ISSN 2079-6226: Proceedings of the 2012 Mechanical Engineering Conference on Sustainable Research and Innovation, Volume 4, 3rd-4th May

2012

Determination of the Shape (k), Scale (c)

Parameters and the Wind Power Density for a
Selected Site in Juja
C. Saoke, J. N. Kamau, R. Kinyua

 generation available, the technology is continuously being

improved for both cheaper and more productive wind turbines
Abstract— Wind is a variable resource in both time and space; it and with this, it can therefore be expected that wind energy
fluctuates so much that it is important to find a model that describes will become even more economically competitive in the
its variation. This is significant in estimating the amount of energy coming decades [2]. In Kenya, this expansion rate has not yet
earnings as well as to optimize the design of the wind turbine. This been realized despite the fact that Kenya is compatible with
work assesses the wind speed distribution by measuring and current wind technology. The main issue is the limited
analyzing the wind speeds at two heights in Juja. The wind speed knowledge on the Kenya wind resource. The meteorological
data of the site is collected and analyzed on the basis of three months station data are quite inadequate since these stations are only
measured hourly time series data. The Weibull distribution model
35 spread all over the country [7], [8], [9]. There is significant
provided a good fit for the recorded wind data. The model is then
used to calculate the Weibull distribution parameters for the speed potential to use wind energy for grid connected wind farms,
variations as the Weibull scale parameter (c) = 5.69 m/s and the isolated grids (through wind-solar hybrid systems) and off-
Weibull shape parameter (k) = 2.06. These parameters are used to grid community electricity and water pumping.
estimate the wind power density of the site which was found to be The equatorial areas are assumed to have poor to medium
131.35 W/m2. Wind rose diagrams depict a large calm condition with wind resource. This might be expected to be the general
a large frequency of wind blowing from the North , north east while pattern for Kenya. However some topography specifics
frequency are lowest from the south and south east directions. (channeling and hill effects due to the presence of the Rift
Valley and various mountain and highland areas) have
. endowed Kenya with some excellent wind regime areas. The
North West of the country (Marsabit and Turkana districts)
Keywords— Power, Scale parameter, Shape parameter, Weibull, and the edges of the Rift Valley are the two large windiest
Wind speed. areas (average wind speeds above 9m/s at 50 m high). The
coast is also a place of interest though the wind resource is
I. INTRODUCTION expected to be lower about 5-7 m/s at 50 m high [4].

T he continuing rise in population and the shift of nations to

industrialization has lead to the increase in energy
demand. However factors such as energy sustainability
Describing wind speed distribution using the Weibull model
Due to the intermittency of wind resource, It is very important
and the gradually emerging consciousness about in the wind industry to be able to describe the spatial and
environmental degradation have provoked priority to the use temporal variation of wind speeds. This is most useful for
of renewable alternative sources such as solar and wind turbine designers who need the information to optimize the
energies [1-3]. Wind energy which one of the renewable design of their turbines with a view of minimizing generating
sources is key to a clean energy future that first used more costs. On the other hand, turbine investors need the
than 3,500 years ago in boats to transport goods in Egypt and information to estimate their income from electricity
to grind seeds to produce flour. Its advantages such as generation. A number of studies in the recent years have
cleanliness, abundance of resource, low cost, sustainability, investigated the fitting of specific distribution to wind speeds
safety and effectiveness in job creation has facilitated its fast for use in practical applications as estimation of wind loads on
growth as an energy source during most of the 1990s, building and power analysis. The two-parameter Weibull
expanding at an annual rate from 25% to 35% [3]. According distribution is one of the mathematical models used to study
to the wind energy outlook report 2007, wind energy is now wind data. It is a probability distribution function used to
one of the most cost effective methods for electricity describe wind speed variations and its expressed
mathematically by the probability f (v) of observing wind
C. Saoke, Department of Physics, JKUAT. Tel no.: +2540722700886; fax: speed v given as in equation 1: [6].
+2546752711; E-mail: winsaoke@gmail.com
k 1
J. N. Kamau, Department of Physics, JKUAT Email: k v   v k 
kamaungugi@yahoo.com. f v     exp     (1)
R. Kinyua, Department of Physics, JKUAT. Email: kinyua@fsc.jkuat.ac.ke c c   c  

1
41
ISSN 2079-6226: Proceedings of the 2012 Mechanical Engineering Conference on Sustainable Research and Innovation, Volume 4, 3rd-4th May 2012

While the corresponding cumulative distribution is

  v k 
F v   1  exp     (2) II. MATERIALS AND METHODS
  c  
Where; The objectives of this paper is to use measured data to
v ≥ 0 is the wind speed (m/s for this study), k > 0 is a shape determine the Weibull shape parameter k and the scale
parameter c (m/s) and further use them to estimate the
parameter (dimensionless), c > 0 is a scale parameter (m/s)
localized wind power density of the site. The wind data
studied is collected using two Davis Vantage Pro weather
The variance of a data set from the Weibull distribution is stations located within Jomo Kenyatta University of
given by equation 3; Agriculture and Technology (JKUAT) in Juja, located 35
kilometers from the Kenyan capital Nairobi. The experimental
  2  set is as in figure 1;
 1  


  k
2
 V  1 (3)
 2 2  
  1   
  k 

Further simplification of equation 3 by Justus, (1978) and

Lysen, (1983) gives the formulars for k and c as in equation
4 and 5

1.086
 
k    (4)
 
V 

1
 0.433  k

c  V  0.568   (5)
 k 
These parameters are very important in wind energy Fig. 1: The set up of equipments.
prediction, k for instance shows how peaked the distribution
is, meaning it determines the shape of the Weibull distribution Wind speeds are measured for a period of three months during
and hence it determines the wind speed variation. Other the months of September, October and November 2010. This
methods that can be employed to evaluate these parameters is based on an hourly series daily at heights of 13 m and 20 m.
include; the maximum likelihood method, having a good fit The data is averaged so as to determine the monthly average
for the discrete cumulative frequency through the log-log wind speeds.
method. Other methods can be obtained from [9]. The data results are grouped with class width of 1 m/s, the
grouped data is then used to statistically model a Weibull
The power density distribution which is finally used to estimate the wind power
density of Juja.
Using values of the Weibull scale and shape parameters the The wind directions are analysed using the Windographer
wind power density of a given site can be calculated from the software to produce the wind rose diagrams as presented in
wind speed data. The wind power density per unit area figures 5 and 6.
analyzed based on a Weibull probability density function is
given by:
1 3  3 III. RESULTS
Pw  c 1   (6)
2  k Table 1: The distribution of wind speeds for statistical
analysis
Meaning multiplying the power of each wind speed with the
probability of each wind speed obtained from the Weibull
curve, then the results gives the distribution of wind energy at
different wind speeds (the power density). Out of this power Velocity
Classes Velocity (fv) Deviation fD2
value (equation 6), the maximum extractable power is subject
to the Beltz limit given by 59.8% of power density [5], [7], (m/s) mid (D)
[9]. points Freq.

2
42
ISSN 2079-6226: Proceedings of the 2012 Mechanical Engineering Conference on Sustainable Research and Innovation, Volume 4, 3rd-4th May 2012

(v) m/s (f) 25 5.03 1011.6 5.39 847.9 22 30

0.0-0.9 0.45 91 40.95 -4.59 1917.19 27 4.89 1011.7 5.38 848.6 20 25

1.0-1.9 1.45 86 124.7 -3.59 1108.37 28 6.47 1011.8 6.72 849.2 20 340

2.0-2.9 2.45 282 690.9 -2.59 1891.68 29 6.39 1012.1 6.76 849.4 21 310

3.0-3.9 3.45 325 1121.3 -1.59 821.63 30 4.28 1012.2 4.75 849.4 20 315

4.0-4.9 4.45 324 1441.8 -0.59 112.78 Av. 4.42 1009.54 4.72 847.04 19.4

5.0-5.9 5.45 194 1057.3 0.41 32.61 Max. 6.47 1012.2 6.76 849.4 22

6.0-6.9 6.45 195 1257.8 1.41 387.67 Min. 2.61 1006.9 2.87 844.1 18

7.0-7.9 7.45 209 1557.1 2.41 1213.89

8.0-8.9 8.45 152 1284.4 3.41 1767.47

Table 3: Daily averages of wind speed, air pressure, direction
9.0-9.9 9.45 90 850.5 4.41 1750.32 and temperature for October 2010
10.0-10.9 10.45 38 397.1 5.41 1112.18

11.0-11.9 11.45 15 171.75 6.41 616.32

12.0-12.9 12.45 5 62.25 7.41 274.54

Height (20 m) Temp.
13.0-13.9 13.45 3 40.35 8.41 212.18 Date Height (13 m) Direction

14.0-14.9 14.45 2 28.9 9.41 177.09 October Air

15.0-15.9 15.45 1 15.45 10.41 108.36 2010 Air Pressure (oC)

Speed pressure Speed
SUM 2012 10142 13504.4 (m/s) (mb) (m/s) (mb)

1 4.37 1006.3 4.75 848.5 20 120

2 4.42 1006.1 4.86 847.7 20 100

4 4.83 1006.1 5.54 846.7 21 140

Table 2: Daily averages of wind speed, air pressure, direction 5 5.25 1006.1 5.47 845.4 21 45
and temperature for September 2010
6 4.8 1006.6 5.14 844.4 20 30

7 4.62 1006.5 5.05 844 19 350

Date Height (13 m) Heigh (20 m) Temp . . . .
Air Air . Direction 8 4.92 1006.5 5.16 844 21 60
Sept Speed pressure Speed pressure (oC)
2010 (m/s) (mb) (m/s) (mb) 11 5.5 1006.4 5.28 844.2 22 10

1 3.5 1007.2 3.6 848.6 18 110 12 5.06 1006.9 6.02 844.7 23 5

2 3.67 1007 3.72 848.1 18 115 13 5.48 1007.2 6.44 845.2 22 350

3 3.42 1006.9 3.64 847.2 20 105 14 6.48 1007.3 6.93 846.2 23 345

4 4.2 1007.1 4.26 846 18 100 15 7.67 1007 7.88 847.1 22 45

5 5.14 1007.7 5.4 845 20 130 18 6.38 1007.3 6.75 847.8 22 30

7 5.08 1007.8 5.48 844.1 21 10 19 6.18 1007.3 6.49 848.2 22 55

8 3.8 1008.3 4.22 844.6 19 5 20 6.1 1007.8 6.49 848.2 21 35

9 3.99 1008.6 4.15 844.9 18 360 21 6.53 1007.6 6.91 847.7 19 60

10 4.74 1008.5 4.84 845.6 18 25 22 5.85 1008.7 6.12 847.1 18 40

13 5.37 1008.7 5.61 846.3 18 40 23 6.11 1008.9 6.49 846.8 20 15

14 4.63 1009.1 4.81 847.1 19 40 25 4.83 1008.8 5.18 846.8 18 320

15 4.14 1009.3 4.91 847.9 20 40 26 6.1 1008.4 6.49 846.8 19 55

16 3.99 1009.6 4.27 848.3 20 40 27 6.1 1008.5 6.49 847.2 19 55

17 3.62 1009.8 4.08 848.1 20 40 28 6.11 1008.3 6.49 847.6 18 85

20 3.85 1010.6 4.23 847.6 20 35 29 4.58 1008.5 4.82 848.1 19 250

21 3.92 1010.8 4.09 847.1 19 35 30 5.71 1008.7 5.9 848.3 21 110

22 4.81 1011.1 5.24 846.9 18 15 Av. 5.58 1007.41 5.96 846.61 20.4

23 4.25 1011.3 4.38 846.7 19 15 Max. 7.67 1008.9 7.88 848.5 23

24 4.78 1011.7 5.35 847.2 20 15 Min. 4.37 1006.1 4.75 844 18

3
43
ISSN 2079-6226: Proceedings of the 2012 Mechanical Engineering Conference on Sustainable Research and Innovation, Volume 4, 3rd-4th May 2012

Table 4: Daily averages of wind speed, air pressure, direction

and temperature for November 2010

Temp
...
Date Height (13 m) Height (20 m) . . Direction
o
Air Air ( C)
Nov. Speed pressure Speed pressure
2010 (m/s) (mb) (m/s) (mb)

1 5.34 1007.2 6 848.6 18 115

2 4.09 1007 4.56 848.1 18 105

3 4.3 1006.9 4.82 847.2 20 100 Fig. 2: The frequency distribution of wind speeds at the site
4 3.62 1007.1 4.05 846 18 130
The frequency distribution graph of wind speed is unevenly
5 4.09 1007.7 4.48 845 20 25 distributed and shows that the highest number of recorded
6 3.86 1007.9 4.3 844.4 20 10 wind speed values lie within the wind speed range of 3.0-3.9
m/s while the lowest number of recorded wind speed values
7 3.74 1007.8 4.17 844.1 21 5
lie within the range of 15 m/s and above.
8 4.56 1008.3 4.95 844.6 19 360

9 4.82 1008.6 5.21 844.9 18 25 The mean wind speed

10 4.05 1008.5 4.39 845.6 18 40

11 4.48 1008.7 4.79 846.3 18 40


V 
fv i i

10142.4
 5.04m / s (7)
12 4.3 1009.1 4.6 847.1 19 40
f i 2012
13 4.17 1009.3 4.43 847.9 20 40

14 4.26 1009.6 4.59 848.3 20 40 The standard deviation is given by:

15 4.67 1009.8 5.03 848.1 20 35
2
16 4.68 1010.6 5.07 847.6 20 35
2 
fD i i

13504.36
 6.71 where
17

18
5.15

4.8
1010.8

1011.1
5.24

4.96
847.1

846.9
19

18
15

15
f i 2012
D is the deviation from the mean wind speed.
19 5.44 1011.3 5.57 846.7 19 15

20 5.9 1011.7 6.05 847.2 20 30 Hence the standard deviation is

22 4.39 1011.6 5.91 847.9 22 25

23 5.08 1011.7 5.36 848.6 20 340    2  6.71  2.59 (8)

24 4.46 1011.8 4.86 849.2 20 310

26 4.9 1011.9 5.15 849.1 19 300 Because of the intemitent nature of wind speeds to assume and
27 4.72 1012.1 4.92 849.3 21 315 use the mean wind speed only to predict the wind energy
potential generally gives very rough estimates. Consequently,
28 3.99 1012 4 849.3 20 20
for effective energy prediction the wind speed data is fitted in
29 4.05 1012 4.3 850 19 25 to Weibull distribution model (which is characterised by
30 4.57 1012.1 4.65 850.1 20 30 Weibull scale factor (c) and the shape parameter (k) for much
better predictions.
Av. 4.52 1009.78 4.98 847.29 19.3

Max. 5.9 1012.1 6.05 850.1 22 The Weibull (shape) k-parameter and the Weibull (scale) c-
Min. 3.62 1006.9 4.05 844.1 18 parameter

The shape parameter was calculated using equation 4, and

the value found to be 2.06.The value obtained indicates a
IV. DISCUSSION peaked distribution i.e, the wind speeds tend to be very close
indication less speed variation. This is evident in the Weibull
Frequency distribution of wind speeds
distribution curve in figure 3. The empirical formula in
The frequency distribution curve of the grouped data is
equation 5 was used to calculate the scale parameter which
presented in figures 2.
was found to be 5.69 m/s. The calculated Weibull shape and

4
44
ISSN 2079-6226: Proceedings of the 2012 Mechanical Engineering Conference on Sustainable Research and Innovation, Volume 4, 3rd-4th May 2012

scale parameters gave a Weibull probability curve as shown Estimation of the Mean Wind Power Density
in figure 3.
From the formula in equation 6, and using Weibull k=2.06,
Scale parameter of 5.69 m/s and air density of 1.1 kg/m3, the
power density was found to be 131.35 W/m2 . It is observed
that the power density is almost twice as much power as the
wind would have if it were blowing constantly at the mean
wind speed of 5.04 m/s. This is because the power density is
obtained from the contribution of every speed with which the
wind flows during the fluctuations, while the power available
in the wind assumes that the wind constantly blows at a
constant speed, this is not practical as sometimes the wind will
be extremely high while other times it will be lower. For a
height of between 10m to 30m which is the range within
which this research was done, the estimated wind power
density lies in the wind power density class 2

Considering the effect of the Beltz limit whose value is 0.593,

Fig. 3: Weibull probability distribution of the wind speeds the maximum extractable power approximates 59.3% of the
theoretical power density. For this site, the maximum
Compared to the frequency distribution graph in figure 2, the
extractable power by a system of unit area operating at its
Weibull probability curve in figure 3 gives a good fit for this
optimum efficiency would be 77.9 W/m2
experimental data and therefore provides the probability of
obtaining a given wind speed at any given time in the study Considering the typical wind turbine design for Ngong, the
site. Vestas V52-850 kW wind turbine with the following
The area under the Weibull curve in figure 3 equals unity specifications (www.thewindpower.net);
since the probability that the wind will be blowing at some
Table 5: Technical characteristics of Vestas V52 wind turbine
wind speed including zero must be 100 per cent.
As can be seen from the graph, the distribution of wind speeds
is skewed to the left showing that there is high probability of
Cut in Wind speed 3 m/s
wind blowing at a wind speed lower than the median of the
horizontal scale. (to the left of the median value ) The curve Cut out wind speed 25 m/s
peaks at wind speeds of 3.5 m/s meaning 3.5 m/s are the most Rated Power 850 kW
probable speeds. Figure 4 gives the cumulative distribution
function of the wind speeds with the curve representing the Rotor diameter 25 m
probability that the wind speed will be smaller, greater than or Rotor swept area 2,123.72 m2
equal to a given wind speed. It is observed that 30% of the
wind speeds recorded will be above the wind speed 3 m/s. Number of blades 3
This result is important especially when analyzing the
percentage wind speed distribution for any given site.
If a wind turbine of the above specifications is installed at our
site in Juja and operates at optimum efficiency, then from the
maximum extractable wind power density for Juja (77.9
W/m2); the amount of power that can be produced is 165 kW.
This value is about a fifth of the production in Ngong station.
Juja therefore has a potential for not only wind power utility
for local community consumption but also has the capacity to
feed considerable amount of power to the national grid if fully
harnessed.
Wind Direction Analysis

The Juja wind rose diagram for the heights of 13 m and 20 m

are presented in figures 5 and 6 respectively. The horizontal
scales represent the frequency of occurrence of wind blowing
from a particular direction sector.

Fig. 4: The Weibull cumulative frequency curve

5
45
ISSN 2079-6226: Proceedings of the 2012 Mechanical Engineering Conference on Sustainable Research and Innovation, Volume 4, 3rd-4th May 2012

Juja Wind Rose Diagram for 13m Height From the diagrams, it is evident that a large frequency of wind
at this site blows from the North , north east. On the other
0
hand the frequency of wind is lowest from the south and south
east directions; while for the rest of the directions the
315 45
frequency is generaly low too. There is a large calm condition
in the wind system of the station such that the maximum bulge
of wind blows from the north to north east diretion It can
therefore be said that the station is in the monsoon wind
system although there could have also been the influence of
the buildings that were closer especially on the southern
direction this could have influence the wind direction,
270 90
moreover, the presence of the dam on the north north east
0 100 200 300 400
direction could also have influence the regime of wind
direction through the breezes.

V. CONCLUSSION

225 135
The following conclusions are made from the results of this
study:
1. The Weibull shape parameter k for the site was found
180 to be 2.06 while the scale parameter c was found to
be 5.69 m/s this depicts that the site has good wind.

Fig. 5: Wind Rose diagram for the Juja wind speeds at 13 m 2. The most prevalent wind speeds were found to be in
height the range of 3.0 to 3.9 m/s speeds.

Juja Windrose Diagram for 20m Height 3. From the wind rose analysis, most winds for the site
were found to be in the North and North East
0
directions

4. The site is found capable of power generation to the

315 45
tune of 165 kW if a typical wind turbine like Vestas
is to be used.

REFERENCES

[1] D. A Bechrakis and P. D Sparis, (2000). Simulation of the wind

speeds at different heights using artificial neural networks. Wind
270 90 Engineering 24 (2), 2000, pp. 127-136.
0 100 200 300 400 [2] A. N Celik, Statistical analysis of wind power density based on the
Weibull and Rayleigh models at the southern region of Turkey.
Renewable Energy 29: 2004, pp. 593-604.
[3] A. Dermibas, Biomass resources facilities and biomass conversion
processing for fuels and chemicals. Energy conversion and
management 42: 2001, pp. 1375-1378
[4] J. N. Kamau, R. Kinyua and J. K. Gathua, 6 years of wind data for
225 135
Marsabit, Kenya average over 14m/s at 100m hub height; an
analysis of the wind energy potential. Renewable Energy 35
(6),2010.
[5] W. H. Kirui, Assessment of the wind and solar energy potential in
180 the central rift valley of Kenya. MSc. Thesis. Egerton University,
Nakuru, Kenya, 2006.
Fig. 6: Wind Rose diagram for the Juja wind speeds at 20 m [6] J. F. Manwell, J. G. McGowan and A. L. Rogers, Wind energy
height explained Theory: design and application, John Wiley & sons Ltd.
West Sussex, United Kingdom, 2009.
[7] S. N. Marigi, Assessment of solar energy resources in Kenya
The wind rose diagrams shows the frequency with which wind Ph.D. Thesis, Moi University, Eldoret, Kenya, 1999.
blows from the different directions; The various arms radiate [8] C. Oludhe and L. J. Ogallo, Statistical characteristics of surface
from the innermost point and the length of each arm represents winds over Kenya. Journal of applied statistics 16: 1989, pp. 331-
343.
the total percentage frequency of time (number of occurrence [9] J. S. Rohatgi and Nelson Vaughn, Wind characteristics; an
of wind in that particular direction) the wind blows from the analysis for the generation of wind power. Alternative Energy
direction concerned. Institute, West Texas, Canyon, Unites States of America, 1994.
[10] M. Stiebler, Wind energy systems for electropower generation:
Green energy and technology. Springer Verlas Heidelberg, Berlin,
2008.

6
46