, a
challenging problem for wind energy. Physica A 338, 187193.
Schertzer, D., Lovejoy, S., Schmitt, F., Chigirinskaya, Y., Marsan, D., 1997. Multifractal cascade dynamics and turbulent intermittency. Fractals 5 (3), 427471.
Schmitt, F.G., 2003. A causal multifractal stochastic equation and its statistical
properties. European Physics Journal B 34, 8598.
Schmitt, F.G., 2007. Gusts in intermittent wind turbulence and the dynamics of
their return times. In: Peinke, J., Schaumann, P., Barth, S. (Eds.), Wind Energy,
Proceedings of the Euromech Colloquium. Springer, pp. 7379.
Schmitt, F., Schertzer, D., Lovejoy, S., Brunet, Y., 1993. Estimation of universal
multifractal indices for atmospheric turbulent speed elds. Fractals 1,
568575.
She, Z.S., Leveque, E., 1994. Universal scaling laws in fully developed turbulence.
Physical Review Letters 72, 336339.
Simiu, E., Scanlan, R.H., 1978. Wind Effects on Structures: An Introduction to Wind
Engineering. Wiley, New York.
Solari, G., Piccardo, G., 2001. Probabilistic 3-D turbulence modeling for gust
buffeting of structures. Probabilistic Engineering Mechanics 16 (1), 7386.
Tennekes, H., Lumley, J.L., 1972. A Fit Course in Turbulence. The MIT Press.
von Karman, T., 1948. Progress in the statistical theory of turbulence. Proceedings
of the National Academy of Sciences of the United States of America 34,
530539.
Vulpiani, A., Livi, R. (Eds.), 2004. The Kolmogorov Legacy in Physics. Springer,
Berlin, Heidelberg, New York.
Wood, A., Chan, G., 1994. Simulation of stationary Gaussian processes. Journal of
Computational and Graphical Statistics 3, 409432.
Yaglom, A.M., 1957. Some classes of random elds in n-dimensional space, related
3441 to stationary random processes. Theory of Probability and Its Applications 2, 273. (3442).
Zhangb, L., Lia, J., Pengb, Y., 2008. Dynamic response and reliability analysis of tall
buildings subject to wind loading. Journal of Wind Engineering and Industrial
Aerodynamics 96, 2540.

Вам также может понравиться