Computers and Chemical Engineering 105 (2017) 56–73

Contents lists available at ScienceDirect

Computers and Chemical Engineering

journal homepage: www.elsevier.com/locate/compchemeng

A computer-aided approach for achieving sustainable process design

by process intensification
Nateetorn Anantasarn a , Uthaiporn Suriyapraphadilok a,∗ , Deenesh K. Babi b,c
a
The Petroleum and Petrochemical College, Chulalongkorn University, 254 Phyathai Road, Patumwan, Bangkok, 10330, Thailand
b
SPEED, Department of Chemical and Biochemical Engineering, Technical University of Denmark, Søltofts Plads, Kongens Lyngby, 2800, Denmark
c
Utility & Solvents, Insulin & Manufacturing 1, Novo Nordisk A/S, Hallas Allé DK-4400 Kalundborg, Denmark

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

Article history: Process intensification can be applied to achieve sustainable process design. In this paper, a systematic,
Received 14 September 2016 3-stage synthesis-intensification framework is applied to achieve more sustainable design. In stage 1, the
Received in revised form 1 February 2017 synthesis stage, an objective function and design constraints are defined and a base case is synthesized. In
Accepted 14 February 2017
stage 2, the design and analysis stage, the base case is analyzed using economic and environmental anal-
Available online 17 February 2017
yses to identify process hot-spots that are translated into design targets. In stage 3, the innovation design
stage, phenomena-based process intensification is performed to generate flowsheet alternatives that
Keywords:
satisfy the design targets thereby, minimizing and/or eliminating the process hot-spots. The application
Process intensification
Phenomena synthesis
of the framework is highlighted through the production of para-xylene via toluene methylation where
Sustainable design more sustainable flowsheet alternatives that consist of hybrid/intensified unit operations are generated
Toluene methylation from the application of phenomena-based process intensification.
Simulation © 2017 Elsevier Ltd. All rights reserved.

1. Introduction
One of the major challenges in the process industry is to achieve
innovative and sustainable process designs to efficiently utilize
raw materials, minimize energy consumption and waste gener-
ation and, improve economic and environmental metrics (Zhang
et al., 2016). Process intensification (PI) can applied in order to
employ hybrid/intensified unit operations for addressing these
challenges. Process intensification has a high potential for perform-
ing process improvement at different scales and thereby, enhancing
process efficiency and economic competitiveness (Stankiewicz and
Moulijn, 2000). Process intensification is defined as the improve-
ment of a process at three different scales, the unit operations
(flowsheet) scale, task (functional) scale and phenomena (molec-
ular) scale (Lutze et al., 2013). In moving from the highest scale
(unit operations) to the lowest scale (phenomena) generated flow-
sheet alternatives are more innovative compared to previous
scales. This is because the search space of unit operations avail-
able for flowsheet generation widens. At the unit operations scale
flowsheet alternatives that are generated consist of well-known
(conventional) unit operations. At the task-scale different tasks
can be combined in different ways to both generate the alterna-
∗ Corresponding author.
E-mail address: uthaiporn.s@chula.ac.th (U. Suriyapraphadilok).
tives at the unit operations scale and alternatives that consist of
hybrid/intensified unit operations. At the phenomena scale, phe-
nomena are connected in multiple ways to generate alternatives
at the previous two scales and alternatives that consist of novel
hybrid/intensified unit operations (Babi et al., 2015). This is shown
in Fig. 1.
According to Stankiewicz and Drinkenburg (2004), the applica-
tion of process intensification and use of hybrid/intensified unit
operations is attractive for achieving sustainable process design
because it has the potential to make new designs smaller (inten-
sified), economically and environmentally attractive, efficient and
safer.
Similar to process synthesis, the methods of process intensifica-
tion can be classified as heuristic, mathematical programming and
hybrid. Heuristic (rule-based) methods have not yet been devel-
oped for the selection of hybrid/intensified unit operations and/or
the intensification of entire processes. Caballero and Grossmann
(2004), Asprion and Kaibel (2010) and Madenoor Ramapriya et al.
(2014) proposed an optimization-based method for the enumer-
ation and selection of divided wall columns. Bravo-Bravo et al.
(2010) proposed an optimization-based method for the selection
and design of extractive-divided wall columns. Song et al. (2015)
proposed a method for the selection of pressure swing absorption
for CO2 capture and storage. Okoli and Adams (2015) proposed
an optimization-based method for the selection of divided wall

http://dx.doi.org/10.1016/j.compchemeng.2017.02.025
0098-1354/© 2017 Elsevier Ltd. All rights reserved.
 N. Anantasarn et al. / Computers and Chemical Engineering 105 (2017) 56–73 57

Fig. 1. Process innovation in moving to lower scales.

Fig. 2. Multi-stage, multi-scale framework for performing process synthesis-intensification.

columns for bio-processes. The presented heuristic and mathemat-
ical methods presented operate at the unit operations scale.
Papalexandri and Pistikopoulos (1994) proposed a hybrid
method for the integration of mass and energy at the task and
phenomena scales. Siirola (1996) proposed the means-ends anal-
ysis that identifies tasks (based on expert knowledge) that satisfy
a set of process specifications. Bessling et al. (1997) proposed a
hybrid method for the selection and design of reactive distillation
columns based on pure compound and mixture analysis. Song et al.
(1998) proposed a hybrid method for the design and analysis of
reactive distillation columns based on reactive residue maps. Kiss
et al. (2007) proposed a hybrid metod for the selection and design
of reactive divided-wall columns based on the analysis of con-
ventional unit operations operating conditions. Rong et al. (2008)
proposed a method where (process) phenomena are represented
by phases, variables that affect the phases, energy sources and
geometry, among others. When these phenomena are identified
for a specific unit operation that has the potential for improve-
ment/replacement, the process phenomena are manipulated in
order to achieve a better design, using a trial and error approach.
Seifert et al. (2012) proposed a modular task concept for process
design where sections of a process are modularized in order to
generate flexible process designs. Peschel et al. (2012) proposed
a hybrid method for generating novel intensified reactor networks
based on elementary process phenomena functions. Holtbruegge
et al. (2014) proposed a method for the synthesis and design of
membrane-assisted, reactive distillation processes based on the
combination of multiple reaction and separations tasks. Leimbrink
et al. (2015) proposed a method for the selection and analysis of
membrane-based separation techniques for post combustion CO2
capture based on flowsheet decomposition at the task (functional)
scale. Lutze et al. (2012, 2013) proposed a hybrid method for per-
forming process synthesis-intensification at the unit operations
and phenomena scales where flowsheets are decomposed into phe-
nomena that are recombined in order to generate flowsheets that
consist of (novel) hybrid/intensified unit operations. The presented
hybrid methods operate at the task and phenomena scale.
In the design/retrofit of new as well as existing processes,
innovation increases in the application of process synthesis-
intensification concepts, more so in moving towards lower scales.
Therefore, there is a need for a flexible and systematic framework
that performs synthesis-intensification at the phenomena scale
thereby managing the complexity of generating a large number of
alternatives from the combination of phenomena in order to find
more sustainable process designs. In this work, a computer-aided,
multi-stage, multi-scale process synthesis-intensification frame-
work proposed by Babi et al. (2015) is applied in order to generate
more sustainable flowsheet alternatives for the production of a
58 N. Anantasarn et al. / Computers and Chemical Engineering 105 (2017) 56–73
Fig. 3. Detailed structure of (a) CSTR, (b) membrane reactor, (c) conventional distillation column and (d) divided wall distillation column.
widely used bulk chemical, para-xylene. Alternatives are gener-
ated both at the unit operations scale and phenomena scale. The
framework consists of three main stages and is shown in Fig. 2.
In stage 1, process synthesis, a base case is first generated at the
unit operations scale based on thermodynamic insights (Jaksland
et al., 1995; Tula et al., 2015). In stage 2, design and analysis, detail
design of the base case is performed and it is analyzed using eco-
nomic and environmental analyses in order to identify process
hot-spots. Process hot-spots are defined as limitations and bottle-
necks of a process that have the highest potential for improvement
and, if minimized/eliminated generates a more sustainable pro-
cess design. In stage 3, the innovation stage, phenomena-based
process intensification is performed where the base case is decom-
posed to tasks and phenomena. Additional beneficial phenomena
are added to the identified phenomena from the base case and these
phenomena are recombined using combination rules to generate
flowsheet alternatives inclusive of hybrid/intensified unit opera-
tions that minimize/eliminate the process hot-spots. Based on a
defined objective function in stage 1 (subject to process constraints
and intensification performance criteria), the flowsheet alternative
that gives the best objective function value while satisfying the con-
straints and criteria is selected as the more sustainable design. In
this paper, the framework work-flow (Babi et al., 2015) is presented
and its application to the production of para-xylene where 3 more
sustainable designs not previously reported has been generated.
2. Sustainable process design framework
The synthesis-intensification problem is defined as follows,
given the desired product, find flowsheet alternatives that consist of
hybrid/intensified (novel) unit operations that produce the desired
product in a sustainable way. The concepts of phenomena-based
synthesis, that is, the generation of flowsheet alternatives from the
combination of phenomena building blocks (PBBs) are presented in
Section 2.1. The computer-aided, multi-stage, multi-scale process
synthesis-intensification framework (Babi et al., 2015) together
with a summary of each step is presented in Section 2.2.
2.1. Phenomena-based synthesis
Phenomena-based synthesis is similar to computer aided
molecular design methods (CAMD, Harper and Gani, 2000) where
 N. Anantasarn et al. / Computers and Chemical Engineering 105 (2017) 56–73 59

Fig. 4. Process synthesis-intensification workflow for achieving sustainable process design.

Table 1
Identification of process hot-spots and design targets.

Analysis output Root causes Process hot-spot Design target

• ECON: High raw • Unreacted raw • Contact problems • Select alternative

material cost material • Heat transfer limits reaction
• SustainPro: Negative • Equilibrium reaction • Activation energy catalyst/pathway
MVA • Difficult separation • Waste minimization
• LCSoft: High AT • Low driving force • Energy minimization

• ECON: High utility • High energy • Difficult separation • Waste minimization

costs consumption for • Low driving force • Energy minimization
• SustainPro: Positive reaction &/or • Azeotrope present
EWC separation
• LCA analysis: High • High energy
carbon footprint consumption for
heating/cooling

*MVA − mass value added, EWC − energy to waste cost, AT − aquatic toxicity.
60 N. Anantasarn et al. / Computers and Chemical Engineering 105 (2017) 56–73
molecules (similar in concept to unit operations) are designed sub-
ject to a set of target (desired) properties and using functional
groups (different combinations of atoms, which are similar in con-
cept to tasks) and/or atoms (similar in concept to phenomena) as
the building blocks. Consider, for example, atoms that can be com-
bined to form functional groups, in the same manner phenomena
building blocks (PBBs) can be combined to form simultaneous phe-
nomena building blocks (SPBs). These SPBs are combined to form
operations in the same way that groups are joined together to form
molecules. By combining building blocks at lower scales (molecules
or PBBs), new molecules and/or new process flowsheets are in
principle obtained. Also, a small set of building blocks are used
for representing multiple molecules (CAMD) as well as flowsheets
(phenomena-based synthesis).
2.1.1. Phenomena building blocks
A set of fundamental building blocks to represent a process is
required, just like atoms, groups, bonds, etc. are required to rep-
resent different molecules. According to Lutze et al. (2013), most
chemical processes can be represented by different combinations
of the following 9 phenomena building blocks: cooling (C), heat-
ing (H), mixing (M), two-phase mixing (2phM), reaction (R), phase
contact (PC), phase transition (PT) and phase separation (PS) and,
dividing phenomena (D). The inlet/outlet stream conditions of the
9 defined PBBs are liquid (L), vapor and liquid (VL), liquid-liquid
(L-L) vapor (V), vapor-liquid-liquid (VLL), solid (S) and solid-liquid
(SL). Consider the separation task of a vapor-liquid mixture. The
PBBs that are associated with the separation task are as follows:
• M – mixing is assumed to inherently occur when 2 or more com-
pounds are in contact
• 2phM – mixing of the two phase, vapor and liquid
• PC(VL) – contact of the vapor and liquid phases occur during the
separation
• PT(VL) – upon contact of the phases, there are transition of
molecules from the vapor phase to the liquid phase and vice versa
• PS(VL) – upon phase transition, there are also separation of
between the vapor phase and liquid phase
Therefore, consider a tray distillation column; the 5 mentioned
PBBs above perform the vapor-liquid separation task and describes
the separation occurring on each tray in the column. Fig. 3 gives
examples of different reaction and separation unit operations and
the PBBs that represent their separation tasks.
2.1.2. Simultaneous phenomena building blocks
In order to use PBBs to identify feasible tasks, PBBs are combined
to form SPBs through the application of combination rules that are
based on thermodynamic insights (Jaksland et al., 1995). Examples
of these combination rules are as follows:
• H and C PBBs cannot be combined to form a H C SPB because this
combination is not thermodynamically efficient
• M, R and C PBBs are combined to form a M(L) = R(L) = C SPB
because it represents a liquid phase, exothermic reaction
• M, 2phM, R, C, PC, PT, PS PBBs are combined to form a
M(VL) = R(L) = 2phM = PC(VL) = PT(VL) = PS(VL) SPB because it rep-
resents a liquid phase, reaction with simultaneous reaction and
separation. Note that such a SPB is beneficial because it provides
a means by which equilibrium limitations can be improved.
Fig. 3 gives examples of different reaction and separation unit
operations and the SPBs that represent them. Three different cat-
egories of SPBs are used: initiator, intermediate and terminator
SPBs. The initiator SPB fulfills the main objective of a task while
the terminator SPB completes a task. An intermediate SPB is an
initiator SPB that is repeated multiple times. Note that an initiator
SPB can either be an intermediate and terminator SPB. Consider the
divided wall column in Fig. 3. The initiator SPB is repeated multiple
times in order to satisfy the specified separation and the termina-
tor SPBs that complete the separation task represent the condenser
and reboiler of divided wall column.
2.1.3. Basic structures
Basic structures are generated from the combination of SPBs and
the feasible tasks that constitute a flowsheet are selected based
on basic structures that can perform the selected task. Basic struc-
tures represent reaction, separation and reaction-separation tasks,
respectively. Fig. 3 gives examples of the basic structures that rep-
resent different reaction and separation unit operations. Therefore,
using Fig. 3 it becomes evident that given a flowsheet (for example,
a base case) one can move from the highest scale (unit opera-
tions) to the lowest scale (phenomena). It also becomes evident that
in performing phenomena-based process intensification, one can
associate basic structures with unit operations therefore, having
selected a particular basic structure for performing (for example) a
separation task, one can identify the potential separation unit oper-
ations. Consider the separation of a ternary mixture. Using Fig. 3,
at the unit operations scale either conventional distillation and
divided wall column are selected because they both are represented
by the same basic structure.
2.2. Computer-aided workflow
The sustainable process synthesis-intensification framework
applied in this work consists of 12 main steps where the first 8
steps are performed at the unit operations scale, and the remain-
ing 4 steps are performed at the tasks and phenomena scales (Babi
et al., 2015). The framework is presented in Fig. 4 followed by a
summary of each step. Note that the output from each step is the
input to the next step and where required, additional information
maybe be required and are therefore, additional inputs.
2.2.1. Step 1: need identification
The objective of this step is to collect production and cost data
for the products. A literature survey is performed in order to obtain
information on the product uses and total annual production rate
that is used to select the production rate. Note that if the process is
existing and improvements would like to be made to the existing
process, the problem statement is related to retrofit.
2.2.2. Step 2: problem definition
The objective of this step is to define the objective function, (pro-
cess) constraints (for example, product purity/production rate) and
(intensification) performance criteria (for example, improvement
in economic and environmental factors). The problem definition is
defined in terms of the production target and product usage. The
objective function is defined terms of profit, or costs (operational
and/or capital), for example, total yearly profit, total annualized
cost, operating cost, etc.
2.2.3. Step 3: reaction identification/selection
The objective of this step is to select the raw materials, reac-
tion path and reaction information for production of the (desired)
product. A literature survey is performed in order to obtain infor-
mation about a feasible reaction path and raw material use for
production of the desired product. If multiple reaction paths exist
then a reaction pathway analysis is recommended (Kongpanna
et al., 2015). Retrieve the following data for a selected reaction
path, reaction equilibrium/kinetic data, catalyst data (homoge-
nous/heterogeneous, selectivity, etc.) and reaction phases. The heat
 N. Anantasarn et al. / Computers and Chemical Engineering 105 (2017) 56–73 61

Table 2
Identification of tasks and phenomena building blocks.

Unit operation Primary task Task type PBBs

CSTR reactor Reaction • L phase • M, R(L)

• Exothermic L-phase • M, R(L), C
• VL phase reaction • M, 2phM, R(L), C, PC(VL),
PT(VL), PS(VL)

Flash Separation • VL separation • M, 2phM, PC(VL), PT(VL),

PS(VL)

Distillation Separation • M, 2phM, PC(VL), PT(VL),

PS(VL)

Decantor Separation • M, 2phM, PC(LL), PT(LL),

PS(LL)

CSTR – continuous stirred tank reactor, L – liquid, VL – vapor-liquid.

Table 3
Binary ratio example.

rij MW Tm Tb RG SolPar VdW VM VP

A/B 2.31 1.00 1.02 1.93 1.53 1.96 1.97 1.7

MW – Molecular weight, Tm – Melting point, Tb – Boiling point, RG – Radius of gyration, SolPar – Solubility parameter, VdW – Van der Waals volume, VM – Molar volume,
VP – Vapor pressure.

Table 4
Process hot-spots translation into desirable phenomena.

Process hot-spots Task Binary ratio MSA required? Desirable PBBs

Activation energy Reaction Grxn No M, H

• Difficult separation Low driving force Separation VM, SolPar, RG No PT(PVL), PT(VV), PS(VV)
• Azeotrope present Separation Tb , VP, Hvap, SolPar Yes PC(VL), PT(VL), PS(VL)

*MSA – mass separating agent, Grxn – Gibbs free energy change, VM – Molar volume, SolPar – Solubility parameter, RG – Radius of gyration, Tb – Boiling point, VP – Vapor
pressure, Hvap – heat of vaporization.

Table 5
Defined constraints (␪) and performance criteria (␸).

Category Objectives

Logical constraints (␪1 ) • First flowsheet task: reaction

• Products, by-products and (un-reacted) raw materials are recovered using
separation tasks

Structural constraints (␪2 ) • Flowsheet consists of reaction and separation tasks

• Combination rules are applied for generating PBBs, SPBs & basic structures
• Use of recycle streams

Operational constraints (␪3 ) • Raw materials are used in their pure state
• Un-reacted raw materials are recovered and recycled
• Production rate: 19,000 kg/hr
• Product purity (p-xylene) = 99.7 wt%

Performance criteria (␸) • Hybrid/intensified unit operations considered

• Improvement in economic factors
• Improvement in sustainability factors
• Improvement in LCA factors
• Flowsheet unit-operations reduction

of reaction is estimated using the ICAS database or via property
prediction, ProPred (Gani et al., 1997).
2.2.4. Step 4: availability of a base case
The objective of this step is to pre-select a base case that must
be verified as a feasible base case (step 5). A literature survey is
performed in order to find a design that utilizes the reaction path
selected in step 3. If a design is found then, it is pre-selected as a
base case and step 5 is entered. If a design is not found then, a base
case must be generated and step 6 is entered.
2.2.5. Step 5: base case feasibility
The objective of this step is to verify the flowsheet structure of
the pre-selected base case. To verify the possibility of using the pre-
selected base case as a base case, the task based synthesis method
proposed by Douglas (1985) is applied which investigates the dif-
ferent steps in the design related to design input/output structure,
reaction, sequence of separation unit operations, heat recovery
units, etc. If the pre-selected base case satisfies the method then,
the pre-selected base case is selected as the base case and step 7 is
62 N. Anantasarn et al. / Computers and Chemical Engineering 105 (2017) 56–73
Table 6
Kinetic data for toluene methylation packed bed reactor (Satelo et al., 1993).
Kinetic constant Value (Frequency factor) Activation
Energy (kJ/mol)
k1 (Eq. (7)) 384.5 mol/(g h atm2 ) 60.4
k2 (Eq. (8)) 1999 mol/(g h atm2 ) 62.4
k3 (Eq. (9)) 250.7 mol/(g h atm) 66.6
k4 (Eq. (10)) 4.11 mol/(g h atm) 26.4
k5 (Eq. (11)) 165.8 mol/(g h atm) 54.2
entered. If the pre-selected base case does not satisfy the method
then a base case must be generated and step 6 is entered.
2.2.6. Step 6: generation of a base case
The objective of this step is to generate a feasible base case if
no (pre-selected) base case is available or if the pre-selected base
case does not satisfy the synthesis method in step 5. A hybrid task,
thermodynamic insight-based method is applied for the genera-
tion of a (feasible) base case (Jaksland et al., 1995). Reaction and
separation tasks are identified and selected based on pure com-
pound and mixture properties. The selected tasks are combined in
multiple ways in order to generate task-based flowsheet alterna-
tives that satisfy the process constraints. The task-based flowsheet
alternatives are translated into unit operation-based flowsheet
alternatives because properties, tasks and separation techniques
are connected since properties are the driving forces that affect
different separation techniques (Jaksland et al., 1995; Tula et al.,
2015). The objective function (defined in step 2) is calculated and
the alternative that gives the best objective function value (subject
to the constraints) is selected as the base case.
2.2.7. Step 7: rigorous simulation
The objective of this step is to perform detail design of the base
case. The base case is simulated using rigorous models in order to
obtained design data of the selected unit operations and, detailed
mass and energy balance data.
2.2.8. Step 8: economic, sustainability and LCA analysis
The objective of this step is to perform economic and envi-
ronmental analyses in order to identify process hot-spots that are
translated into design targets that if matched minimizes and/or
eliminates the process hot-spots.
2.2.8.1. Economic analysis. An economic analysis is performed
using ECON (Gani et al., 1997; Saengwirun, 2011) and is based on
the economic models of Peters et al., 2003.
2.2.8.2. Sustainability analysis. A sustainability analysis is per-
formed using SustainPro (Carvalho et al., 2013) and is an
indicator-based method where critical path indicators are calcu-
lated for the identification of process hot-spots. The paths are either
open paths (OP) or closed paths (CP). An OP is related to in-out
stream paths and a CP is related to recycle stream paths in a pro-
cess. Examples of the calculated indicators are material value added
(MVA), energy to waste cost (EWC) and total value added (TVA). The
MVA indicator provides an estimate of the value added in an OP of a
given compound. Negative values of this indicator indicate that the
compound is being lost within the OP (as waste). The EWC indica-
tor provides an estimate of the energy consumption and compound
treatment costs in OPs and CPs related. Positive values of this indi-
cator indicate high energy consumption within the OP/CP. The TVA
is the difference between MVA and EWC. Negative values of this
indicator show potential for improvements related to the reduction
in operational costs.
2.2.8.3. Life cycle assessment (LCA) Analysis. A life cycle assessment
analysis is performed using LCSoft (Kalakul et al., 2014) and is an
indicator-based method (using cradle to the gate concept for LCA
analysis based on ISO 14040 standard) where a set of environmental
indicator impacts are calculated for identification of process hot-
posts related to, for example, toxicity, waste generation and carbon
footprint. Examples of the calculated indicators are human aquatic
toxicity (AT), global warming potential (GWP), ozone depletion
(ODP) and carbon footprint.
2.2.8.4. Identification of design targets. The calculated indicators
from the economic, sustainability and LCA analysis that have the
highest impact for improvement in the base case are identified
as process hot-spots. The process hot-spots are then translated
into design targets that if matched minimizes and/or eliminates
the process hot-spots. Examples in moving from identified process
hot-spots to design targets are gives in Table 1.
2.2.8.5. Generate alternatives at the unit operations scale. Flowsheet
alternatives are generated at the unit operations scales subject
to the (process/intensification) constraints and design targets. In
order to generate alternatives, process optimization and/or heat
and mass integration are performed. Next, phenomena-based pro-
cess intensification is performed in the remaining 4 steps of the
method in order to generate flowsheet alternatives that consist of
hybrid/intensified unit operations.
2.2.9. IT-PBS-1: process analysis
The objective of this step is to generate the task-based and
phenomena-based flowsheets and analyze flowsheet physiochem-
ical properties
2.2.9.1. Action: Identify tasks and PBBs. Using the PI knowledge-
based (Lutze et al., 2013; Babi et al., 2015) the tasks and phenomena
of a flowsheet are identified. Table 2 gives examples in moving from
the unit operations scale to the task scale and phenomena scale.
2.2.9.2. Pure compound and mixture analysis. The binary ratio is
calculated in order to extract thermodynamic insights related to
pure compound properties (Jaksland et al., 1995). The binary ratio
provides insights into the formation of azeotropes, difficult sep-
arations, miscibility gaps, etc. (Jaksland et al., 1995; Tula et al.,
2015). The binary ratio (rij) of two pure compounds is given as
the quotient of the highest and lowest property value. The total
number of binary pairs for compounds (NC) in a mixture is given
by NC(NC-1)/2. Table 3 gives an example of the binary ratio of AB
compounds. The boiling point ratio is close to one and therefore,
hints that difficult separation between A/B is highly probable and/or
an azeotrope exist between the compounds. The mixture analysis
consists of azeotrope analyses, for example, azeotrope type (het-
erogeneous, homogenous), pressure dependence, and miscibility
analyses for identification of potential phase splits at certain oper-
ating conditions. This information is used in IT-PBS-3 for generation
of flowsheet alternatives.
2.2.10. IT-PBS-2: identification of desirable tasks and PBBs
The objective of this step is to identify desirable PBBs that can
match the design targets thereby minimizing/eliminating the pro-
cess hot-spots. The process hot-spots and, pure compound and
mixture analyses are used to identify desirable tasks and PBBs that
have the highest potential for matching the design targets thereby,
eliminating and/or minimizing the identified process hot-spots in
step 8. In the PI knowledge-based, process hot-spots are associated
with pure compound/mixture properties that are associated with
PBBs. Table 4 gives examples of the association between process
hot-spots and, desirable tasks and PBBs. Consider the separation
 N. Anantasarn et al. / Computers and Chemical Engineering 105 (2017) 56–73 63

Fig. 5. Selected base case flowsheet design of toluene methylation for the production of p-xylene including closed path (CP) and open path (OP) analysis.

of a AB minimum boiling azeotrope and the following identified

process hot-spots; difficult separation, low driving force, azeotrope
present. Using Table 4, the process hot-spots are selected and the
corresponding calculated binary ratios are analyzed (calculated in
IT-PBS-1). If the binary ratio satisfies the thermodynamic insights
upper and lower bounds (Jaksland et al., 1995) then the desirable
PBBs are selected. The total set of PBBs to be used in IT-PBS-3 for
flowsheet generation consists of the PBBs identified in IT-PBS-1 plus
the desirable PBBs identified in this step (IT-PBS-2).

2.2.11. IT-PBS-3: generation of feasible flowsheet alternatives

The objective of this step is to generate all feasible flowsheet
alternatives inclusive of hybrid/intensified unit operations. The
identified set of PBBs (IT-PBS-2) is combined using combination
rules in order to generate feasible SPBs (Lutze et al., 2013; Babi
et al., 2015).
2.2.11.1. Generate task-based superstructure. The minimum num-
ber of required separation tasks (NS) for recovering pure
compounds (NC) at the outlet of a reaction tasks is given by NC-1.
Note that not always is there a requirement to recover all com-
pounds in their pure state (Babi and Gani, 2014). The task-based
superstructure consists of both individual and integrated tasks.
Integrated tasks are the merging of adjacent reaction and separa-
tion tasks and, a separation task with another separation task.
2.2.11.2. Identify basic structures and feasible task-based flowsheet
alternatives. The generated SPBs are combined to form basic struc-
tures that satisfy a task. If the basic structure satisfies a desired task
then the task becomes feasible. Having identified the feasible tasks,
the tasks can be combined to generate feasible task-based flow-
sheets, that is, the feasible task-based flowsheet can be identified
from the generated task-based superstructure. Therefore, multiple
basic structures that perform multiple tasks can be a priori gener-
ated and stored in the PI knowledge(Babi et al., 2015).
2.2.11.3. Identify unit operations-based flowsheet alternatives. The
task-based flowsheet alternatives that constitute the basic
structures are translated to generate the unit operations-
based flowsheet alternatives. Each initiator SPB is associated
with a unit operation and therefore, the conventional and/or
hybrid/intensified unit operation is selected that satisfies the iden-
tified task in the task-based flowsheet.
2.2.11.4. Model based analysis. The generated flowsheet alterna-
tives may consist of both conventional and hybrid/intensified unit
operations. In order to evaluate if the flowsheet alternatives satisfy
Fig. 6. Economic analysis interms of utility costs of the base case (cond. – Condenser
and reb. – reboiler).
both the (process) constraints, performance criteria/design tar-
gets then a model-based analysis of the alternatives are performed
using rigorous models. The analysis can be performed using rig-
orous process simulation or object oriented simulation (ICAS-MoT,
Heitzig et al., 2011) since process simulators do not contain rigorous
equipment models for many hybrid/intensified unit operations. For
object oriented simulation (Babi et al., 2014; Papadakis et al., 2016),
equipment models are either retrieved from a model library (Gani
et al., 1997) or developed and stored in a model library (Fedorova
et al., 2014).
Examples of combination rules for combining PBBs to form SPBs,
generation of the task-based superstructure, identification of basic
structures from the combination of SPBs and the translation of basic
structures into unit operations to generate flowsheet alternatives,
consisting of hybrid/intensified unit operations are provided in the
supplementary material.
2.2.12. IT-PBS-4: comparison and selection of the more
sustainable alternatives
The objective of this step is to evaluate the generated flow-
sheet alternatives for selection of the more sustainable alternative.
The economic and environmental analyses are performed in
order to analyze the generated alternatives. The selected flow-
sheet alternatives are those that satisfy both the (process)
constraints, performance criteria/design targets thereby, mini-
mizing/eliminating the identified process hot-spots in step 8.
Therefore, the selected designs will always be better than the base
case and non-compromising solutions are found. Finally, the objec-
tive function is calculated and each alternative is ranked in order
to select the more, innovative sustainable design.
64 N. Anantasarn et al. / Computers and Chemical Engineering 105 (2017) 56–73

Table 7
Overview of base case design simulation.

Information Value

Feed mole ratio (Toluene:MeOH) 2:1

p-Xylene product (kg/h) 21,385
Energy usage (MJ/h) 677,970
Utility costs ($/year) 8,716,896

 
f x , y, d, z, ␪ =0 (3)

aLB ≤ A1x + A2 y ≤ aUB (4)

Fig. 7. LCA analysis in terms of carbon footprint of the base case (cond. – Condenser
and reb. – reboiler).
3. Application
The framework is highlighted through the application of the
production of high purity para-xylene (p-xylene) via toluene
methylation where more sustainable flowsheet alternatives are
generated that consists of hybrid/intensified unit operations not
previously reported for the production of this bulk-chemical.
3.1. Step 1: need identification
Para-xylene is an important raw material in the plastics and
polyesters industry (Wantanachaisaeng and O’Neil, 2010) and the
estimated total annual production is 35 × 106 tons/year (Dursch
et al., 2009). The specified design production rate is 19,000 kg/hr
(164 × 103 tons/year) with a minimum product purity of 99.7 wt%
purity.
3.2. Step 2: problem definition
The problem is defined as the identification of a more sustain-
able flowsheet for the production of p-xylene and that minimizes
the objective function defined in Eq. (1) subject to Eqs. (2)–(7):
  
crm,i ṁrm,i + cu,j Ėu,j
lowest Fobj =
ṁproduct
Subject to x, y, d, z, ␪withy ∈ {0, 1}, x ≥ 0, d, z, ␪ ∈ R and
 
g x , z, ␪ =0
 
pLB ≤ p x , y, d, ␪ ≤ pUB (5)
 
sLB ≤ s x , y, d, ␪ ≤ sUB (6)
The objective function in Eq. (1) represents the operating costs
where C – costs, m – mass flow, E – energy usage, rm – raw material
and u – utility. The objective function is a function of (x), decision
variables (y), equipment parameters (d), thermodynamic variables
(z) and specifications of process (␪). The synthesis-intensification
problem is formulated as a mixed integer non-linear programming
problem (MINLP) because the objective function and constraints
can be linear and/or non-linear and decision integer variables are
used for selection of alternatives. In order to manage the com-
plexity of solving the MINLP synthesis-intensification problem, a
decomposition-based solution approach is applied that decom-
poses the problem into a set of sub-problems that are solved
according to a pre-defined calculation order (Karunanithi and
Achenie, 2005). The sub-problems require the solution of a sub-set
of equations and the final sub-problem is solved as a set of NLPs or
MILPs. Therefore, flowsheet alternatives are generated by simulta-
neously solving the process model equations, Eqs. (2)–(3), subject
to the (process) constraints. The constraints (Eqs. (4)–(5)) are classi-
fied as logical, structural and operational constraints and are given
in Table 5. The generated flowsheet alternatives are screened using
Eq. (6), intensification performance criteria and given in Table 5. The
objective function (Eq. (1)) is calculated and the flowsheet alter-
natives that gives the best objective function value is selected as
(1) the more sustainable process design. The result from the solution
approach is dependent on the defined problem, the objective func-
tion, constraints and performance criteria, the quality of the input
data, models and model parameters. A global optimal solution is
not guaranteed however, the solution obtained will most likely be
(2) one of the best solutions that can be obtained (Lutze et al., 2013).

Fig. 8. Flowsheet alternative 1.

 N. Anantasarn et al. / Computers and Chemical Engineering 105 (2017) 56–73 65

Table 8
Identified critical paths from the sustainability analysis in the base case.

Path Compound Flow rate (kg/h) MVA (103 $/year) EWC (103 $/year) TVA (103 $/year)

Open path (OP)

OP42 Ethylene 1,547.21 −40,876.96 12.16 −40,889.11

Closed path (CP)

CP12 Toluene 69,288.99 – 5,544.28 –

OP – Open path, CP – Closed path, italic numbers represent a major ranking consideration.

Table 9
Translation of process hot-spots to design targets and improvement alternatives.

Process hot-spots Design targets Improvement Methods

• Difficult separation • Waste minimization Heat integration Pinch analysis

• Low driving forces • Energy minimization Separation sequencing • Driving force
• High energy (utility, carbon • Improvement in economic (optimization) analysis
footprint) consumption in and environmental factors (Bek-Pedersen and
T2, T3, T4 and HX3 Gani, 2003)
• Sensitivity analysis

Ethylene purification Thermodynamic

insights (Jaksland et al.,
1995)

Table 10
Comparison between normal base case and optimized base case.

Performance criteria Base case Flowsheet alternative 1

Overall Heating consumption (MJ/h) 333,134.73 276,837.01

Overall Cooling consumption (MJ/h) 344,838.32 288,290.78
Ethylene purity (wt%) 52.74 99.50
Utility costs ($/h) 1,210.68 828.29
Objective function value (operating cost/p-xylene produced) 0.94 0.90
Carbon footprint (kg CO2 /kg of p-xylene produced) 1.97 1.51

3.3. Step 3: reaction identification/selection C8 H10 –m-Xylene, o-C8 H10 –o-Xylene, H2 O – Water, DTBB
– di-tert-butyl-benzene, TBMX – tert-butyl m-xylene, TBB –
The selected reaction path is a methylation reaction (Breen et al., tert-butyl-benzene
2005) where the raw materials toluene and methanol (MeOH) The kinetic constants and activation energy are given in Table 6.
react to form p-xylene. A magnesium modified ZSM-5 catalyst is
used inside a packed bed reactor (Ashraf et al., 2013). A set of 5 3.4. Step 4: availability of base case
reactions occur inside the reactor; toluene methylation (Eq. (7),
irreversible reaction), methanol dehydration (Eq. (8), irreversible A literature survey found a known base case shown in Fig. 5.
reaction), toluene disproportionation (Eq. (9), irreversible reac- The pre-selected base case consists of 8 unit operations; 4 distil-
tion), p-xylene de-alkylation (Eq. (10), irreversible reaction) and lation columns (T1–T3 and T5), 1 reactive distillation column (T4),
p-xylene isomerization (Eq. (11), irreversible reaction) (Satelo et al., 1 flash vessel (F1), 1 decanter vessel (D1) and 1 reactor (R1). The
1993). Two additional reactions are also presented (Eqs. (12)–(13)) raw materials toluene and methanol are fed at a mole ratio of 2:1
and are presented next in step 4. The main product p-xylene is to R1 at an operating temperature of 400 ◦ C and operating pressure
produced together with the by-products benzene, water, ethylene, of 3 bar. In order to achieve a high conversion of toluene, the ratio
meta-xylene (m-xylene) and ortho-xylene (o-xylene). of catalyst weight (g) and toluene feed (mol/h) is fixed at 2.5. The
outlet of R1 consists of the product p-xylene, by-products benzene,
C7 H8 + CH3 OH → p-C8 H10 + H2 O (7)
water, ethylene, m-xylene, o-xylene together with the un-reacted
2CH3 OH → C2 H4 + 2H2 O (8) raw materials, toluene and methanol. To recovery the p-xylene, a
series of separation unit operations are used. Ethylene is recovered
C7 H8 → 0.5C6 H6 0.5p-C8H10 (9) as the overhead stream in F1. Water and methanol are separated
p-C8 H10 → C7 H8 + 0.5C2 H4 (10) from the aromatics (toluene, benzene and xylenes) using liquid-
liquid separation in D1. Methanol is recovered from water in T1 as
p-C8 H10 → 0.5m-C8 H10 + 0.5o-C8 H10 (11) the overhead product and recycled. The aromatic stream containing
DTBB + m-C8 H10 ↔ TBMX + TBB (12) benzene, toluene and the other xylenes are feed to the separation
system that consists of 3 distillation columns and 1 reactive distilla-
TBB + m-C8 H10 ↔ TBMX + C6 H6 (13) tion column. Xylenes and toluene are recovered from benzene in T2
xTBMX xTBB as the overhead product. Toluene is further purified in T3 (overhead
K= = 0.6 (14) product) and recycled to the reactor. The xylenes (bottoms product
xDTBB xm−x
of T3) are fed to T4 (reactive distillation) where using a solvent mix-
xTBMX xTB
K= = 0.16 (15) ture of di-tert-butyl-benzene (DTBB) and tert-butyl-benzene (TBB),
xTBB xm−x
m-xylene is converted into benzene. The reactions are given in Eq.
Reaction key: CH3 OH – Methanol, C2 H4 –Ethylene, (12)-(13). M-xylene is converted into benzene in order to separate
C6 H6 –Benzene, C7 H8 –Toluene, p-C8 H10 –p-Xylene, m- m-xylene from p-xylene because it is difficult to recover p-xylene
66 N. Anantasarn et al. / Computers and Chemical Engineering 105 (2017) 56–73

Fig. 9. Task based flowsheet of the base case design.

Fig. 10. Phenomena based flowsheet of the base case design.

Table 11
Transformation of the base case flowsheet from unit operation to PPBs level.

Unit operation Task Responsible PBBs

Reactor, R1 Reaction M, C, R
Flash vaporization, F1 Separation 2phM, PC(VL), PT(VL), PS(VL)
Decanter, D1 Separation 2phM, PC(LL), PS(LL)
Distillation column, T1, T2, T3, T5 Separation 2phM, C/H, PC(VL), PT(VL), PS(VL)
Reactive distillation column, T4 Reaction- Separation 2phM, C/H, R,PC(VL), PT(VL), PS(VL)
Total PBBs in the base case R, M (4 types), 2phM, C, H, PT(VL), PC(VL), PS(VL), PS(LL), PT(LL), D (14 PBBs in total)

from m-xylene and o-xylene due to their very similar physiochem-
ical properties. P-xylene is recovered from benzene in T5 as the
bottoms product at a purity of 99.7 wt%.
3.5. Step 5: base case feasibility
default thermodynamic model for all vapor-liquid separations is
the Peng-Robinson equation of state and the thermodynamic model
for modelling the liquid-liquid separation in D1 is the NRTL. An
excerpt of the simulation results (mass and energy) is given in
Table 7.

The pre-selected base was verified using the method of Douglas

(1985) and therefore, selected as the base case. Step 7 is entered. 3.7. Step 8: economic, sustainability and LCA analysis

3.6. Step 7: rigorous simulation 3.7.1. Economic Analysis

The results of the economic analysis are presented in Fig. 6.
The base case is rigorously simulated using Aspen Plus sim- Fig. 6 shows that the reboiler of the columns T2, T3 and T4 and
ulator using equilibrium-based equipment models. The selected heat exchanger HX3 have the highest energy consumption, that is,
 N. Anantasarn et al. / Computers and Chemical Engineering 105 (2017) 56–73 67

Table 12 reduced by 17% and this translates into a carbon footprint reduction
Excerpt of the binary ratio matrix.
of 23% and further minimization of the objective function.
rij Tb RG SolPar VM

Toluene/MeOH 1.14 2.24 1.61 2.74 3.8. IT-PBS-1: process analysis

Toluene/p-Xylene 1.07 1.09 1.02 1.19
p-Xylene/MeOH 1.22 2.44 1.65 3.23
Toluene/Water 1.03 5.65 2.63 4.81 3.8.1. Identify tasks and PBBs
MeOH/Water 1.10 2.52 1.61 1.76 Table 11 lists the identified tasks and PBBs in the base case
p-Xylene/Water 1.10 6.23 2.70 5.71 and, the task-based flowsheet and phenomena-based flowsheet are
Tb – Boiling point, RG – Radius of gyration, SolPar – Solubility parameter, VM – Molar shown in Fig. 9 and Fig. 10, respectively.
volume. Table 11 lists the PBBs present in the base case.

they consume more than the 50% of all heat energy supplied to the
process.
3.7.2. Sustainability analysis
The results of the sustainability analysis are presented in Table 8.
The two main OPs and CPs that have the highest potential for
improvement are listed in Table 8 and highlighted in Fig. 5. In OP42,
the MVA is positive and indicates that the by-product ethylene is
being lost in this path and therefore, should be utilized, that is, sold
as a value-adding by-product that has a positive net contribution
to the profit. In CP 12, the EWC is positive and indicates that there
is high energy consumption in this path. This is confirmed by the
economic analysis because in this path the high energy consuming
unit operations are found, T2, T3, T4 and HX3.
3.7.3. Life cycle assessment (LCA)
The results from the LCA analysis are presented in Fig. 7. As con-
firmed by the economic and sustainability analyses, the path that
contains the unit operations that consume the highest amount of
heat energy in the process (CP12) also has the highest carbon foot-
print. Therefore, it can be concluded that all three analyses must
correspond.
3.7.4. Identify process hot-spots
The identified process hot-spots and how they can be improved
at the unit operations scale is given in Table 9.
3.7.5. Generate alternatives at the unit operations scale
The driving force method for (optimal) sequencing of distillation
columns is applied (Bek-Pedersen and Gani, 2003). From apply-
ing the driving force method it was found that the sequencing of
the distillation columns in the base case satisfied the method. The
recovery of ethylene was investigated and it was found that with
the addition of compressors and an heat exchanger connected to
a flash column, high purity ethylene can be further recovered at a
purity (for re-sale) of 99.5 wt% compared to the base case of 57%.
The design specifications (reflux ratio, distillate to fed ratio) of the
distillation columns were investigated and a sensitivity analysis
applied in order to find the best design specifications. The generated
flowsheet alternative is given in Fig. 8.
The results for flowsheet alternative 1 compared to the base is
given in Table 10. The overall heat energy consumption has been
3.8.2. Pure compound and mixture analysis
Excluding the solvents, there are 8 compounds (toluene,
methanol, p-xylene, m-xylene, o-xylene, ethylene, benzene, water)
and therefore, 28 binary pairs. The pure component properties are
analyzed and an excerpt of the calculated binary ratio matrix is
given in Table 12.
From analysis of mixture properties of the binary pairs of com-
pounds, 7 binary azeotropes are found and given in Table 13.
3.9. IT-PBS-2: identification of desirable tasks and PBBs
From the 7 identified azeotropes, water is present in 4 of them.
Also from the binary pairs toluene/water, MeOH/water and p-
xylene/water, the binary ratio is close to unity and this hints at
difficult separation due to low driving force (Jaksland et al., 1995;
Bek-Pedersen and Gani, 2003). Therefore, it would be beneficial to
remove water upstream the process. Using Table 4 and selecting the
process hot-spot, the selected desirable task for removing water is a
separation task and the selected desirable PBBs are PT(PVL), PT(VV)
and PS(VV). These are selected based on the binary ratio of the
molar volumes for toluene/water, MeOH/water and p-xylene/water
which are greater than 1.5 and hints that the separation of water
is possible using a barrier, for example, a pervaporation membrane
(represented by PT(PVL)) or a vapor permeation membrane (repre-
sented by PT(VV)). The final set of PBBs are those found in IT-PBS-1
plus those identified in IT-PBS-2 (this step), that is, a total of 17.
The final set of PBBs are: R, M (vapor, ideal, flow and rectangular),
2phM, C, H, PT(VL), PC(VL), PS(VL), PS(LL), PT(LL), PT(PVL), PT(VV),
PS(VV) and D.
3.10. IT-PBS-3: generation of feasible flowsheet alternatives
3.10.1. Generate feasible SPBs
The total number of SPBs (feasible and infeasible) is calculated
using Eqs. (16)-(17), where nPBBMax is the maximum number of
PBBs within a SPB, nPBB is the total number of PBBs, NSPBmax is
the total number of SPBs, nPBBE , nPBBM , nPBBD are energy PBBs,
mixing PBBs (vapor, ideal, rectangular and flow) and dividing PBB,
respectively and Eq. (16) gives the maximum number of PBBs that
can be combined to form an SPB (Lutze et al., 2013).
nPBBMax = nPBB − (nPBBE − 1) − (nPBBM − 1) − nPBBD (16)

Table 13
Predicted binary azeotropes (ICAS, Gani et al., 1997).

Binary azeotropes Azeotrope type (max-/minimum boiling) Azeotrope phase (homo-/heterogeneous)

Toluene/MeOH Maximum-boiling Homogeneous

Toluene/Water Maximum-boiling Heterogeneous
MeOH/Benzene Maximum-boiling Homogeneous
Water/p-Xylene Maximum-boiling Heterogeneous
Water/o-Xylene Maximum-boiling Heterogeneous
Water/m-Xylene Maximum-boiling Heterogeneous
Water/Benzene Maximum-boiling Heterogeneous
68 N. Anantasarn et al. / Computers and Chemical Engineering 105 (2017) 56–73

Table 14
Partial list of feasible SPBs.

No. SPB Task

1 M Mixing
2 M = 2phM Mixing
3 M=C Mixing + Cooling
4 M=H Mixing + Heating
5 M=R Mixing + Reaction
6 M=R=C Mixing + Reaction + Cooling
7 M=R=H Mixing + Reaction + Heating
8 M = R = C = PC(VL) = PT(VL) Mixing + Reaction + Cooling
9 M = R = H = PC(VL) = PT(VL) Mixing + Reaction + Heating
10 M = R = 2phM = PC(VL) = PT(VL) Mixing + Reaction + Separation
11 M = R = 2phM = PC(VL) = PT(VL) = PS(VL) Mixing + Reaction + Separation
12 M = R = 2phM = PC(VL) = PT(PVL) = PS(VL) Mixing + Reaction + Separation
13 M = R = C = 2phM = PC(VL) = PT(PVL) = PS(VL) Mixing + Reaction + Cooling + Separation
14 M = R = H = 2phM = PC(VL) = PT(PVL) = PS(VL) Mixing + Reaction + Heating + Separation
15 M = R = 2phM = PT(VV) = PS(VV) Mixing + Reaction + Separation
16 M = R = H = 2phM = PT(VV) = PS(VV) Mixing + Reaction + Heating + Separation
17 M = R = 2phM = PC(LL) = PS(LL) Mixing + Reaction + Separation
18 M = R = C = 2phM = PC(LL) = PS(LL) Mixing + Reaction + Cooling + Separation
19 M = R = H = 2phM = PC(LL) = PS(LL) Mixing + Reaction + Heating + Separation
20 M = 2phM = PC(VL) = PT(VL) Mixing + Phase creation
21 M = 2phM = C = 2phM = PC(VL) = PT(VL) Mixing + Cooling + Phase creation
22 M = 2phM = H = 2phM = PC(VL) = PT(VL) Mixing + Heating + Phase creation
23 M = 2phM = PC(VL) = PT(VL) = PS(VL) Mixing + Separation
24 M = C = 2phM = PC(VL) = PT(VL) = PS(VL) Cooling + Separation
25 M = H = 2phM = PC(VL) = PT(VL) = PS(VL) Heating + Separation
26 M = 2phM = PC(VL) = PT(PVL) = PS(VL) Mixing + Separation
27 M = 2phM = PC(VL) = PT(VV) = PS(VV) Mixing + Separation
28 M = 2phM = PT(VV) = PS(VV) Mixing + Separation
29 M = 2phM = PC(LL) = PS(LL) Mixing + Separation
... ... ...
88 D Stream dividing

 
nPPB,Max
(nPBB − 1)!
 Flowsheet alternatives 2–3 (highlighted red in Fig. 11): A reac-
NSPBmax = +1 (17)
(nPBB − k − 1)!k! tor feed, as in the base case design, of a 2:1 mol ratio of toluene
k=1 and MeOH is used. Water is selected to be removed after the reac-
The calculated number of SPBs using Eq. (17) that can be gen- tion and therefore, a basic structure containing either a PT(PVL) or
erated from the total set of 17 PBBs (identified in IT-PBS-2) is PT(PVV) PBB is selected to perform the separation task (S-1). A basic
3,903,817. The inputs to Eq. (16) are nPBB = 17, nPBBE = 2 (C and H), structure containing a PT(VL) PBB is selected for performing the
nPBBM = 4 (vapor, ideal, flow and rectangular) and nPBBD = 1 and separation task for recovery of ethylene (S-2), then methanol (S-3)
nPBB,Max , is 13. From application of the combination rules a total of and toluene (S-4). For isolation of p-xylene from the other xylenes, a
88 SPBs are found feasible. A list of the feasible SPBs assuming 3 reaction task is selected in order to react away the m-xylene. There-
types of mixing (ideal, flow and rectangular) are listed in Table 14. fore, a basic structure containing R = PC = PT = PS PBBs is selected
for performing the reaction-separation task (RS-1). Finally, a basic
structure containing PT(VL) PBB is selected for performing the sep-
3.10.2. Generate task-based superstructure aration task (S-5) for recovery of p-xylene.
The generated task-based superstructure is shown in Fig. 11 and Flowsheet alternatives 4–5 (highlighted blue in Fig. 11): A reac-
the key for Fig. 11 is given in Table 15. The base case is highlighted tor feed, as in the base case design, of a 2:1 mol ratio of toluene
yellow in Fig. 11. and MeOH is used. Water is selected to be removed after the reac-
tion and therefore, a basic structure containing either a PT(PVL)
3.10.3. Identify basic structures and feasible task-based flowsheet or PT(PVV) PBB is selected to perform the separation task (S-1).
alternatives The separation tasks from flowsheet alternative 2–3 (S-1-S-4) con-
The identified task-based flowsheets are shown in Fig. 12 and tain the same initiator SPBs therefore, the basic structures are
the identified basic structures for the phenomena-based flowsheets combined and the tasks are merged (new S-2). In this combined
are given in Table 16. The generation of the flowsheet alternatives separation task, ethylene, unreacted toluene and methanol and,
is explained as follows: xylenes and benzene are recovered. For isolation of p-xylene from
the other xylenes, a reaction task is selected and a basic struc-
ture containing R = PC = PT = PS PBBs is identified for performing the
Table 15 reaction-separation task (RS-1). Finally, a basic structure contain-
Key for the task-based superstructure. ing PT(VL) PBB is selected for performing the separation task (new
Compound Alphabetic label Boiling point (◦ C) S-3) for recovery of p-xylene.
Flowsheet alternatives 5–6 (highlighted green in Fig. 11): The
Ethylene A 169.45
Methanol B 337.95 starting flowsheets are alternatives 4–5. However, the ethylene
Benzene C 353.15 from S-1 is further investigated to be recovered as high purity
Water D 373.15 ethylene. A basic structure containing a PT(VL) PBB is selected for
Toluene E 383.75 performing the separation task for recovery of high purity ethylene.
p-Xylene F 411.45
The remainder of the flowsheets remain the same.
m-Xylene G 412.25
o-Xylene H 417.65
 N. Anantasarn et al. / Computers and Chemical Engineering 105 (2017) 56–73 69

Table 16
Identified basic structures for performing selected tasks.
70 N. Anantasarn et al. / Computers and Chemical Engineering 105 (2017) 56–73

Table 17
Economic and environmental analysis for the production of p-xylene.

Unit-Ops scale Phenomena scale

Performance criteria Base case 1 3 5 7

Product purity (wt%) 99.72 99.70 99.70 99.74 99.74
By-product purity (wt%) 52.74 99.50 59.73 85.78 99.50
Energy Usage (GJ/h) 677.97 566.13 551.18 289.40 327.72
Utility costs ($/h) 1,216.78 828.29 808.06 550.8 559.08
Fobj (Eq. (1)) 0.94 0.90 0.89 0.83 0.84
Number of unit 8 9↑ 7↓ 5↓↓ 7
Carbon Footprint (eq. kg of CO2 ) 1.97 1.51 1.43 0.99 1.00
HTPI (1/LD50) 3.10E-06 2.50E-06 2.49E-06 1.99E-06 2.01E-06
GWP (CO2 eq.) 0.77 0.65 0.63 0.49 0.50
HTC (kg benzene eq.) 3.70E-04 2.50E-04 2.35E-04 2.32E-04 2.34E-04
HTNC (kg toluene eq.) 9.20E-04 6.10E-04 6.03E-04 5.41E-04 5.45E-04

Fig. 11. Task based superstructure for the production of p-xylene. Base case – Yellow, Alternative 2 – Red, Alternative 4 – Blue, Alternative 5 – Green. (For interpretation of
the references to color in this figure legend, the reader is referred to the web version of this article.)

3.10.4. Identify unit operations-based flowsheet alternatives
The basic structures that perform the different identified tasks
are translated into unit operations. The final flowsheet alternatives
are presented in Fig. 12.
In Fig. 12, flowsheet alternatives 2–3 use either a pervapora-
tion membrane or vapor permeation membrane for the removal of
water. In flowsheet alternatives 3–4, also, a pervaporation mem-
brane or vapor permeation membrane is used for removing water
and a divided wall column is used to recover ethylene, toluene and
methanol and, xylenes and benzene. In flowsheet alternatives 5–6,
a membrane and divided wall column is used for removal of water
and recovery of ethylene (together with toluene and methanol and,
xylenes and benzene) and ethylene is further purified to obtain high
purity ethylene.
methanol and, xylenes and benzene therefore, the selected alter-
native is flowsheet alternative 5.
• Flowsheet alternative 6–7: The selected membrane is a vapor
persmeation membrane and a divided wall column is used to
recover ethylene (together with toluene and methanol and,
xylenes and benzene) which is further purified using a combi-
nation of distillation and flash columns therefore, the selected
alternative is flowsheet alternative 7.
The representation of screening from SPBs to flowsheet alterna-
tives are shown in Fig. 13.
3.11. IT-PBS-4: comparison and selection of the more sustainable
alternatives

3.10.5. Model Based Analysis
The flowsheet alternatives are analyzed and screened using the
local constraints (␪1 ) and operational constraints (␪2 ) defined in
step 2.
• Flowsheet alternatives 2–3: The selected membrane is a vapor
permeation membrane (Bolta et al., 2012; Jonquieres et al., 2002)
because the phase of the mixture at the outlet of the reactor is
vapor therefore, the selected alternative is flowsheet alternative
3 base case design. The benefit of flowsheet alternative 7 is that the
• Flowsheet alternative 4–5: The selected membrane is a vapor
persmeation membrane and a divided wall column is found
feassible for recovery of ethylene, unreacted toluene and
The results of the economic and environmental analysis results
are presented in Table 17.
↓ – Reduction in the number of unit operations, ↑ – Increase in
the number of unit operations
Flowsheet alternative 5 gives the best objective function value
while also having the lowest carbon footprint and number of unit
operations. Flowsheet alternative 3 and flowsheet alternative 7 are
also promising alternatives because they also provide a reduction
in the objective function and carbon footprint compared to the
by-product ethylene is also recovered as a value-added product.
Flowsheet alternative 1 also recovers high purity ethylene but has
a higher number of unit operations but should not be ruled out as an
N. Anantasarn et al. / Computers and Chemical Engineering 105 (2017) 56–73 71

Fig. 12. Translated flowsheet alternatives.

72 N. Anantasarn et al. / Computers and Chemical Engineering 105 (2017) 56–73
Fig. 13. Screening of generated SPBs and flowsheet alternatives gnerated from
phenomen-based process intensifcation.
alternative because the base case did not investigate the recovery
of high purity ethylene.
4. Conclusion
In conclusion, the computer-aided work-flow for performing
process synthesis-intensification has been presented. The frame-
work operates at 3 different scales, the unit operations scale, task
scale and phenomena scale and it has been shown that novel, more
sustainable flowsheet designs are generated by performing process
intensification at the lowest (phenomena) scale. The framework has
been applied to the production of a well-known bulk-chemical,
para-xylene and 4 more sustainable flowsheet alternatives have
been generated. The alternatives include hybrid/intensified unit
operations that provide reductions in operation cost, carbon foot-
prints and environmental impacts.
The next step would be the expansion of the framework in
order to include control issues. Mansouri et al. (2016) has devel-
oped a method for integrated process design-control of intensified
processes. In the integrated method, the control step is the verifi-
cation step because in the synthesis-intensification stages, design
decisions are made based on the process maximum driving forces
related to reaction (Hamid et al., 2010), separation (Bek-Pedersen
and Gani, 2003) and reaction-separation (Daza et al., 2003). There-
fore, in the future work, the framework will be expanded to include
a control stage in order to investigate process dynamics.
Acknowledgements
The work was supported by the Petroleum and Petrochemical
College (PPC), Chulalongkorn University, Thailand and Sustainable
Product Process Engineering, Evaluation and Design (SPEED) of
Chemical and Biochemical Engineering Department, Technical Uni-
versity of Denmark, Denmark.
Appendix A. Supplementary data
Supplementary data associated with this article can be found,
in the online version, at http://dx.doi.org/10.1016/j.compchemeng.
2017.02.025.
References
Ashraf, M.T., Chebbi, R., Darwish, N.A., 2013. Process of p-xylene production by
highly selective methylation of toluene. Ind. Eng. Chem Res. 52 (38),
13730–13737.
Asprion, N., Kaibel, G., 2010. Dividing wall columns: fundamentals and recent
advances. Chem. Eng. Proc. Process Intensif. 49 (2), 139–146.
Babi, D.K., Gani, R., 2014. Hybrid distillation schemes: design, analysis, and
application. In: Gorak, A., Sorensen, E. (Eds.), Distillation. Elsevier, London, pp.
357–381.
Babi, D.K., Lutze, P., Woodley, J.M., Gani, R., 2014. A process
synthesis-intensification framework for the development of sustainable
membrane-based operations. Chem. Eng. Process. Process Intensif. 86,
173–195.
Babi, D.K., Holtbruegge, J., Lutze, P., Gorak, A., Woodley, J.M., Gani, R., 2015.
Sustainable process synthesis–intensification. Comput. Chem. Eng. 81,
218–244.
Bek-Pedersen, E., Gani, R., 2003. Design and synthesis of distillation systems using
a driving-force-based approach. Chem. Eng. Process. 43, 251–262.
Bessling, B., Schembecker, G., Simmrock, K.H., 1997. Design of processes with
reactive distillation line diagrams. Ind. Eng. Chem. Res. 36 (8), 3032–3042.
Bolta, B., Hoang, M., Xie, M., 2012. A review of water recovery by vapour
permeation through membranes. Water Res. 46 (2), 259–266.
Bravo-Bravo, C., Segovia-Hernández, J.G., Gutiérrez-Antonio, C., Durán, A.L.,
Bonilla-Petriciolet, A., Briones-Ramírez, A., 2010. Extractive dividing wall
column: design and optimization. Ind. Eng. Chem. Res. 49 (8), 3672–3688.
Breen, J.P., Burch, R., Kulkarni, M., Collier, P., Golunski, S., 2005. Enhanced
para-Xylene selectivity in the toluene alkylation reaction at ultralow contact
time. JACS 127 (14), 5020–5021.
Caballero, J.A., Grossmann, I.E., 2004. Design of distillation sequences: from
conventional to fully thermally coupled distillation systems. Comput. Chem.
Eng. 28 (11), 2307–2329.
Carvalho, A., Matos, H.A., Gani, R., 2013. SustainPro—a tool for systematic process
analysis: generation and evaluation of sustainable design alternatives.
Comput. Chem. Eng. 50, 8–27.
Daza, O.S., Pérez-Cisneros, E.S., Bek-Pedersen, E., Gani, R., 2003. Graphical and
stage-to- stage methods for reactive distillation column design. AIChE J. 49
(11), 2822–2841.
Douglas, J.M., 1985. A hierarchical decision procedure for process synthesis. AIChE
J. 31 (3), 353–362.
Dursch, T., Khalil, R., Khine, A., Mutahi, F., 2009. Toluene Methylation to
Para-Xxylene. Senior Design Reports. University of Pennsylvania, Department
of Chemical & Biomolecular Engineering.
Fedorova, M., Sin, G., Gani, R., 2014. Computer-aided template for model reuse
development and maintenance. Comput. Aided Chem. Eng. 33, 817–822.
Gani, R., Hytoft, G., Jaksland, C., Jensen, A.K., 1997. An integrated computer aided
system for integrated design of chemical processes. Comput. Chem. Eng. 21
(10), 1135–1146.
Hamid, M.K.A., Sin, G., Gani, R., 2010. Integration of process design and controller
design for chemical processes using model-based methodology. Comput.
Chem. Eng. 34 (5), 683–699.
Harper, P.M., Gani, R., 2000. A multi-step and multi-level approach for computer
aided molecular design. Comput. Chem. Eng. 24 (2–7), 677–683.
Heitzig, M., Gregson, C., Sin, G., Gani, R., 2011. Application of computer-aided
multi-scale modelling framework −aerosol case study. Comput. Aided Chem.
Eng. 29, 16–20.
Holtbruegge, J., Kuhlmann, H., Lutze, P., 2014. Process analysis and economic
optimization of intensified process alternatives for simultaneous industrial
scale production of dimethyl carbonate and propylene glycol. Chem. Eng. Res.
Des. 93, 411–431.
Jaksland, C., Gani, R., Lein, K.M., 1995. Separation process design and synthesis
based on thermodynamic insights. Chem. Eng. Sci. 50 (3), 511.
Jonquieres, A., Clement, R., Lochon, P., 2002. Permeability of block copolymers to
vapors and liquids. Prog. Polym. Sci. 27, 1803–1877.
Kalakul, S., Malakul, P., Siemanond, K., Gani, R., 2014. Integration of life cycle
assessment software with tools for economic and sustainability analyses and
process simulation for sustainable process design. J. Cleaner Prod. 71, 98–109.
Karunanithi, A.T., Achenie, L.E.K., 2005. A new decomposition-based
computer-aided molecular/mixture design methodology for the design of
optimal solvents and solvent mixtures. Ind. Eng. Chem. Res. 44 (13), 4785.
Kiss, A., Pragt, H., van Strien, C., 2007. Computer aided chemical engineering. 17th
European Symposium on Computer Aided Process Engineering vol. 24,
467–472.
Kongpanna, P., Pavarajarn, V., Gani, R., Assabumrungrat, S., 2015. Techno-economic
evaluation of different CO2-based processes for dimethyl carbonate
production. Chem. Eng. Res. Des. 93, 495–510.
Leimbrink, M., Kunze, A.-K., Hellmann, D., Górak, A., Skiborowski, M., 2015.
Conceptual design of post-combustion CO2 capture processes – packed
columns and membrane technologies. Comput. Aided Chem. Eng. 37,
1223–1228.
Lutze, P., Román-Martinez, A., Woodley, J.M., Gani, R., 2012. A systematic synthesis
and design methodology to achieve process intensification in (bio) chemical
processes. Comput. Chem. Eng. 36, 189–207.
Lutze, P., Babi, D.K., Woodley, J.M., Gani, R., 2013. Phenomena based methodology
for process synthesis incorporating process intensification. Ind. Eng. Chem.
Res. 52 (22), 7127–7144.
Madenoor Ramapriya, G., Tawarmalani, M., Agrawal, R., 2014. Thermal coupling
links to liquid-only transfer streams: a path for new dividing wall columns.
AIChE J. 60 (8), 2949–2961.
Mansouri, S.S., Huusom, J.K., Gani, R., Sales-Cruz, M., 2016. Systematic integrated
process design and control of binary element reactive distillation processes.
AIChE J. 62 (9), 3137–3154.
Okoli, C.O., Adams, T.A., 2015. Design of dividing wall columns for butanol recovery
in a thermochemical biomass to butanol process. Chem. Eng. Process. Process
Intensif. 95, 302–316.
 N. Anantasarn et al. / Computers and Chemical Engineering 105 (2017) 56–73
Papadakis, E., Tula, A.K., Anantpinijwtna, A., Babi, D.K., Gani, R., 2016. Sustainable
chemical process development through an integrated framework. Comput.
Aided Chem. Eng. 38, 841–846.
Papalexandri, K.P., Pistikopoulos, E.N., 1994. A multiperiod MINLP model for the
synthesis of flexible heat and mass exchange networks. Comput. Chem Eng. 18
(11–12), 1125–1139.
Peschel, A., Jörke, A., Freund, H., Sundmacher, K., 2012. Model-based development
of optimal reaction concepts for plant wide process intensification. Comput.
Aided Chem. Eng. 31, 150–154.
Peters, M.S., Timmerhaus, K.D., West, R.E., 2003. Plant Design and Economics for
Chemical Engineers. McGraw Hill, New York.
Rong, B.G., Kolehmainen, E., Turunen, I., 2008. Methodology of conceptual process
synthesis for process intensification. Comput. Aided Chem. Eng. 25, 283.
Saengwirun, P., 2011. ECON: a Software for Cost Calculation and Economic
Analysis. The Petroleum and Petrochemical College, Chulalongkorn University,
MSc-thesis.
Satelo, J.L., Uguina, M.A., Valverde, J.L., Serrano, D.P., 1993. Kinetics of toluene
alkylation with methanol over Mg-modified ZSM-5. Ind. Eng. Chem. Res. 32,
2548–2554.
Seifert, T., Sievers, S., Bramsiepe, C., Schembecker, G., 2012. Small scale, modular
and continuous: a new approach in plant design. Chem. Eng. Process. Process
Intensif. 52, 140–150.
73
Siirola, J., 1996. Strategic process synthesis: advances in the hierarchical approach.
Comput. Chem. Eng. 20, S1637–S1643.
Song, W., Venimadhavan, G., Manning, J.M., Malone, M.F., Doherty, M.F., 1998.
Measurement of residue curve maps and heterogeneous kinetics in methyl
acetate synthesis. Ind. Eng. Chem. Res. 37 (5), 1917–1928.
Song, C., Kansha, Y., Ishizuka, M., Fu, Q., Tsutsumi, A., 2015. Conceptual design of a
novel pressure swing CO2 adsorption process based on self-heat recuperation
technology. Chem. Eng. Process. Process Intensif. 94, 20–28.
Stankiewicz, A., Drinkenburg, A.A.H., 2004. Process intensification: history,
philosophy, principles. In: Stankiewicz, A., Moulijn, J.A. (Eds.), Re-engineering
the Chemical Processing Plant: Process Intensification. CRC Press llc, Florida,
pp. 14–40.
Stankiewicz, A., Moulijn, J.A., 2000. Process intensification: transforming chemical
engineering. Chem. Eng. Prog. 96, 22–33.
Tula, A.K., Eden, M.R., Gani, R., 2015. Process synthesis, design and analysis using a
process-group contribution method. Comput. Chem. Eng. 81, 245–325.
Wantanachaisaeng, P., O’Neil, K., 2010. Capturing Opportunities for Para-xylene
Production.
Zhang, L., Babi, D.K., Gani, R., 2016. New vistas in chemical product and process
design. Annu. Rev. Chem. Biomol. Eng. 7, 557–582.