Energies: Analysis of The Brazilian Energy Efficiency Program For Electricity Distribution Systems
Energies: Analysis of The Brazilian Energy Efficiency Program For Electricity Distribution Systems
Energies: Analysis of The Brazilian Energy Efficiency Program For Electricity Distribution Systems
Article
Analysis of the Brazilian Energy Efficiency Program
for Electricity Distribution Systems
Rogério Diogne de Souza e Silva 1, * ID
, Rosana Cavalcante de Oliveira 2 and
Maria Emília de Lima Tostes 1
1 Graduate Program in Electrical Engineering, Federal University of Para, Belem 66075-110, Brazil;
tostes@ufpa.br
2 Laboratory of Artificial Intelligence and Decision Support, Institute for Systems and Computer Engineering,
Technology and Science, 4200 Porto, Portugal; rosana.c.oliveira@inesctec.pt
* Correspondence: rogeriodss@ieee.org; Tel.: +55-91-3201-7634
Abstract: In this paper, a methodology was developed to analyze the results of energy efficiency
programs used in the electricity distribution sector in Brazil. Analyzing the gains obtained through
the investments made, and classifying which actions resulted in the best performance, contributed
to decision making on the best allocation of investments to obtain the greatest energy savings. The
Brazilian Energy Efficiency program was analyzed with a developed non-parametric model, using
the data envelopment analysis method, and the categories of projects with better performance were
determined. A database of the results from 1704 projects, from 2008 to 2016 in the Energy Efficiency
program in Brazil´s electricity distribution sector, was used. The results obtained show that the best
performance was achieved by projects in the industrial and cogeneration categories; however, in
Brazil these constitute only 4.24% of the projects presented and 5.28% of the total investments in
the last eight years, indicating a need to review the regulatory strategies for energy efficiency in
this country.
1. Introduction
Energy efficiency is a high potential and low-cost source that can reduce total energy costs by
about 20%, implementing energy efficiency measures substantially lowers costs compared to the
initiation of new sources of supply and offers an excellent mechanism for the reduction of carbon
emissions [1]. According to the International Energy Agency (IEA), energy efficiency not only saves
energy and reduces emissions, but also has other potential benefits, such as energy security, energy
prices, macroeconomic impacts, industrial productivity, public budget, and health and well-being,
among others [2].
Reducing energy demand has been the main focus of energy efficiency policies in several countries.
This goal can be reached through the improvement of products and process efficiency in energy
consumption on both the demand and supply sides. Energy conservation goals can be attained by
using measures for improving efficiency in appliances, services, and processes that consume energy in
all sectors. Such measures lead to policies that focus on the demand side, for instance, by establishing
patterns for buildings, requisites for efficiency in equipment, and promoting energy management in
industries and other consumption units. However, it is imperative that such tools be well-projected,
linked with realistic goals, and precisely monitored [2].
The use of energy efficiency initially depends on public and/or private strategies that define
incentive actions as a way for market growth, to a point where this is measurable and factual and can
be used for policy making, and consequently for national energy planning as if this were a primary
source of energy. According to the IEA, energy efficiency measures are now recognized as energy
sources. This statement is based on the fact that, in 2010, investment in energy use reduction by
member countries of the IEA, relative to investments between 1974 and 2010, was greater than in any
other single energy source. IEA member countries saved USD $490 per capita on average, and a total
of USD $540 billion was spent on energy in 2015 as a result of energy efficiency gains since 2000, and
1.5 billion tons of CO2 (Gt CO2 ) in 2015, and 13 Gt CO2 since 2000, were prevented from being released
into the environment [3].
Globally, the impacts of electricity distribution company programs, which mainly consist of
demand-side management (DSM), are being seen on the promotion of energy efficiency. However,
on initial analysis, it is illogical that a provider whose product is the commercialization of electricity
would invest in the reduction of the consumption of its product. Ongoing breakthroughs in the market
for electricity, such as the inclusion of competitive markets in distribution, smart grids, and distributed
generation, have signaled changes in the business model of providers, including substantial effects
on the measures of energy efficiency in the sector. Behrangrad [4] evaluated DSM business models in
different electricity market segments of several countries and stated that business models are influenced
by several factors, such as market regulation, power system characteristics, and infrastructure. He
highlighted that such factors actually depend on changes, with the proliferation of infrastructure,
smart grids, and distributed generation, with intermittent renewable energies and storage devices, all
affecting the DSM business models.
The development of regulatory tools is necessary for the success of inserting energy efficiency
strategies into the energy planning of the world’s power systems. Apajahlati et al. [5] stated that
energy conservation measures contribute significantly to the reduction of climate change and energy
insecurity. They evaluated the influence of the 2006/32/UE and 2012/27/UE directives, which
implemented mandatory investment into energy efficiency in Europe. According to Rosenow [6], the
United Kingdom was the first country in Europe to have mandatory measures for energy savings,
starting in 1994. A specific method was followed in the residential sector, which was considered the
sector targeted for energy savings and the reduction of pollutant gas emissions, resulting in electric
power savings in the order of 90 TWh/year. Abrardi et al. [7] proposed an optimal tariff model that
would lead to energy efficiency on the demand side. They indicated that the energy provider or the
regulatory body must propose compatible tariff options for both parties; for instance, if the provider’s
energy efficiency activities have a high impact on the reduction in demand, the consumer must pay
a high demand price, but a lower price on the consumption of electricity. On the other hand, if the
provider does not carry out measures for energy efficiency or if these have a low impact, the price of
the demand tariff will be lower and the price of electricity consumption will be higher. Thollander and
Palm [8] developed an energy efficiency decision support system for industrial energy management,
considering the fact that the most commonly used model for energy management systems, the model
for consumption–product, is insufficient for the understanding of management practices in industrial
energy. They suggested that the success of such measures is due to the use of hybrid models, beginning
with social and technical analyses of the industrial environment, prior to the implementation of a
management system.
Internationally, it is quite evident that energy issues are inseparable from climate change, so much
so that in the 21st Conference of the Parties of United Nations (COP 21), held in Paris in December 2015,
189 countries participated, resulting in 162 Intended Nationally Determined Contributions (INDC),
which consist of proposals for measures and definitions of targets for greenhouse gas reduction.
The accord between countries aims to limit the increase in global average temperature to below 2 ◦ C
compared to pre-industrial levels. The agreement replaces other goals outlined in the Kyoto Protocol
and must be reviewed every five years. However, from among the proposed INDCs, only 143 mention
energy efficiency measures, and among these, only a few dictate new policies for the sector [9].
Thoyre [10] and Slazvik and Csete [11] analyzed how energy efficiency may be used in public policies
Energies 2017, 10, 1391 3 of 19
considering climate change. Thoyre used data from North Carolina, USA, as a reference to compare
36 American states that adopted energy efficiency policies. Thoyre deduced that such projects tend
to emphasize energy efficiency on the demand side, and rarely produce substantial impacts on the
supply side, thus limiting the potential for electricity consumption reduction and consequently, climate
benefits. Slazvik and Csete evaluated the Hungarian energy policy and its role in the economic recovery
plan of the Hungarian government. The authors began with an examination of the National Energy
Strategy 2030, which takes into account climate policy with regard to adaptation and mitigation, and
lists the sources of renewable energies as important tools for reaching strategic goals. The Strategy
intends to increase energy savings as a short-term solution with bearable costs and earn a relatively
fast return on investment.
Calili et al. [12] developed a model for predicting impacts of energy savings in Brazil. Their
results indicate that savings running on 1% electric power per year over a five-year period will result
in profits between USD $70 and 80 million per year and avoids emissions of 3500 MtCO2eq per year.
In contrast, the newly inaugurated USD $7.8 billion Belo Monte hydroelectric power plant had a return
on investment over a period of 30 years, indicating that investments in energy efficiency have a better
return on investment than a new hydroelectric power plant. Brazil proposed emissions reductions of
37% and 43% until 2025 and 2030, respectively, as its INDC. One of the measures for attaining these
goals was a 10% increase in energy efficiency in the country through the National Plan for Energy
(PNEf) [13,14]. Brazil has specific programs for energy savings and efficient use of energy. Of these,
two stand out: The National Program for Energy Conservation (PROCEL) and the Energy Efficiency
Program (PEE). The former is overseen by Electrobras, and the latter by the providers of electric power
managed by the Brazilian Regulatory Electricity Agency (ANEEL). The PROCEL has allowed for
savings of 92.2 TWh for the last 30 years, whereas the last 18 years of PEE projects totaled 9.48 TWh per
year in electricity consumption reduction and 2.95 GW in demand reduction during peak hours [15,16].
Although these programs were relatively successful, Brazil is not in a comfortable position, according
to the American Council for Energy Efficiency Economy (ACEEE), which published a report analyzing
several aspects and indicators. This report lists a ranking of 23 countries in order of their performance
in energy efficiency; Brazil comes in second to last. One of the indicators is the cost per capita for
each country with regard to energy efficiency measures; Brazil invests USD $3.29, whereas Germany,
placed first in the same ranking, invests USD $318.49. Another reason for this large gap is linked to
energy management; Brazil only has 23 companies that are ISO 50.001 certified, whereas Germany has
3402 companies with this certification [17].
In this context, a methodology was developed to analyze the results of energy efficiency programs
applied to the electricity distribution sector, based on Brazil’s situation. Analyzing the gains obtained
through the investments made and classifying which actions obtained the best performance, helped
contribute to deciding which investments to make to obtain greater energy savings. A mathematical
model was constructed using a non-parametric method known as Data Envelopment Analysis (DEA).
DEA is a technique that is applicable to problems with multiple inputs and outputs and is based on
an analysis of the efficiency boundary. Its use has been proposed for energy planning in distribution
systems since it defines indicators that aid in decision making. Several authors use this technique,
in an isolated manner or in hybrid structures, linked with other modeling techniques. Pacudan and
Guzman [18] developed a model using DEA to simulate the effect of energy efficiency policies in
the energy distribution system in the Philippines. Through this model, the authors classified the
performance of 15 energy providers regarding their technical and operational efficiency and the
influence of energy efficiency measures on the demand side. Results indicated that energy savings
greater than 19.8% allow for an increase in operational productivity for the providers.
Chen and Zhang [19] developed a model using DEA linked with Slack-Based Measures (SBM) to
determine the Efficiency Factor of the regional economy in China, considering the use of land, energy,
and greenhouse gas emissions. Alvarez and Rudnick [20] developed a model using DEA, focusing
on input in conjunction with a variable return scale (VRS) to analyze the efficiency of regulatory
Energies 2017, 10, 1391 4 of 19
mechanisms in Italy, Brazil, and Chile and determined the increase in energy efficiency in the electric
distribution grids.
Oh and Hildreth [21] elaborated on a stochastic and deterministic model using stochastic frontier
analysis (SFA) and DEA to measure the effectiveness of initiatives to save energy in the automotive
industry. They presented the best projects for energy consumption reduction. Through a hybrid model
with rough set theory (RST) techniques and data envelopment analysis (DEA), Lee and Tong [22]
determined the efficiency in the transfer of energy in photovoltaic systems. Furthermore, using baseline
data from 30 Chinese provinces, Zeng et al. [23] developed a DEA model to solve an efficient carbon
allocation problem through a license concession for the emission of financial assets mechanism.
The remainder of this paper is structured as follows: in Section 2, Brazilian Energy Efficiency
Programs are discussed; in Section 3, the methodology used is presented; the results of the data
analysis are in Section 4; and in Section 5, the developed model and the results obtained are examined.
Finishing with Section 6, concluding remarks are presented.
600,000 10%
8%
500,000
6%
400,000 4%
2%
GWh
300,000 Electricity
0% Consumed
200,000 -2% Variation
-4%
100,000
-6%
0 -8%
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
Year
Figure 1. Electricity consumption history in Brazil between 1999 and 2015, adapted from Brazilian
Figure
Energy 1. Electricity
Balance consumption
2009, year history in
2008, and Brazilian BrazilBalance
Energy between 1999
2015, and2014.
year 2015, adapted from Brazilian
Energy Balance 2009, year 2008, and Brazilian Energy Balance 2015, year 2014.
Law 9.991/2000 has been adapted to address how resources are allocated and types of priority
Law 9.991/2000 has been adapted to address how resources are allocated and types of priority
projects [28,29]. Other laws and decrees later emerged in Brazil, highlighting the importance of strategy
projects [28,29]. Other laws and decrees later emerged in Brazil, highlighting the importance of
in the public policy formulation for energy efficiency. Table 1 lists the main regulatory tools for energy
strategy in the public policy formulation for energy efficiency. Table 1 lists the main regulatory tools
efficiency that emerged in Brazil.
for energy efficiency that emerged in Brazil.
Table 1. Main energy efficiency regulatory tools for in Brazil.
Table 1. Main energy efficiency regulatory tools for in Brazil.
Year
Year Law or Decree
Law or Decree
1996 9.427:
1996 9.427:Creation
Creation ofof thethe Brazilian
Brazilian Regulatory
Regulatory Electricity
Electricity Agency.
Agency.
9.991: Determines issues relating to investments in research and development (R&D) and
9.991: Determines issues relating to investments in research and development (R&D) and
2000 energy
2000 energy efficiency
efficiency onon behalf
behalf ofof concessionary,
concessionary, permission,
permission, andand authorized
authorized companies
companies in in
thethe
electric power
electric power sector. sector.
10.295: Known as the Law of Energy Efficiency, establishes minimum efficiency indices for
10.295: Known as the Law of Energy Efficiency, establishes minimum efficiency indices for
machines
machines andand appliances
appliances manufactured
manufactured oror marketed
marketed ininthe
thecountry,
country, based
based onon relevant technical
relevant
indicators,
technical which consider
indicators, the useful
which consider thelife of the
useful lifeequipment, also states
of the equipment, alsothat, one
states year
that, oneafter
year
2001 after
2001 publication
publication of the
of thelevels of energy
levels of energy efficiency,
efficiency, a program
a program of targets will
of targets be be
will established
established forfor
its
itsprogressive
progressiveevolution
evolutionand andobliges
obligesmanufacturers
manufacturersand andimporters
importersof ofthe
theequipment
equipmentto totake
takethe
the necessary
necessary measures
measures toto comply
comply with
with thethe maximum
maximum levels
levels ofof energy
energy consumption
consumption and
and minimum
minimum energy efficiency.
energy efficiency.
10.438:Institutes
10.438: Institutes the the PROINFA-
PROINFA- Incentive
Incentive Program
Program forfor Alternative
Alternative Sources
Sources of of Electric
Electric Power
Power ; ;
main
main goal
goal is is
toto increase
increase thethe share
share ofof electric
electric power
power produced
produced byby Autonomous
Autonomous Independent
Independent
2002
2002
Producers
Producers (PIA)
(PIA) (via(viawind
wind power,
power, small
small hydroelectric
hydroelectric power
powerplants, and
plants, andbiomass)
biomass) in in
thethe
National
National Interconnected
Interconnected Electric
Electric System.
System.
4.508:Decides
4.508: Decidesononthe the specific
specific regulations
regulations that
that define
define the the minimum
minimum levels
levels ofof energy
energy efficiency
efficiency
2002 forfor
2002 thethe three-phase
three-phase electric
electric motors
motors of squirrel
of squirrel cagecage induction
induction rotors—produced
rotors—produced domestically
domestically or or
abroad,
abroad, forforsale
saleoror use useininBrazil—of
Brazil—of mandatory
mandatory character.
character.
12.212:Creates
12.212: Creates thethe Social
Social Tariff
Tariff that
that establishes
establishes discounts
discounts forfor consumers
consumers whowho areare classified
classified to to
be low-income paying and determines that concessionaires
be low-income paying and determines that concessionaires and permissions for the and permissions for the
2010
2010
distribution
distribution ofofelectric
electric power
power must
must apply;
apply; at least
at least60%60%of the resources
of the resourcesfrom PEEPEE
from for for
consumer
consumer
units
unitsare funded
are funded byby thetheSocial
SocialTariff.
Tariff.
Resolution
Resolution 482:482: Establishes
Establishes thethe conditions
conditions forfor micro
micro andand mini
mini generation
generation of energy
of energy distributed
distributed
2012
2012
toto the
the electric
electric power
power distribution
distribution system
system andandthethe electric
electric power
power compensation
compensation system.
system.
Currently, in the Brazilian Congress, the law project 1.917/2015 provides free market strategies
for the distribution sector, in which the consumer may choose the desired distribution company and
Energies 2017, 10, 1391 6 of 19
Currently, in the Brazilian Congress, the law project 1.917/2015 provides free market strategies
for the distribution sector, in which the consumer may choose the desired distribution company and
opt for lower priced concessionaires, whose cost reductions for electricity distribution may lead to a
reduction in the final price paid by the consumer.
As a consequence of Law 9.991/2000, the Energy Efficiency Program focuses on promoting
the efficient use of electric power in all economic sectors by means of projects that demonstrate the
importance and economic viability of improving the energy efficiency of equipment, processes, and
final uses of energy. It is aligned with the Brazilian energy planning, present in the National Plan
for Energy Efficiency (PNEf), which sees it as the main source of revenue for energy efficiency in the
Brazilian electricity distribution system [14].
The PEE is mandatory for electricity distribution companies with a market share greater than
1.000 GWh/year, and at least 50% of its resources must be applied to the two consumer classes
with the greatest market share. In profit-seeking consumer units, projects must be conducted via
energy performance contracts, with the total value invested in the project being refunded. However, for
projects in non-profit-seeking consumer units, investments are non-refundable. One of the comparative
advantages of using non-profit-seeking consumer units compared to other forms of financing is that
there is no interest to be collected—only monetary correction. Furthermore, consumers must refund
only the implementation costs. The main goal of the energy performance contract is to avoid the transfer
of public resources to profit-seeking consumer units and reduce the lack of trust in relation to the
success of the energy efficiency measures. Since the payment is conditioned to the success of the chosen
measure, ANEEL adopts the International Protocol for Measurement and Verification of Performance
by Efficiency Valuation Organization (EVO) as a measurement and verification methodology [16].
Consumption classes considered for the PEE are as follows: rural, industrial, trade and services,
residential, public service, public agencies, and public lighting. Projects geared according to these
classes must be contracted via Public Call, which is a mechanism for implementing energy efficiency
measures. The distribution company issues a public notice, requesting energy efficiency project
presentation, wherein competition occurs in a process including auctioning quality and price. The
main goal of these public projects is to motivate the societal participation in general, providing a
greater diversity of projects and incentives for the creation of an energy efficiency market. They are
also aimed at making the decision process for projects and consumers benefited by the PEE more
transparent and democratic, thus promoting greater societal participation [16].
The main choice criterion, among projects participating in the Public Call, is based on the
cost–benefit relation (RCB) of each project (Equation (1)); benefits are considered to be the valuation
of the economized energy and the reduction of demand in peak hours during the project life, and
the costs are all the investments for accomplishing said project. As a general rule, a project will be
approved if it reaches an RCB lesser than or equal to 0.8 [16].
The RCB is represented by
CA T
RCB = (1)
BA T
where CAT is the total annual cost, given in $/year; BAT is the total annual benefit, also in $/year.
3. Methodology
In order to evaluate the Brazilian energy efficiency program, a methodology was used to analyze
the projects, classify and quantify the variables, model the problems and determine the best performing
projects, and this was culminated with recommendations for the improvement of the program.
In the flowchart in Figure 2, the steps of the methodology used are presented.
Energies 2017, 10, 1391 7 of 19
Energies 2017, 10, 1391 7 of 19
Figure
Figure 2. Methodology
2. Methodology flowchart.
flowchart.
The detailed
The steps
detailed of of
steps thethe
methodology areare
methodology asas
follows:
follows:
• • Database:
Database: ToToanalyze
analyze thethe
PEE,
PEE,a database from
a database from2008
2008until
untilMarch
March2016
2016was
was used,
used,comprising
comprising
1704 projects forwarded to the ANEEL by the electric utilities with information
1704 projects forwarded to the ANEEL by the electric utilities with information about about potential
potential
projects. The projects were classified into 14 categories (Appendix A) and 7
projects. The projects were classified into 14 categories (Appendix A) and 7 final uses,final uses, and
andthethe
database
database included
included5 5variables:
variables:saved
savedenergy,
energy,peak
peakdemand
demandreduction,
reduction, total
total cost of project,
cost of project, cost
cost of
of saved
savedenergy,
energy, and RCB. The database is available on the ANEEL website, a
and RCB. The database is available on the ANEEL website, a download address is download
address is available
available in the Supplementary
in the Supplementary MaterialsMaterials
section. section.
• Routine treatment and data filtering: In this step, routines were developed for the treatment
• Routine treatment and data filtering: In this step, routines were developed for the treatment and
and selection of the data to be analyzed, and consisted of the filtering of spurious and
selection of the data to be analyzed, and consisted of the filtering of spurious and inconsistent
inconsistent data, which was due to a failure of data processing by energy concessionaires. Only
data, which was due to a failure of data processing by energy concessionaires. Only the projects
the projects with the data for all the variables were selected, and subsequently, the RCB was
with the data for all the variables were selected, and subsequently, the RCB was adopted with a
adopted with a parameter of less than 2. Above this value, the data of the projects were not
parameter of less than 2. Above this value, the data of the projects were not included, because the
included, because the cut-off limit for acceptance of the project by ANEEL is equal to 1, so the
cut-off limit for acceptance of the project by ANEEL is equal to 1, so the costs of projects with an
costs of projects with an RCB above this value must be assumed by the consumer unit.
RCB above this value must be assumed by the consumer unit.
• Data analysis: This step involved consolidating the data, classified by category and the final use
• Data analysis: This step involved consolidating the data, classified by category and the final use
of the energy, and subsequent analysis. The total value of projects, the energy saved, the
of the energy, and subsequent analysis. The total value of projects, the energy saved, the reduction
reduction in peak demand, the total cost of the projects, the cost of energy saved, and the
in peak demand, the total cost of the projects, the cost of energy saved, and the cost–benefit ratio
cost–benefit ratio are presented for each category. The analysis of the cited variables, in relation
are presented for each category. The analysis of the cited variables, in relation to the number of
to the number of projects carried out per category, was performed. The same analysis was
projects carried out per category, was performed. The same analysis was performed considering
performed considering the final energy use classification.
the final energy use classification.
• Data envelopment analysis modeling: A model was developed using the data envelopment
• Data envelopment analysis modeling: A model was developed using the data envelopment
analysis (DEA) method, treating the problem like a black box, given its unknown production
analysis (DEA) method, treating the problem like a black box, given its unknown production
function, but with the inputs and outputs well known. As a principle, inputs were minimized
function, but with the inputs and outputs well known. As a principle, inputs were minimized and
and outputs maximized. The chosen methodology to model the problem, to identify types of
outputs maximized. The chosen methodology to model the problem, to identify types of measures
measures for energy efficiency, takes into consideration the characteristics of the available
for energy efficiency, takes into consideration the characteristics of the available database that
database that results in the best program. The methodology consists of determining the
results in the best program. The methodology consists of determining the efficiency boundary that
efficiency boundary that defines the most efficient project categories, and consequently the
defines the most efficient project categories, and consequently the measures of energy efficiency,
measures of energy efficiency, composed of the categories with the best input–output relation.
All other categories outside of the boundary curve are classified as a function of their distance
from the curve.
Energies 2017, 10, 1391 8 of 19
composed of the categories with the best input–output relation. All other categories outside of
the boundary curve are classified as a function of their distance from the curve.
DEA is a non-parametric method that uses part-based linear programming to calculate the most
efficient units in a given sample set of decision units. The decision units that form the boundary
also include those considered to be less efficient. Each unit is compared with a linear combination of
units that produce at least as much from each output as the inefficient unit with the smallest possible
quantity of input. The operational efficiency is calculated with a score ranging from 0 to 1, in which
1 is given to the most efficient units that form the boundary curve, to be determined according to
Equations (2) and (3) [30].
∑ v r z r0
MaxE0 = r (2)
∑ u i x i0
i
This is subject to
∑ vr zrj
r
≤ 1; j = 1 . . . n
∑ ui xij (3)
i
vr , ui ≥ 0; r = 1 . . . s; i = 1 . . . m.
Equation (2) corresponds to the objective function, where E represents the efficiency index, given
by the weighted sum of the outputs and divided by the weighted sum of the inputs. In Equation (3),
there are restrictions for optimizing the objective function. vr represents the weights assigned to the
outputs r; ui represents the weights assigned to inputs i; zrj represents the observed quantity of output
r for unit j; xij is the observed quantity of input i for unit j; n is the total number of units under analysis;
s represents the total outputs; and m, the total inputs. The variables E0 , Zr0 , xi0 are relative to the first
decision unit, denominated unit 0. The weights u and v are vectors, obtained by Equations (2) and (3),
which involves finding the values of u and v such that the efficiency measure for which the nth project
is maximized. The model allows total flexibility in the selection of weights, and imposes non-negativity
and feasibility constraints of the decision units.
This method has two outcomes in relation to the returns to scale: the constant returns to scale
(CRS) model, which, by its own denomination, does not alter the returns to scale, and the variable
returns to scale (VRS) model, which considers production efficiency situations with variation of scale
and does not assume proportionality among inputs and outputs [30]. In this work, the VRS model was
selected, in which the variables may be correlated; hence, the results showed the best categories when
a larger number of variables was considered. The DEA constraints can be configured for the specific
characteristics of the problem, referencing the model in relation to the minimization of inputs or the
maximization of outputs [20,30]. This objective function can be evaluated by the two options, and in
this way, models for inputs and outputs were simulated, according to Equations (4) and (5):
s
MaxE0 = ∑ vr zr0
r =1
subject to
s m
∑ v r z r0 − ∑ u i x i0 ≤ 0 . (4)
r =1 i =1
m
∑ u i x i0 = 1
i =1
ui , vr ≥ 0
MinE0
subject to
n
E0 xi0 − ∑ λ j xij ≥ 0; i = 1...m
j =1 (5)
n
−zr0 + ∑ λ j zrj ≥ 0; r = 1...s
j =1
λ j ≥ 0; j = 1...n
The values obtained for E assign an efficiency value for each decision-making unit (DMU). DMUs with
E < 1 are classified as inefficient, if E = 1, the DMUs will lay on the efficiency boundary. The variable λj
is a convexity constant, which represents the weights assigned to the outputs and inputs, guaranteeing
the convexity of the efficiency frontier.
The developed model shows the best type of project to be executed by the electricity distribution
companies, taking into account three main aspects: the lowest cost of energy saved, the largest amount
of energy saved, and the greatest peak demand reduction. The cost of energy saved per project was
selected as the input, the energy saved per project and the peak demand reduction hours per project as
outputs, and the project categories as DMUs, according to Table 2.
• Simulation and analysis: This step involved the simulation and determination of the efficiency
boundary between categories. From the results obtained, the best categories of the project were
identified, and the categories that do not reach the efficiency boundary will be evaluated and the
necessary gains for these to reach the efficiency curve were determined.
25,000
20,000
MWh/year
15,000
10,000
5,000
Figure
Figure 3.
3. Energy
Energy saved
saved per
per project
project for
for each
each category.
category.
We observed that cogeneration had 3.78 times the energy savings per executed project than the
We observed that cogeneration had 3.78 times the energy savings per executed project than
low-income category, which has a greater number of projects executed. The low efficiency of public
the low-income category, which has a greater number of projects executed. The low efficiency
of public agency projects should be discounted, although they represent one quarter of the total
projects analyzed.
Energies 2017, 10, 1391 11 of 19
Energies 2017, 10, 1391 11 of 19
agency projects should be discounted, although they represent one quarter of the total projects
analyzed.
Figure 44 shows
Figure showsa agraph
graphwith
withthethe peak
peak demand
demand reduction
reduction (PDR)
(PDR) per project.
per project. In the In
linethe line
graph,
graph, the highest reduction rates occurred in cogeneration (2357.43 kW/project), low-income
the highest reduction rates occurred in cogeneration (2357.43 kW/project), low-income (2073.94
(2073.94 kW/project),
kW/project), residential
residential (1455.70 (1455.70 kW/project),
kW/project), pilotkW/project),
pilot (843.59 (843.59 kW/project),
and solarand solar heating
heating (380.28
(380.28 kW/project).
kW/project).
2500
2000
1500
kW
1000
500
Figure
Figure 4.
4. Peak
Peak demand
demand reduction
reduction per
per project
project for
for each
each category.
category.
The public lighting and industrial categories were characterized by operation during peak
hours;The public lighting
however, they did andnotindustrial categories
observe this were characterized
correlation by operation
in the data analyzed, with during
averagepeak hours;
reduction
however, they did not observe this correlation in the data analyzed, with average
rates of 292.97 kW/project and 190.81 kW/project, respectively. Several other types of consumers reduction rates of
292.97 have
often kW/project and 190.81
high rates kW/project,
of operation respectively.
at peak Several
hours, such as other
publictypes of consumers
service (water and often have
sewage
high rates of operation at peak hours, such as public service (water and sewage
companies), trade and services (malls, supermarkets, schools, business offices, and others), and companies), trade and
servicesagencies
public (malls, supermarkets, schools, business
(hospitals, universities, schools,offices,
and andpoliceothers), and public
stations), although agencies
there (hospitals,
is a low
universities, schools, and police stations), although there is a low
representation of these categories in the demand reduction results at peak hours. representation of these categories in
the demand
The costreduction
of energy results
savedatbypeak hours. was analyzed using the relationship between the total
category
The cost
investment of energy
realized and saved by category
the energy savings.was analyzed
Figure 5 shows using the relationship
a graph of the cost ofbetween the total
energy saved, in
investment realized and the energy savings. Figure 5 shows a graph
which the industrial category, with 184.92 USD/MWh·year, had the lowest energy cost savings,of the cost of energy saved,
in which the
followed industrial
by rural with category, with 184.92 USD/MWh
219.74 USD/MWh·year, residential ·year,
withhad the lowest
261.34 energy cost
USD/MWh·year, savings,
trade and
followedwith
services by rural with
280.67 219.74 USD/MWh
USD/MWh·year, pilot·year, residential
projects with 282.79with USD/MWh·year,
261.34 USD/MWh ·year,
and trade and
cogeneration
services
with withUSD/MWh·year.
290.05 280.67 USD/MWh The·year,
only pilot projects
project on thewith
supply282.79
sideUSD/MWh ·year, and
cost was 3476.92 cogeneration
USD/MWh·year,
with 290.05among
the largest USD/MWh ·year. The only project on the supply side cost was 3476.92 USD/MWh·year,
the categories.
the largest among the categories.
The categories with the highest number of projects were low-income and public agencies
havingThecosts
categories
of 353.85with the highest number
USD/MWh·year of projects
and 367.41 were low-income
USD/MWh·year, and public agencies having
respectively.
costsAsof 353.85
for theUSD/MWh ·year and
RCB, the results are367.41 in Figure·year,
USD/MWh
detailed respectively.
6, with the rural, cogeneration, industrial,
As for the RCB, the results are detailed in Figure
low-income, and residential categories standing out. Projects with 6, with theanrural,
RCB cogeneration,
above one indicate industrial,
costs
low-income, and residential categories standing out. Projects with an RCB above
greater than the benefits, denoting the failure of actions, at least in regards to energy efficiency. one indicate costs
greater than the benefits,
In addition denoting the
to the categories, the database
failure ofprovides
actions, ata least in regards
classification ofto energyinefficiency.
projects relation to the
final energy use, classified into seven variables, according to Figure 7. The followingin
In addition to the categories, the database provides a classification of projects relation
final energy to use
the
final energy
results standuse,out. classified
Heating had intoeight
sevenprojects
variables,
withaccording
an average to energy
Figure 7. The following
savings per project final energy
of 2398.39
use results stand out. Heating had eight projects with an average energy
MWh/year, lighting with 1247 projects and 2069.18 MWh/year, and electric motors with 133 projects savings per project of
2398.39 MWh/year,
and 1314.61 MWh/year. lighting with 1247 projects and 2069.18 MWh/year, and electric motors with 133
projects and 1314.61 MWh/year.
Energies 2017, 10, 1391 12 of 19
Energies 2017, 10, 1391 12 of 19
Energies 2017, 10, 1391 12 of 19
4000
4000
3500
3500
3000
3000
2500
2500
2000
2000
1500
1500
1000
1000
500
500
0
0
Figure
Figure 5.
5. Cost
Cost of
of energy
energy saved
saved for
for each
each category.
category.
Figure 5. Cost of energy saved for each category.
2.5
2.5
1.93
2.0 1.93
2.0
Benefit
1.5
Benefit
1.5
Cost
1.0 0.86
Cost
3000
3000
2500
2500
2000
2000
MWh/year
MWh/year
1500
1500
1000
1000
500
500
0
0
Figure
Figure 7.
7. Energy
Energy saved
saved per
per project
project for
for final
final energy use.
energy use.
Figure 7. Energy saved per project for final energy use.
900
900
800
800
700
700
600
600
500
kWkW
500
400
400
300
300
200
200
100
100
0
0
Figure 8. Peak demand reduction (PDR) per project for each final energy use.
Figure
Figure 8.
8. Peak
Peak demand
demand reduction
reduction (PDR)
(PDR) per
per project
project for
for each
each final energy use.
final energy use.
Energies 2017, 10, 1391 14 of 19
Energies 2017, 10, 1391 14 of 19
1.6 1.52
1.4
1.2
1.0
Cost benefit
0.84
0.77 0.78
0.8
0.2
0.0
Figure
Figure 9.
9. Cost–benefit
Cost–benefit relationship
relationship for
for each
each final
final energy use.
energy use.
Simulations
Simulations Results
Results and
and Analysis
Analysis
Two
Two simulations
simulationswere
wereperformed,
performed,one
onein in
relation to the
relation inputs
to the and and
inputs the other to thetooutputs.
the other Only
the outputs.
two categories were determined as most efficient in both: industrial and cogeneration, as
Only two categories were determined as most efficient in both: industrial and cogeneration, as shown in
Figure
shown 10.
in Figure 10.
1.2
1.0
Efficiency Frontier
0.8
0.6
0.4
0.2
0.0
Input oriented
Output oriented
In Figure 10, there are 10 inefficient categories: residential, rural, low-income, pilot project,
In Figure
trade and 10, there
services, are 10
public inefficient
service, publiccategories:
agencies,residential, rural, solar
public lighting, low-income,
heating,pilot
and project,
supply trade
side.
and services, public service, public agencies, public lighting, solar heating, and
The difference between input and output shows how much each variable should be increased supply side. The
for
difference
projects tobetween
reach theinput
sameand output shows
performance how much
as projects each
on the variableboundary.
efficiency should be In
increased
Table 5,for
theprojects
results
to
arereach the same
presented, usingperformance as projects
the model with on the efficiency
the minimization boundary. In Table 5, the results are
of inputs.
presented, using the model with the minimization of inputs.
Table 5. Percentage increase required to reach the efficiency boundary using the input-oriented
Table 5. Percentage increase required to reach the efficiency boundary using the input-oriented model.
model.
To
To reach
reachthe
theefficiency
efficiencyboundary,
boundary,the the
costcost
of energy per project
of energy invested
per project in the supply
invested in theside category
supply side
should decrease by 94.50% and the energy saved per project should be increased
category should decrease by 94.50% and the energy saved per project should be increased by by 693.26%, the most
inefficient
693.26%, theDMU.mostForinefficient
the solar heating
DMU. category,
For the the cost
solar of energy
heating should the
category, decrease by 79.51%
cost of energy and the
should
energy
decreasesaved should and
by 79.51% be increased
the energyby 646.01%. The public
saved should lightingby
be increased category
646.01%.should
The reduce the cost
public lighting
of energyshould
category by 55.37% andthe
reduce increase
cost ofthe energy
energy bysaved
55.37%byand
156.35%. The
increase public
the agencies
energy category
saved by should
156.35%. The
reduce the cost of
public agencies energy by
category 49.67%,
should increase
reduce the energy
the cost saved
of energy by by 200.68%,
49.67%, and reduce
increase peaksaved
the energy demand by
200.68%, and reduce peak demand by 2.07%. The residential category has an inefficient DMU that is
closest to the boundary, and should reduce the cost of energy by 5.76% and increase the energy
saved per project by 151.68%.
Energies 2017, 10, 1391 16 of 19
by 2.07%. The residential category has an inefficient DMU that is closest to the boundary, and should
reduce the cost of energy by 5.76% and increase the energy saved per project by 151.68%.
The results obtained with the model that maximizes outputs are presented in Table 6.
The low-income category is closest to the efficiency curve, which needs to increase the energy saved by
278.44%, reduce the peak demand by 13.67%, and reduce the cost of energy by 18.03%. Residential
projects should increase their energy savings by 200.75% and reduce the peak demand by 21.30%.
The public agencies category should decrease the cost of energy by 21.06%, increase the saved energy
by 204.61%, and reduce the peak demand by 1161.06%. The most inefficient DMU was the trade
and services category, requiring a 3035.10% increase in energy saved and a 3031.86% reduction in
peak demand.
Table 6. Percentage increase required to reach the efficiency boundary for each category using the
output-oriented model.
In analyzing the results of projects from previous years, the establishment of weights for the
categories that do not reach the efficiency boundary are suggested; that is, these project categories,
depending on their diminished performance relative to the boundary, must attain better energy savings
and greater reduction during peak hours for their approval. In the category of public agencies and
public services, the resources are non-refundable, without requiring performance contracts, which can
result in lower savings.
With these results, it can be inferred that a redistribution of investments should be applied to the
resources, aiming at greater energy savings and reduced demand at peak hours. Among the categories,
95.76% of the total projects executed and 94.72% of the investments applied to date are ineffective.
6. Conclusions
Based on the obtained results, we verified that the prioritization of project categories may optimize
the reduction in electric power consumption in the Brazilian distribution system. The results indicate
that project categories, and consequently the measures for energy efficiency, are diverse, and it is
paramount that this diversity is considered when prioritizing measures for the Energy Efficiency
Program. It was observed that the categories with better performance only account for 4.24% of the
total projects submitted. The industrial and cogeneration categories were the most efficient, and while
the residential and low-income categories were close, they still require optimization.
Techniques and methodologies used to quantify the depreciation of the measures taken to improve
energy efficiency must be stipulated in the pilot project, as well as the addition of new data variables
to be collected by the ANEEL. Notably, the lifespan of equipment does not define or guarantee the
continuity of energy efficiency initially obtained, and may become indispensable to determine adequate
parameters for measuring the depreciation of results in energy efficiency measurements.
The DSM tariff strategies, such as offering differentiated tariffs based on peak hours, must be
evaluated by the appropriate criteria. Considering only the reduction in consumed energy, the behavior
Energies 2017, 10, 1391 17 of 19
Appendix A
Definition of project categories:
• Low-income: Actions that aim to encourage the efficient use of energy by low-income consumers,
which are imposed by the social tariff of electric energy, for those whose per capita family income
is less than half the Brazilian minimum wage (approximately $140/month).
• Public lighting: Actions in facilities that involve roads, squares, traffic lights, served
by municipalities.
• Pilot project: Unpublished and promising projects that aim to consolidate some energy efficiency
technology or for use on a small scale.
• Trade and services: Actions in consumer units involved in commercial activity or the provision of
services. Included in this category are entities providing non-profit services. It is characterized by
the containment of several consumer units, such as shops, malls, schools, universities, hospitals,
restaurants, non-governmental organizations, etc.
• Industrial: These are projects focused on process optimization, introduction of efficient motors,
and energy management systems, among others, in industrial consumer units served by the
electricity concessionaire, regardless of the size of the industry.
Energies 2017, 10, 1391 18 of 19
• Public agencies: Actions in consumer units of public organizations at the municipal, state,
and federal levels. It serves administrative buildings, schools, universities, and hospitals,
among others.
• Residential: Shares in residential consumer units and condominiums not classified as
low-income projects.
• Rural: Projects in consumer units with rural activity.
• Public service: Actions in consumer units of bodies that provide public services, such as the
collection, treatment, and supply of water, the collection and treatment of sanitary and solid
waste, electricity, and gas, among others.
• Solar heating: Implementing systems for the solar heating of water, instead of processes that use
electricity for the generation of heat.
• Supply side: Projects focused on the area of electricity generation and distribution, primarily by
incentivized sources.
• Cogeneration: Actions in consumer units with the potential to implement energy cogeneration
actions, classified as an incentive source.
• Educational: Projects aimed at spreading the concept of energy efficiency and
sustainable development.
• Municipal energy management: Projects aiming to improve the management of the final use of
electric energy in the municipalities.
© 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).