This paper describes an analysis of historical surface wind data for Eritrea. Winds in this region aredirected by summer and winter monsoons in addition to diurnal land-sea effects. Initial findingsindicate areas of outstanding potential for utility-scale wind development along the southerncoastline, where mean annual wind speeds range from 6 to 8 meters per second. Data recorded atthe Aseb airport between 1977 and 1986 indicate a mean annual wind speed of 9.5 m/s and amean annual wind power density of just under 1 kW/m2. An analysis of Red Sea wind dataimplies that similar potential may be found along the lower 20 km of the Eritrean coastline. Motivations for this study and feasibility of wind power integration in Eritrea are also discussed..
Introduction
Mulugetta and Drake (1996) conducted a study of wind potential for Ethiopia, Djibouti, and Eritrea.1 Their results indicate that the highest potential for wind development is along the Eritrean coastline. Other studies of the region surrounding the Red Sea and Gulf of Aden have also issued promising results (Elliott & Renne, 1987)(Pallabazzer, & Gabow, 1991)(Radwan, 1987). This study provides a more detailed account of the wind potential in Eritrea and discusses the feasibility of wind power integration at the southern Eritrean port city of Aseb..
Study Area
Eritrea occupies the southernmost 1200 km of the Red Sea shoreline, including hundreds of islands near the major ports of Mitsiwa'e in central Eritrea and Aseb in the south.
The capital of Asmera is located in central Eritrea on a plateau at over 2000 meters elevation. This high plateau and the surrounding hillsides are known as the central highlands. To the east and west are the western and coastal lowlands respectively. The western lowlands consist chiefly of desert hills and savanna. To the east, the high plateau drops abruptly to the wide, arid coastal plain, which stretches the entire length of Eritrea. South of Mitsiwa'e, the coastal plain dips to 100 meters below sea level at the Danakil depression, which divides the central highlands in the west from a much smaller range of mountains in the east. This second range parallels the shoreline from the Danakil depression to beyond the Djibouti border, reaching less than 2000 meters at its highest point.
The population of Eritrea is estimated at 2.7 million. Of that number, roughly 20% inhabit Asmera and the high plateau. Although accurate population counts for other major cities are not available, it is estimated that more than half of the Eritrean populace lives in the towns and cities of the highlands.
The main national electric grid stretches from Mitsiwa'e to Asmera, and then continues south to Adi Quala. A second, smaller, national grid is located at Aseb..
Wind patterns in the Red Sea region are controlled to a large extent by two distinct monsoons that are separated by 30 to 45 days each. The Northeast Monsoon dominates the southern half of the Red Sea from November to March. The African equatorial low centered at 10N 30E creates a strong southeast wind that is pulled through the narrow Bab el Mandeb at the southern mouth of the Red Sea. This wind slows as it continues toward Mitsiwa'e where it converges with the northwest wind. Much of the resulting air mass is drawn southwestward across the Eritrean highlands toward the African equatorial low pressure system centered in southern Sudan. During the Southwest Monsoon from May to September, the warm Pakistani low formed at 30N 65E causes surface winds along the entire length of the Red Sea to blow from the northwest, down the length of the Red Sea, and out into the Gulf of Aden (US Navy, 1993).
Study Overview
Our original goal was to use existing data to locate potential sites for utility-scale wind power development in Eritrea. We began our study by searching for existing sources of wind information. In the course of this search we discovered that most of the wind data that had been collected was lost at the end of the Eritrean war of independence in 1991. The three sources of wind information that were found in our initial search are described below..
1. Eritrean Data (1994-1995)
In Eritrea, monitoring efforts were being conducted at 7 stations located in towns and cities scattered throughout the Eritrean highlands. The Eritrean Department of Water Resources began recording continuous 15 minute averages at 7 stations in the highlands and western lowlands in 1994. Accurate mean annual wind speeds could not be derived because data sets spanned less than one year, however rough intervals were estimated from the available data. Data from these stations are summarized in Table 1..
Station Geographical Coordinates MAWS (m/s)
Asmera 15º 17´ N 38º 55´ E 2 - 3
Filfil 15º 36´ N 38º 57´ E 1 - 2
Kerkebet 16º 18´ N 37º 24´ E 2 - 4
Omhajer 14º 20´ N 36º 39´ E 1 - 2
Shambiko 14º 57´ N 37º 53´ E 1 - 2
Sh'eb 15º 51´ N 39º 03´ E 1 - 2
Tsorena 14º 39´ N 39º 13´ E 1 - 2
2. Italian Data (1930-1933)
Intermittent daily averages were recorded at 4 stations along the Italian-built railroad line from Mitsiwa'e to Keren during the Italian occupation. This data is summarized in Table 2..
Station Geographical Coordinates MAWS (m/s)
Asmera 15º 17´ N 38º 55´ E 2.7
Faghena 15º 10´ N 39º 28´ E 2.1
Keren 15º 46´ N 38º 27´ E 1.9
Mitsiwa'e 15º 37´ N 39º 27´ E 3.7
3. U.S. Navy Climatic Study of the Red Sea
A 1982 climatic study of the Red Sea done by the United States Navy mapped monthly averages with 1 degree spatial resolution. The report indicates mean annual speeds of between 6 and 8 m/s for the lower half of the Eritrean coastline.
A synopsis of the historical wind data available is shown in Figure 2..
[Figure : Map of mean annual wind speed estimates from historical data]
These preliminary estimates indicate that wind speeds at the existing meteorological stations in the Eritrean highlands are not adequate for utility wind generation. This is unfortunate because the capital city of Asmera, which has the greatest demand for electricity, is located in this area.
While these initial results seem somewhat discouraging, the highland stations were not erected for wind prospecting purposes and so were not optimally situated for that task. VanBuskirk et al. (1997) suggests that the hills and valleys of the highlands will create areas in which wind speeds are significantly enhanced. The Eritrean Department of Energy is currently collecting additional wind data in the mountain passes near Asmera in an attempt to locate sites with better wind potential.
While initial results do not indicate good wind sites in the Eritrean highlands, the Naval report for the Red Sea indicates mean annual wind speeds of over 8 meters per second just off the southern coast of Eritrea. Although demand at Aseb is not as high as at Asmera, we decided to pursue a detailed study of this area based on the implied wind potential .
Aseb Study
Aseb is a port city of about 50,000 people located at the southernmost tip of the Eritrean coastline.
This area is extremely hot and dry. Summer temperatures from June to September average 40-50ºC and rain is rare. The economy here relies on the port, which is shared with landlocked Ethiopia. Other industries include the salt flats, the national petroleum refinery, and a small fishing industry..
This study will attempt to provide the following for Aseb:
* Mean annual wind speed
* Diurnal and seasonal wind speed patterns
* Annual wind speed frequency distribution
* Potential wind farm capacity factors
* Power in the wind
* Wind roses for peak daytime winds
Data
The Eritrean Civil Aviation Department maintains two stations at Aseb: one at the airport about 10 km northwest of Aseb, and one at the seaport within the city. The airport station is located in a flat, wide-open area at about 10 meters elevation. The seaport station is at the shoreline, surrounded by buildings..
Until the end of Ethiopian occupation in 1991, 3-hourly weather observations collected at Aseb were archived by the United States Air Force Environmental Technical Application Center (USAFETAC) in the DATSAV2 Surface Database (USAFETAC, 1986). Observations reported before 1986 were taken at the airport station, while observations reported after 1986 were taken at the seaport station. At the writing of this paper, weather observations continued at Aseb, but were not reported to USAFETAC. The National Climatic Data Center (NCDC) provided a data set containing thirteen full years (1977-79, 1982-91) of DATSAV2 Surface observations for Aseb. After removing missing data indicators and other inconsistencies, over 5,000 observations remained for each station..
Division of Data
As discussed in the Study Area section, weather patterns throughout the year can be divided in to two distinct seasons. For the purposes of this analysis, the data set was divided into two subsets corresponding to the two seasons. Monthly mean wind speeds were used to assign the data to seasonal subsets. The summer subset contains the months from May to September, while the winter subset contains the months from October through April. A t-test performed on the two subsets shows that the mean wind speeds are significantly different (t=42.8).
Each seasonal subset was further divided according to whether the observations were recorded at the airport or the seaport station. For both seasonal subsets, the differences between the observed wind speeds at the two stations were significantly different (summer t=26.2)(winter t=25.2). The resulting four subsets were analyzed independently. Initial results are presented in Table 4..
AS AW SS SW
N OF CASES 2593 3258 1899 3244
MINIMUM 0.000 0.000 0.000 0.000
MAXIMUM 31.000 42.200 30.000 44.000
MEAN 7.671 12.118 4.819 9.148
VARIANCE 18.265 24.995 13.075 20.830
STANDARD DEV 4.274 4.999 3.616 4.564
STD. ERROR 0.084 0.088 0.083 0.080
SKEWNESS(G1) 0.259 0.144 0.924 0.214
KURTOSIS(G2) 0.194 1.328 1.636 0.919
To improve the accuracy of the analysis, the four data subsets were each divided into eight 3-hour bins. This was done to compensate for the fact that more observations were made during the daytime, when wind speeds are substantially higher than at night..
Here, a problem arose in that records for late evening and early morning
hours do not appear in the data set until after 1986, when the observation
site was moved to the seaport. Of the 16 data subsets for each season,
3 contain insufficient data for analysis: those for 21:00, 00:00 and 03:00
at the airport. Figure 4 illustrates the data divisions and the number
of observations in each cell..
Analysis
Assumptions
To account for the missing data, we assumed similar wind regimes at the two stations. We felt that wind speeds at the seaport would be closely related to wind speeds at the airport, as they are separated by less than 10 kilometers of relatively flat, open land. However, we also expected the more exposed station at the airport to have observed substantially higher wind speeds than the seaport. To provide very rough estimates of the missing data at the airport, therefore, we assumed a linear correlation between the two stations for average wind speed and Weibull parameters..
[map of Aseb]
Calculation of Mean Wind Speeds (diurnal, seasonal, and annual)
To derive the mean annual wind speed at the airport, it was necessary to estimate mean wind speeds for the cells with insufficient data. For each season, mean values were calculated for the 8 cells at the seaport and the 5 cells at the airport containing a sufficient number of data points.
Using regression analysis, linear correlations were derived between the 5 available pairs of mean wind speeds at the two stations (R2 = 0.96). Each of the three missing values for the two seasons were then calculated as µa=m·µs+b, where µa and µs are the mean wind speeds at the airport and seaport respectively, and m and b are the slopes and y-intercepts found in the regression analyses.
Calculation of Wind Speed Frequency Distribution
Due to the uneven distribution of observations among the 3-hour bins, we chose to calculate a distribution for each cell, and then average the distributions to obtain the seasonal and annual wind speed frequency distributions for each site.
As described above, some of the cells for the airport station contained insufficient data to calculate the wind speed distributions. To estimate distributions, we planned to calculate Weibull parameters for the cells with sufficient data, and then use them to estimate the Weibull parameters for the cells with insufficient data. From these estimated parameters, model distributions for these cells could then be constructed..
At this point it became clear that the wind speed database for Aseb contained a large number of calms. This presented a problem because the standard Weibull distribution does not allow for a wind speed of zero. To resolve this problem, we employed a method described by Mulugetta and Drake (1996) in which a Weibull distribution is fit to the observed distribution above 0 m/s, and calms are considered separately. This method tends to provide a more conservative estimate of mean wind speed by assuming that the reported frequency of calms is correct, whereas the standard Weibull distribution assumes that the frequency of calms is zero. This is particularly important in the case of the Aseb data, where the frequency commonly drops abruptly between 0 and 1 m/s, and then follows the expected Weibull distribution closely..To the standard Weibull parameters of k and A is added a variable s denoting the frequency of calms. The equation of the modified Weibull distribution then becomes:
(2)
where t(v) is the non-cumulative wind speed frequency distribution, s is the frequency of calms, v is the wind speed, A is the (?) parameter, and k is the shape parameter, similar to the kurtosis of a normal distribution..
For each season, this modified Weibull model was fit to the wind speed frequency distributions for the 13 cells containing sufficient data. A best fit was obtained by minimizing the root mean square (RMS) between the observed frequency distribution and the Weibull distribution..
The modified Weibull parameters of k, A and s were calculated for each cell. For each parameter, a linear correlation was determined between the two stations for each season. The missing values were then calculated as y = m·x + b, where y and x are the parameters at the airport and seaport respectively, and m and b are the slopes and y-intercepts found in the regression analyses..
Using the modified Weibull method, mean wind speeds are calculated as:
(3)
Most mean 3-hour wind speed values obtained using this equation are within 2% of the observed mean wind speeds calculated above. The greatest difference between values calculated using the two methods was 4.3% at 03:00 at the airport..
[Insert graphs of Modified Weibull fits to 3-hour subsets]
To obtain seasonal wind speed frequency distributions for the airport, the 5 observed and 3 estimated frequency distributions for each season were averaged. Seasonal wind speed frequency distributions at the seaport were simply calculated as the average of the eight observed frequency distributions. Because the summer season has 5 months and the winter season 7, annual distributions for each station were calculated as:
(4)
where F(v) is the annual wind speed frequency distribution, and Fs(v) and Fw(v) are the summer and winter wind speed frequency distributions, respectively.
To summarize the wind regime at the two Aseb sites, the modified Weibull distribution was fit to the annual wind speed distributions to derive Weibull parameters. At both sites, the modified Weibull fit was a substantial improvement over the standard Weibull fit..
[Table: Root Mean Square between observed distribution and model distributions]
Airport Seaport
Modified 1.9E-04 3.6E-04
Standard 3.5E-04 7.6E-04
Calculation of Power in the Wind
Potential Capacity Factor
For future estimates of potential wind farm output, it is useful to derive a theoretical capacity factor. Capacity factor is defined as the ratio of actual energy output to the amount of energy a project would produce if it operated at full rated power for 24 hours per day within a given time period.
The best sites are those with wind regimes that allow the turbines to run at their maximum output the largest fraction of the time. In California wind farms, operating capacity factors for 200-kW turbines are observed to be approximately 22% (Lynette and Gipe 1994).
The potential capacity factor is estimated by
(5)
where v is the wind speed, i is the cut-in speed of the turbine, j is the cut-out speed, F(v) is the frequency distribution and P(v) is the power curve for the turbine.
Potential Wind Power Density
Wind power density is the amount of wind power available to the plane perpendicular to the direction of the wind. In this study we describe wind power density in units of watts per square meter (W/m2). In practice, wind power density is useful in estimating the anticipated electrical output of a wind farm once the total area swept by the wind turbines is known. In this study we calculate the mean annual wind power density as
(6)
where is air density, v is the wind speed, and t(v) is the theoretical wind speed frequency distribution.
Results
Mean Annual Wind Speeds
[Tables with seasonal/diurnal calculations and estimates]
?Figures in parentheses represent estimated values based on linear correlation
[Graph of annual average diurnal wind speeds]
The data analysis indicates that the mean annual wind speed at the airport is 9.5 m/s, 2.5 m/s higher than the mean annual wind speed at the seaport. This difference between the observed mean wind speeds is mainly due to the difference in wind exposure between the two stations. The airport is located outside the city and has excellent wind exposure, whereas the seaport station is located in a developed area within the town..
(Diurnal wind speed pattern)
(Other discussion?)
Theoretical Wind Speed Distribution
[Figure: Modified (3 parameter) Weibull fits to the observed wind speed distributions at the Aseb Airport and Seaport]
(Discussion)
Summary of the differences between results obtained using various methods
(Discussion)
Projected available power
Projected Capacity Factor
Based on the wind speed frequency distribution for the airport station and using the power curve for the Flowind AWT-27 wind turbine3, a capacity factor of 58% was calculated..
Projected Projected
Capacity kWh per
Factor Square Meter
Summer
Winter
Total
Mean Annual Wind Power Density
Potential Annual Energy Production
Coastal Study
Data
U.S. Navy Regional Climatic Study of the Red Sea and Adjacent Waters (1993) and its predecessor Climatic Study of the Red Sea South and Gulf of Aden (1982) were both derived from the Comprehensive Ocean-Atmosphere Data Set (COADS)4. COADS is a global collection of instantaneous marine observations recorded on ships between 1854 and 1993.
Original observations were recorded to the nearest knott, with 0.1 degree spatial resolution. The portion of COADS from 12N to 30N, and from 32E to 44E was obtained from the National Center for Atmospheric Research (NCAR). The data is distributed in two electronic files: the Compressed Marine Reports (CMR) and the Long Marine Reports (LMRF). The earlier set (CMR) covers the period 1854-1969. Until the 1930's, the majority of observations were reported to only the nearest degree. To preserve a spatial resolution of 0.1 degree, this analysis utilizes data recorded after 1940. The later set (LMRF) covers the period from 1969 to 1993 and is recorded entirely with 0.1 degree resolution. After checking the data for missing value indicators and other inconsistencies, the combined data sets contained over 500,000 useable observations..
Analysis
To remove biases that would result from uneven temporal sampling in the data, the data was divided into four 6-hour periods for the two monsoon seasons, producing a total of 8 temporal bins. Utilizing a Gaussian-weighted averaging of the data, a map was produced for each temporal bin. The four periods were then averaged for each season, and the annual average was calculated as the sum of 5/12 of the summer monsoon season average and 7/12 of the Northeast Monsoon season average..
Results
The results of this analysis are shown in Figure 2 which is a map of mean annual wind speeds in the Southern Red Sea. The results show annual mean wind speeds along the lower 200 km of the Eritrean coastline of between 6.5 and 8 meters per second. The high average wind speed just inside the Bab el Mandeb indicates an especially strong potential for continuing strong winds along the shoreline in the area of Aseb..
The general wind speed pattern shown here can be understood by thinking
of the Bab el Mandeb as a narrow conduit for a layer of marine air which
flows from the Gulf of Aden to the Red Sea during the Northeast Monsoon
season. The highest wind speeds are at the point of maximum constriction
of the conduit, while the lowest wind speeds are at the convergence zone
at the widest point in the conduit. As noted in previous work, (VanBuskirk
et. al, 1997) indications of this low-level jet can be seen in cloud patterns
and satellite images..
Discussion and Conclusions
Motivation
Evidence of the detrimental effects of fossil fuel by-products in the atmosphere gives us cause to prevent these emissions (IPCC 1995). The substitution of cleaner energy sources for fossil fuels is an important part of attaining this goal. As the majority of growth in global energy use is expected to occur in developing countries, it is in the interest of the international community that we employ cleaner energy sources to address as much of this growth as is practically possible.
Wind Energy
Wind power is viewed as one of the most promising of the cleaner energy sources. Wind turbines are becoming a common sight in many European countries, especially Denmark, the Netherlands, northern Germany, and Britain. In the United States, over 15,000 machines are installed. In these industrialized nations, wind energy is considered economical at locations where the mean annual wind speed exceeds 6 m/s. As turbine technologies improve, this number is expected to drop to 5.1 m/s (Grubb & Meyer 1993).
Wind energy may be cost effective at even lower wind speeds in locations where fossil fuels must be imported. This makes wind power attractive for many developing countries. India, for example, with an aggressive wind energy program had over 500 MW installed wind capacity by 1995. The modularity of wind turbines allows countries with limited capital to install capacity in relatively small increments. Other benefits for developing countries include a high ratio of employment to installed capacity, safety from international fuel supply interruptions and the ability to acquire low interest loans for emission-free power plants..
Eritrea
The capture of the capital city of Asmera in 1991 ended the Eritrean struggle for independence from Ethiopia. As a result of over three decades of war, famine, and poverty, Eritrea's resources are now largely depleted. The country suffers an energy shortage that limits its economic productivity. Over 80% of the population have no electricity and biomass resources are so scarce that many have no fuel even for cooking. Eighty percent of primary energy demand is met by biomass in this semi-arid region that is already largely deforested..
This young nation, validated by UN-supervised referendum in 1993, has shown admirable foresight by committing itself to a development path that promotes environmental protection and
rehabilitation, while maximizing fiscal and political independence and security. These goals cannot be met by the usual model of development based on imported fossil fuels. A 1995 newsletter prepared by the Secretariat of the National Environmental Management Plan for Eritrea
(NEMP-E) describes the energy crisis in Eritrea and the development policy of the Eritrean government..
Advantages of Wind Power for Eritrea
Some advantages of using wind power in Eritrea include
* Lack of carbon emissions may allow Eritrea to obtain loan subsidies
from industrialized countries..
* Utilization of national energy resources frees Eritrea from complete
dependence on foreign imports..
* Reductions in air, water and soil pollution, prevent environmental
degradation and reduce the necessity for regulatory expenditures in the
long term..
* Development of a clean energy supply provides proof that improving
the standard of living by increasing power production does not require
recapitulating the mistakes of the developed countries..
Feasibility of wind power integration at Aseb
Supply
According to these results, the Eritrean coastline has an outstanding wind resource, particularly in the Aseb area.
* Winds at Aseb are strong enough to generate electricity that is less
expensive than the current electricity supply
* Bidirectional winds allow for closer spacing of wind turbines, which
would further improve the capacity factor of a wind power plant built there
Wind generated electricity at Aseb could cost less than half the electricity
currently supplied by diesel at 6 cents per kWh.6
* The large port facility could aid in the transportation of the necessary
machinery and equipment
* The power plant could take advantage of the already existing electrical
infrastructure
Demand
The electricity demand at Aseb is currently met by several aging diesel generators, many of which are not very efficient and therefore expensive. The power system is generated and administered by three separate institutions: the Aseb Port, the Eritrean Electric Authority (EEA), and the Aseb Petroleum Refinery. The port and refinery generate their own electricity in order to assure a reliable and dependable supply which is under their control. The combined capacity of the power system is approximately 17 MW, though this may soon decrease to 11 MW as the generators at the refinery undergo rehabilitation. The average combined demand of the town, the refinery and the port is 5.3 MW, fluctuating from a low of 4.0 MW during the cooler winter monsoon season to a high of 8.1 MW in the hot summer monsoon season. The power generation system and associated costs are summarized in the following diagram:
*Based on the following prices: gas oil 1.53 Birr/l; EEA fuel oil 0.78
Birr/l; refinery fuel oil 0.735 Birr/kg, where 6.45 Birr = 1 US$ (June
1997)..
* A positive correlation was found between diurnal wind speeds and
demand, i.e. the largest supply is available at the daily peak of demand,
between 9 a.m. and 6 p.m..
* The plant would have community support from the local fishing industry,
which would welcome inexpensive refrigeration
The main factor constraining wind power development near Aseb is its relatively small size.
Grubb and Meyer (1993) reviewed studies of the potential wind energy penetration of utility grids. Those studies considered how much wind energy could be added before operating penalties of an intermittent source became larger than benefits. Although the study results varied, typical simulation results were between 15 and 30%, although with large grids and incorporation of a diversity of out-of-phase intermittent sources, the estimated potential contributed was as high as 45%, without reliance on storage. Taking the conservative end of the range, this suggests that the Aseb utility, without expansion, could accommodate no more than about 3 MW of wind, which could be supplied by 5 to 10 medium size turbines..
One possibility for further exploitation of this plentiful resource is a cooperative effort with Ethiopia in which winter wind power from Aseb is traded for summer hydropower from the Blue Nile region. Another option may be to connect the power system at Aseb to the national grid in the central highlands..Peak demand in summer, peak supply in winter..
Ongoing work
>From March 20 to April 4, 1997, the Department of Energy conducted a field investigation of wind energy potential for the Southeast Eritrean coast. The purpose of this investigation was to better define the areas of high wind potential, and to identify potential demand for intermittent wind power.
It was concluded that for sites near Aseb, the best monitoring location is the airport area. For full exploitation of wind energy potential in Aseb, measures need to be taken to increase the reliability of the city power distribution system, and to increase the integration of the power generation at the city, port, and refinery..
Further funding for this project has been procured from the Global Environment
Facility (GEF), a joint program between the United Nations Development
Program (UNDP), United Nations Environment Program (UNEP) and the World
Bank. A full scale wind prospecting effort will be undertaken under this
grant. Results of the ensuing study are expected after one year..
References
The State of Eritrea - Environment Eritrea Newsletter No. 4
US Air Force DATSAV2 Surface Observations from the National Oceanography and Atmospheric Association, National Climatic Data Center Air Weather Service Global Climatology Division, Federal Climate Complex -- Global, May 1973 to March 1996
Brower, M. C., M. W. Tennis, E. W. Denzler and M. M. Kaplan, Powering the Midwest, Union of Concerned Scientists, 1993..
Elliott, D., Renne, S., Wind Energy Resource Assessment of Egypt, Sixth ASME Wind Energy Symposium presented at the Tenth Annual Energy-Sources Technology Conference and Exhibition, Dallas, Texas, February 15-18m 1987, SED-Vol. 3..
Garbesi, Karina, Kello, Karen & Van Buskirk, Robert (1996). Preliminary wind energy resources assessment for Eritrea, Department of Energy, P.O. Box 5285, Asmera, Eritrea, 54 pp..
Garbesi, Rosen and VanBuskirk. "Wind Power In Eritrea, Africa: A Preliminary Resource Assessment." 1997 American Wind Energy Conference..
Grubb, M. J. and N. I. Meyer (1993). Wind Energy: Resources, Systems, and Regional Strategies. Renewable Energy: Sources for Fuels and Electricity. Washington, D.C., Island Press. T. B. Johansson, H. Kelly, A. K. N. Reddy and R. Williams, ed., 157-212..
Lynette, R. and P. Gipe (1994). Commercial Wind Turbines and Applications. Wind Turbine Technology: Fundamental Concepts of Wind