You are on page 1of 19

Computers and Chemical Engineering 126 (2019) 35–53

Contents lists available at ScienceDirect

Computers and Chemical Engineering

journal homepage:

An integral methodological approach for biorefineries design: Study

case of Colombian coffee cut-stems
Valentina Aristizábal-Marulanda a, Carlos Ariel Cardona Alzate a,∗, Mariano Martín b,∗
Instituto de Biotecnología y Agroindustria, Departamento de Ingeniería Química, Universidad Nacional de Colombia, Manizales campus, Km 07 vía al
Magdalena, Manizales, Colombia
Department of Chemical Engineering, Universidad de Salamanca, Pza. Caídos 1-5, 37008 Salamanca, Spain

a r t i c l e i n f o a b s t r a c t

Article history: This paper presents a generic and integral methodological approach for the design and assessment of
Received 29 July 2018 multiproduct biorefineries. The proposed methodology has been developed based on 2-stage strategy,
Revised 12 February 2019
conceptual design and optimization. In the conceptual design stage, the theoretical data about the biore-
Accepted 30 March 2019
finery are collected in order to build the superstructure. In the optimization stage, the formulated super-
Available online 6 April 2019
structure based on surrogate models, is optimized in order to generate the optimal biorefinery configu-
Keywords: ration where the utilities are produced internally from biomass via a cogeneration system. The proposed
Biorefinery framework is applied to evaluate the optimal exploitation of Coffee Cut-Stems (CCS). The results show
Design that ethanol is the most interesting product following dilute acid pretreatment and the fermentation of
Methodological approach C5 and C6 sugars due to the large yield of the process. However, an alternative scenario where energy is
Conceptual design and optimization the protagonist, results in the selection of power and heat as main products.
© 2019 Elsevier Ltd. All rights reserved.

1. Introduction framework for the preliminary assessment of chemical processes at

the laboratory stage (Moncada et al., 2015; Posada et al., 2013).
A biorefinery is defined as a system that involves assessing and Conceptual design is the methodological approach where hi-
using a wide range of technologies to separate biomass into its erarchy, sequencing and integration concepts are interlinked and
principal constituents (carbohydrates, protein, triglycerides, etc.), their combination carries to a process synthesis approach to obtain
which can be transformed in numerous products (fuels, chemicals, theoretical and practical process models (Douglas, 1988; Smith,
power and heat, materials, and food and feed) (Moncada et al., 2005). This approach has been applied to biorefinery level. For ex-
2016). In the last years, these systems have attracted attention ample, Aristizábal et al. (2016); Cerón et al. (2015); Dávila et al.
in order to maximize the value of the products obtained from (2017); Moncada et al. (2013) used this approach for the concep-
biomass, to increase competitiveness and prosperity of bioindustry, tual design of multiproduct biorefineries from first (i.e., sugarcane,
to reduce the dependency on fossil fuels by producing the same or copoazu fruit and avocado), second (i.e., fresh fruit bunches) and
analogous substances demanded in the current market, to reduce third (i.e., microalgae) generation. This methodology considers dif-
the emission of greenhouse gases and to stimulate regional and ferent scenarios, where technologies, feedstock distribution and in-
rural development (Moncada et al., 2016). Currently, three differ- tegration levels (mass and energy), among others, are taken as
ent approaches are used for the design and analysis of biorefiner- comparative parameters for the economic and environmental as-
ies: conceptual design, optimization and the early-stage method. sessment of the biorefineries.
The conceptual design is based on hierarchy, sequencing and inte- On the other hand, many authors have applied the optimiza-
gration concepts (Douglas, 1988; Botero, 2012; Smith, 2005). The tion methodology to design of all types of biorefineries, such as,
optimization uses superstructure schemes and mathematical mod- (Bao et al., 2011; Bertran et al., 2017; Kelloway and Daoutidis, 2014;
eling (Giarola et al., 2011; Sammons et al., 2007; Zondervan et al., Pham and El-Halwagi, 2011; Ponce-Ortega et al., 2012; Rosen-
2011). The early-stage method employs a multi-objective decision gart, 2017; Zondervan et al., 2011). This approach considers a li-
brary of compact models for the generation of a superstructure of
transformation routes and products (i.e., a tree with all models).
∗ The methodology guarantees the selection of an optimal biore-
Corresponding authors.
E-mail addresses: (C.A. Cardona Alzate), finery configuration assessing different optimization objectives as (M. Martín). maximizing yield and profit, minimizing costs, waste, fixed cost,
0098-1354/© 2019 Elsevier Ltd. All rights reserved.
36 V. Aristizábal-Marulanda, C.A. Cardona Alzate and M. Martín / Computers and Chemical Engineering 126 (2019) 35–53

point of view at early design stages. This methodology allows

Nomenclature to determine the processes sustainability and to diagnose their
strengths and weaknesses. Three sustainability indicators are con-
Variables sidered, (i) economic constrain (EC), (ii) energy related impact of
Z objective function (USD y−1 ) raw materials (EIRM) and (iii) process complexity (PC), which are
C cost consigned in a single index with the help of weighting factors to
f component flow have a fast comparison and to choose the better process (Moncada
S split of mass fraction (sugar) et al., 2015; Posada et al., 2013).
In this sense, this work proposes a general methodological ap-
Binary variables
proach for the design and analysis of biorefineries that promotes
y selection of process unit (binary)
the idea of using suitable tools and strategies to identify, select,
Subscripts and guarantee the existence of a promising biorefinery. The inte-
i component gral methodological approach has been developed based on a 2-
ligno components of lignocellulosic material stage strategy, conceptual design and optimization, focused on the
k unit generation of processing structures of biomass. Existing approaches
Rem remaining component develop similar methodological path in terms of theoretical foun-
r reaction dation, superstructure scheme, use of different technologies for the
n number of distribution stream same product and mathematical modeling (Bertran et al., 2017;
Moncada et al., 2016; Rosengart, 2017; Zondervan et al., 2011). De-
Superscripts spite these similarities, the proposed methodology is presented as
P product an evolution of existing approaches. The approach proposed in this
F feedstock work is based on 2-stage approach where the superstructure is
C chemicals put together using surrogate models for all of the stages based on
Reag reagent component rigorous simulation (Aspen Plus) and/or using experimental data
Ad added component versus typical short cut models. Furthermore, an integrated facility
Util utility is considered that provides the utilities required for it to be self-
ww wastewater management sustainable, a consideration that results in the partial consumption
Inv inversion of raw material that has not previously considered in the litera-
IN unit inlet ture. So far, C5 andC6 sugars were no distinguished as resource
OUT unit outlet for different chemicals. This feature is added to the superstructure
S substrate due to the additional possibilities that can provide to the biorefin-
Liq1 liquid in filter 1 ery. Finally, wastewater treatment is also integrated. This approach
Sol1 solid in filter 1 is used in an example of strategic importance for Colombia, the
R outlet of reaction evaluation of the use of the residues from the coffee production
Liq2 liquid in filter 2 industry. However, the methodology presented is general and can
Sol2 solid in filter 2 be used for any other case.
w waste This work is organized as follows: Section 2 indicates a de-
Liq3 liquid in filter 3 tailed explanation about the general methodological approach for
Sol3 solid in filter 3 the design and analysis of biorefineries. Afterwards, in Section 3,
n reaction number a study case is presented and solved using the proposed method-
Parameters ology and well-known computational tools (i.e., GAMS and Aspen
φ mass fraction of feedstock Plus). Subsequently, the results obtained are explained and dis-
Pr sale price cussed in Section 4. Finally, conclusions and future research works
β utility consumption respect to feedstock flow in the are highlighted in Section 5.
processing stage
μ proportion of added component respect to feedstock 2. Integral methodological approach
α yield
x fraction A general methodological approach is proposed that fuses two
MW molar weight well-known methodologies, conceptual design and optimization,
θ conversion which are articulated through the formulation of a superstructure
γ stoichiometric coefficient problem (Martín and Grossmann, 2012; Moncada et al., 2016). A
δ proportion of added component respect to reference superstructure of alternatives combines continuous (e.g., operating
compound conditions) and discrete variables (e.g., feedstock distribution). Ac-
σ fraction separated of product 1 cording to the decision-making nature of the problem, its approach
ϱ fraction separated of product 2 and solution is based on a problem of Mixed-Integer Nonlinear
M large number of big-M Programming (MINLP). This problem is solved in order to generate
the optimal biorefinery topology.
The superstructure formulation depends on biorefinery goals.
environmental impact, etc. This methodology has the limitation Likewise, the problem formulation requires large amounts of data
that does not allow the selection of various conversion routes taken from various resources, development and assessment of all
that are simultaneously attractive from two or more perspectives alternatives, mathematical model-building of alternatives and so-
(Ponce-Ortega et al., 2012). lution of the mathematical programming problem (Bertran et al.,
Moncada et al. (2015); Patel et al. (2012); Posada et al. 2017). All steps mentioned before are required for each con-
(2013) proposed the use of early-stage method for the biorefineries stituent unit of system that is assessed. Fig. 1 indicates in detail
design, which aims to assess systematically the biorefinery prod- the steps that compose the proposed methodological approach for
ucts to identify those with the best potential from a sustainability biorefineries design. The methodology does not consider loops or
V. Aristizábal-Marulanda, C.A. Cardona Alzate and M. Martín / Computers and Chemical Engineering 126 (2019) 35–53 37

• Step 2: Apply know-how experience, hierarchical decomposi-

tion and sequencing concepts

This step seeks to structure the biorefinery in a logical and in-

terconnected way based on knowledge and experience. This step
is used for the selection of technologies through a hierarchical
decomposition concept, which corresponds to the onion diagram
where the reaction stage is the fundamental part in the process
or system (Sánchez Toro, 2008). This heuristic concept is empha-
sized in a decomposition strategy and alternatives trace, which al-
lows the "quick" location of technological configurations close to
optimal solutions (Sánchez Toro, 2008). In addition, the nature of
this approach allows the definition of the process diagram in a se-
quential and evolutionary way and discards configurations that do
not lead to "good" designs, reducing the number of alternatives
(Sánchez Toro, 2008). In this step, a set of technologies are de-
fined that are used in the superstructure. The maturity level and
information about the available technologies contribute to define
promising and critical processing routes.
• Step 3: Collect experimental results, data from literature,
databases and bibliography

Feedstock characterization, yields, conversions, operating condi-

tions, process configuration, costs of feedstock, reagents and prod-
ucts are data acquired from bibliographic resources, academic or
industrial partners and databases. If some information is not avail-
able, it must be estimated (e.g., equipment cost) on experimentally

2.2. Stage 2: Optimization

The optimization stage has as objective to decide on the topol-

Fig. 1. Generic methodology for the biorefineries design. ogy of the biorefinery by solving the superstructure problem
through mathematical programming.
feedbacks because in the first steps, the user can involve an unde- • Step 4: Propose a superstructure
fined number of feedstocks, technologies and products. These as-
pects generate uncertainty, which is translated in the possibility of The theoretical concepts mentioned in the conceptual design
to have a feedback. However, if the conceptual design stage is ap- stage allow to build a tree-branching (i.e., a superstructure) of al-
propriately carried out, the uncertainty disappears and many pos- ternative processing routes to transform a feedstock into a set
sibilities of biorefinery can be assessed to generated one or two of products, namely, a superstructure that represents a decision-
optimal and integrated biorefinery configurations. making problem (Bertran et al., 2017; Rosengart, 2017). Boxes
(units) denote the process alternatives, while arrows are used to
2.1. Stage 1: Conceptual design connect the viable interactions between boxes. The units are used
to represent a process technology composed by one or more units.
The objective of the conceptual design consists of searching The superstructure components are arranged with respect to pro-
concepts or principles for theoretically structuring the problem. In cessing stages considered in a biorefinery, pretreatment, reaction
this point, the identified problem is analyzed and one or more and/or separation stages.
solutions are synthesized and evaluated with respect to imposed At this point, many biorefinery combinations are considered ac-
constraints. This stage demands high levels of knowledge and inno- cording to aspects such as, supply chain, alternative technologies at
vation, characterized by the dynamic evolution. A comprehensive each stage and feedstock distribution, among others. These consid-
and detailed description of each step for the proposed methodolog- erations must be taken into account for the formulation or solution
ical approach is presented below. of the problem.

• Step 1: Define the general objective, feedstock(s) and set of • Step 5: Developing of the surrogate models
The suitable connection and interaction between collected data
The application of the proposed methodology to a study case in Step 3, allows the generation of mass and energy balances of
is supported in its general objective. The nature of the objective process units that constitute the system. The modeling of process
depends on the problem that is solved. Examples of general ob- units is carried out according to the problem nature.
jectives for which it is applicable this methodology are, the selec- Pretreatment, reaction and separation stages of superstructure
tion of a better configuration of a determined system or processing are modeled using a generic set of equations, through a sequence
routes for one or more products or to choose a feedstock between of process operations (i.e., mixing, milling, cooling/heating, main
a set of feedstocks. In this step, one or a set of feedstocks and reaction, filtration, second reaction, separation). The unit allows
a set of products are defined. Availability, price, logistics, location several inlets and outlets, focusing on the consideration of utili-
and chemical composition are decisive parameters when defining ties, reagents and/or solvents streams. A general scheme and de-
a feedstock. Market demand and feedstock potential allows identi- tailed scheme of inputs and outputs in a process unit are shown in
fying interesting products. Fig. 2A and B, respectively.
38 V. Aristizábal-Marulanda, C.A. Cardona Alzate and M. Martín / Computers and Chemical Engineering 126 (2019) 35–53

Fig. 2. (A) General process unit scheme. (B) Extended unit scheme applied to pretreatment, reaction and separation stages.

The seven process operations that can be considered in a pro- metric coefficients of all involved components as well as conver-
cess unit are described in detail below in order to understand the sion(s) of limiting reagent.
role that every one of them plays. Filter: this process operation represents the separation of a solid
Mixing: It is used to mix flows from different streams or to in- from liquid stream with a split parameter called fraction (repre-
clude components in the main flow such as, reagents and/or added sented by xLiq ).
components (i.e., solvents). If a new component wants to be in- By-products reactor: in this process operation, a secondary
cluded, its flow should be determined based on the mass flow of chemical reaction such as, detoxification (described by a stoichio-
a reference component in the main stream (e.g., kg added com- metric reaction) is carried out. It requires information about stoi-
ponent per kg feedstock) with a parameter called “proportion of chiometric coefficients of reagents and products, and conversion.
added component respect to feedstock” (represented by μ). Separator: this process operation represents downstream pro-
Milling: this operation represents the particle size reduction of cesses such as, distillation, evaporation, crystallization and spray
the feedstock. dry, among others. The separation processes can be coupled in se-
Heat exchanger: It represents the temperature conditioning in quential way (e.g., distillation columns train or crystallization fol-
the inlet streams to main reactor. lowed by spray dry). Recovery percentage of main component is
Main reactor: in this process operation, a chemical transforma- the most important data to model in this process operation. From
tion is carried out (e.g., alkaline hydrolysis, enzymatic hydrolysis these downstream processes two outlet streams are obtained, the
and fermentation, among others). The model requires data about first one includes the interest component and the second one con-
yields or stoichiometric reaction(s) with the respective stoichio- tains the remaining components or wastewater.
V. Aristizábal-Marulanda, C.A. Cardona Alzate and M. Martín / Computers and Chemical Engineering 126 (2019) 35–53 39

Simultaneously to the generation of optimal biorefinery config- 10.13, 14.18 and 1.27% (w/w dry basis), respectively. The moisture
uration, energy requirements are quantified in terms of electricity, content is 8.7% (w/w) (García et al., 2017a).
low pressure steam (LPS), medium pressure steam (MPS) and high
pressure steam (HPS). Once these are quantified, a cogeneration 3.1. Stage 1: Conceptual design
system is coupled to the biorefinery in order to generate a quantity
equal or greater than energy requirements. The addition of cogen- • Step 1: Define the general objective, feedstock(s) and set of
eration system to the optimal biorefinery configuration promotes products
to the generation of optimal and integrated biorefinery configura-
tion. The study case has as goal to determine the most convenient
In addition, the objective function should be defined. The ob- products from an economic perspective in a CCS biorefinery con-
jective function evaluates the system performance based on one or sidering different technologies, feedstock distribution and sugars
more criteria (technical, economic, environmental, among others). use. CCS biorefinery considers one feedstock (coffee cut-stems) and
The selection of the criteria depends on the optimization objec- seven products of different nature, biofuels (ethanol and butanol),
tives (i.e., maximize yield, minimize cost, minimize environmen- biochemicals (xylitol, furfural and lactic acid), biomaterials (poly-
tal impact, etc.). In the case of the model of the case study, see hydroxybutyrate - PHB) and energy (electricity). These products
Appendix A, the objective function considers the annual earnings are selected in order to combine product-driven biorefinery and
from sales of products and annual expenses related to operat- energy-driven biorefinery, giving versatility to proposed biorefinery
ing costs (utilities, reagents and solvents), wastewater management and secure self-supply of required energy (IEA Bioenergy Task42,
cost and capital investment. 2014).
• Step 6: Solving the optimization problem • Step 2: Apply know-how experience, hierarchical decomposi-
tion and sequencing concepts
Due to the fact that the superstructure is a decision-making
problem, it imitates the approach of a Mixed-Integer Nonlinear CCS biorefinery is composed by three processing stages, pre-
Program (MINLP). Although in some situations, it can be reduced treatment, reaction and separation. Additionally, a cogeneration
to a Mixed-Integer Linear Program (MILP or MIP) using linear mod- system (it is explained in Step 5), which supplies the biorefinery
els and/or approximations. As result of the solution of optimization energy requirements, (i.e., electricity and steam) and a wastewater
problem, an optimal topology for biorefinery configuration is gen- treatment system are considered.
erated. The mathematical solution for the superstructure problem Two types of particle size reduction technologies are taken into
is linked to the problem nature. account. The first one has the function to condition the CCS that
enter to the pretreatment stage and the second one aims to condi-
tion the CCS that go to cogeneration system. These units are syn-
3. Study case: coffee cut-stems (CCS) biorefinery
ergistically connected in order to make the correct feedstock dis-
tribution that allows obtaining at least one product adapted with
Each step contemplated in the theoretical description of pro-
a cogeneration system.
posed methodology is applied to the study case of a multiprod-
The biorefinery considers three platforms (glucose, xylose and
uct biorefinery from CCS, as is indicated below. Previous work
syngas), which have the function to articulate the pretreatment
from the group (Triana et al., 2011) shows that CCS are a promis-
stage with reaction and separation stages. Xylose flow can be used
ing raw material for the production of bioproducts in tropical cli-
for the production of xylitol and/or furfural. Meanwhile, glucose
mates because of the wide availability and high yield per hectare.
flow can be distributed for the production of ethanol, PHB, butanol
When there is renovation of coffee plantations by "zoqueo" a
and/or lactic acid. An option is to mix xylose and glucose flows to
high availability of wood is presented, which is called Coffee Cut-
produce ethanol. All these alternatives are possible solutions of the
Stems (CCS) and normally are used as potential energy source
proposed biorefinery. The syngas platform is only used to produce
(Valencia, 20 0 0). The renovation process is carried out in order
heat and power.
to maintain a good and constant production of coffee and in-
In the next step, the methodological details that are used in the
volves cutting the main coffee stem at a height between 20 and
formulation and building of the CCS biorefinery are indicated.
30 cm above the ground (García Velásquez, 2016). CCS biorefin-
ery is located in Department of Caldas (central zone of Colom- • Step 3: Collect experimental results, data from literature and
bia) because this zone is one of the largest coffee producers in databases
Colombia. To determine the CCS availability in Caldas the follow-
ing assumptions are considered: (i) the coffee tree produces as The pretreatment stage considers six different technologies:
fruit the cherry coffee and it is used as feedstock in coffee process- particle size reduction, dilute-acid hydrolysis, alkaline hydrolysis,
ing. (ii) The “pergamino” coffee is the product obtained at the end organosolv, liquid hot water (LHW) and ammonia fiber explosion
of the coffee processing. (iii) The Centro Nacional de Investigación (AFEX). Table 1 indicates the operating conditions for each tech-
de Café (CENICAFÉ) determined that 4.915 kg of cherry coffee are nology used in this stage.
required to produce 1 kg of "pergamino" coffee (Montilla et al., In the reaction stage, furfural, ethanol and butanol can be pro-
2008). (iv) According to CENICAFÉ, 0.6 kg of CCS are obtained per duced through two process alternatives each. In the furfural pro-
kg of processed cherry coffee and (v) every 5 years, CCS is har- duction, dehydration reaction using HCl and H2 SO4 as different
vested (Rodríguez and Franco, 2010). catalysts are considered. Fermentation processes using Zymomonas
In this sense, Caldas had a coffee production of 62,869 ton mobilis and Saccharomyces cerevisiae as microorganisms are taken
in 2014 (Agronet, 2017), corresponding to 309,0 0 0 ton of coffee into account for ethanol production. In the butanol production,
cherry. These values allow determining the productivity of CCS fermentation processes using Clostridium beijerinckii and Clostrid-
per year, which is 37,080 ton, thus, a processing plant capacity of ium acetobutylicum as microorganisms are considered. The produc-
1.176 kg s-1 of feedstock. The CCS biorefinery considers an annual tion of xylitol, PHB and lactic acid only considers one technology.
productivity of 8016 h corresponding to 24 h a day and 334 days Fermentation processes using Candida guilliermondii, Bacillus mega-
a year. CCS present a chemical composition of cellulose, hemicel- terium and Escherichia coli as microorganisms are considered for
lulose, lignin, extractives and ash corresponding to 40.39, 34.01, xylitol, PHB and lactic acid obtaining, respectively. To reduce the
40 V. Aristizábal-Marulanda, C.A. Cardona Alzate and M. Martín / Computers and Chemical Engineering 126 (2019) 35–53

Table 1
Information about process units of pretreatment stage considered in CCS-superstructure.

Technology Operating conditions References

Particle size reduction I 0.4 mm, content moisture <10% and specific energy (kWh/kg) 0.03 (García Velásquez, 2016;
Tumuluru et al., 2014)
Particle size reduction 10 mm, content moisture <10% and Specific energy (kWh/kg) 0.00391 (Di Fulvio et al., 2015; García
II Velásquez, 2016; Vidal and
Martín, 2015)
Dilute-acid hydrolysis Dilute-acid hydrolysis (Aristizábal Marulanda, 2015;
and detoxification García Velásquez, 2016)
2% (v/v) H2 SO4 , 110 °C, 5 h, 1:10 (Solid to liquid ratio)
60 °C, 20 min, pHfinal 6.5
Glucose yield: 0.0574 g glucose/g substrate
Xylose yield: 0.2208 g xylose/g substrate
C a(OH )2 + H2 SO4 → C aSO4 + 2H2 O
CW: 120.77 kg/kg CCS, LPS: 0.43 kg/kg CCS
Alkaline hydrolysis and Alkaline hydrolysis (Oka et al., 2013)
KOH 1 N, 150 °C, 1 h, 1:25 (Solid to liquid ratio)
Glucose yield: 0.0649 g glucose/g substrate
Xylose yield: 0.2667 g xylose/g substrate
30 °C, 10min
3KOH + H3 P O4 → K3 P O4 + 3H2 O
CW: 100.44 kg/kg CCS, LPS: 0.26 kg/kg CCS, MPS: 0.48 kg/kg CCS
Organosolv 50%(v/v) ethanol, 180 °C, 4 MPa, 1 h, 1:10 (Solid to liquid ratio) (Pan et al., 2006)
1.25%(w/w dried wood) H2 SO4
Glucose yield: 0.0055 g glucose/g substrate
Xylose yield: 0.0937 g xylose/g substrate
HPS: 3.25 kg/kg CCS
Liquid hot water 5%(w/v) substrate in water, 180 °C, 4 MPa, 20 min, 500rpm (Yu et al., 2010)
Glucose yield: 0.02 g glucose/g substrate
Xylose yield: 0.825 g xylose/g substrate
HPS: 3.23 kg/kg CCS
Ammonia fiber 180 °C, 700 psi, 4.82 MPa, 30 min, (Balan et al., 2009)
2:1 (ammonia to biomass ratio), 233% moisture (dry biomass basis)
Glucose yield: 0,0 0 019 g glucose/g substrate
Xylose yield: 0,0 0 021 g xylose/g substrate
HPS: 2.10 kg/kg CCS

CW, cooling water.

boundaries of this work, the solids that did not react in the en- considered before pretreatment stage; due to the cogeneration sys-
zymatic hydrolysis unit are not used as an option for electricity tem requires a feedstock size different to the pretreatments (see
generation due to the high levels of energy consumed in the dry- Table 1). In the biorefinery superstructure, each process alternative
ing process in order to remove its moisture content. Table 2 indi- corresponds to one process unit.
cates the operating conditions for process units used in the reac-
tion stage. • Step 5: Developing of the surrogate model
In the separation stage, furfural, ethanol and butanol are pu-
Data reported in Tables 1, 2 and 3 are the base to generate mass
rified by distillation, xylitol through crystallization, PHB by spray
and energy balances for each process unit. The calculation of the
dryer and lactic acid using salts precipitation. Table 3 shows the
mass balance for unit of dehydration using HCl as catalyst for the
description of process units considered in the separation stage in-
furfural production is presented as an example in the Supplemen-
cluding the wastewater treatment unit. Aspen Plus v9.0 is used as
tary material. In this process unit, two types of chemical reactions
tool to generate the mass and energy balances. The information
are carried out: dehydration reaction (main) and neutralization re-
consigned in Tables 1, 2 and 3 is fed to the simulator. Each process
action (secondary). The calculation of mass balances for each pro-
unit is simulated individually.
cess unit is made in a similar way. The optimization problem of
3.2. Stage 2: Optimization superstructure is raised as indicates the generic model shown in
the Appendix A.
• Step 4: Propose a superstructure The proposed cogeneration system aims to supply the energy
requirements of optimal biorefinery configuration, considering the
For the CCS biorefinery case of study, the formulated super- generation of electrical and thermal energy. The technology used in
structure is shown in Fig. 3, which includes 27 process units dis- cogeneration system is known as Biomass Integrated Gasification
tributed in three stages: pretreatment, reaction and separation, and Combined Cycle (BIGCC) (Balat et al., 2009; Rincón et al., 2014),
a cogeneration system. Two types of particle size reduction are which is composed mainly by three stages, gasification, gas tur-
V. Aristizábal-Marulanda, C.A. Cardona Alzate and M. Martín / Computers and Chemical Engineering 126 (2019) 35–53 41

Table 2
Information about process units of reaction stage considered in CCS-superstructure.

Technology Operating conditions References

Enzymatic hydrolysis 50 °C, 48 h, 1:10 (Solid to liquid ratio), Celluclast 1.5 L and Viscozyme L in buffer (Aristizábal Maru-
solution of Sodium Citrate 0.05 M (pH 4.8–5.0) landa, 2015)
Glucose yield: 0.1198 g glucose/g substrate
Xylose yield: 0.0133 g xylose/g substrate
LPS: 11.58 kg/kg xylose and glucose
Fermentation Candida guilliermondii 30 °C, 200 rpm, 48 h (Hernández-Pérez et al.,
Xylitol yield: 0.47 g xylitol/g sugars
79.2% xylose consumption
3C5 H10 O5 → 5C2 H6 O + 5C O2
CW:1702.65 kg/kg xylitol
Dehydration with HCl 180 °C, 11 bar (Martín and
Grossmann, 2016)
CrCl3 catalyst is a molar ratio of 0.06 with the sugars, NaCl added to saturate the
water (35 g of NaCl per 100 g of H2 O), HCl is 0.12 mol/L and butanol is twice the
aqueous phase
Conversion glucose to HMF 67.5%
Conversion xylose to furfural to 56%
Selectivity to glucose 0.85
Selectivity to xylose 0.8
C6 H12 O6 → C6 H6 O3 + 3H2 O
C5 H10 O5 → C5 H4 O2 + 3H2 O
80 °C, 10min
NaOH + HCl → NaCl + H2 O
HPS: 93.99 kg/kg furfural and HMF
Dehydration with H2 SO4 NaCl corresponds to that needed to saturate the water (35 g of NaCl per 100 g of H2 O) (Martín and Grossmann,
and butanol 2016; Moncada et al.,
is twice the aqueous phase
150 °C, 1 bar, 4 h, 8 g/L H2 SO4 as catalyst
Furfural yield: 0.91 g furfural/g xylose
C5 H10 O5 → C5 H4 O2 + 3H2 O
60 °C, 20min
C a (OH )2 + H2 SO4 → C aSO4 + 2H2 O
MPS: 25.23 kg/kg furfural
Fermentation Zymomonas mobilis 33 °C, 80 h (Quintero Suárez, 2011)
(Mariscal Moreno, 2011)
Conversion reducing sugars to ethanol 74.43% (the same for both)
C6 H12 O6 → 2C2 H6 O + 2C O2
3C5 H10 O5 → 5C2 H6 O + 5C O2
In this process unit, glucose and xylose are assumed as unique reducing sugars.
Nutrients (g/100 kg sugars)
Na3 PO4 .12H2 O: 101.2
Urea: 212.2
FeCl3 .H2 O: 1.1
MgSO4 .7H2 O: 1.1
CW: 1521.47 kg/ethanol

Fermentation Saccharomyces cerevisiae 32 °C, 72 h, 170rpm (Moncada et al., 2016b)

Conversion glucose to ethanol 91.3%
C6 H12 O6 → 2C2 H6 O + 2C O2
(Mariscal Moreno, 2011)
Nutrients (g/100 kg sugars)
Na3 PO4 .12H2 O: 101.2
Urea: 212.2
FeCl3 .H2 O: 1.1
MgSO4 .7H2 O: 1.1
CW: 1994.86 kg/kg ethanol
Fermentation Bacillus megaterium 33 °C, 250 rpm, 42 h (Naranjo Vasco, 2014)
Air 12 L/min
0.4 g PHB/g glucose consumed
57.5% glucose consumed
C6 H12 O6 + 0.99O2 + 0.38N H4 N O3 → 4C H1.74 O0.96 N0.19 + 3.28H2 O + 2C O2
CW: 3536.17 kg/kg PHB
(continued on next page)
42 V. Aristizábal-Marulanda, C.A. Cardona Alzate and M. Martín / Computers and Chemical Engineering 126 (2019) 35–53

Table 2 (continued)

Fermentation Clostridium beijerinckii 34 °C, 72 h (Malmierca et al., 2017;

Qureshi and Blaschek,
100% sugars are transformed mainly into butanol and acetone
100C6 H12 O6 →
82.68C4 H10 O + 23.15C3 H6 O + 2.03C2 H6 O + 195.7C O2 + 10.44H2 + 100.6H2 O
Heated up to 70 °C and cool down to 34 °C before recycling the retentate to the
fermenter. Butanol yield in permeate side is 98.67%
CW: 3391.93 kg/kg AB
LPS: 11.68 kg/kg AB
Refrigerant: 3848.94 kg/kg AB
Fermentation Clostridium 37 °C, 72 h (Ramirez, 2018)
Conversion glucose to butanol 45.27%
Conversion glucose to ethanol 3.94%
Conversion glucose to acetone 18.45%
C6 H12 O6 + H2 O → C3 H6 O + 3C O2 + 4H2
C6 H12 O6 → C4 H10 O + 2C O2 + H2 O
C6 H12 O6 → 2C2 H6 O + 2C O2
CW: 4866.37 kg/kg ABE
LPS: 18.50 kg/kg ABE
Refrigerant: 6064.77 kg/kg ABE
Fermentation Escherichia coli 37 °C, 48 h (Ramirez, 2018)
Conversion xylose to lactic acid 72.69%
Total conversion of glucose
C6 H12 O6 → 2C3 H6 O3
3C5 H10 O5 → 5C3 H6 O3
CW: 680.91 kg/kg LA

CW, cooling water.

Fig. 3. Graphical scheme of formulated superstructure for CCS biorefinery.

bine and Heat Steam Recovery Generator (HRSG). Fig. 4 indicates (H2 ), methane (CH4 ) and small amounts of ashes. In pyrolysis zone,
the detailed flowsheet of cogeneration system. The operating con- biomass undergoes into the devolatilization, where it is decom-
ditions, specifications and reactions of each stage are shown in posed into carbon (assumed as char), CO, CO2 , H2, H2 O, light hy-
Table 4. drocarbons (assumed as CH4 ) and heavy hydrocarbons (assumed
The gasification stage contains three reaction zones such as, py- as benzene - C6 H6 ). Then, in the oxidation zone all the compo-
rolysis, oxidation and reduction (García et al., 2017a; García et al., nents from pyrolysis react with oxygen to produce CO2 , CO and
2017b; Sharma, 2011). Syngas is the main product from biomass H2 O. Finally, the produced char in pyrolysis zone passes to the re-
gasification, which it is a mixture of gaseous compounds as car- duction zone, where char gasification takes place to produce CO,
bon monoxide (CO), carbon dioxide (CO2 ), water (H2 O), hydrogen H2 and CH4 . A cleaning system composed by a cyclone and a filter
V. Aristizábal-Marulanda, C.A. Cardona Alzate and M. Martín / Computers and Chemical Engineering 126 (2019) 35–53 43

Table 3
Information about process units of separation stage considered in CCS-superstructure.

Technology Operating conditions References

Xylitol crystallization Concentration of xylitol in broth 16 g/L (De Faveri et al., 2002;
Garcia et al., 2013;
Hernández et al., 2016)
50 °C and 0.16 bar
Concentration of xylitol until 600 g/L
−5 °C, commercial xylitol added 1 g/L to favor the nucleation of xylitol crystals. Crystallization
yield 0.56 and pure degree 1
Recovery 56% of xylitol
LPS: 135.64 kg/kg xylitol
Refrigerant: 99.36 kg/kg xylitol
Furfural purification I Separator (aqueous and organic phase) and two stripping columns. Organic phase is distillated. (Martín and
Grossmann, 2016; De
Rodrigues and
Guirardello, 2012)
Respect to inlet stream to separator, the aqueous phase contains, water ≤94%, butanol ≤1%,
50% of remaining sugars and 100% of NaCl.
Column 1
Reflux ratio at 1.95, 29 stages, feed stage 23, 0.5 bar
Column 2
Reflux ratio at 0.55, 62 stages, feed stage 27, 0.5 bar
Recovery 99.9% of 1-butanol
Recovery 87% of furfural + HMF (94% w/w of purity)
LPS:353.61 kg/kg furfural and HMF
MPS: 561.65 kg/kg furfural and HMF
Furfural purification II Separator (aqueous and organic phase) and two stripping columns. Organic phase is distillated. (Martín and
Grossmann, 2016; De
Rodrigues and
Guirardello, 2012)
Respect to inlet stream to separator, the aqueous phase contains, water ≤94%, butanol ≤1%,
50% of remaining sugars and 100% of NaCl.
Column 1
Reflux ratio at 2, 55 stages, feed stage 22, 0.5 bar
Column 2
Reflux ratio at 0.45, 72 stages, feed stage 39, 0.5 bar
Recovery 99.9% of 1-butanol
Recovery 85% of furfural (96% w/w of purity)
LPS: 239.47 kg/kg furfural
MPS: 380.57 kg/kg furfural
Ethanol purification I and II The concentration of ethanol in water is 5%–6% using Z. mobilis. (García et al., 2017;
Moncada et al., 2016)
The concentration of ethanol in the water is 6%–8% using S. cerevisiae
Column 1 and 2
Reflux ratio at 1.5, 12 stages, feed stage 6, 0.5 bar.
In column 1, the ethanol is concentrated up to 45%–50% w/w and in the second column, the
ethanol is concentrated until the azeotropic point (96% w/w).
Molecular sieves
Recovery 99.8% of ethanol (99.6% w/w of purity).
Utilities I
CW: 600.3 kg/kg ethanol
LPS: 7.27 kg/kg ethanol
Utilities II
CW: 570.41 kg/kg ethanol
LPS: 7.06 kg/kg ethanol
PHB purification The fermentation broth with a PHB and cells concentration of 4.2 and 7.1 g/L, respectively. (Naranjo Vasco, 2014;
Posada et al., 2011)
Digestion and filtration
1:2 NaOCl/cell mass ratio. Enzyme concentration: 2 wt%
H2 O2 -water washing concentration: 1.2 v/v%.
H2 O2 -water washing: mixture ratio 1:1.
90 °C, 0.5 bar and 20.2% w/w PHB purity
Spray dryer
Recovery 99.9% of PHB (99.6% w/w of purity).
CW: 8.05 kg/kg PHB
LPS: 556.88 kg/kg PHB
(continued on next page)
44 V. Aristizábal-Marulanda, C.A. Cardona Alzate and M. Martín / Computers and Chemical Engineering 126 (2019) 35–53

Table 3 (continued)

Acetone and Butanol purification Heat exchanger 1 (Malmierca et al., 2017)

50 °C and 1 bar
Column 1
Reflux ratio at 1.5, 6 stages, feed stage 4, 1 bar
Recovery 95.7% of water in the bottoms (99% w/w of purity).
Heat exchanger 2
62 °C and 0.5 bar
Column 2
Reflux ratio at 2.2, 15 stages, feed stage 5, 0.5 bar
Recovery 95.6% of acetone in the top (99% w/w of purity).
Heat exchanger 3
91.78 °C and 1 bar
Column 3
Reflux ratio at 2, 26 stages, feed stage 22, 1 bar
Recovery 70.8% of butanol in the bottoms (99% w/w of purity).
LPS: 44.09 kg/kg AB
Acetone, Butanol and Ethanol purification The fermentation broth with butanol concentration of 8.65 g/L (Malmierca et al., 2017;
Xie et al., 2015)
Heat exchanger 1
50 °C and 1 bar
Column 1
Reflux ratio at 1.5, 6 stages, feed stage 4, 1 bar
Recovery 95.5% of water in the bottoms (99% w/w of purity).
Heat exchanger 2
62 °C and 0.5 bar
Column 2
Reflux ratio at 2.2, 15 stages, feed stage 5, 0.5 bar
Recovery 96.9% of acetone in the top (99% w/w of purity).
Column 3
Reflux ratio at 1.5, 12 stages, feed stage 6, 0.5 bar
Recovery 90% of ethanol in the top (96% w/w of purity).
Heat exchanger 3
91.78 °C and 1 bar
Column 4
Reflux ratio at 2, 26 stages, feed stage 22, 1 bar
Recovery 87.9% of butanol in the bottoms (99% w/w of purity).
LPS: 51.31 kg/kg ABE
Lactic Acid purification Precipitation reaction (Garcia et al., 2013;
Komesu et al., 2017;
Nakano et al., 2012)
Ca(OH)2 added in aqueous solution (Lime: water ratio 15:85 w/w), 95 °C
2C3 H6 O3 + C a (OH )2 → C a (C3 H5 O3 )2 + 2H2 O
Solid content less than 1% w/w in the liquid fraction
Acidification reaction
H2 SO4 added in aqueous solution (H2 SO4 :water ratio 15:85 w/w), 20 °C
C a(C3 H5 O3 )2 + H2 SO4 → 2C3 H6 O3 + C aSO4
40 °C and 0.04 bar
Recovery 98.8% of lactic acid (80% w/w of purity – food grade).
CW: 36.11 kg/kg LA
Refrigerant: 28.15 kg/kg LA

CW, cooling water.

is used to separate the remaining char, ashes and impurities from hot gases obtained are brought to three steam generators to pro-
produced syngas. duce steam of different qualities, where each one is composed by
The gas turbine uses the syngas to produce power and flue three heat exchangers (HX) with defined functions. The first HX
gases. The intercooled, reheat and regenerative (ICRHR) cycle is increases the water temperature until its saturation point, the sec-
composed of two compressors, one combustor and two turbines ond one provides heat to convert the saturated liquid into satu-
(Razak, 2007). This type of turbines present high thermal efficiency rated steam and the third one increases the steam temperature
and the ICRHR cycle decreases the heat input requirements to the over its saturation point. The mentioned description corresponds
system. The turbine uses atmospheric air at 14.85 °C and 1 bar. to a HRSG.
Both compressors operate at the same pressure ratio in order to
• Step 6: Solving the optimization problem
minimize their work inputs. In the same way, both turbines op-
erate at the same pressure ratio to maximize their work outputs The generic model is formulated as a nonlinear program-
(Razak, 2007). The syngas combustion is carried out in a rich-air ing problem, namely MINLP, which follows the structure pre-
atmosphere to achieve a complete combustion. sented in Appendix B. This uses DICOPT as solver, which is based
The outlet gases from the gas turbine are mixed and com- on three key ideas as, Outer Approximation, Equality Relaxation
busted with more syngas in order to increase the temperature. The and Augmented Penalty (“GAMS. Solver manuals,” 2018). GAMS
V. Aristizábal-Marulanda, C.A. Cardona Alzate and M. Martín / Computers and Chemical Engineering 126 (2019) 35–53 45

Fig. 4. Flowsheet of cogeneration system.

(General Algebraic Modeling System) is used as computer tool 4. Discussion of results

to solve the optimization problem, which combines mathematical
language with conventional programming concepts. In the CCS biorefinery, the model is formulated as a MINLP con-
taining 3243 equations, 7191 continuous variables and 27 binary
decisions. The solutions and results are presented in the next sub-
3.3. Economic aspects sections, with respective description and discussion.

The investment costs are calculated using the software Aspen 4.1. Alternative solutions
Process Economic Analyzer v9 (Aspen Technologies, Inc., USA). The
reference year for the estimation of investment costs is 2015. This When the optimization problem is solved, the model selects
analysis is estimated in US dollars for a 10-year period at an an- ethanol as the main product, where the optimal and integrated
nual interest rate of 20% and working capital of 5% (typical for the biorefinery configuration is composed by the following process
USA economy). Table 5 summarizes the economic data for raw ma- units, size reduction I and II, dilute acid hydrolysis and detoxi-
terial, reagents, solvents, products and utilities. In the supplemen- fication, enzymatic hydrolysis, fermentation using Z. mobilis and
tary material file, the design specifications of all equipments used ethanol purification I. Fig. 5 indicates the flowsheet for the base
in this work are presented. case. In the distribution of the feedstock input flow toward pro-
Generally in process industries, the investment cost for equip- cess units of size reduction I and II, the model determines that the
ment increases as a function of throughput according to the six- mass flows corresponds to 46.68% and 53.31% of CCS total flow,
tenths-factor rule, as described below in Eq. (1) (Peterset al., 2004). respectively. This distribution allows the most suitable interaction
In this work, the investment cost used as reference is calculated between ethanol process and cogeneration system, which the sup-
with the CCS flow fed, namely, 1.176 kg s−1 . Additionally, the sur- plies energy requirements of the biorefinery. In the distribution of
rogate models have been developed for the same flow. However, the sugar platforms (C5 and C6 sugars), the model chooses that
due to the fact that the model that has a generic perspective, the the total xylose and glucose flows produced in pretreatment stage
six-tenths-factor rule is used in order to ensure its applicability to and enzymatic hydrolysis are used for the ethanolic fermentation,
any feedstock flow. it means that, split of mass fraction, S3 and S4 are equal to 1 (see,
 0 , 6 Appendix A). Ethanol biorefinery requires 0.068 MJ s−1 of electri-
Capacity 1 cal energy and 1.184 kg of LPS s−1 . The cogeneration system cov-
Cost Equip. 1 = Cost Equip. 2 ∗ (1)
Capacity 2 ers these demands generating electricity, LPS, MPS and HPS with
flows of, 6.761 MJ s−1 , 1.184, 1.184 and 0.593 kilograms s−1 , respec-
The nonlinear cost-size equations are linearized formulating a tively. The difference between produced and required energy (elec-
piecewise linear approximation (Floudas, 1995). Table 6 indicates trical and thermal) is considered as additional biorefinery products,
the constants of linearized cost functions that represent the in- namely, electricity, MPS and HPS. Additionally, CCS biorefinery con-
vestment cost for each process unit considered in CCS biorefinery. figuration generates a wastewater flow of 0.009 m3 s−1 .
For pretreatment stage, the investment cost equation of each pro- For the base case, the yields for ethanol, LPS, MPS, HPS
cess unit is given as a function of the feedstock flow. Meanwhile, and electricity are 166.52 L, 1888.94 kg, 1888.94 kg, 946.06 kg and
in reaction and separation stages, the investment cost equation is 8308.78 MJ per ton of CCS, respectively. The ethanol yield for the
in function of the main product(s) flow. base case of CCS biorefinery is comparable with values reported
46 V. Aristizábal-Marulanda, C.A. Cardona Alzate and M. Martín / Computers and Chemical Engineering 126 (2019) 35–53

Table 4
Information about process units of cogeneration system.

Technology Operating conditions Reference

Gasification CHAR: C and TAR: C6 H6 (benzene) (García Velásquez, 2016; Sharma, 2011)
600 °C and 1 bar
Fractional CHAR yield
Cellulose: 0.05
Hemicellulose: 0.10
Lignin: 0.55
Empirical mass ratios
7730.317 5019898
YCO/C O2 = e−1.8447896+ T + T 2
YH2 O/C O2 = 1
YC H4 /C O2 = 5 ∗ 10−16 T 5.6
Pyrolytic decomposition
Cel l ul ose : C6 H10 O5 → 0.675CHAR + 1.754CO + 0, 7305C O2 +
0.6498H2 + 1.785H2 O + 0.7636C H4 + 0, 3461T AR
Hemicellulose: C5 H8 O4
→ 1.1CHAR + 1.403CO + 0.5844C O2 + 0.6999H2 + 1.428H2 O +
0.6109C H4 + 0.2169T AR
Lignin : C7.3 H13.9 O1.3 → 5.61CHAR + 0.4561CO + 0.1899C O2 +
5.666H2 + 0.464H2 O + 0.1985C H4 + 0.1409T AR
Air composition (mass fraction): gasifying agent
CO2 : 0.0 0 05, O2 : 0.2314, N2 : 0.7553 and Ar: 0.0128
1.0 kg wood + 1.5 kg air → 2.5 kg syngas
>800 °C and 1 bar
H2 + 0.5O2 → H2 O
CO + 0.5O2 → C O2
C H4 + 1.5O2 → CO + 2H2 O
T AR + 4.5O2 → 6CO + 3H2 O
CHAR + 0.5O2 → CO
<800 °C and 1 bar
C + C O2 ↔ 2CO
C + H2 O ↔ CO + H2
C + 2H2 ↔ C H4
C H4 + H2 O ↔ CO + 3H2
Syngas composition (mole fraction)
CO: 0.2894; CO2 : 0.0537, H2 : 0.2625, H2 O: 0.0368, CH4 :
0.0029 and N2 : 0.3544
Gas turbine (Brayton cycle) Compression (Razak, 2007; Vidal and Martín, 2015)
Pressure ratio 10, isentropic efficiency 85% and air–fuel ratio
6.4 (at 14.85 °C and 1 bar)
875 °C and 97.98 bar
Pressure ratio 0.2 and isentropic efficiency 90%
2.47 kg syngas/kg CCS and 1.2131 kW/kg syngas
Heat recovery steam generator (HRSG) Furnace (Varbanov et al., 2014; Zheng and Furimsky, 2003)
900 °C and 3.6 bar
Low pressure steam (LPS): 145 °C and 3 bar
Medium pressure steam (MPS): 245 °C and 30 bar
High pressure steam (HPS): 285 °C and 60 bar
0.765kg LPS/kg syngas, 0.765kg MPS/kg syngas and
0.383kg HPS/kg syngas

Fig. 5. Flowsheet of the optimal and integrated biorefinery (base case).

V. Aristizábal-Marulanda, C.A. Cardona Alzate and M. Martín / Computers and Chemical Engineering 126 (2019) 35–53 47

Table 5
Costs and selling prices of compounds and utilities.

Compounds Unit Price References

Coffee cut-stems USD/kg 0.024 Calculated

Cooling water USD/kg 0.00125 (García Velásquez, 2016)
H2 SO4 USD/kg 0.094 (ICIS Chemical Pricing, 2010)
Ca(OH)2 USD/kg 0.050 (ICIS Chemical Pricing, 2010)
NaOH USD/kg 0.350 (ICIS Chemical Pricing, 2010)
KOH USD/kg 0.750 (Alibaba, 2018)
Ethanol USD/kg 0.887 (Federación Nacional de Biocombustibles de Colombia, 2018)
NH3 USD/kg 0.386 (ICIS Chemical Pricing, 2010)
Enzyme 1 (Cellulase) USD/kg 0.700 (Moncada et al., 2016b)
CrCl3 USD/kg 2.600 (Alibaba, 2018)
HCl USD/kg 0.200 (Martín and Grossmann, 2016)
NaCl USD/kg 0.100 (Martín and Grossmann, 2016)
Butanol USD/kg 1.230 (Martín and Grossmann, 2016)
HMF USD/kg 2.0 0 0 (Alibaba, 2018)
Furfural USD/kg 2.0 0 0 (Martín and Grossmann, 2016)
NH4 NO3 USD/kg 0.172 (ICIS Chemical Pricing, 2010)
H3 PO4 USD/kg 0.450 (ICIS Chemical Pricing, 2010)
K3 PO4 USD/kg 1.100 (Alibaba, 2018)
Xylitol USD/kg 3.500 (Alibaba, 2018)
Toluene USD/kg 0.850 (ICIS Chemical Pricing, 2010)
PHB USD/kg 3.750 (Naranjo V, 2014)
Lactic acid USD/kg 1.650 (Alibaba, 2018)
NaOCl USD/kg 0.100 (Naranjo Vasco, 2014)
H2 O2 USD/kg 0.400 (Alibaba, 2018)
Enzyme 2 (Burkholderia sp. PTU9) N.R. N.R. N.R.
Acetone USD/kg 0.970 (ICIS Chemical Pricing, 2010)
Na3 PO4 .12H2 O USD/kg 0.410 (Alibaba, 2018)
Urea USD/kg 0.620 (Alibaba, 2018)
FeCl3 .H2 O USD/kg 0.705 (Alibaba, 2018)
MgSO4 .7H2 O USD/kg 0.087 (Alibaba, 2018)
Low pressure steam (LPS) USD/kg 0.00756 (García Velásquez, 2016)
Medium pressure steam (MPS) USD/kg 0.00818 (García Velásquez, 2016)
High pressure steam (HPS) USD/kg 0.00986 (García Velásquez, 2016)
Electricity USD/kWh 0.100 (García Velásquez, 2016)
Ethylene glycol 25% (v/v) USD/kg 0.351 Calculated
Wastewater treatment USD/m3 0.547 (del Villar, 2016)

N.R., nonreported.

in the literature for similar and different process configurations. ternative solution I. The process unit for particle size reduction
For example, Cotana et al. (2014) reported the ethanol produc- presents an electricity requirement of 0.016 MJ s−1 , which is cov-
tion from pine wood chip (softwood) with a process configuration ered by the cogeneration system. The difference between required
composed by pretreatment with steam explosion, enzymatic hy- and generated electricity is taken as main system product. In this
drolysis, fermentation and distillation, obtaining a yield of 134 L of configuration are obtained the following product yields, 1889.46 kg,
ethanol per ton of feedstock. Ko et al. (2012) analyzed hardwoods 1889.46 kg, 946.43 kg and 10,786.88 MJ per ton of CCS of LPS, MPS,
chips (i.e., Acacia confusa Merr. and Eucalyptus globulus) for ethanol HPS and electricity, respectively. García et al. (2017b) indicated for
production with process configuration composed by, steam explo- a stand-alone process with similar characteristics to the exposed
sion pretreatment and simultaneous saccharification and fermen- it in this alternative solution, a yield of 7090 MJ per ton of CCS,
tation (SSF), obtaining yields for ethanol of 89.63 L per ton of aca- which is close to the result obtained. The yield generated in a
cia and 82.64 L per ton of eucalypt. Zabed et al. (2017) indicated biorefinery configuration is lower than in a stand-alone process
potential ethanol yields are 280–285 L per ton for hardwoods. The due to the biorefinery implies the obtaining of additional products,
yields used in acid and enzymatic hydrolysis, and fermentation namely, the feedstock should be distributed for all process units.
correspond specifically to experiments that used CCS as feedstock Meanwhile, the stand-alone process has a total feedstock flow for
that are found in literature. Comparing the obtained and reported electricity generation. The trend indicates that the number of prod-
ethanol yields, is possible to infer that acid hydrolysis is the most ucts is inversely proportional to the electricity yield.
convenient pretreatment than steam explosion, and in the same
way, that the process configuration of separate hydrolysis and fer- 4.2. Economic results
mentation (SHF) is a better option than SSF for this process.
The alternative solution I contemplates electrical and thermal Table 7 shows the costs associated to each alternative solution
energy as main products. Split of mass fraction S3 and S4 are fixed of the CCS biorefinery. Costs of feedstock, reagents and solvents,
to zero in order to avoid ethanol. The distribution of xylose plat- utilities cost, investment cost, wastewater treatment cost and rev-
form can be carried out in S1 or S2 and the distribution of glu- enues contribute to compute the objective function of the mathe-
cose platform can be addressed to S5, S6 or S7 (see, Appendix A). matical model (see, Appendix A). The economic profit for both so-
However, the model indicates that the energy generation is the lutions is negative, namely, both biorefinery configurations are not
most profitable, making the transition to stand-alone process from cost-effective at the processing scale assessed. However, it is pos-
biorefinery configuration. This stand-alone process is composed by sible that the biorefineries have a positive performance at higher
particle size reduction II and cogeneration system, with a process- processing scales. For the base case, the investment cost represents
ing flow of 1.176 kg of CCS s−1 . Fig. 6 shows the flowsheet for al- the highest contribution to the total cost followed by utilities and
48 V. Aristizábal-Marulanda, C.A. Cardona Alzate and M. Martín / Computers and Chemical Engineering 126 (2019) 35–53

Fig. 6. Flowsheet of the optimal and integrated biorefinery (alternative solution I).

Table 6 Table 8
Cost functions (investment cost) for process units of CCS Production costs and economic allocation factors for CCS cases
biorefinery. (see (García-Velásquez et al., 2018) to know the calculation pro-
cedure of economic parameters).
Unit A B
Product Production cost Economic allocation factor
Pretreatment stage
I and II 2618.6 145177 Base case
III 30189 2∗ 106 Ethanol 2.87 USD L−1 0.24
IV 39751 2∗ 106 Electricity 0,115 USD MJ−1 0.70
V 35987 2∗ 106 MPS 0.034 USD kg−1 0.04
VI 25632 1∗ 106 HPS 0.040 USD kg−1 0.02
VII 17820 987973 Alternative solution I
Reaction stage Electricity 0.054 USD MJ−1 0.89
VIII 540644 809991 LPS 0.01461 USD kg−1 0.04
IX 219364 913290 MPS 0.01580 USD kg−1 0.05
X 1∗ 106 3∗ 106 HPS 0.01905 USD kg−1 0.03
XI 810483 2∗ 106
XII 403652 1∗ 106
XIII 1∗ 106 804407
XIV 1∗ 106 412848
XV 1∗ 106 907707 and electricity as products in base case is carried out with the
XVI 2∗ 106 883627 idea of selling them to nearby factories in order to make the most
XVII 450175 671445
of the cogeneration system and reduce the production costs. For
Separation stage the base case, the production cost of ethanol is 2.87 USD L−1 (see,
XVIII 347714 810689
XIX 5∗ 106 8∗ 106
Table 8) and according to Walker (2011), the ethanol production
XX 5∗ 106 1∗ 107 cost from hardwood (Salix) is around 0.57 - 0.84 USD L−1 . As it can
XXI 246301 730423 be seen, the difference between values is significant, however, the
XXII 946708 650157 generation of energy through cogeneration system using biomass
XXIII 4∗ 106 1∗ 106
as fuel cannot be ignored, because this consideration is a plus for
XXIV 7∗ 106 2∗ 106
XXV 4∗ 106 1∗ 106 the CCS biorefinery that makes the process not only more environ-
XXVI 1∗ 106 2∗ 106 mental friendly in comparison with processes that use fossil fuels
Cogeneration system for energy generation but also self-sufficient. If ethanol yield would
XXVII 95963 1∗ 107 be 285 L per ton CCS (potential yield for hardwoods (Zabed et al.,
2017)), the production cost would correspond to 2.37 USD L−1 ,
which represents a reduction of 17.5% over the production cost that
is obtained in base case.
Table 7
Costs associated to each solution of CCS cases (M-USD y−1 ). If the base case has a processing capacity of 360 ton of CCS h−1
(commercial scale) with a yield of 166.52 L of ethanol per ton of
Item Base case Alternative solution I
CCS, the approximate production cost for ethanol, electricity, MPS
Objective function (Economic margin) −23.855 −10.78 and HPS are 0.69 USD L−1 , 0.027 USD MJ−1 , 0.0081 USD kg−1 and
Revenues 7.698 11.56 0.0098 USD kg−1 , respectively. It means that the productions costs
Costs of feedstock, reagents and solvents 4.713 1.01
Utilities cost 11.631 10.90
are lower than sale prices, therefore, the biorefinery presents a
Investment cost 15.067 10.43 promising economic performance with a profit margin of 0.66% ap-
Wastewater treatment cost 0.142 0.00 proximately. In the case of alternative solution I, if the processing
Total costs 31.553 22.34 scale is of 10.6 ton of CCS h−1 (pilot - demonstration scale) the
M-USD, Million dollars. approximate production cost for electricity, LPS, MPS and HPS are
0.027 USD MJ−1 , 0.0074 USD kg−1 , 0.0080 USD kg−1 and 0.0097
USD kg−1 , respectively. From this processing scale, the biorefinery
raw materials cost. These costs are mainly affected by the invest- configuration presents a positive performance with a profit margin
ment cost of cogeneration system (10.149 M-USD y−1 ), the cooling of 1.71% approximately. For larger scales, biorefineries more com-
water cost (5.803 M-USD y−1 ) and enzyme cost (2.989 M-USD y−1 ), petitive are expected.
which represent 60% of total costs. The Biomass Integrated Gasifi- Economic allocation factor is defined as the percentage that
cation Combined Cycle (BIGCC) is used as technology for the co- assumes each product with respect to the total cost of biorefin-
generation system, which involves a large number of equipmentin- ery. For all alternative solutions, base case and I, electricity takes
fluencing directly in the investment cost of both alternative solu- the highest percentage due to the cogeneration system generates
tions. As it can be seen in Table 7, when the cogeneration system a considerable flow of electrical energy in comparison with other
works with a total flow of biomass, the investment cost is 10.43 M- products. In general terms, the economic allocation factor for elec-
USD y−1 . tricity is around 0.70 and 0.89, following ethanol with factor of
Table 8 presents the production costs and economic allocation 0.24 and finally thermal energy products with factors below 0.1
factors for products of each case. The consideration of MPS, HPS (see, Table 8).
V. Aristizábal-Marulanda, C.A. Cardona Alzate and M. Martín / Computers and Chemical Engineering 126 (2019) 35–53 49

5. Conclusions and future research works B. Equality constraints

• Feedstock composition

In this work, a general methodology is proposed for systematic

fi,F ligno = φi,F ligno. f
Eq. (A.9)
and optimal design of biorefineries. The main features consist of
the development of surrogate models and of define simulations or Flow per feedstock component
equations based on models as well as the aim of self-production 
of utilities for integrated exploitation of resources. The methodol-
fF = fi,F ligno Eq. (A.10)
ogy is generic, flexible, versatile and friendly with the user. The
results show that the ethanol is the desirable product due to the Feedstock total flow
high yields and reasonable utilities consumption.
• Mass balance for pretreatment, reaction and separation
As future research works, the authors propose an analysis of
processing scale and a social assessment, a study about the sup-

ply chain of Coffee Cut-Stems (CCS) in Colombia, an experimental IN
fi,k = fi,k (A.11)
assessment of pretreatments and enzymatic hydrolysis to CCS and k
an environmental assessment of biorefineries from CCS using Life
Cycle Assessment (LCA), among others. Inlet unit
fi,k = μi,k . f F (A.12)
Added components for main reaction 1
The authors express their acknowledgments to Departamento  
fi,k = αS1, F
i, ligno,k . f i, ligno
+ fRem Ad,1
+ fi,k (A.13)
Administrativo de Ciencia, Tecnología e Innovación (Colciencias)
call 727 of 2015. The authors express their acknowledgments to Main reaction 1
research group in Process engineering and sustainable products  
fRem = fi,F ligno − αS1, F
i, ligno,k . f i, ligno (A.14)
(PSEM3) of Departamento de Ingeniería Química y Textil of Uni-
versidad de Salamanca, Salamanca – Spain. The authors express Remaining feedstock flow
their acknowledgments to Convocatoria Nacional para el Apoyo a Liq1
fi,k = xLiq1 . fi,k
la Movilidad Internacional de la Universidad Nacional de Colombia
2017–2018. Filter 1
fi,Sol1 F
ligno,k = f Rem (A.16)
Supplementary materials
Solids flow 1
Supplementary material associated with this article can be MW Reag
Reag,1 Ad,1
found, in the online version, at doi:10.1016/j.compchemeng.2019. fi,k = fi,k . (A.17)
MW Ad,1
Reagents for secondary reaction
Appendix A. The model  γi,r,k MWi
fi,k Liq1
= fi,k + fi,k .θr,k .
Ad,1 Ad,1
. (A.18)
γi,r,k MW Ad,1
Secondary reaction
A. Objective function Liq2
fi,k = xLiq2 . fi,k
Z = SProd − C F − C Reag − CUtil − C ww − C Inv (A.1)
Filter 2
fi,k = 1 − xLiq2 . fi,k
SProd = P rkP . fi, k (A.2)
k i Solids flow 2
Product sales Ad,2
fi,k = δi,k . fi,k
C = Pr . f (A.3) Added components for main reaction 2
Feedstock cost MW Reag
fi,k Liq2
= fi,k . (A.22)
CC = P rkReag . fi,k
+ P rkAd . fi,k
(A.4) MWi
k i k i
Reagents for main reaction 2
Reagents and added components cost 
 γi,r,k MWi
fi,k Ad,2
= fi,k Liq2
+ fi,k + fi,k .θr,k .
Liq2 Liq2
. (A.23)
CUtil = βkUtil .PrUtil
k .f
(A.5) γi,r,k MW Liq2
k Util
Main reaction 2
Utilities cost
fi,k P1
= fi,k .σi,k (A.24)
C ww = P rkw . w
fi,k (A.6)
k i Product separation 1
Wastewater management cost w1,OUT P1 P1,Out
 fi,k = fi,k − fi,k (A.25)
C Inv = Invk (A.7)
Waste separation 1

Investment cost
fi,k = δi,k . fi,ligno,k
Invk = Ak . f + Bk (A.8) Added components for main reaction 3
Investment cost, linearized S2
fi,k = αS2, Sol1
i, ligno,k . f i, ligno,k
+ fRem Ad,3
+ fi,k (A.27)
50 V. Aristizábal-Marulanda, C.A. Cardona Alzate and M. Martín / Computers and Chemical Engineering 126 (2019) 35–53

Table A1
Information about the superstructure of CCS biorefinery.

Component Description Component Description

Process units XIX, y19 Furfural purification I

I, y1 Size reduction (0.4 mm) XX, y20 Furfural purification II
II, y2 Size reduction (5–10 mm) XXI, y21 Ethanol purification I
III, y3 Dilute-acid hydrolysis and detoxification XXII, y22 Ethanol purification II
IV, y4 Alkaline hydrolysis and detoxification XXIII, y23 PHB purification
V, y5 Organosolv XXIV, y24 AB purification
VI, y6 Liquid hot water XXV, y25 ABE purification
VII, y7 Ammonia fiber explosion XXVI, y26 Lactic acid purification
VIII, y8 Enzymatic hydrolysis XXVII, y27 Cogeneration system
IX, y9 C. guilliermondii fermentation Sources
X, y10 Dehydration with HCl Src1 Feedstock resource
XI, y11 Dehydration with H2 SO4 Src2 Sulfuric acid resource
XII, y12 Z. mobilis fermentation Src3 Water source
XIII y13 S. cerevisiae fermentation Src4 Calcium hydroxide source
XIV, y14 B. megaterium fermentation Src5 Sulfuric acid source
XV, y15 C. beijerinckii fermentation Src6 Potassium hydroxide source
XVI, y16 C. acetobutylicum fermentation Src7 Water source
XVII, y17 E. coli fermentation Src8 Phosphoric acid source
XVIII, y18 Xylitol purification Src9 Ethanol source
Src10 Water source Src32 Xylitol source
Src11 Sulfuric acid source Src33 Sodium hypochlorite source
Src12 Ethanol source Src34 Enzyme 2 ((Burkholderia sp. PTU9)) source
Src13 Water source Src35 Hydrogen peroxide source
Src14 Water source Src36 Water source
Src15 Water source Src37 Calcium hydroxide source
Src16 Ammonia source Src38 Water source
Src17 Water source Src39 Water source
Src18 Water source Src40 Sulfuric acid source
Src19 Water source Src41 Water source
Src20 Enzyme 1 (Cellulase) source Src42 Water source
Src21 Chromium chloride (III) source Src43 Air source
Src22 Hydrochloric acid source Src44 Air source
Src23 Sodium chloride source Src45 Water source
Src24 Butanol source Mixers
Src25 Sodium hydroxide source Mix1 Liquid streams mixer from pretreatment stage
Src26 Sodium chloride source Mix3 Glucose and xylose mixer
Src27 Butanol source Mix4 Wastewater streams mixer
Src28 Sulfuric acid source Mix5 Required electricity mixer
Src29 Calcium hydroxide source Mix6 Required LPS mixer
Src30 Ammonium nitrate source Mix7 Required MPS mixer
Src31 Air source Mix8 Required HPS mixer
Spl1 Feedstock splitter I
Spl2 Feedstock splitter II
Spl3 Xylose splitter
Spl4 Glucose splitter
Spl5 Glucose splitter to ethanol production
Spl6 Xylose splitter to furfural production

Main reaction 3 Main reaction 3

= fi,Sol1 αS2, Sol1 P2,OUT
fi,k P2
= fi,k . i,k (A.34)
fRem ligno − i, ligno,k . f i, ligno,k (A.28)

Remaining solids flow Product separation 2

w2,OUT P2 P2,Out
fi,Sol3 Sol1 fi,k = fi,k − fi,k (A.35)
ligno,k = f Rem (A.29)
For utilities
Solids flow 3
Liq3 fkUtil1 = βkUtil1 . f F (A.36)
fi,k = xLiq3 . fi,k
Utilities for pretreatment stage
Filter 3
fkUtil2 = βkUtil2 . f F (A.37)
fi,k = δi,k . fi,k
Utilities for main reactor 1
Added components for main reaction 4 fkUtil3 = βkUtil3 . f F (A.38)
Reag,3 Liq3 MW Utilities for separation 1
fi,k = fi,k . (A.32)
fkUtil4 = βkUtil4 . f F (A.39)
Reagents for main reaction 4
Utilities for main reactor 2
 γi,r,k MWi
fi,k = Ad,4
fi,k + Liq3
fi,k + fi,k θ
Liq3 Liq3
. r,k . . (A.33) fkUtil5 = βkUtil5 . f F (A.40)
γi,r,k MW Liq3
V. Aristizábal-Marulanda, C.A. Cardona Alzate and M. Martín / Computers and Chemical Engineering 126 (2019) 35–53 51

Utilities for main reactor 3 Appendix B. MINLP problem structure

fkUtil6 = βkUtil6 . f F (A.41) The general MINLP formulation is the following:

Utilities to separation 2 minx,y z = f (x, y )
fkUtil = fkUtil1 + fkUtil2 + fkUtil3 + fkUtil4 + fkUtil5 + fkUtil6 (A.42)
s.t. h(x, y ) = 0
Total utilities consumption in the unit

C. Inequality constraints g(x, y ) ≤ 0

• Connection equations

fi,k = Sn . fi,k (A.43) x ε n , y ∈ {0, 1}m
Xylose flow rate where:

S1 + S2 + S3 = 1 (A.44) z: represents the objective function (e.g., annualized total cost)

x: is a vector of n continuous variables (e.g., composition, flows,
Mass fraction in xylose distribution
fi,k = Sn . fi,k (A.45) y: is a vector of integer variables (e.g., alternative reagents or
Glucose flow rate h(x, y ) = 0 denotes the m linear or non-linear equality con-
strains (e.g., mass and energy balance)
S4 + S5 + S6 + S7 = 1 (A.46)
g(x, y) ≤ 0 represents the p linear or non-linear inequality con-
Mass fraction in glucose distribution strains (e.g., feasibility constrains and logical constrains)
Sn parameter is the mass fraction of xylose or glucose in the
sugars distribution and its value can be between 0 and 1. The sub-
script can be 1 to 3 for xylose distribution and 4 to 7 for glucose
Agronet, (2017). Estadísticas. Agrícola [WWW Document]. Accessed June 2017. URL
• Logic constraints
Alibaba, (2018). International Prices [WWW Document]. URL
com/ Accessed February 2018.
This type of equations are included in the big-M strategy for Aristizábal, M.V., García, V.C.A., Cardona, A.C.A., 2016. Integrated production of dif-
MILP problems. Big-M term aims to guide the mathematical solver ferent types of bioenergy from oil palm through biorefinery concept. Waste
to a suitable solution considering constrains to evade meaningless Biomass Valorization 7, 737–745. 016- 9564- 7.
Aristizábal Marulanda, V., 2015. Jet Biofuel Production From Agroindustrial Wastes
Through Furfural Platform. Universidad Nacional de Colombia. Departamento de
Ingeniería Química Master thesis.
y3 + y4 + y5 + y6 + y7 = 1 (A.47) Balan, V., Da Costa Sousa, L., Chundawat, S.P.S., Marshall, D., Sharma, L.N., Chamb-
liss, C.K., Dale, B.E., 2009. Enzymatic digestibility and pretreatment degradation
Logical constrain for pretreatment stage products of AFEX-treated hardwoods (populus nigra). Biotechnol. Prog. 25, 365–
In the reaction stage, three logical constraints are considered in 375.
Balat, M., Balat, M., Kırtay, E., Balat, H., 2009. Main routes for the thermo-conversion
order to determine the best technology to produce furfural, ethanol of biomass into fuels and chemicals. Part 2: gasification systems. Energy Con-
and butanol. The variable y is an integer variable that has the val- vers. Manage. 50, 3158–3168.
ues 1 or 0, indicating the selection or activation of a determined Bao, B., Ng, D.K.S., Tay, D.H.S., Jiménez-Gutiérrez, A., El-Halwagi, M.M., 2011. A
shortcut method for the preliminary synthesis of process-technology pathways:
unit and/or stream.
an optimization approach and application for the conceptual design of inte-
grated biorefineries. Comput. Chem. Eng. 35, 1374–1383.
y10 + y11 ≤ 1 (A.48) j.compchemeng.2011.04.013.
Bertran, M.O., Frauzem, R., Sanchez-Arcilla, A.S., Zhang, L., Woodley, J.M., Gani, R.,
Logical constrain for furfural production 2017. A generic methodology for processing route synthesis and design based
on superstructure optimization. Comput. Chem. Eng. 106, 892–910. https://doi.
y12 + y13 ≤ 1 (A.49) org/10.1016/j.compchemeng.2017.01.030.
Cerón, I.X., Higuita, J.C., Cardona, C.A., 2015. Analysis of a biorefinery based on
Logical constrain for ethanol production Theobroma grandiflorum (copoazu) fruit. Biomass Convers. Biorefin. 5, 183–194. 014- 0144- 4.
Cotana, F., Cavalaglio, G., Gelosia, M., Nicolini, A., Coccia, V., Petrozzi, A., 2014. Pro-
y15 + y16 ≤ 1 (A.50) duction of bioethanol in a second generation prototype from pine wood chips.
Energy Proc. 45, 42–51.
Logical constrain for butanol production Dávila, J.A., Rosenberg, M., Castro, E., Cardona, C.A., 2017. A model biorefinery for
To connect the flows to binary decision variable are introduced avocado (Persea americana mill.) processing. Bioresour. Technol. 243, 17–29.
the following expressions.
De, F., Rodrigues, Á., Guirardello, R., 2012. Techno-economic evaluation of large scale
 2.5-dimethylfuran production from fructose. Adv. Chem. Eng. 584. https://doi.
fi,k ≤ M.yk (A.51) org/10.3303/CET1437080.
De Faveri, D., Lambri, M., Converti, A., Perego, P., Del Borghi, M., 2002. Xylitol re-
Constrain activation for feedstock covery by crystallization from synthetic solutions and fermented hemicellulose
hydrolyzates. Chem. Eng. J. 90, 291–298.
fi,k ≤ M.yk (A.52) 00134-1.
Di Fulvio, F., Eriksson, G., Bergström, D., 2015. Effects of wood properties and chip-
ping length on the operational efficiency of a 30 kw electric disc chipper. Croat.
Constrain activation for reagents J. Eng. 36, 85–100.
Ad Douglas, J.M., 1988. Conceptual Design of Chemical Processes.
fi,k ≤ M.yk (A.53) Federación Nacional de Biocombustibles de Colombia (Fedebiocombustibles) [WWW
Document], (2018). Accessed April 2018. URL http://www.fedebiocombustibles.
Constrain activation for added components com/.
52 V. Aristizábal-Marulanda, C.A. Cardona Alzate and M. Martín / Computers and Chemical Engineering 126 (2019) 35–53

Floudas, C.A., 1995. Chapter 7: process synthesis. In: Nonlinear and Mixed-Integer Pan, X., Gilkes, N., Kadla, J., Pye, K., Saka, S., Gregg, D., Ehara, K., Xie, D., Lam, D., Sad-
Optimization, pp. 225–258. dler, J., 2006. Bioconversion of Hybrid Poplar to ethanol and co-products using
GAMS. Solver manuals [WWW Document], (2018). Accessed April 2018. URL https: and organosolv fractionation process: optimization of process yields. Biotechnol.
// Bioeng. 94, 851–861.
García-Velásquez, C.A., Aristizábal-Marulanda, V., Cardona, C.A., 2018. Analysis Patel, A.D., Meesters, K., den Uil, H., de Jong, E., Blok, K., Patel, M.K., 2012. Sus-
of bioenergy production at different levels of integration in energy-driven tainability assessment of novel chemical processes at early stage: application
biorefineries. Clean Technol. Environ. Policy 1–15. to biobased processes. Energy Environ. Sci. 5, 8430.
s10098- 018- 1565- 8. C2EE21581K.
Garcia, A., Egues, I., Sanchez, C., Barta, Z., Labidi, J., 2013. Study of different bio- Peters, M.S., Timmerhaus, K.D., West, R.E., 2004. Chapter 14: materials transfer, han-
processing pathways in a lignocellulosic biorefinery by process simulation. dling, and treatment equipment-design and costs. Plant Design and Economics
Chem. Eng. Trans. 35, 505–510. for Chemical Engineers. McGraw-Hill, pp. 478–500.
García, C.A., Moncada, J., Aristizábal, V., Cardona, C.A., 2017a. Techno-economic and Pham, V., El-Halwai, M., 2011. Process synthesis and optimization of biorefinery con-
energetic assessment of hydrogen production through gasification in the Colom- figuration. AIChE J 58, 1212–1221.
bian context: coffee Cut-Stems case. Int. J. Hydrogen Energy 42, 5849–5864. Ponce-Ortega, J.M., Pham, V., El-Halwagi, M.M., El-Baz, A.A., 2012. A disjunctive pro- gramming formulation for the optimal design of biorefinery configurations. Ind.
García, C.A., Peña, Á., Betancourt, R., Cardona, C.A., 2017b. Energetic and environ- Eng. Chem. Res. 51, 3381–3400.
mental assessment of thermochemical and biochemical ways for producing en- Posada, J.A., Naranjo, J.M., López, J.A., Higuita, J.C., Cardona, C.A., 2011. Design and
ergy from agricultural solid residues: coffee Cut-Stems case. J. Environ. Manage. analysis of poly-3-hydroxybutyrate production processes from crude glycerol.
1–9. Process Biochem 46, 310–317.
García Velásquez, C.A., 2016. Hydrogen Production Through Gasification and Dark Posada, J.A., Patel, A.D., Roes, A., Blok, K., Faaij, A.P.C., Patel, M.K., 2013. Potential
Fermentation. Universidad Nacional de Colombia. Departamento de Ingeniería of bioethanol as a chemical building block for biorefineries: preliminary sus-
Química Master thesis. tainability assessment of 12 bioethanol-based products. Bioresour. Technol. 135,
Giarola, S., Zamboni, A., Bezzo, F., 2011. Spatially explicit multi-objective optimisa- 490–499.
tion for design and planning of hybrid first and second generation biorefiner- Quintero Suárez, J.A., 2011. Design and Evaluation of Fuel Alcohol Production from
ies. Comput. Chem. Eng. 35, 1782–1797. Lignocellulosic Raw Materials Ph.D. thesis. Universidad Nacional de Colombia.
2011.01.020. Departamento de Ingeniería Eléctrica, Electrónica y Computación.
Hernández-Pérez, A.F., de Arruda, P.V., Felipe, M., das, G., de, A., 2016. Sugarcane Qureshi, N., Blaschek, H.P., 1999. Production of acetone butanol ethanol (ABE) by
straw as a feedstock for xylitol production by Candida guilliermondii FTI 20037. a hyper-producing mutant strain of Clostridium beijerinckii BA101 and recov-
Braz. J. Microbiol. 47, 489–496. ery by pervaporation. Biotechnol. Prog. 15, 594–602.
ICIS Chemical Pricing, (2010). Indicative Chemical Prices A-Z [WWW Docu- bp990080e.
ment]. URL info- chemicals- a- z/ Ramirez, Parra, 2018. Evaluation of the biobutanol production from two agroindus-
Accessed February 2018 (accessed 8.20.02). trial wastes generated in the coffee growing region: Plantain peel and milk
IEA Bioenergy Task42, 2014. National BioEconomy Strategies IEA Bioenergy Imple- whey Master thesis. Universidad Nacional de Colombia. Departamento de In-
menting Agreement Countries Bioeconomy Survey 2014. geniería Química.
Kelloway, A., Daoutidis, P., 2014. Process synthesis of biorefineries : optimization of Razak, A.M.Y., 2007. Industrial Gas Turbines: Performance and Operability https://
biomass conversion to fuels and chemicals. Ind. Eng. Chem. Res. 53, 5261–5273.
Ko, C.H., Wang, Y.N., Chang, F.C., Chen, J.J., Chen, W.H., Hwang, W.S., 2012. Potentials Rincón, L.E., Becerra, L.A., Moncada, J., Cardona, C.A., 2014. Techno-economic anal-
of lignocellulosic bioethanols produced from hardwood in Taiwan. Energy 44, ysis of the use of fired cogeneration systems based on sugar cane bagasse in
329–334. south eastern and mid-western regions of Mexico. Waste Biomass Valorization
Komesu, A., Maciel, M.R.W., Filho, R.M., 2017. Separation and purification tech- 5, 189–198. 013- 9224- 0.
nologies for lactic acid – A brief review. BioResources 12, 6885–6901. https: Rodríguez Valencia, N., Zambrano Franco, D., 2010. Avance técnico 393. Los sub-
// productos del café: fuente de energía renovable. Av. Técnicos Cenicafé 8. https:
Malmierca, S., Díez-Antolínez, R., Paniagua, A.I., Martín, M., 2017. Technoeconomic // 0120- 0178.
study of biobutanol AB production. 2. Process design. Ind. Eng. Chem. Res. 56, Rosengart, A., 2017. Process Systems Engineering for Early-Stage Process Develop-
1525–1533. ment: The Case of Sustainable Bio-Derived Adipic Acid PhD thesis. Doctoral Pro-
Mariscal Moreno, J.P., 2011. Evaluation and Selection of Microorganisms for Ethanol gram in Industrial Chemistry and Chemical Engineering. Department of Chem-
Production at Industrial Level Master thesis. Departamento de Ingeniería istry Materials and Chemical Engineering. Politecnico Di Milano.
Química, Universidad Nacional de Colombia sede Manizales. Sammons Jr, N., Eden, M., Yuan, W., Cullian, H., Aksoy, B., 2007. A flexible frame-
Martín, M., Grossmann, I.E., 2016. Optimal production of furfural and DMF from work for optimal biorefinery product allocation. Am. Inst. Chem. Eng. 4, 349–
algae and Switchgrass. Ind. Eng. Chem. Res. 55, 3192–3202. 354.
1021/acs.iecr.5b03038. Sánchez Toro, Ó.J., 2008. Síntesis de esquemas tecnológicos integrados para la pro-
Martín, M., Grossmann, I.E., 2012. Simultaneous optimization and heat integration ducción biotecnológica de alcohol carburante a partir de tres materias primas
for biodiesel production from cooking oil and Algae. Ind. Eng. Chem. Res. 51, colombianas Ph.D. thesis. Universidad Nacional de Colombia. Departamento de
7998–8014. Ingeniería Eléctrica, Electrónica y Computación.
Moncada, J., Aristizábal M., V., Cardona A., C.A., 2016a. Design strategies for sus- Sharma, A.K., 2011. Modeling and simulation of a downdraft biomass gasifier 1.
tainable biorefineries. Biochem. Eng. J. 116, 122–134. Model development and validation. Energy Convers. Manage. 52, 1386–1396.
Moncada, J., Cardona, C.A., Higuita, J.C., Vélez, J.J., López-Suarez, F.E., 2016b. Wood Smith, R., 2005. Chemical Process Design and Integration. McGraw Hill. ed.
residue (Pinus patula bark) as an alternative feedstock for producing ethanol Triana, C.F., Quintero, J.A., Agudelo, R.A., Cardona, C.A., Higuita, J.C., 2011. Analysis of
and furfural in Colombia: experimental, techno-economic and environmental coffee cut-stems (CCS) as raw material for fuel ethanol production. Energy 36,
assessments. Chem. Eng. Sci. 140, 309–318. 4182–4190.
027. Tumuluru, J.S., Tabil, L.G., Song, Y., Iroba, K.L., Meda, V., 2014. Grinding energy
Moncada, J., Cardona, C.A., Rincón, L.E., 2015a. Design and analysis of a second and and physical properties of chopped and hammer-milled barley, wheat, oat, and
third generation biorefinery: the case of castorbean and microalgae. Bioresour. canola straws. Biomass Bioenergy 60, 58–67.
Technol. 198, 836–843. 2013.10.011.
Moncada, J., El-Halwagi, M.M., Cardona, C.A., 2013. Techno-economic analysis for a Valencia, N.R., 20 0 0. Manejo de residuos en la agroindustria cafetera. Seminario In-
sugarcane biorefinery: Colombian case. Bioresour. Technol. 135, 533–543. https: ternacional Gestión Integral De Residuos Sólidos Y Peligrosos Siglo XXI.
// Varbanov, S.P., Klemeš, J., Yen Liew, J., Yow Yong, P., Pan, J., Agulonu, M., Gharaie, A.,
Moncada, J., Posada, J.A., Ramirez, A., 2015b. Early sustainability assessment for po- Perry, M., Zhang, S., Bulatov, N., Smith, I., 2014. Optimal design technologies for
tential configurations of integrated biorefineries. Screening platform chemicals. integration of combined cycle gas turbine power plant with CO2 capture opti-
Biofuels, Bioprod. Biorefining 9, 722–748. mal design technologies for integration of combined cycle gas turbine power
Moncada Botero, J., 2012. Design and Evaluation of Sustainable Biorefineries From plant with CO2 capture. Chem. Eng. Trans. 39, 1441–1446.
Feedstock in Tropical Regions Master thesis). Universidad Nacional de Colombia. 3303/CET1439241.
Departamento de Ingeniería Química. Villar, del A., (2016). Aproximación a Los Costes De Producción Y Valoración Del
Montilla, J., Arcila, J., Aristizábal, M., Montoya, E.C., Puerta, G., Oliveros, C.E., Ca- Agua Regenerada, III Jornada Agua y Sostenibilidad.
dena, G., 2008. Avance técnico 370. Propiedades físicas y factores de conversión Vidal, M., Martín, M., 2015. Optimal coupling of a biomass based polygeneration
del café en el proceso de beneficio. Av. Técnicos Cenicafé 8 Technical Report. system with a concentrated solar power facility for the constant production of
Nakano, S., Ugwu, C.U., Tokiwa, Y., 2012. Efficient production of d-(-)-lactic acid from electricity over a year. Comput. Chem. Eng. 72, 273–283.
broken rice by Lactobacillus delbrueckii using Ca(OH)2as a neutralizing agent. j.compchemeng.2013.11.006.
Bioresour. Technol. 104, 791–794. Walker, G.M., 2011. 125th anniversary review: fuel alcohol: current production and
Naranjo V, J.M., 2014. Design and Analysis of the Polyhydroxybutyrate (PHB) Produc- future challenges. J. Inst. Brew. 117, 3–22.
tion from Agroindustrial Wastes in Colombia Ph.D. thesis. Universidad Nacional 2011.tb00438.x.
de Colombia. Departamento de Ingeniería Eléctrica, Electrónica y Computación. Xie, S., Qiu, X., Yi, C., 2015. Separation of a biofuel: recovery of biobutanol by
Oka, D., Kobayashi, K., Isobe, N., Ogawa, Y., Yokoyama, T., Kimura, S., Kim, U.J., salting-out and distillation. Chem. Eng. Technol. 38, 2181–2188.
Tokuyasu, K., Wada, M., 2013. Enzymatic hydrolysis of wood with alkaline treat- 10.1002/ceat.201500140.
ment. J. Wood Sci. 59, 484–488. 013- 1359- x.
V. Aristizábal-Marulanda, C.A. Cardona Alzate and M. Martín / Computers and Chemical Engineering 126 (2019) 35–53 53

Yu, Q., Zhuang, X., Yuan, Z., Wang, Q., Qi, W., Wang, W., Zhang, Y., Xu, J., Xu, H., Zheng, L., Furimsky, E., 2003. ASPEN simulation of cogeneration plants. Energy Con-
2010. Two-step liquid hot water pretreatment of Eucalyptus grandis to enhance vers. Manage. 44, 1845–1851.
sugar recovery and enzymatic digestibility of cellulose. Bioresour. Technol. 101, Zondervan, E., Nawaz, M., Haan, A.B.D., Woodley, J.M., Gani, R., 2011. Optimal de-
4895–4899. sign of a multi-product biorefinery system. Comput. Chem. Eng. 35, 1752–1766.
Zabed, H., Sahu, J.N., Suely, A., Boyce, A.N., Faruq, G., 2017. Bioethanol production
from renewable sources: current perspectives and technological progress. Re-
newable. Sustainable Energy Rev. 71, 475–501.