Solar Energy Vol. 70, No. 1, pp.

112, 2001
2001 Elsevier Science Ltd

Pergamon PII: S0038 092X( 00) 00126 2 All rights reserved. Printed in Great Britain
0038-092X/ 01/ $ - see front matter
www.elsevier.com/ locate/ solener
ASSESSMENT OF SOLAR ELECTRICITY POTENTIALS IN NORTH AFRICA
BASED ON SATELLITE DATA AND A GEOGRAPHIC INFORMATION
SYSTEM
, ,1 ,

H. BROESAMLE*, H. MANNSTEIN**, C. SCHILLINGS* *** and F. TRIEB***

*NEVAG/ Enersys GmbH, Rheingaustr. 184, D-65203 Wiesbaden, Germany
**Institute of Atmospheric Physics, DLR Oberpfaffenhofen, D-82234 Weling, Germany
***Institute of Technical Thermodynamics, DLR Stuttgart, Pfaffenwaldring 38-40, D-70569 Stuttgart,
Germany
Received 10 February 2000; revised version accepted 31 July 2000
Communicated by RICHARD PEREZ
AbstractSolar thermal power plants will provide a major share of the renewable energy sources needed in
the future. STEPS, an evaluation system for solar thermal power stations, was designed to calculate the
performance of such power stations as a function of direct solar radiation, geographical conditions (land slope,
land cover, distance from cooling water resources, etc.), infrastructure (pipelines, electricity grids, streets etc.)
and the conguration and performance of a selected solar thermal power plant concept. A cloud index derived
from METEOSAT satellite images is used to calculate the direct solar radiation resource. A geographic
information system (GIS) is used to process all the parameters for site assessment. In order to demonstrate the
concept, an analysis of Northern Africa was performed with STEPS providing a ranking of sites with respect to
the potential and cost of solar thermal electricity for a particular power plant conguration. Results were
obtained with high spatial and temporal resolution. 2001 Elsevier Science Ltd. All rights reserved.
1. INTRODUCTION Providing the basis for solar power build out
scenarios.
Solar thermal power plants use concentrated solar
Sensitivity analysis of the performance of solar
radiation in order to generate high pressure steam
thermal power stations regarding site condi-
for electricity generation in conventional steam
tions.
turbines. Because fuel is substituted by solar
STEPS has a clear modular structure (Fig. 1). The
collectors, an additional investment and additional
main module denes the interaction of all other
space at the plant site is required. The evaluation
modules. Temporal and spatial resolution can be
tool STEPS allows the selection and ranking of
varied according to specic needs. Results are
sites for solar thermal power plant construction,
typically shown as maps using a geographic
analysing a large region, country or even a
information system (GIS). The rst application of
continent (Broesamle, 1999). Among others, the
STEPS was the estimation of the potential and
following services can be provided by STEPS.
present cost of solar thermal power plants in
Maps of the direct normal irradiation (DNI)
North Africa. The following topics were treated
resource in high spatial and temporal resolu-
within the study.
tion (best resolution: 2.5 km32.5 km, hourly
Determination of the geographical and
mean values).
meteorological frame conditions.
Assessment of the technical and economic
Determination of all suitable sites for the
potential of solar power generation in a dened
construction of solar thermal power plants.
region.
Development of a map of the direct normal
Ranking and selection of sites for the construc-
irradiation (DNI) for North Africa.
tion of solar thermal power stations.
Calculation of the solar electricity yield per
2
km of land.

Author to whom correspondence should be addressed. Tel.:

Calculation of power generation costs per
149-711-686-2423; fax: 149-711-686-2783; e-mail:
kWh.
franz.trieb@dlr.de
1
The procedure and the results of this analysis are
Former afliation: Geographical Institutes of the University
of Bonn, D-53115 Bonn, Germany. described in the following.
1
2 H. Broesamle et al.
Fig. 1. The modular structure of STEPS.
2. ASSESSMENT OF THE GEOGRAPHIC cultivated land are considered unsuitable for the
FRAME CONDITIONS construction of such plants. Sand deserts are not
considered to be a criteria for exclusion, but may
All available data sets (satellite data, digital elevate the cost.
maps etc.) are processed in a geographic infor- Land cover and land use data sets are created
mation system (GIS). The usually heterogeneous by the US Geological Survey (USGS), the Uni-
data sets have to be transferred into a uniform versity of NebraskaLincoln (UNL) and the
geographic projection and data format. Important European Commission for Research Co-operation.
data for the ranking and evaluation of potential These data can be obtained partially from the
sites of solar thermal power plants are e.g. land Earth Resources Observation System (EROS)
use, land cover, slope and water surfaces. Data Centre via the internet. For the determination
Solar thermal power stations have a relatively of the land cover and land use, the global
big area demand in comparison to conventional ecosystems classication by Olson (1994) is used.
power stations. The specic area demand for a A reduction to 10 classes of land cover has been
2
parabolic trough power station is |1 km per applied.
50 MW of installed electric capacity. Typical sites The land slope should be less than 5% for
are hot, dry regions like deserts or semi-deserts. parabolic trough plants. We used the digital
Surface water, forests, settlements, arable and elevation model (DEM) called GTOPO30 of the
Assessment of solar electricity potentials in North Africa based on satellite data and a geographic information system 3
EROS Data Centre with a spatial resolution of with x 5t / 24 32p and t 5decimal hours of the
nearly 1 km31 km in order to calculate the satellite (UTC). a gives the daily mean tempera-
0
slope. The data were considered to be acceptable ture, a the temperature amplitude, a inuences
1 2
for the rst application of STEPS, but a land slope the width and steepness of the daily temperature
of maximum 2% instead of 5% was selected as wave and a gives the phase shift, which is
3
threshold in order to compensate for the limited dominated by the local solar time. The coef-
resolution. For more detailed project studies, more cients a are tted for every daily period and
n
accurate DEM may be used, as e.g. the Shuttle every pixel using the t from the former time
Radar Topography Mission (SRTM) by NASA interval and the corresponding new cloud free
with a 30 m spatial resolution. The shading of temperature.
mountains, especially in the morning and after- We use the following differences of properties
noon, can also be calculated with DEM data, but of clouds versus surface for a rst cloud detection.
this feature has been neglected in the rst applica- Clouds are cold. The weight of every pixel
tion of STEPS. with a temperature colder than estimated is
Fig. 2 shows some of the results of this reduced proportional to the difference.
2
analysis. In North Africa, 12.6 Mio km full the Clouds move. We compare the data to the
criteria of suitability for the construction of solar previous image and the image of the day
thermal power plants with respect to land slope before. Clouds are colder and show up as local
and land cover. differences.
Surface temperature has a regular daily vari-
ation and depends on the landscape. We com-
3. SOLAR RADIATION RESOURCE
pare the data to the predicted reference tem-
ASSESSMENT
perature image. Clouds again show up as local
differences. The most important parameter for the site
Weather patterns have a larger scale than pixel selection of solar thermal power plants is the
size. We allow for deviations from the pre- direct normal irradiation (DNI). The direct normal
dicted temperatures if they are common within solar irradiation on the ground is described by:
regions of pixels with similar surface prop-
DNI 5E ? t ? t ? t ? t ? t ? t (1) s d
0 R Ozon Gas WV Ae Cl
erties.
For corrections of the visible channel of
where E is the extraterrestrial irradiation. t ,
0 R
METEOSAT, we analysed 1 year of VIS data to
t , t , t , t , t are the transmittance
Ozon Gas WV Ae Cl
extract the distribution of counts with respect to
functions for Rayleigh scattering, ozone absorp-
the solar zenith angle and the angular distance
tion, absorption by uniformly mixed gas (oxygen
between sun and satellite seen from the surface. A
and carbon dioxide), water vapour absorption,
further correction is made to account for atmos-
aerosol extinction and cloud extinction, respec-
pheric inuences like forward and backward
tively. The formula is based on the clear sky
scattering within the atmosphere by dening and
model (Bird, 1984; Iqbal, 1983). For calculating
subtracting a minimum count from the satellite
the DNI in STEPS, we have modied that model,
values (Mannstein et al., 1999). VIS data is
adding a coefcient of transmission (t ) that
Cl
included into the decision process at locations,
takes into account the attenuation of irradiation by
where the cosine of the solar zenith angle is
clouds. We derive a cloud index from the visible
greater than 0.1 (the sun is more than 5.78 over
(VIS) and infrared (IR) image channel of the
the horizon). Similar to the IR analysis, we derive
METEOSAT weather satellite based on self ad-
a reference image, which is in this case not
justing, local thresholds which represent the spa-
variable throughout the day. The VIS images are
tial and temporal variation of the surface prop-
compared against the predicted image and the
erties (Mannstein et al., 1999). We look at long
previous image. The corrected count has to be
time series of METEOSAT IR-data to achieve a
higher then a threshold derived from the predicted
local temperature threshold which is close to the
cloud free scene.
temperature of the cloud-free surface. The refer-
Both IR and VIS information are combined to
ence temperature of land surface is described as a
the nal result by linear interpolation between the
function of time for every pixel:
expected cloud free value and a threshold for a
T 5a 1a cos x 2a 1sin a 3sin x 2a s s s d s dd
0 1 3 2 3
fully cloudy pixel (2408C in the IR and a
corrected count of 150 in the VIS channel). The 10.1 3sin x 2a (2) s dd
3
4
H
.
B
r
o
e
s
a
m
l
e
e
t
a
l
.
Fig. 2. North Africa determination of geographic frame conditions and non-suitable (black) areas.
Assessment of solar electricity potentials in North Africa based on satellite data and a geographic information system 5
higher of both values is given as the result. The from 15% to 215% with respect to the annual
interpolation scheme and the values of these sums of DNI (Broesamle, 1999). It must be
thresholds might be changed, if better validation mentioned that those data sets were obtained from
data are available. In a rst rough assumption, the rough estimates and did cover different time
cloud-index is converted by a linear function into intervals, so the comparison does not necessarily
a coefcient of cloud transmittance. show the quality of the satellite-derived data. All
Not only clouds, but also aerosols have an in all, the satellite data was more complete and
important inuence on solar radiation. We use the had a much better coverage and resolution than
Global Aerosol Data Set by Kopke et al. (1997) the available ground measurements.
to calculate the aerosol transmittance. This data For the analysis of North Africa, METEOSAT
set has a spatial resolution of 58358 and a images from 1998 were used for the calculation of
temporal resolution of two values per year (sum- the direct normal irradiation. STEPS calculates
mer and winter). We took the aerosol optical the DNI for every hour and every location (8760
thickness (AOT) for the wavelengths 0.5 mm and values for each location) with a spatial resolution
0.35 mm and for a relative humidity of 50%. of |535 km. These values are used by the power
These wavelengths are required by the Bird plant simulation module to determine the per-
aerosol transmittance function. We extended the formance and energy yield. The results were also
summer values from June until November, the used to create a map of the yearly sums of solar
winter values from December until May (Hess, direct normal irradiation for North Africa for
1998). 1998 (Fig. 4). A long term climatology of solar
The low resolution of the available aerosol data radiation was not used within this study.
has been considered critical for our application.
Therefore, aerosol transmittances were selected
4. SOLAR POWER PLANT PERFORMANCE
considering only those values relevant for the
main areas of interest (deserts and semi-deserts). In the rst prototype of STEPS, a solar power
Furthermore, aerosol transmittance values were plant simulation model was integrated that repre-
reduced by 20% for every 1000 m of altitude, sents a 200 MW parabolic trough solar electricity
taking into consideration the reduced atmospheric generating system (SEGS) in solar only operation
turbidity at elevated sites. Values for water vapour mode and without thermal energy storage (Fig. 5).
and ozone are taken from the NASAWater Vapour The model is made up of two parts, one that
Project (NVAP) and the NASA Total Ozone simulates the energy balance of the solar eld,
Mapping Spectrometer project (TOMS), respec- and a second that represents the conversion
tively. efciency of the Rankine steam cycle as a func-
The direct solar irradiation obtained by this tion of time. The module calculates the hourly
method shows good agreement (65% with re- thermal power output of the solar eld and the
spect to the annual mean) with data from selected electricity yield of the SEGS from the solar direct
sites derived from WMO-WRDC data, where DNI normal radiation generated in the meteorology
had to be calculated from global horizontal values module for each point of the map. For the
using empirical conversion models (Mannstein et simulation of the collector eld energy output, a
al., 1999). Measured direct radiation data is very simplied stationary model of the physical prop-
poor in the regions in question, so we had to erties and behaviour of the collector is applied.
compare our results to hourly time series from The physical parameters represent the LS-3
1998 measured in Almeria, Southern Spain. With parabolic trough collectors installed in some of
respect to this high quality data set, we observed the plants in California (Table 1). A detailed
errors of less than 65% for monthly sums of DNI description of the performance model and the
(Broesamle, 1999; Mannstein et al., 1999; Schil- related set of parameters can be found in
lings, 1999). On an hourly basis, the coincidence Broesamle (1999).
is very good for clear days, but unsatisfactory for A one-axis tracked parabolic trough collector
cloudy days, although the general daily pattern shows certain losses that depend only on its
and the daily sum of DNI is again represented geometrical structure and on the angle of inci-
quite well (Fig. 3). Other data sets displaying one dence. The following geometric losses are consid-
typical day per month in hourly resolution for ered in the model.
Taroudant and Ouarzazate (Morocco), Ouwairah Cosine losses represented by j consider the
COS
(Jordan), Tahrir (Egypt) and Tenerife (Spain) smaller active area of projection of the collec-
were compared to our results showing differences tor due to non-perpendicular irradiation.
6 H. Broesamle et al.
Fig. 3. Measured and calculated direct normal irradiation on clear and cloudy days in Almeria, Spain (Mannstein et al., 1999).
The ground measured data was kindly provided by Schlaich, Bergermann & Partner, Stuttgart.
The incident angle modier represented by the absorber tubes at the end of each collector
j considers the distortion of the reected row. End losses are described by the intercept
IAM
image of the sun at non-perpendicular incident factor j .
E
angles. Shading losses within the solar eld are de-
Collector end-losses are the portion of the scribed by the shading factor j .
S
sunlight that is reected outside of the range of The second important group of loss mechanisms
Fig. 4. North Africa annual sum of direct normal irradiation (1998).
Assessment of solar electricity potentials in North Africa based on satellite data and a geographic information system 7
Fig. 5. Basic solar thermal power plant conguration simulated by STEPS.
are the optical losses that occur by non ideal thermal losses of the hot collector elements during
reection and absorption of solar radiation. The operation. Losses caused by thermal convection
optical efciency is described by the: are in a rst approximation proportional to the
reectivity of the mirrors r, difference of the mean surface temperature of the
transmission factor of the mirror glass cover absorber tube T (653 K) and the ambient tem-
A
t , perature perceived by the absorber tube exposed
1
optical precision of the mirror surface (quality to the sunlight T (330 K), that both are
amb
factor) g, assumed to be constant during operation. Convec-
transmission factor of the glass tube that tion losses are quantied by the convection loss
2
surrounds the absorber tube t , factor U (W/ m K). Thermal radiation losses are
2
coefcient of absorption of the absorber tube described by a term that is proportional to the
a. difference of the same temperatures but to the
The third group of loss mechanisms considers the power of four. The intensity of thermal radiative
Table 1. Selected properties of the LS-2 and LS-3 parabolic trough collectors
LS-2 LS-3
Aperture 5.00 m 5.76 m
Length SCA (solar collector assembly) 48 m 99 m
Distance between rows 1215 m 1617 m
2 2
Reecting surface per SCA 235 m 545 m
2 2
Convection loss factor 2 W/ m K 2 W/ m K
Diameter of the absorber tube 0.07 m 0.07 m
Concentration ratio 72 82
Reectivity of mirror 0.93 0.93
Coefcient of absorption of absorber tube 0.94 0.96
Coefcient of emission of absorber tube 0.24 0.17
Coefcient of transmission of mirror 0.98 0.98
Coefcient of transmission of glass tube 0.95 0.96
Collector peak efciency 66% 68%
8 H. Broesamle et al.
emission losses is described by the coefcient of of cooling systems, that is dry cooling system,
emission of the absorber tube surface and the evaporation cooling tower and once-through cool-
2 4
Boltzmann constant s (W/ m K ). The simula- ing. The resulting net power is integrated to the
tion model yields the thermal power output of the net annual solar electricity yield E for each
year
solar eld according to the formula: point of the map (Fig. 6).
The simulation showed an acceptable accuracy,
~
Q 5A ? DNI
SF SF
representing well the geometrical effects of the
angle of incidence varying with time and place for p ? U
]]] F ? j h 2 ? T 2T s d
geo? opt A amb
all latitudes between 0 and 408 North and South.
C ? DNI
A comparison to measurements at the original
p ? ? s
4 4
]]] G 2 ? T 2T (3) s d
SEGS in California (Cohen et al., 1999; Dudley A amb
C ? DNI
et al., 1994) showed a very good qualitative and
with the collector area of the solar eld A , the
quantitative agreement with the actual physical SF
factor of concentration of the parabolic trough C,
behaviour of those plants (Fig. 7). The achieved
the direct normal irradiation DNI, the geometric
accuracy of the average daily energy yield of
efciency j 5j j j j and the optical
better than 65% was considered to be sufcient geo IAM S E COS
efciency h 5a r g t t (Table 1).
for the purpose of the rst prototype of STEPS. opt 1 2
The power block model is a very simple
Most of the parameters of the simulation model
formula that considers the nominal conversion
have been considered as constants. Further im-
efciency of the steam cycle, its part load be-
provements of the performance model will be
haviour and the parasitic losses of the power
achieved in the future by considering the in-
plant. The net electric power output of the plant is
uence of varying ambient and operation tem-
peratures. For example, ambient temperatures also
k
~
Q
SF
can be derived from satellite data (Broesamle,
~
]] P 5Q ?h ? 2P 2P
net SF nom Par,SF Par,PB F G
~
Q
1999) and processed by the GIS. Dynamic start-
SF,nom
up behaviour, thermal inertia of the solar eld,
(4)
wind effects, thermal energy storage and hybrid
fuel-solar operation modes are other topics for being h the nominal efciency of the power
nom
~
future enhancement of the simulation model. cycle and Q the rated nominal thermal
SF,nom
power output of the solar eld. The exponent k
describes the partial load behaviour of the power
5. COST ESTIMATE
cycle efciency. P and P represent the
Par,SF Par,PB
The electricity cost C of a solar thermal power parasitic electricity consumption of the solar eld
el
station operating in solar-only mode depends and of the power block, respectively. The
mainly on its investment cost I , the infra- parasitic losses are also a function of load. The
plant
structure cost for connecting the plant to roads model considers a minimum irradiation intensity
2
and the public grid I , the annual running of 200 W/ m for power block start-up. Efciency
inf
expenses of operation and maintenance C , the and parasitic losses are calculated for three types
O&M
2
Fig. 6. North Africa annual solar electricity yield per km of land area.
Assessment of solar electricity potentials in North Africa based on satellite data and a geographic information system 9
Fig. 7. Measured and calculated overall net solar-to-electricity efciency including parasitic losses of SEGS VI on July 1, 1997
(Cohen et al., 1999). The original measured radiation data was used as input to the model. A considerable difference between the
measured and the predicted efciency occurs during start-up, as the model neglects the thermal inertia of the solar eld and the
power block (lower graph). In the real plant, part of the energy used for heating up in the morning is recovered during the day
and in the evening. In this period, the model underestimates plant efciency. Therefore, this source of error compensates itself to
a large extent in the course of a day. As a consequence, the measured and calculated daily electricity yield differs by only 2%.
economic lifetime n, the mean capital interest rate distinction has been made between the different
i, and the net annual solar electricity yield E at economic environments of the countries of the
year
the respective site Maghreb. Especially the cost of infrastructure and
personnel will vary strongly from country to
n
i ? (1 1i )
country. Country-specic parameters (country
]]]] ? (I 1I ) 1C
n
plant inf O&M
(1 1i ) 21
]]]]]]]]]] C 5 (5)
el
E
year
Table 2. Sensitivity of the electricity cost of a 200 MW SEGS
a
in solar only mode (Broesamle, 1999)
The sensitivity of the electricity cost to the
Parameter varied by e.g. 1100% Relative C -
el
variation of selected input parameters is given in
variation
Table 2. Table 3 shows the most important
Annual direct normal irradiation 21.40
Annual cloud index 21.40
economic parameters used as reference for the
Geometrical and optical efciency 21.35
study on Northern Africa. Due to the inuence on
Overall investment 10.95
the power plant efciency, the different types of
Power block investment 10.94
Average interest rate 10.53
cooling systems require different sizes of the solar
Annual aerosol optical thickness 10.41
collector and the corresponding investment in
Thermal losses 10.31
order to yield the same rated power of 200 MW.
Insurance cost 10.08
Cost of operation and maintenance 10.08
Fig. 8 shows the procedure of calculating the
Mean salaries 10.05
infrastructure cost from the distances to the
Atmospheric water vapour 10.04
nearest road, grid and cooling water source de-
Atmospheric ozone 10.02
a
rived from the well known Digital Chart of the
Values indicate relative sensitivities. An absolute 100%
World (DCW). In this rst version of STEPS, no variation may not be realistic for some parameters.
10 H. Broesamle et al.
Table 3. Selected parameters of the economic model for Northern Africa
Type of power plant Parabolic trough, LS-3
Capacity 200 MW
Solar collector aperture area
2
using a dry cooling system 1.228 km
2
using an evaporation cooling tower 1.124 km
2
using a once-through cooling system 1.075 km
Required land area (cost free) |3 times solar collector area
Plant investment (I )
plant
using a dry cooling system 460 Mio USD
using an evaporation cooling tower 420 Mio USD
using a once-through cooling system 405 Mio USD
Infrastructure costs (I )
inf
per km road 185,000 USD
per km high tension grid 125,000 USD
per km pipeline (once-through) 2 Mio USD
per km pipeline (evaporation) 305,000 USD
Operating costs (C )
O&M
Personnel 2.7 Mio USD per year
Operation and maintenance 1% of investment per year
Insurance 1% of investment per year
Economic lifetime (n) 25 years
Average interest rate (i ) 8%
data base) and a more detailed calculation of the tool can be used for the assessment, evaluation
economic performance will be integrated in a and ranking of sites of solar thermal power plant
future version of STEPS. projects, giving project developers, governments,
The resulting solar electricity cost is displayed intergovernmental institutions and other decision
in Fig. 9. Figs. 7 and 9 show the large technical makers a well founded basis for planning and
and economic potential of solar power generation designing the build out of solar power capacity
in North Africa. Theoretically, on less than 1% of world wide. The tool has been applied successful-
the suitable area in North Africa, the total 1997 ly on North Africa, showing the large technical
world electricity demand of 12,000 TWh/ year and economical potential of solar thermal power
could be generated at a cost of less than 12 in this region.
cents/ kWh (price level 1998, solar only opera- It must be pointed out that the results of the
tion, see also Table 4). According to the remain- rst prototype of STEPS shown here should not
ing potential of cost reduction of solar thermal be used directly for investment decisions, as they
power technology, this cost will come down only represent one possible power plant congu-
within a decade to less than 6 cents/ kWh (Ener- ration and are based on low resolution and
modal Engineering Ltd, 1999; Trieb, 1999). With partially insufcient input data (e.g. no radiation
available support from the World Bank and in climatology was used). They are primarily meant
hybrid operation mode, a competitive electricity for demonstrating the concept.
cost can already be achieved today. For in-depth analysis of a countrys solar
thermal power potential and for detailed site
ranking on a feasibility study level, STEPS is
presently enhanced by an extensive country data
6. CONCLUSIONS
base, further GIS data on natural and political
STEPS enables the computer-based assessment
risks, soils, hydrology and natural reserve areas, a
of solar irradiation, geographic frame conditions,
more sophisticated and variable performance
requirements of infrastructure and the expected
model, a more accurate method of cloud and
electricity potential and cost of solar thermal
aerosol assessment, a long term radiation
power stations for large regions, providing results
climatology and other issues. With the recent
with high spatial and temporal resolution. The
Space Shuttle Radar Topography Mission, higher
A
s
s
e
s
s
m
e
n
t
o
f
s
o
l
a
r
e
l
e
c
t
r
i
c
i
t
y
p
o
t
e
n
t
i
a
l
s
i
n
N
o
r
t
h
A
f
r
i
c
a
b
a
s
e
d
o
n
s
a
t
e
l
l
i
t
e
d
a
t
a
a
n
d
a
g
e
o
g
r
a
p
h
i
c
i
n
f
o
r
m
a
t
i
o
n
s
y
s
t
e
m
1
1
Fig. 8. Calculation of infrastructure costs from the distances to roads, grids and cooling water.
12 H. Broesamle et al.
Fig. 9. North Africa solar electricity cost per kWh (200 MW SEGS, solar only, price level 1998).
Table 4. Ranking of areas by the solar thermal power gene- Cohen G. E., Kearney D. W. and Price H. W. (1999) Per-
a ,b
ration cost per kWh in North Africa formance history and future costs of parabolic trough solar
electricity systems. In Proceedings of the 9th International
Power Available Potential
Symposium on Solar Thermal Concentrating Technologies,
generation area in of power
2 Odeillo, France, 1998, J. Phys. IV, EDP Sciences.
costs in 1000 km generation
Dudley V. E., Kolb G. J., Mahoney A. R., Mancini T. R.,
cents/ kWh in TWh
Matthews C. W., Sloan M. and Kearney D. (1994). Test
#12 297.1 37,994
Results of SEGS LS-2 Solar Collector, Sandia, Albuquer-
1213 1107.9 138,047
que, Sandia report SAND94-1884.
1314 2999.4 339,939
Enermodal Engineering Ltd (1999). Cost Reduction Study for
1415 3896.9 412,603
Solar Thermal Power Plants, World Bank/ GEF, Washington
1516 1986.3 203,575
DC.
1617 1340.2 135,973
Hess M. (1998). Personal correspondence.
$17 980.6 93,425
Iqbal M. (1983). An Introduction to Solar Radiation, Academ-
Total 12,608.4 1,361,556
ic Press, Toronto.
a
Kopke P., Hess M., Schult I. and Shettle E. P. (1997). Global Total size of the analysed area in North Africa was 14.3
2
Aerosol Data Set, Max Planck Institute for Meteorology, Mio km .
b
Hamburg, Report no. 243. Suitable area with respect to land cover and land slope was
2
12.6 Mio km . Mannstein H., Broesamle H., Schillings C. and Trieb F. (1999)
Using a METEOSAT cloud index to model the performance
of solar thermal power stations. In Proceedings of the
EUMETSAT Meteorological Satellite Data Users Confer-
resolution topographic data is now available for
ence, Copenhagen.
more detailed siting purposes.
Olson J. S. (1994) Global ecosystem framework-denitions. In
USGS EROS Data Centre Internal Report, USGS EROS
Data Centre, Sioux Falls, SD, 37 S.
Schillings C. (1999). Calculation of the Direct-normal Ir-
REFERENCES
radiation for the using in Solar Thermal Power Stations.
Application of a Physical Clear-sky Model using Satellite
Bird R. E. (1984) A simple solar spectral model for direct-
Data, University of Bonn, Masters thesis.
normal and diffuse horizontal irradiation. Solar Energy
Trieb F. (1999) Competitive solar thermal power stations until
32(4), 461.
2010 the challenge of market introduction. In 9th
Broesamle H. (1999). Solar Thermal Power Stations. Locali-
International Symposium on Solar Thermal Concentrating
zation and Assessment of the Potential with the Planning
Technologies, Odeillo, France, 1998, J. Phys. IV, EDP
Tool STEPS, University of Vechta, Doctoral thesis; in
Sciences.
German.