- Email: [email protected]
Spatial analysis methodology applied to rural electrification
ARTICLE IN PRESS
Renewable Energy 31 (2006) 1505–1520 www.elsevier.com/locate/renene
Spatial analysis methodology applied to rural electrification J. Amadora, J. Domı´ nguezb, a
Department of Electric Engineering, EUTI, UPM, Ronda de Valencia, E-28012 Madrid, Spain b Renewable Energies Division, CIEMAT, Av. Complutense 22, E-28040 Madrid, Spain Received 21 March 2005; accepted 3 September 2005 Available online 11 November 2005
Abstract The use of geographical information systems (GISs) in studies of regional integration of renewable energies provides advantages such as speed, amount of information, analysis capacity and others. However, these characteristics make it difficult to link the results to the initial variables, and therefore to validate the GIS. This makes it hard to ascertain the reliability of both the results and their subsequent analysis. To solve these problems, a GIS-based method is proposed with renewable energies for rural electrification structured in three stages, with the aim of finding out the influence of the initial variables on the result. In the first stage, a classic sensitivity analysis of the equivalent electrification cost (LEC) is performed; the second stage involves a spatial sensitivity analysis and the third determines the stability of the results. This methodology has been verified in the application of a GIS in Lorca (Spain). r 2005 Elsevier Ltd. All rights reserved. Keywords: GIS; LEC; Geographic distribution; Spatial sensitivity analysis; Renewable energy sources; Rural areas
1. Introduction Usually, in rural electrification geographical information system (GIS), the allocation of the potential of various technologies in the studied area is determined by comparing their LEC values [1–5]. Each isolated household will ‘‘belong’’ to the technology offering the lowest LEC in the point corresponding to that household. In this way, the potential of each Corresponding author. Tel.: +34 91 3466041; fax: +34 91 3466604.
E-mail address: [email protected] (J. Domı´ nguez). 0960-1481/$ - see front matter r 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.renene.2005.09.008
ARTICLE IN PRESS 1506
J. Amador, J. Domı´nguez / Renewable Energy 31 (2006) 1505–1520
technology in the studied area can be expressed by the number of households, or by the surface area in which such a technology offers the best cost per kWh throughout the life of the installation. But the real potential of each technology involved will be determined not only by minimizing the cost per kWh but also by the difference between its own costs and those of the other technologies which are in most direct competition with it. Obviously, a technology’s potential does not depend only on its cost per kWh, but also on other matters related to the problems of regional integration of renewable energies [6]. Furthermore, the influence that various parameters may have on these costs will be of vital importance, since a slight variation in a parameter might change the results significantly. For these reasons, a spatial sensitivity analysis methodology is proposed to determine the influence of various parameters on the results, linking them clearly with the initial variables. This method is structured in three stages: 1. Determination of the main influencing parameters by a sensitivity analysis of each LEC technology. 2. Spatial sensitivity analysis of the potential of the studied zone with respect to the parameters found in the previous stage, selecting the most significant variables in the distribution of the rural electrification potential. 3. Study of the spatial behaviour of the variables of the previous stage in order to determine the ‘‘stability’’ of the result obtained. The following points describe the application of this method to the rural electrification study of Lorca (Murcia, Spain) carried out by a GIS developed by the authors [7] from an initial version of Solargis [8–11], GIS prepared by several Research Institutions in the framework of the JOULE II Program of the European Union. 2. First stage: sensitivity analysis of LEC The method consists of applying a classic procedure considering the risk in the economic analysis of the investment selection: an analysis of the sensitivity of project costs with respect to variations of the technical and economical parameters. This type of analysis is conducted by establishing the value of all the parameters except the one being considered, which it is made to vary about a central value, usually the ‘‘best’’ estimate of that parameter. Figs. 1–6 show the results of the sensitivity analysis for each technology considered in the GIS. In these graphs, the influence of the variation of the parameter values with respect to the reference value can be observed. It must be taken into account that the slopes of the resulting curves cannot be compared directly, since the meaning of the variation of each parameter is different, as well as the probability that each variation will occur. As a rule, the causes of variation of the various parameters considered can be grouped into four categories:
Uncertainty in the establishment of the renewable resources, in this case solar radiation and wind speed. Evolution of the technology, causing changes in the efficiency and equipment lifetime. Changes in the socio-economic scenario that affect consumption, load density and discount rate, among other parameters.
ARTICLE IN PRESS J. Amador, J. Domı´nguez / Renewable Energy 31 (2006) 1505–1520
1507
1.5 Solar radiation 1.4 Battery life 1.3
lecpv
System efficiency 1.2 System investment 1.1 Consumption
Discount rate Efficient receivers cost
1
System life
0.9
0.8 0.5
1
1.5
Fig. 1. Photovoltaic LEC sensitivity. This graph shows the value variation influence of the different parameters regarding a reference value. The most significant parameters in the PV system are renewable resource, efficiency and battery.
1.5 1.4
Battery life
Battery Autonomy
1.3 Wind speed 1.2
System life
lecwt
Efficiency
System investment
1.1 Consumption
Discount rate
1
Turbine high
0.9 0.8 0.7 0.6 0.5
1
1.5
Fig. 2. Wind LEC sensitivity. This graph shows the value variation influence of the different parameters regarding a reference value. The most significant parameters in the wind system are renewable resource, and battery. In the study area, wind resources are not relevant.
ARTICLE IN PRESS J. Amador, J. Domı´nguez / Renewable Energy 31 (2006) 1505–1520
1508
1.5 Consumption
1.4
lec diesel
1.3 1.2
O&M cost
System life
1.1
System investment
Battery life
Diesel price 1
Discount rate
0.9 0.8 0.7 0.5
1
1.5
Fig. 3. Individual diesel LEC sensitivity. This graph shows the value variation influence of the different parameters regarding a reference value. The most significant parameters in the domestic diesel system are consumption and costs.
Changes in the energy market, specifically regarding equipment investment costs, fuel price and cost per kWh. The margins of variation considered with respect to the case of reference have been:
Renewable resources: 725%. Demand: 750%. Efficiencies: moderate variations. Investment costs, fuel and kWh prices: large variations.
The study of Figs. 1–6 (for greater clarity of the graphs, the parameters with the smallest influence are not shown), considering the probability of variation of the various parameters, allows one to determine the ones with the greatest influence on each corresponding LEC technology (6–8 parameters have been selected per technology). 3. Second stage: spatial sensitivity analysis 3.1. Spatial sensitivity graphs A ‘‘spatial sensitivity graph’’ is determined for the parameters selected in the first stage (a graph of this type is shown for demand [4]). It expresses the variation of the potential for rural electrification with each parameter of the studied area. The vertical axis of these
ARTICLE IN PRESS J. Amador, J. Domı´nguez / Renewable Energy 31 (2006) 1505–1520
1509
1.7 1.6 Load density
1.5
lec electric net
1.4 1.3 MT length 1.2 1.1 MT investment 1 0.9 Discount rate
Consumption
0.8 0.7 0.5
1
1.5
Fig. 4. Connection to grid LEC sensitivity. This graph shows the value variation influence of the different parameters regarding a reference value. The most significant parameters in the grid-connected system are load and consumption.
2
1.8 Consumption
lec central diesel
1.6
1.4 System investment 1.2
O&M cost Diesel price
System life Load density
1
BT line investment Discount rate
0.8
0.6 0.5
1
1.5
Fig. 5. Central diesel LEC sensitivity. This graph shows the value variation influence of the different parameters regarding a reference value. The most significant parameters in the central diesel system are consumption and cost.
ARTICLE IN PRESS J. Amador, J. Domı´nguez / Renewable Energy 31 (2006) 1505–1520
1510
2
1.8 Wind speed
lec wind-diesel
1.6
1.4 Consumption
1.2
Diesel inversion
Turbine efficiency 1
Load density Discount rate System life
0.8
0.6
0.5
1
1.5
Fig. 6. Wind–diesel LEC sensitivity. This graph shows the value variation influence of the different parameters regarding a reference value. The most significant parameters in hybrid system are consumption and renewable resource.
graphs represents the potential of each technology as a percentage (e.g., the number of households corresponding to that technology) and the horizontal axis represents the variation of the various parameters. These graphs are calculated from AML programs (ArcInfoTM programming language) that permit applying GIS to different values of the variables. The application of spatial sensitivity graphs allows ordering the variables selected in the first stage according to their influence on the result. As a rule, these variables can be classified into three influence groups: great, moderate and negligible. The result obtained for Lorca is the following:
Parameters with a great influence: J demand, J storage life, J investment cost of photovoltaic system, J investment cost of central diesel system, J diesel price, J solar radiation. Parameters with a moderate influence: J investment cost of LV lines, J photovoltaic system efficiency, J electrical grid rates, J CO2 tax, J discount rate,
ARTICLE IN PRESS J. Amador, J. Domı´nguez / Renewable Energy 31 (2006) 1505–1520
1511
Parameters with a negligible influence: J wind resource, J investment cost in HV lines, J investment cost in the wind system, J autonomy days of the photovoltaic system storage, J autonomy days of the wind system storage, J others.
The variables in the first group should be analysed individually. Conclusions are extracted from the spatial sensitivity graph for the study considered and, in consequence, one or more meaningful cases are chosen. Maps and numerical data are obtained for these cases, which are then compared to those of the reference case. For the two variables with the greatest influence a complete analysis is performed. The spatial sensitivity graph is shown considering only the four variables with the greatest influence. 3.2. Parameters with a great influence on the result 3.2.1. Demand Growing demand produces a drastic decrease of photovoltaic systems, losing competitivity strongly with respect to central diesel and slightly with respect to connection to the grid. Once photovoltaic systems have reached 10% of the regional potential, diesel domestic systems can compete with them; this is, for areas with low load densities far from the grid. Another important result is that connection to grid is stabilized at 30%. This is the only parameter that can lead to the disappearance of the renewable technologies potential. Demand decreases will significantly affect to the potential distribution of each technology. A 10% decrease produces a competitive loss for central diesel of 25%, but only a 5% loss for connection to the grid. This trend continues as the demand decreases, so that central diesel ‘‘disappears’’ and ‘‘connection to grid’’ loses its competitiveness slowly. The potential of wind systems is not affected by the drop in demand. To complete the spatial sensitivity graph information, a map showing the distribution of the technologies and a table of numerical results (corresponding to the case of a demand 40% higher than the reference demand) are provided (Fig. 7). 3.2.2. Storage life This parameter powerfully influences the result due to the cost of storage, representing a very important percentage of the investment cost, particularly for renewable energy facilities. A lifetime 20% higher increases the photovoltaic potential by approximately 30% compared to central diesel. Higher values would mean that photovoltaic systems exceed a 75% rural electrification potential, which would imply the near disappearance of central diesel. From this moment, onward photovoltaic systems continue to grow, albeit more slowly due to the fact that there are areas with high load densities. These results show the importance of a good system design and maintenance, particularly for the storage system. It also justifies the use of certain regulators that charge the batteries in an efficient and secure way, as well as the use of ‘‘tubular’’ batteries.
ARTICLE IN PRESS 1512
J. Amador, J. Domı´nguez / Renewable Energy 31 (2006) 1505–1520
Fig. 7. Spatial influence demand in the Lorca area (spatial sensitivity graph, map and numerical results for the case of reference and for a demand 40% higher). Demand is the most relevant parameter in the study area because PV system is the best renewable option.
The spatial sensitivity graph information is completed with the technology distribution map and the numerical results table, corresponding to a lifetime 40% greater than in the case of reference (Fig. 8). 3.2.3. Investment cost of the photovoltaic system The greatest variations of the photovoltaic system take place about 8000 h/kWp, specifically to 6000 and 10,000 h/kWp, where the curves defining the potential change
ARTICLE IN PRESS J. Amador, J. Domı´nguez / Renewable Energy 31 (2006) 1505–1520
1513
Fig. 8. Spatial influence of the storage lifetime in the Lorca area (spatial sensitivity graph, map and numerical results for the case reference and for a 6-year storage lifetime). In addition to demand, battery life is very important in order to improve the renewable rate in Lorca.
their slopes. Therefore, a 9000 h/kWp value is considered located within the areas which change most. An investment cost of 6000 h/kWp causes the specific weight duplicity of photovoltaic potential, corresponding to a massive installation program of the municipality electrification technology.
ARTICLE IN PRESS 1514
J. Amador, J. Domı´nguez / Renewable Energy 31 (2006) 1505–1520
3.2.4. Investment cost of the central diesel system A complete observation of this graph (Fig. 9) and the previous one shows the direct competition of the photovoltaic and central diesel systems; connection to grid is maintained with a percentage always near one-third of the total potential, while wind systems have a symbolic participation, although practically inalterable in normal conditions, around 0.7%. 3.2.5. Fuel price Fuel price shows a similar influence as that of the previous parameters: competition between photovoltaic and central diesel systems, with connection to grid and wind systems maintained, their participation being independent of parameter variations. Furthermore, in this case the value taken as reference, 0.4 h/l (it is assumed that a rural area will have access to agricultural type gas–oil), is represented in the area of the curves with a sharp slope. This means that fuel price variations will have a greater effect in this stage than in others. Although the current trend implies an increase of prices, it is not possible to establish the absolute stability of this trend [12]. 3.2.6. Solar radiation The variation of solar radiation due to the competition between photovoltaic and central diesel affects the results similarly; otherwise, the connection to grid and wind system 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 3000
11000
100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 550
0.7 0.8 0.9
100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 0.75
5000 7000 9000 Fotovoltaic system investment
100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 0.1 0.2 0.3 0.4 0.5 0.6
Diesel price (Euros/litre)
1
770 990 1210 1430 Central diesel investment (Euros/kW)
0.85
0.95 1.05 Solar radiation (0/1)
1.15
1.25
Fig. 9. Spatial sensitivity graph in the Lorca area: parameters with great influence. Next to demand and battery life, others parameters such as system investment, price of resources are also important for renewable energy integration, considering the great importance of PV system for this area.
ARTICLE IN PRESS J. Amador, J. Domı´nguez / Renewable Energy 31 (2006) 1505–1520
1515
remain constant. Again, the value considered (in this case, being a spatial variation parameter, the values are represented in 0/1) is represented in the centre of the section with the greatest variation, oscillating 715% with respect to the reference value. This graph (Fig. 9) shows the convenience of an accurate knowledge of the solar radiation as, e.g., a 710% error in this value leads to a 725% relative variation of the photovoltaic potential. This graph demonstrates the coherence of the procedure, since solar radiation variations affect both its associated technology and the one competing with it, due to the great homogeneity of this resource in the studied area. 3.3. Parameters with a moderate influence on the result 3.3.1. Investment cost of the LV lines This parameter influences the above analysis in a similar way. These costs are related to the length of the lines; this factor has not been considered in the procedure, as there was no information on this relationship, but in any case, this would soften the influence of this parameter, which would have a smaller importance than shown in the spatial sensitivity graph (Fig. 10). The shape of this graph is determined by the household density. For LV lines with investment costs less than 10,000 h/km, the potential distribution remains practically constant, as the photovoltaic area has a density of almost one. For costs higher than 20,000 h/km, there is a new stabilization of the potential distribution, albeit smaller, as in this case high demands per pixel are reached. 3.3.2. Efficiency of the photovoltaic system This parameter repeats the behaviour described above, although less markedly. The value of the photovoltaic system efficiency, influenced by problems such as voltage drops in the lines, losses in the energy transformation in storage, etc., can be improved without requiring great changes in the current state of photovoltaic technology. Moreover, it must be considered that the other technologies are also evolving. For these reasons, this parameter is not included among those with a great influence, although it is very convenient to establish its mean value strictly for the reliability of the result. 3.3.3. Electrical grid rates Electrical grid rate increases can produce a total loss of competitiveness of connection to grid, but these increases must be very great for this to happen. In this sense, if the kWh cost for the user is doubled the connection to grid participation will be hardly reduced. The fact that electrical grid rates are being gradually reduced in all Europe in the past few years is much more important, and this short-term trend will continue. Again, however, great variations of the electrical grid rate are required for significant repercussions; e.g., a reduction to a third of its cost only produces a 5% increase in the total potential of connection to the grid. 3.3.4. CO2 tax The variation of CO2 tax in the Lorca area (Fig. 11) contradicts the previsions of renewable potential improvement when considering the external costs through ecological taxes. Considering the forecasts of higher tax values for CO2 emissions during 2010 [13], the photovoltaic potential will increase by less than 10%. This parameter value, CO2 tax, has been considered as ‘‘0’’ for the reference case.
ARTICLE IN PRESS J. Amador, J. Domı´nguez / Renewable Energy 31 (2006) 1505–1520
1516
00% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 5000
100% 80% 60% 40% 20%
10000
15000
20000
25000
0% 0.6
LT electric line cost 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 0.05
0.2
0.35
0.5
0.65
0.8
0.95
100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 0
Electric rate (Euros/kWh)
0.65 0.7 0.75 0.8 Fotovoltaic system efficiency
0.8
10 20 30 40 50 60 70 80 90 100 CO2 tax (Euros/Ton)
Fig. 10. Spatial sensitivity graph in the Lorca area: parameters with moderate influence. These parameters have a moderate influence in the study case. The influence of local variables is very important with respect to global parameters such as taxes and price in renewable energy projects.
3.4. Parameters with a negligible influence on the result 3.4.1. Discount rate In the potential evolution of this parameter, a contradiction is observed with what is usually asserted: ‘‘high discount rates help to conventional systems with respect to renewable ones, owing to their smaller investment costs’’. This is true for facilities connected to the grid; for isolated systems, if a detailed analysis is conducted as done in the present study, the conclusion is reached that conventional systems have higher investment costs than renewable ones. In any case, this influence is small, about 5%, affecting mainly central diesel technology. 3.4.2. Wind resource Variations in wind resource do not affect the potential allocation, inversely to the case for the solar resource. Variations above 10% only influence wind technology slightly. This is because the competitive wind potential area for this type of applications is small for Lorca. Wind resource variations around 25% do not significantly alter the result. The conclusion is that, with a distribution of wind conditions as in Lorca, it is not necessary to know the wind potential accurately in the entire territory to conduct a rural electrification study. However, it is convenient to confirm the potential of the better wind areas.
ARTICLE IN PRESS J. Amador, J. Domı´nguez / Renewable Energy 31 (2006) 1505–1520 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 0.02
0.03
0.04 0.05 Dicsount rate
0.06
0.07
0.08
100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 0.75
0.85
0.95
1.05
1517
1.15
1.25
Wind speed (0/1)
100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 1000 2000 3000 4000 5000 6000 7000 8000 Wind system investment (Euros/kW)
Fig. 11. Spatial sensitivity graph in the Lorca area: zero influence parameters. As noted, the great importance of the PV systems on a local scale joined to the low wind resources implied that parameters such as discount rate or others related with wind systems do not have any influence.
3.4.3. Other parameters The investment cost of HV lines or the investment cost of the wind system affect the results very slightly. Other parameters, such as the autonomy days of photovoltaic storage, the hub height of wind turbine and the efficiency of wind systems, although having a meaningful influence in the corresponding technology LEC, are characterized by spatial sensitivity graphs that are practically steady and do not affect the result. 4. Third stage: result stability 4.1. Effects of simultaneous parameter variation To guarantee the stability of the result, both the influence of each parameter separately and the influence of the simultaneous variations of the variables with a greater influence must be known. The problem is to establish the way in which this simultaneous variation should be outlined. These changes could be established as random in value and sign or, on the contrary, according to the expected short-term evolution. The second option has been used, as it is considered to fit better the philosophy of analysis followed until now and, in fact, it is simply a variation of the accomplished stages analysis developed by other authors
ARTICLE IN PRESS 1518
J. Amador, J. Domı´nguez / Renewable Energy 31 (2006) 1505–1520
[5] to exploit the GIS potential. The ‘‘combined effects’’ considered for the case of Lorca are described below. 4.2. Moderate evolution
20% higher demand. Cost of photovoltaic systems investment of 8000 h/kWP. Storage life of 6 years. Fuel price of 0.5 h/l.
The situation outlined refers to the most probable evolution of the parameters in a period of 1 or 2 years. It could also correspond to an error in the estimation of parameters, due to determinant actors unknown to the researcher in the studied case. The result obtained hardly varies with respect to the case of reference: It increases the photovoltaic potential by 2% and that of the grid by 1.5%, reducing that of central diesel by almost 4%. 4.3. Fast evolution
40% higher demand. Cost of photovoltaic systems investment of 6000 h/kWP. Storage life of 8 years. Fuel price of 0.5 h/l.
This corresponds to a situation of economic growth in the region, which translates to an important increase in demand. Furthermore, photovoltaic investment demand costs decrease strongly due to an expansion of renewable technologies and the meaningful increase of the battery lifetime. Fuel cost is maintained the same as in the previous case. In this case, there is a strong increase of the photovoltaic potential, reaching almost the threefourths of the regional potential, due to central diesel systems, as usual, maintaining hardly 4% of the potential. Grid connection maintains a percentage of about 25%. The fast evolution case has been studied, taking into account that the photovoltaic investment costs will only decrease 8000 h/kWP, resulting in that although photovoltaic systems will lose a 14% considering central diesel, they will maintain a meaningful importance, with almost the fourth of the total potential in the Lorca area. 4.4. Conclusions with respect to the stability of the result As indicated in the previous paragraphs, it can be concluded that the potential rural distribution of electrification for Lorca, in the ‘‘case of reference’’, presents a great stability for the optional connection to grid. The remaining potential is distributed between the photovoltaic systems and central diesel, particularly depending on the photovoltaic investment cost. Wind systems maintain a symbolic, but inalterable, representation. Solar radiation has not been considered in the combined effects study. This is one of the variables with a greatest influence on the result, as the sense of its variation is different from that of the other great influencing parameters. On the other hand, the sense and magnitude of its influence, in the cases previously studied, is immediately deduced from its corresponding spatial sensitivity graph.
ARTICLE IN PRESS J. Amador, J. Domı´nguez / Renewable Energy 31 (2006) 1505–1520
1519
5. Conclusions GIS has many advantages as a support tool for rural electrification plans with renewable energies. The many variables that must be considered, the fact of considering mean annual values and the study of the lifetime of the facilities, as well as the analysis capacity, could imply that linking the results to the initial variables will be difficult. Furthermore, a small variation of only one parameter can produce meaningful changes in the result, with the consequent uncertainty for planners who may use this tool. A ‘‘spatial sensitivity analysis methodology’’ is proposed, permitting to determine the parameters that have a greater influence on the method’s results, as well as the manner in which these parameters affect said result, in order to establish its reliability as a function of the ‘‘degree of spatial sensitivity’’ shown by the cost per kWh of each technology. Firstly, the variables with a greater influence in the LEC of each technology are determined by conventional sensitivity analysis and spider graphs. Then a spatial sensitivity analysis is conducted for all the detected variables, translated into a spatial sensitivity graph and into one or several maps. This spatial analysis produces two results. It determines the variables with a greater influence on the rural electrification plan (the variables with a greater influence for the practical case studied have turned out to be demand, storage life, photovoltaic system investment cost, fuel price and solar radiation). And, in addition, it establishes the manner in which this influence is accomplished so that the effects of this action can be anticipated, such as for subsidies, price increases, or even mistakes in the estimates of parameters, etc., considering their influence on the result. All of this, together with the analysis of the variation of several parameters, is used to determine the stability of the results, which allow increasing the control about these, and delimiting the error risk assumed by planners who may use this tool.
References [1] Alonso-Abella M, Mukadam K. Present situation in South Africa with respect to implementation of solar home systems. In: Renewable energy development European conference and APAS-RENA contractors meeting. Florencia: EDIFIR; 1996. [2] Ariza F, Lo´pez J, Lo´pez A. Territorial competitiveness of the stand alone photovoltaic systems versus grid electricity supply. A method and a study based on geographical information systems. Solar Energy 1997;61(2):107–18. [3] Monteiro C, et al. Evaluation of electrification alternatives in developing countries-the Solargis tool. In: MELECON ’98, ninth Mediterranean electrotechnical conference. New York: IEEE; 1998. p. 1037–41. [4] Muselli M, et al. Computer-aided analysis of the integration of renewable-energy systems in remote areas using a geographical-information system. Applied Energy 1999;63(3):141–60. [5] Vandenbergh P, et al. Integration of renewable energies for rural electrification in a Tunisian region. In: European photovoltaic solar energy conference. Barcelona: WIP; 1997. p. 2530–4. [6] Amador J, Domı´ nguez J. Renewable energies regional integration. Ciemat Technical Reports 2000;943:38. [7] Amador J, Domı´ nguez J. Application of geographical information systems to rural electrification with renewable energy sources. Renewable Energy 2005;30(12):1897–912. [8] Solargis Team. Solargis handbook. Brussels: Directorate General XII of the European Commission; 1996. [9] Domı´ nguez J. Evaluation of potential locations for electricity decentralized generation systems with renewable energy. In: Juaristi J, Moro I, editors. Models and systems of information in geography. Vitoria: Universidad del Paı´ s Vasco; 1996. p. 211–7.
ARTICLE IN PRESS 1520
J. Amador, J. Domı´nguez / Renewable Energy 31 (2006) 1505–1520
[10] Neirac F, et al. Solargis project: integration of renewable energies for decentralised electricity production in regions of EU and developing countries. In: 14th European photovoltaic solar energy conference, vol. II. Munich: WIP & H.S. Stephens & Associates; 1997. [11] Mayer D, et al. Sustainable development of areas: decentralized electrification. Applied Solar Energy 1998;34(3):10–8. [12] Moody-Stuart M. The energy supplies in the XXI century. Ingenierı´ a Quı´ mica 1998. [13] Foro Nuclear. French frame to mitigate the greenhouse gas emissions, including a direct tax on the emissions of carbon. Foro Nuclear 2000;408.
Renewable Energy 31 (2006) 1505–1520 www.elsevier.com/locate/renene
Spatial analysis methodology applied to rural electrification J. Amadora, J. Domı´ nguezb, a
Department of Electric Engineering, EUTI, UPM, Ronda de Valencia, E-28012 Madrid, Spain b Renewable Energies Division, CIEMAT, Av. Complutense 22, E-28040 Madrid, Spain Received 21 March 2005; accepted 3 September 2005 Available online 11 November 2005
Abstract The use of geographical information systems (GISs) in studies of regional integration of renewable energies provides advantages such as speed, amount of information, analysis capacity and others. However, these characteristics make it difficult to link the results to the initial variables, and therefore to validate the GIS. This makes it hard to ascertain the reliability of both the results and their subsequent analysis. To solve these problems, a GIS-based method is proposed with renewable energies for rural electrification structured in three stages, with the aim of finding out the influence of the initial variables on the result. In the first stage, a classic sensitivity analysis of the equivalent electrification cost (LEC) is performed; the second stage involves a spatial sensitivity analysis and the third determines the stability of the results. This methodology has been verified in the application of a GIS in Lorca (Spain). r 2005 Elsevier Ltd. All rights reserved. Keywords: GIS; LEC; Geographic distribution; Spatial sensitivity analysis; Renewable energy sources; Rural areas
1. Introduction Usually, in rural electrification geographical information system (GIS), the allocation of the potential of various technologies in the studied area is determined by comparing their LEC values [1–5]. Each isolated household will ‘‘belong’’ to the technology offering the lowest LEC in the point corresponding to that household. In this way, the potential of each Corresponding author. Tel.: +34 91 3466041; fax: +34 91 3466604.
E-mail address: [email protected] (J. Domı´ nguez). 0960-1481/$ - see front matter r 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.renene.2005.09.008
ARTICLE IN PRESS 1506
J. Amador, J. Domı´nguez / Renewable Energy 31 (2006) 1505–1520
technology in the studied area can be expressed by the number of households, or by the surface area in which such a technology offers the best cost per kWh throughout the life of the installation. But the real potential of each technology involved will be determined not only by minimizing the cost per kWh but also by the difference between its own costs and those of the other technologies which are in most direct competition with it. Obviously, a technology’s potential does not depend only on its cost per kWh, but also on other matters related to the problems of regional integration of renewable energies [6]. Furthermore, the influence that various parameters may have on these costs will be of vital importance, since a slight variation in a parameter might change the results significantly. For these reasons, a spatial sensitivity analysis methodology is proposed to determine the influence of various parameters on the results, linking them clearly with the initial variables. This method is structured in three stages: 1. Determination of the main influencing parameters by a sensitivity analysis of each LEC technology. 2. Spatial sensitivity analysis of the potential of the studied zone with respect to the parameters found in the previous stage, selecting the most significant variables in the distribution of the rural electrification potential. 3. Study of the spatial behaviour of the variables of the previous stage in order to determine the ‘‘stability’’ of the result obtained. The following points describe the application of this method to the rural electrification study of Lorca (Murcia, Spain) carried out by a GIS developed by the authors [7] from an initial version of Solargis [8–11], GIS prepared by several Research Institutions in the framework of the JOULE II Program of the European Union. 2. First stage: sensitivity analysis of LEC The method consists of applying a classic procedure considering the risk in the economic analysis of the investment selection: an analysis of the sensitivity of project costs with respect to variations of the technical and economical parameters. This type of analysis is conducted by establishing the value of all the parameters except the one being considered, which it is made to vary about a central value, usually the ‘‘best’’ estimate of that parameter. Figs. 1–6 show the results of the sensitivity analysis for each technology considered in the GIS. In these graphs, the influence of the variation of the parameter values with respect to the reference value can be observed. It must be taken into account that the slopes of the resulting curves cannot be compared directly, since the meaning of the variation of each parameter is different, as well as the probability that each variation will occur. As a rule, the causes of variation of the various parameters considered can be grouped into four categories:
Uncertainty in the establishment of the renewable resources, in this case solar radiation and wind speed. Evolution of the technology, causing changes in the efficiency and equipment lifetime. Changes in the socio-economic scenario that affect consumption, load density and discount rate, among other parameters.
ARTICLE IN PRESS J. Amador, J. Domı´nguez / Renewable Energy 31 (2006) 1505–1520
1507
1.5 Solar radiation 1.4 Battery life 1.3
lecpv
System efficiency 1.2 System investment 1.1 Consumption
Discount rate Efficient receivers cost
1
System life
0.9
0.8 0.5
1
1.5
Fig. 1. Photovoltaic LEC sensitivity. This graph shows the value variation influence of the different parameters regarding a reference value. The most significant parameters in the PV system are renewable resource, efficiency and battery.
1.5 1.4
Battery life
Battery Autonomy
1.3 Wind speed 1.2
System life
lecwt
Efficiency
System investment
1.1 Consumption
Discount rate
1
Turbine high
0.9 0.8 0.7 0.6 0.5
1
1.5
Fig. 2. Wind LEC sensitivity. This graph shows the value variation influence of the different parameters regarding a reference value. The most significant parameters in the wind system are renewable resource, and battery. In the study area, wind resources are not relevant.
ARTICLE IN PRESS J. Amador, J. Domı´nguez / Renewable Energy 31 (2006) 1505–1520
1508
1.5 Consumption
1.4
lec diesel
1.3 1.2
O&M cost
System life
1.1
System investment
Battery life
Diesel price 1
Discount rate
0.9 0.8 0.7 0.5
1
1.5
Fig. 3. Individual diesel LEC sensitivity. This graph shows the value variation influence of the different parameters regarding a reference value. The most significant parameters in the domestic diesel system are consumption and costs.
Changes in the energy market, specifically regarding equipment investment costs, fuel price and cost per kWh. The margins of variation considered with respect to the case of reference have been:
Renewable resources: 725%. Demand: 750%. Efficiencies: moderate variations. Investment costs, fuel and kWh prices: large variations.
The study of Figs. 1–6 (for greater clarity of the graphs, the parameters with the smallest influence are not shown), considering the probability of variation of the various parameters, allows one to determine the ones with the greatest influence on each corresponding LEC technology (6–8 parameters have been selected per technology). 3. Second stage: spatial sensitivity analysis 3.1. Spatial sensitivity graphs A ‘‘spatial sensitivity graph’’ is determined for the parameters selected in the first stage (a graph of this type is shown for demand [4]). It expresses the variation of the potential for rural electrification with each parameter of the studied area. The vertical axis of these
ARTICLE IN PRESS J. Amador, J. Domı´nguez / Renewable Energy 31 (2006) 1505–1520
1509
1.7 1.6 Load density
1.5
lec electric net
1.4 1.3 MT length 1.2 1.1 MT investment 1 0.9 Discount rate
Consumption
0.8 0.7 0.5
1
1.5
Fig. 4. Connection to grid LEC sensitivity. This graph shows the value variation influence of the different parameters regarding a reference value. The most significant parameters in the grid-connected system are load and consumption.
2
1.8 Consumption
lec central diesel
1.6
1.4 System investment 1.2
O&M cost Diesel price
System life Load density
1
BT line investment Discount rate
0.8
0.6 0.5
1
1.5
Fig. 5. Central diesel LEC sensitivity. This graph shows the value variation influence of the different parameters regarding a reference value. The most significant parameters in the central diesel system are consumption and cost.
ARTICLE IN PRESS J. Amador, J. Domı´nguez / Renewable Energy 31 (2006) 1505–1520
1510
2
1.8 Wind speed
lec wind-diesel
1.6
1.4 Consumption
1.2
Diesel inversion
Turbine efficiency 1
Load density Discount rate System life
0.8
0.6
0.5
1
1.5
Fig. 6. Wind–diesel LEC sensitivity. This graph shows the value variation influence of the different parameters regarding a reference value. The most significant parameters in hybrid system are consumption and renewable resource.
graphs represents the potential of each technology as a percentage (e.g., the number of households corresponding to that technology) and the horizontal axis represents the variation of the various parameters. These graphs are calculated from AML programs (ArcInfoTM programming language) that permit applying GIS to different values of the variables. The application of spatial sensitivity graphs allows ordering the variables selected in the first stage according to their influence on the result. As a rule, these variables can be classified into three influence groups: great, moderate and negligible. The result obtained for Lorca is the following:
Parameters with a great influence: J demand, J storage life, J investment cost of photovoltaic system, J investment cost of central diesel system, J diesel price, J solar radiation. Parameters with a moderate influence: J investment cost of LV lines, J photovoltaic system efficiency, J electrical grid rates, J CO2 tax, J discount rate,
ARTICLE IN PRESS J. Amador, J. Domı´nguez / Renewable Energy 31 (2006) 1505–1520
1511
Parameters with a negligible influence: J wind resource, J investment cost in HV lines, J investment cost in the wind system, J autonomy days of the photovoltaic system storage, J autonomy days of the wind system storage, J others.
The variables in the first group should be analysed individually. Conclusions are extracted from the spatial sensitivity graph for the study considered and, in consequence, one or more meaningful cases are chosen. Maps and numerical data are obtained for these cases, which are then compared to those of the reference case. For the two variables with the greatest influence a complete analysis is performed. The spatial sensitivity graph is shown considering only the four variables with the greatest influence. 3.2. Parameters with a great influence on the result 3.2.1. Demand Growing demand produces a drastic decrease of photovoltaic systems, losing competitivity strongly with respect to central diesel and slightly with respect to connection to the grid. Once photovoltaic systems have reached 10% of the regional potential, diesel domestic systems can compete with them; this is, for areas with low load densities far from the grid. Another important result is that connection to grid is stabilized at 30%. This is the only parameter that can lead to the disappearance of the renewable technologies potential. Demand decreases will significantly affect to the potential distribution of each technology. A 10% decrease produces a competitive loss for central diesel of 25%, but only a 5% loss for connection to the grid. This trend continues as the demand decreases, so that central diesel ‘‘disappears’’ and ‘‘connection to grid’’ loses its competitiveness slowly. The potential of wind systems is not affected by the drop in demand. To complete the spatial sensitivity graph information, a map showing the distribution of the technologies and a table of numerical results (corresponding to the case of a demand 40% higher than the reference demand) are provided (Fig. 7). 3.2.2. Storage life This parameter powerfully influences the result due to the cost of storage, representing a very important percentage of the investment cost, particularly for renewable energy facilities. A lifetime 20% higher increases the photovoltaic potential by approximately 30% compared to central diesel. Higher values would mean that photovoltaic systems exceed a 75% rural electrification potential, which would imply the near disappearance of central diesel. From this moment, onward photovoltaic systems continue to grow, albeit more slowly due to the fact that there are areas with high load densities. These results show the importance of a good system design and maintenance, particularly for the storage system. It also justifies the use of certain regulators that charge the batteries in an efficient and secure way, as well as the use of ‘‘tubular’’ batteries.
ARTICLE IN PRESS 1512
J. Amador, J. Domı´nguez / Renewable Energy 31 (2006) 1505–1520
Fig. 7. Spatial influence demand in the Lorca area (spatial sensitivity graph, map and numerical results for the case of reference and for a demand 40% higher). Demand is the most relevant parameter in the study area because PV system is the best renewable option.
The spatial sensitivity graph information is completed with the technology distribution map and the numerical results table, corresponding to a lifetime 40% greater than in the case of reference (Fig. 8). 3.2.3. Investment cost of the photovoltaic system The greatest variations of the photovoltaic system take place about 8000 h/kWp, specifically to 6000 and 10,000 h/kWp, where the curves defining the potential change
ARTICLE IN PRESS J. Amador, J. Domı´nguez / Renewable Energy 31 (2006) 1505–1520
1513
Fig. 8. Spatial influence of the storage lifetime in the Lorca area (spatial sensitivity graph, map and numerical results for the case reference and for a 6-year storage lifetime). In addition to demand, battery life is very important in order to improve the renewable rate in Lorca.
their slopes. Therefore, a 9000 h/kWp value is considered located within the areas which change most. An investment cost of 6000 h/kWp causes the specific weight duplicity of photovoltaic potential, corresponding to a massive installation program of the municipality electrification technology.
ARTICLE IN PRESS 1514
J. Amador, J. Domı´nguez / Renewable Energy 31 (2006) 1505–1520
3.2.4. Investment cost of the central diesel system A complete observation of this graph (Fig. 9) and the previous one shows the direct competition of the photovoltaic and central diesel systems; connection to grid is maintained with a percentage always near one-third of the total potential, while wind systems have a symbolic participation, although practically inalterable in normal conditions, around 0.7%. 3.2.5. Fuel price Fuel price shows a similar influence as that of the previous parameters: competition between photovoltaic and central diesel systems, with connection to grid and wind systems maintained, their participation being independent of parameter variations. Furthermore, in this case the value taken as reference, 0.4 h/l (it is assumed that a rural area will have access to agricultural type gas–oil), is represented in the area of the curves with a sharp slope. This means that fuel price variations will have a greater effect in this stage than in others. Although the current trend implies an increase of prices, it is not possible to establish the absolute stability of this trend [12]. 3.2.6. Solar radiation The variation of solar radiation due to the competition between photovoltaic and central diesel affects the results similarly; otherwise, the connection to grid and wind system 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 3000
11000
100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 550
0.7 0.8 0.9
100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 0.75
5000 7000 9000 Fotovoltaic system investment
100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 0.1 0.2 0.3 0.4 0.5 0.6
Diesel price (Euros/litre)
1
770 990 1210 1430 Central diesel investment (Euros/kW)
0.85
0.95 1.05 Solar radiation (0/1)
1.15
1.25
Fig. 9. Spatial sensitivity graph in the Lorca area: parameters with great influence. Next to demand and battery life, others parameters such as system investment, price of resources are also important for renewable energy integration, considering the great importance of PV system for this area.
ARTICLE IN PRESS J. Amador, J. Domı´nguez / Renewable Energy 31 (2006) 1505–1520
1515
remain constant. Again, the value considered (in this case, being a spatial variation parameter, the values are represented in 0/1) is represented in the centre of the section with the greatest variation, oscillating 715% with respect to the reference value. This graph (Fig. 9) shows the convenience of an accurate knowledge of the solar radiation as, e.g., a 710% error in this value leads to a 725% relative variation of the photovoltaic potential. This graph demonstrates the coherence of the procedure, since solar radiation variations affect both its associated technology and the one competing with it, due to the great homogeneity of this resource in the studied area. 3.3. Parameters with a moderate influence on the result 3.3.1. Investment cost of the LV lines This parameter influences the above analysis in a similar way. These costs are related to the length of the lines; this factor has not been considered in the procedure, as there was no information on this relationship, but in any case, this would soften the influence of this parameter, which would have a smaller importance than shown in the spatial sensitivity graph (Fig. 10). The shape of this graph is determined by the household density. For LV lines with investment costs less than 10,000 h/km, the potential distribution remains practically constant, as the photovoltaic area has a density of almost one. For costs higher than 20,000 h/km, there is a new stabilization of the potential distribution, albeit smaller, as in this case high demands per pixel are reached. 3.3.2. Efficiency of the photovoltaic system This parameter repeats the behaviour described above, although less markedly. The value of the photovoltaic system efficiency, influenced by problems such as voltage drops in the lines, losses in the energy transformation in storage, etc., can be improved without requiring great changes in the current state of photovoltaic technology. Moreover, it must be considered that the other technologies are also evolving. For these reasons, this parameter is not included among those with a great influence, although it is very convenient to establish its mean value strictly for the reliability of the result. 3.3.3. Electrical grid rates Electrical grid rate increases can produce a total loss of competitiveness of connection to grid, but these increases must be very great for this to happen. In this sense, if the kWh cost for the user is doubled the connection to grid participation will be hardly reduced. The fact that electrical grid rates are being gradually reduced in all Europe in the past few years is much more important, and this short-term trend will continue. Again, however, great variations of the electrical grid rate are required for significant repercussions; e.g., a reduction to a third of its cost only produces a 5% increase in the total potential of connection to the grid. 3.3.4. CO2 tax The variation of CO2 tax in the Lorca area (Fig. 11) contradicts the previsions of renewable potential improvement when considering the external costs through ecological taxes. Considering the forecasts of higher tax values for CO2 emissions during 2010 [13], the photovoltaic potential will increase by less than 10%. This parameter value, CO2 tax, has been considered as ‘‘0’’ for the reference case.
ARTICLE IN PRESS J. Amador, J. Domı´nguez / Renewable Energy 31 (2006) 1505–1520
1516
00% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 5000
100% 80% 60% 40% 20%
10000
15000
20000
25000
0% 0.6
LT electric line cost 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 0.05
0.2
0.35
0.5
0.65
0.8
0.95
100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 0
Electric rate (Euros/kWh)
0.65 0.7 0.75 0.8 Fotovoltaic system efficiency
0.8
10 20 30 40 50 60 70 80 90 100 CO2 tax (Euros/Ton)
Fig. 10. Spatial sensitivity graph in the Lorca area: parameters with moderate influence. These parameters have a moderate influence in the study case. The influence of local variables is very important with respect to global parameters such as taxes and price in renewable energy projects.
3.4. Parameters with a negligible influence on the result 3.4.1. Discount rate In the potential evolution of this parameter, a contradiction is observed with what is usually asserted: ‘‘high discount rates help to conventional systems with respect to renewable ones, owing to their smaller investment costs’’. This is true for facilities connected to the grid; for isolated systems, if a detailed analysis is conducted as done in the present study, the conclusion is reached that conventional systems have higher investment costs than renewable ones. In any case, this influence is small, about 5%, affecting mainly central diesel technology. 3.4.2. Wind resource Variations in wind resource do not affect the potential allocation, inversely to the case for the solar resource. Variations above 10% only influence wind technology slightly. This is because the competitive wind potential area for this type of applications is small for Lorca. Wind resource variations around 25% do not significantly alter the result. The conclusion is that, with a distribution of wind conditions as in Lorca, it is not necessary to know the wind potential accurately in the entire territory to conduct a rural electrification study. However, it is convenient to confirm the potential of the better wind areas.
ARTICLE IN PRESS J. Amador, J. Domı´nguez / Renewable Energy 31 (2006) 1505–1520 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 0.02
0.03
0.04 0.05 Dicsount rate
0.06
0.07
0.08
100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 0.75
0.85
0.95
1.05
1517
1.15
1.25
Wind speed (0/1)
100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 1000 2000 3000 4000 5000 6000 7000 8000 Wind system investment (Euros/kW)
Fig. 11. Spatial sensitivity graph in the Lorca area: zero influence parameters. As noted, the great importance of the PV systems on a local scale joined to the low wind resources implied that parameters such as discount rate or others related with wind systems do not have any influence.
3.4.3. Other parameters The investment cost of HV lines or the investment cost of the wind system affect the results very slightly. Other parameters, such as the autonomy days of photovoltaic storage, the hub height of wind turbine and the efficiency of wind systems, although having a meaningful influence in the corresponding technology LEC, are characterized by spatial sensitivity graphs that are practically steady and do not affect the result. 4. Third stage: result stability 4.1. Effects of simultaneous parameter variation To guarantee the stability of the result, both the influence of each parameter separately and the influence of the simultaneous variations of the variables with a greater influence must be known. The problem is to establish the way in which this simultaneous variation should be outlined. These changes could be established as random in value and sign or, on the contrary, according to the expected short-term evolution. The second option has been used, as it is considered to fit better the philosophy of analysis followed until now and, in fact, it is simply a variation of the accomplished stages analysis developed by other authors
ARTICLE IN PRESS 1518
J. Amador, J. Domı´nguez / Renewable Energy 31 (2006) 1505–1520
[5] to exploit the GIS potential. The ‘‘combined effects’’ considered for the case of Lorca are described below. 4.2. Moderate evolution
20% higher demand. Cost of photovoltaic systems investment of 8000 h/kWP. Storage life of 6 years. Fuel price of 0.5 h/l.
The situation outlined refers to the most probable evolution of the parameters in a period of 1 or 2 years. It could also correspond to an error in the estimation of parameters, due to determinant actors unknown to the researcher in the studied case. The result obtained hardly varies with respect to the case of reference: It increases the photovoltaic potential by 2% and that of the grid by 1.5%, reducing that of central diesel by almost 4%. 4.3. Fast evolution
40% higher demand. Cost of photovoltaic systems investment of 6000 h/kWP. Storage life of 8 years. Fuel price of 0.5 h/l.
This corresponds to a situation of economic growth in the region, which translates to an important increase in demand. Furthermore, photovoltaic investment demand costs decrease strongly due to an expansion of renewable technologies and the meaningful increase of the battery lifetime. Fuel cost is maintained the same as in the previous case. In this case, there is a strong increase of the photovoltaic potential, reaching almost the threefourths of the regional potential, due to central diesel systems, as usual, maintaining hardly 4% of the potential. Grid connection maintains a percentage of about 25%. The fast evolution case has been studied, taking into account that the photovoltaic investment costs will only decrease 8000 h/kWP, resulting in that although photovoltaic systems will lose a 14% considering central diesel, they will maintain a meaningful importance, with almost the fourth of the total potential in the Lorca area. 4.4. Conclusions with respect to the stability of the result As indicated in the previous paragraphs, it can be concluded that the potential rural distribution of electrification for Lorca, in the ‘‘case of reference’’, presents a great stability for the optional connection to grid. The remaining potential is distributed between the photovoltaic systems and central diesel, particularly depending on the photovoltaic investment cost. Wind systems maintain a symbolic, but inalterable, representation. Solar radiation has not been considered in the combined effects study. This is one of the variables with a greatest influence on the result, as the sense of its variation is different from that of the other great influencing parameters. On the other hand, the sense and magnitude of its influence, in the cases previously studied, is immediately deduced from its corresponding spatial sensitivity graph.
ARTICLE IN PRESS J. Amador, J. Domı´nguez / Renewable Energy 31 (2006) 1505–1520
1519
5. Conclusions GIS has many advantages as a support tool for rural electrification plans with renewable energies. The many variables that must be considered, the fact of considering mean annual values and the study of the lifetime of the facilities, as well as the analysis capacity, could imply that linking the results to the initial variables will be difficult. Furthermore, a small variation of only one parameter can produce meaningful changes in the result, with the consequent uncertainty for planners who may use this tool. A ‘‘spatial sensitivity analysis methodology’’ is proposed, permitting to determine the parameters that have a greater influence on the method’s results, as well as the manner in which these parameters affect said result, in order to establish its reliability as a function of the ‘‘degree of spatial sensitivity’’ shown by the cost per kWh of each technology. Firstly, the variables with a greater influence in the LEC of each technology are determined by conventional sensitivity analysis and spider graphs. Then a spatial sensitivity analysis is conducted for all the detected variables, translated into a spatial sensitivity graph and into one or several maps. This spatial analysis produces two results. It determines the variables with a greater influence on the rural electrification plan (the variables with a greater influence for the practical case studied have turned out to be demand, storage life, photovoltaic system investment cost, fuel price and solar radiation). And, in addition, it establishes the manner in which this influence is accomplished so that the effects of this action can be anticipated, such as for subsidies, price increases, or even mistakes in the estimates of parameters, etc., considering their influence on the result. All of this, together with the analysis of the variation of several parameters, is used to determine the stability of the results, which allow increasing the control about these, and delimiting the error risk assumed by planners who may use this tool.
References [1] Alonso-Abella M, Mukadam K. Present situation in South Africa with respect to implementation of solar home systems. In: Renewable energy development European conference and APAS-RENA contractors meeting. Florencia: EDIFIR; 1996. [2] Ariza F, Lo´pez J, Lo´pez A. Territorial competitiveness of the stand alone photovoltaic systems versus grid electricity supply. A method and a study based on geographical information systems. Solar Energy 1997;61(2):107–18. [3] Monteiro C, et al. Evaluation of electrification alternatives in developing countries-the Solargis tool. In: MELECON ’98, ninth Mediterranean electrotechnical conference. New York: IEEE; 1998. p. 1037–41. [4] Muselli M, et al. Computer-aided analysis of the integration of renewable-energy systems in remote areas using a geographical-information system. Applied Energy 1999;63(3):141–60. [5] Vandenbergh P, et al. Integration of renewable energies for rural electrification in a Tunisian region. In: European photovoltaic solar energy conference. Barcelona: WIP; 1997. p. 2530–4. [6] Amador J, Domı´ nguez J. Renewable energies regional integration. Ciemat Technical Reports 2000;943:38. [7] Amador J, Domı´ nguez J. Application of geographical information systems to rural electrification with renewable energy sources. Renewable Energy 2005;30(12):1897–912. [8] Solargis Team. Solargis handbook. Brussels: Directorate General XII of the European Commission; 1996. [9] Domı´ nguez J. Evaluation of potential locations for electricity decentralized generation systems with renewable energy. In: Juaristi J, Moro I, editors. Models and systems of information in geography. Vitoria: Universidad del Paı´ s Vasco; 1996. p. 211–7.
ARTICLE IN PRESS 1520
J. Amador, J. Domı´nguez / Renewable Energy 31 (2006) 1505–1520
[10] Neirac F, et al. Solargis project: integration of renewable energies for decentralised electricity production in regions of EU and developing countries. In: 14th European photovoltaic solar energy conference, vol. II. Munich: WIP & H.S. Stephens & Associates; 1997. [11] Mayer D, et al. Sustainable development of areas: decentralized electrification. Applied Solar Energy 1998;34(3):10–8. [12] Moody-Stuart M. The energy supplies in the XXI century. Ingenierı´ a Quı´ mica 1998. [13] Foro Nuclear. French frame to mitigate the greenhouse gas emissions, including a direct tax on the emissions of carbon. Foro Nuclear 2000;408.