Chemical Engineering and Processing 104 (2016) 219242

Contents lists available at ScienceDirect

Chemical Engineering and Processing:

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

Distillation technology and need of simultaneous design and control: A

review
V. Aneesha , Rahul Antonya , Ganesh Paramasivanb , N. Selvarajua,*
a
Department of Chemical Engineering, National Institute of Technology Calicut, Calicut, Kerala 673601, India
b
Advanced Propulsion Research Group, Vikram Sarabhai Space Centre, Thiruvananthapuram, Kerala 695022, India

A R T I C L E I N F O A B S T R A C T

Article history:
Received 19 August 2015 During the whole of history, distillation has been the most extensive separation method. Regardless of its
Received in revised form 15 March 2016 exibility and simplicity, distillation technology is highly energy consuming. Modernistic distillation
Accepted 24 March 2016 technologies based on intensication process has more ascendency than its conventional, not in terms of
Available online 31 March 2016 less energy consumption but it reduces the capital investment cost and improved efciency. In order to
meet the demands of modern society, a drastic development in the eld of separation is required. Here
Keywords: this article rst pointing towards the overview of latest trends in distillation technology with process
Distillation technology intensication principles. In order to meet the control of modern chemical process, simultaneous design
Optimisation
and control is proposed over the conventional design and control methods. Second, the article is pointing
Mixed integer dynamic optimisation
to various optimisation techniques used for simultaneous design and control of chemical processes.
Simultaneous design and control
Signicant development in the area of simultaneous design and control is categorised into two. 1)
Controllability indicator based foundation 2) optimisation based foundation. In the simultaneous design
and control of chemical process, rst discussion started with the motivation and signicance of
simultaneous design and control based optimisation technique, then pointing towards the existing
methodologies with theoretical algorithms and its merits and demerits.
2016 Elsevier B.V. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
2. Advanced distillation technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
2.1. Heat integrated distillation column (HIDC) . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
2.2. Membrane distillation (MD) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
2.3. Cyclic distillation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
2.4. Cryogenic distillation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
2.5. Reactive distillation column . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
3. Optimisation based simultaneous design and control . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
3.1. Need and importance of simultaneous design and control . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
3.2. Upto the minute of optimisation based simultaneous design and control . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
3.3. Controllability index based optimisation approach (CIOA) . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
3.4. MIDO based approach for simultaneous design and control . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
3.4.1. Framework for simultaneous process control design . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228
3.4.2. Framework for model predictive control . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
3.5. Framework for simultaneous design and control under uncertainty . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
3.6. Robust based optimisation approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
3.6.1. Quadratic lyapunov function based approach (QLf-based approach) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234
3.6.2. Singular structured value based approach (SSV based approach) . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234

* Corresponding author at: Department of Chemical Engineering, National

Institute of Technology Calicut, Kozhikode, Kerala 673601, India.
E-mail address: selvaraju@nitc.ac.in (N. Selvaraju).

http://dx.doi.org/10.1016/j.cep.2016.03.016
0255-2701/ 2016 Elsevier B.V. All rights reserved.
220 V. Aneesh et al. / Chemical Engineering and Processing 104 (2016) 219242

3.6.3. Hybrid worst case (HWC) approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234

3.6.4. Constructive nonlinear dynamics (CND) based approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234
3.7. Embedded control based optimisation approach (ECOA) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234
3.7.1. Framework of embedded control based optimisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
3.8. Observer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
3.9. Black box optimisation approach (BBOA) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236
4. Challenges and future directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
5. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238

1. Introduction main principle of PI is clearly dened in [11] as: 1. Maximising the

A physical separation of liquid mixtures with difference in
boiling point or volatility of the essential components is called
Distillation. It is a widely used separation process in the current era
and its history starts from a beverage factory in China (800 BCE)
[1]. It is a batch wise process by heating a liquid mixture in a pot
and condensing the vapour with the help of a condenser. The
process of distillation taking place continuously (Fig. 1) in vertical
empty columns linked with piping, heat exchangers, vessels,
pumps and supporting structures [25]. By considering the feed
stream and product, distillation requires a decrease in entropy so it
is not spontaneous in nature. For making the process thermody-
namically possible additional heat is required. Due to the
irreversible losses with respect to the pressure drop, mass transfer
and heat transfer, the efciency of distillation is very low [69].
Development in the chemical process industry transformed
towards the high energy efcient separation technologies. By the
development of process intensication (PI) the overall cost is
reduced and improves the sustainability of industrial process. PI is
a process design methodology, which improves the process
exibility, quality of the product and inherent safety [10]. The
intra and inter molecular event effectiveness 2. Give equal
processing experience to each and every molecule 3. Maximise
the synergistic effects 4. Maximise the surface area to which force
apply and optimise the driving force.
Chemical processes are nonlinear system continuously affected
by external disturbances and process uncertainty which leads to
process variability. Even though algorithms can be used to reduce
process variability and design, also have a huge impact on closed
loop behaviour of the system. Conventional chemical plant design
is fully based on steady state calculations and synthesis
investigation by considering controllability only after the process
design variables have been determined. Thus independent
consideration of controllability and exibility is taking place. This
sequential approach is inadequate to handle large variations in
controllability and exibility.
Based on the above, it has been proposed to optimise the
chemical process design simultaneously. Thus the closed loop
performance of the system dynamics are analysed together with
system design and control. Design and control related degrees of
freedom is used to determine the optimal operating state and unit
capacity of the system. The problem is addressing trade-off

Fig. 1. Schematic diagram showing continuous distillation column.

 V. Aneesh et al. / Chemical Engineering and Processing 104 (2016) 219242
between conicting design and control objectives. The problem
has different visible features such as control structure selection,
process synthesis, parametric uncertainty and worst case scenario
assessment. There is no unied structure to address all these
aspects. Recent development in this area have formulated this
problem with in a dynamic programming framework, may cover
almost all aspects of the problem and signicant computational
effort required.
This methodology introduces a new approach to integrate the
design and control that lessen some of the complex computational
burdens. The gist of the method is based on the application of
robust control tools to estimate bounds on the stability, dynamic
operability, controllability and feasibility of the process to be
designed. The application of nonlinear optimisation problem with
these robust control tool boxes are very easy to handle than the
dynamic programming based formulations. This formulation is
fully based on a cost function and is dened in terms of physically
measurable process variables. Process steady state economics are
calculated from the capital and operating cost functions. Operating
cost is related to utility and process steams cost and capital cost has
direct link with process unit capacities. The dynamic deviations of
the process variables with respect to a nominal operating
condition are also assigned a cost that is added as an additional
term into the cost function. To measure the dynamic deviations in
the process variables, different approaches has been used referred
before this as process variability. Process variability related cost
within the problem has excluded from the controllability based
methodologies but have considered as a controllability constraint
within the analysis.
2. Advanced distillation technologies
The efciency of a distillation process can be improved by using
intelligent column design. In this type of approaches we are
considering minimum use of resources over the safety and process
lifetime [12,13]. The intelligent design of a column involves the
Fig. 2. Heat Integrated Distillation Column (HIDC): different type of congurations.
221
approaches in the following levels 1. Materials used for construc-
tion and resources 2. Design of process and congurations 3.
Internal design optimisation. The main objective of distillation is to
obtain maximum yield with very minimum amount of energy,
normally achieved with the help of sequencing the column and
modifying the column congurations. Degree of freedom of the
tray design is higher than packing design.
The following section deals with the latest technologies used in
the distillation process, with high output yield and excellent
industrial impacts. Beyond that each and every distillation column
modelling and optimisation changing and improving day to day
with varieties of control and design of processed and applications.
2.1. Heat integrated distillation column (HIDC)
This type of distillation column provides high potential for
energy saving by using the whole rectifying section act as a heat
source and the stripping part act as a heat sink instead of using a
single point heat source and sink [14]. The work input is provided
from the top outlet of the stripper section by the use of a
compressor and the heat pump cycle is closed by the valve ashing
the rectier bottom outlet. This type of distillation has highest
energy saving possibility than any other distillations. The capital
investment is very high compared with the conventional methods
because of its complex congurations and additional compressors
[15]. Experimental and theoretical studies proved that HIDC has
higher energy efciency than conventional methods. Process can
be operated very smoothly without any tuning difculties [15,16
21]. The HIDC congurations are not yet implemented in industrial
applications. The important issues we want to consider for the
design of HIDCs are exibility to change the operating point, effect
of an external disturbance or an external component and dynamics
and control processes. Therefore, achieve a settlement between
system design economics and operations appears to be very
important and it has to be handled with atmost caution [17,20,21].
Among the design and operating variable, the most signicantly
222 V. Aneesh et al. / Chemical Engineering and Processing 104 (2016) 219242
affected variable which disturb the performance of the system are
compression ratio and heat transfer coefcient. Recent research
works shows more good congurations like multi-tube and multi
shell, plate type or structured HIDC systems [1618]. Basic
schematic of HIDC is shown in Fig. 2.
The development in the conguration of HIDC was so far limited
to the separation of binary mixtures. A multi-component mixture
separation with HIDC [20,22,23] is an important and challenging
topic for future work. An often ignored issue related to HIDC is
dependence on compressor, installing a complex and expensive
rotating device to ensure smooth and continuous operation of
column in the case of breakdown of running plant is not practical.
By any mean, it is to be anticipating that with increase in energy
prices, Heat assisted distillation column and heat integrated
distillation column will become more economically competing in
the future. To make a break down in this account, Super HIDC is the
rst commercial conguration so far.
2.2. Membrane distillation (MD)
Membrane separation technology controls a great promise in
advancing desalination and water treatment to making better the
efciency of impurity removal as well as augment water supply
[2426]. Microltration, Ultra-ltration, Nanolration, Reverse
Osmosis and Membrane Distillation are the commonly used
membrane technologies for the application of waste/water
treatment [27,28]. In the eld of membrane based desalination
and water/waste water treatment, Membrane Distillation (MD) has
many attractive features compared with others [29,30]. Tremen-
dous development is taking place in the eld for novel membrane
fabrication [31], designing more efcient modules [32] and
Fig. 3. Membrane Distillation Congurations.
optimise the MD process [33] pointing towards the theoretical
and experimental studies [3436]. It is a thermally driven
distillation process and separation is enabled due to phase
changes. A hydrophobic membrane provides a barrier for liquid
phase. The driven force is achieved by the differences in the partial
vapour pressure, which is commonly triggered by the temperature
differences.
Commonly used membrane distillation are direct contact
membrane distillation(DCMD), air gap membrane distillation
(AGMD), sweeping gas membrane distillation (SGMD), vacuum
membrane distillation (VMD), permeate gap membrane distilla-
tion (PGMD), vacuum multi-effect membrane distillation (VMMD)
[3739]. The advantages of membrane distillation are easy to get a
pure product, highly efcient, no chemicals used, very less cost to
avoid clogging and simple membrane structure. The main
disadvantage of membrane distillation process is high initial cost,
sediment clogging and reduced power penetration due to
sediment effect.
DCMD (Direct contact membrane distillation): is the simplest
technique in membrane distillation [40], in this membrane has
direct contact with both hot and cold uids. In this case heat loss
due to conduction through membrane is very high compared with
other technique. This technique is very suitable for the applications
with water as the major permeate and aqueous solution does not
contain any non-volatile element [41].
AGMD (Air gap membrane distillation): In AGMD permeates gap
lies in between the membrane and a cooled walling, the gap is
lled with air and the evaporator channel is similar to DCMD. The
feed has direct contact with hot side of the membrane only. It can
be used to remove the volatile component from aqueous solutions
because permeate uid and membrane surface has no direct
 V. Aneesh et al. / Chemical Engineering and Processing 104 (2016) 219242
Table 1
Advantages and Disadvantages of Membrane Distillation.
Mode Pros
Direct contact
Easiest and simplest
membrane
High gained output ratio
distillation
Distillate ux is more stable than other methods

Most acceptable mode to remove volatiles
Air-gap
Lower distillate ux than other modes
membrane
Lowest thermal loss
distillation
No pore wetting risk

Less tendency of membrane fouling
Vacuum
High distillate ux
membrane
Ability to remove volatility
distillation
No pore wetting

Lower thermal polarization
Sweeping gas
Lowest thermal polarization
membrane
No pore wetting risk
distillation
Better future perspective for aqueous streams
contact. According to the air gap between the permeate side and
the condensing surface, heat transfer resistance get increased, so
less heat is lost during the operation [42].
PGMD (Permeate-gap membrane distillation): It is the advanced
version of direct contact membrane distillation. In this a third
channel is introduced by a non-permeable repulse. The advantage
of Permeate gap membrane distillation is separation of coolant and
the distillate, by using a condensation surface the permeate
channel is separated from the condenser channel. It uses the feed
water directly as cooling liquid inside the module thus only one
heat exchanger is required to heat the feed [42]. The main
drawback of PGMD is very poor heat conduction to condenser wall
from the membrane surface is due to less ow velocity of the
permeate in the permeate gap.
SGMD (Sweeping gas membrane distillation): It is also known as
air stripping membrane distillation. Same as in AGMD, SGMD uses
a path conguration with an empty gap on distillate side. With the
help of an external condenser, vapour condensation taking place
outside the membrane distillation module. With this process we
can distillate volatile substances with low surface tension. The
advantages of SGMD compared with AGMD is the considerable
reduction of barrier to mass transport through forced stream but in
this a small volume of permeate diffuses in large volume of
sweeping gas thus it requires a larger condenser to function
[37,40]. We can overcome the above drawback by adding a coolant
wall to permeate channel and maintain the temperature in the
desired rate by ushing air [42]. TSGMD (Thermostatic sweeping
gas membrane distillation) is a combined process of SGMD and
AGMD
Vacuum Membrane Distillation (VMD): VMD is fully based on air
gap conguration. The vapour is passed through the membrane
and it is sucked out by a vacuum pump at the permeate channel.
Condensation is taking place outside the module. The main
advantage is heat loss due to conduction is minimised [31,43]. It
can be used for removing organic components from liquid
solutions.
Vacuum multi-effect membrane distillation (VMEMD): Similar to
MED, VMEMD also provides energy saving at lower investment
cost. It consists of a steam raiser and evaporation-condensation
stages. Permeate is produced in each evaporationcondensation
stage and in the condenser. The exchange of heat produced by each
external source is taking place in steam raiser. Water present in the
steam raiser is lower than the ambient [43]. Evaporation-
condensation stages are composed of alternative hydrophobic
223
Cons

Highest thermal polarization

Distillate quality is highly depend on the pore wetting

Relatively lower ux than VMD

Mainly used for aqueous solutions

Additional resistance due to air gap for mass transfer

Complex design

Lowest gained output ration

Higher fouling

Lower selectivity in volatiles

An external vacuum and condenser is required in distillate side

Complex system

Heat recovery is difcult

Sweep gas stream cleaning and drying required
membrane and foil pp (polypropylene) frames. The input feed is
given to stage1 of the module. The feed ows serially through the
E-C (evaporation-condensation) stage and it is ejected at the last
stage of the module as a saline solution [43].
Flux, volatility and the permeate composition are the normally
used key features for the selection of membrane distillation. To
remove a volatile organics or dissolved gases from an aqueous
solution, normally SGMD and VMD are good. The most versatile
membrane distillation technique is AGMD and membrane
distillation technique with less number of equipment and easy
to operate is DCMD. The above two methods are best if water is the
permeate ux. Membrane parameters also affect the membrane
distillation techniques. Commonly affected parameters are mem-
brane material, liquid entry pressure, membrane thickness,
porosity and pore tortuosity [37,40,42,44,45]. The basic congura-
tion of membrane distillation is shown in Fig. 3. Advantages and
disadvantages of each method are also described in Table 1.
2.3. Cyclic distillation
It is another important trend in the development of distillation
technology with improved performances, enhances the distillation
efciency through pseudo-steady state operation and it results an
energy saving upto 50% [4648]. Cyclic distillation can be obtained
by the stepwise periodic or controlled cycling or both together.
Vapour ow period and the liquid ow period are the two
important key operation of cyclic distillation. While considering
the throughput of conventional distillation column, the through-
put of the controlled cyclic distillation is twice higher than
conventional method. But we want to consider the practical
implementation difculties of cyclic distillation. According to the
literature reviews, it can be concluded as; there are so many
advantages for cyclic distillation compared with conventional
methods. They are smaller pressure drops on trays, due to high
separation efciency they have increased quality of products,
column can be operated with maximum load, high throughput,
reduced energy requirement, high tray efciency etc. [48]. The
basic working principle of cyclic distillation is shown in Fig. 4.
2.4. Cryogenic distillation
It is one among the method used for air separation, separates
atmospheric air into oxygen and nitrogen (primary components). It
co-produces argon also. The application of cryogenic distillation is
224 V. Aneesh et al. / Chemical Engineering and Processing 104 (2016) 219242

Fig. 4. Cyclic Distillation: Liquid Period (Centre) alternating with vapour period (left and right).

coal gasication and steelmaking [49]. In cryogenic liquication
process, pure gas can separated from air by cooling the air until it
become liquid. After that it is distillates at their boiling points. This
process can produce highly puried gases but it is energy intensive.
To obtain good efciency in the process a very closely tight
integration of heat exchanger and separation column is required
[5052].
2.5. Reactive distillation column
Reactive distillation is a process in which reaction as well as
separation taking place in a single vessel. Consider a chemical
process with two reactants (A and B) producing two products (C
and D) which is in liquid phase and is reversible.
A+B$C+D

Fig. 5. (a) Reactive Distillation Column; (b) Reactive dividing wall column.
 V. Aneesh et al. / Chemical Engineering and Processing 104 (2016) 219242
For the working of reactive distillation we want to remove the
product from the reactants using distillation process. It implies
product should be heavier or lighter than reactants. An ideal case
for the above statement is, one product is heaviest and the other
should be lighter. Column pressure is the one of the most
important design parameter of reactive distillation column.
Pressure effects are one of the strongly marked one in RD column
than a conventional distillation column. In the case of normal
distillation, low column pressure is selected so that we can made
separation easier. Low pressure suggested a low temperature
coolant and low reux drum temperature. In RD column, column
temperature affects both chemical kinetics and phase equilibrium.
RD column is differing from the conventional columns by the
product compositions and reaction conversion specications.
Many designs adjust degrees of freedom of the system to obtain
these specications. Another design aspect of RD column is tray
holdup. This tray holdup has no effect on tradition column design.
It will not affect the steady state condition of the column but it will
affect the dynamics of the system. In RD column reaction rate is
directly proportional to the tray holdup in each tray therefore tray
holdup is very important. The main limitations of RD columns are
temperature mismatch, unfavourable volatilities, slow reaction
rate and other restrictions like RD column is limited to liquid phase
reaction and the dry up of tray by high exothermic reactions [12].
The schematic of reactive distillation process is shown in Fig. 5.
Over 2000 articles and 800 US Patents for reactive distillation
were published during the last 40 years [12]. Fig. 6 represents how
RD column works with reaction types. The most important
application of RD column: hydrolysis, polyesterication, ether-
ication, hydration and dehydratino, (trans-) esterication, (trans-)
alkylation, isomerisation, (de-)hydrogenation, amination, conden-
sation, chlorination, nitration [5355].
Control of reactive distillation column is a challenging task due
to its nonlinear nature and complex interactions (between vapour
Fig. 6. Reaction Types for RD.
225
liquid equilibrium and chemical reactions). Commonly used
control methods in reactive distillation column ranging from
Proportional Integral controller to Advanced model predictive
control such as dynamic matrix control (DMC), Advanced DMC,
Quadratic DMC, Robust multivariable predictive control and
Generalised predictive control. High attention is required for the
design of an optimal control and controller performance assess-
ment with optimal performance, energy saving and cost effective-
ness. The main objective of the controller is to maintain the
product purity within the desired range.
Backstepping based control actions are also carried out in the
eld of engineering [56,57]. Conventional chemical processes are
having sequential design approach. These designs are mainly based
on the steady state economic calculations and synthesis of
economic structures. Sometimes it shows poor dynamic operabili-
ty in the case of uncertainty and disturbances. By the above,
investigation on interactions between process design, control and
operability to improve the dynamic performance is essential. The
work in the area of simultaneous design and control of RD column
is very limited. Simultaneous design and control of ethylene glycol
reactive distillation is described in [58]. The application of Genetic
Algorithm in the area of reactive distillation column in described in
[59].With minimum usage of reboiler, how to do optimisation of a
thermally coupled RD is described in [60]. and optimisation
methods for a RD process is described in [61].
3. Optimisation based simultaneous design and control
In the current era, chemical process design and process control
are two separate areas assisting process developments. Design and
control decisions are made separately against the common
objectives of depleting the effect of disturbances and uncertainty
to offer robust plant operations. Nowadays it is not a desirable
situation because these types of control actions are very difcult to
226 V. Aneesh et al. / Chemical Engineering and Processing 104 (2016) 219242
control. As a consequence of this problem, a different way to take
controllability related issues in the process plant design has been
developed and stated in the articles during the past few decades
and is called Integrated the design and control or Simultaneous
design and control.
The modern chemical processes are highly integrated and the
process is interacted together to obtain high efciency of heat/
mass transfer and decreases the capital investment cost. The main
characteristics of the integrated process are two; 1.the process is
nonlinear in nature 2. The integrated process is a complex network
(process is interacted with more than one number of intercon-
nected systems) [6264]. Conventional approach to the design of
chemical process is a sequential one, in which sequence by
sequence design is evaluated and the output of the rst sequence is
used as the input to the second sequence [65]. Initially all the
process is designed in the steady state conditions. After establish-
ing the main system we are designing the controllers for the plant.
This may lead to poor dynamic operability in the case of external
disturbances and uncertainties. In this case, developer needs to
improve the performance and dynamic characteristics of the
control system to obtain an optimum performance [66].
3.1. Need and importance of simultaneous design and control
When going through the word Simultaneous Design and
Control, design of a chemical process means process characteristics
like owsheet topologies, design, operating parameters and
nominal values based on the steady state conditions and on the
other hand, the word control is pointing towards an optimum
control solution of the chemical process based on control system
design. Under the effect of external disturbances, the theoretical
and plant wide process has different optimum solutions. There are
some conicts between the inherent characteristics of the
chemical process and design parameters. When the input feed
and process specications are xed, the process shows some highly
nonlinear characteristics including limit cycle, oscillations, hyster-
esis, input/output multiplicities etc. A methyl methacrylate
polymerisation reactor problem is used [67] to describe the
conict between inherent characteristics and design parameters.
With the increase in volume of reactor the process becomes
unstable, i.e. if we have a volume of reactor V = 0.2 the process
shows open loop stability and minimum phase behaviour at the
operating points. While increase in the volume V = 0.5, the process
become unstable and shows non-minimum phase behaviour at the
operating point [68,69]. To solve the above instability a complex
control action is also required. It is very clear that increasing the
volume of the reactor means capital investment cost is also very
high. From the above it is very transparent that, the design
parameters pointing to the most economic steady state conditions
are not the best from the nonlinear analysis of practical cases.
3.2. Upto the minute of optimisation based simultaneous design and
control
As per the works already carried out on the simultaneous
design and control, we can classify the methods into two
categories. They are a) design based on controllability indicators
b) design based on optimisation. If we are looking towards the
work done based on controllability indicators, a linear and
nonlinear model based systems are used for the design [70,71].
The main advantages of these types of designs are controllability
indicators are very easily computed on nominal models but it has a
disadvantage: the model give accurate information on some
specic steady state conditions only, not in the entire state of the
process [7278]. Also this approach fails to provide accurate
information on start-up and shutdown. In order to overcome the
drawback of controllability based approach a new method called
optimisation based approach is introduced.
To remain combative, we need to design a plant which should
be at a low cost with acrimonious safety requirements and
unsparing environmental regulations. The plant should meet
seductive dynamic performance criteria, controllability analysis
and tuning specications. Due to the complications afliated with
this problem, few methodologies addressing simultaneous design
and control are proposed. Accustomed approaches to simulta-
neous design and control can be proximately grouped as
Optimisation based approach, Heuristic based approach, Evolu-
tionary based approach and combinational of all above [7982].
Simultaneous design and control approaches listing issues such as
1.parameter uncertainty and robust design under external
disturbances 2. Control structures selection and tuning parameters
3. Select and formulate good optimisation approach and algorithm
[83,84]. Simultaneous design and control problem can be stated as:
Minimize cost function
s.t.
Process modelling equations,
Control Algorithm equations,
Process Stability constraints,
Process output variability constraints,
Process feasibility constraints,
The formulation of the problem can be written as


min F 1 a; b; c
Fa
; a; b; c 1
a; b; c F 2 a; b; c
Subjected to
f a; b; c; z 0 2
ha; b; c 0 3
ga; b; c  0 4
aL  a  aU 5
az0 a0 6
cz0 C 0 7
Where, a is a vector of state variables, b is a vector of process
variable and c is a vector of design variables. The vector of initial
conditions of state variable and design variables are represented as
a0 and c0The objective function (Eq. (1)) to be decrease includes
the combination of capital and operating cost (F1) and the
controllability measures (F2). The set of differential and algebraic
equality constraints are represented here with the notation f. Eqs.
(3)(5) are used for expressing additional requirements for the
process performance like equality and inequality constrains or
paths. Classication of existing techniques addressing simulta-
neous design and control is shown in Fig. 7.
3.3. Controllability index based optimisation approach (CIOA)
An economic cost function is dened for designing one
economically optimal chemical process that can run in dynamic
mode within the range of nominal operating points. The cost
function is a combination of process steady state economics and an
economic cost associated with controllability indicators. This
 V. Aneesh et al. / Chemical Engineering and Processing 104 (2016) 219242 227

done. Based on the economic potential and static controllability

Fig. 7. Classication of optimisation based simultaneous design and control
techniques.
method takes the benets of controllability indicators to calculate
the closed loop dynamic performance of the process. Normally
used controllability indexes are integral square error(ISE), Singular
value decomposition, relative gain array, worst case disturbance,
gap metric, closed loop controllability and Eigen value of Jacobian
matrix. Morari [85] developed a multi-objective optimisation
technique with a linear model using controllability index as
integral square error. The objective of the work is to minimise
economic cost and control cost. With the same objective and
optimisation technique, Palazoglu [86] developed a nominal linear
model with controllability index as singular value decomposition.
Pioneer work in the eld of controllability index based optimisa-
tion was listed in Table 2.
For checking controllability index based optimisation approach,
different types of optimisation techniques have been proposed. In
the way that through taking into dynamic control performance in
the contour of matrix norms or dynamically determined control
errors. Floudas [90,101] proposed one methodology for analysing
the design and control interaction as a multi-objective mixed
integer optimal control problem. Therefore simultaneous making
of process design, controller structures and controller tuning can
characteristics for evaluating process ow conguration and
control structures [102], introduced new method [103]. proposed
a new framework for optimisation; in this case, economic
optimisation was stated based on the process design variables,
designing parameters and external disturbances affecting the
process. Different types of optimisation techniques developed and
are commonly used with controllability index based optimisation
approach are multi-objective optimisation, state steady-state
optimisation, mixed integer nonlinear programming, gap metric
minimise and iterative dynamic decomposition [104109].
In the foregoing methods, controllability indicators are treated
as a must of for optimisation problems. Process design and control
aspects are integrated in this methodologies but it fails to treat the
dynamic behaviour of the plant and competing objectives.
Together with, methods check thoroughly the closed loop
controllability properties. With the above limitations, this method
fails further developments with nonlinear models [97
103,110,111]. The application of multi-objective optimisation based
simultaneous design and control to check process may nish in a
suboptimal design, on account of the use of appropriate selection
of cost function and linearity assumptions. A coordinate strategy
pointing towards the integration of model predictive control with
process design scheme was developed by Brengel and Seider [112].
This work evaluates the economic objectives and alternate design
framework for controllability checking based on ISE (integral of
squared error) for a closed loop system with high external
disturbances. In this method, controlling of the process is takes
place with complete utilisation of the mathematical model and
advanced model based control, are the key advantages of the work
compared with Table 2.
3.4. MIDO based approach for simultaneous design and control
While designing a process the need of checking/considering the
operability is a widely accepted one in industry and academia. A
number of optimisation technologies have been introduced to
address this issue. Simultaneous design and control of a chemical
process is treated as a mixed integer dynamic optimisation with
both continuous and discrete parameters in the same framework.
A hand off dynamic optimisation based approach has been
proposed for considering the need of process operability issues
at the beginning of the process design. Mainly mixed integer

Table 2
Pioneer works in the eld of controllability indicator based SDC.

Authors Type of model for Controllability index Objective function Optimization

optimization
Morari [85] Linear ISE Minimise an economic and control cost Multi objective optimization
Palazoglu [86] Nominal Linear SVD Minimise an economic and control cost Multi objective optimization
Palazoglu [87] Nominal Linear SVD Minimise an economic and control cost Steady state optimisation (2 stage)
Figueroa [88] Nominal Linear Structured SVD Maximise an economic objective Multi objective optimization
Perkins [89] Nonlinear WCD Minimise design cost Steady state optimisation
Floudas [90] MINLP Models RGA Minimise capital, operating and control cost Multi objective optimization &
MINLP
Pistikopoulos MINLP Models Structural Controllability Minimise total anual cost and operating cost Multi objective optimization &
[91,92] MINLP
Perkins [93] Nonlinear WCD Minimise operating cost & cost of control MILP
instrument
Zeng [94] Nominal Linear Controllability index Minimise annualized cost steady state optimisation
Lee [95] Nominal Linear Gap metric Minimise gap metric Gap metric minimise
Jorgensen [96] Nominal Linear RGA Minimise closed loop error Steady State Optimisation
Blanco [97,98] Nonlinear Eigen Value of Jacobian Minimise the capital and operating cost. Steady State Optimisation
Matrix
Alhammadi [99] Nonlinear RGA Minimise economic and control cost Multi objective optimization and
MINLP
Douglas [100] Nonlinear SSV Minimise economic cost Steady State optimization

ISE, integral of the squared error; SVD, singular value decomposition; RGA, relative gain array; WCD, worst case deviation.
228 V. Aneesh et al. / Chemical Engineering and Processing 104 (2016) 219242
dynamic optimisation approach is used in batch process synthesis
[113115], solvent design in batch process [116], optimisation of
continuous or discrete systems [117,118], biochemical process
[119,120], reduction in kinetic mechanism [121,122] etc. Up-to-
date development in synthesis, design, operation and control of
process have created the need of strong and advanced numerical
algorithms for optimising a system [123125].
Mixed Integer dynamic optimisation problems are coming in
consideration when the binary decisions are getting into a dynamic
optimisation problem. According to optimisation views, dynamic
optimisation problems without binary variables are unintegrated
into: simultaneous approach and sequential approach. Simulta-
neous approach includes orthogonal collocation of nite elements
and multiple shooting. Sequential approaches are Control vector
parameterisation and Adaptive control vector parameterisation. In
sequential approach control variable discretization is taking place
and in simultaneous both control and state variable discretization
is taking place. Thus the differential algebraic equation is
converted to nonlinear programming and optimisation taking
place. A mixed integer nonlinear programming is converted to
MINLP (mixed integer nonlinear programming) and can be solved
by branch & bound algorithm (BBA), outer approximation (OA) and
generalised bender decomposition (GBD). Pioneer works listing
the discretization and solution to MINLP is listed in Table 3.
While comparing with continuous dynamic optimisation,
nding a solution for MIDO approach is very complex because
of the presence of binary variables. But due to the availability of
advanced dynamic optimisation algorithms, MIDO is widely
admired by the researchers. Orthogonal collocation on nite
element is the common methodology widely accepted by the
researches, which used for the discretization of MIDO approach
because both control and state variables are discretized thus.
Finding a reliable and feasible solution using orthogonal colloca-
tion is not that much simple because after discretization problem
generates a large number of variables and equations. Thus number
of parameters to be optimised is also increased in large scale
problems. Therefore, orthogonal collocation is pull back to small
scale problems.
Avaraam [117] and Mohideen [131] solved MIDO with the help
of orthogonal collocation of nite elements and outer optimisation
and GBD respectively. Nevertheless, linearization of
Table 3
Methods addressing Mixed Integer Dynamic Problem.
Authors Method of discretization
Androulakis [126] Orthogonal collocation on nite elements
Giovanoglou [127]
Avraam [128]
Biegler [129]
Bahri [130]
Mohideen [131]
Dimitriadis [132]
Flores-Tlacuahuac [133]
Flores-Tlacuahuac [134]
Terrazas-Moreno [135]
Terrazas-Moreno [136]
Chaowei Liu [137]
Oldenburg [138] Adaptive control vector parameterization
Bush [139]
Chatzidoukas [140] Control vector parameterisation
Hirmajer [141]
Bansal [142]
Bansal [143]
Allgor [125]
Sharif [144]
Sager [145] Direct multiple shooting
Sager [146]
Prata [147] Hybrid discretization method
nonconvexities at impracticable points in full discretization
method can reduce the algorithm to an often improvised scheme
when used in concurrence with a convex MINLP solver [114]. By the
use of control vector parameterisation and sensitivity arguments
for solving primal problems with xed binary variables, MIDO is
solved [101,125,142]. Thenceforth, an adjoint problem is analysed
to deliver dual information required to develop a relaxed master
problem, which sets a new framework for the next primal problem
and is solved using reduced space method. It has a reward of
reduced space method for solving the primal problem and has a
disadvantage of quite expensive intermediate adjoint problem. To
overcome the limitations Bansal [148] introduced a new approach,
which treats integers with a simplied master problem using
control vector parameterisation and GBD. Based on this MIDO is
divided into a sequence of primal problem and relaxed master
problem. But this problem excludes large area of feasible region
due to nonconvexities of constraints and a suboptimal solution in
the feasible region.
In sum-up, solving mixed integer dynamic optimisation is of
two stages of work: Converting MIDO problem into MINLP
problem and treatment of MINLP through algorithms. Stochastic
and deterministic global optimisation is used to reduce the
incidence of convergence to suboptimal problem. Deterministic
method banks either on decomposition strategy or on branch and
bound method [118]. Chachuat [124] developed a novel decompo-
sition approach for all general class of MIDO approaches. But it can
nd a global solution for MIDO but it fails to meet the total tally of
discrete alternatives in the dynamic optimisation sub-problems.
Stochastic method proposes a genetic algorithm for MINLP solver
results a global solution with relative efciency [149152]. But
these methods cannot guarantee global optimality and fails to
produce the results with highly constrained problems.
To eliminate the drawback of controllability index based
optimisation approach, a number of MIDO approaches are
introduced. The new approach combines control structure selec-
tion, tuning parameters with process design.
3.4.1. Framework for simultaneous process control design
The problems deals with the integration of process design,
control and operability are conceptually posed to minimise the
Methodology used for solving MINLP
Variant 2 GBD
Variant 2 GBD
Outer Approximation
Outer Approximation
Outer Approximation
Generalised Bender Decomposition
Generalised Bender Decomposition
Branch and Bound Algorithm
Using CPLEX to solve MINLP
SBB solver in GAMS
DICOPT based on outer approximation
CEPLEX and CONOPT in GAMS
CPLEX for MILP, Outer approximation
Outer Approximation(Augmented Penalty Extension)
GBD, CPLEX in GAMS for MILP
MISQP
Variant 2 GBD
Generalised Bender Decomposition
Screening model technique
Outer approximation
Outer convexication, relaxation
Combination of Integral approximation and branch and bound.
Outer approximation
 V. Aneesh et al. / Chemical Engineering and Processing 104 (2016) 219242 229

expected annual cost. A general algorithm for the above problem is Step3: Test the control and design from step2 for all the possible
summarised as follows [153]. ranges of the uncertain parameters by solving dynamic feasibility
Step1: Choose initial set of conditions for the uncertain test problem.
parameters
x maxu minuv maxl2q;z20;zf  gl  13
Step2: For the present set of conditions, calculate the optimum
points for the process control and design by solving Mixed Integer Subject to
Dynamic Optimisation
f a :f d :f 0 : 0 14
min X
ns

i
i i i i i
w i :fx z ; x z ; x a z ; va z ; vu ; u z ; d; d ; z xd; d  0, achievable operation can be guaranteed dynamically
d; d; u1v ; u2v ; ::::uns
v i0
f f f f f f

for all uvalues within the range. Algorithm abort in this case;
8
alternatively the solution denotes a critical condition that is added
Subject to with the existing set of conditions before going to step2.


A common mixed integer dynamic optimisation (MIDO)
i
f xl z; xi z; xi z; vi z; vi ; u z; d; d 0 9 formulation similar to:
d a u
minuv ;d fx
zf ; xzf ; xa zf ; uv ; d; d; zf 15

Subjected to
i
f a xl z; xi z; xia z; vi z; viu ; u z; d; d 0 10
0 f a x
z; xz; xa z; uv ; d; d; z 16

i
gxl z; xi z; xia z; vi z; viu ; u z; d; d  0 11 0 f a xz; xa z; uv ; d; d; z 17

i 0 f 0 x
z0 ; xz0 ; xa z0 ; uv ; d; d; z0 18
f 0 xi z0 ; xi z0 ; xa i z0 ; va i z0 ; vu i ; u z0 ; d; d 0 12

z0  z  zf ; i 1; 2; 3; ::::ns
Where the x denotes the differential state variables of the system,
xa for algebraic variables like thermodynamic variables, internal
ow rate, d 2 Y  f0; 1gNd denotes the binary variables for the
process,v denotes the vector of manipulating variables, uv is set
points for controllers, d is design variables but these variables
cannot be modied during the operations, i is the index set for the
conditions of U(t) (uncertain parameter) which is both time
varying and time invariant, ns is number of initial conditions, wi,
i = 1 . . . .ns are the discrete probability of the conditions
X ns
( wi 1), fd represents the mass and energy balance differential
i1
equations, fa represents the algebraic equations to hydraulic and
thermodynamic equations, fo is the initial condition of the
system,f represents the performance economic index.
0  gx
zf ; xzf ; xa zf ; uv ; d; d; zf ; z0  z  zf 19
The practical implementation of the above problem with a PID
controller is done in [154]. One of the main features of MIDO
approach is process nonlinearity, uncertainty, external disturbance
with time. Hence for a set of nominal working conditions; it is easy
to detect the scenario that generates the largest possible process
output error called Worst Case Scenario. Simultaneous design and
control problem is solved with predicted worst case scenario. Huge
number of methodologies addressing the eld of dynamic
optimisation based simultaneous design and control and the
application of these methodologies are proposed, few pioneer
works are listed in Table 4.
The main drawback of MINLP is that, there is no accurate
guidance for changes in the process and control design [143]. A
linear model is used in back off framework. But it can be used with
nonlinear models also with proper investigations. In effect with

Table 4
Simultaneous Design and Control based on Mixed Integer Dynamic Optimisation.

Author Application Controller Key Features

Mohideen [142] Ternary Distillation Column PI Mixed integer stochastic optimal control formulation, multi period decomposition
approach
Bahri [130] Two series CSTR PI Back off minimisation
Bansal [155] Binary Distillation PI [142] in a rigorous distillation model
Ross [156,157] High purity industrial distillation PI IMC Simplied structure of process control with integer decision combinations
column
Kookos [158] Evaporator System Multivariable Stochastic Mixed Integer Dynamic Optimisation
PI
Bansal [148] Distillation column PI Multicomponent Dynamic Optimisation
Panjwani [159] Reactive Distillation PI High delity dynamic model usage under uncertainty
Asteasuain [160] Polymerisation Reactor Multivariable SDC for optimal grade transition operation
PI
Lopez-Negrete [161] Binary Distillation PI Fully discretized approach
Nikolaos [162] CHP system PI Dynamic design and regulatory control optimisation
Yunfei Chu [163] Batch Process PI MINLP and GBD
Pu Li [164] Oil Spill Multivariable Dynamic Optimisation, Fast, Efcient and Cost effective
PI
Vida Meidanshashi Semi-continuous distillation Multivariable equation oriented gPROM and 1.03CPU hours
[165] PI
230 V. Aneesh et al. / Chemical Engineering and Processing 104 (2016) 219242
parametric uncertainties and disturbances, Mohideen [166]
introduced a new framework to obtain integrated process and
control system design with dynamic performance under optimal
conditions. Later Mohideen introduced a new and advanced model
with joint benets of control vector parameterization and
orthogonal collocation on nite element method [167]. A block
representation of software implementation of above stated one is
shown in the Fig. 8. In the primary problem, differential equations
of the model are substituted by discrete time implicit equations.
The evaluation gradients are calculated from efcient adjoint
based arguments and dynamic optimisation is done only over
reduced space of control variable [168].
In order to reduce the computational complexities, a new
method with simple procedure without the calculation of dual
variables and adjoint information to take a shot at MIDO and it is
based on variant 2 of the benders decomposition (v2-GBD)
methodology [148]. This method is independent with type of
methods used for solving problems. The only guarantee benet of
this method is that it can be converged to optimum condition
under convexity conditions [169]. Normally most of the MIDO
based simultaneous design and control methods make use of
conventional Proportional Integral (Derivative) controllers with
single input single output (SISO) systems and do not consider
environmental safety and operational constraints [170]. Along
with conventional PI controller, multivariable PI controllers are
also used for simultaneous design and control because of the easy
tuning. But the multivariable PI produces higher rated perfor-
mance than the conventional one [158]. Introduction of MPC
(model predictive control) to simultaneous design and control
improves the process economics and controllability [83].
In a uid catalytic cracker unit, Loeblein and Perkin [171,172]
introduced an online optimisation based MPC controller. Taking
into consideration the parametric controller has the ability to
eliminate the online optimisation, Sakizlis proposed a new method
for advanced explicit parametric model based controller into SDC
framework using binary distillation column and an evaporator
[83,170]. In this case author compared both conventional sequen-
tial approach and simultaneous design and control approach and
concluded that through simultaneous design and control total cost
Fig. 8. Prototype software implementation of simultaneous design and control.
obtained is very less than the sequential approach. Implementa-
tion of a parametric controller to the design structure enables the
direct adaptation of stability performance criteria in the controller
deduction state [170].
Based on control vector parameterization, dynamic optimisa-
tion based simultaneous design and control approach only
discretized the variables to be controlled and integrate the
differential algebraic equation models at each looping. The
advantages of optimisation based simultaneous design and control
in binary distillation, high purity industrial separation plants are
high protability, better controller performance and efcient
transaction of constraints. But conventional sequential approach
fails to produce better results with unstable steady states. For
solving a MIDO problem with highly nonlinear system, full
discretization method requires additional attention on initialisa-
tion and in this model uncertainties are not taken into consider-
ation. To solve a large scale process, proper initialisation and
decomposition strategies are required to keep the CPU timing in a
reasonable level.
In the beginning stages, the simultaneous design and control
problem is solved by mixed integer dynamic optimisation with
realistic events and on the basis of nonlinear dynamic model the
methods show highly nonlinear characteristics at the optimum
region. The main impediment of this method is the computational
complexities blended with nonlinear dynamic optimisation
problems. Mixed Integer Nonlinear Programming results from
the full discretization methods pointing to some special charac-
teristic features like few integer variables and large nonlinear
programming problems. Therefore, MINLP problems want to
handle NLPs that is very much complex to solve. By the tight
and complex computational requirements this system is impracti-
cable for industrial uses. But the controller tuning obtained using
the MIDO approach shows good stability in the closed loop
feedback unstable process [173,174].
A distillation based process intensication technique conceived
for the purication of desired products from a ternary mixture is
called semi-continuous separation. Comparing with conventional
continuous process, semi-continuous process has less annualized
cost. A methodology is proposed in [165], for obtaining the optimal
design parameters like number of column stages, feed tray location
and optimum tuning parameters for PI controller. With the help of
deterministic optimisation packages using MIDO based gPROMS
software is used for the simultaneous design and control of a semi-
continuous process and method resulted a minimised annual cost
than its conventional methods.
Integrated design and control of reactive distillation column is
presented in [175]. Simpler graphical approach method which is
very close to non-reactive process such as McCabe-Thiele method
and driving force approach are used here. Analytical solutions and
closed and open loop simulations are used here to analyse the
reactive distillation column process at its highest driving force.
According to the analysis the reactive distillation column is less
sensitive to disturbances in the feed at its highest driving force and
has an ability to reject external disturbances.
A two-tier approach is proposed in [176], where a solution to
obtain optimal steady state operating point using economic based
optimisation problem and to minimise the transition time to new
operating point is optimised using dynamic optimisation. By the
analysis of active constraints, the plant design limitations are
inferred and are represented in an industrial nitrogen plant. Other
recent works contributing optimisation of chemical plant are [177
179].
3.4.2. Framework for model predictive control
Models used in model predictive control (MPC) are designed to
represent a dynamic system and common dynamic characteristics
 V. Aneesh et al. / Chemical Engineering and Processing 104 (2016) 219242
are large time delay and higher order dynamics, which cannot be
able to handle by PID controller, can easily control by MPC. Changes
in the dependent variable of the real dynamics, and process
variable focus and limit to calculate further variations in the
dependent variables. MPC system typically works with the initial
changes in the independent variable to be achieved and repeats
with the solution till the next changes required. MPC is included as
an advanced control strategy in SDC. In the literature, MPC based
approach reported for optimal design and control, deals only the
optimal sizing of the processing unit i.e. a xed process owsheet is
used for entire calculations. A model predictive control scheme is
proposed in this integrated process design and control method.
The algorithm consist of an online calculation of the anticipated
control moves by the solution of constrained optimisation problem
that reduces the tracking errors in the future time [180,181]. The
latest works in this area are [182184]. A linear discrete state space
model is selected as a process and is obtained by the rst principle
nonlinear model of the process [185]. The main advantage of the
proposed control scheme is the ability of handling constraints and
multivariable systems. At each sampling time, the model predic-
tive control algorithm is solving as follows:
! Based on the control effort for pairing manipulated and controlled
XHp HX
C 1
2 2
min k yt ijt  ysp kR k Dut ijt kQ
Du
i1 i0
s.t.
xt 1jt Ap; u0 ; ysp xtjt Bp; u0 ; ysp utjt 20
ytjt Cp; u0 ; ysp xtjt
umin < ut ijt < umax ; i 0; 1; 2; :::::::; Hc  1
ymin < ut ijt < ymax ; i 0; 1; 2; :::::::; Hp
Dumin < Dut ijt < Dumax ; i 0; 1; 2; :::::::; Hc  1
Where, t denotes the sampling point, yt ijtis a vector quantity
representing the output prediction at time t + i with a calculation
up to t. A predicted state with time t + i with measurement up to t is
represented in a vectorxt ijt. Dut ijtis a vector of manipu-
lated variable moves with measurement up to t. umin and umax are
the lower and upper boundaries of the manipulated variable.
ymin and ymax are the lower and upper boundaries of the output
variable.HC and Hp are the control and prediction horizon. R and Q
are the positive denite matrices representing the weight of
control moves and set-point tracking errors respectively. A,B and C
are the state-space matrices of the linear model with process
design parameters p, steady state condition of the manipulated
variable (u0 ) and the set-pointysp . In this case matrices Q and R are
the time independent diagonal matrices. Other advanced control
strategies considering only the dynamic feasibility and asymptotic
stability of the process using MINLP. The MPC based system will
additionally deal the robust feasibility analysis and robust
asymptotic stability analysis, are formulated as a complex convex
problem and are effectively solved.
Under the assumption of random time-varying variables, robust
dynamic feasibility analysis estimates huge worst case variability
for the process variable in the process feasibility constraints. This is
one important advantage of the work [186] compared to the
others. Previous works deal with optimization, included structural
decisions in the analysis and it requires a time-dependent function
with unknown parameters. The algorithm used in MPC is very
general. It is independent of user dened disturbance function and
has the ability to handle the minimum and maximum frequency
magnitude bounded realization in the disturbance. Implementa-
tion of an MPC based simultaneous design and control requires the
231
identication of an uncertain model working around the nominal
operation region and it enables the formulation of robust stability
and feasibility test. The critical disturbance trajectories calculated
from the nonlinear model may result in constraint violations
because of the deviation of critical time-discrete realization of the
MPC based model from those obtained through full nonlinear
closed loop process model. The direct identication of critical
disturbance trajectory is a complex task for even a simple process.
The critical disturbance trajectory can be able to compute using an
uncertain impulse response model which adds an additional
computational cost to this method may become signicant for a
complex nonlinear system with a large number of constraints. The
ongoing method may show conservative design congurations due
to the usage of the uncertain process model. The present method
may not come to the point excessively moderate designs through
this precise aspect of the method depends on the trim of
complexity and nonlinearity. A new approach is proposed in
MPC based control structure to take care of both centralized and
decentralised control strategies together for solving optimal
control structure of the process. System dynamic performance
and constraints can be calculated from a robust variability test.
variables, two different optimisation approaches with different
calculations are used for the calculation of communication cost. A
nonlinear MPC (NMPC) and real time optimisation (RTO) based
approach for an ideal reactive distillation column is presented in
[187], aiming optimal inferential control of product concentration
in the face of reactant ow disturbances results a better optimum
operating point. An MPC based work for waste water treatment is
presented in [188]. In [189], how the deterministic and stochastic
optimal design work with reactive distillation column is stated.
3.5. Framework for simultaneous design and control under
uncertainty
Dynamic systems are always affected with rapid and unimag-
inable variations in the process characteristics and model
parameter uncertainties. These changes affect the operability
and performance of the process. The pivotal objection in the case of
simultaneous design and control is the assessment of the largest
variability (worst case) in the process variable due to uncertainty
in the system parameter and realizations in the disturbances
(worst case scenario).Distribution analysis in the worst case
scenario is discussed in [190]. For the above analysis, a new
method called probabilistic approach is introduced in the presence
of random perturbations. The evaluation is done with user dened
probability limits. Conventional approaches in simultaneous
design and control are conservative and expensive in design but
this method deals a exibility to assign probabilities to worst case
variability. By the above reason, the current method deals with
multiple process design and each evaluation is taking place at
different level of signicance.
Another worst case approach included with process synthesis
and control structure decision for an optimal control of dynamic
system under uncertainty is presented in [191]. The new method
integrates the dynamic exibility and feasibility in a single
optimization formulation. This reduces the cost of design and
checks the robustness in the presence of magnitude bounded
perturbations. A stochastic based simultaneous design and control
under uncertainty for a chemical process is described in [192]. For
getting a feasible and stable optimal process design, here one
advanced model based control scheme such as Model Predictive
Control (MPC) is used. The present method adopts linear
constrained model predictive control strategy to maintain the
manipulated and controlled variables within the limits in the
presence of disturbances. A methodology with robust feasibility
232 V. Aneesh et al. / Chemical Engineering and Processing 104 (2016) 219242
and stability analysis is invented under the category of design and
control under uncertainty [193]. It is a convex mathematical
problem; therefore, it does not require a solution of mixed integer
nonlinear programming to evaluate the robust stability and
feasibility. To evaluate the worst case deviation, a norm bounded
metric based structured singular value analysis is employed.
Robust stability test is taking place with the help of Lyapunov
theory and the process guarantees asymptotic stability.
Normally for worst case analysis, critical realization with
disturbance is considered. Latest development in this area is, by
considering the critical frequency of the inuenced disturbance
[194]. It is used to quantify the worst case variability in the
controlled variable via frequency response analysis. The key
feature of the work is, it explicitly addresses scheduling, design and
control simultaneously. Process disturbances are specied as
sinusoidal signals at critical frequency.
Under the category of process disturbance and uncertainty, a
new method is introduced in simultaneous design and control
[195] using power series expansion (PSE). The key principle of the
method is pull out from the optimal steady state design. The
method is pointing towards the calculation of various optimal
design and control parameters by solving a set of optimization
problems using power series expansion of the mathematical
expressions. These approximations are used to calculate the
optimal design parameters and operating conditions that is
essential for a dynamically feasible and economically attractive
process. The key challenge in the above method is, to determine
the magnitude of pull out needed to accommodate the feasible
operation of the process in the disturbance and uncertainty.
Integration of design and control, scheduling and design and
Control and scheduling, all have been core challenges. A drastic
improvement in this area has been achieved over the past few
years but there is no proper method or algorithm for such
integration is not commercially available. In [196], present a
software platform that enables integrated design and control. A
prototype software which is used for the representation, modelling
and simulation of integrated design, PAROC(PARametric Optimi-
sation and Control) [197], is shown in the Fig. 9, having features
like high delity dynamic model representation with sensitivity
analysis, parameter estimation and MIDO capability, model
approximation method toolbox, multi parametric program solvers,
Fig. 9. PAROC framework.
state-space modelling capability for scheduling and advanced
toolkits for Model Predictive Control and reactive scheduling.
3.6. Robust based optimisation approach
Previous work on controllability index based optimisation
approach and mixed integer dynamic optimisation approach,
process cost and control cost are incorporated into objective
function. Nonlinear dynamic model is used by MIDO approach
generates difcult and intractable problems. All the works are
pointing only towards the process uncertainty not to robustness of
the process.
By taking care of the above statements, a new approach called
Robust Control is considered with the rst order system plus time
delay linear one [198,199]. The idea behind the working of robust
based optimisation approach is seeking the worst case scenario via
robust control theory tools. The function of design and objective
variables are integrated together with operating cost and capital
cost. One of the newness in the methodologies is the cost of
variability in control variable and model error under model
uncertainty has been quantied specially. Constrains in the
optimisation framework is the robust performance criteria with
robust stability term and performance term related to closed loop
variability.
The proposed method is composed of an inner loop optimisa-
tion and an outer loop optimisation. The main use of the inner and
outer loops is to blow up the variability cost and attenuate the
capital and operation cost respectively. While working with IMC
and MPC, the proposed method pointing to only one particular
disturbance prole. By this, it is not good, cannot generate a work
capable of handling all external disturbances and uncertainty in
process [199].
The framework for robust based optimisation approach is
shown below:
The objective can be represented as follows:
min
CC d ; u OC ss dv ; uv C v dv ; uv ; p 30
dv ; uv ; p ss v v
s.t.
edv ; uv ; p 0
; p  0
idv ; uv
 V. Aneesh et al. / Chemical Engineering and Processing 104 (2016) 219242 233

Where dv is the design variable, uv is the operating variable, pis the Where v is the frequency of unmeasured disturbance w and T is
controller tuning parameter, CC ss is the capital cost at steady state the sampling period. Therefore, the transfer matrix can be dened:
conditions, OC ss is the operating cost at steady state condition, C v is h i
the process variability cost due to imperfect control actions, exis GetwT I  Ms  f n K MPC Mt etwT 1 f n K MPC 38
n ny
p nd
the equality constraints and ixis the inequality constraints.
The process variability cost, C v dv ; uv ; pcan be calculated based In the above model, disturbance is a sinusoidal wave with a specic
on a nominal model with uncertainty and model predictive control frequency. Assuming the disturbance is periodic and is used to
algorithm as follows: calculate the maximum output variability of the frequency
The nonlinearity of the process can be calculated from the response. Put i = 1 in the above equation and using
model uncertainty (i;j;h ). Usually for non-linear system, step ~
Yi w YiandY~ Dw, the largest variability for a linear
changes with different magnitude are used for the response process denoting a nonlinear system is calculated as follows:
analysis. For each step response coefcients, i;j;h as follows: X
m
max max max
Y k;1  j1 DYi;j;h Z etwT jjWj; li;j;h < li;j;h 39
norm
f i;j;h  f i;j;h li;j;h
k1
i;j;h norm 31
f i;j;h
A scalar value is that pointing towards the bound over the
norm
Where f i;j;h ; f i;j;h are actual and nominal step response coefcients ~
variability of output k at time step i = 1 is denoted here asY
max
k;1 .
for a combination of j input, i output at a time interval of h. The DYi;j;h is the transfer matrix in the rst ny rows of the Eq. (40) and W
largest deviation from the nominal value and lower/upper limit is represents the amplitude of the disturbance. The summation term
the largest model uncertainty, is normally represented asmax i;j;h in the above equation shows the prediction at each step is given by
up lw
andf i;j;h /f i;j;h are the lower and upper boundary coefcients. A the product of transfer function vector and the disturbance vector
w of length m.
robust model, f n , can be dened to detail the nonlinear dynamics of
nom If a sinusoidal signal with different frequencies is considered for
the process as a function of nominal linear model f n . Robust
the calculation of maximum variability as a function of frequency,
model f n has elements f i;j;h such that:
is represented as:
D nom
f i;j;h f i;j;h 1 i;j;h 32 k Y k2
k D k1 supDjv sup 40
v W2L2 k W k2
With respect to the Model Predictive Control algorithm, it is
assumed that the measured setpoint and disturbance vectors are The principle used for the computation of maximum output
zero, i.e. f d Ddi 1 0andRi 1 0. Where R is a vector of set magnitude of the frequency response is used to compute the
points and Ddis the measured disturbance at a time step of i-1. maximum variability in manipulated variables during its closed
Error at time t will be, loop operation. Based on the MPC algorithm, unmeasured
disturbances can be obtained as:
~  1  wi
ei Mt Yi 33
h i1 
DuZ  
K MPC Mt In
n  Ms  f n K MPC Mt Z 1
nom nom
~
where Yiis the prediction output vector, Mt is the truncated shift f n K MPC  IP
P
wZ
matrix and w represents the unmeasured disturbance vector at a 41
time interval of i.
The unmeasured disturbance vector is, The maximum value of manipulated variable move Dumax
j;h
can
h be calculated from the transfer function in the above equation as
~i
wi 1 Iny xny :::::::: Iny xny T yi  y 34
h follows:
where h is the prediction horizon and Iny xny is an identity matrix X
m
Dumax
j;h maxj DU
d;h;1 Z e
twT
jjWj  uh;max  uh;nom 42
with dimensionny xny . lj;h;1
d1
The model used for the analysis is,
where Dumax is a scalar value and it species a boundary over the
~ M
Yi ~ s Yi
~  1 f nom Dui  1 f nom Ddi  1 35
h;l
n d maximum variability of manipulated variable h at time i = 1.The
~ s is the shift matrix vector, f nom is the nominal step
where M maximum or the largest manipulated variable change is bounded
n
nom as,
response model vector, f d is the nominal step response model
vector due to measured disturbance, Duis the manipulated uh;norm Dumax
h;l  uh;max 43
variable at time step i-1.
Substituting the unmeasured disturbance vector and error at The robust stability condition developed [200] in terms of
~ structured singular value condition referred as #D analysis [201] is
time I in the system model, Yican be expressed as,
represented as:
Yi ~  1  f K MPC wi
~ Ms  f K MPC Mt Yi 36
n n #D Mtv  1 44
Rearrange the above equation and take the z transform, the where M(tv) is an interconnection matrix and is dened using the
transfer matrix between predictions to disturbance can be following equations:
expressed as,


#i 1 #i
~
YZ M 45
I  Ms  f n K MPC Mt Z 1 f n K MPC 37 yD #D
wZ
A sinusoidal signal is assumed for the unmeasured disturban-
ces. To compute a bound over the maximal value of variability in #D Du yD
the prediction output vector, the discrete frequency response can The above equation is a function of the Model Predictive Controller.
be used. Here, we remove the initial transient by settingZ elvT . Vector of manipulated variable is denoted here by#i. #D and
234 V. Aneesh et al. / Chemical Engineering and Processing 104 (2016) 219242
yD are the dummy variables and can be eliminated by combining
the above equations.
The quantity #D is dened as:
#D 1 M minfDjdetI  MD 0f orDl 2 Dg 46
Dl
The uncertainty element Du is a diagonal matrix with scalar
elements.
2 3
D1 . . . 0
. ..
Du 6
4 .. } .
7
5 47
0 . . . Dn nu ny
Each Dl in the above matrix represents a perturbation and it is
bounded by the condition, & D  1, where & denotes the
maximum singular value of the matrix D. This type of problems
(Structured singular value) can be easily calculated with the help of
Robust Control Toolbox in MATLAB.
3.6.1. Quadratic lyapunov function based approach (QLf-based
approach)
In this case, to represent the closed loop nonlinear dynamic
behaviour of the system, a linear state space model with a
predened range of model parameters is used [202]. The obtained
robust model is used to check the robustness and the limits of
worst case deviation in process variables in response to the
external disturbances. The formulated nonlinear constrained
optimisation problem is applied to a mixing tank process and
the obtained output is compared with the MIDO approach [166],
shows a level of magnitude faster than MIDO based simultaneous
approach. The main drawbacks of this approach are, nonlinear
process equations and control algorithms do not appear speci-
cally within the mathematical calculations, the process parameter
uncertainty and disturbance have uniform probability but process
variable follows a uniform distribution function, robust design is
cautious.
3.6.2. Singular structured value based approach (SSV based approach)
To seek the time dependent element in the disturbance
(produces large variation in the output), a new technique based
on the structured singular value is adapted. A robust FIR (nite
impulse response) model is used for the analysis of the worst case
scenario. Process asymptotic stability is checked with Lyapunov
function and included a LMI equation for optimisation constrains.
Identifying a robust model which produces optimal output at
nominal operating points is necessary for the analysis of process
stability at each and every iteration. In the case of large systems, it
is very complex and computationally expensive. To avoid the above
drawbacks, a new model is invented [203]. Two primary tests are
conducted for checking the stability of the system at nominal
operating states. To check the transient behaviour of the closed
loop nonlinear system a discrete closed loop linear impulse model
is used and for checking the system stability, quadratic Lyapunov
function based stability criteria is used.
Integrated process control and design with uncertain model
parameters and disturbances are discussed in [204]. A robustness
checking for bioreactor and continuous stirred tank reactor is
discussed in [205].
3.6.3. Hybrid worst case (HWC) approach
For calculating output variability and related variable con-
straints, analytical bound based approaches are commonly used.
By considering the conservatism used in the analytical bound, a
hybrid worst case integrating the analytical calculation of worst
case disturbance and dynamic simulation based SDC has been
introduced [100,206]. Structured Singular Value (SSV) norm is used
for the realization in disturbance variable and worst value in a
parameter and to nd the maximum variability in the plant and
variables, dynamic simulation in the closed loop systems are taken
into consideration. Application of robust based approach in the
chemical process which with stand to external disturbances and
process uncertainties results less computational burden than
MIDO based approach. Hence, computational efciency of hybrid
worst case approach makes this approach an efcient practical tool
used for SDC. Similar to other approaches, it also has certain pull
backs like: (1) the design is fully depending on the difference
between the nonlinear plant dynamics and the uncertain model,
makes the system conservative; (2) Through process knowledge is
required to initiate suitable initial values because of nonconvex
nonlinear constrained optimisation problem and variations in the
starting point results various design and control solution. (3) The
process model assumes xed owsheet and control design. (4)
Only conventional PI controller is implemented and tested in this
approach addressing simultaneous design and control. (5)
Application of robust based approach in a higher order process
with high nonlinearities required additional stability test because
the process is described by large set of nonlinear differential
equations.
3.6.4. Constructive nonlinear dynamics (CND) based approach
For a detailed understanding of quantitative nonlinear dynam-
ics of the plant, analysis of sensitivity, singularity and numerical
bifurcation analysis are commonly used. In numerical bifurcation
method process plant analysis is only takes place not any of the
plant synthesis. A novel approach is proposed by Marquardt based
on conservative nonlinear dynamics, which is take caring synthesis
and analysis of the plant [204,207213]. The proposed design
structure allows the formulation of constraints on asymptotic
process behaviours. Process feasibility, stability and dynamic
performance are the advantages of this method compared with the
conventional methods. In the space of process parameters all the
constraints are integrated as a nonlinear boundary which
separates desirable and undesirable behaviour of the chemical
plant. This framework is successfully tested for designing a process
[209,210], robust controller tuning [214,215] and SDC [216219].
This process is applied to a continuous mixed-product-removal
(MSMPR) crystallisation process by Grosh [216] for the analysis of
interaction between economic and dynamic performance con-
straints under uncertainty. Application of CND-based approach in
Tennessee Eastman process and crystallisation process has high
potential to overcome more complex and realistic scenarios. This
method requires a prior knowledge about parametric uncertainty.
Only PI controller is applied to the framework to manipulate the
process and model assumes xed owsheet and control design.
3.7. Embedded control based optimisation approach (ECOA)
Embedded control based optimisation approach is used to
reduce the combinatorial complexity of simultaneous design and
control [220224]. To manage the problem size comparatively
smaller, control decision and design decisions are separated. The
entire framework is decomposed into two sub-problems. The rst
sub-problem seeks the optimum design decisions that make sure
the process performance under uncertainty and the second sub-
problem is based on design decision, determines the dynamic
exibility of the process. This approach reduces the combinatorial
complexities. The main objective of this approach includes
conservation laws, control algorithms, safety, equipment and
production requirement, expected operating cost and capital cost.
This optimisation problem is given in detail by Malcolm [220]. The
entire system working is based on simpler adaptive state space
model and a linear quadratic regulator. The regulators are used to
 V. Aneesh et al. / Chemical Engineering and Processing 104 (2016) 219242 235

of discredited time boundaries. The quality of the system is

improved by the observer. At last, the optimal control actions are
taking place at regulator and it is represented in the Fig. 11.
A description regarding the embedded control optimisation is
follows:
Identier: A robust and fast algorithm which is completely
perfect for the dynamic input output data to represent the model in
state space representation is shown in Eq. (48). At a suitable time,
interval sampling of the nonlinear system model gives the required
input output data. Adaptive identication of the nonlinear model
gives the solution for Least Square tting problems. Very
inexpensive and cheap sequential algorithms are available for
the analysis [226228]. Eq. (48) correlates the present and past
states of the system with control actions through the estimated
^ Here the present state is denoted asxk, the past states are
vectorb.
denoted as xt  1; xt  2; :::::and the control actions
^ composed of scalars a
asut  1; ut  2; ::::. The row vector bis
and b.
xt Axt  1 But  1 48
The above equation can be written in a most compact form as,

Fig. 10. Hierarchy decomposition for exible design optimization. ^T

xt ztb 49
r

With zt xt  1::::  xt  nut  1:::::ut  n and b^

 
compute the best control action by that reducing a cost function. a1 ::::::an b1 :::::::::bm
The state space model is obtained using least-squares method The solution of the least square tting problem in the below
[218] or moving horizon estimation [223,224]. While applying this equation gives the desired state space model.
technique on a polymer reactor, better results are produced X r Xr
2
compared with MIDO approach. But for a highly nonlinear process, minDLSQ xt  ^
2
xt ^T
xt  ztb 50
^
b t1 t1
steady state identications are more complex. So the use of this
technique is limited to simple state space identication problems. From the previous model and new observations, a recursive
An advanced algorithm is developed by Patel called Modied ^ also possible. The nal formulae for the
calculation of the model bis
Linear Quadratic Regulator (mLQR) [225] for solving highly
model is given below,
nonlinear model.
^ b ^ T^
br r1 Sr t r xt  t r br1 with Sr
3.7.1. Framework of embedded control based optimisation T
Sr1 tr tr Sr1
There are two hierarchical optimisation loop present in the Sr1 51
1 tTr Sr1 tr
formation of embedded control algorithm, are Master Design
Optimisation Loop and Embedded Control Optimisation [220]. WherePr denotes the estimation of covariance of the error matrix.
Master Design Optimisation Loop: This stage deals with the
initial design guess and uncertain scenario as input as shown in 3.8. Observer
Fig. 10. At master optimisation level, the design decisions dd are
considered and are improved to increase the probabilistic cost The model identication gives a full state space representation
function. The system dynamics are designed using a set of of the system in each step with less computational efforts as shown
differential equations and the uncertainty sources are described above. The linear state space model is suggested to predict the
by mathematical methods, listed in Table 5. system response with past control moves of the disturbances. In a
Embedded Control Optimisation: It has three elements as real-time system, it reduces the noise. Here, it is an equation based
shown in Fig. 10: 1. System Identication 2. State prediction and 3. model so there are no noise calculations. This improves the
Optimal control. The system identication step is mapping theoretical background of the technique because it reduces
dynamically with equation based process model into a linear modelling errors. Using a Kalman lter, it improves the predictions
state space model and this model is executed in every time period between the nonlinear system model and its linear state space

Table 5
Categories of variables in Integrated Design and Control.

Variable type Symbols Type Example

Design dd Discrete Structural decisions: connectivity
Continuous Reactor volume and column length
Control C Discrete Control conguration and type
Continuous Controller tuning parameters and set-point
System state xs Continuous Temperature, Pressure and Composition
Uncertainty u Time variant Uncertainty of parameters
& t Time invariant Seasonal changes affecting variations
236 V. Aneesh et al. / Chemical Engineering and Processing 104 (2016) 219242

Fig. 11. Proposed embedded feedback controller with identier, observer and regulator.

approximation. Basic prediction step equations are shown below with xed integer values. MINLP problems are nonconvex and fail
as, to locate global solution with local optimisation technique. Many
assumptive methods are endorsed to solve optimisation problems
^
xt1 At ^
xt Bt ut 52
[229235] where the integer and continuous parameters are
treated simultaneously through selected algorithms like particle
swarm optimisation, Tabu search based algorithm and genetic
S T
t1 At St At Gt algorithms. Advanced algorithms like ant colony optimisation and
fuzzy modelling/control algorithms are also proposed for the
The measurement correction steps are,
development of simultaneous design and control of process
K t S T  T
t C t C t St C t Rt
1
53 [231,232,235]. Proposed Black box optimisation methods have
the ability to reduce the combinatorial complexity. But if we are
comparing this technique with normal technique it lacks the
 
^ xt K t yt  C t ^
xt ^ xt 54 theoretical background.
A fuzzy model is used to calculate a nonlinear process in a large
operating range. First, generate a local model, is a state space
St S 
t  K t C t St 55 model obtained at each operating points. The fuzzy logic is used to
interconnect these models into a single model called global model.
The process noise is accounted by the covariance matrix G. The The controlling of these fuzzy models are very easy because it is
covariance of the measurement error is denoted by the matrix R designed with the same fuzzy theory. Local approximation is
and in this case it is set to zero. The current weight for the adopted to construct a T-S fuzzy model for the nonlinear process
prediction error is denoted byK t . The matrix C is used to relate the shown below,
measured variable to the state variable as in Eq. (53).
Regulator. The best control action is measured with the help of a x
t Axt  Cx4 t Bdut ht 58
regulator. It is a relatively very easy task for a linear state space Where B(d) is the function of design parameter vector d.
model system compared with the previous steps and is achieved The nonlinear term in the above model is rewritten as
with the help of a Linear Quadratic Regulator (LQR), which selects
the control actions to reduce a cost function. The objective x4 t zt:xt 59
accounts for the loss of performance according to the matrix P and
The nonlinear process model is represented by fuzzy model as:
punishes the need of control action by the matrixj. With the help
Model rule s:
of these data regulator chooses one optimal combination of the
IF x1;1 t x1;1;s and . . . . . . xmp;mp t xmp;mp;s
control actions in terms of the gain matrixK . It guaranteed the
THEN x
A  C S xt Bdut ht; s 1; 2; 3; :::::; r
stability of the system and overcomes the challenge of control
Where xi;j;S is a fuzzy set,C S is equal to C.
variable pairing.  T
zv ; zv diag z1;1;S ::::::zmp;mp;S ,can be calculated in each rule
X
1
min

J xTt Pxt uTt jut 56 according to the denition shown below:
h
K
0
zt diag T 31;1 t:::::T 3i;p1 tT 3i;p tT 3i1;1 tT 3i1;2 t:::::::
u! K xt
s.t xt1 Axt Bu t y Cxt~ T 3m;p tT ; T r tis the reference with r is the number of model rules.

with ut ut1 ut and xt xt  Y sp (57)
Final state of the fuzzy system is as follows:
where Y sp is the set-point.
The main advantage of this method is the separation of model X
r
x
t wS xtA  C S xt Bdut htt 60
equations for design and uncertainty from the control strategies.
S1
The decomposition of the problem reduces the size and complexity
of the master problem. This method reduces the combinatorial Where,
complexity of integrated design and control and improves Y
mo
structural and parametric decisions. mS xt Nh;h;S X h;h t
h1
mS xt 61
3.9. Black box optimisation approach (BBOA) wS xt
X
r
mS xt
Several methods are introduced for addressing simultaneous S1
design and control, which can formulate at MINLP (Mixed Integer
The term Nh;h;S X h;h tis the grade of membership of xh;h tis
Nonlinear Programming) problems. These problems can be solved
by general benders decomposition, outer approximation or branch the Sth rule.
& bound algorithm. All these methods solve a nonlinear problem
 V. Aneesh et al. / Chemical Engineering and Processing 104 (2016) 219242 237

Fig. 12. Fuzzy model based fuzzy control-block diagram.

The robust tracking performance under external disturbance is
achieved by using H1 performance:
ZtS ZtS
 T  2
e tQetdt  eT 0Pe0 l wtT wtdt 62
o 0
Where, e(t) is the tracking error and tsis the end time. The
weighting matrix is denoted as0  Q C TQ C Q 2 Rmp
mp , an un- energy system, pharmaceutical manufacturing process and mir-
known matrix is denoted as0  P 2 Rmp
mp , wtis equal to
wt xr t x
r tT ; g  0 is an attenuation level. A fuzzy model
based fuzzy control is shown in the following Fig.12.
4. Challenges and future directions
Bidirectional development in the methodologies and its
application is the main advantage of the eld of Optimisation
Based Simultaneous Design and control. Signicant progress
achieved in the eld of optimisation based simultaneous design
and control produces relevant strategies and algorithms for solving
the chemical problem but it still lacks its efciency, predictability,
safety and secure behaviour for large nonlinear chemical systems.
Still the theoretical and practical implementation fails to meet
with the criteria. Advanced controllers always produce better
results than its conventional controllers. They offer more safe and
economic operations for the unsafe and safe chemical process.
Model Predictive Control (MPC) is the most commonly used
controller in the industries because of its capability of handling
multivariable systems. Implementing advanced controllers to a
chemical process improves the process protability and process
dynamic performance. There are no simultaneous design and
control methodologies simultaneously considering the ow sheet
congurations and operating point optimisation with robust
stability under the disturbance conditions. Certain, highly nonlin-
ear chemical process shows undesirable characteristics like
instability and non-minimum phase behaviour which would affect
the overall system performance under xed ow sheet. Therefore
it is very necessary to include process ow sheet structure and
stability within the optimisation. It will produce more energy
efcient design with desirable performance on controllability,
stability, exibility and robustness. Existing simultaneous design
and control strategies are relatively applied to small chemical
processes like distillation column, waste-water treatment process
and reactor operation etc. Further research on this area should
focus to extend its application to certain novel processes involving
complex chemical processing system, biomedical system, hybrid
oscale chemical process etc. [236].
While applying Mixed Integer Dynamic Optimisation (MIDO)
based simultaneous design and control to a complex chemical
process, the result is a stochastic and large scale MINLP (Mixed
Integer Nonlinear Programming) problem which is very difcult to
handle, hence, more efcient numerical solution is required, needs
further development. The main burden of the application of
optimisation based simultaneous design and control is its
computational complexity i.e. it requires a huge computation
effort to solve the problems. To prevent the development of an
economically undesirable design under uncertainty and make the
application of simultaneous design framework for an above stated
model, a new rigorous, robust and efcient method for MINLP is
required. A steady state process optimisation strategy [237] gives a
well efcient and robust stability guaranteed results for a complex
chemical process. Regardless of signicant improvements in the
eld of optimisation based simultaneous design and control,
favourable matter of fact industrial applications are infrequent. The
rst in rank justication for the developed techniques addressing
optimisation not piercing industrial practice is largely on account
of the deciency of commercial software packages. The developed
methodologies should be applied to a systematic computer aided
design to prosper a user friendly interface software tool which may
raise the possibility of SDC.
5. Conclusion
Based on the process intensication principle, distillation
technology has a lot of advantages like high energy efciency
and cost reduction. Membrane distillation gives grounds of
expectations for separation process as well as purication process.

Table 6
Summary of optimisation based SDC approaches.

Approach Model Performance indices Controller Control structure Flowsheet structure Stability analysis
Controllability index based Linear CN, RGA, ISE PI, MPC Yes Yes No
Mixed integer dynamic optimisation Nonlinear ISE, WCD PI, MPC Yes Yes No
Robust Based Linear Worst deviations PI, MPC No No Yes
Embedded control Linear Dynamic exibility PI No No Yes
Black Box Nonlinear ISE PI Yes Yes No

CN, condition number; ISE, integral squared error; RGA, relative gain array; WCD, worst case deviation; PI, proportional integral; MPC, model predictive control.
238 V. Aneesh et al. / Chemical Engineering and Processing 104 (2016) 219242
But it still lacks some industrial standards. To make it to industrial
level further optimisations required. The main advantages of
membrane distillation is low operating temperature, high rate of
rejection for salts and colloids, high concentration solutions are
possible to use in the feed and low hydrostatic pressure. But still
membrane distillation is not used in industrial applications
because of uncertain energy and economic cost and membrane
distillation modules are not available commercially. Low energy
requirement, increased column throughput and high separation
process are the advantages of cyclic distillation. This method gives
a new concept for old distillation process. This technique has good
process control (separation phase movement has more degree of
freedom). Special operator training and safety requirements are
required in the case cyclic distillation, because cyclic distillation is
a pseudo- steady state operation. Reactive distillation technology
made the distillation process more tremendous with low operating
and investment cost on equilibrium limited systems.
The backbone of optimisation based simultaneous design and
control is the reciprocation between the application area and the
theoretical domain. The imperfection in the sequential approach
highlights the necessity and advantages of simultaneous design
and control. Here the discussion already ended with 5 methods like
controllability index based, mixed integer dynamic optimisation,
robust theory based, embedded control based, black box optimi-
sation approaches. Controllability index based optimisation
approach is developed using controllability indexes like RGA, ISE
etc., normally used with linear models but it fails to deal with the
nonlinear models because of the lack of nonlinear dynamic
analysis and closed loop controllability checking capabilities. To
avoid the above drawbacks MIDO based approach came with
combined control structure selection and tuning parameters with
process design. This method has the capability to deal with both
linear and nonlinear systems. But before applying this method
with nonlinear systems prior investigation are required on
parameter uncertainty and external disturbances. Based on the
robust control theory a new optimisation approach developed with
worst case scenario called Robust based optimisation approach.
Quadratic Lyapunov function based approach deals with robust-
ness and worst case scenario of linear systems. Application of QLf
based approach has certain difculties with nonlinear systems
such as complex calculations are required for nonlinear process
equations and control algorithms. Summary of optimisation based
simultaneous design and control approaches are listed in Table 6.
Acknowledgments
The authors are grateful to the Editor and Reviewers for their
valuable comments and suggestions which improved the quality of
the manuscript.
References
[1] M.R. Resetarits, M.J. Lockett, Distillation, Enc. Phys. Sci. Technol., 3rd edn.,
(2003) , pp. 547559.
[2] H.Z. Kister, DistillationDesign, McGraw-Hill, USA, 1992.
[3] H.Z. Kister, DistillationOperation, McGraw-Hill, USA, 1992.
[4] H.Z. Kister, Distillation Troubleshooting, John Wiley & Sons, Hoboken, New
Jersey, 2006.
[5] A.A. Kiss, Advanced Distillation TechnologiesDesign, Control and
Applications, Wiley, Chichester, 2013.
[6] D.C. Freshwater, Thermal economy in distillation, Trans. Inst. Chem. Eng. 29
(1951) 149160.
[7] R. Taylor, R. Krishna, Multi-Component Mass Transfer, Wiley, New York, 1993.
[8] G.M. de Koeijer, S. Kjelstrup, Application of irreversible thermodynamics to
distillation, Int. J. Thermodyn. 7 (2004) 107114.
[9] F.B. Petlyuk, Distillation Theory and Its Application to Optimal Design of
Separation Units, Cambridge University Press, New York, 2004.
[10] D. Reay, C. Ramshaw, A.A. Harvey, Process IntensicationEngineering for
Efciency, Sustainability and Flexibility, 2nd edn., Oxford, Butterworth-
Heinemann, 2013.
[11] T. van Gerven, A. Stankiewicz, Structure, energy, synergy, time the
fundamentals of process intensication, Ind. Eng. Chem. Res. 48 (2009)
24652474.
[12] L. William Luyben, Cheng-ChingYu, Reactive Distillation: Design and Control,
John Wiley and Sons, Hoboken, New Jersey, 2008.
[13] D.R. Summers, M. Pilling, Be smart about column design, Chem. Eng. Prog.
(2012) 3238.
[14] A.A. Kiss, S.J. Flores Landaeta, Infante Ferreira CA: towards energy efcient
distillation technologies  making the right choice, Energy 47 (2012) 531
542.
[15] M. Nakaiwa, K. Huang, A. Endo, T. Ohmori, T. Akiya, T. Takamatsu, Internally
heat-integrated distillation columns: a review, Chem. Eng. Res. Des. 81
(2003) 162177.
[16] T. Fukushima, M. Kano, S. Hasebe, Dynamics and control of heat integrated
distillation column (HIDiC), J. Chem. Eng. (Japan) 39 (2006) 10961103.
[17] A. de Rijke, Development of a Concentric Internally Heat Integrated
Distillation Column (HIDiC). PhD Thesis, TU Delft, The Netherlands, 2007.
[18] O.S.L. Bruinsma, T. Krikken, J. Cot, M. Saric, S.A. Tromp, Z. Olujic, A.I.
Stankiewicz, The structured heat integrated distillation column, Chem. Eng.
Res. Des. 90 (2012) 458470.
[19] K. Matsuda, K. Kawazuishi, Y. Kansha, C. Fushimi, M. Nagao, H. Kunnikiyo, F.
Masuda, A. Tsutsumi, Advanced energy saving in distillation process with
self-heat recuperation technology, Energy 36 (2011) 46404645.
[20] K. Matsuda, K. Iwakabe, M. Nakaiwa, Recent advances in internally heated
integrated distillation columns for sustainable developments, J. Chem. Eng.
Jpn. 45 (2012) 363372.
[21] Process System Enterprise Ltd., gPROMS Process Model Library
documentation, 2011.
[22] K. Iwakabe, M. Nakaiwa, K. Huang, T. Nakanishi, A. Rosjorde, T. Ohmori, A.
Endo, T. Yamamoto, Energy saving in multicomponent separation using an
internally heat-integrated distillation column, Appl. Therm. Eng. 26 (2006)
13621368.
[23] B. Kiran, A.K. Jana, A.N. Samanta, A novel intensied heat integration in
multicomponent distillation, Energy 41 (2012) 443453.
[24] C. Feng, K.C. Khulbe, T. Matsuura, R. Farnood, A.F. Ismail, Recent progress in
zeolite/zeotype membranes, J. Membr. Sci. Res. 1 (2015) 4972.
[25] W.L. Ang, A.W. Muhammed, N. Hilal, C.P. Leo, A review of the applicability of
integrated/hybrid membrane process in water treatment and desalination
plants, Desalination 363 (2015) 218.
[26] S.S. Shenvi, A.M. Isloor, A.F. Ismail, A review on RO membrane technology;
developments and challenges, Desalination 368 (2015) 1026.
[27] M. Padaki, R. Surya Murali, M.S. Abdullah, N. Misdan, A. Moslehyani, M.A.
Kassim, N. Hilal, A.F. Ismail, Membrane technology enhancement in oil-water
separation: a review, Desalination 357 (2015) 197207.
[28] A.W. Mohammed, Y.H. Teow, W.L. Ang, T.Y. Chung, D.L. Oatley-Radcliffe, N.
Hilal, Nanoltration Membranes review: recent advances and future
prospects, Desalination 356 (2015) 226254.
[29] M.M.A. Shirazi, A. Kargari, M.J.A. Shirazi, Direct contact membrane
distillation for sea water desalination, Desalination 49 (2012) 368375.
[30] M.M.A. Shirazi, A. Kargari, M. Tabatabaei, Evaluation of commercial PTFE
membrane in desalination by direct contact membrane distillation, Chem.
Eng. Process. Process Intensif. 76 (2014) 1625.
[31] M. Khayet, Membranes and theoretical modelling of membrane distillation
a review, Adv. Colloid Interface Sci. 164 (2011) 5688.
[32] E. Drioli, A. Ali, F. Macedonio, Membrane distillation: recent developments
and perspectives, Desalination 356 (2015) 5684.
[33] P. Wang, T.S. Chung, Recent Advances in membrane distillation process:
membrane development conguration design and application exploring, J.
Membr. Sci 475 (2015) 3956.
[34] E. Guillen-Burrieza, A. Ruiz-Agurrie, G. Zaragoza, H.A. Arafat, Membrane
fouling and cleaning in long term plant scale membrane distillation
operation, J. Membr. Sci. 468 (2014) 360372.
[35] A. Ali, P. Aimar, E. Drioli, Effect of module design and ow patterns on
performance of membrane destillation process, Chem. Eng. J. 277 (2015)
368377.
[36] S. Nejati, C. Boo, C.O. Osuji, M. Elimelech, Engineering at sheet microporous
PVDF lm of membrane distillation, J. Membr. Sci. 492 (2015) 355363.
[37] A. Alkhudhiri, N. Darwish, N. Hilal, Membrane distillation a comprehensive
review, Desalination 287 (2012) 218.
[38] J.P. Mericq, S. Laborie, C. Cabassud, Evaluation of systems coupling vacuum
membrane distillation and solar energy for seawater desalination, Chem.
Eng. J. 166 (2011) 596606.
[39] A. Hausmann, P. Sanciolo, T. Vasiljevic, M. Weeks, M. Duke, Integration of
membrane distillation into heat paths of industrial processes, Chem. Eng. J.
211 (2012) 378387.
[40] A.S. Hassan, H.E.S. Fath, Review and assessment of the newly developed MD
for desalination processes, Desalin. Water Treat. 51 (2013) 574585.
[41] A. Burgoyne, M.M. Vahdati, Direct contact membrane distillation, Sep. Sci.
Technol. 35 (2000) 12571284.
[42] L.M. Camacho, L. Dumee, J.H. Zhang, J.D. Li, M. Duke, J. Gomez, S. Gray,
Advances in membrane distillation for water desalination and purication
applications, Water 5 (2013) 94196.
[43] B. Jiao, A. Cassano, E. Drioli, Recent advances on membrane processes for the
concentration of fruit juices: a review, J. Food Eng. 63 (2004) 303324.
[44] H. Susanto, Towards practical implementations of membrane distillation,
Chem. Eng. Process. 50 (2011) 139150.
 V. Aneesh et al. / Chemical Engineering and Processing 104 (2016) 219242
[45] M.S. El-Bourawi, Z. Ding, R. Ma, M. Khayet, A framework for better
understanding membrane distillation separation process, J. Membr. Sci. 285
(2006) 429.
[46] R.A. Gaska, M.R. Cannon, Controlled cycling improves various processes, Ind.
Eng. Chem. 53 (1961) 630631.
[47] J.R. McWhirter, M.R. Cannon, Controlled cycling distillation in a packed-plate
column, Inst. Eng. Chem. 5 (1961) 632634.
[48] V.N. Maleta, A.A. Kiss, V.M. Taran, B.V. Maleta, Understanding process
intensication in cyclic distillation systems, Chem. Eng. Process. 50 (2011)
655664.
[49] Higman, Christopher, van der Burgt, Maarten, Gasication 2 (2008) 324.
[50] R.E. Latimer, Distillation of air, Chem. Eng. Prog. 63 (1967) 3559.
[51] R. Agrawal, Synthesis of distillation column congurations for a
multicomponent separation, Ind. Eng. Chem. Res. 35 (1996) 1059.
[52] W.F. Castle, Air separation and liquefaction: recent developments and
prospects for the beginning of the new millennium, Int. J. Refrig. 25 (2002)
158172.
[53] A.A. Kiss, Advanced distillation technologies-design, Control and
Applications, Wiley, Chichester, 2013.
[54] K. Sundmacher, A. Kienle, Reactive Distillation: Status and Future Directoins,
Wiley-VCH, Weinheim, 2003.
[55] A.A. Kiss, C.S. Bildea, A review on biodiesel production by integrated reactive
separation technologies, J. Chem. Technol. Biotechnol. 87 (2012) 861879.
[56] R. Monroy-Loperena, J. Alvarez-Ramirez, Backstepping based cascade control
scheme for batch distillation, AlChE J. 9 (2004) 21132129.
[57] V. Aneesh, Rahul Antony, M.S. Giri Nandagopal, Ganesh Paramasivan, N.
Selvaraju, Resilient controller design availing recursive technique rooted
Lyapunov Function for a dynamic system, Microsyst. Technol. (2015) 112,
doi:http://dx.doi.org/10.1007/s00542-015-2733-y.
[58] J.A. Heath, I. Kookos, J.D. Perkin, Process Control structure selection based on
economics, AlChE J. 46 (2000) 1998.
[59] L.K. Tun, H. Matsumoto, Application methods for genetic algorithm for the
search of feed position in the design of a reactive distillation process,
Procedia Comput. Sci. 22 (2013) 623.
[60] M. Vzquez-Ojeda, J.G. Segovia-Hernndez, S. Hernndez, A. Hernndez-
Aguirre, R. Maya-Yescas, Optimization of thermally coupled reactive
distillation arrangements with minimum use of reboilers, Ind. Eng. Chem.
Res. 51 (2012) 5856.
[61] Juan Gabriel Segovia-Hernandez, Salvador Hernandez, Adrian Bonilla
Petriciolet, Reactive Disitllation:- A review of optimal design using
deterministic and stochastic techniques, Chem. Eng. Process. 97 (2015) 134
143.
[62] J. Morud, S. Skogestad, Effects of recycle on dynamics and control of chemical
processing plants, Comput. Chem. Eng. 20 (1996) S883S888.
[63] W.D. Seider, J.D. Seader, D.R. Lewin, Product and Process Design Principles:
Synthesis, Analysis, and Design, John Wiley and Sons, Inc., New York, 2008.
[64] A.A. Kiss, C.S. Bileda, A.C. Dimian, P.D. Iedema, State multiplicity in CSTR-
separator-recycle polymerization systems, Chem. Eng. Sci. 57 (2002) 535
546.
[65] J.M. Douglas, Conceptual Design of Chemical Processes, McGraw-Hill, New
York, 1988.
[66] S. Skogestad, Postlethwaite, Multivariable Feedback Control, John Wiley &
Sons, Inc, New York, 1996.
[67] M. Daoutidis, C. Soroush, Kravaris Feedforward/feedback control of
multivariable nonlinear processes, AIChE J. 36 (1990) 14711484.
[68] Z.H. Yuan, B.Z. Chen, J.S. Zhao, Phase behaviour analysis for industrial
polymerization reactor, AIChE J. 57 (2011) 27952807.
[69] Z.H. Yuan, H.Z. Wang, B.Z. Chen, J.S. Zhao, Operating zone segregation of
chemical reaction systems based on stability and non-minimum phase
behaviour analysis, Chem. Eng. J. 155 (2009) 304311.
[70] D.R. Lewin, Interaction of design and control, Proceedings of the 7th IEEE
Mediterranean Conference on Control and Automation; June 28-30, Haifa,
Israel, 1999.
[71] Z.H. Yuan, B.Z. Chen, J.S. Zhao, An overview on controllability analysis of
chemical processes, AIChE J. 57 (2011) 11851201.
[72] P. Psarris, C.A. Floudas, Dynamic operability of MIMO systems with time
delays and transmission zeros, 1. Assessment. Chem Eng Sci. 46 (1991) 2691
2707.
[73] L.T. Narraway, J.D. Perkins, Interaction between process design and process
control: economic analysis of process dynamics, J Process Control 1 (1991)
243250.
[74] Z.H. Yuan, N. Zhang, B.Z. Chen, J.S. Zhao, Systematic controllability analysis for
chemical processes, AIChE J. (2011).
[75] Z.H. Yuan, B.Z. Chen, J.S. Zhao, Effects of manipulated variables selection on
the controllability of chemical process, Ind. Eng. Chem. Res. 50 (2011) 7403
7413.
[76] K.J. Trickett, Towards Quantifying Inverse Response in Nonlinear Systems [Ph.
D Thesis], University of London, 1994.
[77] I. Karafyllis, A. Kokossis, On a new measure for the integration of process
design a control: the disturbance resiliency index, Chem. Eng. Sci. 57 (2002)
873886.
[78] M. Morari, Effect of design on the controllability of chemical plants, IFAC
Workshop on Interactions between Process Design and Process Control;
September 78, London, UK, 1992.
239
[79] W.L. Luyben, The need for simultaneous design education, in: P. Seferlis, M.C.
Georgiadis (Eds.), The Integration of Process Design and Control, Elsevier, Ltd,
Amsterdam, The Netherlands, 2004, pp. 1041.
[80] J.E. Rijnsdorp, P. Bekkers, Early integration of process and control design, IFAC
Workshop on Interactions between Process Design and Process Control;
September 78, 1992, London, UK, 1992, pp. 1722.
[81] P. Seferlis, M.C. Georgiadis, The Integration of Process Design and Control,
Elsevier, 2004, 2016.
[82] T.J. Mcavoy, Synthesis of plantwide control systems using optimization, Ind.
Eng. Chem. Res. 38 (1999) 29842994.
[83] V. Sakizlis, J.D. Perkins, E.N. Pistikopoulos, Recent advances in optimization-
based simultaneous process and control design, Comput. Chem. Eng. 28
(2004) 20692086.
[84] L.A. Ricardez, H.M. Sandoval, P.L. Budman Douglas, Integration of design and
control of chemical processes: a review of the literature and some recent
results, Annu. Rev. Contr. 33 (2009) 158171.
[85] M. Morari, A. Lenhoff, Design of resilient processing plants: I. Process design
under consideration of dynamic aspects, Chem. Eng. Sci. 37 (1982) 245258.
[86] A. Palazoglu, Y. Arkun, A multiobjective approach to design chemical plants
with robust dynamic operability characteristics, Comput. Chem. Eng. 10
(1986) 567575.
[87] A. Palazoglu, Y. Arkun, Design of chemical plant with multiregime
capabilities and robust dynamic operability characteristics, Comput. Chem.
Eng. 11 (1987) 205216.
[88] J. Figueroa, J. Romagnoli, G. Barton, Robust control indicators in process
design and control, IFAC Workshop on Interactions Between Process Design
and Control, Londo, UK, 1992, pp. 140144.
[89] S.P. Walsh, J.D. Perkins, Integrated design of efuent treatment systems, IFAC
Workshop on Interaction Between Process Design and Process Control,
London, UK, 1992, pp. 107112.
[90] M.L. Luyben, C.A. Floudas, Analyzing the interaction of design and control-1:
A multiobjective framework and application to binary distillation synthesis,
Comput. Chem. Eng. 18 (1994) 933969.
[91] K.P. Papalexandri, E.N. Pistikopoulos, Synthesis and retrot design of
operable heat exchanger networks: part 1: Flexibility and structural
controllability aspects, Ind. Eng. Chem. Res. 33 (1994) 17181737.
[92] K.P. Papalexandri, E.N. Pistikopoulos, Synthesis and retrot design of
operable heat exchanger networks: part II: Dynamic and control structure
consideration, Ind. Eng. Chem. Res. 33 (1994) 17381755.
[93] J.D. Perkins, S.P. Walsh, Optimisation as a tool for design/control interactions,
Comput. Chem. Eng. 20 (1996) 315323.
[94] A. Zheng, R.V. Mahajanam, A quantitative controllability index, Ind. Eng.
Chem. Res. 38 (1999) 9991006.
[95] L.P. Lee, Decentralized control design for nonlinear multi-unit plants: a gap
metric approach, Chem. Eng. Sci. 55 (2000) 37433758.
[96] J.B. Jorgensen, S.B. Jorgensen, Towards automatic decentralized control
structure selection, Comput. Chem. Eng. 24 (2000) 841846.
[97] A.M. Blanco, J.A. Bandoni, Interaction between process design and process
operability of chemical processes: an eigenvalue optimization approach,
Comput. Chem. Eng. 27 (2003) 12911301.
[98] A.M. Blanco, J.A. Bandoni, Eigenvalue optimization-based formulationsfor
nonlinear dynamics and control problems, Chem. Eng. Process. 46 (2007)
11921199.
[99] H.Y. Alhammadi, J.A. Romagnoli, Process design and operation incorporating
environment, protability, heat integration and controllability consideration,
in: P. Seferlis, M.C. Georgiadis (Eds.), The Integration of Process Design and
Control, Elsevier, 2004, pp. 264305.
[100] L.A. Ricardez-Sandoval, H.M. Budman, P.L. Douglas, Simultaneous design and
control: a new approach and comparisons with existing methodologies, Ind.
Eng. Chem. Res. 49 (2010) 28222833.
[101] C. Schweiger, C.A. Floudas, Interaction of design and control: optimization
with dynamic models, in: W. Hager, P. Pardalos (Eds.), Optimal Control:
Theory, Algorithms and Applications, Kluwer Academic Publishers, 1997, pp.
388435.
[102] M.F. Meeuse, R.L. Tousain, Closed-loop controllability analysis of process
designs: application to distillation column design, Comput. Chem. Eng. 26
(2002) 641647.
[103] P. Seferlis, J. Grievink, Process design and control structure screening based
on economic and static controllability criteria, Comput. Chem. Eng. 25 (2001)
177188.
[104] V. Amte, S.H. Nistala, S.M. Mahajani, R.K. Malik, Optimisation based
conceptual design of reactive distillation for selectivity engineering, Comput.
Chem. Eng. 48 (2013) 209.
[105] C.S. Bildea, G. Romuald, E. Sanchez-Ramirez, J.J. Quiroz-Ramirez, J.G. Segovia-
Hermandez, T. Kiss, Optimal design and control of novel process for di-n-
pentyl ether production, J. Chem. Technol. Biotechnol. 90 (2015) 9921001.
[106] I. Guerra, R. Lamanna, S. Revollar, An activated-sludge-process application of
integrated design and predictive control with instantaneous linearization,
IEEE Mediterranean Conference on Control and Automation (2012).
[107] M. Sharifzadeh, N.F. Thornhill, Integrated design and control using a dynamic
inversely controlled process model, Comput. Chem. Eng. 48 (2013) 121134.
[108] M. Sharifzadeh, N.F. Thornhill, Optimal selection of control structure using a
steady-state inversely controlled process model, Comput. Chem. Eng. 38
(2012) 126138.
[109] M. Sharifzadeh, Integration of process design and control: a review, Chem
Eng Res. Des. 91 (2013) 25152549.
240 V. Aneesh et al. / Chemical Engineering and Processing 104 (2016) 219242
[110] P. Seferlis, J. Grievink, Optimal design and sensitivity analysis of reactive
distillation units using collocation models, Ind. Eng. Chem. Res. 40 (2001)
16731685.
[111] E. Ekawati, P.A. Bahri, The integration of the output controllability index
within the dynamic operability framework in process system design, J
Process Control 13 (2003) 717727.
[112] D.D. Brengel, W.D. Seider, Coordinated design and control optimization of
nonlinear processes, Comput. Chem. Eng. 16 (1992) 861886.
[113] R.J. Allgor, P.I. Barton, L.B. Evans, Optimal batch process development,
Comput. Chem. Eng. 20 (1996) 885896.
[114] R.J. Allgor, L.B. Evans, P.I. Barton, Screening models for batch process
development, Chem. Eng. Sci. 54 (1999) 41454164.
[115] T.K. Bhatia, L.T. Biegler, Dynamic optimization in the design and scheduling of
multiproduct batch plants, Ind. Eng. Chem. Res. 35 (1996) 22342246.
[116] A. Giovanoglou, C.S. Adjiman, E.N. Pistikopoulos, Optimal solvent design for
batch separation based on economic performance, AIChE J. 49 (2003) 3095
3109.
[117] M.P. Avraam, N. Shah, C.C. Pantelides, A decomposition algorithm for the
optimization of hybrid dynamic processes, Comput. Chem. Eng. 23 (1999)
451454.
[118] P.I. Barton, C.K. Lee, Design of process operations using hybrid dynamic
optimization, Comput. Chem. Eng. 28 (2004) 885896.
[119] P. Dua, V. Dua, E.N. Pistikopoulos, Optimal delivery of chemotherapeutic
agents in cancer, Comput. Chem. Eng. 32 (2008) 99107.
[120] C.L. Chen, H.W. Tsai, Model-based insulin therapy scheduling: a mixed-
integer nonlinear dynamic optimization approach, Ind. Eng. Chem. Res. 48
(2009) 85958604.
[121] I.P. Androulakis, Kinetic mechanism reduction based on an integer
programming approach, AIChE J. 46 (2000) 361371.
[122] L. Petzold, W. Zhu, Model reduction for chemical kinetics: an optimization
approach, AIChE J. 45 (1999) 869886.
[123] R.J. Allgor, P.I. Barton, Mixed-integer dynamic optimization, Comput. Chem.
Eng. 21 (1997) 451456.
[124] B. Chachuat, A.B. Singer, P.I. Barton, Global mixed-integer dynamic
optimization, AIChE J. 51 (2005) 22352253.
[125] R.J. Allgor, P.I. Barton, Mixed integer dynamic optimization. I-problem
formulation, Comput. Chem. Eng. 23 (1999) 567584.
[126] I.P. Androulakis, Kinetic mechanism reduction based on an integer
programming approach, AIChE J. 46 (2000) 361371.
[127] A. Giovanoglou, C.S. Adjiman, E.N. Pistikopoulos, Optimal solvent design for
batch separation based on economic performance, AIChE J. 49 (2003) 3095
3109.
[128] M.P. Avraam, N. Shah, C.C. Pantelides, A decomposition algorithm for the
optimization of hybrid dynamic processes, Comput. Chem. Eng. 23 (1999)
s451s454.
[129] L.T. Biegler, An overview of simultaneous strategies for dynamic
optimization, Chem. Eng. Process. 46 (2007) 10431053.
[130] P. Bahri, J.A. Bandoni, J.A. Romagnoli, Integrated exibility and controllability
analysis in design of chemical processes, AIChE J. 43 (1997) 9971015.
[131] M.J. Mohideen, J.D. Perkins, E.N. Pistikopoulos, Optimal synthesis and design
of dynamic systems under uncertainty, Comput. Chem. Eng. 20 (1996) s895
s900.
[132] V. Dimitriadis, E.N. Pistikopoulos, Flexibility analysis of dynamic systems,
Ind. Eng. Chem. Res. 34 (1995) 44514462.
[133] A. Flores-Tlacuahuac, L.T. Biegler, Simultaneous mixed-integer dynamic
optimization for integrated design and control, Comput. Chem. Eng. 31
(2007) 588600.
[134] A. Flores-Tlacuahuac, I.E. Grossmann, Simultaneous cyclic scheduling and
control of tubular reactors: single production lines, Ind. Eng. Chem. Res. 49
(2010) 1145311463.
[135] S. Terrazas-Moreno, A. Flores-Tlacuahuac, I.E. Grossmann, Lagrangean
heuristic for the scheduling and control of polymerization reactors, AIChE J.
54 (2008) 163182.
[136] S. Terrazas-Moreno, A. Flores-Tlacuahuac, I.E. Grossmann, Simultaneous
design, scheduling, and optimal control of a methyl-methacrylate continuous
polymerization reactor, AIChE J. 54 (2008) 31603170.
[137] Chaowei Liu, Jie Fu, Qiang Xu Simultaneous Mixed-integer dynamic
optimization for environmentally benign electroplating, Comput. Chem. Eng.
35 (2011) 24112425.
[138] J. Oldenburg, W. Marquardt, Mixed-logic dynamic optimization applied to
batch distillation process design, AIChE J. 49 (2003) 29002917.
[139] J. Bush, J. Oldenburg, W. Marquardt, Dynamic predictive scheduling of
operational strategies for continuous processes using mixed-logic dynamic
optimization, Comput. Chem. Eng. 31 (2007) 574587.
[140] C. Chatzidoukas, E.N. Pistikopoulos, C. Kiparissides, A hierarchical
optimization approach to optimal production scheduling in an industrial
continuous olen polymerization reactor, Macromol. React. Eng. 3 (2009)
3646.
[141] T. Hirmajer, E. Balsa-Canto, J.R. Banga, Mixed integer nonlinear optimal
control in systems biology and biotechnology: numerical methods and a
software toolbox, Proceedings of 9th International Symposium on Dynamics
and Control of Process Systems, Leuven, Belgium, 2010, pp. 300305.
[142] V. Bansal, J.D. Perkins, E.N. Pistikopoulos, Simultaneous design and control
optimization under uncertainty, Comput. Chem. Eng. 24 (2000) 261266.
[143] V. Bansal, J.D. Perkins, E.N. Pistikopoulos, A case study in simultaneous design
and control using rigorous: mixed-integer dynamic optimization models,
Ind. Eng. Chem. Res. 41 (2002) 760778.
[144] M. Sharif, N. Shah, C.C. Pantelides, On the design of multi-component batch
distillation columns, Comput. Chem. Eng. 22 (1998) S69S76.
[145] S. Sager, H.G. Bock, G. Reinelt, Direct methods with maximal lower bound for
mixed-integer optimal control problems, Math. Program. 118 (2009) 109
149.
[146] S. Sager, Reformulations and algorithms for the optimization of switching
decisions in nonlinear optimal control, J Process Control 19 (2009) 1238
1247.
[147] A. Prata, J. Oldenburg, W. Marquardt, Integrated scheduling and dynamic
optimization of grade transitions for a continuous polymerization reactor,
Comput. Chem. Eng. 32 (2008) 463476.
[148] V. Bansal, V. Sakizlis, R. Ross, J.D. Perkins, E.N. Pistikopoulos, New algorithms
for mixed integer dynamic optimization, Comput. Chem. Eng. 27 (2003) 647
668.
[149] K.H. Low, E. Sorensen, Simultaneous optimal design and operation of
multivessel batch distillation, AIChE J. 49 (2003) 25642576.
[150] K.H. Low, E. Sorensen, Simultaneous optimal design and operation of
multipurpose batch distillation, Chem. Eng. Process. 43 (2003) 273289.
[151] K.H. Low, E. Sorensen, Simultaneous optimal conguration: design and
operation of batch distillation, AIChE J. 51 (2005) 17001713.
[152] W. Wongrat, A. Younes, Control vector optimization and genetic algorithms
for mixed-integer dynamic optimization in the synthesis of rice drying
processes, J. Franklin Inst. 348 (2011) 13181338.
[153] E.N. Pistikopoulos, V. Sakizlis, Simultaneous design and control optimization
under uncertainty in reaction/separation systems, AIChE J. 98 (2002) 223
238.
[154] G. Paramasivan, A. Kienle, A reactive distillation case study for decentralized
control system design using mixed integer optimization, Proceedings of 20th
European Symposium on Computer Aided Process Engineering, Naples, Italy,
2010, pp. 565570.
[155] V. Bansal, R. Ross, J.D. Perkins, Interactions of design and control: double-
effect distillation systems, Dycops 5, Corfu, Greece, 1999, pp. 679684.
[156] R. Ross, J.D. Perkins, E.N. Pistikopoulos, Optimal design and control of a high-
purity industrial distillation system, Comput. Chem. Eng. 25 (2001) 141150.
[157] R. Ross, V. Bansal, J.D. Perkins, Optimal design and control of an industrial
distillation system, Comput. Chem. Eng. 23 (1999) s875s878.
[158] I.K. Kookos, J.D. Perkins, An algorithm for simultaneous process design and
control, Ind. Eng. Chem. Res. 40 (2001) 40794088.
[159] P. Panjwani, M. Schenk, M.C. Georgiadis, E.N. Pistikopoulos, Optimal design
and control of a reactive distillation system, Eng. Optimiz. 37 (2005) 733
753.
[160] M. Asteasuain, A. Brandolin, A. Bandoni, Integration of control aspects and
uncertainty in the process design of polymerization reactors, Chem. Eng. J.
131 (2007) 135144.
[161] D.R. Lopez-Negrete, A. Flores-Tlacuahuac, Integrated design and control
using a simultaneous mixed-integer dynamic optimization approach, Ind.
Eng. Chem. Res. 48 (2009) 19331943.
[162] A. Nikolaos Diangelakis, M. Amit Manthanwar, E.N. Pistikopoulos, A
framework of design and control optimisation: application on CHP system,
Comput. Chem. Eng. 34 (2014) 765770, doi:http://dx.doi.org/10.1016/B978-
0-444-63433-7.50112-7.
[163] Yunfei Chu, Fengqi You, Integrated Scheduling and dynamic optimisation for
network batch process, Comput. Chem. Eng. 33 (2014) 523528, doi:http://
dx.doi.org/10.1016/B978-0-444-63456-6.50088-0.
[164] Pu Li, Bing Chen, Baiyu Zhang, Lian Jing, Jisi Zhang, Monte Carlo, Simulation
based dynamic mixed integer nonlinear programming for supporting oil
recovery and device allocation during offshore oil spill responses, Ocean Cost
Manage. 89 (2014) 5870, doi:http://dx.doi.org/10.1016/j.
oceancoaman.2013.12.006.
[165] Vida Meidanshashi, A. Thomas Adams II, Integrated Design and control of
semi-continuous process using mixed integer nonlinear dynamic
optimisation, Comput. Chem. Eng. 37 (2015) 11811186, doi:http://dx.doi.
org/10.1016/B978-0-444-63577-8.50042-5.
[166] M.J. Mohideen, J.D. Perkins, E.N. Pistikopoulos, Optimal design of dynamic
systems under uncertainty, AIChE J. 42 (1996) 22512272.
[167] M.J. Mohideen, J.D. Perkins, E.N. Pistikopoulos, Towards an efcient
numerical procedure for mixed integer optimal control, Comput. Chem. Eng.
21 (1997) 457462.
[168] R. Pytlak, Runge-kutta based procedure for the optimal control of differential
algebraic equations, J. Optim. Theory Appl. 97 (1999) 675705.
[169] C.A. Floudas, Nonlinear and Mixed-Integer Optimization: Fundamentals and
Applications, Oxford University Press, 1995.
[170] V. Sakizlis, J.D. Perkins, E.N. Pistikopoulos, Parametric controllers in
simultaneous process and control design optimization, Ind. Eng. Chem. Res.
42 (2003) 45454563.
[171] C. Loeblein, J.D. Perkins, Structural design for on-line process optimization: I.
Dynamic economics of MPC, AIChE J. 45 (1999) 10181029.
[172] C. Loeblein, J.D. Perkins, Structural design for on-line process optimization: II.
Application to a simulated FCC, AIChE J. 45 (1999) 10301040.
[173] M.J. Mohideen, J.D. Perkins, E.N. Pistikopoulos, Robust stability
considerations in optimal design of dynamic systems under uncertainty, J.
Process Control 7 (1997) 371385.
 V. Aneesh et al. / Chemical Engineering and Processing 104 (2016) 219242
[174] V. Sakizlis, Design of Model-Based Controller via Parametric Programming
[Ph.D Thesis], Imperial College, London, UK, 2003.
[175] Seyed Kjobsted Huusom, M. John Woodley, Raqual Gani, Integrated process
design and control of reactive distillation process, IFAC-Papers Online 48-8
(2015) 11201125.
[176] Yanan Cao, L.E. Christopher Swartz, Michael Baldee, Stephane Blowin,
Optimisation based assessment of design limitations to air separation plant
agility in demand response scenario, J. Process Control 33 (2015) 3748.
[177] Yao-Hsien Chung, Hao-Yeh Lee, Chang Liang Chen, Design and control of
reactive distillation process for the production of methyl valerate, Ind. Eng.
Chem. Res. 55 (2016) 13471360, doi:http://dx.doi.org/10.1021/acr.
iecr.5b03495.
[178] Yao-Hsien Chung, Tzu-Hsuan Peng, Hao- Yeh Lee, Chang-Liang Chen, Lung
Chen I, Design and control of reactive distillation system for the esterication
of Levulinic Acide and N-butanol, Ind. Eng. Chem. Res. 54 (2015) 33413354,
doi:http://dx.doi.org/10.1021/ie500660h.
[179] Miguel Angel Gutirrez-Limna, Antonio Flores-Tlacuahuacb, E. Ignacio
Grossman, A reactive optimisation strategy for simultaneous planning,
scheduling and control of short period continuous reactor, Comp. Chem. Eng.
84 (2016) 507515.
[180] P. Vega, R. Lamanna de Rocco, S. Revollar, M. Franciso, Integrated design and
control of chemical process-part I: Revision and Classication, Comput.
Chem. Eng. 71 (2014) 602617.
[181] P. Vega, R. Lamanna de Rocco, S. Revollar, M. Franciso, Integrated design and
control of chemical process-part II: An illustrative example, Comput. Chem.
Eng. 71 (2014) 618635.
[182] M. Francisco, S. Revollar, P. Vega, R. Lamanna, Simultaneous synthesis, design
and con-trol of processes using model predictive control, Proceedings of
International Symposium on Advanced Control of Chemical Processes (Ad
Chem) (2009).
[183] M. Francisco, P. Vega, H. Alvarez, Robust integrated design of processes with
terminal penalty model predictive controllers, Chem. Eng. Res. Des. 89 (2011)
10111024.
[184] K. Snchez-Snchez, L. Ricardez-Sandoval, Simultaneous design and control
under uncertainty using model predictive control, Ind. Eng. Chem. Res. 52
(13) (2013) 48154833.
[185] J.M. Maciejowski, Predictive Control with Constraints, Prentice Hall, United
Kingdom, 2002.
[186] Kelvyn B. Sanchez-Sanchez, Luis A. Ricardez-Sandoval, Simultaneous design
and control under uncertainty using model predictive control, Ind. Eng.
Chem. Res. 52 (2013) 48154833.
[187] Jayaram Valluru, P. Jalesh Purohit, C. Sachin Patwardhan, M. Sanjay Mahajan,
Adaptive optimisation control of ideal reactive distillation column, IFAC-
Papers Online 48-8 (2015) 489494.
[188] G. Arayo Pimentel, A. Rapcport, A. Vande Woumer, Nonlinear Model
predictive control of waste water treatment process tted with submerges
membrane bioreactor, IFAC-Papers Online 488 (2015) 12461251.
[189] Juan Gabriel Segovia-Hernandez, Salvador Hernandez, Adrian Bonilla
Petriciolet, Reactive Distillation: a review of optimal design using
deterministic and stochastic technique, Chem. Eng. Process. 97 (2015) 134
143.
[190] A. Luis, Ricardez Sandoval, Optimal design and control of dynamic systems
under uncertainty: a probabilistic approach, Comput. Chem. Eng. 43 (2012)
91107.
[191] Kelvyn Sanchez, L. Ricardez Sandoval, Simultaneous Process synthesis and
control design under uncertainty: a worst case performance approach, AIChE
J. 59 (7) (2013) 24972514.
[192] S. Bahakim, L. Ricardez-Sandoval, Simultaneous Design and MPC based
control for dynamic systems under uncertainty: a Stochastic approach,
Comput. Chem. Eng. 63 (17) (2014) 6681.
[193] M. Trainor, V. Ginnakeas, C. Kiss, L.A. Ricardez Sandoval, Optimal process and
control design under uncertainty: a method worth robust feasibility and
stability analyses, Chem. Eng. Sci. 104 (2013) 10651080.
[194] P. Bhushan Patil, Eduardo Maria, L.A. Ricardez Sandoval, Integration of
scheduling, design and control of multiproduct chemical processes under
uncertainty, J. AIChE 61 (8) (2015) 24562470.
[195] Siddharth Mehta, A. Luis Ricardez Sandoval, Integration of design and control
of dynamic systems under uncertainty: a new back off approach, Ind. Eng.
Chem. Res. 55 (2016) 485498, doi:http://dx.doi.org/10.1021/acs.
iecr.5b03522.
[196] N. Efstratios Pistikopoulos, A. Nikolaos Diangelakis, Towards the integration
of process design, control and scheduling: are we getting closer? Comput.
Chem. Eng. (2016), doi:http://dx.doi.org/10.1016/j.
computchemeng.2015.11.002.
[197] N. Efstratios Pistikopoulos, A. Nikolaos Diangelakis, Richard Oberdiect, M.
Maria Papathanasiou, Loana Nascu, Muxin Sun, PAROC-an integrated
framework and software platform for the optimisation and advanced model
based control of process system, Chem. Eng. Sci. 136 (2015) 115138, doi:
http://dx.doi.org/10.1016/j.ces.2015.02.030.
[198] N. Chawankul, H. Budman, P.L. Douglas, The integration of design and control:
IMC control and robustness, Comput. Chem. Eng. 2 (2005) 261271.
[199] N. Chawankul, L.A. Ricardez-Sandoval, H. Budman, Integration of design and
control: a robust control approach using MPC, Can. J. Chem. Eng. 85 (2007)
433446.
241
[200] R. Zonovello, H.M. Budman, Model predictive control with soft constraints
with application to Lime Kiln Control, Comput. Chem. Eng. 23 (1999) 791
806.
[201] M. Morari, E. Zariou, Robust Process Control, Prentice Hall, New Jersey, 1989.
[202] L.A. Ricardez- Sandoval, H. Budman, P.L. Douglas, Simultaneous design and
control of process under uncertainty: a robust modelling approach, J. Process
Control 18 (2008) 735752.
[203] L.A. Ricardez-Sandoval, H. Budman, P.L. Douglas, Simultaneous design and
control of chemical processes with application to the Tennessee Eastman
process, J. Process Control. 19 (2009) 13771391.
[204] Johannes Gerhard, Wolfgang Marquardt, Martin Monnigmann, Normal
vector on critical manifold for robust design on transient process in the
presence of fast disturbances, SIAM J. Appl. Dyn. Syst. 7 (2) (2008) 461490.
[205] A. Diego Munoz, Johannes Gerhard, Wolfgang Marquardt, A normal vector
approach for integrated process and control with uncertain model
parameters and disturbances, Comput. Chem. Eng. 40 (2012) 2022012.
[206] L.A. Ricardez-Sandoval, H. Budman, P.L. Douglas, A methodology for the
simultaneous design and control of large-scale systems under process
parameter uncertainty, Comput. Chem. Eng. 35 (2011) 307318.
[207] M. Monnigmann, W. Marquardt, Normal vectors on manifolds of critical
points for parametric robustness of equilibrium solutions of ODE systems, J.
Nonlinear Sci. 12 (2002) 85112.
[208] M. Monnigmann, W. Marquardt, Steady-state process optimization with
guaranteed robust stability and feasibility, AIChE J. 49 (2003) 31103126.
[209] M. Monnigmann, W. Marquardt, Steady-state process optimization with
guaranteed robust stability and exibility: application to HAD reaction
section, Ind. Eng. Chem. Res. 44 (2005) 27372753.
[210] M. Monnigmann, W. Marquardt, A hybrid approach for efcient robust design
of dynamic systems, SIAM Rev. 49 (2007) 236254.
[211] J. Gerhard, M. Monnigmann, W. Marquardt, Steady state optimization with
guaranteed stability of a tryptophan biosynthesis model, Comput. Chem. Eng.
32 (2008) 29142919.
[212] W. Benedikt, J. Gerhard, W. Marquardt, Robust Optimisation with normal
vectors on critical manifolds of disturbance-induced stability loss, J.
Nonlinear Sci. 21 (2011) 5792.
[213] D. Kastsian, M. Monnigmann, Robust optimization of xed points of
nonlinear discrete time systems with uncertain parameters, SIAM J. Appl.
Dyn. Syst. 9 (2011) 357390.
[214] J. Hahn, M. Monnigmann, W. Marquardt, A method for robustness analysis of
controlled nonlinear systems, Chem. Eng. Sci. 59 (2004) 43254338.
[215] J. Hahn, M. Martin, W. Marquardt, On the use of bifurcation analysis for robust
controller tuning for nonlinear systems, J. Process Control 18 (2008) 408
420.
[216] R. Grosch, M. Monnigmann, W. Marquardt, Integrated design and control for
robust performance: application to an msmpr crystallizer, J. Process Control
18 (2008) 173188.
[217] J. Gerhard, Normal Vectors for the Robust Design of Dynamic Systems [Ph.D
Thesis], RWTH Aachen University, 2009.
[218] R. Grosch, Integrated Design and Control of Continuous Suspension
Crystallization [Ph.D Thesis], RWTH Aachen University, 2008.
[219] D.A. Munoz, J. Gerhard, W. Marquardt, Integrated process and control design
by normal vector approach: application to the Tennessee-eastman process,
Proceedings of 21th European Symposium on Computer Aided Process
Engineering, Chalkidiki, Greece, 2011, pp. 668672.
[220] A. Malcolm, L.B. Zhang, B.A. Ogunnaike, A.A. Linninger, Integrating systems
design and control using dynamic exibility analysis, AIChE J. 53 (2007)
20482061.
[221] J. Moon, S. Kim, A.A. Linninger, Embedded control for optimizing exible
dynamic process performance, Proceedings of 10th International Symposium
on Process Systems Engineering, Salvador, Brazil, 2009, pp. 12511256.
[222] J. Moon, S. Kim, A.A. Linninger, Embedded control for optimizing exible
dynamic process performance, Ind. Eng. Chem. Res. 50 (2011) 49935004.
[223] S. Kim, A.A. Linninger, Integration of design and control for a large scale
owsheet, Proceedings of 20th European Symposium on Computer Aided
Process Engineering, Naples, Italy, 2010, pp. 12791284.
[224] J. Moon, S.S. Kim, A.A. Linninger, Integrated design and control under
uncertainty: embedded control optimization for plantwide processes,
Comput. Chem. Eng. 35 (2011) 17181724.
[225] J. Patel, K. Uygun, Y.L. Huang, A path constrained method for integration of
process design and control, Comput. Chem. Eng. 32 (2008) 13731384.
[226] T.C. Hsia, System Identication, Lexington Books, Lexington, 1977.
[227] W.F. Ramirez, Process Control and Identication, Academic press, San Diego,
2016, pp. 1994.
[228] A. Gelb, Applied Optimal Estimation, MIT Press, Cambridge, 1974.
[229] S. Revollar, R. Lamanna, P. Vega, Algorithmic synthesis and integrated design
for activated sludge processes using genetic algorithms, 15th European
Symposium on Computer Aided Process Engineering, Barcelona, Spain, 2005,
pp. 739744.
[230] S. Revollar, M. Francisco, R. Lamanna, Genetic Algorithms for the Synthesis
and Integrated Design of Processes Using Advanced Control Strategies,
Proceedings of DCAI2008, 2016, pp. 205214.
[231] O. Exler, L.T. Antelo, J.R. Banga, A Tabu search-based algorithm for mixed
integer nonlinear problems and its application to integrated process and
control system design, Comput. Chem. Eng. 32 (2008) 18771891.
242 V. Aneesh et al. / Chemical Engineering and Processing 104 (2016) 219242
[232] M. Schluter, J.A. Egea, J.R. Banga, An extended ant colony optimization
algorithm for integrated process and control system design, Ind. Eng. Chem.
Res. 48 (2009) 67236738.
[233] S. Revollar, P. Vega, M. Francisco, Multi-objective genetic algorithms for the
integrated design of chemical processes using advanced control techniques,
20th European Symposium on Computer Aided Process Engineering, Naples,
Italy, 2010.
[234] S. Revollar, M. Francisco, R. Lamanna, Stochastic optimization for the
simultaneous synthesis and control system design of an activated sludge
process, Lat. Am. Appl. Res. 40 (2010) 137146.
[235] X.J. Lu, H.X. Li, X. Yuan, PSO-based intelligent integration of design and
control for one kind of curing process, J. Process Control 20 (2010) 11161125.
[236] M. Alvarado-Morales, K.M.A. Hamid, R. Gami, A model based methodology
for simultaneous design and control of a bioethanol production process,
Comput. Chem. Eng. 34 (2010) 20432061.
[237] Y.J. Chang, N.V. Sahinidis, Steady state process optimization with guaranteed
robust stability under parametric uncertainty, AIChE J. 57 (2011) 33953407.