You are on page 1of 5

Assignment 1: Mean wind speed and the Weibull distribution

David Schillebeeckx s143138


December 10, 2014

The invalid wind speed data, where u = 99.99 m/s, are first removed from all the columns.
1. The mean wind speed and standard deviation are calculated by using the mean- and std-functions
in Matlab, respectively. This gives following result:
U = 8.2356 m/s,
U = 3.9079 m/s.
2. To plot the pdf of the wind speed, a histogram is first created using the hist-function with 500
bins. This sorts all the elements of the wind speed into 500 equally spaced bins. The function
gives back two vectors: an x-vector with the location of the 500 bins and a f-vector with the
number of wind speed elements in each bin.
In order to get the pdf, these values need to be normalized by dividing the f-vector by the area
under the curve, which is determined by integration using the trapz-function.
The pdf of the wind speed, fU (u), is shown in Figure 1.
0.14
0.12

fU (u)

0.1
0.08
0.06
0.04
0.02
0
0

10

15
20
Wind speed, u [m/s]

25

30

35

Figure 1: The pdf of the wind speed measured in Sprog.

3. The cdf of the wind speed can be obtained from its pdf using following equation:
Z x
FU (u) =
fU (u0 )du0 .

This integration is performed in Matlab using the trapz-function. The result is shown in Figure 2.

1
0.9
0.8
0.7

FU (u)

0.6
0.5
0.4
0.3
0.2
0.1
0
0

10

15
20
Wind speed, u [m/s]

25

30

Figure 2: The cdf of the wind speed measured in Sprog.

4. The A and k parameters are estimated in three different ways:


Using the wblfit-function of Matlab. This function uses the wind speed data as input and
returns the two Weibull parameters. However, to use this function the 0-values have to be
changed to very small positive values (e.g. E-9). This method renders following parameters:
A = 9.1838,
k = 2.0855.
Using the calculated mean, U , and standard variation, U , the following equations are
solved w.r.t. A and k:
U = A(1 + 1/k),

2
2
U = A (1 + 2/k) 2 (1 + 1/k) .
This method renders following parameters:
A = 9.2987,
k = 2.2278.
With the pdf, the third non-central moment, 3 , can be calculated using the following
equation:
Z +
3 (u) =
u3 fU (u)du.

Using this third non-central moment and the mean, U = 1 , the following equations are
solved w.r.t. A and k:
U = A(1 + 1/k),
3 = A3 (1 + 3/k),
This method renders following parameters:
A = 9.2984,
k = 2.2381.
2

5. To compare the three different Weibul distributions they are plotted together with the measurement data in Figure 3.
0.14

Measurements
wblfit
U and U
1 and 3

0.12

Probability density

0.1
0.08
0.06
0.04
0.02
0
0

10

15
20
Wind speed [m/s]

25

30

35

Figure 3: The pdf of the measurement data (blue bars) compared with three different Weibull distributions
(solid coloured lines).

A good agreement between the measured data and the fitted Weibull distribution is obtained at
lower and higher wind speeds. However, near the peak of the histogram the probability density
is underestimated by all the Weibull distributions.
To verify which method to determine the A and k Weibull parameters gives the best agreement
with the measurement data the R2 -values are calculated for each method. R2 is calculated as
follows:
P
(yi fi )2
2
R = 1 Pi
,
)2
i (yi y
where yi are the measured values and fi the values obtained from the fitted model. The following
R2 -values are obtained:
Method
wblfit
U and U
X and 3

R2
0.9572
0.9713
0.9717

From these R2 -values and Figure 3, it can be seen that the best A and k estimations are obtained
by fitting the first and third non-central moment (red line).
6. Analysing the wind direction data, one can see that the direction is first measured at a height
of 67.5 m and thereafter at 70 m. From these direction data the wind speeds are divided into
12 directional sectors of 30 wide and centered on 0 , 30 , ..., 330 . For each of these sectors the
pdf of the measured wind speeds and the Weibull pdf parameters (by fitting the first and third
non-central moment) are determined and plotted.
From these different plots one can see that the wind speed is highly directional which results in
a large difference of the Weibull parameters for different directional sectors.

Direction = 0 , A =7.78 , k =1.93

Direction = 30 , A =7.57 , k =1.90

0.14

0.14
Measurements
Weibull pdf

0.12

0.12

0.1

0.1

0.08

0.08

fU (u)

fU (u)

Measurements
Weibull pdf

0.06

0.06

0.04

0.04

0.02

0.02

0
0

10
15
20
Wind speed, u [m/s]

25

0
0

30

Direction = 60 , A =7.60 , k =1.96

10
15
Wind speed, u [m/s]

20

25

Direction = 90 , A =8.41 , k =2.19

0.25

0.16
Measurements
Weibull pdf

Measurements
Weibull pdf

0.14

0.2
0.12
0.1
fU (u)

fU (u)

0.15

0.1

0.08
0.06
0.04

0.05
0.02
0
0

10
15
Wind speed, u [m/s]

20

0
0

25

Direction = 120 , A =9.75 , k =2.42

10
15
Wind speed, u [m/s]

20

25

Direction = 150 , A =8.83 , k =2.21

0.14

0.16
Measurements
Weibull pdf

Measurements
Weibull pdf

0.14

0.12

0.12

0.1

fU (u)

fU (u)

0.1
0.08

0.08

0.06
0.06
0.04

0.04

0.02
0
0

0.02

10
15
20
Wind speed, u [m/s]

25

30

0
0

10
15
Wind speed, u [m/s]

20

25

Direction = 180 , A =8.87 , k =2.07

Direction = 210 , A =9.83 , k =2.38

0.14

0.14
Measurements
Weibull pdf

0.12

0.12

0.1

0.1

0.08

0.08

fU (u)

fU (u)

Measurements
Weibull pdf

0.06

0.06

0.04

0.04

0.02

0.02

0
0

10
15
20
Wind speed, u [m/s]

25

0
0

30

Direction = 240 , A =9.93 , k =2.46

10
15
20
Wind speed, u [m/s]

25

30

Direction = 270 , A =10.07 , k =2.62

0.14

0.16
Measurements
Weibull pdf

Measurements
Weibull pdf

0.14

0.12

0.12

0.1

fU (u)

fU (u)

0.1
0.08

0.08

0.06
0.06
0.04

0.04

0.02

0.02

0
0

10
15
20
Wind speed, u [m/s]

25

0
0

30

10

Direction = 300 , A =10.19 , k =2.35

15
20
Wind speed, u [m/s]

35

0.14
Measurements
Weibull pdf

Measurements
Weibull pdf

0.12

0.12

0.1

0.1

0.08

0.08

fU (u)

fU (u)

30

Direction = 330 , A =8.38 , k =1.97

0.14

0.06

0.06

0.04

0.04

0.02

0.02

0
0

25

10

15
20
Wind speed, u [m/s]

25

30

35

0
0

10
15
Wind speed, u [m/s]

20

25

You might also like