Chemical Engineering and Processing 84 (2014) 113

Contents lists available at ScienceDirect

Chemical Engineering and Processing:

Process Intensication
journal homepage: www.elsevier.com/locate/cep

Local and global process intensication

Jean-Francois Portha a,b , Laurent Falk a,b, , Jean-Marc Commenge a,b
a
b

Universit de Lorraine, Laboratoire Ractions et Gnie des Procds, UMR 7274, Nancy F-54001, France
CNRS, Laboratoire Ractions et Gnie des Procds, UMR 7274, Nancy F-54001, France

a r t i c l e

i n f o

Article history:
Received 19 December 2013
Received in revised form 19 April 2014
Accepted 5 May 2014
Available online 13 May 2014
Keywords:
Process intensication
Process system engineering
Process retrot
Debottlenecking

a b s t r a c t
The present paper aims at proposing a complementary view of process intensication (PI) based on the
concepts of local intensication and global intensication. Local intensication is dened here as the
classical approach of PI based on the use of techniques and methods for the drastic improvement of the
efciency of a single unit or device. Some examples are given to illustrate that local process intensication
presents several limitations when compared to holistic overall process-based intensication, named
global intensication. Indeed, when PI focuses on single units (reactors, separators, hybrid separators,
etc.), the strong interactions among all units within the process are ignored and the impact of local
intensication of a single unit can be very limited, resulting in weak improvement of the whole process.
This paper identies that process intensication is broader than technical improvement of devices or
processes and has to consider several drivers such as economics, safety, eco-efciency and sustainability
to fulll the key objectives in designing new plants and retrotting existing units.
2014 Elsevier B.V. All rights reserved.

1. Introduction
Process intensication (PI), formally regarded as process
improvement strategy, consists, according to Stankiewicz and
Moulijn [1], in novel equipment, processing techniques, and process development methods that, compared to conventional ones,
offer substantial improvements in (bio) chemical manufacturing
and processing. It is related to new and innovative technologies
which replace large expensive and energy-intensive equipment
by devices which are smaller, less costly and more efcient [24].
Among the different strategies, a well-established approach is the
design of multifunctional devices, which merge several unit operations, e.g. reaction and separation into one apparatus [2]. However,
recent approaches in PI predominantly focus on equipment and
more specically on microstructured technologies, providing a high
surface-to-volume ratio, hence increasing mass and heat transfer
by several orders of magnitude [5]. The revolution-like manners
in which microtechnology develops (and as which it is sometimes
praised) manifest themselves in the vast number of patents in this

Abbreviations: API, active pharmaceutical ingredient; CSTR, continuous stirred

tank reactor; GHG, greenhouse gases; LCA, life cycle assessment; PI, process
intensication; PFR, plug ow reactor; PR, process retrot; PSE, process system
engineering; ROI, return on investment time; RSR, reactor separator recycle.
Corresponding author at: Universit de Lorraine, Laboratoire Ractions et Gnie
des Procds, UMR 7274, Nancy F-54001, France. Tel.: +33 383175094.
E-mail address: laurent.falk@univ-lorraine.fr (L. Falk).
http://dx.doi.org/10.1016/j.cep.2014.05.002
0255-2701/ 2014 Elsevier B.V. All rights reserved.

eld of research. Every year, numerous patents are reported for

microtechnology-related areas. Focusing on microreactors, more
than 1500 patent publications in different elds of chemical applications already exist, annually rising in number by more than 250
new patents [6]. Then, from the numerous studies published in the
literature, it is clear that PI focuses on the equipment improvement,
which can be dened as a local approach of process improvement
strategy.
On the other hand, process system engineering (PSE), according
to Grossmann and Westerberg [7], aims at improving decisionmaking for the creation and operation of the chemical supply chain,
which deals with the discovery, design, manufacturing, and distribution of chemical products. From this denition, PSE can be
considered as a global approach of process improvement strategy.
Recently, Moulijn et al. [8] felt that it was timely to attempt to
better dene the place of PI in relation with other chemical engineering disciplines, such as PSE. Based on the preliminary approach
by Grossmann and Westerberg [7] and Marquardt et al. [9], they
proposed to dene PI in conjunction with PSE. The focus and action
of PSE take place along the product creation chain, as a top-down
approach from the enterprise to the molecules, while the focus and
action of PI are on chemical engineering aspects of the process units
separately. PI has a more creative than integrating character and
primarily aims at higher efciency of individual steps in that chain,
for instance by offering new mechanisms, materials, and structural
building blocks for process synthesis.
In addition, the scales considered are different; PSE focuses less
on the scale of molecules, sites and (nano-) structure, whereas PI

J.-F. Portha et al. / Chemical Engineering and Processing 84 (2014) 113

explicitly includes this level but often pays less attention to the
highest levels. It is clear that PI has consequences for the longitudinal action of PSE; for instance, development and application of
a reactive separation can inuence the PSE over the whole chain,
from molecule to site, if not to enterprise.
The improvement of processes must also be examined in the
context that a major portion of the chemical industry has matured.
Most chemical plants were built at a time when prot margins could
be kept large and thus were not typically designed to be the most
efcient from an energy and raw material perspective. Nowadays,
competitive pressures have greatly increased the need for more
efcient processes claiming for the redesign and the modernization of existing facilities. Grossmann et al. [10] estimated that, in
the end of the 80s, 7080% of all process design projects in the western countries dealt with the redesign, i.e. retrot design of existing
facilities.
Also, the successful commercialization of specialty chemicals
requires the ability to redesign processes quickly, to respond to
changes in new technology and to the short life cycle of new products. According to Grossmann and Westerberg, the development
of retrot design strategies is a more difcult problem than the
design of new processes because it includes a far greater number of
alternatives than the grassroots problem, due to the need to evaluate and use existing equipment. The nature of the retrot problem
is by nature complex and multidimensional. They propose a list of
technical objectives as the increase of the throughput by debottlenecking and by higher conversion of feedstocks, the processing of a
new feedstock and the improvement of the quality of the product. Sustainability is also considered by the increase of process
safety, the reduction of the environmental impact of an existing process, the reduction of energy input per unit of production,
and the higher operability of the process (exibility, controllability).
Concerning economics and the major constraints of process
retrot (PR), several approaches can be developed. The change of
the operating conditions of the process can enable to keep the
same implemented equipment which is obviously the least costly
in terms of investment. The change of architecture of the process by changing the piping that connects the devices is another
alternative. For example, with respect to the cost of purchasing
a new column, repiping typically incurs very modest costs. It is
also possible to keep the process owsheet intact but to change
the equipment sizing, sometimes in ways that the external physical dimensions of the equipment are not altered. Such changes
could include installation of new tube bundles inside existing
heat-exchanger shells, closer packed trays or even packing inside
columns. And nally, the last approach is the addition of new equipment to reach the objectives.
From the denition proposed by Grossmann et al. [10], it
is obvious that PR shares numerous keywords (improvement of
conversion and yield, reduction of energy consumption and environmental impact, safety considerations, etc.) with process system
engineering and process intensication. This large overlapping
between those three concepts shows also that PR is concerned by
local and global intensication and that synergies between them
may exist.
However, as stated by El-Halwagi and co-workers [11], there are
some limitations in most of the previous works focused on unitbased intensication when compared to holistic overall processbased intensication.
El-Halwagi and co-workers identied process integration as a
holistic and systematic framework for intensication where, however, process intensication has a broader scope. They dened
two main classes for intensication: single-unit intensication
and plant intensication. Unit intensication refers to the previous denition of process intensication. On the other hand, plant

intensication focuses on the improvement of the global process:

maximize the throughput, minimize inventory, or minimize utility
materials and feedstock. In case of plant intensication, units that
will be intensied are not pre-specied and more than one unit can
be intensied simultaneously.
The present paper aims at proposing a complementary view of
process intensication based on the concepts of local intensication and global intensication. Local intensication is dened here
as the classical approach of PI based on the use of techniques and
methods for the drastic improvement of the efciency of a single unit or device (reactors, separators, mixers, exchangers, etc.),
to overcome specic limitations that can be related to thermodynamics, kinetics, heat or mass transfer and energy supply. It
mainly focuses on the technical improvement of the performances
of equipment but the interactions among all units within the process are ignored and the impact of intensifying a single unit on the
rest of the process is not considered.
First, global (or overall) intensication has a more general
view on the whole process, considering rst a multi-dimensional
approach consisting in the simultaneous improvement of several
units. The process is improved by inventory and utility minimization and by throughput maximization. Process intensication is
achieved using the classical methods of local intensication and
heat and mass integrations meaning that a complex owsheet or
architecture should be designed to increase the overall process efciency. The impact of a local change will have an effect on the entire
process due to the strong interactions between units and should
therefore be studied at the whole process scale.
Secondly, global intensication possesses a multi-dimensional
aspect where different drivers (economic, safety, eco-efciency and
sustainability) are included in the strategy. There are some limitations in most of the previous PI works as they focused mainly
on technical drivers but did not develop an holistic view, omitting
to include the different drivers. This is not the case with retrot
design that recently included various methods to evaluate and
reduce the environmental impact of chemical processes. Sun et al.
[12] proposed the formulation of a multi-objective optimization
problem to determine sustainable chemical process designs taking into account economic, environmental and societal aspects.
El-Halwagi and co-workers [13] presented a multi-objective optimization procedure for the recycle and reuse networks including
the environmental implications of the discharged wastes using
life-cycle assessment. More recently, they developed [11] a rst
attempt to couple an intensication strategy with a multi-objective
optimization problem, but the mathematical functions used to represent intensication lacked for realism. In a recent study, Gani and
co-workers [14] presented the development of a software tool and
its application to chemical processes, based on the implementation
of an extended systematic methodology for sustainable process
design.
Further work is still needed to combine process retrot, process intensication and process system engineering to dene an
intensication strategy which takes into account both the local
approach of PI and the global approach of PSE considering the
whole process by multi-objective optimization to propose sustainable and intensive chemical process designs. The strategy
should answer the following questions. Which equipment should
be intensied in a process? What is the impact of local process
intensication of a device on the overall process performance?
How should new process architecture be achieved? Which optimization criteria should be chosen? What are the criteria of safer
processes?
In the present paper, some examples will be given to illustrate that the classical approach of process intensication based on
single-unit intensication presents several limitations when compared to holistic overall process-based intensication.

J.-F. Portha et al. / Chemical Engineering and Processing 84 (2014) 113

Step 1

Step 2

Step 3

Step 4

Table 1
Comparison of two intensication strategies for a unique extensity in a cascade of
two reactors.

Step 5

Fig. 1. Linear process composed of 5 steps, units or devices in series.

2. Multi-units aspects: impact of local intensication on

the overall efciency of a process
In most studies dedicated to local intensication, the objective is
to improve the reactor or the separator without taking into account
the consequences on the whole process. This section illustrates
the impact of improving single equipment on the whole process
performance.

Base case
Intensication of the rst unit
Intensication of the second unit

(1)

In this paragraph, N process units or devices Ei are considered

in series (1 i N). Each one is characterized by an efciency  and
G0 denotes the outlet extensity ux of equipment Ei (Fig. 1):
i =

Gi1 Gi
G0

(2)

The global efciency of this serial arrangement of devices is

obviously a combination of the local efciencies of all the devices or
steps. The relationship between the local and the global efciency
is a key information to select the local step or device which has to
be intensied to improve the performance of the global process.
For a complex process, this relationship is not straightforward, but
in some specic cases, the relationship can be analytically derived
and general rules can be identied. Two cases are selected here,
for process units in series, with respect to the type of extensity
considered. The rst case arises when the same extensity is considered along the process. For example, when several continuous
stirred tank reactors in series are used to perform the conversion
of a specic molecule, the extensity ux is the molar ow rate of
the reactant and the efciency corresponds to the conversion rate
of that reactant. Typical cases concern depollution where the conversion of the pollutant must be maximal and the conversion of a
specic reactant, whose concentrations in the nal product must
be minimized to get the right end-use specications.
The global efciency can be dened by considering inefciencies. Indeed, the overall inefciency is equal to the product of the
local inefciencies:
1 g =

GN
G1 G2
GN
=

= (1 1 )(1 2 ). . .(1 N )
G0
G0 G1
GN1

Unit 2
2 (%)

Overall
efciency
g (%)

10
20
10

50
50
60

55
60
64

N


ng = 1

(1 i )

(4)

Let us consider a batch or a continuous process unit E with

several steps or units. In each of them a specic operation (i.e.
transformation or transfer) is carried out. The objective here is to
establish the relationship between the efciency of one specic
device (for a continuous process) or step (for a discontinuous process) and the efciency of the whole process in different conditions
and to derive some intensication strategies. The concept of extensity (for example the number of moles) is essential because the
case study is different when identical or different extensities are
considered in the analysis.
The general denition of the efciency of a unit or a step corresponds to the ratio of the transformed or transferred extensity ux
to the maximal extensity ux which can be transformed or transferred in this unit or operation step. The denition is identical for an
entire process. The efciency  is given by the following equation
where G0 is the inlet extensity ux and G1 is the outlet extensity
ux:
G0 G1
G0

Unit 1
1 (%)

As a consequence, the expression of the overall efciency is:

2.1. Equipment efciency versus process efciency

=

(3)

A numerical example is given in Table 1 for a chemical reaction.

The objective is to convert the reactant by using two continuous
stirred tank reactors in series. The considered extensity and the
equipment efciency correspond here to reactant ow rate and
reactant conversion, respectively. In this example, the rst and the
second units have an efciency of 10% and 50% (base case). Which
unit should be improved uppermost? Intuitively, the intensication
of the step with the lowest efciency appears as the most promising strategy in order to achieve the highest overall benet. But the
results show that, when the local efciency of one step is increased
by 10% while keeping the efciency of the other step constant, the
step with the highest efciency should be intensied rst in order
to reach the highest global efciency.
The second case arises when different extensities are considered along the process, as for instance the conversion of several
molecules in a multiple-steps reaction scheme A B C. In this
case, the overall efciency is simply equal to the product of the
local efciencies:
g =

N


i

(5)

A two-step continuous process is considered as an example presented in Table 2. In the rst stage, the reaction (a) is performed,
whereas reaction (b) is carried out in the second stage. Both stages
are not only reactors but can include other units, such as separation
units. Extensities and denitions of efciencies for each stage are
given in Table 2 with a numerical example showing that the global
efciency is equal to the product of the local efciencies.
This simple illustration shows that, depending on the considered
extensity, the intensication strategy, i.e. the selection of the step
or unit which must be intensied to improve the performance of the
whole process, can be very different. In Table 2, the most promising
strategy in order to achieve the highest overall efciency consists
in intensifying the stage having the lowest efciency contrary to
the previous example in Table 1. The previous relations show that
the global efciency of a process is also a function of the number
of steps or units in the process. Then, an intensication strategy
would also be the reduction of the number of steps in the process.
2.2. Reduction of the number of steps
The three main drivers for commercial production of chemicals
are overall yield, productivity and cost. As indicated previously, the
overall yield of a process is a function of the local efciencies and
the impact of efciency improvement then depends on the chosen step. In ne chemistry, the productivity of a chemical route
mainly depends on the labor cost; indeed, an intense labor cost
occurs when many procedures are performed manually by workers. The transition from a batch to a continuous process or the use
of advanced process control may be solutions in order to reduce

J.-F. Portha et al. / Chemical Engineering and Processing 84 (2014) 113

Table 2
Comparison of two intensication strategies for different extensities in a two stages process.
Stage 1

Stage 2

Whole process

Reaction performed
Extensity ux (mol/s)
Efciency

A B (a)
FA
F
1 = 1 A0

B C (b)
FB
F
2 = 1 B0

A C (c)
FC
F
g = 1 C0

Molar ow rates (base case)

FA0 = 100 mol/s

1
50%
60%
50%

Base case
Intensication of the rst stage
Intensication of the second stage

the labor cost. The cost driver is mainly dependent on the chemical
route. By a rough estimation, the decrease of manufacturing costs
is proportional to the reduction of the number of synthesis steps.
As mentioned previously, the other driver is logically the overall
yield. However, it is important to point out that, economically, an
increase of the yield for one individual step does not have the same
impact whether the step is located close to the nal product or at
the beginning of the synthesis route [15]. This also explains why,
very often, chemical reactors operate with higher dilution conditions when they are closer to the nal product. Higher dilution
softens reaction conditions and often favors selectivity and yield
with respect to throughput. Conversely, at the beginning of the
synthesis route the costs of chemicals are generally low, and the
throughput/productivity becomes the driving force.
The example of chemical synthesis of an intermediate product
through a multistage organometallic reaction followed by a workup step is considered here and has been studied by Roberge et al.
[16]. Conventional stirred tank reactors, limited by heat and mass
transfer, must very often operate with diluted reactants in order
to maximize the selectivity. But these diluted conditions are also
very detrimental to the reaction time which also contributes to the
degradation of selectivity. Therefore, protection and de-protection
steps are often used in ne chemical routes, which are considered
as additional steps of the process. The use of intensied reactors
(micro heat exchanger-reactors) can be used to overcome this bottleneck. In the study of Roberge, different scenarios are studied and
compared for a campaign producing ve tons of the intermediate.
Table 3 presents the reduction of the synthesis route from seven to
ve chemical steps. Batch reactors, where the slowest reaction is
the coupling reaction (bottleneck), have been replaced by a continuous micro-reactor. In this perspective, micro-reactor technology
may have a signicant impact by allowing the decrease of residence
time and the use of a higher temperature by safely performing hazardous reactions. As the workup operations remain unchanged in
both cases, the bottleneck of the whole process is then no longer
the coupling reaction, but the downstream step. Results show that
the intensied process enables to increase both the total throughput and the global yield, and to reduce the number of operators
resulting in higher economical performances of 10%.
The yield has a larger effect on cost than throughput which
is explained by the large cost contribution from the starting
material since the intermediate product lies closer to the active
Table 3
Overview of assumptions and gain for two scenarios in commercial production [16].

Campaign size (tons)

Batch assets (6 m3 )
CAPEX (M$)
Operators
Throughput (kg/min)
Bottleneck
Gain in yield (%)
Economical gain (%)

Batch process

Continuous and batch

5
7
0
3.5
1.7
Coupling reaction

5
5
<1
2.8
2.1
Distillation
5
10

FB0 = 50 mol/s
2
25%
25%
35%

FC = 12.5 mol/s
g = 1 2
12.5%
15%
17.5%

pharmaceutical ingredient (API) than the raw materials. But for an

intermediate at the beginning of the synthesis sequence, this situation may be reversed and the throughput would become the critical
cost factor of the process. In this particular case, distillation is the
bottleneck that would limit the throughput and minimize the effect
of reactors intensication, showing that PI requires an overall view
of the process.

2.3. Reactionseparation with recycling

In most chemical processes, reactors are sequenced with separators where the desired product is removed from the unconverted
reactant that is recycled back to the reactor. The reactor-separatorrecycle (RSR) system presents an optimal operating point between
the two extrema of a highly efcient reactor coupled with a poorly
efcient separator and reversely. The optimal design strategy of RSR
congurations has been extensively studied, as the superstructurebased approach [17] and the theory of the attainable region
[1820]. Kokossis and Floudas proposed a general superstructure
for isothermal system divided into a reactor network (isothermal
CSTR or PFR in series, in parallel or in series-parallel) and a separator network (distillations column) that are both interconnected
(recycles from the separators back to the reactors for example).
The superstructure should account for the entire potential interconnections between the equipment. They use economic (maximal
prot. . .) and technical (yield. . .) objective functions. Glasser and
Hildebrandt developed the graphical concept of attainable region
that consists in combining a curve representing the characteristic
equation (depending on the RSR system) with the attainable region
boundary curve to nd the optimal operating point. The objective functions are based on the maximum yield. Matallana et al.
[21] have recently studied the dynamic aspects of a RSR system.
They compared results obtained with an economic objective function and an operability-based one. Results highlight that a conict
between both objectives occurs. The reactor corresponding to the
optimal economic design presents a smaller volume and a higher
temperature than the one resulting from operability design. This
reactor operates close to the boundaries of the domain of attraction
meaning that the reactor is more sensitive to disturbances. In the
following example, the question of intensication of RSR system
will be addressed.

2.3.1. Global intensication strategy of a reactionseparation

with recycling
In the present illustration, apparatuses are only characterized
by their efciency. The objective is to identify reactor and separator efciencies that will maximize the overall process performance
without taking into account economic aspects. A RSR system with
an isothermal reactor is considered and presented in Fig. 2. The
conguration where the recycled stream is obtained at the top of
the separator device (not represented here), is strictly equivalent
considering the following developments.

G3

The reactor (equipment 1) efciency (i.e. conversion) is given by

the following expression:
1 = 1

G1
G0 + G3

(8)

The overall system efciency is represented with respect to local

efciencies in Fig. 3. When the separator exhibits zero efciency,
the overall efciency is, as expected, equal to the reactor efciency.
It can be noticed that, for a given overall efciency, several couples
of efciencies (reactorseparator) are solutions of Eq. (8). Indeed, a
high overall efciency can be obtained either with a low separator
efciency and a high reactor efciency, or reversely a high separator efciency and a low reactor efciency. However, in most cases,
high global efciency cannot be obtained at low reactor conversion
because the reaction product must be generated rst.
0.99

.43
000..2
1
05
00.

0. 9 0.95
0. 0.7
0. 8
5

0.9

0.95

0.99
99
0.

0.9

0.
95

0.
6

0 01

0.8

0.
9

0.8

0.
7

0.5
0.4

0.95

0.9

0.7

0.8

0.6

0 01

0.4

0.99

0.2

0.3

0. 1
0.05

Separator efficiency

0.7

0.5

0.3
0.4

0.3

0.4
0.5
0.6
Reactor efficiency

0.7

0.95

0.2

0.8

0.99

0.7

0.1

0.8

0.6

0.05

0.1

0.2

0.3

0.2
0.1

0.8

0. 5

0.3

0.4

0.
7

0. 5

0.7

0.9

0.3

0. 6
0.8

0.5
0.6

0. 7

0.7

0.3

0.8

0.9

Reactor intensification recommended

0.9

0. 95

0.9

0. 95

0.1

0.2

0.3

0.4
0.5
0.6
Reactor efficiency

0.95

0.7

0.8

0.9

Fig. 4. Preferential intensication strategies to be applied to the reactor or separator

with respect to their respective efciencies (dotted lines present the values of the
elasticity of these strategies).

(7)

1
1 2 (1 1 )

0.5

8
0.

0.1

G3
G2
=1
G1
G1

0.6

1. 5

(6)

The overall process efciency g can be calculated with respect

to the two previous local efciencies and by using a material balance on the separator. The following expression can be derived:
g

5
0.9

0.5
0.4

0.2

0.2

where the sum G0 + G3 represents the extensity ux at reactor inlet

and G1 the extensity ux at reactor outlet. The separator (equipment 2) efciency is dened as the ratio between the recycled
extensity ux G3 over the extensity ux G1 at the separator inlet.
The separator efciency is given by the expression:
2 =

0.6

5
0.9

Fig. 2. Representation of an isothermal reactorseparator system with recycling.

Separator efficiency

0.7

0.1

1
Separator
intensification
recommended
1

0.4

G0

0.8

.5

5
4
3

0.
6

G1

10

0.
9

0.9

G2

0.9
5

J.-F. Portha et al. / Chemical Engineering and Processing 84 (2014) 113

0.9

Fig. 3. Overall RSR efciency as a function of reactor (equipment 1) and separator

(equipment 2) efciencies.

The relative impact of unit intensication on the whole process

efciency should then be quantied. The study can be performed by
using a sensitivity analysis as for instance by calculating the elasticity of the overall process efciency. The elasticity with respect to
the efciency of unit i represents the relative intensication impact
of unit i on the global process efciency, on a percentage scale: an
elasticity equal to 0.5 means that a relative 1% increase in unit efciency i will induce a relative 0.5% increase in the overall process
efciency. Expression of elasticity e/i is given by the following
equation:
e/i = lim

/

 i /i

i d
%
=
 di
%i

(9)

The calculations of reactor elasticity and separator elasticity give

some additional information. Using the above relations, these elasticities can be calculated with respect to each equipment, enabling
to build elasticity maps, similar to Fig. 3. These maps indicate the
relative impact of local intensication strategies that would either
focus on the reactor or on the separator. Then, by overlapping
two maps representing both elasticities, it is possible to establish
a global map with two areas: one area where the reactor is the
limiting equipment and one area where the separator is the limiting
equipment toward overall process efciency. These zones are presented in Fig. 4 where the border between both areas is drawn as
a thick black dotted line. Thin dashed lines represent the iso-value
curves of the elasticity of the overall efciency with respect to each
strategy and the units efciencies.
This gure illustrates that the choice of intensication strategies, even for this simple units layout, is not intuitive and can easily
yield to non-optimal decisions. Whatever the separator efciency
below 50% (bottom half of the gure), strategies focusing on the
reactor intensication should be favored since they have a more
signicant impact on the overall efciency. In this area, elasticity
values are above 0.5, indicating that a 10% relative increase in the
reactor efciency will yield to a minimum 5% increase in the overall efciency. For most layouts, the elasticity is below unity, which
indicates the absence of synergetic effects between the units. On
the contrary, in the upper left-hand side corner of the map that corresponds to highly efcient separators coupled to low-conversion
reactors, separator intensication strategies are favorable and elasticity may reach values far above unity. In this area, a small increase

J.-F. Portha et al. / Chemical Engineering and Processing 84 (2014) 113

of unit operations into multifunctional process units, respectively.

In the examples presented here, it is shown that simple changes
of the owsheet, as recycling or heat and mass integration can be
used to intensify a process, even with conventional non-intensied
devices.

106

Specific energy consumption

(Joule / mole produced)

difficult separation =1.5

gain 40 %

3.1. Reactant injection strategies

easy separation =3

105

gain 25%

Conventional
reactor

Intensified reactor

104
0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

reactor conversion (XA)

Fig. 5. Impact of the intensication of a reactor on the energy consumption of
downstream separation by distillation. Inuence of the difculty of separation.

in the separator efciency yields to a large increase in the process

efciency. From a practical point of view, one should not forget
that this necessary increase in the separator efciency can require
a large capital expenditure if no new technology is considered.
2.3.2. Reactionseparation with recycling: impact of the reactor
performance on the energy consumption
The modes of action offered by the local intensication of the
reactor concerns the improvement of reaction efciencies with
a better control of operating conditions for efciency and selectivity gains [3,22], but also lead to reduced energy consumption
during the downstream separation operation. Indeed, a high reactor conversion increases the concentration of desired product at
separator inlet, leading, for a distillation column, to a lower reux
ratio, as well as a lower energy consumption in the boiler. This case
corresponds to a debottlenecking approach when the volumes of
equipment are xed. This is illustrated for a simple process composed of a continuous reactor where a rst-order reaction takes
place followed by an existing distillation column comprised of 20
theoretical plates [23]. The purity of the desired product at the column head is xed to 99%. Depending on the feed composition of the
column (reactor outlet), the power supplied at the boiler and the
recycling rate are adjusted to satisfy the composition constraints
at the column outlet. The results illustrated in Fig. 5 show that the
reactor intensication implies an energy saving. This gain depends
on the volatility of the species, indicated by the separation difculty (i.e. an easy separation corresponds to a relative volatility
= 3 and a difcult separation corresponds to a relative volatility
= 1.5; enthalpy of vaporization = 25 kJ/mol).
The results illustrated in Fig. 5 show that the energy gain is signicantly different depending on the difculty of separation. For
an increase of about 30% of the reaction efciency, the gains in
energy are about 2540%.
3. Process intensication by owsheet architecture
improvement
Among the different strategies of intensication, the possibility
of changing the architecture of the process and the use of a network
of devices are very often forgotten or not even considered as PI.
However, a great deal of process intensication is linked with the
sequence of the process operations and the potential integration

For each reactant fed to the reactor, there are several injection
strategies [24] to choose among one-shot (batch), step function
(continuous feed), pulsed feed injection and staged injection of the
feed (in the sense of time or space). The choice between a continuous and a batch operation is probably one of the most important
parameters. Batch processes are more appropriate for high-value
products such as ne chemicals (high operating costs, improved
exibility) whereas continuous processes are more advantageous
for industrial-scale production (low operating and labor costs,
higher production, safer production). For example, in the work of
Lomel et al. [25], the case of the Grignard reaction was considered.
It has been shown that the transition from a batch to a continuous
process presented a considerable reduction of operating times by
using a continuous micro-reactor in which mixing and heat transfer
were intensied.
Steady-state operation is the easiest conguration to operate
a continuous process but is not necessarily the optimal one. The
variation of the feeding ow rate (pulsed reactant injection) can
increase the process performance. This strategy can be benecial for
conversion or selectivity when the reaction kinetics is non-linear
or when reaction orders are different [24].
In xed bed operation with periodic ow reversal (reverse ow
reactor), the top and the bottom part of the catalyst bed play the
role of regenerative heat exchangers for feed and efuent, allowing
exothermal reactions to be operated auto-thermally. The conguration is well suited for exothermal reactions, for instance the
catalytic combustion of undesired compounds (volatile organic
compounds) [26]. The reverse ow operation for methanol synthesis has also been investigated in forced unsteady conditions
using a network of three catalytic xed bed reactors with periodical change of the feed position [27]. This conguration is called
ring reactor and corresponds to the principle of a simulated moving bed. Methanol synthesis is performed from CO, CO2 and H2 over
a commercial CuZnAl catalyst at 50 bars and 300 C. The objective was to save energy by using the thermal storage capacity of
the catalyst bed, and by applying an optimal temperature prole to
improve thermodynamic conditions for exothermal equilibriumlimited reactions. In this way, numerical simulations have shown
that a periodic unsteady regime and auto-thermal processing are
possible (for a given range of optimized switching times) because
no energy is needed except the moderate feed preheating at a temperature of 100 C. It has been demonstrated that this conguration
achieves a better conversion rate than the equivalent reverse-ow
conguration by avoiding the inconvenient phenomenon of washout.
The technique of staged injection of a reactant corresponds
to the addition of at least one of the reactants in a distributed
manner along the reactor (reactant dosing). This technique can
be led through an inter-stage feeding reactor system (discrete
feeding) or in a plug-ow membrane reactor (distributed feeding, which is a generalization of the rst conguration). Membrane
reactors have mainly been considered for the improvement of
equilibrium-limited reactions by selective product removal. For
instance, Kiwi-Minsker et al. [28] have developed a two-zone
micro-structured system: the zones are separated by a membrane
allowing to eliminate the hydrogen produced by the propane dehydrogenation, implying the shift of the thermodynamic equilibrium,
and hydrogen is oxidized by air providing heat for the endothermic

J.-F. Portha et al. / Chemical Engineering and Processing 84 (2014) 113

dehydrogenation. For a network of irreversible reactions, optimal

local reactant concentrations are essential for a high selectivity
toward the target product. If undesired consecutive reactions can
occur, it is usually advantageous to avoid backmixing. The possibility to enhance selectivity and yield in network of parallel and
serial irreversible reactions has been theoretically investigated by
Hamel et al. [29] and Thomas et al. [30] in an isothermal tubular
membrane reactor. It is demonstrated that the reaction order of
the individual reactions have to fulll some requirements (reaction order of the reactant in main reaction lower than the one in
the side reaction, case of heterogeneously catalyzed partial oxidation) if reactant dosing is used. Then, it is advantageous to lower
the reactant concentration by means of dosing; the corresponding decrease in concentration can be compensated by increasing
residence time.
3.2. Process intensication by heat and mass integration
3.2.1. Complex reactor networks for multistep reactions
optimization
Intensication requires to design the optimal conguration of
process units and to reach the optimal operating conditions for
each process unit. The optimal conguration also means that heat
and mass transfers should be managed judiciously thanks to heat
integration and optimal mass recycling policy. As indicated above,
methods have been developed as the attainable region method
(Glasser, Hildebrandt and co-workers Lakshmanan and Biegler
[45]) and the superstructure method (Kokossis, Floudas and coworkers). The reactor network superstructure consists in taking
into account all ow and mixing patterns with different feeding,
recycling and bypassing strategies around ideal CSTRs and PFRs
(modeled by ve CSTRs) in series, parallel, series-parallel congurations. Different techno-economic objective functions may be
considered to perform the optimization. The complex reactor network has been determined for isothermal operation rst [31] and
then for non-isothermal operation [32]. The examples presented
hereafter demonstrate that the optimal congurations maximizing
the yield are not a unique reactor but a network of several reactors.
The optimization of complex chemical schemes requires a reactor network to maximize products selectivity. In order to increase
this selectivity, local intensication strategy consists in the adjustment of the operating conditions, such as temperature, pressure
and concentration, to selectively change the kinetics of the reactions scheme, within the same process architecture. However, in
the following illustration, local intensication of chemical conditions in a specic reactor may lead to different process structures
involving different types of reactors. Chitra and Govind [33] considered the classical Van de Vusse reaction scheme and determined
that two optimal reactors structures, based on continuous stirred
tank and plug ow reactor, can be proposed to maximize the yield
of B as a function of the characteristic time of the three reactions.
The Van der Wusse reaction scheme can be written as:
1

Fig. 6. Impact of the intensication of chemical reactions on the process structure

delivering the maximal selectivity (according to [33]).

lead to more complex structures and that architecture modication

is also an intensication lever.
The following example concerns the isothermal synthesis of pinene whose reaction scheme can be written as:
AE
AD
BD

(11)

C  2D
2A C
Reactions are either rst-order or second-order reaction. The
objective function intends to maximize the selectivity of C over
D. The optimal conguration consists in one PFR and one CSTR
connected in a complex way as described in Fig. 7. The maximal
selectivity is obtained with a complex structure combining several
reactors with recycling and specic injection points. This example
shows that a simple architecture is not sufcient in order achieve
high performances.
The methodology of reactor network superstructure was then
extended to non-isothermal homogeneous operation [32] and for
non-isothermal multiphase reactor networks [34]. The example of

100 L/h of pure A

ABC

(10)

2AD
Fig. 6 illustrates the two reactors structures for different ratios
of reaction times tR1 /tR2 and tR1 /tR3 . Let us consider the intensication of initial working conditions, denoted by point A in the gure:
it is possible to increase the yield of B from 0.3 to 0.6 by decreasing
tR1 /tR2 and decreasing tR1 /tR3 (scenario B) or by decreasing tR1 /tR2
and increasing tR1 /tR3 (scenario C). These two strategies results in
two different process structures: scenario B can be performed with
a single plug ow reactor whereas scenario C requires using a cascade of a stirred tank followed by a plug ow reactor. This simple
example shows that local intensication of a specic device can

PFR
VPFR= 0.01 L

10.8 L/h

CSTR
VCSTR= 5000 L
Fig. 7. Reactor network for -pinene synthesis [31].

J.-F. Portha et al. / Chemical Engineering and Processing 84 (2014) 113

Adiabatic
20 C

210 C

Configuration 1 -

20 C

20 C

= 16700 10 kg

Adiabatic
210 C

Configuration 2 -

= 208 10 kg

20 C

Adiabatic

Adiabatic
210 C

Configuration 3 -

210 C

Isotherm

210 C

Configuration 4 -

= 1.4 10 kg

= 354 kg

210 C

Configuration 5 Fig. 8. Different intensication scenarios of a catalytic reaction by heat and mass integration.

naphthalene oxidation is considered as an exothermal reaction. The

catalyzed gas phase reaction between naphthalene A and oxygen
(always in great excess so that oxygen concentration is supposed
to be constant) is represented by the following scheme:
ABCD
AC

(12)

The objective is to maximize the yield of phthalic anhydride C.

The optimal conguration found consists in a cascade of four PFR
with four intermediate heat exchangers in order to remove the heat
excess. The feed stream is split between the rst (53.8%) and the
second stage (46.2%) and a recycle loop is added. An optimal prole
of removed heat is determined by optimization in order to nd a
maximum yield equal to 99.9% [32].
3.2.2. Reactor intensication by heat and mass integration
network
The case study of exothermal reactions is a typical case of PI
for which energy savings could be performed by judiciously using
the energy produced by the chemical reactions. Among the several
possibilities which can be used as mass recycling, heat recycling,
combined heat and mass recycling, PI can also lead to a drastic
minimization of volume of the reactors. The following example is an
illustrative case study that shows how intensication by heat and
mass integration can reduce drastically the catalyst amount of an
exothermal reaction. The irreversible reaction A + B P is assumed,
P denoting the products. The feed is supplied at 20 C, whereas reaction cannot be performed above 210 C for safety reasons (risk of
solvent vaporization). The catalyst is assumed to operate in kinetic
regime. Different process congurations are studied and presented
in Fig. 8, with the objective of saving energy and decreasing catalyst
mass to reach a conversion of 96%.
Adiabatic reactors are considered with different ways to recycle
mass and/or heat at reactor inlet in order to kinetically initiate the
chemical reaction. The optimal solution is to recycle both mass and
heat (conguration 4) because it enables to preheat the feed and to
decrease the global mean residence time. A drastic decrease in the

catalyst mass is observed. An isothermal reactor at 210 C (conguration 5) would obviously present the best productivity from the
point of view of the catalyst mass.

3.2.3. Process intensication by recycling

The following illustration is taken from the precipitation domain
used in the treatment of wastes. Classical installations are composed of a continuous stirred tank reactor (precipitator) followed
by a solidliquid separation (clarier) device. Because the concentration of the compound to eliminate is generally very low,
precipitation reactors have very large volumes. The decrease in
the reactor volume is a key parameter to reduce the cost of the
precipitation process, and Plasari and Muhr [35] proposed new
process strategies based on relationships between process parameters and reactor volume. Three different tested congurations are
presented in Fig. 9: the classical conguration described above
(Fig. 9a), a recycling strategy (Fig. 9b) where an important part of the
sludge is recycled from the clarier to the reactor and an integrated
precipitator-settler (Fig. 9c) which consists in a perfectly-mixed
precipitation zone coupled with a clarier.
The recycling and the integration strategy enables to drastically
increase the kinetics of the limiting step, which is the particles
growth, by increasing the concentration of particles and then
the specic surface of solid per unit volume. From population
balances and by taking into account some hypotheses (particles
produced only by nucleation, growth rate independent of particle
size, agglomeration and breakage negligible), the authors derive
the expression of the mean residence time for each conguration.
It is then possible to establish the following equations relating the
recycle ratio R = qR /Q and the liquidsolid separation ratio  = Q/q
to the volumes:
Precipitation with recycling
Intergrated precipation

Vrecy
1

Vclass
1+R

Vint
1

Vclass


(13)

J.-F. Portha et al. / Chemical Engineering and Processing 84 (2014) 113

B
Settling
Q-q

V
Precipitation
q
Classical configuration (a)

Q-q

Settling
Q-q

V
q

Precipitation

V
qR

Precipitation with recycling (b)

Integrated configuration (c)

Fig. 9. Congurations for precipitation and settling [35].

4. Intensication drivers

ROItime =

Itotal
Gannual

(14)

where Itotal denotes to the total amount of investment (D ) and

Gannaul the annual nancial gain (D /year). This annual nancial
gain is equal to the product of the annual nominal production
(tons/year), the product price (D /ton) and the productivity gain
associated with the intensied technology (%). From this relationship, the production gain necessary to pay off an investment of
100 kD in 3 years is presented in Fig. 10 as a function of the price of

Tons/year
50
1

30

50

35

3
2

10

25

4
5

20

20

3
50

100

10

250

5
1

4
5

10

5 6

10
16
Price (/kg)

50
100
25

3
4
5

20

500

The elds of applications of intensication are very vast and

the disparities among the various examples are sometimes signicant. It is therefore difcult to derive general laws on the economic
gains in operation. It is, however, possible to approach the problem in a simple way by estimating the return on investment (ROI)
time on the investment coming from the nancial gain contributed
by the productivity gain associated with the technological change

20

40

15

4.1. Economic drivers

10

45

100

Global intensication is not only a multi-units approach but

also a multi-dimensional strategy which has to consider the different drivers as economic, safety, eco-efciency and more generally
sustainability. The importance of economic considerations in the
design of process equipment and chemical plant facilities to maximize protability is no longer to be demonstrated. However, most
local intensication studies focus on the technical performance
only, without considering the cost of the new intensied technology. It is often the case that projects based on nice improved
systems have been abandoned due to economic reasons. More
generally, not only production costs but other criteria of process
sustainability have to be considered, as environmental problems,
health and safety hazard. Nevertheless, such integrated assessment
of these elements into process intensication has not been generally addressed in the previous approaches. Several examples are
presented here to show that it is necessary to develop a more
global approach of PI to solve the complicated issues of sustainability.

(transition from the conventional process to the intensied process). This estimation can be made by the following simple
calculation:

Production gain (%)

where Vclass is the reactor volume of the classical conguration (a),

Vrecy is the reactor volume of the conguration with recycling (b)
and Vint is the perfectly mixed precipitation volume of the integrated conguration (c). These equations demonstrate that the
reactor volume can be drastically reduced if the recycle ratio or
the liquidsolid separation ratios are large.

40

10
20
50
63

10
20
100

Fig. 10. Production gain required for an investment of 100 kD in 36 months based
on the price of the manufactured product and for different production capacities
(tons/year) [23].

10

J.-F. Portha et al. / Chemical Engineering and Processing 84 (2014) 113

Table 4
Scenarios for the chlorination process [17].
Scenario 1
Reactor network

Separation network
Prot (M$)
Annual cost (k$)
Yield
Conversion

Scenario 2

the product and the production capacity [23]. It is found that technological innovation mostly favors the pharmaceutical processes
(products with high added value and low tonnage) because even
with a small overall gain in the productivity by a few percent, there
is a short time to return on investment. The innovation for intermediate products at a few euros per kilogram and higher annual
tonnage requires a substantially higher overall productivity gain.
For a higher investment of several hundred kilo-euro (to use the gure, either the value of the nominal productivity at constant gain,
or the gain at constant nominal productivity must be multiplied
accordingly), Fig. 10 indicates that the expected gain on the entire
production line should not be incremental but should at least touch
several dozen percent, which is a much more difcult objective.
The simple hypothesis made here considers that the nancial gain
is borne entirely by the productivity gain and shows the need to
nd other gains on operating costs in order to reduce the ROI time.
However, Becht et al. [36] showed that Capex reduction is a
major driver of process innovation. Their rough calculation was
based on an overall Capex of 10 MD for setting an intensied technology, and a return on investment time of 10 years. Even for an
economically pessimistic scenario, an overall net present value of
0 D , the positive differential cash ow has to amount 1.63 MD /year
(for a constant interest rate of 10%) meaning that the new technology has to provide an economic net advantage of 163 D /ton if the
production is 10 kt/year. Even for a rather poor project with a zero
NPV, this simple example shows that the target is pretty ambitious.
Techno-economic analysis is a key issue in process design and
process intensication. As illustrated in the following example, various economic scenarios lead to different technical solutions. This
implicitly shows that reversely, the way how improvement or local
intensication of equipment is performed may strongly inuence
the nal protability strategy. A complex techno-economic analysis, taking into account Capex and Opex, has been performed by
Kokossis and Floudas [17] for benzene chlorination (production of
50 kmol/h of chlorobenzene). The reaction network is given by:
C6 H6 + Cl2 C6 H5 Cl + HCl
C6 H5 Cl + Cl2 C6 H4 Cl2 + HCl

Scenario 3

2 CSTR in series
Medium recycling
ratio
2 distillation columns (identical for the 3 scenarios)
1.224
1.166
321
293
0.314
0.46
0.326
0.501

3 CSTR in series
High recycling ratio

(15)

Authors have underlined that the optimization of various objective functions imply various process networks. The objective
functions were, respectively:
- to maximize the total prot of the plant (scenario 1),
- to minimize the annualized venture cost of the plant (scenario 2),
- to maximize the overall yield (ratio of the molar ow rate of
desired product over the molar ow rate of feed reactants) (scenario 3).
The results, given in Table 4, show that for each objective
function, different process congurations are obtained. Indeed,
maximizing the overall yield (scenario 3) will not lead to the maximal prot because yield does not actually take into account the
separation network and the recycling rate. As expected, a PFR is the

1 PFR
Low recycling ratio

0.667
785
0.565
0.617

optimal reactor for a consecutive reaction network with respect to

the yield criterion. The annualized cost and the prot (scenarios
1 and 2) provide similar results with process based on CSTR presenting a lower Capex and Opex. These results show the importance
of taking into account the coupling of the separation and the reactor network and the need for a systematic methodology to extract
optimal solutions.
4.2. Safety drivers
It is commonly accepted that process intensication is closely
linked to equipment miniaturization. The resulting minimization
of the inventory of hazardous material is so that the safety and
environmental consequences of loss of containment are reduced
in the event of a large leak from the process equipment, leading to the concept of inherently safer process by intensication.
However, according to Luyben and Hendershot [37], the direct link
between intensied process and inherent safety is not always true
and a more global analysis of the process has to be performed. They
claimed that these small hold-ups may also be very sensitive to disturbances which can cause rapid changes within the process, and
which may be detrimental to safety and product-quality constraints
before corrective action can be taken. The intensied process may
be less robust to changes in the internal and external environment,
creating a conict with the concept of inherently safer process
design. Among their different examples, they considered the comparison of the irreversible and exothermal benzene nitration in two
alternatives scenarios, with the same objective of 96% conversion:
C6 H6 + HNO3 C6 H5 NO2 + H2 O

(16)

The rst scenario is based on a large CSTR and the second considers two small CSTRs, each reactor being equipped with a jacket to
cool the uid. The two small reactors can be viewed as an intensied
version of the large reactor because the higher surface to volume
ratio enables to increase the cooling heat ux per reacting volume
of the reactor. The two corresponding scenarios are presented in
Table 5. Clearly, in absence of dynamics consideration, the intensication principles would favor scenario 2 because the inventory of
dangerous materials is smaller and the size of equipment is smaller
meaning a lower capital cost, and less required coolant.

Table 5
Scenarios for benzene nitration process [37].

Conguration
Volume of each reactor
(m3 )
Liquid holdup (%)
U (BTU/h/ft2)
A (ft2)
Temperature deviation ( F)
(response to a step of
benzene ow)

Scenario 1

Scenario 2

1 CSTR
122

2 CSTR in series
14

100
150
1230
+0.2 and 0.4

75
150
289
+1.7 and 3.6

J.-F. Portha et al. / Chemical Engineering and Processing 84 (2014) 113

Table 6
Results of acetaldehyde production [11].
Variable

Base
case

Scenario 1
(no new
units)

Scenario 2
(with new
units)

Reaction temperature (K)

Reux ratio (3rd column)
Ethanol inventory (tons)
Process yield

600
3.5
11,240
0.589

580
5
7100
0.315

610
3
7080
0.648

The response of each conguration to a step in the benzene ow

is studied (increase to 120% and decrease to 80% with respect to the
nominal value of 100%). Results show that the maximal temperature deviations are larger when two small CSTRs are used. A change
of the inlet ow rate implies a larger variation of concentrations in a
small CSTR than in a large CSTR due to different space times (lower
inertia for a small space time). If the controller response is not sufciently fast, the kinetics of reaction being very sensitive to variation
of concentrations, the system of two CSTRs is much more instable,
despite of large heat transfer capacities.
Despite several advantages under steady-state operation, the
rst small CSTR presents much poorer dynamics results due to a
lower liquid hold-up. The closed-loop response is more oscillatory, the temperature deviations being about an order of magnitude
larger in the small reactor than in the large one.
4.3. Inventory drivers
The minimization of the inventories of reactants and products
is a key issue in factories which is related to safety and economic
considerations because stocks of chemicals are often hazardous
and expensive. Then, global intensication aims at the compromise
between the maximization of the throughput for a given process,
the inventory minimization, or the minimization of utility materials
and feedstock, as proposed by El-Halwagi and co-workers [44].
The production of acetaldehyde through ethanol oxidation,
studied by Ponce-Ortega et al. [11], is considered here as an illustration case to reduce process inventory. The main difference here
with respect to the classical superstructure optimization is that the
process intensication approach considers the minimization of the
inventory for a given production, whereas the classical superstructure optimization usually considers the maximization of the prot
or the minimization of the total cost. The owsheet is composed of
a catalytic reactor, two scrubbers to remove some gases from the
industrial exhaust (using ethanol as a solvent), and a network of
three distillation columns to separate acetaldehyde from the unreacted ethanol, from light organic wastes and from water. About
11,240 tons of ethanol, that is a hazardous and ammable liquid, are
stored for two weeks. Whereas the classical superstructure-based
optimization usually considers the maximization of the prot, the
process intensication approach, developed by the authors, is used
to minimize ethanol inventory. To reach this objective, two scenarios were taken into account: one scenario without new intensied
equipment and one scenario with an intensied reactor and a distillation column. The main results are presented in Table 6. The
addition of new intensied units in scenario 2 implies a higher process yield but does not provide a signicant reduction of ethanol
inventory with respect to scenario 1. By manipulating the reux
ratio of the third column in scenario 1, the ethanol recovery rate
is higher leading to ethanol savings. The ethanol quantity stored is
decreased by approximately 37%.
4.4. Environmental drivers
The environmental impact of a chemical process is closely
related to social, technical and economic issues. The decrease of

11

environmental impacts is generally costly but can also generate

savings due to a decrease of energy consumption for example.
A typical example concerns the reduction of greenhouse gases
(GHG) emissions in a factory. To reach the given constraints, some
investments are necessary in order to save energy (optimized heat
exchanger network, use of more efcient equipment, recovery of
low level energy, etc.) whose production is often the main source
of GHG emissions. Energy saving is also benecial in the long term.
The main tool to determine the environmental impact of a process
is life cycle assessment (LCA). This methodology has been widely
described in the literature and is now considered as a tool for process design, selection and optimization [38]. LCA is a standardized
method divided in four main steps [39]:
- Denition of the scope of the study (objectives and system boundary).
- Life cycle inventory analysis (determination of mass and energy
input and output uxes by mass and energy balances).
- Life cycle impact assessment [calculation of environmental
impacts based on mass and heat uxes with an impact assessment
method, for example Eco-Indicator [40]].
- Results interpretation (improvement analysis).
The results of LCA are subjected to the quality of the collected
data, to the hypotheses considered in each scenario, to the chosen
impact assessment method and to the selected allocation procedure when co-products are obtained. LCA has been used in the
literature coupled with process integration to analyze a biofuels
production process [41], or coupled with exergy analysis to analyze
a petrochemical process [42]. The coupling between environmental
analysis, economic analysis and process integration techniques has
been recently developed by different research teams as described
below. The environmental method consistently used in these methods is LCA. Ponce-Ortega et al. [13] establish a multi-objective
optimization model for the synthesis of recycle and reuse mass
integration networks. Sources (inlet ow rate), interceptors (representing each available unit with conversion factor based on physical
and chemical properties) and sinks (exit ow rate or waste stream
discharged to environment) are dened and used to build the problem together with constraints. An environmental objective function
is measured by an impact evaluation method (Eco-Indicator 99)
and an economic objective function considers fresh sources and
treatment costs. Solution is represented by a set of Pareto optimal
solutions and a network is also obtained. This approach requires
mathematical models to represent all process alternatives, making
its application difcult and time consuming.
An other methodology for systematic process analysis and generation of design alternatives has been developed by Carvalho et al.
[14] for continuous or discontinuous processes. The corresponding tool, called SustainPro, determines a set of mass and energy
indicators, establishes operational and design targets and identies process alternatives matching the design targets. The starting
point is to dene the process specications (prices, conditions of
operation, and process owsheet) of the base case. Then, indicators are calculated to identify process bottleneck. A general safety,
economic and environmental evaluation is performed to take into
account these issues in the design. Finally, design alternatives are
generated and each alternative is evaluated using environmental
impact tools and safety indices with respect to the reference design
(base case). This method is especially useful for process retrot. This
methodology has been applied to the production of an intracellular
enzyme: the -galactosidase. The basic process operates in a batch
mode and is divided into three main sections: fermentation, primary recovery and purication. By applying the methodology, two
main solutions were found: water purication or water recycling.
As water purication was not economically viable, the recycle of

12

J.-F. Portha et al. / Chemical Engineering and Processing 84 (2014) 113

water was chosen. An important decrease of eutrophication and

acidication indexes was observed and the investment was fully
recoverable making the new design an economically sustainable
alternative.

5. Conclusion
In the present contribution, several examples have been given
to illustrate that the classical approach of process intensication
based on single-unit improvement (local intensication) presents
several limitations when compared to holistic overall processbased intensication (global intensication). In fact, a joint local
and global intensication strategy needs to be developed, based
on a two-step approach. Firstly, global intensication approach
can be performed to get a drastic improvement of the whole
performance of the global process. The appropriate selection of
the elementary unit operations that have to be intensied is one
technique that can be used. Optimization of process architecture
by using reactant injection strategies, heat and mass integration or reactants recycling is another complementary approach.
Secondly, local intensication by classical techniques, such as
microstructuring to avoid heat or mass transfer limitations, use
of innovative driving forces or multi-functional systems, can be
performed to reach the expected technical performances of the
devices.
In addition, it has been shown that process or device improvement is not only based on technical criteria but several drivers such
as economic, safety, eco-efciency and sustainability have to be
considered to fulll the key objectives in designing improved processes and new plants or in retrotting existing units. For instance,
enhanced process safety is submitted to PI. As the inventory and
ows of hazardous substances are lowered, the process risk is
typically reduced. Additionally, conservation of natural resources,
including better utilization of mass and energy, may be linked to
PI.
The objective of the paper was not to oppose local and global
intensication but to show the complementary aspects between
both methods. Local intensication is essential because it considers the physical and chemical phenomena taking place in a system.
Nevertheless, the focus is also made at the scale of unit operation
without considering the interactions between them. Global intensication techniques will consider these interactions and constitute
then a relevant tool that can validate the addition of intensied
equipment in a global process.
Given the complexity of a typical process and the various
objectives, several authors [11,14] have already developed an
effective framework for generating and pursuing valid opportunities for process intensication. In both approaches, Gani and
El-Halwagi groups propose to couple an intensication strategy
with a multi-objective optimization problem where the units that
will be intensied are not pre-specied and more than one unit
can be intensied simultaneously and where additional intensied
units can be considered in the strategy. Optimization criteria are
based on the previously cited drivers (economic, safety, sustainability, etc.). Following the same objectives, Freund and Sundmacher
[43] have developed a methodology not based on the classical unit
operation concept but on the basic functional principles encountered in a chemical process. Process intensication is led at three
levels: the phase level, the unit level and the plant level and interactions between levels are taken into account. In these works,
models of performance of the different devices have an important
role to play in this framework. The results obtained by the two
approaches are promising but however the challenge is to propose
now more reliable models that enable to consider local intensication of equipment.

Appendix A. Nomenclature

Latin letters
e/i
elasticity taking into account the impact of i variations
on 
Fj
molar ow rate of component j (mol/s)
molar ow rate of component j at unit inlet (mol/s)
Fj0
G
extensity ux
Gannual annual nancial gain (D /year)
total amount of investment (D )
Itotal
number of process units
N
actual owing volume (m3 s1 )
q
Q
actual owing volume (m3 s1 )
recycle ratio
R
V
unit volume (m3 )
Greek letters
i
efciency of a process or an equipment i
g
overall efciency

liquidsolid separation ratio
References
[1] A. Stankiewicz, J.A. Moulijn, Process intensication: transforming chemical
engineering, Chem. Eng. Prog. 96 (2000) 2234.
[2] R. Jachuck, Process intensication for responsive processing, Trans. IChemE 80
(2002) 233238.
[3] J.I. Walpot, TNOs work on intensication: practical examples, J. Chem. Technol.
Biotechnol. 78 (2003) 236240.
[4] C. Tsouris, J. Porcelli, Process intensication has its time nally come? CEP
Mag. 99 (2003) 5055.
[5] S. Becht, R. Franke, A. Geisselmann, H. Hahn, Micro process technology as a
means of process intensication, Chem. Eng. Technol. 30 (2007) 295299.
[6] V. Hessel, C. Knoblauch, H. Lwe, Review on patents in microreactor and microprocess engineering, Recent Pat. Chem. Eng. 1 (2008) 116.
[7] I.E. Grossmann, A.W. Westerberg, Research challenges in process systems engineering, AIChE J. 46 (2000) 17001703.
[8] J.A. Moulijn, A. Stankiewicz, J. Grievink, A. Gorak, Process intensication and
process system engineering: a friendly symbiosis, Comp. Chem. Eng. 32 (2008)
311.
[9] W. Marquardt, L. Von Wedel, B. Bayer, Perspectives on lifecycle process modeling, AIChE Symp. Ser. 96 (2000) 192214.
[10] I.E. Grossmann, A.W. Westerberg, L.T. Biegler, Retrot Design of Processes,
Carnegie Mellon University, Engineering Design Research Center, Department
of Chemical Engineering, 1987.
[11] J.M. Ponce-Ortega, M.M. Al-Thubaiti, M.M. El-Halwagi, Process intensication:
new understanding and systematic approach, Chem. Eng. Proc.: Process Intensication 53 (2012) 6375.
[12] L. Sun, J. Pan, A. Wang, A multi-objective process optimization procedure under
uncertainty for sustainable process design, in: Proceedings2nd International
Conference on Bioinformatics and Biomedical Engineering, ICBBE08, 2008, pp.
43734376.
[13] J.M. Ponce-Ortega, F.W. Mosqueda-Jimnez, M. Serna-Gonzlez, A. JimnezGutirrez, M.M. El-Halwagi, A property-based approach to the synthesis of
material conservation networks with economic and environmental objectives,
AIChE J. 57 (2011) 23692384.
[14] A. Carvalho, H.A. Matos, R. Gani, SustainPro a tool for systematic process
analysis, generation and evaluation of sustainable design alternatives, Comp.
Chem. Eng. 50 (2013) 827.
[15] T.Y. Zhang, Process chemistry: the science, business, logic and logistics, Chem.
Rev. 106 (2006) 25832595.
[16] D.M. Roberge, B. Zimmermann, F. Rainone, M. Gottsponer, M. Eyholzer, N. Kockmann, Microreactor technology and continuous processes in the ne chemical
and pharmaceutical industry: is the revolution underway? Org. Proc. Res. Dev.
12 (2008) 905910.
[17] A.C. Kokossis, C.A. Floudas, Synthesis of isothermal reactor-separator-recycle
systems, Chem. Eng. Sci. 46 (1991) 13611383.
[18] D. Hildebrandt, D. Glasser, Attainable region and optimal reactor structures,
Chem. Eng. Sci. 45 (1990) 21612168.
[19] D. Glasser, D. Hildebrandt, Reactor and process synthesis, Comp. Chem. Eng. 21
(1997) 775783.
[20] X. Wang, C. Wang, S. Jin, H. Chen, Optimal operating conditions for reactorseparator-recycle system, J. Sci. Ind. Res. 65 (2006) 867872.
[21] L.G. Matallana, A.M. Blanco, J.A. Bandoni, Design of reactor-separator-recycle
systems based on the optimization of estimations of the domain of attraction,
Ind. Eng. Chem. Res. 50 (2011) 67696776.

J.-F. Portha et al. / Chemical Engineering and Processing 84 (2014) 113

[22] D.M. Mohunta, Process intensication process for improving protability and
remaining competitive, 2013, http://www.ccdcindia.com/index2.php?act=
service&sub=1
[23] L. Falk, The place of intensied processes in the plant of future, in: J.P. Dal Pont
(Ed.), Process Engineering and Industrial Management, ISTE & Wiley, London,
2012, pp. 401430.
[24] R. Krishna, S.T. Sie, Strategies for multiphase reactor selection, Chem. Eng. Sci.
49 (1994) 40294065.
[25] S. Lomel, L. Falk, J.M. Commenge, J.L. Houzelot, K. Ramdani, The micro-reactor:
a systematic and efcient tool for the transition from batch to continuous
process? Chem. Eng. Res. Des. 84 (2006) 363369.
[26] K. Ramdani, R. Pontier, D. Schweich, Reverse ow reactor at short switching
periods for VOC combustion, Chem. Eng. Sci. 56 (2001) 15311539.
[27] S.A. Velardi, A.A. Barresi, Methanol synthesis in a forced unsteady-state reactor
network, Chem. Eng. Sci. 57 (2002) 29953004.
[28] L. Kiwi-Minsker, O. Wolfrath, A. Renken, Membrane reactor microstructured
by lamentous catalyst, Chem. Eng. Sci. 57 (2002) 49474953.
[29] C. Hamel, S. Thomas, K. Schdlich, A. Seidel-Morgenstern, Theoretical analysis
of reactant dosing concepts to perform parallel-series reactions, Chem. Eng. Sci.
58 (2003) 44834492.
[30] S. Thomas, S. Pushpavanam, A. Seidel-Morgenstern, Performance improvements of parallel-series reactions in tubular reactors using reactant dosing
concepts, Ind. Eng. Chem. Res. 43 (2004) 969979.
[31] A.C. Kokossis, C.A. Floudas, Optimization of complex reactor networks. I:
Isothermal operation, Chem. Eng. Sci. 45 (1990) 595614.
[32] A.C. Kokossis, C.A. Floudas, Optimization of complex reactor networks. II: Nonisothermal operation, Chem. Eng. Sci. 49 (1994) 10371051.
[33] S.P. Chitra, R. Govind, Yield optimization for complex reactor system, Chem.
Eng. Sci. 36 (1981) 12191225.
[34] V.L. Mehta, A.C. Kokossis, Nonisothermal synthesis of homogeneous and multiphase reactor networks, AICHE J. 46 (2000) 22562273.

13

[35] E. Plasari, H. Muhr, Developments in precipitation engineering for the process

intensication in the environmental protection and other purication industrial activities, Chem. Eng. Trans. 11 (2007) 6570.
[36] S. Becht, R. Franke, A. Geisselmann, H. Hahn, An industrial view of process intensication, Chem. Eng. Proc.: Process Intensication 48 (2009)
329332.
[37] W.L. Luyben, D.C. Hendershot, Dynamic disadvantages of intensication in
inherently safer process design, Ind. Eng. Chem. Res. 43 (2004) 384396.
[38] A. Azapagic, Life cycle assessment and its application to process selection,
design and optimization, Chem. Eng. J. 73 (1999) 121.
[39] J.B. Guine, M. Gorre, R. Heijungs, G. Huppes, R. Kleijn, A. de Koning, R.
van Duin, M.A.J. Huijbregts, Handbook on Life Cycle Assessment. An Operational Guide to the ISO Standard, Kluver Academic Publishers, Dordrecht, The
Netherlands, 2002.
[40] M. Geodkoop, R. Spriensma, The Eco-Indicator 99. A damage oriented method
for life cycle impact assessment. Methodology report and manual for designers.
Technical Report, 2001, Pr Consultants, Amersfoort, The Netherlands.
[41] P. Chouinard-Dussault, L. Bradt, J.M. Ponce-Ortega, M.M. El Halwagi, Incorporation of process integration into life cycle assessment for the production of
biofuels, Clean Technol. Environ. Policy 13 (2011) 673685.
[42] J.F. Portha, S. Louret, M.N. Pons, J.N. Jaubert, Estimation of the environmental impact of a petrochemical process using coupled LCA and exergy analysis,
Resour. Conserv. Recy. 54 (2010) 291298.
[43] H. Freund, K. Sundmacher, Towards a methodology for the systematic analysis
and design of efcient chemical processes. Part 1: From unit operations to elementary process functions, Chem. Eng. Proc.: Process Intensication 47 (2008)
20512060.
[44] M.M. El-Halwagi, A.A. Hamad, G.W. Garrison, Synthesis of waste interception
and allocation networks, AIChE J. 42 (1996) 30873101.
[45] A. Lakshmanan, L.T. Biegler, Synthesis of optimal chemical reactor networks,
Ind. Eng. Chem. Res. 35 (1996) 13441353.