A thermoelectric solar plant is mainly composed of the following components: a solar collector concentrator in which the collection and the concentration are performed through the reflection or diffraction of the light; the absorber, which absorbs light and transfers heat to a thermal fluid; a heat storage system; a steam generator sys- tem; and a conventional system of thermal energy conversion into electricity (Figure 1). Basically, in the fol- lowing four concepts, the collector sections can be differentiated: Figure 1(a) and Figure 1(b) are linear and bi- dimensional concentrators, respectively, and Figure 1(c) and Figure 1(d) are tri-dimensional ones. Concentra- tors Figure 1(b) and Figure 1(d) have reflective components known as heliostats in which each mirror reflects the light for a linear or volumetric recipient, while concentrators Figure 1(a) and Figure 1(c) are image formers.

The recent evolution of solar thermoelectric plants began in the 1980s and can be summarised in the following highlights [1] :

1980s

· 1981—The solar towers known as SOLAR I (10.0 MW, EUA), CESA I (1.2 MW Spain), THEMIS (2.5 MW, France), EURELIOS (1.0 MW, Italy) and NIO (1.0 MW, Japan) were connected to an electric network.

· 1984—Solar Electric Generating System I (SEGS), a thermoelectric solar plant with a parabolic cylinder concentrator with 13.8 MW power, starts operating.

1990s

· 1990—Nine SEGS-type plants were implanted in commercial bases (SEGS 1, 13.8 MWe; SEGS 2 to 7, 30 MWe and SEGS 8 and 9.80 MWe), totaling 354 MWe.

· 1991—The main developer of SEGS, Solar Light, declares bankruptcy.

· 1996—The solar tower, SOLAR II (Barstow, USA), a 10 MW plant, starts operation and exhibits storage viability with molten salts.

· 1990-2000—Despite the interruption of the installation of large commercial systems, basic research continued.

2000s

· 2004—A Parabolic Disc with a Stirling engine of 150 kW was implemented at Sandia Laboratory, the Spanish government implements a decree, stimulating the purchase of 200 MW of thermoelectric solar energy with a guaranteed fee.

· 2006—Installation of the P&D (1 MW) plant of cylinder parabolic technology in Saguaro, USA.

· 2007—Implementation in Spain of the commercial plant tower type PS10 and Nevada I of 60 MW with cylinder parabolic technology.

Start of the commercialization growth

· 2008—Andasol I (Spain, 50 MW), the cylinder parabolic power plant, is commissioned; also, Kimberlina (EUA, 5 MW), the first thermoelectric solar plant with a Fresnel linear concentrator, was installed.

· 2009—In Spain: the PS20 (20 MW solar tower), cylindrical parabolic power plants in Puertollano, Andasol II and La Risca were built, each of 50 MW and Linear Fresnel technology PE1 of 1.4 MW. In the U.S., the Sierra Sun Tower (5 MW) was built.

· 2012—The cumulative capacity of solar thermal power plants installed worldwide was 2236 MW, with the majority of the plants using cylindrical parabolic concentration technology (95%).

The plants of the SEGS (Solar Electric Generation Systems) were installed at the end of the last century, representing the largest (354 MWe) and most successful example of installed solar thermal technology using cylindrical parabolic concentrators. The nine SEGS plants were built at three different places in the Mojave Desert in California (USA), between 1984 and 1991, and continue to be in commercial operation, demonstrating its technical and commercial reliability.

After a break in the installation of such systems for approximately 15 years, there was a vigorous resumption of this technology in the mid-2000s. In 2012, the cumulative amount of solar thermal power plants located worldwide was of 2236 MW, with the majority being of cylindrical parabolic concentration technology (95%). Estimates of the thermal solar plants under construction or announced publicly are discrepant due to the use of different criteria, such as different period consideration and the absence of updating the changes in the project in terms of the power produced. According to [2] , approximately 2590 MW were under construction and 4508 MW were announced in the U.S., 1080 MW in Spain, and 4386 MW in the rest of the world. Carefully considering these numbers, it is reasonable to say that the pace of growth, the rate of accumulation of experience, and the scale gains in the coming years will be spectacular. Therefore, based on the learning curve prepared by [3] (Figure 2), it is likely that in the next 10 years, the cost of solar thermal energy would be equal to the cost of energy from the conventional network.

The state of Minas Gerais has used almost all of the large water resources for hydroelectric power generation. Nevertheless, the options of energy imports, the use of other local renewable sources (PCH, wind and solar) or the creation of conventional thermal power plants (with fuel oil, gas, coal, or nuclear) remain undeveloped. The main barriers to importing energy or creating nuclear and conventional thermal power plants are as follows: 1) Amazon hydroelectricity import costs are much higher than the generation itself, besides the environmental problems that add uncertainty to the activity; 2) fossil fuels that have increasing prices; 3) environmental problems and “invisible” subsidies (externalities), which are increasingly questioned by society; and finally 4) in the case of nuclear energy, the risk of catastrophic accidents.

With the prospect of technological and commercial maturity of large-scale solar thermoelectric plants in the 2020s, it would be appropriate for the electrical system in Minas Gerais to follow the evolution of this technology, to perform R&D in detail, and to evaluate the potential of the solar resource available in its territory. In addition to its renewable aspect and its low environmental impact, the installation of solar thermal power plants is important regionally, due to the following aspects: a) it improves the “mix” of the Brazilian electricity generation sector and thus increases the security of the electrical system; b) the solar resource is the strongest in the dry period in the region, and therefore, it has complementary characteristics to the regional hydroelectric system; c) the solar resource is the maximum in the highest consumption period (summer); d) it has a characteristically distributed generation (another safety factor of the electrical system); e) it can be installed in semiarid regions with low population density and non-competitive lands for other nobler uses (for example, agriculture); and f) it generates employment and income.

The use of GIS in renewable energy that began in the 1990s went through considerable progress, and, as a result, various decision support tools were developed [4] . The pioneering work regarding the use of GIS for CSP (Concentrated Solar Power) is due to [5] who analysed Northern Africa, providing a ranking of sites with respect to the potential and the cost of solar thermal electricity for a particular power plant configuration. Recently, this type of study has become widespread: [6] for the Southwest of the USA, [7] for South Africa, [8] for Oman, [9] for Burkina Faso, [10] for Australia [11] for India, and [12] based on the combination of a Geographic Information System (GIS) and multi-criteria decision making (MCDM), for Spain.

The assessment and identification of site using GIS are revolutionising the decision-making process in the world and decisively contribute to the rapid growth of the CSP technology implementation. Thus, in this work, the identification and mapping of the most promising places in Northeastern Brazil was performed so that the country may soon start the process of implementing CSP solar technology on a large scale.

Spatial data in GIS are composed of graphical and descriptive data. The former are responsible for the graphical records, which correspond to definite locations in the real world, and they can be stored or represented in the formats of a vector or matrix files (raster). The vector structure foresees the existence of three distinct spatial bodies represented in the geographic primitives: point, line and polygon. The raster, however, is a numerical structure represented by stored images in matrix format. The matrix is composed of a set of pixels (cells) disposed in lines (x) and columns (y). The matrix position (x, y) defines the image coordinates and exhibits an associated grey level.

The descriptive data are structured by sequential lines (records) or indexed files and include the attributes that describe the bodies of these objects. The descriptive data qualify the graphic data by means of attributes and are, generally speaking, stored in database table structures and controlled by the Database Managing System (DBMS).

The spatial data have four inseparable characteristics [13] : 1) geographic position; 2) associated attributes; 3) topologic relations; and 4) temporal reference. The geographic position responds to the “where is it?” question, and the attribute is a spatial data component that aims to describe and characterise the graphic feature, responding to the “what is it” and “what does it refer to” questions; the topologic relations describe the spatial relationships of the object and its peripheries, and the temporal reference involves the data behaviour assessment throughout time.

The GIS applications are usually in the form of two large spatial data: a) Geofields, which are used to represent spatial distributed values, such as, soil type, topography and mineral content; and b) Geographic objects, which are individualised and have a specific identification. The geographic objects (or geo-objects) have non- spatial attributes stored in a database. These objects might be associated with several graphic representations, such as schools, municipalities and farms.

The objective of the spatial analysis is to measure properties and relationships considering the spatial location of the studied phenomenon [15] . One of the main types of spatial analysis is field algebra. Field algebra (also known as map algebra) consists of a spatial analysis procedure set in GIS, which produces new data from the manipulation functions applied to it. The elements of the field algebra operate on maps, associating each place of a determined studied area to a quantitative or qualitative value. The maps are treated as individual variables, and the calculations defined on those variables are applied homogenously in all points of the map. According to [16] , these calculations may be grouped into three classes: punctual, neighbours, and zone.

The punctual calculations result in numerical or thematic geo-fields, whose values are a function of the associated values to the same place by one or more representations of the other geo-fields. These calculations occur over maps (scanning of a numerical model of the land, for example) or over spatial sets (Boolean operations, for example), whether mathematical or transformational (pondering, scanning, among others).

The neighbor calculations act on a geo-field influenced by the neighbour dimension and shape (mask). During the calculation, the mask dislocates, applying to the geo-field the values of the attribute delimited by the mask. In the computational environment, the most used mask shape is that of a cell matrix. The spatial filtering of the geo-field, using image representations and slope calculations from an altimetry geo-field, is an example of neighbour operations using a mask in the cell matrix format [16] .

In the zone calculations in field algebra, the value of each geographic position of the geo-field destination depends on the value of the attribute in all of the geographic positions that comprise the region in the initial geo- field. Unlike the neighbour transformations, where each geographic position had its own neighbour, represented by a mask that dislocates over the data, in the zone transformation, the regions are static and do not dislocate over the geographic studied region. Examples of such calculations are the calculations of zone majority and zone diversity.

A solar plant, whether thermal or photovoltaic, with concentration requires high incidence of direct normal solar irradiation because it uses (almost exclusively) only the direct portion of the solar irradiation falling onto its surface. According to Equation (1), the solar irradiation incident on the collection plane of the concentrator (I_c) is determined by the direct normal component (I_b) plus the diffuse component (I_d) divided by the concentration (C). Then, in the solar plant where there is a concentration ratio of 50 < C < 100, which is typical, for example, of the commercial plant of cylindrical parabolic technology, the contribution of the diffuse component is negligible, that is, I_b - I_c. Therefore, to determine the best location of these plants, it is essential to characterise the direct normal solar irradiation.

For low concentration systems or plants, the direct normal solar irradiation is less critical, but even so, the prerequisite is still a good level of global solar irradiation, which usually occurs in regions where there is also too high incidence of direct solar irradiation.

A solar plant for the production of solar thermal electric is usually a high concentration system of solar irradiation, and therefore, favorable sites for their installation must have a solar irradiation >2100 kWh/(m²∙year), or >5.75 kWh/m²∙day annual average. For flat photovoltaic systems (without concentration), the requirement is lower: the global solar irradiation should be >2000 kWh/(m²∙year), i.e., >5.5 kWh/m²∙day annual average. These values are slightly higher than the ones usually used for the preparation of feasibility studies for the installation of photovoltaic or thermoelectric plants in the world.

The state of Minas Gerais has vast regions fulfilling the aforementioned criteria. In Figure 5, it can be seen that half of the state, more precisely the entire western side, has a daily global solar irradiation average in the range of 5.5 to 6.5 kWh/m²∙day. For the daily normal direct solar irradiation, the annual average for the entire western side of the state satisfies this condition, as shown in Figure 6.

The cylindrical parabolic concentrators require relatively large areas. A solar power plant consisting of 80 MWe

requires approximately 500,000 m² (without considering the thermal storage) to accommodate the arrangement of the collectors. The scaling factor derived from the experiences of the construction of solar power plants indicates that it is advantageous to install several adjacent plants, thereby forming a solar park. Thus, the minimum area availability could be 2 - 8 km² for successive modular installation of these plants.

The topography of the land where the solar plant will be located determines the acceptability of the site according to its impact on the relative cost for the preparation and levelling of the land. This location must be as flat as possible (with slopes less than 2%) but sufficient to allow the natural drainage of the land.

Regarding the visual horizon of the field of collectors, only obstructions (such as hills, trees, towers, or others) whose visual angles are less than 100 are allowed.

In Figure 7, the slopes in the state are shown. Continuous regions of low slope (<3%) can be identified (“patches”) in the Northwest of Minas Gerais, in the region of Januária, and in Janaúba and Capão Redondo. The solar photovoltaic plants are much less demanding in terms of slope because they are more modular. The slopes in the range of 3% - 8% are acceptable; thus, some additional regions in the Northwest, the Central Western region of Belo Horizonte, and the extreme West satisfy such prerequisites.

In the implementation of a solar plant, permits and any restrictions on land use should be considered. Permissions relate to the issues of the contractual relationship between the landowners and the project developers, while the restrictions refer to the intended use of the ground. In this regard, it is noted that in urban areas or urban expansion, Environmental Protected Areas (EPAs, Figure 8), areas of food production (arable, Figure 9), and former slave (quilombolas) and indigenous territories, for example, are considered unsuitable regions for installing solar plants. The dry and arid regions are considered typical for the installation of these plants.

AGRICULTURAL POTENTIAL MAP

Natural disasters, such as earthquakes, strong gusts of wind, sand storms, hail, lightning, and other factors, are potential damaging factors or disruptive factors to the operation of solar power plants. Moreover, soil quality should also be considered.

The requirements for connection to the electrical grid for a solar plant that uses parabolic cylindrical collectors are similar to other steam power plants. A plant that produces 80 MWe of power, for example, must have 230 kV of transmission lines. In addition to charge capacity, another important aspect should be considered: the distance between the solar plant and transmission lines. The actual costs with the construction of new transmission lines are generally very high and depend on the voltage level of the line and its length. Thus, the solar plant must be positioned as close as possible to the transmission lines. Figure 10 shows the transmission lines that cross the state of Minas Gerais.

A 50 MWe parabolic cylinder power plant operating 350 days a year and 10 hours per day, for example, uses approximately 500,000 m³ of water or 1500 m³/day. This water is required for the cooling towers (90%), steam generation in a power cycle (8%), and the cleaning of the mirrors (2%). The typical flow for the cooling tower is 320 m³/h. The water must also have adequate quality to prevent oxidation and incrustations of the equipment.

In general, the state of Minas Gerais is well supplied with water. Consequently, this aspect will not be restrictive for the installation of large solar power plants. Note that the solar photovoltaic plants require only a small amount of water, mainly for cleaning of the modules and for removing the dust and other contaminants.

Fuels or other backup energy sources are required for the hybrid operation of the solar plant (solar resource + fuel). In solar cylindrical parabolic plants abroad, natural gas is used as a backup fuel. In the semiarid region of Minas Gerais, recent announcements indicate a good occurrence of gas, which is very promising.

Access to the site is relevant to transport large and fragile equipment (glass mirrors). The criteria for classifying the access are the widths of the roads, the road surface quality and the possibilities for manoeuvring large vehicles. According to the Department of Highways of Minas Gerais the state has 8957 km of federal highways (only 667 km unpaved) and 26,604 km of state highways (7238 km unpaved, Figure 11). As far as the rail network is concerned, Minas Gerais has 5080 km of railways, and the major companies operating in the sector are Centro-Atlântica, MRS Logistica S.A., and Estrada de Ferro Vitoria a Minas. Furthermore, Minas Gerais has navigable rivers, such as the Paranaíba, Paracatu, Velhas, Paraopeba and São Francisco rivers.

Considering the relevant aspects mentioned above, and taking as a reference the limits of micro-regions defined by the IBGE, it can be said that the classification of the most promising areas in the state of Minas Gerais are those shown in Figure 12: