Renewable Energy 35 (2010) 1298–1302

Contents lists available at ScienceDirect

Renewable Energy
journal homepage: www.elsevier.com/locate/renene

Data Bank

6 years of wind data for Marsabit, Kenya average over 14 m/s

at 100 m hub height; An analysis of the wind energy potential
J.N. Kamau a, R. Kinyua a, *, J.K. Gathua b
a
Department of Physics, Jomo Kenyatta University of Agriculture and Technology, Nairobi, Kenya
b
Department of Physics, Kenyatta University, Nairobi, Kenya

a r t i c l e i n f o a b s t r a c t

Article history: The wind energy potential for generation of electricity and for domestic water pumping has been
Received 11 April 2009 investigated for Marsabit, Kenya. Marsabit (37 580 N, 2 190 E) lies in Eastern province approximately
Accepted 9 October 2009 560 km from Nairobi. Wind data from the Kenya Meteorological department for the period 2001–2006
Available online 13 November 2009
has been used to study the Diurnal, monthly and inter-annual variability using empirical methods
including the Power law and Weibull statistics. Average wind speeds greater than11 m/s at a height of
Keywords:
10 m are prevalent in this area. The available power density at a height of 100 m is between 1776 W/m2
Energy potential
and 2202 W/m2 which is in the wind class range of 7 and 8. The maximum extractable power density at
Weibull parameters
Wind speed 100 m varied between 1417 W/m2 and 1757 W/m2. Values of Weibull parameters k (dimensionless) and
Wind frequency c (m/s) ranged between 2.5–3.05 and 11.86–12.97 respectively. Wind Rose analysis revealed no marked
Wind class variation in wind direction and frequency throughout the year (mean direction between 150 and 160
Wind Rose degrees with highest standard deviation of 33.5 degrees). From the analysis, the site was found suitable
for grid connected power generation and also for other stand-alone generators that can be used for water
pumping and battery charging.
Ó 2009 Elsevier Ltd. All rights reserved.

1. Introduction Wind energy is the world’s fastest growing energy source,

Energy resource is one of the most important and continuous
issues of our time. It determines the form and rate of economic
growth. The energy demand is increasing continuously with pop-
ulation and industrialization and one of the ways to fulfill this
demand is to use renewable energy sources, that is, technologies
that can be used in future without irreparably or irreversibly
damaging the ecosystem [1].
In Kenya, the cost of oil based imports has increased by over
1000% in the past three decades. The cost of electricity alone has
more than doubled in the past year. However the generation of
electricity (hydro, thermal and geothermal) has been growing
slowly in Kenya and is currently at about 1200 MW. Furthermore,
due to the high component of hydropower (about 700 MW), the
country always faces chronic shortages of power during the dry
season. Therefore there is need to develop alternative energy
sources such as solar and wind energy to satisfy the growing
demand without harming the environment.
* Corresponding author. Tel.: þ254 722330488; fax: þ254 067 52164.
E-mail address: kinyua@fsc.jkuat.ac.ke (R. Kinyua).
expanding at annual rates of between 25 and 35% [2]. Wind
capacity world-wide increased by 26% in 2007 hitting the 100 GW
mark by July 2008 [3]. Recent interests in wind energy technology
have resulted in bigger turbines, larger rotor diameters and reduced
costs that are quite competitive with other sources. Currently,
Kenya has only 5.1 MW of wind power connected to the grid.
This paper examines the wind energy potential for Marsabit
(37 580 N, 2 190 E), a rural town in the northern part of Kenya with
possibilities of feeding the power obtained into the national grid
and also for pumping water for domestic use, irrigation and for
pasture in the surrounding areas. The specific objectives include:
1. Characterizing the various wind parameters viz., average speed
(diurnal monthly and inter-annual variability), Weibull shape
parameter, k (dimensionless) and scale parameter, c (m/s).
2. Performing Wind Rose analysis and determining the available
power density for the site.
The data used in this study consists of averaged time series wind
speeds for the period between 2001 and 2006. The wind speed data
was recorded at the standard Anemometer height of 10 m by the
Kenya Meteorological Department (KMD). Wind direction was also
recorded for the same period.

0960-1481/$ – see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.renene.2009.10.008
 J.N. Kamau et al. / Renewable Energy 35 (2010) 1298–1302 1299

Table 2
Monthly average values of Weibull shape parameter k (dimensionless) and scale
parameter c (m/s) for Marsabit, Kenya 2001–2006.

Month Parameters 2001 2002 2003 2004 2005 2006

Jan k 2.077 3.155 2.896 2.112 2.190 2.783
c 9.590 10.180 9.240 10.610 11.380 12.180
Feb k 3.486 3.610 2.708 1.884 2.894 2.694
c 14.300 11.760 11.110 12.180 13.040 13.080
Mar k 2.739 2.315 3.390 2.453 3.107 3.364
c 14.180 9.720 11.970 12.220 12.750 12.080
Apr k 2.278 2.427 2.741 2.204 3.316 3.184
c 11.040 9.760 11.550 10.380 12.770 10.930
May k 2.909 2.063 2.428 2.175 2.520 3.800
c 12.590 10.940 13.790 11.390 11.180 12.270
Jun k 3.278 3.038 2.367 3.459 3.416 4.042
c 12.600 12.700 12.480 14.870 13.900 13.360
Jul k 3.769 3.610 3.278 3.717 3.572 6.113
c 13.370 14.470 12.810 13.950 14.420 13.030
Aug k 3.379 3.768 3.363 4.260 3.708 3.149
c 14.140 16.030 12.140 14.430 13.460 12.080
Sep k 3.176 3.392 3.605 3.365 2.909 4.180
c 13.830 15.520 14.600 14.420 13.250 13.510
Oct k 3.072 2.866 3.297 3.047 4.229 2.795
c 12.220 12.670 11.670 13.020 13.610 10.640
Nov k 3.310 2.902 4.446 2.182 2.805 3.056
Fig. 1. Inter-annual probability distribution function for Marsabit, Kenya 2001–2006. c 10.510 10.290 11.330 10.030 10.770 9.440
Dec k 3.310 2.114 3.348 2.391 3.098 2.274
c 11.310 8.400 10.620 9.840 11.540 9.140
All k 2.817 2.533 2.835 2.502 2.984 3.053
2. Methodology c 12.510 11.970 12.000 12.360 12.720 11.860

Wind energy resource is highly variable both in space and time;   k1   k 
therefore, to understand the characteristics of the resource various k v v
f ðvÞ ¼ exp  (1)
parameters were considered. c c c

2.1. Probability distribution function (PDF) where f (v) is the probability of observing wind speed v, c (m/s) is
the Weibull scale parameter and k is the dimensionless shape
Due to the variability of the wind speed it is found useful to plot factor.
the wind speed probability distribution function (the percentage of The cumulative probability function of the Weibull distribution
time that the wind spends at each speed) to understand the char- is given by
acter of the variation. One of the commonly used functions for this   k 
purpose is the Weibull distribution [4], [5] given by v
FðvÞ ¼ 1  exp  (2)
c

Table 1
To determine k and c requires a good fit of Equation (2) to the
Monthly average wind speeds v (m/s) and standard deviation s (m/s) for Marsabit, recorded discrete cumulative frequency function [5]. Taking the
Kenya 2001–2006. natural logarithm of both sides of Equation (2) twice gives
Month Parameters 2001 2002 2003 2004 2005 2006
Jan v 8.530 9.140 8.240 9.420 10.130 10.850
s 4.253 3.071 3.094 4.662 4.773 4.261
Feb v 12.860 10.590 9.880 10.910 11.640 11.730
s 3.993 3.192 3.940 5.708 4.460 4.655
Mar v 12.630 8.600 10.760 10.870 11.400 10.890
s 4.959 3.973 3.542 4.744 4.150 3.665
Apr v 9.750 8.640 10.290 9.200 11.480 9.820
s 4.583 3.830 4.054 4.423 3.833 3.419
May v 11.220 9.730 12.250 10.100 9.980 11.100
s 4.226 4.856 5.335 4.896 4.180 3.197
Jun v 11.290 11.350 11.050 13.380 12.510 12.150
s 3.807 4.032 4.989 4.310 3.974 3.400
Jul v 12.080 13.060 11.480 12.590 13.040 12.110
s 3.622 3.855 3.775 3.856 4.033 2.260
Aug v 12.680 14.490 10.890 13.160 12.170 10.840
s 4.272 4.131 3.611 3.476 3.710 3.861
Sep v 12.370 13.930 13.160 12.960 11.790 12.270
s 4.313 4.470 4.003 4.223 4.507 3.485
Oct v 10.920 11.280 10.470 11.640 12.430 9.540
s 3.886 4.168 3.432 4.176 3.262 3.669
Nov v 9.430 9.170 10.340 8.890 9.640 8.460
s 3.131 3.425 2.556 4.284 3.682 3.101
Dec v 10.150 7.440 9.550 8.720 10.400 8.160
s 3.303 3.725 3.092 3.920 3.665 3.744
All v 11.150 10.620 10.700 10.990 11.380 10.650
Fig. 2. Inter-annual monthly variation of average wind speed for Marsabit, Kenya
s 4.286 4.487 4.038 4.718 4.171 3.837
2001–2006.
1300 J.N. Kamau et al. / Renewable Energy 35 (2010) 1298–1302
Fig. 3. Inter-annual variability of the average wind speed by hour of day Marsabit,
Kenya 2001–2006.
lnh  ln½1  FðvÞi ¼ k lnðvÞ  k ln c
Plotting lnC  ln½1  FðvÞD against ln (v) presents a straight line
whose gradient is k and the y-intercept is k ln c from which c can
be calculated.
2.2. Wind direction
A Wind Rose is a polar plot that represents the percentage of the
time that the wind direction falls within the sector of the compass.
This is important for determining the type of turbine to be used in
a particular location and also its orientation to the wind stream.
Where the wind direction is shown over a period, the value
represents the vector sum for the period. We show the direction for
standard anemometer height of 10 m.
2.3. Wind speed variation with height
As stated before, the standard anemometer height is 10 m.
However, in order to harvest wind energy, wind speeds at heights
greater than 10 m are required. Therefore, equations that predict
the wind speeds at one height in terms of the measured speed at
another height are required. The power law model has received
extensive application in the field of wind energy basically to esti-
mate the hub height wind speeds at various potential sites [6,7,8].
The formula takes the form
 a
v2 z2
¼ (4)
v1 z1
where v2 is the extrapolated wind speed at height z2 and v1 is the
measured speed at z1.
The exponent a depends on such factors as nature of terrain
(surface roughness), wind speeds and temperature [9]. For
neutral stability the exponent value of 1/7 has widely been
chosen as a good representative of the prevailing conditions
[10]. We used the power law to predict the wind speeds at the
various heights.
2.4. Wind power density
The maximum power, P, available from the wind
1
P ¼ rAv3 (5)
2
where r is the density of air and A is the swept area of the rotor. The
actual amount would be less since all available energy is not
extractable.
Monthly or annual wind power density per unit area, Pw of a site
based on Weibull probability density function can be expressed as
[5]
 
1 3 3
Pw ¼ rc 1 þ (6)
2 k
The two significant parameters k and c have been shown [11] to
be related to the mean value of the wind speed v as
 
1
v ¼ cG 1 þ (7)
k
The maximum extractable power PE (W/m2), by a system
working at its optimum efficiency, is limited by the power coeffi-
cient called the Betz limit whose value is 16/27 or 0.593 [12] and is
therefore given by
(3)
PE ¼ 0:593P (8)
This power coefficient makes the maximum extractable power
approximately 59.3% of the theoretical power density.
3. Results and discussion
In this study wind speed data for Marsabit for the six year period
from 2001 to 2006 was analyzed. Based on this data, wind speeds
were processed using WindographerÓ software program to obtain
distribution curves and Wind Rose plots. Calculations were then
made to obtain Weibull parameters k and c, the mean wind speeds
and the mean power density.
Fig. 1 shows the probability density function (PDF) for the years
2001–2006. All the plots show a similar profile of a near normal
distribution around the modal value of 11 m/s. The mean annual
values of wind speeds range between 10.6 and 11.4 m/s whilst the
k values range between 2.50 and 3.10 showing very small spatial
variation.
Tables 1 and 2 show the monthly average wind speeds v m/s and
their corresponding standard deviations. The lowest monthly average
wind speed of 7.440 m/s was registered in December 2002 whilst
August 2002 recorded the highest average wind speed of 14.490 m/s.
The lowest standard deviation of 2.556 m/s was recorded in
November 2003 whilst the highest standard deviation of 5.708 was
recorded in February 2004. The inter-annual monthly variations are
illustrated by Fig. 2. It is clear from the plots that the months of June,
July, August and September registered the highest average wind
speed. This can be attributed to the north-easterly wind currents that
arise due to the northern hemisphere summer (June–August). For
most years February and March registered high averages mainly due
to the south-easterlies resulting from the northern hemisphere
winter (December–February). The month to month variation
exhibited in Fig. 2 may be attributed to channeling effects of wind due
the many hills around this area [13]. The inter-annual variability in the
wind speeds could be associated with the inter-annual variation in
the monsoonal wind characteristics. Fig. 3 illustrates the inter-annual
variability of the average wind speed by hour of day. All the years
under consideration show a similar profile with the lowest average
 J.N. Kamau et al. / Renewable Energy 35 (2010) 1298–1302 1301

Average of Wind Power Density Marsabit 2002

0°
350° 10°
340° 20°

330° 30°

320° 40°

310° 50°

300° 60°

290° 70°

280° 80°

270° 90°

260° 100°

400 W/m²
250° 110°

240° 120°

800 W/m²
230° 130°

220° 140°

210° 150°
1,200 W/m²
200° 160°
190° 180° 170°

Average of Wind Power Density 2006

0°
350° 10°
340° 20°

330° 30°

320° 40°

310° 50°

300° 60°

290° 70°

280° 80°

270° 90°

260° 100°

500 W/m²
250° 110°

240° 120°

1,000 W/m²
230° 130°

220° 140°

210° 150°
1,500 W/m²
200° 160°
190° 180° 170°

Fig. 4. a: Average power density by direction for Marsabit, Kenya 2002. b: Average power density by direction for Marsabit, Kenya 2006.
1302 J.N. Kamau et al. / Renewable Energy 35 (2010) 1298–1302

Table 3 4. Conclusions
Summary of annual average wind speeds, Weibull shape parameter k (dimension-
less), Weibull scale parameter c (m/s), available power density P10 at height 10 m
(W/m2), available power density P50 at height 50 m and P100 at height 100 m, wind
The following main conclusions can be drawn from this study:
class, maximum extractable power density PE (W/m2), mean direction of the wind
( ) and the standard deviation sd ( ) of the wind direction. 1. The average wind speed for the site studied was found to range
Parameter 2001 2002 2003 2004 2005 2006
from 10.62 to 11.38 m/s. These are high wind speeds throughout
the six years giving power densities of between 903 and
v(10) (m/s) 11.15 10.62 10.70 10.99 11.38 10.65
k 2.81 2.53 2.83 2.50 2.99 3.05 1119 W/m2 at 10 m. Power law calculation gave available power
c (m/s) 12.51 12.97 12.00 12.36 12.72 11.86 densities of between 1776 and 2202 W/m2 at a height of 50 m
v100 (m/s) 15.4 14.7 14.8 15.1 15.7 14.8 and between 1776 and 2202 W/m2 at 100 m height. The
P10 (W/m2) 1097 1021 967 1119 1121 903 maximum extractable power at a height of 100 m was found to
P50 (W/m2) 2157 2007 1901 2199 2202 1776
Wind class 8 8 7 8 8 7
be between 1417 and 1757 W/m2. Increasing the hub height will
P100 (W/m2) 2900 2710 2550 2959 2963 2390 have a corresponding increase in the extractable power even
PE (W/m2) 1719 1607 1512 1754 1757 1417 though it will also come with an increased sophistication and
Mean direction ( ) 155.1 156.0 158.9 158.6 155.7 153.5 cost in the construction of a support tower.
sd ( ) 25.3 26.9 24.9 29.4 28.5 33.5
2. The yearly Weibull parameters k (dimensionless) and c (m/s)
ranged from 2.05 to 3.05 and 11.86 to 12.96 m/s respectively.
3. Wind Rose analysis showed that the wind direction was fairly
consistent from between 150 and 160 which is generally from
Table 4 the South to South East for most of the years.
Wind class at a height of 50 m. 4. This site was found suitable for power generation both for grid
Wind speed (m/s) Wind power density (W/m2) Wind class
connection and also for stand-alone activities such as water
pumping and battery charging.
<5.6 <200 1
<6.4 <300 2
<7.0 <400 3 It is anticipated that the study can give useful insights to the
<7.5 <500 4 stake-holders in wind energy technology.
<8.0 <600 5
<8.8 <800 6
<11.9 <2000 7
References

speed of 7.5 m/s occurring between 15 and 16 h (GMT). This corre-
sponds to the sunset hours in Kenya. High wind speeds of above 11 m/s
are prevalent during the nighttime hours and peaking to speeds
of above 13 m/s during the early morning hours. This enhanced
nocturnal resource may be attributed to the reduced vertical
mixing of low momentum near-surface air due to the large
thermal stability during the night hours.
Fig. 4a and b illustrate the average power density by direction for
years 2002 and 2006 respectively. Both years registered a similar
average wind speed but recorded different power densities at
a height 10 m of 1021 W/m2 and 903 W/m2 respectively. It can be
seen from both figures that year 2006 had a higher dispersion of the
wind speeds than 2002, the standard deviation of the wind direction
being 33.5 and 26.5 degrees respectively (Table 3). However, all the
years show a mean direction of between 150 and 160 degrees with
the highest standard deviation of 33.5 registered in year 2006.
A summary of the calculated parameters including the maximum
extractable power density PE (W/m2) at 100 m is shown in Table 3.
Table 4 shows the classification of wind speed and available
wind power density at a height of 50 m [14]. Values greater than
2000 W/m2 are recorded as class 8 which makes the site an
excellent wind resource.
[1] Ozgener O, Ulgen K, Hepbasli A. Wind and Wave Power Potential. Power,
Energy Sources 2004;26:891–901.
[2] Shika S, Bhatti TS, Kothari DP. Wind energy in India. Shifting paradigms and
challenges ahead. J Energy Eng 2004:67–74.
[3] IEA Wind 2007 Annual Report. www.ieawind.org.
[4] Seguro JV, Lambert TW. Modern estimation of the parameters of the Weibull
wind speed distribution for wind energy analysis. J Wind Eng Ind Aerodyn
2000;85:75–84.
[5] Akpinar EK, Akpinar S. An analysis of the wind energy potential of Elazing,
Turkey. Int J Green Energy 2004;1(2):193–207.
[6] Mikhail AS, Justus CG. Comparison of height extrapolation models and
sensitivity analysis. Physical climatology for solar and wind energy. Singapore:
World Scientific; 1986. 377–397.
[7] Oludhe C, Ogallo LJ. Vertical variation of wind power. J Discovery Innovation
1990;2:73–80.
[8] Lun IYF, Lam JC. A study of Weibull parameters using long term wind obser-
vations. Renew Energy 2000;20:145–53.
[9] Golding E. The generation of electricity by wind power. New York: Halsted
Press; 1976.
[10] Damarais AG. Wind speed profiles at Brookhaven National Laboratory. J Appl
Meteorol 1959;16:181–90.
[11] Ulgen K, Hepbasli A. Determination of Weibull parameters for wind energy
analysis of Izmir, Turkey. Int J Energy Res 2002;26:495–506.
[12] John WT, Antony DW. Renewable energy resources. Cambridge London: The
University Press; 1987.
[13] Asnani GC, Kinuthia JH. A newly found Jet in northern Kenya (Turkana Channel).
In: First international conference on southern meteorology, July–August 1984,
Brazil, pp. 337–380.
[14] Wind Resource Assessment Handbook. Fundamentals for conducting
a successful monitoring program. AWS Scientific Inc.; 1997. p. 14.