R. Van Buskirk⁽¹⁾, K. Garbesi⁽²⁾ and K. Rosen⁽²⁾

Like winds in California, the most consistently strong winds in Eritrea are produced by strong horizontal temperature gradients during the dry season (the time of the Northeast Monsoon in Eritrea). Regional pressure gradients during this season are produced primarily by the Central Asian High and the Equatorial African Low (Naval Oceanography Command, 1993). Dry air descending out of Asia cools due to evaporation when it reaches the Indian Ocean and the Gulf of Aden. This produces a cool humid inversion layer over these bodies of water. Synoptic pressure gradients then push the cool air to the West from the Gulf of Aden, but the westward motion is blocked by the Ethiopian highlands, which produce a topographic barrier that is consistently above 2000 meters high. This topographic barrier diverts the air in the inversion into the Red Sea and to the North into Central and Northern Eritrea and Northern Sudan. There, the cool sea air flows inland where it is heated and either rises into the Ethiopian Highlands or flows to the Southwest into the Equatorial Africa Low. Due to the low latitude and the relatively short distance scale of wind variations (10's to 100's of kilometers), Coriolis forces are weak and frictional forces often dominate near the land surface, producing conditions where wind velocity vectors are aligned with pressure gradients.

Winds in Eritrea are produced by a combination of synoptic-scale gradients, inversion layer dynamics, and local differential heating. Synoptic scale gradients, as indicated by monthly average pressure maps, are moderate (1 - 0.5 millibar/100 kilometers). Synoptic gradients combined with differential heating tend to produce wind velocities that are relatively low (mean velocities less than 4 m/s, and typically 1-3 m/s throughout the Eritrean interior). This is indicated by measurements at stations which do not have winds enhanced by inversion-layer dynamics and channeling.

.

Figure 1 is a NOAA channel 4 satellite image from April 1996 during largely clear sky conditions at approximately 3 p.m. local time. The contrast in the image has been enhanced to sharply illustrate the boundary between the cool eastern air mass and the warmer Western air. This boundary occurs along the Eastern-most crest of the highlands. In visible images this crest also becomes the Western boundary of the Eastern Escarpment fog bank.

There are different types of inversion layers that can exist. We will make a distinction between a Surface Boundary Layer (SBL) inversion and a Planetary Boundary Layer inversion. The SBL inversion is typically a near-ground nighttime inversion which is a few tens of meters thick and which is heated and eliminated during the day. The PBL inversion, which will be the main topic of our discussion regarding wind meteorology, is the inversion at the top of the mixed PBL. This inversion layer is a few hundred to a couple thousand meters thick and below the inversion the air is well mixed during the day.

The PBL inversion layer can be identified and characterized through a combination of satellite images, cloud indicators, and temperature data. Infrared satellite images identify air masses with different temperatures and humidities. Also, a fog layer can assist in identifying the upper layers of the cold humid air mass and where the air experiences moderate uplift. Meanwhile, along the Eastern Escarpment in Eritrea temperature data show differences in potential temperature for nearby stations at different altitudes providing further indications of the persistent PBL inversion layer.

Visible Meteosat images can be used to examine the motions in the inversion layer by illustrating low-cloud patterns in the Southern Red Sea.

Once it is recognized that synoptic pressure gradients are weak and a strong inversion layer exists that flows through complex topography, it becomes necessary to analyze inversion layer dynamics as a possible source of high-wind conditions. Such an inversion flowing through a topographic barrier into a warmer basin of air will have the largest flows through the lowest passes. Furthermore, steep slopes on the lee side of the passes will provide additional acceleration of the flow. Hence the areas most intensely investigated for potential wind energy sites have been highland passes with leeward slopes along the cold-warm air interaction line.

Using the above criteria, the Department of Energy has been undertaking occasional spot-measurements of sites along the interaction line where warm western air interacts with cooler eastern air (see figure 1). It has been observed that several sites in this zone have daytime wind speeds in excess of 10 m/s during the high-wind season (compared to 5-7 m/s for Asmera), but sites which are clearly feasible for utility-scale generation have not yet been found.

Investigations have focused on two areas located approximately 40 km North and South of the Eritrean capital of Asmera: Adi Teklezan and Dekemhare respectively. These sites indicate mean daytime peak winds of 9 - 11 m/s (approximately 1.5 to 2 times the velocity in Asmera) during the high-wind season at 5 m elevation.

Dekemhare appears to have winds which correlate to some extent with those in Asmera. Correlations with spot measurements give a relation of V_Dekemhare = 0.9 V_Asmera + 3 m/s during Easterly winds and V_Dekemhare = 1.1 V_Asmera + 1 m/s during Westerly winds for measurements taken during the day and evening (until 20:00). The r-squared values of these correlations are 0.6 and 0.8 respectively. Measurements taken with a recently installed automated data-logging station in Asmera indicate mean wind speeds in Asmera of approximately 3 m/s at 12 m height. The Dekemhare wind sites might be feasible if either sites with improved topographic enhancement are found, or higher measurements (at 30 m or 40 m height) indicate substantially higher wind speeds.

Adi Teklezan is located near deeper passes through the highlands which channel larger flows from the Red Sea basin to the Western lowlands. During Westerly or Southerly winds in Asmera, winds in Adi Teklezan are of similar magnitude. But when winds are Easterly, daytime winds in Adi-Teklezan are approximately 50% higher than those in Asmera. In addition to this, when moderate night-time Easterly winds exist in Asmera, strong night-time winds may exist at Adi Teklezan. Recent spot measurements have shown one-hour mean night-time wind in Adi Teklezan can be as high as 10 m/s at 5 m height when the night wind speed in Asmera at 12 m height is below 4 m/s. One week of continuous monitoring has indicated a mean wind speed of 5.4 m/s at 5 m during a period when winds in Asmera were 2.7 m/s at 12 m. If measurements at larger heights can yield a 20% increase in wind speeds, then this site may be marginally feasible. The lack of correlation with Asmera, and the existence of strong night-time winds warrants further investigation of the winds in this area.

As described above, the inversion at the Eritrean Eastern Escarpment ranges from 1500 to 2500 meters. Meanwhile, during the winter, airflow in the Red Sea is from the Southeast. This means that mountains along the Eritrean and Yemeni coast, which rise to above 2000 meters, can channel the inversion layer as it flows from the Gulf of Aden into the Southern Red Sea. This channeling -- because of mass conservation -- forces the flow to accelerate through constrictions in the Southern Red Sea.

For a detailed assessment of wind patterns we analyzed over 3000 instantaneous wind velocity readings for each of two stations near Aseb, slightly south of the point of maximum constriction. These stations show similar wind patterns but have winds of significantly different magnitudes. Both sites are near Aseb or slightly to the west. One site (station #1) has an estimated mean wind speed of 9.2 m/s at 10 meters, while the other has a mean speed of 6.9 m/s at the same height. Both sites show distinct high and low-wind seasons where wind speeds are approximately 4 m/s lower during the low-wind season. Meanwhile figure 4 shows the average daily wind cycle for station #1 for both the high-wind and low-wind seasons. It is seen that during the high-wind season, the wind peaks in early afternoon at values above 12 m/s and appears to be lowest in very early morning.

We provide a preliminary evaluation of wind energy potential at both sites (a more detailed evaluation will have to await the results of current monitoring efforts). To evaluate the wind energy potential, the relevant statistics are described by the power density function [Troen & Peterson, pg 97]. Meanwhile, for purposes of economic evaluation of potential projects, we also estimated the approximate capacity factor of turbines that might be installed in such a site. For the capacity factor estimate, we used the power curve for a Flowind AWT-27 rated at 250 kW to provide estimated capacity factors. In these calculations, we obtained the following values for the Weibull distribution parameters k and A, total power density (Pd), and capacity factor (Cf):

  k A Pd (W/m²) Cf

Station #1 High-wind 3.0 12.9 1210 0.71

Low-wind 1.9 8.5 460 0.35

Station #2 High-wind 2.5 10.1 670 0.48

Low-wind 1.3 5.1 170 0.14

Annual 1.6 8.6 467 0.36

In performing these calculations, data was binned into 3-hour intervals. Then the statistical distributions were analyzed for each interval and the average distribution was used for the aggregate calculation. This eliminates bias due to uneven sampling. For stations #2, there was sufficient data for all hours, while for station #1 there was insufficient data for the hours 3:00, 21:00 and 24:00. To interpolate for these hours the linear correlation between the k and A values of station #1 and station #2 were calculated for the available data and this relation was used to estimate the values of k and A for the missing hours.

In reality the capacity factor may also depend on the maintenance schedule and the turbine downtime at a wind power site, but we calculate the raw capacity factor integrating the turbine power curve with the velocity density function. Note that very high capacity factors are estimated for site #1 indicating an annual average capacity factor exceeding 50%. Compared to international average capacity factors of approximately 25%, this indicates that wind energy production at an enhanced South Red Sea site may cost half as much per kilowatt-hour as the international average (all other factors being equal).

Considering that these calculations were performed with 10 m wind data with no extrapolation to hub height, wind speeds and capacity factors at a hub heights of 30 m to 40 m may be even greater. Furthermore, the low daily and statistical variation (indicated by k > 2 for large A) means that turbines should operate a large fraction of the time at designed wind speeds. It is apparent from such data that Eritrea likely has wind energy sites near Aseb of the highest quality.

In Eritrea, the most promising wind energy sites exist by virtue of channeling and local enhancement of flows in a wintertime inversion layer. This inversion layer originates in the Gulf of Aden, flows to the Northwest into the South Red Sea and proceeds across the Northern tier of the Ethiopian highlands into the Eritrean and Sudanese lowlands. The inversion layer flows are channeled and accelerated in two areas: first at the mouth of the Red Sea where the point of maximum constriction is at Mt. Ramlu in Eritrea, and then through passes in the Eritrean Highlands.

The South Red Sea channeling produces excellent wind sites with wind power densities exceeding 850 W/m² at 10 m. Wind turbines situated at such sites could possibly be operated with capacity factors exceeding 50%. Highland sites are more marginal because of the nighttime decrease of wind velocities. This means that while daily peak winds may often exceed 10 m/s, rough estimates of the average wind speed yield 5-7 m/s at 5 m. indicating that these are marginal but possibly economical sites.

Wind power development along the Southeast Eritrean coast would likely be very economical, either yielding low electricity costs for government-lead development, or potential high returns for private investors. For power transmitted to the central national grid, transmission costs would require economies of scale that would dictate a minimum project capacity exceeding 50 MW. Since this is comparable to current national demand, such a national scale project would have to wait perhaps 5 to 15 years for demand (currently growing at 10% per year) to grow to the level where it could absorb such a large increase in supply.

Nonetheless, whether we contemplate a near-term three megawatt development or a long-term 50 megawatt development, it is to the benefit of both Eritrea and interested international partners to aggressively pursue wind energy development at this juncture.

This work was undertaken at the initiative of the Government of Eritrea with collaborative assistance from San Jose State University, U.S.A., sectoral support from the State of Lower Saxony, Germany, and private support from Dullas Engineering, U.K.