8/7/2019 Multi Objective Assessment of Rural Electrification in Remote Areas With Poverty Considerations 2009 Energy Policy
1/13
Multi-objective assessment of rural electrification in remote areas withpoverty considerations
Diego Silva, Toshihiko Nakata
Department of Management Science and Technology, Graduate School of Engineering, Tohoku University, Aoba-yama 6-6-11-815, Sendai 980-8579, Japan
a r t i c l e i n f o
Article history:
Received 28 July 2008
Accepted 30 March 2009Available online 12 May 2009
Keywords:
Renewable energy system
Energy access
Multi-objective assessment
a b s t r a c t
Rural electrification with renewable energy technologies (RETs) offers several benefits to remote areas
where diesel generation is unsuitable due to fuel supply constraints. Such benefits include
environmental and social aspects, which are linked to energy access and poverty reduction in less-favored areas of developing countries. In this case, multi-objective methods are suitable tools for
planning in rural areas. In this study, assessment of rural electrification with renewable energy systems
is conducted by means of goal programming towards fuel substitution. The approach showed that, in
the Non-Interconnected Zones of Colombia, substitution of traditional biomass with an electrification
scheme using renewable energy sources provides significant environmental benefits, measured as land
use and avoided emissions, as well as higher employment generation rates than diesel generation
schemes. Nevertheless, fuel substitution is constrained by the elevated cost of electricity compared to
traditional biomass, which raises households energy expenditures between twofold to five times higher
values. The present approach, yet wide in scope, is still limited for quantifying the impact of energy
access improvements on poverty reduction, as well as for the assessment of energy systems technical
feasibility.
& 2009 Elsevier Ltd. All rights reserved.
1. Introduction
The Millennium Development Goals (MDGs), an initiative of
the UN looking forward the achievement of eight global targets by
nations by the year 2015, placed first on the list the reduction of
poverty by half by the year 2015 (UN, 2000). The achievement of
this goal within rural areas of developing countries is highly
correlated with energy access, a fact pointed out by the United
Nations in the UN Millennium Declaration and the International
Energy Agency in its World Energy Outlook, as well as by many
other organizations around the world (IEA, 2002). Improvement of
energy access conditions can be associated with two main
aspects: increase of electrification rates and substitution of
traditional biomass. As of year 2000, there were over 1.6 billionpeople, with no access to electricity. More than 99% of these
people come from developing countries and four out of five live in
rural areas, the majority of them living under the poverty line. In
addition, people relying on traditional biomass for space heating
and cooking reaches 2.4 billion (IEA, 2004).
Among the rural population, people living in remote areas are
especially vulnerable. Many of these areas have poor road and
public utilities infrastructures, but possess unexploited renewable
resources. Thus, decentralized electrification schemes with re-
newable energy technologies (RETs) based on local resources are a
suitable solution to satisfy energy needs. This fact contrasts with
actual unsustainable electrification schemes with diesel genera-
tion or extension of the electric grid. In spite of the relevance of
RETs for energy supply in remote rural areas, low diffusion rates of
this kind of technologies and failure of technology transfer
programs for rural electrification are common in developing
countries (Department of Economic and Social AffairsUnited
Nations, 2001; Green, 1999).
The reason why RETs have not been successful in rural areas so
far, has been attributed in part to the lack of integrated evaluation
approaches in rural electrification planning. Such approachesallow the inclusion of economic, social and environmental aspects
of technologies into the planning process, which is fundamental to
address poverty as a multi-dimensional issue. They also permit
addressing certain barriers to the diffusion of RETs in rural areas of
developing countries (Painuly, 2001). Among the integrated
evaluation approaches existent, multi-objective decision-making
(MODM) methods like goal programming (GP) are gaining
popularity in energy planning issues. MODM methods permit
the inclusion of multiple and often conflicting targets into
optimization schemes. With these methods it is also possible to
incorporate factors that cannot be expressed in comparable
units, a common limitation when addressing poverty issues in
ARTICLE IN PRESS
Contents lists available at ScienceDirect
journal homepage: www.elsevier.com/locate/enpol
Energy Policy
0301-4215/$- see front matter& 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.enpol.2009.03.060
Corresponding author. Tel./fax: +8122 7957004.
E-mail addresses: [email protected] (D. Silva),
[email protected] (T. Nakata).
Energy Policy 37 (2009) 30963108
http://www.sciencedirect.com/science/journal/jepohttp://www.elsevier.com/locate/enpolhttp://dx.doi.org/10.1016/j.enpol.2009.03.060mailto:[email protected]:[email protected]:[email protected]:[email protected]://dx.doi.org/10.1016/j.enpol.2009.03.060http://www.elsevier.com/locate/enpolhttp://www.sciencedirect.com/science/journal/jepo8/7/2019 Multi Objective Assessment of Rural Electrification in Remote Areas With Poverty Considerations 2009 Energy Policy
2/13
technology evaluation frameworks (Pohekar and Ramachandran,
2004). Nevertheless, research examples deploying MODM meth-
ods for energy system analysis focusing in rural planning
addressing poverty issues are scarce. Furthermore, there are few
studies dealing with modeling of optimum mix of energy
resources using a decentralized energy approach (Hiremath et
al., 2007).
The purpose of the research is to study the introduction of
renewable energy systems as a possible alternative in the futurefor rural electrification towards the substitution of traditional
biomass in developing countries by means of multi-objective
approach including multiple aspects. The integrated evaluation of
the system is used to highlight the multiple benefits offered by
RETs in remote rural areas. Goal programming is applied to assess
energy system performance with respect to electricity generation
cost, employment generation, land use, and avoided CO2 emis-
sions for the case of the Non-Interconnected Zones (NIZ) of
Colombia. The introduction of electricity for cooking may be
considered unsuitable for rural areas of developing countries, and
other alternatives such as improved cooking stoves may be more
appropriate in the short term. However, the simultaneous
inclusion of different aspects in the assessment process by means
of the application of the MODM methodology may show the
benefits from the use of electricity from renewables in the long
term overlooked by other approaches.
2. Energy access and poverty alleviation
The concept of poverty has evolved as a multi-dimensional
issue that, in addition of low income level, it is linked to several
aspects of human life, like deficient health levels and failure to
obtain access to medical assistance and safe water and food
provision. Thus, poverty alleviation can be interpreted as the
improvement of living conditions towards the achievement of
minimum basic human needs, as it has been outlined in the
targets set within the MDGs. In spite that energy is not explicitlymentioned in the MDGs and is not always considered a basic need,
energy is necessary to conduct human activities and provides
basic services for human life such as heat for cooking, space
heating and illumination. Moreover, energy provision enhances
the productivity of agricultural and industrial activities that, in
the end, may translate in higher income opportunities. Based on
the above, it has been acknowledged that access to reliable and
affordable energy plays an essential role in the achievement of the
MDGs (Modi et al., 2005). Accordingly, a minimum level of energy
consumption per capita can be associated to a condition above the
poverty line (Spreng, 2005).
Improvement of energy access conditions requires not only the
attainment of a reliable energy supply but also the transition to
modern energy forms like LPG and electricity. Traditional fuels,such as charcoal, straw, wood, agricultural wastes and dung,
which are termed traditional biomass, are generally not
commercially traded and are used extensively and inefficiently
by the poor, resulting in negative impacts on the people and the
environment. For example, people in rural areas of developing
countries, especially women and children, spend much of their
time for collection of firewood. Moreover, burning of biomass in
inefficient stoves is a major cause of indoor smoke pollution,
which claims each year the lives of over 1.6 million women and
children according to the World Health Organization (IEA, 2004).
Currently, in rural areas of the world more than 1.3 billion people
have no access to electricity and more than 2.1 billion people rely
on the use of traditional fuels to cover energy needs (IEA, 2004,
2006). Therefore, substitution of traditional biomass represents an
important aspect of energy access, in particular for rural areas in
developing countries.
The relationship between energy access and poverty in the
energy planning literature is scarcely addressed. Most of the
studies have approached the issue founded in the concept of
energy poverty, which describes the condition where people
cannot afford access to a sustainable energy supply. Pachauri et al.
developed the energy access-consumption matrix, a two-
dimensional indicator to measure energy poverty in the Indianhousehold sector for the period 19832000, incorporating non-
commercial energy sources and differentiating the population
according to the income level and to the type of energy source as
well as its consumed amount (Pachauri et al., 2004). Kemmler
et al. worked on the definition of indicators capable to describe
poverty and sustainable development in a multi-dimensional way
through energy consumption data, proposing the access-adjusted
useful energy as a suitable indicator (Kemmler and Spreng, 2007).
Kanagawa et al. analyzed the impacts of energy access for cooking
in rural households by designing an energy-economic model
introducing the opportunity cost associated with wood collection
and the exposure to particulate matter (Kanagawa and Nakata,
2007). The same authors extended their analysis to cover the
impact of lighting in literacy rates (Kanagawa and Nakata, 2008).
In the case of remote rural areas of developing countries,
improvement of energy access conditions by means of rural
electrification and fuel substitution is directly related to the
efficient utilization of local energy resources with RETs. Although
there are several barriers to the successful introduction of these
technologies, some of these barriers can be overcome highlighting
the benefits provided by RETs to rural communities in comparison
to conventional electrification schemes based on diesel genera-
tion. To this end, MODM methods offer a convenient approach for
assessment of technologies given that they allow the simulta-
neous inclusion of several attributes even with different units.
3. Multi-objective approach for assessment of rural
electrification
An MODM method is deployed in order to evaluate renewable
energy systems using local energy resources for rural electrifica-
tion and substitution of traditional biomass in localities of
Colombia without access to the electricity grid, referred to as
NIZ. The energy system for electrification with RETs is designed to
meet two targets. First, to satisfy electricity needs according to the
demand rates observed in the NIZ, with local renewable resources
instead of diesel generation, which is currently the main supply
alternative in the target area. The second purpose of the energy
system is to extend electricity supply in order to cover energy
demand for cooking purposes in households and to substitute
with electric stoves traditional stoves of low conversion efficiency
using firewood. This second target is referred to as fuel substitu-tion throughout the paper.
3.1. The Non-Interconnected Zones of Colombia
The NIZ of Colombia comprehends remote areas outside the
national electric grid, or National Interconnected System (NIS).
Most of the regions within the NIZ have the lowest GDP per capita
levels in the country. The NIZ, depicted in Fig. 1, extend over
756,531 km2, which is nearly 66% of countrys territory surface.
The population reaches 1,524,304, accounting for only 4% of
countrys population, and is highly dispersed as there are
approximately 2 inhabitants/km2, while countrys average is
38 inhabitants/km2. Around 88% of people in the NIZ live in
rural areas. Although average electrification rates of Colombia
ARTICLE IN PRESS
D. Silva, T. Nakata / Energy Policy 37 (2009) 30963108 3097
8/7/2019 Multi Objective Assessment of Rural Electrification in Remote Areas With Poverty Considerations 2009 Energy Policy
3/13
were over 90% as for the year 2003 (Deparatmento Nacional de
Planeacion, 2007), almost two thirds of people in NIZ have no
access to electricity. The average electricity price is twice the price
charged in the NIS, with only half of service hours. Diesel
generation plants supplies almost all the electricity, and only a
few small hydro power plants and PV systems operate in the NIZ.
Fuels are over 60% more expensive than the capitals price due to
restrictions on their transportation and security issues. Wood and
LPG are the most important energy fuels for cooking, with a userate of 41% and 35% respectively (Unidad de Planeacion Minero
Energetica, 2000a; Zapata and Bayona, 2000). Given the
remoteness of localities and the extensive availability of
renewable resources within the NIZ, RETs are a primary
potential solution to support their development. The Unidad de
Planeacion Minero Energetica (UPME), institution in charge of
national energy planning issues in Colombia, has referred in this
regard in the national energy plan (Unidad de Planeacion Minero
Energetica, 2003). Nevertheless, inexistence of policies or
regulations encouraging the promotion of renewable energies
has raised the necessity to revise methodologies for assessment of
technologies based only on costs, so as to include environmental
and social factors (Ruiz and Rodrguez-Padilla, 2006). Table 1
summarizes the characteristics of the NIZ. Villages are categorized
into three groups of localities according to the size of population
as large, medium and small, following the typology proposed by
the UPME.
The importance and application of MODM methods to aid in
rural energy planning issues in Colombia have been referenced in
previous researches (Henao Piza, 2004). The UPME developed an
information and analysis tool to support the decision makers in
energy planning in the NIZ considering an integral analysis, using
indicators to estimate the performance of alternatives in several
aspects such as costs, level of adoption of technologies andenvironmental factors (Unidad de Planeacion Minero Energetica,
2000b; Zapata and Bayona, 2000). The result of this study foresees
that diesel generation is the most suitable technology for
supplying electricity in the majority of NIZ localities, and that
RETs are suitable only for localities with less than 200 inhabitants
(Zapata and Bayona, 2000). In addition, the evaluation of
appropriate rural electrification alternatives towards improve-
ment of communitys livelihoods within the NIZ has been
analyzed using a multi-criteria approach (Cherni et al., 2007).
The authors suggest that a micro-hydro power plant and a hybrid
system combining this alternative with a diesel generator are the
most appropriate energy generation options for a remote rural
area of 400 inhabitants. An initial insight of the use of a multiple
objective approach in the analysis of renewable energy systems
has been given for rural electrification in the NIZ (Silva and
Nakata, 2008).
3.2. Renewable energy system for the NIZ
The proposed energy system is designed considering the three
locality types outlined above, in accordance with the character-
ization made by the UPME. The energy system, presented in Fig. 2,
is a representation of the elements involved in the supply and
consumption of energy in a single locality. This system is
composed by three main groups of elements, namely the energy
resources representing the sources of primary energy; the energy
conversion technologies symbolizing the power plants for the
supply of electricity; and energy demand sectors making use ofthe electricity supplied in different devices in order to satisfy
certain needs.
Renewable energy resources which can be harnessed locally in
NIZ are considered in contrast to diesel fuel. Distribution of fuels
within the NIZ is highly constrained. This results in high fuel
prices which rises the costs of electricity supply, since in these
areas the service is almost totally supplied by diesel generators
(Unidad de Planeacion Minero Energetica, 2000a). Therefore
utilization of local resources is fundamental to boost sustain-
ability of electricity supply. Energy resources geographical
distribution is not included in the analysis, and average avail-
ARTICLE IN PRESS
Fig. 1. The Non-Interconnected Zones of Colombia.
Table 1
Characterization of the NIZ by localities.
Large Medium Small
Population range 4500 200500 o200
Population per locality (assumed) 1000 500 200
Total population 344,526 145,066 38,128
Electricity demand per house (kWh/month/household) 68 25 23
Total electricity demand (GWh/yr) 83.7 11.4 2.8
Firewood demand for cooking (kWh/month/household) 1932 2286 1189
Total cooking energy demand (GWh/yr)a 239 104 14
Total final energy demand (GWh/yr)b 323 115 17
Share of energy for cooking in total energy demand (%) 74 90 84
Calculated based on data from Unidad de Planeacion Minero Energetica (2000a).a Conversion efficiency of 10% assumed, LHV for firewood 16.7MJ/kg.b
Sum of electricity demand and cooking energy demand.
D. Silva, T. Nakata / Energy Policy 37 (2009) 309631083098
8/7/2019 Multi Objective Assessment of Rural Electrification in Remote Areas With Poverty Considerations 2009 Energy Policy
4/13
ability values corresponding to the NIZ are used (Silva and Nakata,2008).
Electricity conversion technologies under consideration corre-
spond to different small-scale power plants, namely a steam
turbine coupled with direct combustion of firewood (Dir.Comb.-
Firewood), a steam turbine coupled with direct combustion of
organic wastes (Dir.Comb.Waste), a gas engine using biogas
produced by anaerobic digestion of organic wastes (Dig.Biogas-
Comb.), solar photovoltaic (PV) panels, windmills (Wind), and
small hydro power plants (Hydro). These technologies are
combined to form a single energy supply unit for a single locality.
They, together with the energy resources employed by them, are
designated as the renewable energy system. A diesel generation
scheme, composed by a diesel generator using diesel fuel, is used
to establish the baseline of comparison for evaluation of the
renewable energy systems performance with respect to multiple
attributes. The main features of the technologies are summarized
in Table 2.
Energy demand includes the energy for cooking and for electric
appliances in the residential sector, and energy for electric
appliances in other sectors, namely industrial, commercial and
institutional sectors. Energy demand for cooking is approximated
to the final energy quantities resulting from the combustion of
firewood in traditional stoves with 10% energy conversionefficiency, taking into account current consumption levels of this
fuel in the NIZ. Energy for other purposes corresponds to average
electricity consumption levels observed in all sectors of the NIZ.
For purposes of the analysis each locality type is considered
homogeneous, neglecting any differences by income or energy
consumption levels among the population. The population under
study corresponds to human groups living together in small towns
or villages excluding peasants living in farms.
3.3. Case setting
The study considers the baseline and the two cases explained
below:
(a) Baseline for comparison: energy for cooking purposes is
supplied by traditional biomass and electricity demand in all
sectors, equivalent to actual consumption rates in the NIZ, is
supplied by a diesel generation scheme.
(b) No fuel substitution (NFS): energy for cooking purposes is
supplied by traditional biomass and electricity demand is
supplied by the renewable energy system.
(c) Total fuel substitution (TFS): traditional stoves are replaced by
electric stoves and electricity is supplied by the renewable
energy system, including demand for cooking.
3.4. Integrated assessment of the renewable energy system
Performance of the rural energy system is evaluated with
respect to economic, social and environmental aspects, by means
of four attributes.
(a) Electricity generation cost: given that low income of population
may prevent the introduction of the energy system, cost of
electricity supplied can be used as a parameter for the need of
subsidies or other mechanisms to adjust electricity prices.
(b) Employment generation: jobs needed for operation of the
renewable energy system are an income opportunity to some
of the inhabitants of the NIZ; thus, employment generation is
used to relate system introduction with poverty alleviation.
(c) Land use: construction of power plants may represent a
negative impact on the local environment of remote ruralareas in developing countries, since it may interfere with
other important uses for land like agriculture and habitat
conservation.
(d) Avoided emissions: the amount of emissions of CO2 avoided by
the displacement of diesel generation electrification schemes
serves to highlight one characteristic advantage of introducing
renewable energy systems; avoided emissions can be used as
a factor to promote environmental acceptability of renewable
energy systems based on local resources.
The integrated evaluation of the renewable energy system
considers the benefits granted in these four attributes by systems
introduction. Benefits are estimated as the percentage deviation
with respect to a goal value set to each attribute, by means of the
ARTICLE IN PRESS
Outcomes of the analysis
Energy access in remote rural areas
Feasibility of fuel substitution
with local energy resources
(No fuel substitution)(Total fuel substitution)
ElectrificationElectrification &
fuel substitution
Target area
Non Interconnected Zones of Colombia
Type 1
(> 500 people)
Type 2
(200-500 people)
Type 3
(< 200 people)
Technology alternatives
Diesel generation
(used as basis
for comparison)
Renewable energy system
(energy supply with
local resources)
Alternative for rural electrification
Methodology
Goal programming
(four goals, four priority structures)
Benefits
(deviation from goal's values)
System's configuration
System's performance
Fig. 2. Energy system for rural areas of Colombia.
D. Silva, T. Nakata / Energy Policy 37 (2009) 30963108 3099
8/7/2019 Multi Objective Assessment of Rural Electrification in Remote Areas With Poverty Considerations 2009 Energy Policy
5/13
following equation:
bj gj Aj
gj
! 100; j felectricity cost; land useg (1)
bj Aj gj
gj
! 100; j femployment generation; av: emissionsg
(2)
where j is the goal, bj is the benefit with respect to attribute j, Aj is
the performance of renewable energy system with respect to
attribute j and gj is the goal with respect to attribute to attribute j.Goals values and their meaning are explained in the following
subsection. The integrated assessment of the system includes the
design of the energy system. For this purpose a specific MODM
method called goal programming is utilized. Different to other
MODM methods used in energy planning which can only consider
a limited number of combinations of technologies defined by the
decision maker with anticipation, such as the analytical hierarchy
process and many other multi-criteria methods, GP allows for the
estimation of an optimal mix of technologies, indicated by the
system configuration, in a continuous range of alternatives. A
graphic representation of the multi-objective approach proposed
in the study is presented in Fig. 3.
3.5. Goal programming formulation
GP is an MODM method very close to optimization methods,
and is considered an extension of linear programming (LP).1 In
addition to an objective function and a set of inequalities
representing the constraints of the problem, in GP formulations
functions for multiple objectives with respect to different
attributes are introduced and formulated as goal constraints.
The right-hand side values in these inequalities correspond to
constant target values or goals. Each goal constraint equation
involves a pair of deviation variables which indicate the deviation
of the goal constraints with respect to their respective goals. The
objective function in GP is to minimize these deviations
(Schneiderjans, 1995). The objective function can be expressed
as a weighted sum or as an order of preferences. The latter case,
known as a preemptive GP, refers to a GP formulation where the
optimization procedure is performed minimizing deviation vari-
ables with respect to each attribute in the sequence specified by
the order of preferences in a priority structure. In the study
preemptive GP is applied in order to obtain the optimal
configuration of the system, using the electricity to be supplied
by each technology as the decision variables. The order of
preferences considered is presented in a further subsection.
Equations of the GP formulation deployed in the study are
presented in the Appendix A.
GP has been used in rural energy planning studies looking at
energy resource allocation at local and regional level (Rama-
nathan and Ganesh, 1995; Kanniappan and Ramachandran, 2000)
and at country level (Mezher et al., 1998). Calculations were
performed with LINDOs software. This is a computer tool
commonly used for solving linear programming models. Its
application has been referenced for the optimum utilization of
renewable energy sources in a remote area (Akella et al., 2007).
3.6. Goal constraints and goals
In the study goals represent evaluation guidelines for theperformance of renewable energy systems. Performance values of
the baseline energy system are used as goals for the formulation.
Models goal constraints and goal values used for this study are
explained below.
(a) Electricity generation cost (US$/kWh): this goal constraint looks
for an electricity generation cost value lower or at most equal
to that of diesel generation, which translates in the mini-
mization of the overachievement of the electricity costs goal
value (US$0.136/kWh for large localities, US$0.209/kWh for
medium localities and US$0.216/kWh for small localities).
Economic performance of the system is calculated based on
operation and maintenance costs, capital cost, and resource
cost, considering a life time of 20 years for technologies and a
ARTICLE IN PRESS
Table 2
Main features of energy conversion technologies.
Technology Diesel Dir.comb. firewood Dir.comb. waste Dig.biogas comb. PV Wind Hydro
Scale (kW)b 30/75/150a 32/79/182 1/3/7 1/2/4 22/54/124 2/6/13 11/26/61
Efficiency (%) 30a 28c 28c 35d 11.3c 45d 80d
Capital cost (US$/kWp)a 300 3000 3000 500 8000 3200 4500
O&M cost (USb/kWh) 1.9/5.8/6.1n 4.74c 4.74c 5.5d 0.15c 1.0c 1.5c
Employees (jobs/GW)f 0.83/0.27/0.25o 1 1 6 2.7 2 1.4
Land use (m2
/kW) 182h
45.1c,i
45.1c,i
144j
48.5c,k
267.7e,l
50m
Emissions (kg-CO2/kWh)g 0.262 0.273 0.273 0.305 0 0 0
a Unidad de Planeacion Minero Energetica (2000a).b Plant size calculated according to resource availability and electricity demand.c EPRI and US Department of Energy (1997).d IEA (1997).e Micro-grid verification test facilities, Tokyo Gas.f For operation and maintenance, Moreno and Lopez (2008).g For plant operation, Mezher et al. (1998).h Calculated based on land required for fuel storage (Unidad de Planeacion Minero Energetica, 2000a).i A scaling factor of 5 is assumed.j Iwate biogas plant, Iwate prefecture, Japan.k Device area is 20 m2.l Spacing between windmills equal to rotor diameter (5.5 m).m Assumed value.n Equivalent to 10% of capital cost.o
Calculated based on value from Mezher et al. (1998).
1 Linear programming (LP): planning of activities to obtain an optimal result,
its application generally involves the problem of allocating limited resources to
competing activities in a best (i.e. optimal) way (Hillier and Lieberman, 2001).
D. Silva, T. Nakata / Energy Policy 37 (2009) 309631083100
http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-8/7/2019 Multi Objective Assessment of Rural Electrification in Remote Areas With Poverty Considerations 2009 Energy Policy
6/13
10% discount rate. The cost of electric stoves is omitted, and it
is assumed that households will purchase these devices
independently. Same costs for transmission and distribution
of electricity are considered for all technologies and, thus, also
disregarded in the analysis.
(b) Employment generation (jobs/kWh): seek an employment level
at least equal to or higher than that of diesel generation,
equivalent to 5.130 107 jobs/kWh according to Mezher et al.
(1998). Employment generation levels correspond to values
estimated by Moreno and Lopez (2008) for operation and
maintenance of electricity generation plants and are utilized
irrespective of plant scale, disregarding possible numericaldifferences for low-scale plants. These values are only mean-
ingful as relative quantities to differentiate employment
generation levels among the technologies being considered.
(c) Land use (m2/kWh/yr): achieve a plant area no greater than the
required for the storage of the amount of diesel needed for the
supply of electricity (0.0415 m2/kWh/yr). This attribute mea-
sures the efficient use of land and it is calculated based on
land requirements for plant installation.
(d) Avoided CO2 emissions (kg-CO2/kWh): avoid a quantity of CO2emissions equivalent to the emissions produced by a diesel
generation scheme (0.36kg-CO2/kWh). Emissions are calcu-
lated with respect to CO2 emissions generated by a diesel
generation plant. In addition to emissions contributed by
system operation those stemming from the deforestation
associated with land use requirements are also accounted,
measured based on the average rate of CO2 sequestration per
square meter in tropical zones. Equivalent CO2 emissions due
to methane production from the decay of organic wastes are
taken into account for direct combustion of waste and biogas
technologies.
3.7. Constraints
The models constraints are the total electricity demand andthe maximum resource availability in each NIZ locality type. Data
regarding the electricity demand and the availability of energy
resources are listed in Table 3.
3.8. Order of preferences
The configuration and outcomes respect to the attributes of the
energy systems obtained are analyzed according to four priority
structures, listed in Table 4, each one describing a different order
of preferences set for the objective functions. They represent the
possible views of the decision maker towards the assessment of
rural electrification projects. The first two priority structures serve
to assess whether there is a conflict between economic
performance and employment generation. The latter two
ARTICLE IN PRESS
Electricity
distribution
Small -Hydro
Dir.Comb .Firewood
Windmill
Dir.Comb .
Waste
PV
Organic
Resid . Waste
Wind
resource
Solar
resource
Hydro
resource
Firewood
resource
Dig.Biogas
Comb .
Org.Resid .Waste : organic residential waste
Dir .Comb . : direct combustion
Dig .BiogasComb .: anaerobic digestion and combustion of biogas
Cooking
Electricity
demand
Residential
Electric
appliances
Electric
stove
Traditional
stove
Energy
resourcesEnergy demand
Energy using
devices
Energy
conversion
technologies
Electricity
demand
Electric
appliances
Other sectors
Diesel
generationDiesel fuel
RENEWABLE ENERGY SYSTEM
DIESEL GENERATION SCHEME
Fig. 3. Multi-objective approach for energy access issues in rural areas.
D. Silva, T. Nakata / Energy Policy 37 (2009) 30963108 3101
8/7/2019 Multi Objective Assessment of Rural Electrification in Remote Areas With Poverty Considerations 2009 Energy Policy
7/13
8/7/2019 Multi Objective Assessment of Rural Electrification in Remote Areas With Poverty Considerations 2009 Energy Policy
8/13
US$0.158/kWh. This occurs as a consequence of the introduction
of lower-performance technologies in the system configuration at
the expense of the depletion of Wind and Hydro resources. Fuel
substitution requires the introduction of direct combustion of
firewood (Dir.Comb.Firewood) but still with small shares, below
30% of the total electricity supply.
In localities with population between 200 and 500 people
(Medium) benefits were obtained in avoided emissions and
employment generation in the case of no fuel substitution as
can be seen from Fig. 5b. Transition to electricity put into evidence
a clear trade-off between employment generation and electricitycosts. For total fuel substitution employment benefits rise sharply
from over 50% to more than 100% of the baseline value. In
contrast, system performance in the other three attributes decline
to negative values, in particular the electricity cost, which
increases from around b12 to b23 per kWh. This cost increase
is a result of the large electricity output required for fuel
substitution, which must be covered by expensive technologies
such as PV or direct combustion of firewood as long as resources
for cheaper technologies are depleted. This electricity supply
increase is more than ten times bigger than electricity demand
without fuel substitution, different to other locality types where
total substitution of traditional biomass represents only four to six
times the original electricity demand values. For the Cost and Job
priority structures benefits for avoided emissions were 6%. Thisnegative value, although indicates underachievement of goals
value, represents no lower performance than that of diesel
generation as the absolute value of the avoided emissions was
0.34kg-CO2/kWh.
For the smallest localities (Small) electrification without fuel
substitution brought along considerable increases in employment
generation and avoided emissions, as illustrated in Fig. 5c. Energy
system configuration for this type of localities included Wind,
Hydro and biogas (Dig.BiogasComb.) technologies, similar to
configurations for no fuel substitution in Large and Medium
localities. Substitution of traditional biomass resulted in similar
employment generation levels, while benefits in other attributes
decreased, in particular avoided emissions benefits considerably
reduced from 0.93 kg-CO2/kWh to less than 0.60 kg-CO2/kWh.
Nevertheless, share of technologies in electricity supply remained
almost unchanged.
The results for all the three locality types showed no variability
with respect to the priority structure chosen when no fuel
substitution is considered. Slight variations were present in
priority structures emphasizing environmental aspects (Land
priority and Emissions priority) for total substitution of traditional
biomass, resulting in inferior cost benefits and increased benefits
in other attributes. However, attainment of positive or negative
benefits was more correlated to whether fuel substitution
occurred or not, rather than being linked to the priority structurechosen.
4.2. Sensitivity analysis
In all types of localities renewable energy systems perfor-
mance declined in some aspects with total substitution of
traditional biomass, in particular the electricity cost. Therefore,
sensitivity analysis was conducted in order to inspect system
performance through benefits at rates of substitution between 0%
and 100%. A sample set of graphs showing trends in benefits by
attribute for Large localities are presented in Fig. 6. Table 6
summarizes the maximum levels of fuel substitution for which
positive benefits may be achieved in each attribute.Sensitivity analysis showed that in Large localities there is a
marked trade-off between benefits in electricity cost and employ-
ment generation. When higher priority is given to costs negative
benefits in electricity cost are obtained only for total (100%) fuel
substitution rates. On the other hand, employment generation
benefits fall under negative values from a 30% fuel substitution
rate onwards. For other priority structures this relation maintains,
but in an opposite way, showing negative values for benefits in
electricity costs from a 30% rate and positive employment benefits
for the whole range of fuel substitution rate. Land use and avoided
emissions benefits remained positive for all fuel substitution
rates, achieving maximum values at 80% for the former, and at 10%
and 20% for the later. Therefore, it can be said that if surpassing
electricity cost levels of diesel generation are to be avoided up to a
ARTICLE IN PRESS
Table 5
Performance outcomes for renewable energy system.
Attribute Locality Goal value No fuel
substitutionaTotal fuel substitution
Priorities (a),
(b), (c), (d)
(a) Cost
priority
(b) Job
priority
(c) Land
priority
(d) Emissions
priority
Electricity cost (US$/kWh) Large 0.136 0.136 0.146 0.158 0.158 0.158
Medium 0.209 0.21 0.326 0.326 0.443 0.443Small 0.216 0.212 0.226 0.226 0.238 0.238
Total 0.147 0.147 0.190 0.199 0.226 0.226
Employment (107 jobs/kWh) Large 5.13 5.15 4.55 5.14 5.14 5.14
Medium 5.13 7.65 11.20 11.20 15.37 15.37
Small 5.13 8.21 8.33 8.33 8.67 8.67
Total 5.13 5.53 6.21 6.64 7.61 7.61
Land use (m2/kWh/yr) Large 0.041 0.034 0.024 0.023 0.023 0.023
Medium 0.041 0.042 0.053 0.053 0.046 0.046
Small 0.041 0.041 0.043 0.043 0.041 0.041
Total 0.041 0.035 0.031 0.031 0.029 0.029
Avoided emissions (kg-CO2/kWh) Large 0.36 0.46 0.38 0.41 0.41 0.41
Medium 0.36 0.47 0.34 0.34 0.46 0.46
Small 0.36 0.93 0.56 0.56 0.57 0.57
Total 0.36 0.47 0.38 0.40 0.43 0.43
a Results in this case were the same for the four priority structures considered.
D. Silva, T. Nakata / Energy Policy 37 (2009) 30963108 3103
8/7/2019 Multi Objective Assessment of Rural Electrification in Remote Areas With Poverty Considerations 2009 Energy Policy
9/13
90% fuel substitution rate can be pursued. If costs are disregarded
then total fuel substitution is feasible. But, if benefits in both cost
and employment generation are desired only a 20% of fuel
substitution is possible.
The contrasting results in costs and employment benefits were
present again in Medium localities. Fuel substitution rates lower
or equal to 50% allowed positive electricity cost benefits. However,employment benefits over this fuel substitution rate grew
considerably from 50% to more than 100%, at the expense of
increasing electricity costs which surpassed goals value up to
60%. Regarding land use, maximum fuel substitution rates of 70%
for Cost and Job priorities, and 80% for other priority structures,
allowed positive benefits. Emissions benefits turned negative only
at total fuel substitution rates for priority structures centered on
non-environmental attributes. Nevertheless, these negative values
still represent gains for the system performance since the zero
benefit value represents a positive emission reduction amount
equivalent to the goals value (0.360 kg-CO2/kWh).
For Small localities, focusing on electricity cost benefits
permits a maximum 80% of fuel substitution for Cost and Job
priorities, and a maximum of 40% for other priority structures.
Targeting the reduction of land use rates enables up to 50% of fuel
substitution for Cost and Job priorities, and up to 100% of fuel
substitution for Land and Emissions priorities. Employment
generation and emissions benefits kept positive values over the
entire range of fuel substitution.
5. Discussion
Among the four attributes considered, the electricity cost
presented the major limitation for the achievement of a favorable
performance of renewable energy systems in the NIZ of Colombia.
Different to other attributes, negative values for benefits were
more likely in electricity cost, and only partial substitution of
traditional biomass was possible without rising electricity costs
over diesel generation electricity cost. Nevertheless, fuel substitu-
tion can still be afforded at high rates for Large and Small localities
depending on the priority structure. In general, viability of fuel
substitution is more likely for smaller localities as a result of lower
energy consumption rates compared to availability of energy
resources. If the cost of electricity is taken as an indicator of
ARTICLE IN PRESS
-100
0
100
0
100
-100 0 1000100
Electricity cost
benefits
Employment
generation
benefits
Land use
benefits
Avoided
emissions
benefits
-100
0
100
0
100
-100 0 1000100
Electricity cost
benefits
Employment
generation
benefits
Land use
benefits
Avoided
emissions
benefits
-100
0
100
0
100
-100 0 100 2000100200
Electricity cost
benefits
Employment
generation
benefits
Land use
benefits
Avoided
emissions
benefits
No fuel substitution Total fuel substitution
Fig. 5. Benefits as goal deviation rates: (a) Large localities; (b) Medium localities; (c) Small localities.
D. Silva, T. Nakata / Energy Policy 37 (2009) 309631083104
8/7/2019 Multi Objective Assessment of Rural Electrification in Remote Areas With Poverty Considerations 2009 Energy Policy
10/13
energy expenditure, it is clear that household expenditure rises as
a greater amount of fuel is replaced by electricity, given the low
price of traditional biomass, or more specifically, of firewood. The
electricity cost, which ranged between US$0.136/kWh to
US$0.216/kWh, is considerably higher than firewood price, which
is US$0.062/kWh (considering a conversion efficiency of 10% for
traditional firewood stoves). Furthermore, these costs are higher
than the average price of electricity being supplied in the NIZ,
corresponding to US$0.168/kWh (Unidad de Planeacion Minero
Energetica, 2000a). In terms of household energy expenditures,
fuel substitution imply an increase in electricity cost proportional
to the firewood consumption rates observed in each locality, as
can be seen in Table 7. For example, for Large localities energy
expenditure almost doubles, while for Medium localities
expenditures increase three-fold to five-fold. Initially, this
economic drawback resulting from fuel transition may be seen
as a hardship in the short term. The payback from reducing both
indoor air pollution and firewood collection time, translated into
lesser health risks and more free time for other activities, may not
be economically evident but in the mid or long term. However,
ARTICLE IN PRESS
-20
-10
0
10
20
0 20 40 60 80 100
Electric
itycostbenefit(%)
Fuel substitution rate (%)
Cost priority Job priority Land priority Emissions priority
0
20
40
60
80
100
0 20 40 60 80 100
Landusebenefit(%)
Fuel substitution rate (%)
-20
-10
0
10
20
0 20 40 60 80 100
Employmentgenerationbenefit(%)
Fuel substitution rate (%)
0
20
40
60
80
100
0 20 40 60 80 100
Avoidedemiss
ionsbenefit(%)
Fuel substitution rate (%)
Fig. 6. Sensitivity analysis results for Large localities benefits: (a) electricity cost benefits; (b) employment generation benefits; (c) land use benefits; (d) avoided emissions
benefits.
Table 6Maximum rates of fuel substitution obtained from the sensitivity analysis.
Locality Attribute Maximum rate of fuel substitution (%)
(a) Cost priority (b) Job priority (c) Land priority (d) Emissions priority
Large Electricity cost 90 20 20 20
Employment 20 100 100 100
Land use 100 100 100 100
Avoided emissions 100 100 100 100
Medium Electricity cost 50 50 50 50
Employment 100 100 100 100
Land use 70 70 80 80
Avoided emissions 90 90 100 100
Small Electricity cost 80 80 40 40
Employment 100 100 100 100Land use 50 50 100 100
Avoided emissions 100 100 100 100
D. Silva, T. Nakata / Energy Policy 37 (2009) 30963108 3105
8/7/2019 Multi Objective Assessment of Rural Electrification in Remote Areas With Poverty Considerations 2009 Energy Policy
11/13
this is not to mean that fuel transition is completely unacceptable
for people dwelling in remote areas. Rural people are likely to
allocate savings for higher energy expenditures and accept the
financial risk involved if technology adoption is considered in the
planning of rural electrification, namely by the inclusion of
households into the decision-making process.
Regulatory policies based on financial mechanisms, such as
subsidies in the electricity tariff or in the initial costs of
technologies through international aid among others, may be
needed to improve sustainability of electrification with renew-
ables. Another approach could be the transaction of other benefits
directly as cost reductions in the electrification project. This
alternative may prove helpful for avoided emissions and land use
benefits, which could be internalized by means of a Clean
Development Mechanism (CDM) as part of carbon mitigation
and carbon sequestration initiatives. Regarding actual implemen-
tation of the CDM in the NIZ of Colombia, it has been reported that
actual transaction costs of carbon emissions limits feasibility of
projects working under the CDM only to large electricity outputs,like in highly populated areas or a group of localities served by a
common plant. Thus, other instruments derived from climate
change international agreements, such as special climate change
fund (SCCF), have to be considered in order to promote
electrification in the NIZ at a larger extent based on carbon
reduction initiatives (Oficina Colombiana para la Mitigacion del
Cambio Climatico (OCMCC), 2003). In addition, other alternatives
for fuel substitution need to be considered. For example,
introduction of improved cooking stoves using firewood instead
of electric stoves is a promising alternative in many countries, as
these devices can be manufactured by people living in rural areas,
increasing income generation opportunities. However, it has to be
remarked that the total substitution of traditional biomass for
electricity is rather a process than a one-single-step transition. Ithas been found that several poor families in developing countries
that gain access to electricity use the service selectively,
principally for lighting and communication devices. In this regard
three main determinants in the transition to modern forms of
energy have been identified: the fuel availability, affordability and
cultural preferences (IEA, 2002). Accordingly, the shift from
traditional biomass to electricity is expected to be encouraged
by the continuous rise of income rates among rural people.
Employment generation was used in the multi-objective model
in order to introduce a factor directly related to the impact that
energy access may have on poverty reduction. However, the
quantification of this attribute was surrounded by several
uncertainties. As a consequence, the model deployed in the study
is too limited for the assessment of impacts of energy access on
income poverty. Instead, the model is more suitable to address the
impact on other poverty aspects, such as energy supply infra-
structure development and substitution of traditional biomass.
The most significant benefit from fuel substitution with
renewables was the reduction of greenhouse gases, estimated by
means of the avoided emissions. Table 7 summarizes these
benefits in absolute values and taking into account emissions
from combustion of firewood in traditional stoves. CO2 emissions
reduction accounted for over 190,000 t-CO2/yr in Large localities,
68,00077,000 t-CO2/yr in Medium localities, and 14,000 t-CO2/yr
in Small localities. Total emissions reductions, which summed
between 268,000 and 287,000 t-CO2/yr, are larger than actual CO2emissions reported in the NIZ due to the consumption of fuels,
equivalent to 116,983t-CO2/yr (Oficina Colombiana para la
Mitigacion del Cambio Climatico (OCMCC), 2003). Nevertheless,
these figures are relatively small compared to the entire emissions
in the Colombian energy sector, which summed more than 55
million t-CO2 by the year 1994 (Instituto de Hidrologa and
Meteoreloga y Estudios Ambientales (IDEAM), 2001).Recognizing economic and social benefits regarding the
creation of jobs, the reduction of energy expenditures and a
better utilization of energy sources provided by the substitution of
traditional biomass, may allow the inclusion of RETs implementa-
tion in local development programs. In that sense, the success of
RETs penetration in rural areas of Colombia and, in general, of
developing countries, requires the commitment of the national
government and the international community: the former,
through the creation and enforcement of regulatory policies
promoting the growth of RETs market; and the later by means of
cooperation programs facilitating technology transfer towards the
use renewable energy resources.
The application of GP under an energy systems analysis context
provides a more straightforward method to integrate and assessquantitatively different aspects of energy technologies in rural
communities. For example, compared to the sustainable livelihoods
approach used by Cherni et al. (2007), the application of GP has a
stronger quantitative foundation, since the set of indices deployed in
that study are constructed based on the views of rural people rather
than on exact measurements, for example, of the energy resources
and the energy demand rates. Thus, the process governing the
selection of technologies in that study involves a considerable level
of uncertainty. In addition, the methodology proposed by Cherni et
al. (2007) is able to evaluate only a set of eight alternatives
previously defined, reducing the possibility to evaluate several
combinations of technologies as energy supply alternatives. Never-
theless, such an approach is more complete in the estimation of the
impacts on rural communities given that it includes a broader range
ARTICLE IN PRESS
Table 7
Energy expenditure and emissions reduction for diesel and renewable energy schemes.
Feature Locality Diesel generation Renewable energy system
No fuel
substitution
Total fuel
substitution
No fuel substitutiona Total fuel substitution
Priorities (a), (b), (c),
(d)
(a) Cost
priority
(b) Job
priority
(c) Land
priority
(d) Emissions
priority
Energy expenditure (US$/yr/household)
Large 255 427 255 457 496 496 496Medium 232 634 232 988 988 1341 1341
Small 148 366 147 383 383 404 404
Emissions reduction (103 t-CO2/
yr)bLarge 0 65.3 38.8 187 196 196 196
Medium 0 28.4 5.3 67.6 67.6 77.3 77.3
Small 0 3.9 2.6 13.5 13.5 13.6 13.6
Emission factor for traditional biomass equivalent to 0.273kg-CO2/kWh was used.a Results in this case were the same for the four priority structures considered.b Emission reductions calculated as [(avoided emissions)(electricity supply)+(emissions traditional biomass)(substituted traditional biomass as final energy)].
D. Silva, T. Nakata / Energy Policy 37 (2009) 309631083106
8/7/2019 Multi Objective Assessment of Rural Electrification in Remote Areas With Poverty Considerations 2009 Energy Policy
12/13
8/7/2019 Multi Objective Assessment of Rural Electrification in Remote Areas With Poverty Considerations 2009 Energy Policy
13/13
International Energy Agency (IEA), 2002. Energy and poverty. In: World EnergyOutlook 2002. International Energy Agency, Paris. Available: /http://www.iea.org//textbase/nppdf/free/2000/weo2002.pdfS.
International Energy Agency (IEA), 2004. Energy and development. In: WorldEnergy Outlook 2004. International Energy Agency, Paris. Available: /http://www.iea.org//textbase/nppdf/free/2004/weo2004.pdfS.
International Energy Agency (IEA), 2006. Energy for cooking in developingcountries. In: World Energy Outlook 2006. International Energy Agency, Paris.Available: /http://www.iea.org/textbase/nppdf/free/2006/weo2006.pdfS.
Kanagawa, M., Nakata, T., 2007. Analysis of the energy access improvement and itssocio-economic impacts in rural areas of developing countries. Ecological
Economics 62 (2), 319329.Kanagawa, M., Nakata, T., 2008. Assessment of access to electricity and the socio-
economic impacts in rural areas of developing countries. Energy Policy 36 (6),20162029.
Kanniappan, P., Ramachandran, T., 2000. Goal programming model for sustainableelectricity production from biomass. International Journal of Energy Research24, 118.
Kemmler, A., Spreng, D., 2007. Energy indicators for tracking sustainability indeveloping countries. Energy Policy 35 (4), 24662486.
Mezher, T., Chedid, R., Zahabi, W., 1998. E nergy resource allocation using multi-objective goal programming: the case of Lebanon. Applied Energy61 (4),175192.
Modi, V., McDade, S., Lallement, D., Saghir, J., 2005. Energy services for theMillennium Development Goals. World Bank, UNDP, New York.
Moreno, B., Lopez, A.J., 2008. The effect of renewable energy on employment. Thecase of Asturias (Spain). Renewable and Sustainable Energy Reviews 12 (3),732751.
Oficina Colombiana para la Mitigacion del Cambio Climatico (OCMCC)Ministeriode Ambiente, Vivienda y Desarrollo Territorial, Instituto de Planificacion y
Promocion de Soluciones Energeticas (IPSE), 2003, Estrategia para laimplementacion del mecanismo de desarrollo limpio en zonas no interconec-tadas (in Spanish). Bogota, Colombia.
Pachauri, S., Mueller, A., Kemmler, A., Spreng, D., 2004. On measuring energypoverty in Indian households. World Development 32 (12), 20832104.
Painuly, J.P., 2001. Barriers to renewable energy penetration: a framework foranalysis. Renewable Energy 24, 7389.
Pohekar, S.D., Ramachandran, M., 2004. Application of multi-criteria decisionmaking to sustainable energy planninga review. Renewable and SustainableEnergy Reviews 8, 365381.
Ramanathan, R., Ganesh, L.S., 1995. Energy resource allocation incorporatingqualitative and quantitative criteria: an integrated model using goal program-ming and AHP. Socio-Economic Planning Sciences 29 (3), 197218.
Ruiz, B.J., Rodrguez-Padilla, V., 2006. Renewable energy sources in the Colombianenergy policy, analysis and perspectives. Energy Policy 34 (18), 36843690.
Schneiderjans, MarcJ., 1995. Goal Programming: Methodology and Applications.
Kluwer Academic Publishers, Massachusetts.Silva, D., Nakata, T., 2008. Renewable technologies for rural electrification in
Colombiaa multiple objective approach. International Journal of EnergySector Management 2 (1), 139154.
Spreng, D., 2005. Distribution of energy consumption and the 2000 W/capitatarget. Energy Policy 33 (15), 19051911.
Unidad de Planeacion Minero Energetica, 2000a. Establecimiento de un planestructural, institucional y financiero, que permita el abastecimiento energe-tico de las Zonas No Interconectadas con participacion de las comunidades y elsector privado. Centros poblados, caracterizacion energetica y agrupacion (inSpanish). UPME, Bogota, Colombia.
Unidad de Planeacion Minero Energetica, 2000b. Linea base geo-referenciada parala formulacion del plan de suministro energetico en las Zonas No Inter-conectadas de Colombia (in Spanish). UPME, Bogota, Colombia.
Unidad de Planeacion Minero Energetica, 2003. Plan energetico nacional,Estrategia energetica integral, Vision 20032020 (in Spanish). UPME, Bogota,Colombia.
United Nations, 2000. United Nations Millennium Declaration. Resolution. UN,
New York.Zapata, J., Bayona, L., 2000. Nuevo esquema de organizacion para el suministroenergetico en las Zonas No Interconectadas de Colombia (in Spanish).Escenarios y EstrategiasUPME. UPME, Bogota, Colombia.
ARTICLE IN PRESS
D. Silva, T. Nakata / Energy Policy 37 (2009) 309631083108
http://www.iea.org//textbase/nppdf/free/2000/weo2002.pdfhttp://www.iea.org//textbase/nppdf/free/2000/weo2002.pdfhttp://www.iea.org//textbase/nppdf/free/2004/weo2004.pdfhttp://www.iea.org//textbase/nppdf/free/2004/weo2004.pdfhttp://www.iea.org/textbase/nppdf/free/2006/weo2006.pdfhttp://www.iea.org/textbase/nppdf/free/2006/weo2006.pdfhttp://www.iea.org//textbase/nppdf/free/2004/weo2004.pdfhttp://www.iea.org//textbase/nppdf/free/2004/weo2004.pdfhttp://www.iea.org//textbase/nppdf/free/2000/weo2002.pdfhttp://www.iea.org//textbase/nppdf/free/2000/weo2002.pdf