Fonseca 2021

Applied Energy 290 (2021) 116746
Contents lists available at ScienceDirect
Applied Energy
journal homepage: www.elsevier.com/locate/apenergy
Sustainability analysis for the design of distributed energy systems: A

multi-objective optimization approach
Juan D. Fonseca a, c, *, Jean-Marc Commenge b, Mauricio Camargo a, Laurent Falk b, Iván D. Gil c
a
Équipe de Recherche sur les Processus Innovatifs (ERPI), Université de Lorraine, 8 rue Bastien Lepage, 54000 Nancy Cedex, France
b
Laboratoire Réactions et Génie des Procédés (LRGP), Université de Lorraine, 1 rue Grandville, BP 20451, 54001 Nancy Cedex, France
c
Grupo de Procesos Químicos y Bioquímicos, Department of Chemical and Environmental Engineering, Universidad Nacional de Colombia – Sede Bogotá, Carrera 30 45-
03, Bogotá, Colombia
H I G H L I G H T S G R A P H I C A L A B S T R A C T
• Inherent safety index of the energy sys

tem was evaluated.
• Sustainability assessment for the design
of energy systems.
• Four multi-objective optimization prob
lems were addressed.
• Relationships among the sustainability
indicators were established.
• Pareto solutions were explored in terms
of design and operation variables.
A R T I C L E I N F O A B S T R A C T
Keywords: The design of sustainable energy systems requires to enlarge the analysis beyond the traditional boundaries for
Renewable energy including the economic, environmental, and societal needs and constraints in the decision-making process. In this
Hybrid energy system regard, this work investigates the conceptual design of distributed energy systems by means of a multi-objective
Inherent safety index
optimization strategy to simultaneously address the economic, environmental, and social aspects in the energy
Hydrogen
Sustainability assessment
system design. Initially, the water consumption and the inherent safety indicators were introduced and evaluated
Decision-making through two single-objective optimization problems to enhance the analysis of the environmental and social
dimensions of sustainability. Then, a framework including the total annualized cost, CO2 emissions, water
consumption, grid dependence, and inherent safety index was used to perform the multi-objective analysis. To
carry out a thorough and comprehensive analysis, four optimization problems including different combinations
of the sustainability indicators were proposed and solved. The compromise among the objective functions was
identified, and the obtained Pareto sets were explored for elucidating the changes in the design and operating
conditions across the non-dominated solutions. According to results, the cost of energy can range between 0.37
and 0.63 €/kWh, the CO2 emissions can vary between 10.6 and 68.5 kgCO2/MWh, and the water consumption
can be between 27.8 and 70.2 m3H2O/GWh depending on the evaluated objective. Moreover, it was determined
* Corresponding author at: Équipe de Recherche sur les Processus Innovatifs (ERPI), Université de Lorraine, 8 rue Bastien Lepage, 54000 Nancy Cedex, France.
E-mail address: jdfonsecag@unal.edu.co (J.D. Fonseca).
https://doi.org/10.1016/j.apenergy.2021.116746
Received 28 October 2020; Received in revised form 11 February 2021; Accepted 23 February 2021
Available online 5 March 2021
0306-2619/© 2021 Elsevier Ltd. All rights reserved.
J.D. Fonseca et al. Applied Energy 290 (2021) 116746
that the safety of the energy system can be improved by increasing the use of the water electrolysis pathway to
produce hydrogen and by reducing the capacity of the hydrogen storage unit.
Nomenclature u Decision variables

t Time
Hm
2,El Mass flow rate of hydrogen from electrolyzer (kg/s) SH2 Energy stored in hydrogen form (kWh)
LHV H2 Low heating value of hydrogen (kJ/kg) BioD Biomass available for digestion (kg/year)
H2 O R Mass flow rate of water to the reformer (kg/s) SB Energy stored in the battery (kWh)
ADin Input biomass to the anaerobic digester (kg/s)
Greek letters
H2 OFC Mass flow rate of water from the fuel cell (kg/s)
φ Source of electricity for demand
WC Water consumption of the energy system (kgH2O/year)
θ Source of hydrogen
Rout Output power of the reformer (kW)
α Source of electricity for electrolysis
FCin Input power to the fuel cell (kW)
δ Electricity storage option
MWH2 Molecular weight of hydrogen (kg/kmol)
γ Source of methane for reforming
MWH2O Molecular weight of water (kg/kmol)
ψ El Water consumption in electrolysis (kgH2O/kgH2)
IC Chemical safety index
ψ AD Water consumption in anaerobic digestion (kgH2O/kg
IP Process safety index
biomass)
IFL Flammability score
IEX Explosiveness score Abbreviations
ITOX Toxicity score DES Distributed energy systems
ICOR Corrosiveness score HAZOP Hazard and operability
IR Reaction score P&ID Piping and instrumentation diagram
II Inventory score PIIS Prototype index of inherent safety
IT Temperature score ISI Inherent safety index
IPr Pressure score EISI Enhanced inherent safety index
k Technology index CISI Comprehensive inherent safety index
J Performance criterion
Taking this into account, it is noted that sustainability requires a holistic

approach, and to enlarge the analysis out of the traditional system
1. Introduction boundaries by including the economic, environmental and societal
needs and constraints in the decision-making process [10,11]. There
The current situation due to the coronavirus pandemic (Covid-19) fore, the sustainable design of energy systems requires to simultaneously
has represented great impacts on global health, economy, energy use integrate multiple and often contradictory criteria, i.e. to address a
and CO2 emissions. Indeed, because of these unprecedent circumstances, multi-objective optimization problem.
the economic curtailment has led to a decline in energy demand of 3.8% Considering the aforementioned aspects, sustainable energy systems
with respect to the first quarter of 2019. In this respect, according to the should have four main characteristics: cost-efficiency, reliability, safety,
perspectives of the International Energy Agency, the annual energy and environmental-friendliness [12,13]. In principle, distributed energy
demand will decrease between 4 and 6%. In the same line, it is expected systems (DES) seem to be a suitable alternative to the before mentioned
the CO2 emissions push down by 8% relative to the 2019 [1]. challenges, since they fit well with the features of sustainable energy
Nevertheless, before the exceptional circumstances linked to this systems. DES rely on the energy consumption close to the generation
sanitary crisis, global energy demand grew by 2.3%, and CO2 emissions site, so that energy efficiency is increased as the energy losses across
had increased by 1.7% in 2018 [2]. This was mainly driven by the long-distance transmission lines are avoided [12]. Distributed genera
constant expansion of world economy and population, and the growing tion also promotes the use of local and renewable resources, thus it
demand for heating and cooling in some parts of the world. Indeed, in enhances self-sufficiency and energy security, and it offers an alternative
2018 around 20% of the increase in energy consumption was the to provide energy with low-carbon emissions [12,14].
consequence of some extreme weather conditions (cold and hot snaps)
[2]. This situation represents a significant concern for the energy sector,
since the raising emissions trends of the last few years are not aligned 1.1. Literature review
with the Paris agreement for keeping the global temperature rise below
2 ◦ C [3]. Consequently, as a response to this scenario, a global energy A diverse range of research has been published addressing the design
transformation is underway to satisfy the continuous increase in the of DES by using multi-objective optimization approaches. Broadly, these
energy consumption and to meet the world environmental goals. This works differ in the method used to solve the optimization problem and in
necessary change lies upon three main pillars: (i) to increase the share of the objective functions evaluated. For instance, Gabrielli et al. per
renewable resources, (ii) to rise the usage of low-carbon electricity as formed the analysis including the total annualized cost and the annual
end-use energy form, and (iii) to deploy the distributed generation of CO2 emissions by means of the epsilon-constraint method. In fact, they
energy [3–5]. carried out the energy system design under both, deterministic [15] and
Furthermore, there is also a need to develop a global industry in a uncertain [16] scenarios for considering the unpredictable behavior of
sustainable way [6,7], so that activities of the present do not compro the weather conditions and energy demands. Jing et al. used the same
mise the ability of future generations to meet their own needs [8,9]. method and objective functions, but they include benefit allocation
Broadly, sustainable development promotes the balance of three aspects: constraints inspired from cooperative Game Theory [17]. Falke et al.
economic success, social acceptance, and environmental protection [8]. proposed to decompose the optimization problem into three stages to
2
reduce the mathematical complexity, and they included the life cycle Table 1
assessment to evaluate the emissions of CO2 equivalents [18]. Ren et al. Objective function, criteria and optimization approach commonly used for the
also evaluated the annual cost and CO2 emissions as objective functions, design of distributed energy systems. (Ec) economic, (Env) environmental, (Tec)
but they focused upon the operation of the energy system and employed technical criteria.
the compromise method to solve the multi-objective optimization Authors Objective Criteria Optimization Reference
problem [19]. Meanwhile, other works have employed the weighted Function Approach
sum method for transforming the problem into a single-optimization Gabrielli Total cost – Ec – Env Epsilon-constraint [15]
case. Di Somma et al. used this approach for the design of energy sys et al. CO2 emissions
tems by evaluating the total cost and the exergy efficiency as criteria Jing et al. Total cost – Ec – Env Epsilon-constraint [17]
CO2 emissions
[20]. Dorotić et al. also used that method to address optimization Falke et al. Total cost - CO2 Ec – Env NSGA-II [18]
problems including cost, CO2 emissions, and exergy destruction as the equivalents
objective functions [21,22]. In another study, Dufo-López and Bernal- Ren et al. Total cost – Ec – Env Compromise method [19]
Agustín employed a multi-objective evolutionary and genetic algorithms CO2 emissions
Di Somma Total cost – Ec – Tec Weighted sum method [20]
for the design of energy system considering the net present cost, the CO2
et al. exergy
emissions and the unmet load as obective functions [23]. efficiency
Alvarado et al. proposed an optimization model for the selection and Dorotić Cost – CO2 Ec – Env Weighted sum method [21,22]
operation of technologies by incorporating real-time energy pricing and et al. emissions - - Tec
demands, fuel flexibility, and projections for various parameters [24]. exergy
Dufo-López Cost – CO2 Ec – Env Evolutionary [23]
Moreover, they considered the total cost and CO2 emissions as objective
et al. emissions – - Tec algorithm (multi-
functions and used the weighted sum method to solve the multi- unmet load objective)
objective optimization. Mayer et al. provided a design framework for Alvarado Cost – CO2 Ec – Env Weighted sum method [24]
hybrid energy systems [25]. Besides, they used a genetic algorithm to et al. emissions
Mayer et al. Cost – CO2 Ec – Env NSGA-II [25]
solve single and multi-objective optimization problems considering the
emissions
net present cost and the environmental footprint (life cycle assessment) (LCA)
as objectives. Hou et al. adressed the multi-objective optimization of Hou et al. Cost – CO2 Ec – Env Epsilon-constraint [26]
DES by considering three different operational strategies namely (i) heat emissions
load following, (ii) electrical load following, and (iii) load characteristic Karmellos Cost – CO2 Ec – Env Epsilon-constraint [27]
et al. emissions
matching following [26]. They employed the epsilon-constraint method
Yan et al. Life cycle cost - Ec – Env Multidisciplinary [28]
to analyze the minimization of the total cost and CO2 emissions. Like LCA design optimization
wise, Karmellos and Mavrotas used the same approach and objective (MDO)
functions, but they focused on comparing two methodologies for the
design of DES. The former to simultaneously size the available tech
since they are not sustainable [33–37]. Accordingly, water consumption
nologies, and the latter to select technologies with a predefined size
[27]. Recently, Yan et al. used an approach based on multidisciplinary appears as a suitable indicator for evaluating the performance of the
energy system and for enhancing the analysis of the environmental
design optimization (MDO) to determine the best combination of tech
nologies for different scenarios (building types and climate zones) in dimension of sustainability.
On the other hand, even though the energy system design is pri
terms of economic and environmental impact [28].
Table 1 presents a summary of the literature review drawing the marily a technical challenge, social aspects can become the most
important factors for its successful implementation in the community
objectives, criteria and the approaches used by researchers for the
design of DES. As depicted, the analysis is focused on economic and [29]. In general, the social dimension is associated to the impacts on the
environmental criteria by evaluating objective functions such as the quality of life, which lies on two main features: equity and health. Social
total cost and the CO2 emissions. Moreover, in most cases the optimi equity involves the level of fairness and inclusiveness for energy re
zation approach is based on transforming the original problem into a sources distribution [38]. Indeed, this issue is included in one of the
single-objective one by using the weighting and epsilon-constraint sustainable development goals proposed by United Nations (Goal 7),
methods. These methods are simple and effective for solving problems which consists in ensuring access to affordable, reliable, sustainable and
with few objectives. Nevertheless, to build an evenly distributed set of modern energy for all [39]. Otherwise, the health aspect refers to the
non-dominated solutions, the optimization problem must be solved potential pollution, accidents, injuries, or fatalities derived from energy
many times since a single point is obtained at a time. generation. In this respect, the pollution issue commonly falls into
environmental dimension, as it is quantified with variables such as gas
emissions, particulate matter, or contamination of water. Meanwhile,
1.2. Novelty of this paper
occupational accidents and public hazards are related with the inherent
risk derived from the operation of conversion technologies (e.g. tem
As noted from the previous literature review, the design of energy
perature and pressure conditions), and the properties of chemical
systems is predominantly performed based upon economic and envi
compounds (e.g. flammability, toxicity) employed within energy sys
ronmental criteria. Indeed, the multi-objective analysis has been focused
tems [35–38,40,41]. In this line, inherent safety analysis is recognized as
on two main issues: cost and CO2 emissions. In such a way, the obtained
a proper indicator for evaluating the potential impacts on the health
results leave aside the impact of additional indicators, and especially the
during the energy system design, since the selection of conversion
influence of the social aspects on the design of energy systems. Thus, to
technologies and process conditions could be rated according to their
address this issue, it is necessary to develop methodological tools that
intrinsic properties [42,43].
enable to analyze these socio-technical systems within the framework of
Considering the foregoing, this work aims to perform a sustainability
sustainable development, i.e. considering their economic, environ
analysis for the design of energy systems. In such a way, the contribution
mental, and social aspects [29–31].
of this study is twofold: (i) two new indicators, namely water con
On the one hand, billions of people worldwide still currently lack of
sumption and inherent safety index, are introduced as objective func
water for drinking and hygiene purposes [32]. Besides, water shortages
tions to perform the design of energy systems; and (ii) the trade-offs
due to climate change effects, make the use of water another factor of
among a set of sustainability indicators are investigated through a multi-
supreme importance for assessing industrial activities [33]. In fact, high
objective optimization approach to obtain some general insights for the
water consumptions in energy conversion processes should be avoided
3
conceptual design of DES. Accordingly, the remainder of this document 2. Methodology

is structured as follows. First, two single-objective optimization prob
lems focused on the water consumption and inherent safety indicators 2.1. Energy system description
are addressed for studying the impact of the objective function on the
design and operating conditions of the energy system. Then, these two The energy system being considered is depicted in Fig. 1, which
indicators are added to the framework developed by Fonseca et al. [44], corresponds to that one previously studied by Fonseca et al. [44]. As
wherein the total annualized cost, the CO2 emissions and the grid observed, the energy sources include solar and biomass resources, in
dependence were used as the indicators for the energy system design. addition with the possibility of importing electricity and natural gas
Considering that the results of a multi-objective optimization problem from the main grid. These energy inputs are converted and/or managed
with five objectives would be hard to analyze, the sustainability analysis within the system to satisfy electricity and hydrogen demand. For
is decomposed into four multi-objective optimization problems. These tackling the mismatch between the energy availability from renewables
optimization cases include different combinations of the five sustain and the demand, the surplus electricity is stored either in an electrical
ability indicators by simultaneously addressing three objectives at most. battery or in hydrogen form by using the power-to-power pathway
This strategy enables to obtain a thorough understanding about the re (electrolyzer - pressurized tank - fuel cell). Meanwhile, hydrogen is
lationships among the objective functions since the results can be obtained by means of water electrolysis or steam methane reforming
analyzed based on a three-dimension plot. Thereafter, the set of Pareto processes. In the latter, the reforming reactor can be fed either by nat
solutions is explored and analyzed for identifying the changes in the ural gas from the network or by biomethane from the biomass digestion
design and operating conditions across the optimization results. Finally, process.
the main findings are summarized and conclusions are drawn.
2.2. System modeling and case study
The mass and energy balances, as well as the performance of the
Fig. 1. Schematic representation of the analyzed energy system. ( ) Electricity, ( ) Hydrogen, ( ) Water, ( ) CO2.
4
energy conversion and storage units are described according to the en In these expressions, Hm 2,El represents the amount of hydrogen pro
ergy system model presented in reference [44]. Broadly, the system duced by water electrolysis, H2 OR the water required by reforming
model has four main features: process, ψ AD the water consumption of the anaerobic digestion process,
and ADin the input biomass to the digester. Moreover, H2 OFC represents
1. It is a pseudo-steady state model, so that the time-dependence is the water obtained from the operation of fuel cell, Rout is the output
taken into consideration but process units with instantaneous re power from the reformer, FCin the input power to the fuel cell, and
sponses are assumed. Hence, accumulation within the energy con LHV H2 is the low heating value of hydrogen. Meanwhile, MWH2O and
verters is neglected. Moreover, the evolution of the energy stored is MWH2 are the molecular weight of water and hydrogen, respectively.
addressed by discretization of the temporal variable. In literature it is reported that water corresponds to 15–40% of the
2. The operational conditions of energy converters are fixed; thus the total mass of the digester [46]. Accordingly, in this work is considered a
input/output variables are related by means of constant efficiencies water consumption factor of 20%, i.e. 0.25 kgH2O/kg biomass.
and linear expressions.
3. There are no energy losses across the connection lines. • Inherent Safety:
4. The model is deterministic, and all the sizes of equipment are
considered as continuous variables. Power plants involve hazardous materials and processes that can
lead to accidents, and negatively impact the health of workers and the
Moreover, the case study is a hypothetical neighborhood of 1500 wellbeing of communities. Therefore, aiming to avoid or reduce those
inhabitants located at Marseille-France. The yearly electricity and accidents, a variety of methods can be implemented for evaluating the
hydrogen demands are 4080.6 and 731.5 MWh, respectively. Further safety issue during the whole life cycle of the project [47]. Broadly, these
details about the profiles of weather conditions (solar irradiance and methods differ in the aspects considered, the required information, and
ambient temperature), and energy demands can be found in [44]. Also, the output data type [43]. For example, some of the most common and
the corresponding set of performance and cost parameters is presented popular approaches are the Hazard and Operability method (HAZOP),
in that reference. the Dow chemical/fire and exposure indices (C&EI and F&EI), and the
Mond index [43,48]. However, these methods are not suitable for con
2.3. Single-objective optimization ceptual design stage, since they rely upon information from basic and
detailed engineering such as the P&ID (piping and instrumentation di
As mentioned in the introduction of this paper (Section 1), one of the agram) of the process [43,47,48].
contributions of this research lies on enhancing the sustainability eval Moreover, some methodologies are based upon the assessment of the
uation of the energy system by including two new performance criteria. inherent safety of the processes. The essence of this perspective is to
Therefore, the first step consisted in performing these single-objective avoid and/or eliminate hazards rather than controlling them through
optimizations. In the next sections, the objective functions and the add-on systems [49–52]. Indeed, these approaches rely on the fact that
formulation of the optimization problem are described. the potential risks of a process are related with the intrinsic character
istics of the chemical substances and operation units [52,53]. Inherent
2.3.1. Objective functions safety comprises four main principles: intensification, substitution,
The optimization objectives proposed in this work are the annual attenuation, and simplification. Intensification or minimization refers to
water consumption and the inherent safety index. Accordingly, their quantities of materials and size of equipment within the plant. Thus,
corresponding mathematical formulation is presented as follows. safer processes are those ones with lower number of hazardous sub
stances and smaller operation units. Substitution relies on replacing
• Water Consumption: hazardous materials for safer ones, e.g. using non-flammable and/or
non-toxic refrigerants or solvents instead of flammable and/or toxic
According to the energy system structure (Fig. 1), water is consumed compounds. Attenuation or moderation seeks to modify process condi
in three operations: electrolysis (Equation (1)), steam methane reform tions (e.g. temperature, pressure, concentration) for avoiding flammable
ing (Equation (2)), and anaerobic digestion. In the latter, water is mixed limits or reactions close to runaway temperatures. Meanwhile, simpli
with the biomass for obtaining the slurry that is fed to the digester. fication lies on the fact that processes with less equipment lead to fewer
Nevertheless, water is also obtained from the operation of fuel cell. In opportunities for error, so that simpler plants are inherently safer
fact, this water is commonly recycled and reused in the electrolyzer. [42,51,52].
Considering the foregoing, the inherent safety assessment is typically
2H2 O→2H2 + O2 (1)
quantified as the contribution of two sub-indices: the chemical and the
CH4 + 2H2 O ↔ 4H2 + CO2 (2) process inherent safety index. The former includes properties such as
heat of reaction, flammability, explosiveness, and toxicity. In contrast,
Thus, the net water consumption (WC) of the energy system can be the latter focuses on process conditions such as pressure and tempera
described by means of Equation (3). It is worthy to note that from the ture. In this respect, a variety of methods have been proposed to evaluate
stoichiometry of water electrolysis reaction (Equation (1)), 9 kg of water the inherent safety of processes. For example, Edwards and Lawrence
are needed to produce 1 kg of hydrogen. However, in practice around 11 proposed the prototype index of inherent safety (PIIS) which is mainly
kgH2O/kgH2 are required [45], which is represented by the factor ψ El . focused on the selection of raw materials and reaction steps [54].
Moreover, Equations (4) and (5) represent the water required in the Nonetheless, this is a reaction-oriented method and is not suitable for
reforming reactor and the water produced from fuel cell, respectively. the safety evaluation of the whole process plant. In another work,
∫ tf ∫ tf ∫ tf ∫ tf Heikkilä proposed the inherent safety index (ISI), which allows com
WC = ψ El m
H2,El dt + H2 OR dt + ψ AD ADin dt − H2 OFC dt (3) parison between different process alternatives [51]. Despite that, this
method has some limitations because it is based on the worst-case sce
t0 t0 t0 t0
(
Rout (t)
)(
2MW H2O
) nario and does not consider neither the quantity of materials nor the
H2 OR (t) = (4) amount of equipment within the process. Indeed, the evaluation of the
LHV H2 4MW H2
worst-case scenario could lead to similar results even if considerably
( )(
FCin (t) MW H2O
) different processes are analyzed [53,55].
H2 OFC (t) = (5) To overcome this issue, Li et al. proposed the enhanced inherent
LHV H2 MW H2
5
safety index (EISI), which includes all the chemicals and their amount by ( )
multiplying the flow rate with the severity factor (e.g. explosiveness or SB (t0 ) = SB tf , periodicity (14)
toxicity) [53]. Additionally, for the process inherent safety index, the
0 ≤ SH2 (t) ≤ SH2,max , hydrogen storage (15)
scores of the equipment are multiplied by their number and are added all
together. In a similar direction, recently Gangadharan et al. introduced ( )
SH2 (t0 ) = SH2 tf , periodicity (16)
the comprehensive inherent safety index (CISI) by adopting and object-
oriented approach [55]. In this method, each equipment corresponds to ∫ tf
a separate entity so that the inherent safety index is calculated for in ADin dt ≤ BioD , biomass available (17)
dividual operation units. Moreover, this approach considers the severity t0
of reactions based on the combination of chemicals, and a connection

0 ≤ u(t) ≤ 1; u ∋ {φ1 , φ2 , θ1 , α1 , δ1 , γ1 }, optimization variables (18)
score between two units as a function of their individual safety scores.
In this work, the inherent safety for the energy system design is ∑
I
quantified based upon the comprehensive inherent safety index. How Si (t) = 1S ∋ {φ, θ, α, δ, γ}, consistency (19)
ever, the severity of reaction and the connection score are not included. i
The former because the score of reaction is considered as a function of

the heat of reaction (already included), and the latter because that score where, J represents each objective function and h the energy system
could change depending on the judgement and/or experience of the model, which is a function of the decision variables (u), the state vari
method user, as stated by Gangadharan et al. [55]. Accordingly, the total ables (x), the parameters (p), and the time (t). Moreover, the optimiza
inherent safety index for evaluating the energy system can be calculated tion problem also includes path (Equations (13) and (15)) and endpoint
by means of Equation (6). Moreover, Equation (7) describes the indi (Equations (14) and (16)) constraints on the level of energy stored in the
vidual equipment safety index, which includes the contribution of the battery (SB ) and in the pressurized tank (SH2 ). The former to avoid
chemical and process indices. The former is described by Equation (8) negative values in the storage level, and the latter to ensure periodic
and the latter through Equation (9). Note that a lower index corresponds behavior over the evaluated time horizon. Besides, Equation (17) rep
to a safer system structure. resents a constraint for limiting the biomass used in the digestion process
according to the amount available (BioD ). Meanwhile, Equation (18)
∑
K
denotes the set of optimization variables, which correspond to the
IST = IE,k (6)
fractions φ1 ,φ2 , θ1 , α1 , δ1 and γ1 . Further details about the optimization
k
approach and the energy system model can be found in reference [44].
IE,k = IC,k + IP,k (7) The optimization problems were solved considering one year as time
horizon and using time steps of 12 h, i.e. Δt = 12 h, t0 = 0 h and tf =
∑N ∑N 8760 h. Likewise, the upper limits for the storage technologies are 30
(IFL,n + IEX,n + ITOX,n + ICOR,n )Fn + IR,k Fn
IC,k = n n
(8) and 300 MWh for the battery and the pressurized tank, respectively.
1000
Besides, the amount of biomass available (BioD ) is 255.6 ton/year. This
IP,k = II,k + IT,k + IPr,k (9) value was estimated by considering the domestic waste generated per
inhabitant (568 kg/year), the fraction of organic matter that can be used
In those expressions, IST is the total inherent safety index, IE,k is the
in the digestion process (30%), and the size of the studied neighborhood
individual equipment safety index, IC,k is the chemical index, and IP,k is
(1500 inhabitants) [56,57]. Meanwhile, the photovoltaic surfaces
the process index of the equipment k. Additionally, IFL,n , IEX,n , ITOX,n and
evaluated in this work are 7500 and 10000 m2. These values represent
ICOR,n are the flammability, explosiveness, toxicity and corrosiveness
scenarios with an overcapacity of 50 and 100% with respect to the size
scores, respectively. Fn represents the mass flow rate, IR,k is the reaction
required to cover the peak electricity demand [44].
sore, the subscript n denotes each chemical substance through the
equipment k and the factor 1000 is the basis flow rate. Meanwhile, II,k ,
IT,k and IPr,k correspond to the inventory, temperature, and pressure 2.4. Multi-objective optimization
scores, respectively. Detailed information for computing the inherent
safety index is presented in section S1 of the supplementary material. In As stated in the introduction section, the multi-objective analysis
this formulation is noted that the traditional scores for the inventory proposed in this work considers the water consumption and the inherent
index (II,k ) depicted in Table S7 are conceived to assess large-scale units. safety index in addition to the total annualized cost, CO2 emissions and
Nevertheless, as distributed energy systems are small-scale plants, in this grid dependence indicators previously used by Fonseca et al. [44].
work a new scale was proposed for the evaluation of that index. All the Accordingly, five objective functions are included within the multi-
details of this analysis are presented in section S2 of the supplementary objective optimization framework for the energy system design under
material. the sustainability dimensions. The principles, criteria and indicators of
the proposed framework are presented in Fig. 2.
2.3.2. Optimization problem The results of an optimization problem including five objective
Considering the two objective functions previously described, two functions would be hard to both visualize and analyze. Moreover, as
independent optimization problems are stated to minimize the water presented in the introduction of this document (Section 1.2), one of the
consumption and the inherent safety index. The mathematical formu goals of this work is to obtain a thorough understanding about the re
lation of these problems was built following the approach developed by lationships among the objective functions and to identify some general
Fonseca et al. [44], which is presented in Equations (10)–(19). insights for the conceptual design of DES. Accordingly, with the aim of
performing this analysis, four multi-objective optimization problems are
Minimize J1 (u, x, t) = WC (10) proposed and solved. These problems include different combinations of
the indicators to assess the sustainability of the energy system. The
J2 (u, x, t) = IST (11)
proposed cases cover the three dimensions of sustainability and simul
taneously consider three objectives at most, which leads to results that
Subject to h = f (u(t), x(t), p, t ), system model (12)
can be clearly drawn in a three-dimension plot. Moreover, this strategy
0 ≤ SB (t) ≤ SB,max , battery storage (13) also enables to carry out a comprehensive analysis about the trade-offs
of the objectives to support the subsequent decision-making process.
Table 2 depicts how the indicators are grouped for addressing the
6
Fig. 2. Framework of principles, criteria and indicators for the sustainability assessment of energy systems.
was used because it enables to simultaneously obtain the entire set of

Table 2
non-dominated solutions (i.e. only one optimization run is required),
Multi-objective optimization problems and corresponding indicators for the
which provides more information to the decision-maker [58,60]. Be
energy system design.
sides, this evolutionary algorithm does not depend on continuity, de
Problem Indicators rivative conditions and initial points, therefore it is suitable for solving
P1 Cost - CO2 emissions / Grid dependence problems that could be difficult to address with deterministic methods
P2 Cost - Water consumption - Grid dependence [60]. Moreover, the ready availability and effectiveness of the algorithm
P3 Cost – Water consumption - Safety
are well documented, as it has been extensively used to solve multi-
P4 Safety – Cost / CO2 emissions
objective optimization problems [58,60–62]. In this work, the NSGA-II
algorithm was employed through the gamultiobj function within MAT
multi-objective optimization. As noted, each one of the problems in LAB® software.
volves at least two dimensions of sustainability. Thus, problem P1 in
cludes two sub-problems, i.e. the cost against the CO2 emissions and the 3. Results and analysis
cost against the grid dependence. This was selected as the first case
because it covers the indicators previously analyzed in the literature. 3.1. Single-objective optimization
Meanwhile, problems P2 and P3 simultaneously include the three di
mensions of sustainability and the two indicators introduced in this Water consumption and inherent safety index were evaluated to
work (water consumption and inherent safety). Finally, problem P4 also enhance the analysis of the environmental and social dimensions of
comprises two sub-problems to complete the analysis among all the sustainability. Then, Fig. 3 shows the optimal flowsheet of the system by
indicators, i.e. the safety against the cost, and the safety against the CO2 optimizing those objectives.
emissions are studied. The remainder of the formulation is identical to Regarding the minimization of water consumption, the obtained
that employed in the single-objective optimization problems, i.e. system configuration indicates that the electrolyzer is only powered by
Equations (12)–(19). PV electricity, and that all the hydrogen produced by this process is sent
A common strategy for solving multi-objective optimization prob to the pressurized tank, as observed in Fig. 3a. Therefore, the whole
lems is to transform the original problem into a single-objective one. demand of hydrogen is supplied by the steam methane reforming pro
That is the case of the weighting, epsilon-constraint, and global criterion cess. Also note that the anaerobic digester is not included within the
methods. In general, such methods are simple and effective for solving system, hence, only methane from the main network is supplied to the
problems with few objectives. However, they require to solve the opti reformer reactor.
mization problem many times, since only one point of the Pareto set is On the one hand, the fact of sending all electrolytic hydrogen to the
obtained at a time. Moreover, as the number of objectives increases, it is storage tank can be explained by analyzing the Equation (3). In that
more difficult to select suitable values for the weighting and epsilon- expression, it is noted that the employment of the system electrolyzer-
constraint methods [58,59]. tank-fuel cell offers the possibility of recovering the water produced in
Taking this into account and the number of sustainability indicators the fuel cell to be reused in the electrolyzer. In contrast, if such hydrogen
included in the proposed framework (5 objective functions), the Non- is sent to supply the demand, there is no way of recuperating the water,
dominated Sorting Genetic Algorithm (NSGA-II) was selected for solv and therefore its net consumption will increase. On the other hand, the
ing the proposed multi-objective optimization problems. This algorithm selection of reforming instead of electrolysis process lies on the
7
Fig. 3. Optimal configurations of the distributed energy system. Objective functions (a) water consumption, (b) inherent safety index.
stoichiometry of reactions. Thus, by considering the stochiometric

Table 3
relation of each process, water electrolysis requires 9 kgH2O/kgH2,
Optimization results of the distributed energy system for minimizing the water
whereas the reforming reaction needs 4.5 kgH2O/kgH2. Consequently,
consumption (WC) and the inherent safety index (IST).
as the objective is to minimize the water consumption, steam reforming
Variable PV area = 7500 m2 PV area = 10000 m2
process provides a better performance than water electrolysis.
Otherwise, concerning the optimization of the inherent safety index, WC IST WC IST
the obtained results indicate that the whole demand of hydrogen must LCOE (€/kWh) 0.57 0.54 0.63 0.58
be supplied by the water electrolysis process, as depicted in Fig. 3b. TAC (M€/year) 2.7 2.6 3.0 2.8
Thus, neither the reformer nor the anaerobic digester is included within CAPEX (% TAC) 92 90 94 92
OPEX (% TAC) 8 10 6 8
the energy system. This happens because one of the principles of the
inherent safety lies on the simplification of the process, so that a safer CO2 emissions (ton/year) 329.5 124.7 310.3 89.3
process is the one with less amount of equipment. In such a way, as the Grid emissions (%) 33 100 29 100
Process emissions (%) 67 71
electrolyzer is already installed for converting the surplus electricity into
– –
Biogenic (ton/year) – – – –
hydrogen, it is used to produce all the required hydrogen. In fact, even
Water consumption (m3/year) 134.1 276.7 193.4 313.6
though the electrolyzer was not employed as part of the storage system,
Electrolysis (%) 26 100 51 100
it would be the preferred option for obtaining hydrogen instead of the Reforming (%) 74 – 49 –
reformer. This fact can be verified by comparing the operating condi Digestion (%) – – – –
tions of the two processes. In this respect, the electrolyzer typically
Grid Dependence (%) 49 44 42 30
operates at 80 ◦ C and 4 MPa, whereas the reforming reaction is carried Imported electricity (%) 39 100 29 100
out at 800 ◦ C and 3 MPa. Imported natural gas (%) 61 – 71 –
Table 3 presents the main results for the five sustainability indicators Inherent Safety 23.0 12.1 23.8 13.4
when the water consumption and the inherent safety index are mini Chemical Index 1.8 1.5 2.5 2.2
mized. Moreover, these results also depict the impact of the PV surface Process Index 21.2 10.6 21.3 11.2
on the objective functions. In this respect, it is noted that the perfor
mance of both indicators improves as the area of PV decreases.
emissions and grid dependence indicators were taken from the reference
Regarding the water consumption, it occurs because a larger PV surface
[44].
leads to a higher amount of surplus electricity, and consequently more
water is required for converting such excess of electricity into hydrogen
3.2.1. Problem 1: Cost - CO2 emissions/grid dependence
for its storage. Similarly, the energy system becomes safer as the size of
Fig. 5 depicts the Pareto fronts for the two multi-objective optimi
the PV decreases, since smaller equipment are needed for energy con
zation problems: (i) economic-environmental (Fig. 5a), and (ii)
version and storage.
economic-social (Fig. 5b). Initially, such problems were solved consid
ering different sizes of the population for the genetic algorithm. In this
3.2. Multi-objective optimization respect, populations between 500 and 4000 individuals were evaluated
to identify the impact of this variable on the Pareto solutions. The cor
As aforementioned, the multi-objective optimization framework responding results are presented in Figs. S2 and S3 of the section S3 in
proposed in this work includes five objective functions. Therefore, the supplementary material. Accordingly, a population of 3000 in
aiming to summarize the results from the single-objective optimization, dividuals was selected, since for populations larger than 2000 in
and to ease the analysis of the multi-objective optimization problems, dividuals, no significant impact of this variable was observed on the
Fig. 4 depicts the obtained energy system configuration from the inde optimal set of solutions.
pendent assessment of each one of the five objective functions. It is First, note that the points A-A′ correspond to the best performance
noteworthy that the flowsheets corresponding to the economic, CO2 from the economic point of view, the points B-B′ represent the best
8
Fig. 4. Optimal configurations of the distributed energy system from single-objective optimization. Objective functions (a) total annualized cost, (b) CO2 emissions,
(c) water consumption, (d) grid dependence, (e) inherent safety index.
9
Fig. 5. Pareto fronts for the distributed energy system design. (a) CO2 emission - cost, (b) grid dependence – cost. Photovoltaic surface ( ) 7500 m2, ( ) 10000 m2.
(A – A′ ) optimal cost, (B – B′ ) optimal emission, and (C – C′ ) optimal grid dependence.
solution from the environmental perspective, and the points C-C′ invest between 2.3 and 2.5 M€, it would be better to install a PV surface
represent the minimal grid dependence. Thus, the obtained Pareto fronts of 10000 instead of 7500 m2, since for the same cost lower CO2 emis
reflect the competitive behavior of both pair of objectives, since as the sions and grid dependence could be obtained.
total annualized cost decreases, the CO2 emissions and the grid depen Comparing the energy system configuration for the extreme points of
dence get worse. the Pareto front, i.e. those obtained from the mono-objective optimi
Moreover, Fig. 5 also depicts the influence of the PV surface on the zation (Fig. 4), it is noted that the difference of the energy system
Pareto solutions. In this respect, a larger area of PV enables to achieve structure lies on the source of methane for the reforming process. Thus,
lower CO2 emissions and grid dependence, but it also entails a greater when the economic objective is addressed, all the methane is imported
economic cost. In general, these results provide a wide range of solutions from the network (Fig. 4a). In contrast, if the objective is to reduce the
that can be further evaluated for the decision-maker to their imple CO2 emissions or the grid dependence, the reformer reactor is fed by
mentation. Interestingly, the results show that there is a zone wherein methane from the anaerobic digestion process (Fig. 4b and d).
the PV surface of 7500 m2 will not be competitive against the area of Aiming to elucidate the differences in the design and operating
10000 m2. This occurs for the solutions between the points B and M conditions throughout the Pareto fronts, Fig. 6 presents the installed size
because all of them could be improved in at least one criterion by using a and the use of some energy sources as a function of the total annualized
PV surface of 10000 m2. Mathematically, this implies that the solutions cost. As noted in Fig. 6a and b, there is a direct relation between the
within the MB line are dominated by those obtained with a PV area of energy storage capacities and the total cost of the system. Therefore, the
10000 m2 (line A′ B′ ). In such a way, if the decision-maker is willing to best configurations from the economic perspective correspond to those
Fig. 6. Change of design and operation conditions across the Pareto fronts. Photovoltaic surface ( ) 7500 m2 and ( ) 10000 m2. (a) Battery, (b) pressurized tank, (c)
imported natural gas, and (d) biomass consumption. (A – A′ ) optimal cost, (B – B′ ) optimal emission, (C – C′ ) optimal grid dependence.
10
with the smallest storage units. Meanwhile, Fig. 6c and d depict the lowest grid dependence requires the greatest water consumption. Ac
change in the source of methane for the reforming process. As observed, cording to optimization results, it is also noted that for a given value of
the best performance for the CO2 emissions and grid dependence in the economic indicator (TAC), there is a wide range of possible energy
dicators is obtained without importing methane from the grid and by system structures and/or operating policies (Figs. 7b, d, 8b and d). Then,
using the maximum amount available of biomass. Then, as the economic for a fixed cost, the selection of the preferred alternative would require
performance improves, the imported natural gas increases and the defining a value for the water consumption or the grid dependence.
biomass gets unused, but at the cost of higher CO2 emissions and grid Figs. 7c and 8c depict the relation between these two objectives. In this
dependence. Interestingly, note that whilst the storage capacities are the respect, the decision would depend on the specific context conditions
most influencing variables on the cost criterion, the source of methane and/or decision-maker preferences considering the reliability of the grid
has the biggest impact on the emission and grid dependence issues. This and the availability of water. For instance, if the energy system is
fact can be observed by analyzing Figs. 5 and 6. In such a way, going developed in an isolated location or with regular energy outages, the
from low emission to low cost, it is observed that the most significative grid dependence indicator must be privileged to assure the access to
alteration on the slope of the Pareto curves corresponds to an important energy. Meanwhile, if the project is implemented in an arid zone, or in a
change in the curves of the imported natural gas and the biomass con place with low water resources, the indicator related to the water con
sumption. Roughly, this occurs at a value of 1.9 M€ for a PV surface of sumption will be the most important in the selection of the energy
7500 m2, and at 2.2 M€ when the area of PV is 10000 m2. Moreover, it is system.
also noted that the biomass utilization does not seem to have a great Fig. 9 presents the comparison of the Pareto solutions for the two
impact on the economic indicator (Fig. 6d). evaluated areas of PV. In those figures, the points A-A′ , B-B′ and C-C′
represent the optimal value of the economic, environmental, and social
3.2.2. Problem 2: Cost - water consumption - grid dependence objectives, respectively. In general, the results suggest that the total
The second multi-objective optimization problem consisted in the annualized cost and water consumption indicators improve as the PV
simultaneous minimization of the total annualized cost, water con surface decreases. Conversely, larger areas of PV favor the energy au
sumption and grid dependence. Initially, the problem was solved tonomy. In this respect, note that as the PV surface gets larger, the
considering different sizes of population for the genetic algorithm. Ac surplus of electricity increases, and consequently, there is a need of
cording to the obtained results, a population of 3000 individuals was bigger units for energy storage. For the economic and environmental
selected, since above 2000 individuals, no significant effect on the objectives, this implies higher investment cost due to the size of
Pareto solutions was observed. Results of such optimizations are pre equipment, and more elevated use of water for converting the surplus
sented in Fig. S4 of the supplementary material. Accordingly, Figs. 7 and electricity into hydrogen through the electrolyzer. In contrast, as the
8 depict the obtained Pareto sets considering a PV surface of 7500 and area of PV becomes larger, the grid dependence indicator is enhanced,
10000 m2, respectively. Such figures include a 3-dimension represen since in such a case, there is a greater amount of energy available from
tation (Figs. 7a and 8a), and the 2-dimension projections for the three renewable sources, and therefore less energy must be imported from the
evaluated objectives. grid.
Results show the trade-offs and the relationships among the objective Comparing the optimal configuration from the economic (Fig. 4a)
functions. In fact, the competition among the three indicators can be and environmental perspectives (Fig. 4c), note that both structures
observed. Thus, the minimum water consumption implies the poorest involve the same set of equipment for energy conversion and storage.
yield in the cost and self-sufficiency indicators. In the same line, the Indeed, according to the flowsheets, the only difference is in the sources
Fig. 7. Pareto solutions for minimizing the total annualized cost, the water consumption and the grid dependence considering a photovoltaic surface of 7500 m2. ( )
3-dimension representation ( ) 2-dimension projections.
11
Fig. 8. Pareto solutions for minimizing the total annualized cost, the water consumption and the grid dependence considering a photovoltaic surface of 10000 m2.
( ) 3-dimension representation ( ) 2-dimension projections.
Fig. 9. 2-dimension projection of the Pareto solutions for minimizing the total annualized cost, the water consumption, and the grid dependence. Photovoltaic
surface ( ) 7500 m2, ( ) 10000 m2. (A – A′ ) optimal cost, (B – B′ ) optimal water consumption, (C – C′ ) optimal grid dependence.
12
for supplying the hydrogen demand. Thus, whilst the economic opti and inherent safety indicators. As before, the first step consisted in
mization includes the possibility of providing hydrogen from water performing the optimization by considering different sizes of population
electrolysis, the water consumption indicator suggests that all the for the genetic algorithm. In this case, the evaluated values were 500 and
hydrogen must be supplied via steam methane reforming. In this respect, 2000 individuals, and according to the obtained results, no considerable
Fig. 10 shows the change in the design and operation parameters by effect of this variable on the set of Pareto solutions was observed. Such
going from the economic (A-A′ ) to the environmental (B-B′ ) indicator results are depicted in Fig. S5 of the supplementary material. Conse
across the Pareto solutions. On the one hand, as observed in Fig. 10a and quently, the results corresponding to a population of 2000 individuals
b, the economic optimum corresponds to the configurations with the were selected for analyzing the multi-objective optimization problem.
smallest size of the storage units. This happens because such devices Then, Figs. 11 and 12 present the obtained Pareto solutions considering
represent about 60–65% of the CAPEX of the system [44]. a PV surface of 7500 and 10000 m2, respectively.
On the other hand, note that moving towards the optimal value of the In general, results show the compromise among the objective func
water consumption requires to increase the capacity of energy storage tions and their contradictory behavior. As depicted in Fig. 11(12)b and
(Fig. 10a and b) and the amount of natural gas imported from the 11(12)d, there are at least two feasible energy system configurations or
network (Fig. 10c). Regarding the energy storage issue, it can be operating conditions that yield the same performance in the economic
explained by the fact that when the whole system electrolyzer-tank-fuel indicator. In such a way, the choice of the most suitable option involves
cell is employed, there is a possibility to recover the water produced in the evaluation of the trade-off between the water consumption and
the fuel cell to be reused in the electrolyzer. Consequently, the net water safety indicators, which is presented in Fig. 11(12)c. In those figures, the
consumption is lower than that obtained by sending the hydrogen to competition between these two objectives can be noted since the
supply the demand. Besides, the requirement of methane for the improvement of the safety index requires to increase the use of water.
reforming process increases as moving towards the minimal water Moreover, Fig. 13 illustrates the obtained Pareto solutions for the
consumption (points B and B′ ), since the hydrogen supplied by water two assessed areas of PV. In those figures, the points A-A′ , B-B′ and C-C′
electrolysis gets lower. represent the optimal value of the economic, environmental, and social
Moreover, the best performance for the grid dependence indicator is objectives, respectively. As mentioned in a previous multi-objective
obtained by feeding the reforming reactor with only the biomethane optimization problem (P2), the total annualized cost and water con
produced from the anaerobic digestion process (Fig. 4d). Accordingly, sumption indicators improve as the PV surface decreases. Likewise, the
this configuration corresponds to that without importing natural gas safety indicator also gets better for the smallest size of PV, as shown in
from the network and with the highest use of biomass, as presented by Fig. 13b and 13c. As aforementioned (P2), as the surface of PV increases,
the points C-C′ in the Fig. 10c and d. bigger units are required for converting the surplus electricity into
hydrogen and for energy storage. Consequently, this leads to an energy
3.2.3. Problem 3: cost – water consumption - safety system with higher costs, more elevated consumption of water, and
In this optimization problem, the three dimensions of sustainability more hazardous conditions.
are evaluated by means of the total annualized cost, water consumption The comparison between the results of the economic and water
Fig. 10. Change of design and operation conditions across the Pareto solutions. Photovoltaic surface ( ) 7500 m2 and ( ) 10000 m2. (a) battery, (b) pressurized
tank, (c) imported natural gas, and (d) biomass consumption. (A – A′ ) optimal cost, (B – B′ ) optimal water consumption, (C – C′ ) optimal grid dependence.
13
Fig. 11. Pareto solutions for minimizing the total annualized cost, the water consumption and the inherent safety considering a photovoltaic surface of 7500 m2. ( )
3-dimension representation ( ) 2-dimension projections.
Fig. 12. Pareto solutions for minimizing the total annualized cost, the water consumption and the inherent safety considering a photovoltaic surface of 10000 m2.
( ) 3-dimension representation ( ) 2-dimension projections.
14
Fig. 13. 2-dimension projection of the Pareto solutions for minimizing the total annualized cost, the water consumption, and the inherent safety. Photovoltaic
surface ( ) 7500 m2, ( ) 10000 m2. (A – A′ ) optimal cost, (B – B′ ) optimal water consumption, (C – C′ ) optimal inherent safety.
Fig. 14. Change of design and operating conditions across the Pareto solutions. Photovoltaic surface ( ) 7500 m2 and ( ) 10000 m2. (a) battery, (b) pressurized
tank, (c) imported natural gas, and (d) imported electricity. (A – A′ ) optimal cost, (B – B′ ) optimal water consumption, (C – C′ ) optimal inherent safety.
15
consumption indicators was already discussed in the optimization supplementary material. According to those results, a population of
problem P2. Meanwhile, note that the optimization of the safety index 2000 individuals was selected for analyzing the multi-objective opti
implies a new alternative for the energy system configuration (Fig. 4e). mization problems. Fig. 15 depicts the obtained Pareto front for both
In this regard, aiming to reduce the amount of equipment, the safety optimization cases and the two surfaces of PV evaluated.
indicator suggests a process flowsheet without the reforming reactor. Results show the antagonistic behavior of inherent safety index with
This result can be explained by two facts: (i) because of the intense respect to both the total annualized cost and the CO2 emissions.
operating conditions of the reforming process (temperature and pres Therefore, as the hazardous of the energy system decreases, the cost of
sure), and (ii) because the electrolyzer is already installed as a part of the the plant (Fig. 15a) and the emissions (Fig. 15b) become higher. Fig. 15
hydrogen storage system. Likewise, as the reformer is not used, also shows the influence of the PV surface on the Pareto fronts. In this
connection with the natural gas network is not required, since all the respect, note that the inherent safety improves as the area of PV de
hydrogen is obtained via water electrolysis. Additionally, according to creases. This happens because a smaller PV surface enables to reduce the
the results, the electrolyzer must be powered by both the PV and the size of the energy conversion and storage units, which favors the safety
electricity grid. of the system.
Fig. 14 illustrates the changes in design and operating conditions As observed by comparing the Fig. 4a, b and e, the topology of the
across the Pareto solutions. Going from the economic (A – A′ ) to the system mainly differs in the energy sources employed for supplying the
safety optimum (C – C′ ), it is noted that the capacity of the battery in demand of hydrogen. Thus, the economic objective focuses on produc
creases (Fig. 14a), whereas the size of the pressurized tank gets smaller ing hydrogen from two process options: (i) the steam methane reforming
(Fig. 14b). This happens because in improving the safety of the process, (using the gas network), and the water electrolysis powered by PV
the system seeks to reduce the inventory of material (hydrogen) as much (Fig. 4a). Meanwhile, the CO2 emissions indicator suggests including the
as possible, and consequently the battery becomes used more for storing anaerobic digester for obtaining biogas and subsequently perform the
the surplus electricity. Also, by moving towards the safest energy system reforming process (Fig. 4b). Besides, in this case, the electrolyzer is
configuration, the reforming reactor gets unused, and hence its installed powered by both the electricity grid and the PV. Unlike, the inherent
capacity decreases (Fig. 14c). At the same time, the electrolyzer is safety index does not include the reforming process within the energy
gradually more used for supplying the demand of hydrogen, and system (Fig. 4e).
consequently, the requirement of importing electricity from the grid Aiming to elucidate the relation of the optimization objectives in
increases, as depicted in Fig. 14d. terms of design and/ or operating conditions, Fig. 16 presents the evo
Interestingly, as shown in Fig. 13b and c, the greatest change on the lution of these variables across the Pareto fronts. As mentioned in the
safety index occurs between the economic and social points, i.e. lines AC optimization problem P3, the relation between the cost and safety in
and A′ C′ of those figures. Indeed, note that there is a change in the slope dicators can be explained through two process aspects: the size of the
of the curves in Fig. 13c after the points A and A′ . This can be explained units for energy storage, and the sources of hydrogen. In this regard, as
by the results depicted in Fig. 14b and c. As observed, whilst the capacity noted in Fig. 16a, the economic objective attempts to reduce the ca
of the pressurized tank decreases across the lines BA and B′ A′ , the size of pacity of the battery because of its high investment cost. In contrast, the
the reformer remains constant. This implies that in that zone only the safety indicator focuses on reducing the amount of hydrogen stored
size of the pressurized tank affects the safety index, which translates into (Fig. 16c), and the size of the reformer because of its intense operating
the smaller effect presented in Fig. 13b and c. Conversely, in the region conditions (Fig. 16e). Therefore, moving from the safest to the
represented by lines AC and A′ C′ , the capacity of both the pressurized economical energy system requires to increase the capacity of the
tank and the reformer reactor decreases, which leads to a bigger impact pressurized tank, to install and expand the capacity of the reforming
on the safety of the system. reactor, and to decrease the size of the battery.
On the other hand, the right column in the Fig. 16 shows the changes
3.2.4. Problem 4: Safety – cost/CO2 emissions of the energy system when the CO2 emissions and the safety indicators
This case comprises two multi-objective optimization problems: (i) are addressed. Again, the safety index seeks to reduce capacity of the
the inherent safety against the total annualized cost, and (ii) the inherent pressurized tank, whereas the environmental indicator requires a large
safety against the CO2 emissions. As before, these problems were solved storage capacity for exploiting the energy production from renewables
by using a variety of sizes of population for the genetic algorithm. The (Fig. 16b). Moreover, as the safety of the system increases, the electro
values considered were 1000, 2000 and 3000 individuals. Results of lyzer becomes the preferred alternative to obtain hydrogen. Accord
these optimizations are presented in Figs. S6 and S7 of the ingly, the amount of electricity imported from the grid gets higher, as
Fig. 15. Pareto fronts for optimizing the total annualized cost, the CO2 emissions, and the inherent safety. (a) safety - cost, (b) safety – CO2 emissions. Photovoltaic
surface ( ) 7500 m2, ( ) 10000 m2. (A – A′ ) optimal cost, (B – B′ ) optimal emission, and (C – C′ ) optimal inherent safety.
16
Fig. 16. Change of design and operating conditions across the Pareto fronts. Photovoltaic surface ( ) 7500 m2 and ( ) 10000 m2. Left column (a,c,e) cost vs inherent
safety, right column (b,d,f) CO2 emissions vs inherent safety. (A – A′ ) optimal cost, (B – B′ ) optimal emission, and (C – C′ ) optimal inherent safety.
depicted in Fig. 16d. Thus, as the CO2 emissions grow, the anaerobic negative and positive impacts, respectively. For instance, a larger sur
digester and reformer reactor get unused, and hence the use of biomass face of PV enables the reduction of CO2 emissions and grid dependence
decreases (Fig. 16f). as a higher amount of renewable-based electricity is available. However,
this also leads to a more expensive and potentially more hazardous
4. General insights for the design of DES energy system since bigger units for energy conversion and storage are
needed. Besides, the performance of the water consumption indicator
Throughout the previous section, different multi-objective optimi gets worse because the amount of water required for converting the
zation problems were addressed for the energy system design consid surplus electricity into hydrogen also increases.
ering the sustainability dimensions. Besides, the obtained Pareto sets Regarding the alternatives for electricity storage, it is noted that both
were explored for studying the compromises among the criteria, and the options enhance the indicators of the environmental dimension and
evolution of the process variables across the non-dominated solutions. promote the independence of the system for the energy supply. On the
Accordingly, aiming to summarize those findings, this section is dedi one hand, using energy storage technologies favors the exploitation of
cated to state the main trends and to identify some general insights for renewable-based electricity, since they enable to deal with the mismatch
the conceptual design of distributed energy systems. between electricity production and consumption. Thereby, CO2 emis
Fig. 17 presents the impact of the analyzed process variables on the sions and grid dependence indicators are improved because a lower
sustainability indicators. It is worth bearing in mind that the perfor amount of energy must be imported from the grid. On the other hand,
mance of the indicators improves as their corresponding value de the electrical battery does not require water for its operation, and the
creases. In such a way, ascending and descending arrows represent power-to-power system allows the recovery of water from the fuel cell to
17
Fig. 17. Impact of increasing the process variables on the sustainability indicators. (a) the use of biomass entails anaerobic digestion and reforming processes, (b) the
gas network is used for steam methane reforming process, (c) grid is used for electrolysis of water.
be reused in the electrolyzer. Consequently, they have a positive impact of the energy system design to enhance the environmental and social
on the water consumption indicator. Additionally, the obtained results aspects. Then, these were added to a framework considering the total
also indicate that the economic aspect is improved by reducing the size annualized cost, the CO2 emissions, and the grid dependence to perform
of the storage units, which evidences the competition among the sus the multi-objective optimization. Altogether, four multi-objective opti
tainability indicators. Moreover, concerning the inherent safety of the mization problems were formulated and solved for different combina
system, results suggest reducing capacity of the pressurized tank as it tions of the sustainability indicators. From multi-objective optimization
represents the accumulation of a potentially hazardous material. results, the relationships among the objective functions were estab
Otherwise, as observed in Fig. 17, the impact of the technologies for lished, and a wide spread of plausible system configurations and oper
obtaining hydrogen was also assessed. These alternatives include the use ating conditions were obtained. Broadly, results reflect the compromise
of biomass through the anaerobic digestion and the subsequent and the antagonistic behavior among the sustainability criteria. Thus,
reforming process, the steam methane reforming of natural gas from the according to multi-objective optimization results, it was determined the
network, and the electrolysis of water by using electricity from the grid. competition between the total annualized cost and the CO2 emissions
In this regard, it is worth to note that no technology offers a solution able and grid dependence objectives. In such a way, for a given surface of PV,
to simultaneously get the best performance of all the sustainability the total annualized cost decreases as the performance in the emissions
criteria. For instance, using biomass enables to reduce the CO2 emissions and self-sufficiency indicators gets worse. Also, outcomes indicate that
and the dependence of the main grid, but at the cost of a higher in the water consumption and the grid dependence are contradictory in
vestment, a more elevated water consumption, and riskier process dicators. Likewise, the obtained results depict the competition between
conditions. Meanwhile, the gas network represents a better alternative the inherent safety index with both the cost and emissions objectives.
from the economic point of view as the anaerobic digestion step is not From this analysis, the following outcomes can be drawn:
required, and because of the low price of the natural gas. However, this
pathway also leads to CO2 emissions from fossil fuels, it corresponds to • Depending on the objective function, the cost of energy range be
an option highly reliant on the main grid, and it entails the risk associ tween 0.37 and 0.63 €/kWh and the CO2 emissions vary between
ated to the intense process conditions of the reforming reaction. In this 10.6 and 68.5 kgCO2/MWh.
respect, it is noted that the safety of the process can be improved by • The best performance for the CO2 emissions (10.6 kgCO2/MWh) and
increasing the use of electricity from the grid for producing hydrogen. grid dependence (20%) indicators entails to install the largest ca
Nonetheless, in such a way, the performance of the other indicators pacity of the storage units and the highest amount of equipment
becomes worse. within the energy system. Consequently, this scenario leads to the
energy system with the most elevated water consumption (70.2
5. Conclusions m3H2O/GWh) and the most hazardous conditions.
• The water consumption can be reduced to around 28 m3H2O/GWh
In this work a multi-objective optimization analysis for the design by producing the hydrogen by means of steam reforming of methane.
and operation of energy systems considering the sustainability di Nevertheless, this also implies an energy system structure with the
mensions was carried out. Moreover, two new indicators, namely water highest cost (0.63 €/kWh), CO2 emissions (68.5 kgCO2/MWh), and
consumption and inherent safety index, were included for the evaluation grid dependence (50%).
18
• The safest energy system conditions are obtained by producing the [11] Bakshi BR, Fiksel J. The quest for sustainability: Challenges for process systems
engineering. AIChE J 2003;49:1350–8. https://doi.org/10.1002/aic.690490602.
whole hydrogen through the water electrolysis route and by reducing
[12] Rong A, Lahdelma R. Role of polygeneration in sustainable energy system
the size of the pressurized tank as much as possible. Indeed, this development challenges and opportunities from optimization viewpoints. Renew
configuration entails to install a PV area of 7500 m2 and leads to a Sustain Energy Rev 2016;53:363–72. https://doi.org/10.1016/j.rser.2015.08.060.
cost of energy of 0.54 €/kWh, an emission factor of 26 kgCO2/MWh, [13] Alanne K, Saari A. Distributed energy generation and sustainable development.
Renew Sustain Energy Rev 2006;10:539–58. https://doi.org/10.1016/j.
and a water intensity of 57.5 m3H2O/GWh. rser.2004.11.004.
[14] Adil AM, Ko Y. Socio-technical evolution of Decentralized Energy Systems: A
Otherwise, the set of Pareto solutions was explored and analyzed for critical review and implications for urban planning and policy. Renew Sustain
Energy Rev 2016;57:1025–37. https://doi.org/10.1016/j.rser.2015.12.079.
identifying the changes in the design and operating conditions across the [15] Gabrielli P, Gazzani M, Martelli E, Mazzotti M. Optimal design of multi-energy
optimization results. Overall, these results could support the subsequent systems with seasonal storage. Appl Energy 2018;219:408–24. https://doi.org/
decision-making process since they depict the trade-off among the sus 10.1016/j.apenergy.2017.07.142.
[16] Gabrielli P, Fürer F, Mavromatidis G, Mazzotti M. Robust and optimal design of
tainability dimensions, and the impact of any decision in terms of design multi-energy systems with seasonal storage through uncertainty analysis. Appl
and operation variables. Energy 2019;238:1192–210. https://doi.org/10.1016/j.apenergy.2019.01.064.
The proposed framework could be easily adapted and used for the [17] Jing R, Wang M, Liang H, Wang X, Li N, Shah N, et al. Multi-objective optimization
of a neighborhood-level urban energy network: Considering Game-theory inspired
design of energy systems in different context conditions, considering multi-benefit allocation constraints. Appl Energy 2018;231:534–48. https://doi.
other criteria for evaluation, or including different technological units org/10.1016/j.apenergy.2018.09.151.
and energy forms. Moreover, as a perspective, the integration of the [18] Falke T, Krengel S, Meinerzhagen AK, Schnettler A. Multi-objective optimization
and simulation model for the design of distributed energy systems. Appl Energy
decision-maker preferences into the energy system design through a
2016;184:1508–16. https://doi.org/10.1016/j.apenergy.2016.03.044.
decision-aid making tool is envisaged. This would enable to classify the [19] Ren H, Zhou W, Nakagami K, Gao W, Wu Q. Multi-objective optimization for the
obtained Pareto sets and to select the most suitable alternative for its operation of distributed energy systems considering economic and environmental
implementation. aspects. Appl Energy 2010;87:3642–51. https://doi.org/10.1016/j.
apenergy.2010.06.013.
[20] Di Somma M, Yan B, Bianco N, Graditi G, Luh PB, Mongibello L, et al. Multi-
CRediT authorship contribution statement objective design optimization of distributed energy systems through cost and
exergy assessments. Appl Energy 2017;204:1299–316. https://doi.org/10.1016/j.
egypro.2017.03.706.
Juan D. Fonseca: Conceptualization, Methodology, Formal analysis, [21] Dorotić H, Pukšec T, Duić N. Multi-objective optimization of district heating and
Software, Investigation, Writing - original draft. Jean-Marc Com cooling systems for a one-year time horizon. Energy 2019;169:319–28. https://doi.
org/10.1016/j.energy.2018.11.149.
menge: Conceptualization, Formal analysis, Software, Writing - review
[22] Dorotić H, Pukšec T, Duić N. Economical, environmental and exergetic multi-
& editing, Supervision. Mauricio Camargo: Conceptualization, Writing objective optimization of district heating systems on hourly level for a whole year.
- review & editing, Supervision, Resources, Funding acquisition. Lau Appl Energy 2019;251:113394. https://doi.org/10.1016/j.apenergy.2019.113394.
[23] Dufo-López R, Bernal-Agustín JL. Multi-objective design of PV-wind-diesel-
rent Falk: Conceptualization, Writing - review & editing, Supervision,
hydrogen-battery systems. Renew Energy 2008;33:2559–72. https://doi.org/
Resources, Funding acquisition. Iván D. Gil: Conceptualization, Writing 10.1016/j.renene.2008.02.027.
- review & editing, Supervision. [24] Cedillos Alvarado D, Acha S, Shah N, Markides CN. A Technology Selection and
Operation (TSO) optimisation model for distributed energy systems: Mathematical
formulation and case study. Appl Energy 2016;180:491–503. https://doi.org/
10.1016/j.apenergy.2016.08.013.
Declaration of Competing Interest [25] Mayer MJ, Szilágyi A, Gróf G. Environmental and economic multi-objective
optimization of a household level hybrid renewable energy system by genetic
The authors declare that they have no known competing financial algorithm. Appl Energy 2020;269:115058. https://doi.org/10.1016/j.
apenergy.2020.115058.
interests or personal relationships that could have appeared to influence [26] Hou J, Wang J, Zhou Y, Lu X. Distributed energy systems: Multi-objective
the work reported in this paper. optimization and evaluation under different operational strategies. J Clean Prod
2021;280:124050. https://doi.org/10.1016/j.jclepro.2020.124050.
[27] Karmellos M, Mavrotas G. Multi-objective optimization and comparison framework
Acknowledgment for the design of Distributed Energy Systems. Energy Convers Manag 2019;180:
473–95. https://doi.org/10.1016/j.enconman.2018.10.083.
This work was supported partly by the French PIA project « Lorraine [28] Yan J, Broesicke OA, Tong X, Wang D, Li D, Crittenden JC. Multidisciplinary design
optimization of distributed energy generation systems: The trade-offs between life
Université d’Excellence », reference ANR-15-IDEX-04-LUE » cycle environmental and economic impacts. Appl Energy 2020;284:116197.
https://doi.org/10.1016/j.apenergy.2020.116197.
[29] Weinand JM, Scheller F, McKenna R. Reviewing energy system modelling of
Appendix A. Supplementary material
decentralized energy autonomy. Energy 2020;203:117817. https://doi.org/
10.1016/j.energy.2020.117817.
Supplementary data to this article can be found online at https://doi. [30] Campos-Guzmán V, García-Cáscales MS, Espinosa N, Urbina A. Life Cycle Analysis
org/10.1016/j.apenergy.2021.116746. with Multi-Criteria Decision Making: A review of approaches for the sustainability
evaluation of renewable energy technologies. Renew Sustain Energy Rev 2019;104:
343–66. https://doi.org/10.1016/j.rser.2019.01.031.
References [31] Azapagic A, Stamford L, Youds L, Barteczko-Hibbert C. Towards sustainable
production and consumption: A novel DEcision-Support Framework IntegRating
Economic, Environmental and Social Sustainability (DESIRES). Comput Chem Eng
[1] International Energy Agency (IEA). Global Energy Review 2020; 2020.
2016;91:93–103. https://doi.org/10.1016/j.compchemeng.2016.03.017.
[2] International Energy Agency (IEA). Global Energy & CO2 Status Report; 2019.
[32] United Nations. Sustainable Development Goals. Goal 6 Ensure Access to Water
[3] International Renewable Energy Agency (IRENA). Global Energy transformation:A
Sanit All 2020. https://www.un.org/sustainabledevelopment/water-and-sanita
roadmap to 2050. Abu Dhabi; 2018.
tion/ [accessed May 16, 2020].
[4] International Energy Agency (IEA). Status of Power System Transformation 2019:
[33] Evans A, Strezov V, Evans TJ. Assessment of sustainability indicators for renewable
Power System Flexibility; 2019.
energy technologies. Renew Sustain Energy Rev 2009;13:1082–8. https://doi.org/
[5] REN21. Renewables 2019 Global Status Report. Paris; 2019.
10.1016/j.rser.2008.03.008.
[6] Liu P, Georgiadis MC, Pistikopoulos EN. An energy systems engineering approach
[34] Ruiz-Mercado GJ, Smith RL, Gonzalez MA. Sustainability indicators for chemical
for the design and operation of microgrids in residential applications. Chem Eng
processes: I. Taxonomy. Ind Eng Chem Res 2012;51:2309–28. https://doi.org/
Res Des 2013;91:2054–69. https://doi.org/10.1016/j.cherd.2013.08.016.
10.1021/ie102116e.
[7] Mavromatidis G, Orehounig K, Bollinger LA, Hohmann M, Marquant JF, Miglani S,
[35] Shaaban M, Scheffran J. Selection of sustainable development indicators for the
et al. Ten questions concerning modeling of distributed multi-energy systems. Build
assessment of electricity production in Egypt. Sustain Energy Technol Assessments
Environ 2019;165:106372. https://doi.org/10.1016/j.buildenv.2019.106372.
2017;22:65–73. https://doi.org/10.1016/j.seta.2017.07.003.
[8] Dimian AC, Bildea CS, Kiss AA. Integrated Design and Simulation of Chemical
[36] Moslehi S, Arababadi R. Sustainability Assessment of Complex Energy Systems
Processes, 2nd ed. Amsterdam: Elsevier B.V; 2014.
Using Life Cycle Approach-Case Study: Arizona State University Tempe Campus.
[9] Smith R. Chemical Process Design and Integration. John Wiley & Sons, Ltd; 2005.
Procedia Eng 2016;145:1096–103. https://doi.org/10.1016/j.
[10] Ruiz-mercado G, Cabezas H. Sustainability in the Design, Synthesis and Analysis of
proeng.2016.04.142.
Chemical Engineering Processes. Elsevier - Butterworth-Heinemann; 2016.
19
[37] Abu-Rayash A, Dincer I. Sustainability assessment of energy systems: A novel [50] Jafari MJ, Mohammadi H, Reniers G, Pouyakian M, Nourai F, Torabi SA, et al.
integrated model. J Clean Prod 2019;212:1098–116. https://doi.org/10.1016/j. Exploring inherent process safety indicators and approaches for their estimation: A
jclepro.2018.12.090. systematic review. J Loss Prev Process Ind 2018;52:66–80. https://doi.org/
[38] International Atomic Energy Agency (IAEA), United Nations Department of 10.1016/j.jlp.2018.01.013.
Economic and Social Affairs, International Energy Agency (IEA), EUROSTAT, [51] Heikkilä AM. Inherent safety in process plant design. An index-based approach.
European Environment Agency. Energy Indicators for Sustainable Development: Helsinki University of Technology; 1999.
Guidelines and Methodologies; 2005. [52] Kletz T, Amyotte P. Process plants: A handbook for inherently safer design. Second.
[39] United Nations. Sustainable Development Goals. Goal 7 Afford Clean Energy 2020. Boca Raton: CRC Press; 2010. 10.1201/9781439804568.
https://www.un.org/sustainabledevelopment/energy/ [accessed May 12, 2020]. [53] Li X, Zanwar A, Jayswal A, Lou HH, Huang Y. Incorporating exergy analysis and
[40] Santoyo-Castelazo E, Azapagic A. Sustainability assessment of energy systems: inherent safety analysis for sustainability assessment of biofuels. vol. 50; 2011.
Integrating environmental, economic and social aspects. J Clean Prod 2014;80: 10.1021/ie101660q.
119–38. https://doi.org/10.1016/j.jclepro.2014.05.061. [54] Laurence D. Quantifying inherent safety of chemical process routes. Loughborough
[41] Stamford L, Azapagic A. Sustainability indicators for the assessment of nuclear University of Technology; 1996.
power. Energy 2011;36:6037–57. https://doi.org/10.1016/j.energy.2011.08.011. [55] Gangadharan P, Singh R, Cheng F, Lou HH. Novel Methodology for Inherent Safety
[42] Park S, Xu S, Rogers W, Pasman H, El-Halwagi MM. Incorporating inherent safety Assessment in the Process Design Stage. Ind Eng Chem Res 2013;52:5921–33.
during the conceptual process design stage: A literature review. J Loss Prev Process https://doi.org/10.1021/ie303163y.
Ind 2020;63:104040. https://doi.org/10.1016/j.jlp.2019.104040. [56] (ADEME) Agence de l’Environnement et de la Maîtrise de l’Energie. MODECOM
[43] Roy N, Eljack F, Jiménez-Gutiérrez A, Zhang B, Thiruvenkataswamy P, El- 2017 Campagne nationale de caractérisation des déchets ménagers et assimilés.
Halwagi M, et al. A review of safety indices for process design. Curr Opin Chem Eng Angers; 2019.
2016;14:42–8. https://doi.org/10.1016/j.coche.2016.07.001. [57] (ADEME) Agence de l’Environnement et de la Maîtrise de l’Energie. Déchets
[44] Fonseca JD, Commenge J-M, Camargo M, Falk L, Gil ID. Multi-criteria optimization Chiffres-clés. Angers; 2018.
for the design and operation of distributed energy systems considering [58] Pandu Rangaiah G. Multi-objective Optimization: Techniques and Applications in
sustainability dimensions. Energy 2021;214. https://doi.org/10.1016/j. Chemical Engineering. Singapore: World Scientific Publishing; 2009. 10.1007/
energy.2020.118989. 978-1-84800-382-8_2.
[45] Association française pour l’hydrogene et les piles à combustible (AFHYPAC). [59] Diwekar U. Introduction to Applied Optimization, 2nd ed. Springer; 2008.
Production D’Hydrogene Par Electrolyse De L’Eau. vol. Fiche 3.2; 2019. 10.1007/978-0-387-73669-3.
[46] Angelonidi E, Smith SR. A comparison of wet and dry anaerobic digestion processes [60] Coello Coello CA, Lamont GB, Veldhuizen DA Van, Goldberg DE, Koza JR.
for the treatment of municipal solid waste and food waste. Water Environ J 2015; Evolutionary Algorithms for Solving Multi-Objective Problems. 2nd ed. Springer;
29:549–57. https://doi.org/10.1111/wej.12130. 2007. 10.1007/978-0-387-36797-2.
[47] Athar M, Shariff AM, Buang A. A review of inherent assessment for sustainable [61] Pandu Rangaiah G, Bonilla-Petriciolet A. Multi-Objective Optimization in Chemical
process design. J Clean Prod 2019;233:242–63. https://doi.org/10.1016/j. Engineering: Developments and Applications. John Wiley & Sons, Ltd; 2013.
jclepro.2019.06.060. 10.6028/jres.090.045.
[48] Song D, Yoon ES, Jang N. A framework and method for the assessment of inherent [62] Cui Y, Geng Z, Zhu Q, Han Y. Review: Multi-objective optimization methods and
safety to enhance sustainability in conceptual chemical process design. J Loss Prev application in energy saving. Energy 2017;125:681–704. https://doi.org/10.1016/
Process Ind 2018;54:10–7. https://doi.org/10.1016/j.jlp.2018.02.006. j.energy.2017.02.174.
[49] Rahman M, Heikkilä AM, Hurme M. Comparison of inherent safety indices in
process concept evaluation. J Loss Prev Process Ind 2005;18:327–34. https://doi.
org/10.1016/j.jlp.2005.06.015.
20

Fonseca 2021

Uploaded by

Copyright:

Available Formats

Fonseca 2021

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Fonseca 2021

Uploaded by

Copyright:

Available Formats

Applied Energy 290 (2021) 116746

Contents lists available at ScienceDirect

Sustainability analysis for the design of distributed energy systems: A

• Inherent safety index of the energy sys

Nomenclature u Decision variables

Taking this into account, it is noted that sustainability requires a holistic

conceptual design of DES. Accordingly, the remainder of this document 2. Methodology

The mass and energy balances, as well as the performance of the

of reactions based on the combination of chemicals, and a connection

The former because the score of reaction is considered as a function of

was used because it enables to simultaneously obtain the entire set of

stoichiometry of reactions. Thus, by considering the stochiometric

You might also like

Fonseca 2021

Uploaded by

Copyright:

Available Formats

Fonseca 2021

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Fonseca 2021

Uploaded by

Copyright:

Available Formats

Applied Energy 290 (2021) 116746

Contents lists available at ScienceDirect

Sustainability analysis for the design of distributed energy systems: A

• Inherent safety index of the energy sys­

Nomenclature u Decision variables

Taking this into account, it is noted that sustainability requires a holistic

conceptual design of DES. Accordingly, the remainder of this document 2. Methodology

The mass and energy balances, as well as the performance of the

of reactions based on the combination of chemicals, and a connection

The former because the score of reaction is considered as a function of

was used because it enables to simultaneously obtain the entire set of

stoichiometry of reactions. Thus, by considering the stochiometric

You might also like

• Inherent safety index of the energy sys