Beruflich Dokumente
Kultur Dokumente
Applied Energy
journal homepage: www.elsevier.com/locate/apenergy
h i g h l i g h t s
a r t i c l e i n f o a b s t r a c t
Article history: Bioethanol from sugarcane can be produced using first-generation (1G) or second-generation (2G) tech-
Received 10 February 2016 nologies. 2G technologies can increase the capacity of production per sugarcane mass input and are
Received in revised form 24 July 2016 expected to have a key role in future reductions of environmental impacts of sugarcane bioethanol. A
Accepted 28 July 2016
hybrid Input-Output (IO) framework is developed for Brazil coupling the System of National Accounts
and the National Energy Balance, which is extended to assess Greenhouse Gas (GHG) emissions. Life-
cycle based estimates for two sugarcane cultivation systems, two 1G and eight 2G bioethanol production
Keywords:
scenarios, are coupled in the IO framework. A multi-objective linear programming (MOLP) model is formu-
Input-output analysis
Hybrid modeling
lated based on this framework for energy-economic-environmental analysis of the Brazilian economic sys-
Multi-objective linear programming tem and domestic bioethanol supply in prospective scenarios. Twenty-four solutions are computed: four
Multi-sectoral economy-energy- ‘‘extreme” solutions resulting from the individual optimization of each objective function (GDP, employ-
environment models ment level, total energy consumption and total GHG emissions - 1G scenario), ten compromise solutions
Lignocellulosic bioethanol minimizing the distance of the feasible region to the ideal solution (1G, 1G-optimized and prospective 1G
+ 2G scenarios), and ten solutions maximizing the total bioethanol production (1G, 1G-optimized and
prospective 1G + 2G scenarios). Higher diesel oil and lubricants consumption in the mechanical harvesting
process has counterbalanced the positive effects of more efficient trucks leading to higher energy con-
sumption and GHG emissions. Lower overall employment level in the 1G + 2G scenarios is achieved such
that policies linked to reabsorption of sugarcane cutters in alternative activities are positive. Indirect
effects from maximizing the bioethanol production increase the total energy consumption and the GHG
emissions thus requiring efficiency measures and fossil energy substitution by cleaner sources. The
integrated- or country-based analysis of the whole economic system has complemented the process
design and process-based analysis, contributing to identify direct and indirect effects that can offset the
benefits. Direct and indirect effects on the whole economic system have to be considered in policies
and technological choices for prospective bioethanol production, since positive direct effects of 1G + 2G
plants can be counterbalanced by indirect impacts on other sectors, mainly from chemicals in the process.
Ó 2016 Elsevier Ltd. All rights reserved.
⇑ Corresponding author at: Federal Institute of Mato Grosso, Campus Diamantino, Rodovia Roberto Campos, 78400-970 Diamantino, Brazil.
E-mail address: ariovaldo.carvalho@dmt.ifmt.edu.br (A.L. de Carvalho).
1
http://www.uc.pt/efs.
2
http://www.uc.pt/en/org/inescc.
3
http://www.ifmt.edu.br.
http://dx.doi.org/10.1016/j.apenergy.2016.07.122
0306-2619/Ó 2016 Elsevier Ltd. All rights reserved.
A.L. de Carvalho et al. / Applied Energy 181 (2016) 514–526 515
Acronyms Notation
1G first generation bioethanol ba anhydrous bioethanol
1G + 2G combined and integrated 1G and 2G technologies bh hydrous bioethanol
2G second generation bioethanol ci change in inventories
CH4 methane debt public debt
CO2 carbon dioxide ec total energy consumption
EH enzymatic hidrolysis emp employment level
EIO-LCA economic input-output life cycle analysis exp exports
GDP Gross Domestic Product gdp gross domestic product
GHG Greenhouse Gases gdpcurr gross domestic product at current prices
IO Input-Output gfcf gross fixed capital formation
IOA Input Output Analysis ggb public administration global balance
IO-MOLP input-output multi-objective linear programming ghg total greenhouse gas emissions
ISO International Organization for Standardization gva gross value added
LCA Life Cycle Assessment imp imports
LP Linear Programming pc public consumption
MOLP Multi-objective Linear Programming rc resident consumption
NO2 nitrous oxide sc sugarcane
NPISH non-profit institution serving households ts taxes less subsides on products
PROALCOOL Brazilian Alcohol Program ydcurr resident’s disposable income at current prices
R$ Brazilian Real
WIS water insoluble solids
restrict the analysis, since upstream and downstream impacts that Some studies have developed linear programming (LP) models
can propagate through the economy are neglected. The cost and coupled with the IO framework for different purposes [50–52].
time required for (full and exhaustive) LCA studies are a shortcom- However, Multi-objective Linear Programming (MOLP) models
ing of this methodology, which has also motivated the research of coupled with IO framework can provide a more thorough assess-
methods that simplify the analysis while still providing consistent ment of different axes of evaluation of potential policies, enabling
information to satisfy the overall assessment goals [23]. In this to explore the trade-offs between competing objectives through
sense, the environmental extended input-output models can be a the computation of non-dominated solutions, i.e. solutions for
useful approach to overcome some weaknesses of LCA studies, which no other feasible solution exists improving the value of a
since the overall environmental impact of an entire sector of the given objective function without worsening the value of, at least,
economy can be accounted for. other objective function.
Input-Output Analysis (IOA) has been traditionally used to IO-MOLP models have been applied to study the impacts of
study the inter/intra-relationships among different sectors in the national and regional policies on the employment, water pollution,
economic system, describing the relationship between the inputs energy requirements, carbon dioxide (CO2) emissions, foreign trade
used and the outputs produced. Some authors have proposed dif- balance, etc. [53–56]. There are also IO-MOLP models using hybrid
ferent approaches to modify the Input-Output (IO) models to frameworks and external expansions of the IO model to assess en
account for environmental impacts using generalized IO models ergy-environment-economic-social objectives [57–60]. These
[24], economic ecological-models [25] and commodity by industry models are used as basis for the extended hybrid IO-MOLP model
models [26]. An external expansion of the IO framework can also herein presented and applied to the Brazilian economic system,
be made to incorporate the environmental impacts, assuming a which will be briefly outlined in the next section.
proportional relation between the output of the sectors and the The literature review revealed a lack of studies using the IO-
corresponding impact levels [27]. IO models have also been MOLP framework encompassing economic, energy, environmental
adapted for the analysis of the energy sector (e.g., [28–31]) and (especially GHG emissions) and social (especially employment
energy and environmental interactions (e.g., [32,33]). Some studies levels) spheres coupled with LCA estimates for energy commodi-
have developed basic and extended IO models to assess the inter- ties analysis, as well as for the assessment of the 2G bioethanol
actions of the Brazilian bioethanol sector with the overall Brazilian sector specifically devoted to the Brazilian economic system.
economic system for different purposes: to assess the impacts from Therefore, a comprehensive approach is proposed for prospective
sugar and bioethanol exports [34,35]; to estimate the impacts from analysis of the economic system and specific energy commodities,
new bioethanol factories [36]; to prospective studies of demand which applies extended and hybrid IO frameworks coupled with
increase [37]; to assess the impacts on socioeconomic indicators multi-objective modeling and LCA estimates. The aim is to assess
of different prospective technological mixes [38–40]. However, the impacts on the total energy consumption, GHG emissions, sev-
most of these studies have just performed an economic analysis eral economic variables and employment level associated with the
and, in a small number of cases, an analysis of the employment bioethanol production according to the current and prospective
level. Unfortunately, these studies have not deeply assessed the technologies (1G and 2G) in the horizon of 2018. The integration
energy and GHG aspects related to the bioethanol sector and the of LCA and hybrid IO-MOLP models herein proposed improves
whole economic system. Other studies using the IO framework the scope of analysis of energy products, especially 2G bioethanol.
have also addressed environmental and energy issues. Compeán This methodological framework allows incorporating different pro-
and Polenske [41] developed an environmentally extended interre- cesses into the model and expanding the boundaries of the analysis
gional input-output model using hybrid units accounting for to the entire economic system such that direct and indirect effects
energy- and carbon-intensities to analyze the direct and indirect in an integrated- or country-basis analysis are accounted for. Addi-
impacts of different bioethanol production technologies. However, tionally, the model overcomes the lack of studies integrating IO-
this study has a limited number of sectors and fuels, the model MOLP models devoted to energy products, particularly for the anal-
does not include the employment analysis and the impacts from ysis of 2G bioethanol systems, as well as for the integrated assess-
2G bioethanol are not assessed. Watanabe et al. [42] applied the ment of energy, environmental, economic and social spheres. Such
so-called Economic Input-Output Life-Cycle Analysis (EIO-LCA) to comprehensive modeling and analysis has not been developed pre-
the assessment of economic and GHG emissions of different 1G viously, especially applied to the Brazilian bioethanol system. This
and 2G sugarcane based bioethanol production technologies in methodological framework can be adapted to other economic sys-
Brazil. However, this study did not address the effects on the tems, countries and target sectors to assess the trade-offs between
employment level, as well as the impacts on the energy intensities those axes of evaluation and support the design of appropriate
and on the composition of the national energy mix. The EIO-LCA is policies. The methods used to estimate the sugarcane and bioetha-
one of the methods utilized to integrate the IOA with the LCA of nol scenarios as well as the hybrid IO model formulated in this
specific sectors, such as hybrid approaches [27,43]. study are presented in Section 2. Some illustrative results are pre-
Alternatively, there are some studies, although not linked with sented in Section 3. Conclusions and future developments are
the bioethanol system, which have used IO hybrid models to assess drawn in Section 4.
the energy sector (see e.g., [44,45]) and to assess the Brazilian eco-
nomic system (see e.g., [46–48]). It is important to refer that ‘‘hy-
brid” does not refer here to the linkage between IOA and LCA 2. Materials and methods
methodologies, but to the combination of physical and monetary
units into the transaction and technical coefficients matrices of 2.1. Sugarcane scenarios
the IOA. In this approach the basic IO model is adjusted and new
rows and columns are included in the transaction matrix to allo- The first step to build up the Extended Hybrid IO-MOLP model
cate the energy sectors (or commodities), substituting flows in is to construct the sugarcane sector (or disaggregate it from the
monetary units by flows in physical units for the energy sectors sector in which it appears), which is not explicitly available in
(or commodities) into the inter-sector transaction and final the Brazilian IO tables. For this purpose, the methodology proposed
demand matrices. These new matrices are then utilized to generate in [40] to estimate the average utilization and costs of the main
new technical coefficient matrices, which are then applied to the inputs (such as diesel oil, lubricants, agriculture inputs, salary,
IO model (see [46,49]). etc.) in the life cycle of bioethanol was jointly used with the total
A.L. de Carvalho et al. / Applied Energy 181 (2016) 514–526 517
the sugarcane cultivated using average agricultural technology of Inputs 2G-A 2G-B 2G-C 2G-D 2G-E 2G-F 2G-G 2G-H
Brazil in 2009, and a future sugarcane scenario considering opti- Sugarcane 0.45 0.43 0.43 0.39 0.38 0.33 0.35 0.33
mized conditions of cultivation, mechanical harvesting (with straw + leaves
removal and delivery to the bio-refinery) and transportation. Enzymes 0.11 0.12 0.13 0.19 0.21 0.26 0.21 0.26
Acid 0.02 0.03 0.03 0.02 0.03 0.03 0.03 0.03
Base 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.02
2.2. Bioethanol scenarios
a
The scenarios 2G-A to 2G-H represent the values for combined 1G + 2G plants
without mixing the material streams. The 1G process (autonomous distillery)
Results from [12] have been used in our study as input data to
comprises the traditional Melle-Boinot steps, while the 2G bioethanol production
eight scenarios regarding current and prospective 1G and 2G tech- from sugarcane bagasse and leaves using separate hydrolysis and fermentation
nologies. The simulations of [12] have considered that 1G plants [12].
utilize sugarcane as raw material (165 dry t/h), fresh water
(310 m3/h) and Bagasse to boiler (74 dry t/h), producing 46 m3/h
of bioethanol, 370 m3/h of Vinasse and 77 MWel of electricity.
The integration of 1G and 2G bioethanol production was carried the outputs of other sectors in fixed proportions in order to pro-
out sequentially investigating eight scenarios (2G-A to 2G-H) for duce its own single output and the inputs into a sector are propor-
process variables found to have significant impact on the plant tional to the total output of that sector [61,49]. The basic IO
design and capital cost. Firstly, a base case (Scenario 2G-A, repre- relationship is:
senting the combined 1G + 2G plant from [12]) is considered. Four x¼A xþy ð1Þ
combinations of enzyme dosage and enzymatic hydrolysis (EH)
residence time were investigated in scenarios 2G-B, 2G-C, 2G-D where A is a matrix of inter-sectoral direct requirements (or techni-
and 2G-E to identify the scenario with lowest bioethanol selling cal coefficients matrix) in which each element aij represents the
price. The resulting scenario 2G-E was then used as the starting value of input required from sector i to produce one monetary unit
point for the evaluation of leaves addition (Scenario 2G-F), heat worth output of sector j (i = 1, . . . , n; and j = 1, . . . , n), x is the vector
integration (Scenario 2G-G) and leaves addition together with heat of total outputs and y is the vector of the final demand.
integration (Scenario 2G-H) (see Table 1). In this step, the energy flows in the Brazilian National Energy
The estimates of [12] for eight different configurations of com- Balance [62] are incorporated into the 2009 Brazilian IO table
bined 1G + 2G bioethanol production plants (2G-A to 2G-H) were [63] by considering artificial sectors (see also [57,58]). For this pur-
used to calculate the cost of the main inputs utilized in each com- pose some adjustments and inclusion of new rows and columns in
bined 1G + 2G process to produce a monetary unit of bioethanol the IO table are necessary in order to incorporate the different
(presented in Table 2). Those estimates were used to compute energy sectors (or commodities). Thus a new inter-sectoral (or
eight different sets of technical coefficients for the Brazilian transaction) matrix and vectors for final demand and total output
bioethanol sector weighted by the total bioethanol produced in are obtained, in which energy flows are considered in physical
the 1G + 2G plants for each scenario (see Appendix A). These sets quantities of energy (tons of oil equivalent, toe) and all non-
of technological coefficients of bioethanol production were used energy sector flows are measured in monetary units. The technical
to substitute the technical coefficient of the bioethanol sector into coefficients for the inter-sectoral matrix and final demand vectors
the 2009 Brazilian technical coefficient matrix in each optimiza- with hybrid units (monetary/physical, monetary/monetary, physi-
tion. Moreover, as referred to previously, two sugarcane produc- cal/physical or physical/monetary) are then obtained from this
tion scenarios will be considered and also inserted into the new IO framework. The adjustments performed in the IO frame-
technical coefficient matrix according to the scenarios to be opti- work provide: a square matrix with 112 activity sectors split into
mized. Therefore each bioethanol plant uses different sugarcane 53 economic sectors, 5 energy producing sectors, 5 artificial sectors
types as input: for the current scenario (1G) the current sugarcane used for distributing the energy consumed by each means of
scenario is considered, whereas for 1G-optimized and 1G + 2G (2G- transportation and 49 artificial sectors for energy commodities; 6
A to 2G-H) scenarios the future sugarcane scenario is considered. column vectors with the components of final demand (exports –
exp; public consumption – pc; resident consumption – rc; gross
2.3. Hybrid IO-MOLP model fixed capital formation – gfcf; and changes in inventories – ci); 1
column vector for competitive imports (considered for energy
The third step to build the Extended Hybrid IO-MOLP model commodities only); and 6 row vectors for the primary inputs
consists in rearranging the IO tables to obtain technical coefficients (wages, gross mixed income, gross operating surplus, other
in hybrid units. The IO model assumes that each sector consumes production taxes and other production subsides).
Table 1
Summary of the conditions for the 1G + 2G scenarios investigated (based on [12]).
Parameters 1G + 2G scenarios
2G-A 2G-B 2G-C 2G-D 2G-E 2G-F 2G-G 2G-H
1G heat Integration Yes Yes
Leaves Yes Yes
WIS (%)a 7 7 7 7 7 7/7 7 7/14
Enzyme dosage Low Low Low Highb Highb Highb Highb Highb
Hydrolysis time (h) 72 72 48 72 48 72 72 72
Glucose yield from EH, (%)a 47 47 42 73 66 73/96 73 73/76
Steam Dryer No Yes Yes Yes Yes Yes Yes Yes
1G + 2G bioethanol, L/dry t of sugarcane 329 341 337 360 360 435 379 420
a
When two values are given the first is for bagasse and the second for leaves.
b
The high enzyme loading is twice the low enzyme loading.
518 A.L. de Carvalho et al. / Applied Energy 181 (2016) 514–526
An external expansion of the IO model is made to estimate GHG The employment level (emp) is obtained by using labor gross
emissions from energy combustion, industrial processes, agricul- productivity coefficients for each sector.
ture activities, waste management, wastewater treatment and dis- The total energy consumption for energy purposes (ec) is
charge, and fugitive emissions. In this step, which is based on the obtained from the sum of national and imported energy excluding
IPCC [64] methodology, carbon dioxide (CO2), methane (CH4) and the energy consumed for non-energy purposes, i.e. the energy
nitrous oxide (N2O) emission factors are used in combination with commodities that are used as raw materials (e.g., primary fuels
the activity level of specific sectors and final demand components. used to produce secondary fuels, products for the chemical indus-
These estimates give a vector with the environmental impacts per try, etc.). Specific technical coefficients are applied to the interme-
unit of output of the sectors and the final demand, considering the diary consumption and final demand.
corresponding Global Warming Potential (100-year horizon: 25 for
CH4 and 298 for NO2) relative to CO2 [65]. 2.5. Objective functions
Finally, the MOLP model based on IOA proposed by [57] is used
as basis and adapted to the Brazilian reality. In this model a set of The objective functions considered in the model are: maximiza-
443 variables is considered including: the total output and exports tion of GDP (at 2009 basic prices – due to the most recent data
(at basic and consumer prices) of each sector (a total of 336 vari- available) (gdp), maximization of the employment level (emp),
ables); competitive imports of energy commodities (49 variables); minimization of the total energy consumption for energy purposes
several economic variables (58 variables). The model includes (ec) and minimization of the total GHG emissions (ghg).
(internal) coherence constraints derived from the IOA and other Firstly, each objective function was optimized individually for
sets of constraints associated with the structure of the economic the 1G scenario to obtain four distinct non-dominated solutions
system, employment and energy consumption, which are briefly (solutions 1–4). These solutions provide an overview of the range
described below. Further details about the MOLP model can be of variation of the objective values within the non-dominated
found in [59]. region. Then, for the 1G scenario, the solution that minimizes the
Tchebyscheff distance to the ideal solution is computed (solution
5), which allows obtaining a ‘‘balanced” compromise non-
2.4. Model constraints dominated solution to the MOLP model since this surrogate scalar
function enables to obtain the non-dominated solution that mini-
Coherence constraints are used to determine that the interme- mizes the maximum deviation to the best value reachable in the fea-
diate consumption and final demand of each activity sector shall sible region for each objective function (which cannot be obtained
not exceed the corresponding total amount available from national simultaneously since the objective functions are conflicting).
production and competitive imports. The same procedure is then applied for the 1G-optimized (sub-
The Gross Domestic Product (GDP, expenditure approach) is stituting the 2009-average coefficient vector of the sugarcane sec-
computed considering the final demand minus imports at FOB tor by the optimized one, solution 6) and prospective technologies
(free on board) prices (including tourism). The GDP (production (1G + 2G) scenarios (substituting the 2009-average coefficient vec-
approach) is computed by the sum of gross value added (gva) tor of the sugarcane sector by the optimized one – in all scenarios –
and the total of taxes less subsides on products (ts) that are not and the coefficient vector of the bioethanol sector by the corre-
included in the production. sponding 1G + 2G vector – in each scenario, solutions 7–14). In this
The gross value added is given by the sum of wages, gross step, two additional constraints are included in the model in order
mixed income, gross operating surplus, other production taxes to establish the values of GDP (at 2009 basic prices) and the total
minus other production subsides. (anhydrous + hydrous) bioethanol production at the same levels
Taxes less subsidies on goods or services are calculated for the obtained in the 1G scenario. This procedure enables to compare
intermediate consumption and final demand items. the impacts due to the specific technologies on the other objectives
The model also establishes some assumptions for several con- (and selected decision variables) assuring the same economic per-
sumption relations: the households’ consumption on the territory formance and bioethanol production level.
includes the consumption on the territory by resident and non- In order to compare the potential of these new technologies in
resident households; the residents’ consumption includes the con- increasing the total bioethanol production, as well as the impacts
sumption of households and Non-profit Institutions Serving on the economic, environmental, energy and employment levels,
Households (NPISH); the resident households’ consumption on the maximization of the total (anhydrous + hydrous) bioethanol
the territory is linearly dependent on the available income; and production in each of the 10 different scenarios (1G, 1G-optimized,
the tourism imports is given as a proportion of the residents’ 2G-A, 2G-B, 2G-C, 2G-D, 2G-E, 2G-F, 2G-G and 2G-H) are considered
household consumption. (solutions 15–24). Table 3 summarizes the objective functions and
The GDP at current prices (gdpcurr) is estimated considering the scenarios considered in the 24 non-dominated solutions obtained.
components of GDP (expense approach) at constant prices and the The MOLP model has been supplied with realistic data gathered
corresponding deflators. Additionally, the consumption of goods from several sources [12,62–64,66–68]. The IO framework has been
and services by the public administration at current prices and compiled using a workbook structure of multiple linked spread-
the gross fixed capital formation (gfcf) at current prices are exoge- sheets (using Microsoft Excel), while the optimizations was per-
nously defined. formed using Opensolver (OS; http://opensolver.org). The results of
The residents’ disposable income at current prices (ydcurr) is all optimizations and the discussion will be presented in the next
computed by subtracting the public administration and (non- section.
financial and financial) corporations’ disposable incomes from
the National Disposable Income. 3. Results and discussion
Public debt (debt) is given by the summation of the previous
period debt with the symmetrical value of the public administra- 3.1. Individual optimization of each objective function
tion’s global balance, plus an adjustment variable.
Public administration’s global balance (ggb) is computed by Tables 4 and 5 outline the results for the objective functions and
subtracting the public administration’s expenditures from the pub- main variables of 4 non-dominated solutions corresponding to the
lic administration’s revenues. individual optimization of each objective function (GDP, employ-
A.L. de Carvalho et al. / Applied Energy 181 (2016) 514–526 519
ment level, total energy consumption and total GHG emissions) for
the 1G scenario. The 2009 values of each objective function and
Sol.
24
x
selected variables are included in Tables 4 and 5 for comparison
purposes.
Sol.
23
x
3.2. Minimization of the distance of the feasible region to the ideal
Sol.
solution
22
x
Tables 6 and 7 outline the results of the objective functions and
Sol.
21
x
main variables of 10 non-dominated compromise solutions
obtained by minimizing the distance of the feasible region (using
Sol.
20
x
for the 1G, 1G-optimized and prospective 1G + 2G scenarios (solu-
Sol.
x
Solutions that minimize the distance of the feasible region to
the ideal solution present objective function values that are well
Sol.
18
employment objectives;
an increase of approximately 1.11–1.26 million of employees
Sol.
14
nario. Although the more efficient and higher load capacity trucks
Sol.
reliable methodology and data shortage, the model has not consid-
2
2G-H
2G-D
2G-G
2G-A
2G-B
2G-C
2G-E
2G-F
Bioethanol
Bioethanol
Bioethanol
Bioethanol
Bioethanol
Bioethanol
Bioethanol
Bioethanol
Bioethanol
Bioethanol
Optimized
Sugarcane
Max Total
Max emp
Max gdp
Distance
Min ghg
Scenarios
functions
Bioethanol
Sugarcane
Table 4
Values of the objective functions in the non-dominated solutions computed in the individual optimization of each objective function.
Notation Description Units 2009 Max gdp Max emp Min ec Min ghg
a
gdp Gross Domestic Product MR$ 3,239,404 4,351,337 4,351,337 4,133,118 4,126,511
emp Employees Emp. 103 41,208 55,200 55,319 52,883 52,766
ec Energy consumption – total ktoe 205,679 273,740 273,526 256,800 257,254
ghg GHG emissions – total Gg CO2eq 2,406,175 2,738,023 2,752,640 2,660,362 2,659,322
The values in bold represent the ideal solution of each objective function.
a
Constant 2009 prices.
Table 5
Values of the main variables in the non-dominated solutions computed in the individual optimization of each objective function.
Notation Description Units 2009c Max gdp Max emp Min ec Min ghg
a
sc Sugarcane MR$ 21,098 26,612 26,703 24,569 25,050
ba Bioethanol Anhydrous GL 7014 8836 8922 7014 8325
bh Bioethanol Hydrous GL 19,089 24,929 24,633 22,521 23,357
rc Resident́s consumption MR$a 1,979,751 2,668,603 2,574,038 2,475,401 2,494,412
pc Public consumption MR$a 687,001 922,816 922,816 922,816 922,816
ci Changes in inventories MR$a 7471 10,082 10,221 9527 9405
gfcf Gross fixed capital formation MR$a 585,317 861,608 861,608 816,599 786,297
imp Imports MR$a 360,847 436,793 436,188 411,438 408,272
exp Exports MR$a 355,653 345,186 439,284 339,267 340,663
ts Taxes MR$a 445,025 593,996 592,700 558,856 557,783
gva Gross value added MR$a 2,794,379 3,757,340 3,758,636 3,574,262 3,568,728
gdpcurr GDP current MR$b 3,239,404 6,735,909 6,776,766 6,447,358 6,486,833
ydcurr Residents’ disposable income MR$b 2,091,051 4,726,263 4,558,783 4,384,091 4,417,761
ggb Public adm. global balance MR$b 149,309 130,120 139,629 245,756 245,774
debt Public debt MR$b 2,167,856 4,558,698 4,568,207 4,674,334 4,702,954
a
Constant 2009 prices.
b
Current 2018 prices.
c
The values for gdpcurr, ydcurr, ggb and debt in the column 2009 are in current 2009 prices.
Table 6
Values of the objective functions in the non-dominated solutions computed by minimizing the distance of the feasible region to the ideal solution.
Notation Units Sol. 5 Sol. 6 Sol. 7 Sol. 8 Sol. 9 Sol. 10 Sol. 11 Sol. 12 Sol. 13 Sol. 14
1G 1G Opt. 2G - A 2G - B 2G - C 2G - D 2G - E 2G - F 2G - G 2G - H
gdp MR$a 4,236,658 4,236,658 4,236,658 4,236,658 4,236,658 4,236,658 4,236,658 4,236,658 4,236,658 4,236,658
emp Emp. 103 54,024 54,024 54,029 54,024 54,024 54,016 54,013 53,998 54,005 54,000
ec ktoe 264,410 264,410 264,379 264,409 264,405 264,453 264,474 264,588 264,519 264,581
ghg Gg CO2eq 2,695,334 2,695,337 2,695,190 2,695,333 2,695,314 2,695,541 2,695,641 2,696,177 2,695,853 2,696,145
a
Constant 2009 prices.
Table 7
Values of the main variables in the non-dominated solutions computed by minimizing the distance of the feasible region to the ideal solution.
Notation Units Sol. 5 Sol. 6 Sol. 7 Sol. 8 Sol. 9 Sol. 10 Sol. 11 Sol. 12 Sol. 13 Sol. 14
1G 1G Opt. 2G - A 2G - B 2G - C 2G - D 2G - E 2G - F 2G - G 2G - H
sc MR$a 25,536 25,536 25,021 24,423 24,473 23,455 23,152 21,565 22,238 21,739
ba GL 8439 8439 8445 8444 8444 8443 8485 8502 8461 8516
bh GL 23,867 23,867 23,862 23,862 23,862 23,863 23,822 23,805 23,846 23,791
rc MR$a 2,551,100 2,551,101 2,549,135 2,549,557 2,549,827 2,551,814 2,552,048 2,551,951 2,551,637 2,552,113
pc MR$a 922,816 922,816 922,816 922,816 922,816 922,816 922,816 922,816 922,816 922,816
ci MR$a 9870 9870 9875 9874 9874 9869 9479 9238 9700 9108
gfcf MR$a 861,608 861,608 861,608 861,608 861,608 861,608 861,608 861,608 861,608 861,608
imp MR$a 425,656 425,656 426,333 426,260 426,309 426,363 426,093 425,689 425,965 425,681
exp MR$a 336,660 336,660 339,306 338,811 338,589 336,653 335,758 335,209 336,262 334,911
ts MR$a 575,572 575,574 575,636 575,661 575,664 575,732 575,729 575,776 575,754 575,768
gva MR$a 3,661,086 3,661,083 3,661,022 3,660,997 3,660,994 3,660,926 3,660,929 3,660,882 3,660,904 3,660,890
gdpcurr MR$b 6,544,865 6,544,865 6,544,865 6,544,865 6,544,865 6,544,865 6,544,865 6,544,865 6,544,865 6,544,865
ydcurr MR$b 4,518,159 4,518,160 4,514,679 4,515,426 4,515,904 4,519,422 4,519,838 4,519,666 4,519,110 4,519,952
ggb MR$b 193,215 193,239 193,232 193,190 193,152 192,900 192,912 192,962 192,980 192,946
debt MR$b 4,621,794 4,621,818 4,621,811 4,621,769 4,621,731 4,621,478 4,621,490 4,621,540 4,621,559 4,621,525
a
Constant 2009 prices.
b
Current 2018 prices.
A.L. de Carvalho et al. / Applied Energy 181 (2016) 514–526 521
the employment level increases or has similar values in solu- All solutions achieved the upper limit established for the resi-
tions 7–9 (achieving the highest values in scenario 2G-A, repre- dent’s consumption (2.75% higher than the values obtained in
senting an increase of approximately 5.19 thousand employees the optimization of GDP), total exports and GFCF.
compared to the 1G scenario); All bioethanol scenarios have negative impacts in terms of
the employment level decreases in solutions 10–14 (achieving employment level. The minimum value is obtained in solution
the lowest values in scenario 2G-F, representing a reduction of 22 that is 5.58% lower than in the solution optimizing the
almost 25.80 thousand employees compared to the 1G scenario). employment level. This fact can be explained by the negative
influence on the outputs of the public services (the sectors pre-
These differences regarding the energy consumption, GHG senting the major output decreases) due to the lowest level of
emissions and employment level can be explained by the higher public consumption in those solutions.
use of chemicals, especially enzymes (see Table 1), in the scenarios All 1G + 2G scenarios provided improvements in terms of total
10–14 in relation to scenarios 7–9. This increases the output of the bioethanol production, achieving the highest level in solution
chemical sector leading to direct and indirect effects on the energy 24 with a surplus of 15.47 ML of bioethanol compared to the
consumption (and consequently on the GHG emissions), as well as 1G scenario (solution 15).
on the overall employment level. Even with a higher bioethanol production the total sugarcane
Another important characteristic is that all 1G + 2G scenarios dis- requirements are reduced in all 1G + 2G scenarios, achieving
play a reduction in the total sugarcane outputs, achieving the lowest the lowest value in solution 22 (MR$ 23,790).
value in the scenario 2G-F (MR$ 21,565), validating the assumption The values of imports and taxes in all scenarios increase
that less sugarcane is demanded in combined 1G + 2G plants in rela- between 1.78–1.84% and 2.34–2.40%, respectively, compared
tion to 1G plants to produce the same bioethanol amount. to the solution optimizing the GDP.
It is possible to verify in Tables 6 and 7 that even fixing the GDP
and the total bioethanol values in those optimizations (solutions 5– An important characteristic of these solutions is that even
10), the economic structure is remarkably dissimilar in each solu- obtaining highest values (in absolute terms) of public debt, the
tion (the same occurs for the amount of hydrous and anhydrous ratio between the public debt and GDP are lower than in the solu-
bioethanol produced). The main characteristics of those solutions tions optimizing the energy consumption and GHG emissions. The
include: an overall increase of exports, imports, taxes and gross average values in solutions 15–24 represent 67.7% of the GDP,
value added in the 1G + 2G scenarios compared to the 1G scenario, while in the solutions optimizing the energy consumption and
while the values of public administration’s global balance (in abso- GHG emissions this ratio achieves 72.5%.
lute terms) and public debt have a marginal overall decrease. The major drawbacks of solutions 15–24 are the negative
impacts on the total energy consumption and GHG emissions,
which achieved in the best case (solution 17) 284,500 ktoe and
3.3. Maximization of the total bioethanol production 2,803,701 Gg of CO2eq, respectively. This can be explained by the
influence of higher production of bioethanol on the overall produc-
The results regarding the maximization of the total (anhydrous tion of all sectors (demonstrated in higher levels of gross value
+ hydrous) bioethanol production in each scenario (solutions 15– added) and exports, as well as direct and indirect effects from
24) are presented in Tables 8 and 9. It is possible to verify that: higher consumption of chemical products in the 1G + 2G scenarios.
Table 8
Values of the objective function in the non-dominated solutions considering the maximization of the bioethanol production in each scenario.
Notation Units Sol. 15 Sol. 16 Sol. 17 Sol. 18 Sol. 19 Sol. 20 Sol. 21 Sol. 22 Sol. 23 Sol. 24
1G 1G Opt. 2G - A 2G - B 2G - C 2G - D 2G - E 2G - F 2G - G 2G - H
ba + bh GL 35,885 35,885 35,894 35,895 35,895 35,898 35,899 35,900 35,897 35,901
Table 9
Values of selected variables in the non-dominated solutions considering the maximization of the bioethanol production in each scenario.
Notation Units Sol. 15 Sol. 16 Sol. 17 Sol. 18 Sol. 19 Sol. 20 Sol. 21 Sol. 22 Sol. 23 Sol. 24
1G 1G Opt. 2G - A 2G - B 2G - C 2G - D 2G - E 2G - F 2G - G 2G - H
gdp MR$a 4,276,212 4,276,211 4,276,609 4,276,671 4,276,693 4,276,899 4,276,986 4,277,124 4,276,972 4,277,155
emp Emp. 103 52,252 52,252 52,256 52,251 52,253 52,248 52,247 52,234 52,238 52,237
ec ktoe 284,614 284,614 284,500 284,572 284,550 284,617 284,620 284,833 284,793 284,782
ghg Gg CO2eq 2,803,884 2,803,884 2,803,701 2,803,946 2,803,985 2,804,657 2,804,900 2,805,540 2,804,993 2,805,578
sc MR$a 28,143 28,143 27,554 26,902 26,955 25,840 25,519 23,790 24,518 23,983
ba GL 9703 9703 9777 9777 9781 9799 9809 9805 9787 9814
bh GL 26,182 26,182 26,117 26,118 26,114 26,098 26,090 26,095 26,110 26,087
rc MR$a 2,741,955 2,741,955 2,741,955 2,741,955 2,741,955 2,741,955 2,741,955 2,741,955 2,741,955 2,741,955
pc MR$a 687,001 687,001 687,001 687,001 687,001 687,001 687,001 687,001 687,001 687,001
ci MR$a 9031 9032 8392 8397 8356 8202 8121 8157 8309 8081
gfcf MR$a 861,608 861,608 861,608 861,608 861,608 861,608 861,608 861,608 861,608 861,608
imp MR$a 444,605 444,606 444,848 444,781 444,800 444,746 444,741 444,568 444,568 444,612
exp MR$a 439,284 439,284 439,284 439,284 439,284 439,284 439,284 439,284 439,284 439,284
ts MR$a 607,924 607,925 608,061 608,089 608,091 608,160 608,182 608,264 608,203 608,262
gva MR$a 3,668,287 3,668,286 3,668,548 3,668,582 3,668,602 3,668,740 3,668,804 3,668,860 3,668,769 3,668,893
gdpcurr MR$b 7,145,234 7,145,224 7,145,641 7,145,657 7,145,676 7,145,793 7,145,844 7,145,885 7,145,768 7,145,918
ydcurr MR$b 4,856,175 4,856,175 4,856,175 4,856,175 4,856,175 4,856,175 4,856,175 4,856,175 4,856,175 4,856,175
ggb MR$b 264,535 264,535 263,934 263,833 263,799 263,472 263,336 263,112 263,348 263,064
debt MR$b 4,834,887 4,834,887 4,834,887 4,834,887 4,834,887 4,834,887 4,834,887 4,834,887 4,834,887 4,834,887
a
Constant 2009 prices.
b
Current 2018 prices.
522 A.L. de Carvalho et al. / Applied Energy 181 (2016) 514–526
interval programming techniques to reach robust recommenda- and PTDC/SEN-TRA/117251/2010 and doctoral grant SFRH/
tions compatible with a range of input parameters. BD/42960/2008. This work has been developed under the Energy
for Sustainability Initiative of the University of Coimbra.
Acknowledgements
Appendix A
This work has been supported by the Portuguese Foundation for
Science and Technology under projects UID/MULTI/00308/2013 See Table A1.
Table A1
Technical coefficients for the 1G and 2G scenarios investigated (R$/toe or toe/toe).
Table A1 (continued)
do Brasil de desenvolvimento – XIV Encontro regional de economia, 16 – 17 [60] Carvalho A, Antunes C, Freire F, Henriques C. A multi-objective interactive
July, Banco do Nordeste do Brasil, Fortaleza, Brazil. Available at:Available approach to assess economic-energy-environment trade-offs in Brazil. Renew
from: <http://www.bnb.gov.br/content/aplicacao/ Sust Energy Rev 2016;54:1429–42. http://dx.doi.org/10.1016/j.
eventos/forumbnb2009/docs/construcao.pdf> [accessed 1 March 2014]. rser.2015.10.064.
[48] Santiago F. Um modelo econométrico + insumo-produto para a previsão de [61] Leontief W. Input-output analysis. In: Leontief W, editor. Input-output
longo prazo da demanda de combustíveis no Brasil MSc economics. New York (USA): Oxford University Press; 1985. p. 19–40. 1986.
Thesis. Brazil: Universidade Federal de Juiz de Fora; 2009. [62] Ministério das Minas e Energia (MME). Balanço Energético Nacional,
[49] Miller R, Blair P. Input-output analysis: foundations and extensions. 2nd Ministério das Minas e Energia, Brazil Available at:Available from: <https://
ed. New York (USA): Cambridge University Press; 2009. ben.epe.gov.br/downloads/relatorio_final_ben_2010.pdf>2010 [accessed 10
[50] Moulik T, Dholakia B, Dholakia R, Ramani K. Energy planning in India: the June 2013].
relevance of regional planning for natural policy. Energy Policy 1992;20 [63] Guilhoto J, Sesso Filho U. Estimação da Matriz Insumo-Produto Utilizando
(9):836–46. http://dx.doi.org/10.1016/0301-4215(92)90120-Q. Dados Preliminares das Contas Nacionais: Aplicação e Análise de Indicadores
[51] Hristu-Varsakelis D, Karagianni S, Pempetzoglou M, Sfetsos A. Optimizing Econômicos para o Brasil em 2005. Econ Technol 2010;23:53–62. http://dx.
production with energy and GHG emission constraints in Greece: an input– doi.org/10.2139/ssrn.1836495.
output analysis. Energy Policy 2010;38(3):1566–77. http://dx.doi.org/ [64] Intergovernmental Panel on Climate Change (IPCC). IPCC guidelines for
10.1016/j.enpol.2009.11.040. national greenhouse gas inventories, intergovernmental panel on climate
[52] Tan R, Aviso K, Barilea I, Culaba A, Cruz Jr J. A fuzzy multi-regional input– change/institute for global environmental strategies (IGES). Japan: Hayama;
output optimization model for biomass production and trade under resource 2006.
and footprint constraints. Appl Energy 2012;90:154–60. http://dx.doi.org/ [65] Intergovernmental Panel on Climate Change (IPCC). Climate change 2007 – the
10.1016/j.apenergy.2011.01.032. physical science basis. New York (USA): Intergovernmental Panel on Climate
[53] Cho C. The economic-energy-environmental policy problem: an application of Change, Cambridge University Press; 2007.
the interactive multiobjective decision method for Chungbuk Province. J [66] Instituto Brasileiro de Geografia e Estatística (IBGE). Matriz Insumo-Produto
Environ Manage 1999;56(2):119–31. http://dx.doi.org/ Brasil 2000/2005, Instituto Brasileiro de Geografia e Estatística, Brazil; 2008.
10.1006/jema.1999.0264. [67] Ministério da Ciência e Tecnologia (MCT). Segunda Comunicação Nacional do
[54] Hsu G, Chou F. Integrated planning for mitigating CO2 emissions in Taiwan: a Brasil à Convenção-Quadro das Nações Unidas sobre Mudança do Clima,
multi-objective programming approach. Energy Policy 2000;28(8):519–23. Ministério da Ciência e Tecnologia, Brazil Available at:Available from: <http://
http://dx.doi.org/10.1016/S0301-4215(00)00006-9. www.mct.gov.br/upd_blob/0213/213909.pdf>2010 [accessed 15 October
[55] Chen T. The impact of mitigating CO emissions on Taiwan’s economy. Energy 2013].
Econ 2001;23:141–51. http://dx.doi.org/10.1016/S0140-9883(00)00060-8. [68] International Monetary Fund (IMF). World economic outlook database,
[56] Kravtsov M, Pashkevich A. A multicriteria approach to optimization of the international monetary fund Available at:Available from: <http://www.imf.
gross domestic product. Autom Remote Control 2004;65(2):337–45. http://dx. org/external/pubs/ft/weo/2013/02/weodata/index.aspx>2013 [accessed 10
doi.org/10.1023/B:AURC.0000014730.28129.e5. January 2014].
[57] Oliveira C, Antunes C. A multiple objective model to deal with economy- [69] Steuer R, Choo E. An interactive weighted Tchebycheff procedure for multiple
energy-environment interactions. Eur J Oper Res 2004;153:370–85. http://dx. objective programming. Math Program 26, 326–344. http://dx.doi.org/10.
doi.org/10.1016/S0377-2217(03)00159-0. 1007/BF02591870.
[58] Oliveira C, Antunes C. A multi-objective multi-sectoral economy-energy- [70] Bickneel K, Ball R, Cullen R, Bigsby H. New methodology for the ecological
environment model: application to Portugal. Energy 2011;36(5):2856–66. footprint with an application to the New Zealand economy. Ecol Econ 1998;27
http://dx.doi.org/10.1016/j.energy.2011.02.028. (2):149–60.
[59] Carvalho A, Antunes C, Freire F, Henriques C. A hybrid input-output multi- [71] Cruz L. A Portuguese energy-economy-environment input-output model:
objective model to assess economic-energy-environment trade-offs in Brazil. policy applications Ph.D. Thesis. Keele (UK): Keele University, School of
Energy 2015;82:769–85. http://dx.doi.org/10.1016/j.energy.2015.01.089. Politics International Relations and the Environment; 2002.