[Computer Aided Chemical Engineering] Integrated Design and Simulation of Chemical Processes Volume 35 || Process Intensification

CHAPTER

PROCESS INTENSIFICATION 1010.1 INTRODUCTIONIn the European Roadmap of process intensification (PI), PI is defined as a set of innovative principles

applied in process and equipment design, which can bring significant benefits in terms of process and

chain efficiency, lower capital and operating expenses, higher quality of products, less wastes and im-

proved process safety. There are several books, reviews and research papers that address topics related

to PI in the chemical industry (Stankiewicz, 2003; van Gerven and Stankiewicz, 2009; Sanders et al.,

2012; Boodhoo and Harvey, 2013; Reay et al., 2013).

It is worth noting that on the macroscale of reactor and plant, the classic concept of unit operations

(one function per unit) cannot take into account the positive effect of integration. For example, in re-

active separation processes the combination of reaction and separation can increase the conversion to

100% in case of reversible reactions, by taking advantage of the Le Chatelier principle—pulling the

equilibrium by the continuous removal of products, instead of the classic push of the equilibrium by

using an excess of reactants. Not surprisingly, Freund and Sundmacher (2008) claimed that knowledge

of the existing apparatuses that perform unit operations immediately narrows our creativity in search

for new solutions, and they proposed to shift from unit apparatuses to functions. A function (or a fun-

damental task) describes what should happen and not how it should happen. Some examples of func-

tions include: mass movement, chemical reaction, mixing, separation, heat transfer, phase change,

temperature change, pressure change, form change, etc.

The main principles of PI were recently described in the research paper of van Gerven and

Stankiewicz (2009), as follows:

1. Maximise the effectiveness of intra- and inter-molecular events: This principle is primarily about

changing the kinetics of a process, which is actually the root of low conversions and selectivities,

unwanted side-products and other issues.

2. Give each molecule the same processing experience: When all molecules undergo the same history,

the process delivers ideally uniform products with minimum waste. The meso- and micro-mixing,

temperature gradients, macroscopic residence time distribution, dead zones or bypassing play an

important role. For example, a plug-flow reactor with gradientless, volumetric heating (e.g. by

means of microwaves) is clearly much closer to the ideal described by this principle as compared to

a continuous stirred-tank reactor with jacket heating.

3. Optimise the driving forces at every scale and maximise the specific surface area to which theseforces apply: This principle is about the transport rates across interfaces. The resulting effect of the

Computer Aided Chemical Engineering. Volume 35. ISSN 1570-7946. http://dx.doi.org/10.1016/B978-0-444-62700-1.00010-3

© 2014 Elsevier B.V. All rights reserved.397

http://dx.doi.org/10.1016/B978-0-444-62700-1.00010-3

driving forces (e.g. concentration difference) needs to be maximised, and this is done by

maximising the interfacial area, to which that driving force applies. Increased transfer areas (or

surface-to-volume ratios) can be obtained by moving from mm to mm scales of channel diameters,

for example, a circular micro-channel of 400 mm in a microreactor delivers a specific area of

15,000 m2/m3. While impressive, this figure is lower than what is encountered in natural systems:

for example, capillary veins are 10 mm in diameter, with specific areas of ca. 400,000 m2/m3.

4. Maximise the synergistic effects from partial processes: Synergistic effects should be required andutilised, whenever possible and at all possible scales. Such utilisation occurs in form of multi-

functionality on the macroscale, as, for example, in reactive separation units, where the reaction

equilibrium is shifted by removing the products in situ, from the reaction environment.

These principles are not entirely new to chemical engineering, but in PI they are seen as explicit goals

aimed to be reached by intensified processes. Moreover, the PI interpretation of these principles often

goes beyond the boundaries of the classical chemical engineering approach. A completely intensified

process is successful in realising all these PI principles, by making use of one or more fundamental

approaches in four domains (van Gerven and Stankiewicz, 2009): spatial (structure), thermodynamic

(energy), functional (synergy) and temporal (time). In addition, it is worth noting that the PI technol-

ogies also adhere to the guiding principles for the conceptual design of safe chemical processes, thus

providing inherent safety or safety-by-design:

• Avoid: no extra chemicals, no solvent, no strip gas, no extra vessels, no extra pumps.

• Small: reduced holdups, low number of equipment units and inter-connections.

• Control: continuous processing, inherent process control (e.g. boiling systems).

Within the development of PI technologies, two main directions can be distinguished (Stankiewicz,

2003; van Gerven and Stankiewicz, 2009; Reay et al., 2013):

1. PI equipment○ Chemical reactors: spinning disc reactor (SDR), static mixer (SM) reactor, microreactor and

monolithic reactor.○ Equipment for non-reactive systems: rotating packed bed (RPB), centrifugal absorber, SM and

compact heat exchanger (CHE).

2. Process intensification methods○ Multi-functional reactors: heat-integrated reactor, reactive separation processes (reactive

distillation (RD)/stripping/absorption/extraction/crystallisation, as well as membrane reactors),

reactive comminution, reactive extrusion and fuel cells.○ Hybrid separations: dividing-wall column (DWC), membrane distillation, pervaporation,

membrane adsorption and adsorptive distillation.○ Alternative energy sources: solar energy, microwave, ultrasound, electric field and centrifugal

field.○ Other methods: supercritical fluids, plasma technology, periodic operation.

This chapter describes hereafter only selected PI equipments and technologies, namely, the ones that

became successful stories at industrial scales. For example, reactive separation processes improve the

production efficiency by integrating reaction and separation into a single unit that allows the simulta-

neous production and removal of products, therefore enhancing the productivity and selectivity,

398 CHAPTER 10 PROCESS INTENSIFICATION

reducing the energy use, eliminating the need for solvents, intensifying the mass and heat transfer, and

ultimately leading to high-efficiency systems. Reactive separation processes are among the most prom-

ising as they can bring substantial process economic benefits.

10.2 PROCESS INTENSIFICATION EQUIPMENT10.2.1 COMPACT HEAT EXCHANGERSHeat exchangers are widely used in the chemical process industry for both heating and cooling. A com-

pact heat exchanger (CHE) is a piece of equipment built for efficient heat transfer from one medium to

another, being characterised by large heat transfer area-to-volume ratio (minimum 300 m2/m3), high

heat transfer coefficients (up to 5000 W/m2K), small flow passages, and laminar flow. CHE have dense

arrays of finned tubes or plates and are widely used to achieve large heat rates per unit volume, par-

ticularly when at least one of the two fluids is a gas (Kays and London, 1998; Hesselgreaves, 2001). The

type and size of heat exchangers can be tailored to suit a particular process depending on the type of

fluid, phase, composition, temperature, pressures, density, viscosity and other physical properties

(Shah et al., 1990). A large section of compact and non-tubular heat exchanger can be found in the

8th edition of Perry’s Chemical Engineers’ Handbook (Green and Perry, 2008).

The most common types of CHE are summarised hereafter:

• Plate heat exchangers (PHE) use metal plates to transfer heat between two fluids which are exposed

to a much larger surface area, being spread out over the plates. The thin, corrugated plates used in

PHE are gasketed, welded or brazed together depending on the application. The plates are

compressed together in a rigid frame to form an arrangement of parallel flow channels with

alternating hot and cold fluids (Figure 10.1, left). The increase and reduction of heat transfer area is

made through the addition or removal of plates from the stack.

Plate finned-tube heatexchanger

Cout

Cout

Hin

Hin

Hout

Hout

Cin

Cin

Plate and frame heatexchanger

Spiral heat exchanger

FIGURE 10.1

Schematics of a plate and frame heat exchanger (left), plate finned-tube heat exchanger (centre) and spiral heat

exchanger (right).

39910.2 PROCESS INTENSIFICATION EQUIPMENT

• Plate-fin heat exchangers (PFHE) use plates and finned chambers to transfer the heat between fluids

(Figure 10.1, centre). A PFHE is made of layers of corrugated sheets separated by flat metal plates

that create a series of finned chambers. The hot and cold fluid streams flow through alternating

layers of the HE and are enclosed by side bars at the edges. The fins also serve to increase the

structural integrity of the PFHE, allowing it to withstand high pressures while providing an

extended heat transfer area. PFHE can operate with any combination of gas, liquid and two-phase

fluids. Moreover, it can accommodate the heat transfer between multiple process streams by

using a variety of fin heights and types as different entry and exit points for each stream. The main

types of fins are: plain (simple straight-finned triangular or rectangular), herringbone (fins are

placed sideways to provide a zig-zag path), serrated and perforated (cuts and perforations in the finsto enhance flow distribution and heat transfer). Some disadvantages of PFHE include the prone

to fouling and difficult mechanical cleaning.

• Spiral heat exchangers (SHE) consist of a pair of flat surfaces that are coiled to form the two

channels in a counter-flow arrangement, for example, helical/coiled tube configuration

(Figure 10.1, right). SHE is highly efficient in using the space, thus having a small footprint and low

capital costs. The most common application is handling slurries. There are three main types of

flow patterns in a SHE: (1) spiral-spiral flow used for all heating and cooling service, (2) spiral-cross

flow (one fluid is in spiral flow and the other in a cross flow) used for condenser and reboiler

purposes and (3) distributed vapour—spiral flow that can condense and subcool in the same unit.

10.2.2 STATIC MIXERSStatic mixers (SMs) are precision engineered devices for the continuous mixing of fluids without the

need of moving parts. SM can be used to mix liquid and gas streams, disperse gas into liquid or blend

immiscible liquids. The energy required for the efficient mixing of fluids comes from the pressure drop

through the SM’s elements. Details about design and applications can be found elsewhere (Edward,

2004). Themain benefits of SMs are: small volumes, lowmaintenance, simple installation and cleaning

and excellent reliability.

Two main types of SMs are available on the market and largely used at industrial scale:

1. Housed-elements type: Consists of mixer elements contained in a cylindrical (tube) or squared

housing. As the streams move through the mixer, the static elements continuously blend the fluid

materials. The degree of mixing depends on several variables: fluids properties, inner diameter of

the tube, the number of elements and their design (Thakur et al., 2003). Helical elements can

simultaneously produce patterns of flow division (number of striations produced being 2N, where Nis the number of elements in mixer) and radial mixing in order to reduce or eliminate radial

gradients in temperature, velocity and material composition (Figure 10.2, left).

2. Plate-type mixer: In the plate-type design, mixing is accomplished through intense turbulence in the

flow. The corrugated plate SM is capable of mixing low viscosity liquids, blending gases,

dispersing immiscible liquids and creating gas–liquid dispersions with a very high degree of mixing

in a short length. Many geometric configuration variables—such as the number of layers,

corrugation angle, spacers betweenmixing elements—can be used to impact the intensity of mixing

created within any given pipe diameter and create mixing solutions that are difficult to achieve with

any other technology.


Nowadays, SMs are used for a wide range of applications, such as chemical processing, wastewater

treatment, acid–base neutralisation, oxidation and bleaching, gaseous reactant blending, blending of

multi-component drugs, fertiliser and pesticide preparation, steam injection, organic-aqueous disper-

sions, oil and gas industry (Thakur et al., 2003; Edward, 2004).

10.2.3 MICROREACTORSMicroreactors—also known as micro-structured reactors or micro-channel reactors—are very small

devices, with channel dimensions of less than 1 mm, in which chemical reactions take place

(Figure 10.2, right). They are used in microprocess engineering, along with other devices involving phys-

ical processes (e.g. micro-heat exchangers and micro-distillation). Microreactors are devoted mainly to

processes at smaller production rate, as encountered in analytical chemistry, production of pharmaceu-

ticals and biochemicals, specialty chemicals and polymers, carrying out hazardous reactions or special

organic synthesis. Once the process development is solved for a single device, the scale-up to larger

throughput is simply solved bymultiplying the number of devices (numbering-up or scale-out). Examples

of industrial applications and processes using micro-devices can be found in specialised monographs

(Ehrfeld et al., 2000; Reschetilowski, 2013; Wirth, 2013).

Microreactors are a valuable tool for chemists and engineers, providing significant benefits:

• Continuous operation (typically) that allows the subsequent processing of unstable intermediates

with better selectivity and avoids batch workup delays and batch-to-batch variations.

• Different concentration profile as compared to batch reactors due to continuous operation and

mixing. Since in a microreactor the reactants are mixed almost instantly, none of them will be

Micronit microreactor (6 ml)

Bottom thickness 0.7 mm

Top thickness 1.1 mm

Channel width 150 mm

Channel depth 150 mm

Internal volume 6 ml

Materials all-glass

Depiction of how flow division and radial mixing can occur in a static mixer

Flow division

Radial mixing

FIGURE 10.2

Flow division and radial mixing that occur in a static mixer (left). Micronit (www.micronit.com) glass

microreactor—6 ml, 150 mm channel width and depth (right).


http://www.micronit.com

exposed to a large excess of the other reactants. Depending on the reaction mechanism, this may be

an advantage or disadvantage, so one must be aware of such different concentration profiles.

• Very small reaction times, of order of magnitude of seconds, which offer the possibility of isolating

valuable intermediate species in consecutive-type reactions.

• Because of the very small inventory, the microreactors are particularly adapted for carrying out

difficult organic synthesis processes involving hazardous and toxic reactants.

• Microreactors have high heat exchange coefficients, in the range of 1–500 MW/m3K, so they can

remove heat much more efficiently than classic vessels due to the high area-to-volume ratio. The hot

spots and duration of high temperature exposure also decrease significantly. Better control of the

reaction rates is possible as local temperature gradients are much smaller than in classic batch vessels.

Heating and cooling is much quicker, allowing a larger range of operating temperatures.

• Pressurisation of materials is typically easier than in traditional batch reactors, hence it is possible

to perform reactions at higher rates by raising the temperature beyond the boiling point of the

solvent, and improve the dissolution of gasses within the liquid flow stream.

Nonetheless, microreactors also suffer from several problems associated to their small scale, such as:

bad toleration of particles leading to clogging, corrosion, shorter residence time, pulsating flow, scaling

up issues to other types of vessels.

Microreactors can involve not only liquid–liquid systems but also solid–liquid systems (e.g. chan-

nel walls coated with a solid catalyst), and they are generally applied in combination with photochem-

istry, electrosynthesis, polymerisation, multi-component reactions, and purification of the product.

There are many hardware suppliers, offering various microreactors depending on the application focus:

• Ready-to-run (turnkey) systems: Used for new chemical synthesis schemes, for high throughput of

10–100 experiments per day and production scales ranging from 10 mg/exp to 1 kt/y.

• Modular (open) systems: Used for continuous process engineering layouts, where a measurable

advantage is anticipated. Multiple process layouts can be assembled on a scale ranging from 1 g/exp

to 100 kg/day. The engineering findings provide targets capacity of single-product plants.

• Dedicated developments: In the search of novel synthesis technologies, the manufacturers and

scientists set-up investigation and supply schemes, model a desired contacting pattern or spatial

arrangement of matter, and establish the overall application analytics until the critical initial

hypothesis can be validated and further confined (Wirth, 2013).

10.2.4 HIGH-GRAVITY TECHNOLOGYThe essence of high-gravity technology (HiGee) technology is replacing the gravitational field by a

high centrifugal field achieved by rotating a specially shaped rigid bed, typically a disc with an eye

in the centre. The higher mass-transfer coefficients and higher flooding limits allow the use of high

surface-area packing. In this way, the momentum, heat and mass transfer can be tremendously inten-

sified (Ramshaw, 1983). RPB technology is enabling PI in absorption, distillation, multiphase reactors

(e.g. trickle bed reactors), and production of micro- and nano-particles and ultrafine emulsions

(Rao et al., 2004; Reay et al., 2013; Reddy et al., 2006).

Another technology based on centrifugation is the Spinning Disk Reactor (SDR) that operates on

the principle of thin, wavy film flow generated when a liquid is introduced at the centre of a horizontal

disc surface rotating at high speeds. The important features of the thin films are highly sheared films


(typical thicknesses of 50–300 mm), very short residence times (few seconds) that are easy to control,

intense micro-mixing and plug-flow characteristics. The speed of rotation can reach 4000 rpm creating

centrifugal field accelerations of up to 1000�g at the edge of a typical disc (10 cm diameter). Heat and

mass-transfer studies have shown that high heat and mass-transfer coefficients can be achieved in an

SDR: convective film heat transfer coefficients of 20 kW/m2K, mass-transfer coefficients up to

kL¼30�10�5 m/s (liquid side) and kG¼12�10�8 m/s (gas side). These enhanced transport rates pro-

vides the proper environment in a SDR for fast and highly exothermic reactions.

HiGee distillation is using the RPB concept in a high-gravity field (100–1000�g) technology—claiming HETP values as low as 1–2 cm, about three to six times higher throughput and a volume re-

duction of two to three orders of magnitude as compared to that of conventional packed columns (Rao

et al., 2004; Stankiewicz and Moulijn, 2004; Wang et al., 2011; Reay et al., 2013).

Figure 10.3 shows a simplified schematic diagram of a typical RPBwith a vertical axis (Wang et al.,

2011; TU Dortmund website, 2013). The rotor is an annular, cylindrical packed bed housed in a casing

and driven by a motor. The liquid is fed onto the packing at the inner periphery, through a stationary

distributor located at the eye of the rotor. The liquid leaves the packing as a shower of droplets, col-

lected by the casing wall and runs downwards along the walls by the action of gravity, leaving the

casing. The gas/vapour is tangentially introduced into the casing, entering into the packing at the outer

periphery, and is forced to flow radially inward by pressure driving force. The gas/vapour leaves the

packing at the eye of the rotor through the outlet pipe (Wang et al., 2011).

Distillation in rotatingpacked bed (RPB)

HiGee distillationsystem with 2´ RPB

Liquid inlet

Vapour inlet

Vapour outlet

Liquid outlet Rotor shaft

Dynamic seal

Liquid distributor

Packed rotor

Casing

Feed

Stripping

Bottoms

Distillate

Rectification

Reflux

Condenser

Reboiler

Vapour

Liquid

VapourLiquid

FIGURE 10.3

Rotating packed bed (left) and HiGee distillation (right).


The main component of HiGee devices is the rotor, the structure of which determines the charac-

teristics of different devices. There are several types of rotating beds, described in more detail else-

where (Rao et al., 2004; Stankiewicz and Moulijn, 2004; Reay et al., 2013; Kiss, 2014): waveform

discs, helical rotating bed, multistage spraying rotating bed, RPB with blade packing, rotating zig-

zag bed (RZB) and rotating split packing bed.

The vapour–liquid counter-current flow is horizontal in case of HiGee distillation and not vertical,

as typical for conventional operation. Basically, this means that the capacity depends on the height of

the rotor, while the separation efficiency is given by the diameter of the rotor—in contrast to classic

distillation where the diameter gives the capacity, while the height of the column gives the separation

efficiency. Replacing the vertical gravitational force by the centrifugal force has many important ben-

efits, such as (Reay et al., 2013; Kiss, 2014):

• Very high volumetric mass-transfer coefficients, leading to reduced size of the equipment. Compact

equipment is also convenient for installation, troubleshooting and maintenance.

• Gas flow velocity can be dramatically increased and the tendency to flood is reduced, thus higher

hydraulic capacity is possible.

• The rotor is practically self-cleaning and it does not suffer from plugging, which is especially

beneficial to treatments of fouling and solid-containing systems.

• Larger driving force of liquid flow due to high rotational speed allows the use of non-Newtonian or

very viscous Newtonian fluids (stripping monomers, polymers solvents).

• Micro-mixing at molecular scale is extremely intensified (useful to make nano-particles).

• The gas–liquid contact in RPB is characterised by low liquid holdup, thus the time required to reach

steady state operation is drastically reduced. Moreover, the short liquid residence time contributes

to avoiding the decomposition of heat-sensitive materials (e.g. thermally unstable).

• The thinner liquid film and small inventories, favour processing of valuable materials.

• The RPB unit is unaffected by moderate disturbance in orientation, which allows its use in ship-

mounted or portable units where conventional distillation is not an acceptable option.

10.2.5 CYCLIC DISTILLATIONCyclic distillation (CyDist) emerged as an important trend for improving distillation performance by PI,

namely, by enhancing the separation efficiency through pseudo-steady-state operation based on separate

phase movement (SPM) and providing up to 50% energy savings (Maleta et al., 2011). Basically, cyclic

operation can be achieved by controlled cycling, stepwise periodic operation, a combination of these two,

or by stage switching. Controlled cycling appears to be the simplest scheme and it is therefore the pre-

ferred option. The cyclic operation was demonstrated on columns equipped with various types of inter-

nals: plates (brass, mesh-screen, bubble cap, sieve, packed-plate) and trays with sluice chambers.

Essentially, a CyDist column has an operating cycle consisting of two key operation parts: (1)

A vapour flow period, when vapour flows upwards through the column and liquid remains stationary

on each plate and (2) A liquid flow period, when vapour flow is stopped, reflux and feed liquid are sup-

plied, and liquid is dropped from each tray to the one below—as shown in Figure 10.4 (Patrut et al.,

2014a). As a result of cyclic operation, the achievable throughput is typically over two times higher

and lower vapour flow rates are necessary to achieve a certain purity. Moreover, the cyclic mode of op-

eration allows larger liquid holdups that can be beneficial for RD concepts, such as catalytic cyclic


distillation Figure 10.4 (Patrut et al., 2014b). Nonetheless, the limitations of cyclic operationmust be also

taken into account. The application of cyclic operation to vacuum distillation seems rather difficult, and

the performance enhancement critically depends on the complete separation between the liquid and va-

pour flow periods. However, the more recently proposed sluice-chamber trays seem to avoid the limi-

tations of simple trays (Maleta et al., 2011).

10.3 DIVIDING-WALL COLUMNThe separation of a zeotropic ternary mixture (ABC) typically requires a direct or indirect sequence of

at least two distillation columns. For some mixtures (e.g. when B is the major component and the split

between A and B is just as easy as the split between B and C), the direct separation sequence has an

inherent thermal inefficiency due to the remixing occurring for the mid-boiling component B—as il-

lustrated in Figure 10.5 (left). Note that a certain amount of energy is used to separate B to a maximum

concentration, but B is not removed at this point of high purity—it is actually remixed and diluted to a

lower concentration at which it is removed in the bottoms, together with heavy C.

A more energetically favourable alternative configuration that avoids the remixing of internal

streams is the so-called Petlyuk distillation or fully thermally coupled distillation columns (Petlyuk,

2004). However, note that a Petlyuk setup is not always the best choice, as there are mixtures and

Vapour

To reboiler

To condenser

Liquid

Vapour

Feed

To condenser

Liquid

Vapour

Feed

To condenser

Liquid

To reboiler To reboiler

Feed

A B C

2

1

3

NF

NT

NT-1

NT-2

FIGURE 10.4

Schematics illustrating the working principle of cyclic distillation: (A) vapor-flow period; (B) liquid flow-period;

(C) beginning of a new vapor-flow period.

40510.3 DIVIDING-WALL COLUMN

conditions that may favour implementation of alternative configurations. Although a fully thermally

coupled system always has the lowest minimum vapour flow, the energetic optimum strongly depends

on the feed composition. As illustrated in Figure 10.5 (right), certain conventional arrangements (e.g.

direct or indirect sequence) provide greater energy savings for lower contents of middle boiling com-

ponent in the feed, symmetric distribution of high and low boiling components, as well as large dif-

ferences in relative volatilities (Dejanovic et al., 2010).

In Petlyuk setup, the mixture is submitted firstly to pre-fractionation in two liquid and vapour

streams and then to the separation of the three components in the main column. The pre-fractionator

and the main column are linked by vapour and liquid flows, while the required condenser and reboiler

units are attached only to the main column. DWC is a practical implementation that allows further

equipment integration and cost savings by integrating the two columns of a Petlyuk configuration into

a single shell—as shown in Figure 10.6 (Kaibel, 1987; Asprion and Kaibel, 2010; Dejanovic et al.,

2010; Yildirim et al., 2011).

Compared to classic columns with a side draw, a DWC is capable of delivering higher purity side

product. The partition wall helps in avoiding the contamination between the feed side and the side draw

section of the column. More details about the specific internals of DWC are available at equipment

suppliers (www.montz.de), in review papers (Olujic et al., 2003; Yildirim et al., 2011) and specialised

monographs (Kiss, 2013a,b).

DWC technology found recently great appeal in the chemical industry—with Montz and BASF as

leading companies—because it offers major benefits: 25–40% lower energy requirements (when the feed

contains at least 20% of the mid-boiling component), high purity for all product streams, reduced main-

tenance costs, small footprint and up to 30% lower investment costs due to the reduced number of equip-

ment units (Dejanovic et al., 2010; Kiss, 2013a,b). Nonetheless, one should keep in mind that DWC has

Component B Profile in column 1

Component B Profile in column 2

Remixing occurs in column 1

Mole fraction

Col

umn

heig

ht

Top

0 1.0

Bottom

ABC

A

C

BC

BA

1 2

0.2 0.4 0.6 0.8 1.00

DS – Direct sequenceIS – Indirect sequenceSS – Side stripperSR – Side rectifier

aA/B = aB/C = 2.0

DS

IS

SS

SR

PetlyukDWC

0.2

0.4

0.6

0.8

0

1.0 B

A

C

Mole fraction

Mol

e fr

actio

n

B

A

C

FIGURE 10.5

Remixing ofmid-boiling component occurs in a direct-sequence arrangement (left). Optimality regions of different

configurations on the composition space (right).


http://www.montz.de

some limitations as well. DWC is operated at the same pressure, which can reduce its cost effectiveness

and practical applicability. For example, one cannot combine in a DWC unit, a distillation taking place at

vacuum with one at ambient or elevated pressure. Another drawback is that a DWC unit is larger in size

(diameter and height) as compared to any single column of a conventional separation sequence—which

in certain cases can render the application of DWC as impractical. In addition, the energy required by

DWC has to be supplied and rejected at the highest and the lowest temperature levels. This can reduce

the overall economy, since more expensive utilities must be used (Dejanovic et al., 2010).

10.3.1 DWC CONFIGURATIONSThe original DWC concept was further extended to other useful configurations—such as top and bot-

tom split columns, Kaibel column (two side-products) or even multi-partitioned DWC, as shown in

Figure 10.7 (Ghadrdan et al., 2011; Yildirim et al., 2011). DWC units can also be applied for the sep-

aration of more than three components. The number of possible configurations grows accordingly with

the increasing number of components.

For a three-component separation, two different DWC configurations can be applied. The first type

(Figure 10.6) is the most common (Yildirim et al., 2011): the dividing wall and the feed and side draws

are placed close to the middle of the column. The second configuration employs a bottom or top split

column (Figure 10.7): the wall is located at the lower or at the upper part of the column.

The bottom split column is referred to as split shell column with common overhead section and

divided bottoms section, while the top split column is known as split shell column with divided over-

head section and common bottoms section (Yildirim et al., 2011). Moreover, the wall can be shifted

from the centre towards the column walls and can have diagonal sections as well. In the Kaibel column

ABC

C

B

A

Dividing wall

Pre-fractionationsection

Main column

Liquid split

Vapour split

ABC B1 2

A

C

FIGURE 10.6

Petlyuk configuration (left) and dividing-wall column (right).


configuration, the separation is performed in a shell with one dividing wall while the two mid-boiling

products accumulate at the right side of the dividing wall. The configuration with only one wall is less

thermally efficient and it can be improved by using more dividing walls. This multi-partitioned setup is

referred to as the Sargent arrangement. In spite of the theoretical studies carried out so far, no industrial

application has been reported so far (Dejanovic et al., 2010). In the Agrawal arrangement, the feed

enters the middle partition of the DWC (Yildirim et al., 2011).

Moreover, the DWC technology is not limited to ternary separations alone, but it could be used also

in azeotropic, extractive, and RD—as shown in Figure 10.8 (Kiss, 2013a,b).

ABCD

I II III

A

B

C

DD

FIGURE 10.7

Alternative configurations: bottom and top split column, Kaibel column and multi-partitioned DWC (Sargent and

Agrawal arrangements).

DWCDWC

A,B

B

A, B, E

A,E

B

A

A,B

S

A B

S

ABC

C

B

A

AA--DWCDWCRR--DWCDWC EE--DWCDWC

FIGURE 10.8

Dividing-wall column used for reactive distillation (R-DWC), azeotropic distillation (A-DWC) and extractive

distillation (E-DWC).


10.3.2 CONSTRUCTION ASPECTSA proper selection of column internals is necessary to achieve efficient heat and mass transfer and,

hence, the required purity. Dividing wall columns can be equipped with trays or packing. Typically,

the selection criteria are similar to those for conventional distillation columns, but the wall construction

is different for tray and packed columns. Generally, tray DWC is easier to build and the dividing wall

that is welded on the column can strengthen the shell stability. The construction of a packed DWC is

more complex. Recently, the non-welded wall technology was developed. Using unfixed walls, the

column design becomes much simpler. Other benefits include fewer manholes and lower weight, since

the manufacturing requires less metal. The revamping of conventional columns becomes faster, sim-

pler and cheaper, too. Adopting non-welded partition wall enabled a significant increase of the appli-

cations with possibilities for revamping existing columns—so this can be considered as a milestone in

the implementation of DWC technology (Kiss, 2013a,b).

10.3.3 DESIGN, CONTROL AND APPLICATIONSThe design and control of DWC are nowadays quite well established, more information being available

elsewhere (Dejanovic et al., 2010; Kiss and Bildea, 2011; Kiss, 2013a,b). Several shortcut and detailed

methods are described in literature for the design of DWC, and rigorous DWC simulations can be per-

formed with commercial process simulators (Aspen Plus, ChemCAD, HYSYS, ProSim). The optimal

design of a DWC requires adequate models and computer-based simulations. However, commercial

process simulators do not include particular subroutines for DWC units. The so-called decomposition

method simplifies the design problem, as the existing DWC configuration is replaced by a sequence of

conventional distillation columns.

The literature reveals that there are several design methods available that concern mostly ternary

separations. However, they can be relatively easily extended to cover cases with more components.

When designing a DWC system for separation of a three-component feed into three products, the num-

ber of degrees of freedom (DoF) (i.e. design parameters) increases as compared to that required in case

of designing conventional configurations of two columns in series—where the two columns can be

optimised independently of each other (Dejanovic et al., 2010). The required design parameters for

a DWC are number of stages in six different sections (e.g. common top and bottom sections, sections

above and below the feed stage and side draw, respectively), vapour split ratio, liquid split ratio, reflux

ratio, heat input in the reboiler and the side-product flow rate.

The initial steps are similar to designing conventional columns: selecting column configuration

(e.g. number of stages, partition wall length, feed and side-stream locations) and operating pressure,

combined with an appropriate VLE model. However, the next steps differ considerably, including

establishing initial DWC configuration, shortcut or detailed design, stage and reflux requirements, op-

timisation, equipment sizing, and process control system—these steps are explained in more details

elsewhere (Kiss, 2013a,b). The following list of heuristics provides good initial estimates for both

shortcut and detailed simulations:

• Design a conventional two-column system as a base case (e.g. in-/direct sequence).

• Take the total numbers of stages for DWC as 80% of the total number of stages required for the

conventional two-column sequence: NDWC¼0.8 (N1+N2).

• Place the partition (i.e. dividing wall) in the middle third of the column (e.g. 33–66% H).


• Set the internal flow rates in the DWC, as determined by the reboiler or condenser duty, at 70% of

the total duties of two conventional columns: QDWC¼0.7 (Q1+Q2).

• Use equalised vapour and liquid splits (rL¼0.5, rV¼0.5) as initial values.

However, this is only enough to have a good start with the simulation, as it requires a lot of tuning in

order to achieve optimal DWC design. Carrying out rigorous DWC simulations requires certain expe-

rience and is rather computationally demanding, depending on the configuration chosen to represent

the DWC (e.g. 1–4 column models) and the modelling approach (shortcut or detailed).

Another simple yet effective design method is based on a graphical treatment of the minimum en-

ergy represented by normalised vapour flow, as a function of the feed distribution. The method is based

on the Underwood’s equations, using these assumptions: constant molar flows, infinite number of

stages, and constant relative volatilities (Halvorsen and Skogestad, 2003, 2011). The required input

parameters are the feed composition, the feed quality expressed by the liquid fraction, relative vola-

tilities and desired product purities or recoveries. The Underwood equations are then used to determine

the minimum vapour (Vmin) and liquid flows needed to perform all the binary separations of a specified

feed mixture. These are in fact product splits occurring in each section of the column, assuming an

infinite number of theoretical stages. In practise, the infinite number of stages can be approximated

by setting the number of stages for each simulation to at least 4 Nmin as calculated by the Fenske equa-

tion. This value is confirmed by simulations showing that no decrease of reboiler heat duty can be

achieved by further increasing the number of stages. Note that the shortcut method Fenske–

Underwood–Gilliland is explained in Chapter 20.

The Vmin diagram conveniently shows the vapour and liquid traffic needed in every column section,

which can directly serve as basis for the column design. The basic claim of the Vmin method is that the

minimum vapour flow required for the separation of n components feed into n pure products in any

arrangement corresponds to that required for the most difficult binary split (shown as the highest peak

in the Vmin diagram)—getting all the other separations ‘for free’. The number of stages can be prelim-

inary considered as two times the minimum number, as determined using the Fenske equation. The

Vmin diagram plots the vapour flow rate above the feed (V/F) versus the net flow of the top product

(D/F) per unit of feed. For each given pair (D/F, V/F), all the other properties are completely deter-

mined, such as all component recoveries and product compositions. The feed enthalpy condition is

given by the liquid fraction (q) in the feed stream.

Figure 10.9 plots the Vmin diagram for an equimolar ternary system (benzene–toluene–xylene or

ABC). The Vmin diagram shows how the feed components of a ternary feed (ABC) are distributed

to the top and bottom products in a simple two-product ‘infinite stage’ distillation column as a function

of the operating point (D/F, V/F). For values of V/F above the upper boundary following the three peaks

in the diagram (0,0-PAB-PAC-PBC-1,0), the column is over-fractionating meaning that valuable energy

is wasted. Note that the point located at x,y¼1,0 in Figure 10.9 is more generally defined as 1,(1�q)but in this particular case a saturated liquid feed stream was considered (q¼1).

The values at the peaks (PAB, PAC, PBC) give the vapour flow for the corresponding sharp neighbour

component splits. The knots are Vmin for the so-called preferred splits where a sharp split between twokey components is specified, while allowing intermediate components to be distributed. Only the sharp

split between each possible pair of key components must be solved in order to find the diagram for a

multi-component feed. For the example shown in Figure 10.9, only three points are needed: PAB: sharp

A/B split, PBC: sharp B/C split and PAC: sharp A/C split. PAC is the preferred split that is the minimum


energy operating point for a sharp separation between the heavy and light keys while the intermediate

distribute to both column ends. At any operating point at or above the V-shaped PAB-PAC-PBC, a sharp

A/C split is obtained but with higher energy than the one required at the exact point PAC (Halvorsen and

Skogestad, 2011). Note that the minimum energy required for the ternary separation in a DWC corre-

sponds to the highest peak (i.e. PBC in Figure 10.9).

In spite of the clear advantages of DWC and the steady increase of industrial applications, the

spreading of DWC was still limited until a few years ago to just a few companies. One reason for this

was the insufficient insight on the operation and control of a DWC—this lack of knowledge making

most companies reticent to large-scale implementations. Although more difficult to control as com-

pared to conventional columns, the recent studies and the industrial experiences indicate that the con-

trol of DWC units is in fact satisfactory (Kiss and Bildea, 2011). Dynamic simulations can be used

additionally to provide insight into the dynamic behaviour of the DWC system, and a valuable guidance

for choosing the right control strategy. The paper of Kiss and Bildea (2011) gave an overview of the

available control strategies for DWC, varying from the classic three-point control structure and PID

controllers in a multi-loop framework to model predictive control and other advanced control strategies

(LQG, LSDP, H1 and m-synthesis). The available results show that MIMO controllers perform better

than multi-loop PID controllers. However, among the decentralised multivariable PI-structured con-

trollers, LSV and DSV are the best control structures being able to handle persistent disturbances in

reasonably short times (Kiss and Bildea, 2011).

0

0.2

0.4

0.6

0.8

1

1.2

0 0.2 0.4 0.6 0.8 1

V/F

/ (-

)

D/F / (-)

PAB

PBC

PAC

ABC / ABC

AB / ABCA / ABC ABC / BC ABC / C

Preferred split sharp

Sharp split A / BC

AB / BC

Sharp split AB / C

BSD

1−q

V Top V PF V SS

FIGURE 10.9

Vmin diagram of a BTX ternary system (equimolar mixture).


Ever since its first industrial application in 1985, DWC moved from a conceptual to a proven tech-

nology, steadily growing in number and size of applications (Kaibel, 1987; Yildirim et al., 2011).

Table 10.1 conveniently summarises the current DWC applications available at applied research

and industrial scale (Kiss, 2013a,b).

10.4 REACTIVE DISTILLATIONIn RD, significant benefits can be realised by using the reaction to improve separation (e.g. overcoming

azeotropes, reacting away contaminants, avoiding difficult separations) or by using separation to

Table 10.1 Reported Applications of Dividing-Wall Columns (Yildirim et al., 2011; Kiss, 2013a,b)

Applied Research and Industrial Applications Additional Information/Remarks

Ternary separations

Benzene–toluene–xylene fractionation ExxonMobil

Separation of hydrocarbons from Fischer–Tropsch synthesis unit Linde AG, tray column, H¼107 m, D¼5 m

Separation of benzene from pyrolysis gasoline Uhde, 170,000 m/year feed capacity

Separation of C7+aromatics from C7+olefin/paraffin UOP, five DWC, trap tray

Mostly undisclosed systems BASF/Montz, over 70 columns, D¼0.6–4 m,

P¼2 mbar to 10 bar

Undisclosed systems Sumitomo Heavy Ind., Kyowa Yuka, Sulzer

Chemtech Ltd. (20 DWC), Koch–Glitsch (10

DWC)

Multi-component separations

Recovery of four component mixtures of fine chemical

intermediates

BASF/Montz, single wall column, H¼34 m,

D¼3.6 m, deep vacuum

Integration of a product separator and an HPNA stripper UOP, five product streams

Retrofit of conventional columns to DWC

Recovers mixed xylenes from reformate motor gasoline Koch–Glitsch, D¼3.8–4.3 m, tray column,

over 50% energy savings

Separation of (iso)paraffins. Production of isohexane Koch–Glitsch

Separation and purification of 2-ethylhexanol (2-EtH) Dual operation possible

Reactive DWC

Esterification, trans-esterification, etherification Rate-based model/Aspen Tech ACM, Aspen

Plus, Aspen HYSYS or Pro/II

Azeotropic DWC

Ethanol dehydration Entrainer: cyclohexane, n-pentane

Extractive DWC

Separation of toluene and non-aromatics withN-formyl-morpholine Uhde, 28,000 m/year feed capacity

Crude butadiene from a crude C4 using N-methyl-pyrrolidone

(NMP) as solvent

BASF, both trays and packing

Bioethanol dehydration Ethylene glycol used as solvent


improve reactions (e.g. overcoming chemical equilibrium limitations, and improving selectivity)—

maximum effect being achieved when both aspects apply. Mathematical modelling including

rate-based description can be found in the paper of Taylor and Krishna (2000). Theoretical issues

and applications are described by various specialists in the monographs edited by Sundmacher and

Kienle (2003) and Sundmacher et al. (2005). Design, simulation and control issues are handled in

the book of Luyben and Yu (2008), while new developments are described in the monograph of

Kiss (2013a,b). Remarkable, over 1100 articles and 800 US patents on RD were published during

the past 40 years, covering in total over 235 reaction systems (Luyben and Yu, 2008).

In a RD process, the reaction and distillation take place in the same piece of equipment, the reactants

being converted with the simultaneous separation of the products and internal recycle of unused reac-

tants. This implies that the boiling points of the products must be different from those of the reactants,

preferably highest and lowest in order to remove the products as top and bottom streams. Since both

operations occur simultaneously in the same unit, there must be a proper match between the conditions

required for reaction and separation. However, the application of RD is somewhat limited by con-

straints, such as common operation range for distillation and reaction in terms of temperature and pres-

sure, favourable boiling point sequence (product should be the lightest or heaviest component, while

side or by-products the mid-boiling ones) and difficulty in ensuring sufficient residence time by the LV

traffic for completing the reaction. The major constraint is set by the LV phase equilibrium on the

chemical reaction, which takes place in liquid phase. By this effect, the actual concentration of reac-

tants in the liquid phase is smaller than in a pure liquid-phase reaction. Accordingly, the reaction rate

should be increased by higher temperature and higher pressure, which in many cases is not practical.

The best solution is boosting the reaction rate by means of a catalyst. Therefore, employing a (solid)

catalyst is usually inevitably in RD. For this reason RD is often designated in industry by the term cat-

alytic distillation. The catalyst developed for RD should be much faster than for operating a homoge-

neous liquid-phase reactor, in order to bring the reaction rate compatible with the residence time

resulting from the hydraulics. An example is the zeolite catalyst used in alkylation reactions.

RD setups may consist of multiple catalyst systems, gas and liquid traffic over the catalyst, sepa-

ration, mass flow, and enthalpy exchange—all of them being optimally integrated in a single proces-

sing unit, a key feature of PI. By continuously removing the products, RD makes it possible to use only

the stoichiometric reactants ratio (neat operation) and to pull the equilibrium to high conversions

(Luyben and Yu, 2008). This is in contrast to the typical practise of using an excess of one of the re-

actants to push the equilibrium towards the desired products, at the penalty of having to recover and

recycle the unreacted reactant (Kiss, 2013a,b).

10.4.1 RESIDUE CURVE MAP REPRESENTATIONRD is characterised by the simultaneous occurrence of chemical and phase equilibrium (C&PE). This

should be the starting point of a feasibility analysis. Useful insights of a C&PE can be found by graph-

ical representations, as Residue Curves Maps. Here, we present only the general frame. For more the-

ory, the reader may consult the books of Stichlmair and Fair (1998) and Doherty and Malone (2001), as

well as Dimian and Bildea (2008).

Let us consider the general equilibrium reaction:

nAA + nBB + � � �$ nP + nRR+ � � � orXc

i¼1

niAi ¼ 0 with nt ¼Xc

i¼1

ni (10.1)

41310.4 REACTIVE DISTILLATION

The chemical equilibrium constant formulated by means of activities is given by the following

expression:

Keq ¼ avPP avRR � � �avAA avBB � � � ¼

xvPP xvRR � � �xvAA xvBB � � �

gvPP gvRR � � �gvAA gvBB � � � ¼

Yi

xigið Þvi ¼KxKg (10.2)

The composition can be expressed with respect to a reference species k as follows:

xi ¼ xi0 vk�vtxkð Þ+ vi xk�xk0ð Þvk�vtxk0

(10.3)

This relation describes the so-called stoichiometric lines (Stichlmair and Fair, 1998), which help the

graphical representation, and the introduction of transformed variables. These lines converge into a

pole p, whose location is:

xi ¼ viXi

vi¼ vivt

(10.4)

Note that when the number total of moles does not change by reaction (nt¼0), the stoichiometric lines

are parallel.

A residue curve characterises the evolution of the liquid composition in a vessel during a batch-wise

RD experiment. The residue curve map is obtained by considering different initial mixture composi-

tions. For non-reactive mixtures the RCM is obtained by solving the following differential equation:

dxidx

¼ xi�yi (10.5)

where x¼H/V is a ‘warped-time’ defined as the ratio of molar liquid holdup H by the molar vapour

rate V, while xi and yi are vapour and liquid compositions, respectively. A similar representation based

on distillation lines describes the composition on successive trays of a distillation column with infinite

number of stages at infinite reflux (1/1 analysis). In contrast with relation describing the stoichio-metric lines, the distillation lines may be obtained by algebraic computations involving series of bubble

and dew points, as follows:

xi,1 ! y�i,1 ¼ xi,2 ! y�i,2 ¼ xi,3 ! yi,3 . . . (10.6)

Figure 10.10 (left) shows the construction of a distillation line for an ideal ternary mixture in which

A and C are the light (stable node) and the heavy (unstable node) boilers, while B is an intermediate

boiler (saddle). The initial point xi,1 produces the vapour yi,1* which by condensation gives a liquid with

the same composition such that the next point is xi,2¼yi,1* , etc. Accordingly, the distillation line de-

scribes the evolution of composition on the stages of a distillation column at equilibrium and total re-

flux from the bottom to the top. The slope of a distillation line is a measure of the relative volatility of

components. The analysis by RCM or DCM leads to similar results.

When a chemical reaction takes place the residue curves can be found by the equation:

dxidx

¼ xi�yi +Da vi�vTxið ÞR (10.7)


where Da is the Damkohler number given by the ratio of the characteristic process time H/V to the

characteristic reaction time 1/r0. The reaction rate r0 is the reference value at the system’s pressure

and at an arbitrary reference temperature, usually the lowest or the highest boiling point. For catalytic

reactions, r0 includes a reference value of the catalyst amount. R is the dimensionless reaction rate

R¼ r/r0. The kinetics of a liquid-phase reaction is described as a function of activities:

r¼ kY

avii

� �pr�

Yavjj

� �eq=Keq

� �(10.8)

Thus, the parameter Da is a measure of the reaction rate, but its absolute value cannot be taken as the

basis for comparing different systems.

Analogous with the procedure presented before, RD lines can be computed by a series of dew and

bubble points incorporating the chemical equilibrium, as follows:

xi,1 ! y�i,1 $ xeqi,2 ! y�i,2 $ xeqi,3 ! y�i,3 . . . (10.9)

Graphical construction of RD lines at equilibrium is shown in Figure 10.10 (right) for the reversible

reaction A+B$C. The initial point xi,1, at chemical equilibrium produces a vapour yi,1* , which by con-

densation and equilibrium reaction gives the liquid with the composition xi,2. This is found by crossingthe stoichiometric line passing through yi,1* with the chemical equilibrium curve. Then the liquid xi,2produces the vapour yi,2* , and so on. Similarly the points 11, 12, 13, . . . N show the situation in which

the mixture becomes richer in B and poorer in A and C. Figure 10.10 (right) emphasises a particular

position where phase equilibrium and stoichiometric lines are co-linear. The liquid composition re-

mains unchanged because the resulting vapour, after condensation, is converted into the original

Xi,11

Xi,2

Xi,3

y*i,1

y*i,2

y* i,3

Xi,1 y*i,az

C B

A

Xi,12 Xi,13

y*i,12

y* i,13

Xi,az

°°

°

°°

°°

VL equilibrium

C B

A

Chemical equilibrium

1°°

°

°°

°°

2

3

1112

13

z

Xi,2

Xi,3

Xi,4y*i,2

y*i,3

y*i,4

Xi,1y*i,1

A

Xi,5

C B

A

FIGURE 10.10

Construction of the distillation lines for non-reactive (left) and reactive mixtures (right).


composition. This point is a potential reactive azeotrope, but when the composition satisfies chemical

equilibrium too it becomes a true reactive azeotrope.Figure 10.11 illustrates the construction of a RCM for the reversible reaction A+B$C for which

the relative volatilities are in the order 3/2/1 and the equilibrium constant Kx¼6.75. The physical and

RD lines may be obtained simply by computation in Microsoft Excel using the above equations. Note

that in this case the starting point, liquid with composition (0.1, 0.l, 0.8), is not at chemical equilibrium.

The coordinates of the triangle are in normal mole fractions. It may be observed that after a short

straight path the RD line superposes the chemical equilibrium curve. The same trend is observed also

when starting from other points. Figure 10.11 illustrates also graphically the formation of a reactive

azeotrope as the point where a particular stoichiometric line becomes tangential to the non-reactive

residue curve and intersects simultaneously the chemical equilibrium curve. For more than three spe-

cies, a change of variables appears useful in order to reduce the dimensionality of a graphical repre-

sentation. More generally, the composition of a reacting system characterised by cmolar fractions can

be reduced to (c�1) new composition variables by the following transformation:

Xi ¼ nkxi� nixknk� ntxk

(10.10)

The reference k component should be preferably a product.

The kinetics of a reaction rate has a substantial influence on residue curve maps. Distillation bound-

aries and physical azeotropes can vanish, while other singular points due to kinetic effects might ap-

pear. The influence of the kinetics on RCM can be studied by integrating the equation for finite Danumbers. In addition, the singular points satisfy the relation:

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1C

A

Distillation lines

Reactive distillation line

Chemical equilibrium

Stoichiometric lines

Reactive azeotrope

B

A+B<=>C

a

1

2

3

4

5

b

π

FIGURE 10.11

Residue curves for a ternary mixture involving the equilibrium reaction A+B$C.


DaR¼ xi�yixi� nTxi

(10.11)

For example, Thiel et al. (1997) showed the influence of the reaction rate on the RCM of the mixture

isobutene/methanol/MTBE at 8 bars. Several RCM plots are shown: (a) The physical phase equilib-

rium (Da¼0) with two physical azeotropes, MeOH/MTBE (x¼0.55) and MeOH/IB (x¼0.08), the

first saddle and the second unstable node, both linked by a distillation boundary. Two distillation fields

appear above and below the separatrix in whichMeOH andMTBE are stable nodes. (b) The situation at

Da¼10�4 when the reaction is slow. The upper distillation region is not affected by reaction, contrary

near to the MTBE corner. The stable point moves from MTBE to a new position containing about

76.5% MTBE, 20% MeOH and 3.5% IB. (c) Raising Da to 2�10�4 leads to a situation in which

the saddle point disappears and all trajectories point-out to the methanol vertex; the reaction rate be-

comes dominant. (d) ForDa¼1, when the reaction rate is high enough, the trajectories are collected by

the chemical equilibrium curve.

Note that the influence of the kinetics of a chemical reaction on the vapour–liquid equilibrium is very

complex. As a consequence, evaluating the kinetic effects on residue curve maps is of great importance

for conceptual design of RD systems. However, it can be appreciated that in practise the reaction rate is

fast enough such that the chemical equilibrium is reached quickly. The RCM simplifies considerably.

But even in this case the analysis may be complicated by the occurrence of reactive azeotropes.

10.4.2 MODELLING REACTIVE DISTILLATIONThe simulation of RD processes can consider two types of fundamental models: equilibrium stage

models (EQ) and non-equilibrium stage modelling (NEQ). The merits and disadvantages of each ap-

proach, as well a direct comparison, are discussed in the landmark paper of Taylor and Krishna (2000)

that is recommended for deeper study. The EQ modelling can be formulated at two levels:

• Simultaneous phase and chemical equilibrium.

• Phase equilibrium with chemical kinetics.

The full equilibrium model requires only thermodynamic knowledge. RCM can greatly help to high-

light the range on feasible design in term of pressure, temperature, and separation of products. The

simulation is in general easy, but care should be paid to the accuracy of thermodynamic properties,

phase equilibrium and chemical equilibrium. Thus, the model based on phase and chemical equilibrium

allows a rapid assessment of the feasibility of a RD process (Noeres et al., 2003).

The simulation becomes more realistic adding the knowledge of the chemical kinetics. The progres-

sion of the reaction on each stage can be followed, and consequently the number of theoretical stages for

achieving a target conversion can be obtained. A key parameter in the kinetic approach is the reaction

holdup. Accordingly, the selection of internals and hydraulic pre-design are necessary. This topic will be

discussed in the next section. It should be stressed that an accurate knowledge of the reaction rate ex-

pression is necessary, which can be extrapolated over the interval of composition and temperature. This

is a central point in RD and a major source of failure. The reaction rate must be expressed adequately,

either on pseudo-homogeneous basis (volume), or per mass of catalyst. Moreover, using concentration

instead of activities could introduce large errors, when highly non-ideal mixtures are handled, namely,

when containing water.


In the NEQ modelling, the intensity of the interfacial mass transfer in liquid and vapour phases are

counted for, by using the Maxwell–Stefan equations. The availability of specific correlations for cal-

culating the mass-transfer coefficients is necessary, which in turn depend on the selected internals. The

potential accuracy of this approach is paid by a much more elaborated procedure that needs customised

programming. The comparison with experiments showed that the NEQ modelling gives good results if

the accurate model parameters are employed.

10.4.3 DESIGN OF RD PROCESSESRD sets specifications on both product compositions and reaction conversion. Consequently, the DoF

in a RD column must be adjusted to accomplish these specifications while optimising an objective

function such as the total annual cost (TAC). The following specifications are usually required:

• Column pressure and pressure drop on column or stage. Setting the pressure is constrained by the

temperature of top and bottoms, more specifically by the available of on-site cold and hot

utilities. If the bottom temperature is excessive, the solution may be the dilution with reactant,

which will be recovered separately. In general, working at the highest acceptable temperature is

recommended, because of its accelerating effect on the reaction rate.

• Number of stages, feed locations of reactants and exit points of the product streams. In this way, the

RD configuration is set-up in term of rectification, reaction and stripping sections.

• Top distillate (liquid, vapour, mixed) or bottom product, absolute or ratio values.

• Condenser and reboiler types.

• Reflux or boilup ratio.

• Holdup distribution on the reactive stages.

Unlike conventional distillation, the choice of internals for RD is much more limited (Krishna, 2002).

Figure 10.12 presents some examples of catalytic packing. The most important are:

• Catalytic Raschig rings with surface-coated catalyst.

• Catalyst bales, formed by wrapped wire sheets filled with catalyst.

• Structured packing. The elements have the shape of sandwiches manufactured from corrugated wire

gauze sheets hosting catalyst bags, assembled as cylinders or rectangular boxes. Conceptually,

the packing structure consists of alternating catalyst bags and open channel spaces. For ensuring

higher efficiency of combined reaction and diffusion the catalyst particle should be as small as

possible, but large enough for acceptable pressure drop. Particle diameter of about 0.8–1 mm gives

a good compromise. The advantages of structured packing are uniform flow conditions with

minimum backmixing and maldistributions; good radial dispersion, an order of magnitude better

than in conventional packed beds, ensuring a longer residence time; and efficient maintenance

and replacement of catalyst by service procedures.

Among the commercial offer, one can mention Katapak-S® manufactured by Sulzer ChemTech

(Goetze et al., 2001), and Multipak® supplied by Julius Montz (Hoffman et al., 2004).

Concerning the hydraulic design, we illustrate the method by using the structured packing Multi-

pak®, which was fully characterised by correlations for hydraulics and mass transfer (Hoffman et al,

2004). The elements may be shaped as cylinders or boxes, such to cover the entire cross sectional area

of the column. The geometrical parameters are the catalyst volume fraction c, the void fraction E and


the specific surface area s (m2/m3). As it can be seen in Figure 10.13, they vary considerably with the

column diameter. Thus, over the range 50 mm to 4 m column diameter, the catalyst volume fraction

takes values from 0.18 to 0.38 forMultipak-I, while forMultipak-II from 0.28 to 0.56. The void fraction

diminishes accordingly. The specific area varies between 300 to 400, and 250 to 320 m2/m3,

respectively.

In contrast with the sizing procedure of distillation columns, which is based on the vapour phase

load at the flooding point, the hydraulic computation of a RD column is based on the liquid-phase flow.

From physical viewpoint, the structure can be viewed as assembled by catalyst bags, which is the re-

action space, and open channels, through which the two-phase mixture flows in counter-current. The

catalyst must be completely wet by liquid, which happens at the ‘load point’. At lower flow rate the

wetting is partial, while at higher load the excess liquid overflows in the open channels and does not

take part in reaction. Therefore, a RD column should be operated slightly above the load point. The

vapour phase velocity is not important as long as flooding does not occur. The liquid velocity should

be examined with respect to residence time distribution, which should approach closely a plug flow.

Liquid loads of 5–20 m3/m2/h seem the most practical, with optimum around 10 m3/m2/h. The liquid

velocity ULP at the load point can be determined using the following relations:

Spherical basketsCylindrical container for

catalyst particles Wire-gauze envelopes

Gas / vapour

Top

Catalyst bales by Chemical Research and Licensing

Front

Gas / vapour

Liquid

Horizontally disposed gutters

Catalystparticles

Structured catalyst sandwiches by Sulzer and Koch–Glitsch

FIGURE 10.12

Type of internals (packing) employed in RD (Krishna, 2002; Kiss, 2013a,b).


U2LP ¼

e3CBgdp1� eCBð ÞxCB

with xCB ¼160

ReLP+

3:1

Re0:1LP

and ReLP ¼ ULPdprL1� eCBð Þ�L

(10.12)

where ReLP is the Reynolds number and xCB the friction factor, while the other notations are ECB voidfraction catalyst bag, dp particle diameter, rL liquid density, and �L liquid viscosity.

The liquid holdup hl, is given by three terms: catalyst bags, open channels and wire gauze. The same

holds for pressure drop calculation. The most important is the contribution of the catalyst bags, for

which the following simple relation holds:

hl,CB ¼ceCB (10.13)

Experimental laboratory data for Multipak-II given in Figure 10.14 show that the liquid holdup is prac-

tically independent to the gas load over a large range. As a typical value, at a mean superficial liquid ve-

locity of 11 m/h the holdup is about 0.22 m3 liquid/m3 packing. On the other hand, for goodmass transfer

the gas load expressed by theF-factorF¼UgffiffiffiffiffiffirG

pshould be in the range 0.5–1.5 Pa0.5. The pressure drop

varies proportionally with the gas load. A value of 2 mbar/m is recommended for preliminary design.

The last element in analysis is the number of theoretical stages per metre (NTSM). Laboratory ex-

periments indicate a value of three to six for Multipak. On the other hand, NTSM of two or three seems

more appropriate from industrial viewpoint. With the above considerations, the following shortcut

method for the hydraulic designing of a RD column can be formulated:

1. Estimate a mean volumetric liquid flow rate for operation.

2. Assume an initial value for the superficial liquid velocity at the ‘load point’ (ULP): recommended

10 m3/m2/h.

0.1

Column diameter (m)

Voi

d fr

actio

n, c

ata

lyst

vol

ume

frac

tion

(-)

Spe

cific

sur

face

are

a (m

2 /m

3 )

Multipak® -I

0

100

200

300

400

Catalyst volume fraction

Void fraction

A B

Specific surface area

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

1 40.05 0.1

Column diameter (m)V

oid

frac

tion,

ca

taly

st v

olum

e fr

actio

n (-

)

Spe

cific

sur

face

are

a (m

2 /m

3 )

Multipak® -II

0

100

200

300

400

Catalyst volume fraction

Void fraction

Specific surface area

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

1 40.05

FIGURE 10.13

Characteristic geometric data of Multipak-I (left) and Multipak-II (right) structured packings (Hoffman et al.,

2004).


3. Assume an initial value for the NSTM.

4. Determine the column diameter. Knowing the packing specifications, estimate the volume of

packing and the catalyst holdup per reaction stage.

5. Introduce the above values in simulation, in which the reaction rate is expressed in units compatible

with the holdup, mass or volumetric.

6. Determine the total number of reactive stages needed to achieve the target conversion. Pay attention

to the profiles of temperatures, concentrations, and reaction rate. Extract liquid and gas flows, as

well as fluid properties.

7. Recalculate the load point velocity, the liquid holdup from the above information by using specific

correlations and diagrams. Check the hydraulic design by selecting packaging with similar

characteristics.

8. Verify if the gas load and the pressure drop are within optimal region.

Afterwards, check all values and repeat the points 4–8 until acceptable values are achieved.

Note that the mentioned paper of Hoffman et al. (2004) handles also the topic of mass-transfer cor-

relations and the use of NEQ models. The good agreement between modelling and experiments for

methyl acetate synthesis in a pilot plant demonstrates the applicability of the NEQ approach to the

design of industrial RD processes.

10.4.4 APPLICATIONS OF REACTIVE DISTILLATIONTable 10.2 lists the most important applications: (trans-)esterification, hydrolysis, etherification, hy-dration and dehydration, (trans-)alkylation, isomerisation, (de-)hydrogenation, amination, condensa-

tion, polyesterification, chlorination, nitration—all being equilibrium limited (Sundmacher and Kienle,

2003; Kiss, 2013a,b). Figure 10.15 illustrates some RD configuration alternatives, ranging from a con-

ventional RD column to reactive DWC, and RD columns combined with a pre-reactor, side reactors or

even membrane separation units (Kiss, 2013a,b).

0 0.3 0.5

100

101

1 2 30.1

0.2

0.3

1

Liquid load uL (m3/m2h)

Liquid loaduL (m

3/m2h)

Gas load (Pa0.5)

Pre

ssur

e dr

op (

mba

r/m

)

Gas load (Pa0.5)

Liqu

id h

oldu

p h L

(-)

2

2.8 0.0 5.5

29.0

11.15.5 11.1

20.7 20.729.0

FIGURE 10.14

Liquid holdup (left) and pressure drop (right) for Multipak-II of 100 mm diameter (Hoffman et al., 2004).


Table 10.2 Main Industrial Applications of Reactive Distillation (Kiss, 2013a,b)

Reaction Type Catalyst/Internals

Alkylation

Alkyl benzene from ethylene/propylene and benzene Zeolite b, molecular sieves

Amination

Amines from ammonia and alcohols H2 and hydrogenation catalyst

Carbonylation

Acetic acid from CO and methanol/dimethyl ether Homogeneous

Condensation

Diacetone alcohol from acetone Heterogeneous

Bisphenol-A from phenol and acetone N/A

Trioxane from formaldehyde Strong acid catalyst, zeolite ZSM-5

Esterification

Methyl acetate from methanol and acetic acid H2SO4, Dowex 50, Amberlyst-15 N/A

Ethyl acetate from ethanol and acetic acid Katapak-S

2-Methyl propyl acetate from 2-methyl propanol and acid Cation-exchange resin

Butyl acetate from butanol and acetic acid H2SO4, Amberlyst-15, metal oxides

Fatty acid methyl esters from fatty acids and methanol H2SO4, Amberlyst-15, metal oxides

Fatty acid alkyl esters from fatty acids and alkyl alcohols Ion-exchange resin bags

Cyclohexyl carboxylate from cyclohexene and acids

Etherification

MTBE from isobutene and methanol Amberlyst-15

ETBE from isobutene and ethanol Amberlyst-15/pellets, structured

TAME from isoamylene and methanol Ion-exchange resin

DIPE from isopropanol and propylene ZSM 12, Amberlyst-36, zeolite

Hydration/dehydration

Mono ethylene glycol from ethylene oxide and water Homogeneous

Hydrogenation/dehydrogenation

Cyclohexane from benzene Alumina supported Ni catalyst

MIBK from benzene Cation-exchange resin with Pd/Ni

Hydrolysis

Acetic acid and methanol from methyl acetate+water Ion-exchange resin bags

Acrylamide from acrylonitrile Cation exchanger, copper oxide

Isomerisation

Iso-paraffins from n-paraffins Chlorinated alumina and H2

Nitration

4-Nitrochlorobenzene from chlorobenzene+nitric acid Azeotropic removal of water


CDTECH (CB&ILummusTechnology), themajor commercial RD technology provider, has licenced

until now more than 200 commercial scale processes operated worldwide at capacities of 100–3000 ktpy

for production of ethers (MTBE, TAME, ETBE), hydrogenation of aromatics and light sulphur hydro-

desulphurisation, ethyl benzene and isobutylene production. Sulzer ChemTech also reports several indus-

trial scale applications such as: synthesis of ethyl, butyl andmethyl acetates, hydrolysis ofmethyl acetate,

synthesis of methylal, and fatty acid esters production (Harmsen, 2007, 2010; Kiss, 2013a,b).

10.4.5 FEASIBILITY AND TECHNICAL EVALUATIONThe technical feasibility and economical attractiveness of RD processes can be evaluated using the

schemes recently proposed by Shah et al. (2012). The proposed framework for feasibility and technical

evaluation of RD allows a quick and easy feasibility analysis for a wide range of chemical processes.

Basically, the method determines the boundary conditions (e.g. relative volatilities, target purities,

equilibrium conversion and equipment restriction), checks the integrated process constrains, evaluates

the feasibility and provides guidelines to any potential RD process application. Providing that a RD

RD with pre-reactor and side reactors

RDC

RX

RX

EQ

Reactive distillation column

RDC

Feed heavy

Feed light

Reactive zone(catalyst)

Rectifying zone

Stripping zone

Distillate

Bottoms

Reactive dividing-wall column

R-DWC

Dividing wall

PrefractionatorMain column

FIGURE 10.15

Reactive distillation configurations.

Table 10.2 Main Industrial Applications of Reactive Distillation (Kiss, 2013a,b)—cont’d

Reaction Type Catalyst/Internals

Trans-esterification

Ethyl acetate from ethanol and butyl acetate Homogeneous

Diethyl carbonate from ethanol and dimethyl carbonate Heterogeneous

Vinyl acetate from vinyl stearate and acetic acid N/A

Unclassified reactions

Monosilane from trichlorosilane Heterogeneous

Methanol from syngas Cu/Zn/Al2O3 and inert solvent

DEA from monoethanolamine and ethylene oxide N/A

Polyesterification Autocatalytic


process is indeed feasible, a technical evaluation is performed afterward in order to determine the tech-

nical feasibility, the process limitations, working regime and requirements for internals as well as the

models needed for RD. This approach is based on dimensionless numbers such as Damkohler and Hatta

numbers, while taking into account the kinetic, thermodynamic and mass-transfer constrains (Shah

et al., 2012). Note that the Damkohler number is the ratio of characteristics residence time (H0/V)to characteristics reaction time (1/kf), and the Hatta number is the ratio of the maximum possible con-

version in the film to the maximum diffusion transport through the film (Kiss, 2013a,b).

10.5 EXAMPLES10.5.1 TERNARY SEPARATION OF HYDROCARBONSThis simple example illustrates the design and simulation of a DWC used for the ternary separation of a

mixture consisting of pentane (A), hexane (B) and heptane (C), at a flow rate of F¼100 kmol/h with

the composition xA,F¼0.2, xB,F¼0.6 and xC,F¼0.2. The required product purities considered here are

xA,1¼xB,2¼xC,3¼0.99.

Figure 10.16 shows the separation of the ternary mixture by the usual direct sequence. A simple

mass balance gives, for each column, the flow rate and composition of product streams, together with

the recoveries of key components. This allows calculation of the minimum number of trays (Nmin) and

of the minimum reflux ratio (Rmin) by the Underwood–Fenske method, as implemented by the DSTWU

Qc = -0.46 Gcal/h

Qr = 0.49 Gcal/h Qr = 0.86 Gcal/h

R = 2.73 R = 1.06

Qc = -0.86 Gcal/h

ABC

A

C

BC1 2

B

NT = 25

NF = 11

A: 20 kmol/hB: 60 kmol/hC: 20 kmol/h

A: 19.7kmol/hB: 0.2kmol/h

A: 0. 3 kmol/hB: 59.8 kmol/hC: 20 kmol/h

A: 0.3 kmol/hB: 59.6 kmol/hC: 0.3 kmol/h

B: 0.2 kmol/hC: 19.7 kmol/h

NT = 19

NF = 11

FIGURE 10.16

Direct sequence for the separation of pentane–hexane–heptane (results obtained by shortcut method).


shortcut model from Aspen Plus. Heuristics recommend a reflux ratio R¼1.2 Rmin, which roughly cor-

responds to NT¼2 Nmin. The DSTWU model also gives the location of the feed tray (NF) the reboilerduty (Qr) and the condenser duty (Qc). A summary of the main results is also included in Figure 10.16.

The model used to obtain a preliminary design of the DWC consists of three conventional distil-

lation columns, as illustrated by Figure 10.17. The columns correspond to the main sections of a DWC

(pre-fractionator, main-column section above the side-stream, main-column section below the side

stream). In the first step, we aim to find for each section the number of trays, the location of the feed

tray and an estimation of the reflux and vapour flow rates. We assume that the amount of light com-

ponent A found in the side stream FA,S¼0.3 kmol/h is given by two equal contributions: the amount

A found in the bottoms of the pre-fractionator and the amount of A coming from the top section of the

main column, due to finite separation efficiency. A similar argument holds for the amount of C present

in the side stream. These considerations give the split of the key components A and C in the pre-

fractionator. Therefore, we have all the data required to design the pre-fractionator. Using again the

DSTWU shortcut model, the number of trays N¼2 Nmin¼14 the reflux ratio R¼1.2 Rmin¼0.61

and the location of the feed tray NF¼7 are obtained. For these values, it is rather straightforward

to obtain the distribution of the non-key component B between the pre-fractionator distillate and

the bottom streams. Moreover, the values of the liquid reflux and vapour boilup are also found. At this

A

C

B

3

ABC

1

2

B

AB

BC

NT = 14

NF = 7

R = 0.61

NT = 20NF = 10R = 1.48

NT = 24NF = 12R = 1.33


A: 19.7 kmol/hB: 0.2 kmol/h







FIGURE 10.17

Distributed sequence for the separation of pentane–hexane–heptane (results obtained by shortcut method).

42510.5 EXAMPLES

point, we have all the data necessary to calculate the recoveries of the key components for the two

sections of the main column. Therefore, these sections can also be designed. The results obtained after

applying this procedure are summarised in Figure 10.17.

A base case design for the DWC column is simply obtained by combining the two sections of the

main column. We also observe that the number of trays in the pre-fractionator side of the dividing wall

is different from the number of trays on the main-column side of the dividing (14 vs. 22). While such

construction can be achieved in practice, a simpler solution is to have the same number of trays on both

sides. Therefore, the size of the pre-fractionator is increased.

For rigorous simulation of the DWC column using the RADFRACmodel from Aspen Plus, the pre-

fractionator condenser and reboiler are removed—the necessary liquid reflux and vapour boilup being

provided by the side streams L and V of the main column (see Figure 10.18). These streams can be

initialised with values obtained in the previous step. No specifications are needed for the pre-

fractionator. As usual, the main column is specified in terms of reflux ratio, distillate and side streams

(S, L and V). After convergence, three design specifications are added, namely, the purities of the dis-

tillate, side-stream and bottoms products, with reflux ratio, distillate and side-stream flow rates as ma-

nipulated variables.

Finally, an optimisation block is added, where the reboiler duty is the objective function to be mini-

mised, while the vapour and liquid split L and V are the decision variables. The results obtained after






A: 19.99 kmol/hB: 0.2 kmol/h



Qr = 0.98 Gcal/h

Qc = -0.94 Gcal/h

A

C

BABC

PF

MC

10

20

32

1

22

1

44

AB

BC

Liq.

Vap.

FIGURE 10.18

Petlyuk equivalent of a DWC (results obtained by rigorous simulation—the reboiler duty was minimised by using

the liquid and vapour flow rates as decision variables).


optimisation are presented in Figure 10.18. It can be observed that, for the same number of tray, the re-

quired duty was reduced from 1.35 Gcal/h (direct sequence) to 0.98 Gcal/h (DWC). The design can be

further optimised by considering the TAC as an objective function, and adding number of trays and

the location of the feed and side streams as decision variables. This is a mixed integer nonlinear program-

ming problemwhich is somewhat more difficult to solve. Therefore, it is left as an exercise for the reader.

It should be remarked that a column with a side stream is often used to separate a ternary mixture

using only one unit. However, in contrast to the DWC, a side-stream column is able to provide only two

products with high purity (e.g. distillate and bottoms), with the drawbacks of lower purity of the third

product (side stream), lower recoveries of the high-purity products, higher energy requirements and

larger column diameter.

10.5.2 BIOETHANOL DEHYDRATIONThe bioethanol production relies on several processes: corn-to-ethanol, sugarcane-to-ethanol, basic and

integrated lignocellulosic biomass-to-ethanol. The raw materials undergo several pre-treatment steps

and then enter the fermentation stage where bioethanol is produced. A common feature is the produc-

tion of a diluted stream (typically in the range of 5–12%wt ethanol) that needs to be further concen-

trated beyond the azeotropic composition, to over 99–99.8%wt ethanol—as shown in Figure 10.19, left

(Kiss, 2013a,b). Several energy demanding separation steps are required to reach the purity target,

mainly due to the presence of the binary azeotrope ethanol–water (95.63%wt ethanol).

PDC

EDC

PDC-TOP

Ethanol

EDC-BTMFeed

Water

Solvent

SRC

Water

Solvent

PDC – pre-concentration distillation columnEDC – extractive distillation columnSRC – solvent recovery column

(ethylene glycol)

(recycle)

1

30

21

1

17

4

11

1

16

85–12%

92–94%

>99.9%

Sugarcane Corn Lignocellulosicbiomass

Pre-treatment

Saccharification

Fermentation

Distillation(Bioethanol pre-concentration)

5–12 wt% Bioethanol

92–94 wt% Bioethanol

Bioethanol dehydration

Bioethanol >99.8%wt

Water

Water

Pre-treatment

FIGURE 10.19

Block flow diagram of the bioethanol production from various feedstock (left). Bioethanol dehydration by

conventional extractive distillation (right).

42710.5 EXAMPLES

The first step is carried out in a pre-concentration distillation column (PDC) that concentrates

bioethanol from 5–12% up to 92.4–94%wt. The second step is the ethanol dehydration up to concen-

trations above the azeotropic composition. Although extractive distillation presents relatively high en-

ergy costs, it is still the option of choice in case of large scale production of bioethanol fuel.

The feed considered here is the diluted stream (10%wt) obtained by fermentation, at a production

rate of 100 ktpy (12,500 kg/h ethanol). This is distilled to near-azeotropic composition (93.5%wt) and

then dehydrated to a purity of over 99.8%wt ethanol to comply with all standards. Aspen Plus simu-

lations were performed using the rigorous RADFRAC unit for distillation. NRTL property method was

used due to the presence of a non-ideal mixture containing polar components. Figure 10.19 (right)

shows the conventional ED process used for bioethanol dehydration (Kiss and Ignat, 2012). The con-

ventional ED sequence consists of three distillation units: PDC, extractive distillation column (EDC)

and solvent recovery column (SRC)—three column shells, three condensers and three reboilers in total.

The first column (PDC) in the sequence has the function to separate water as bottom stream and a near-

azeotropic composition mixture as distillate—sent afterward to the second column (EDC). In the EDC

unit, ethylene glycol—used as a high boiling solvent—is added on a stage higher than the feed stage of

the ethanol–water mixture. Due to the presence of the EG solvent, the relative volatility of ethanol–

water is changed such that their separation becomes possible. Pure ethanol is collected as top distillate

product of the EDC, while the bottom product contains only solvent and water. The solvent is then

completely recovered in the bottom of the third column (SRC), cooled and then recycled back to

the EDC. An additional water stream is obtained as distillate of the SRC unit. Heat integration may

be also employed in order to recover the heat from the solvent recycle stream.

The numbers on the columns shown in Figure 10.19 (right) indicate the top, bottom and feed stage.

Notably, PDC is the largest column, being also over three times more energy intensive as the rest of the

columns combined. The reason for this unbalance is the extremely large amount of water that is sep-

arated in this unit—due to the high concentrating factor (>9). In terms of temperature, the PDC exhibits

a much lower temperature span as compared to the EDC and SRC units. The composition profiles are

also a confirmation of the previous description of the flowsheet. Table 10.3 presents the design param-

eters of the optimised conventional sequence (Kiss and Ignat, 2012; Kiss, 2013a,b).

Since all the distillation columns of the conventional ED sequence operate at atmospheric pressure,

the use of a DWCwas explored as an attractive alternative. Special attention was paid to combining the

column sections such that there is also a match of temperatures. However, commercial process sim-

ulators do not include particular subroutines for DWC units. The so-called decomposition method sim-

plifies the design problem, as the DWC configuration can be replaced by a sequence of conventional

distillation columns. Therefore, two coupled RADFRAC units were used in Aspen Plus, as the ther-

modynamically equivalent of the E-DWC.

The results of the conventional distillation sequence were used as a starting point for the E-DWC

simulations described hereafter, providing initial estimates for all design variables. Note that both the

conventional and E-DWC alternatives were optimised in terms of minimal energy demand using

the sequential quadratic programming (SQP) method. The approach minimises the total heat duty

of the sequence, constraint by the required purity and recovery of the bioethanol product and solvent,

using several decision variables: total number of stages, feed-stage, side-stream location, partition wall

size and location, solvent flowrate, reflux ratio, liquid and vapour split. The SQP optimisation method

and the effective sensitivity analysis tool from Aspen Plus were used in the E-DWC optimisation pro-

cedure illustrated in Figure 10.20 (left)—extended from the effective design method proposed by


Dejanovic et al. (2010) for standard DWC units. The objective of the optimisation is to minimise the

total reboiler duty required, as follows:

Min Qð Þ¼ f NT,NF,NS,NSS,NDWS,NDWC,SFR,RR,V,rV,rLð ÞSubject to y

!m

� x!

m (10.14)

where NT is the total number of stages, NF is the feed stage, NS is the feed stage location for the solvent,

NSS is the side-draw stage (where applicable), NDWS is the number of dividing-wall stages, NDWC is the

location of the dividing wall, SFR is the solvent-to-feed ratio, RR is the reflux ratio, V is the boilup rate,

rL and rV are the liquid and vapour split, while ym and xm are vectors of the obtained and required pu-

rities for the m products. The design problem is a complex optimisation problem with both continuous

(SFR, RR, V, rV, rL) and discrete (NT, NF, NS, NSS, NDWS, NDWC) decision variables. In order to de-

termine the optimal ratio between the energy cost and the number of stages, an additional objective

function was used, Min NT (RR+1) that approximates very well the minimum of total annualised cost

of a conventional distillation column (Kiss, 2013a,b).

Table 10.3 Design Parameters of an Optimal Conventional Sequence for Bioethanol Separation

(Kiss and Ignat, 2012; Kiss, 2013a,b)

Design Parameters PDC EDC SRC Unit

Total number of stages 30 17 16 –

Feed stage number 21 11 8 –

Feed stage of extractive solvent – 4 – –

Column diameter 3.4 1.5 0.9 m

Operating pressure 1 1 1 bar

Feed composition (mass fraction)

Ethanol:water 0.1:0.9 0.935:0.065 – kg/kg

Water:solvent – – 0.039:0.961 kg/kg

Feed flowrate (mass)

Ethanol 12,500 12,494 1.25 kg/h

Water 112,500 868.5 852.2 kg/h

Solvent 0 20,793 20,784 kg/h

Reflux ratio 2.9 0.17 0.6 kg/kg

Reboiler duty 23,882 5574 1454 kW

Condenser duty �13,626 �3440 �865 kW

Ethanol recovery – 99.94 – %

Water recovery 99.98 – 99.98 %

Solvent (EG) recovery – – 99.91 %

Purity of bioethanol product – 99.80 – %wt

Purity of water by-product 99.99 – 98.6 %wt

Purity of ethylene glycol recycle – – 99.99 %wt

42910.5 EXAMPLES

However, the standard DWC configurations require more energy than the conventional sequence

thus being economically unattractive. This is because a huge amount of water (about 90% of the feed

stream) must be evaporated and removed as side stream or top distillate. The lesson learned is that water

must be produced as bottom product in order to avoid its complete evaporation.

Figure 10.20 (right) shows the conceptual design of the proposed E-DWC that combines three dis-

tillation units into just one column. In this column, the feed side (pre-fractionator) acts as the PDC unit.

Water is removed as liquid side stream, but an additional side reboiler is required in order to return the

required amount of water vapours to the column. The liquid feed stream is fed on top of the pre-

fractionator side, thus serving as a reflux to the PDC section. The vapour leaving the feed side of

the E-DWC has a near azeotropic composition. Solvent is added at the top of the E-DWC, this section

acting as the EDC unit of the conventional sequence. Ethanol is separated here as high-purity top dis-

tillate and removed as main product. The liquid flowing down the top section (EDC) is collected and

distributed only to the (SRC) side opposite to the feed side (pre-fractionator) and further down the bot-

tom of the E-DWC. This complete redistribution of the liquid flow is required to avoid the presence and

loss of solvent on the feed side (PDC section). In the SRC section, the solvent is separated as bottom

product and then recycled in the process.

Initialization

ChangeNT,NF, NS, NSS, NDWS, NDWC, V, RR, SFR

NT,NF, NS, NSS, NDWS, NDWCV, RR, SFR, rL, rV

Optimal designConverged DWC profiles and stage requirements

No

Adjust rL and rV

Min QR

Min N(RR+1)

Yes

No

Yes

Ethanol

(recycle)

Water

Feed

rV

rL

E-DWC

Solvent

1

1

17

18

34

42

4

35

17

(ethylene glycol)

Solvent

EDC

PD

C

SR

CFIGURE 10.20

Procedure for the optimal design of an extractive dividing-wall column (left). Bioethanol dehydration by extractive

distillation: conventional process versus DWC.


Figure 10.21 plots the temperature as well as the vapour and liquid composition profiles in the

E-DWC, while the key parameters of the optimal design are presented in Table 10.4 (Kiss and

Ignat, 2012). Remarkable, the temperature difference between the two sides of the wall is very low

(less than 20 �C)—such conditions being feasible for practical application.

Moreover, high purity and recovery is obtained for all three products of the extractive DWC: eth-

anol as top distillate, water as side product and EG solvent as recovered bottom product. In contrast to

the well-known DWC configuration, the side stream is collected here from the same (feed) side of the

column, not the opposite. Figure 10.21 clearly illustrates the changes in the vapour and liquid compo-

sition along the column, these being in line with the functional task of each part of the column: PDC on

the feed side, EDC in the common top part of the column, and SRC on the bottom common section of

the column. It is also worth noting that the diameter of the E-DWC unit is only slightly lower than the

diameter of the PDC unit of the conventional sequence, although it does require some additional stages.

In practise, this means that the revamping of existing plants is possible by re-using the existing PDC

unit (i.e. add more stages by extending the height of the column or by using a more efficient structured

packing). The single-step E-DWC alternative is the most efficient in terms of energy requirements

allowing energy savings of 17% (specific energy requirements of only 2070 kWh/tonne bioethanol)

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 5 10 15 20 25 30 35 40 45

Mas

s fr

actio

n (-

)

Mas

s fr

actio

n (-

)

Tem

pera

ture

(�C

)

Ethanol

Water

Ethyleneglycol

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1Ethanol

Water

Ethyleneglycol

Liquid composition Vapour composition

40

60

80

100

120

140

160

180

200

220

0 5 10 15 20 25 30 35 40 45

0 5 10 15 20 25 30 35 40 45

Stage (-)

Stage (-)Stage (-)

Pre-fractionator (PF)

Main column

FIGURE 10.21

Temperature and composition profiles in the E-DWC (dotted line means the pre-fractionator or feed side of the

column).

43110.5 EXAMPLES

while also being the least expensive in terms of capital investment and operating costs (about 17%

lower TACs).

10.5.3 DIMETHYL ETHER PRODUCTIONDimethyl ether (DME) is of great interest due to its use as clean fuel for diesel engines or in combustion

cells, as a precursor to organic compounds, as well as a green aerosol propellant. Currently, DME is typ-

ically produced bymethanol dehydration taking place at temperatures of 250–400 �C and pressures up to

20 bars. Different types of solid acid catalysts can be used, but g-alumina is the preferred one due to its

thermal stability, mechanical resistance, high surface area and catalytic properties (Muller and Hubsch,

2005). The current industrial process involves a gas-phase reactor (70–80% conversion of methanol),

followed by a direct sequence of two distillation columns that deliver high-purity DME (over 99.99%wt).

Table 10.4 Design and Process Parameters of an Optimised E-DWC for Single-Step Bioethanol

Separation and Dehydration—100 ktpy Plant (Kiss and Ignat, 2012; Kiss, 2013a,b)

Design Parameters Value Unit

Total number of stages 42 –

Number of stages pre-fractionator side 17 –

Feed stage on pre-fractionator side 1 –

Feed stage of extractive solvent (main-column side) 4 –

Side stream withdrawal stage 17 –

Wall position (from–to stage) 18–34 –

Column diameter 3.35 m

Operating pressure 1 bar

Feed stream flowrate (mass)

Ethanol–water mixture 125,000 kg/h

Solvent 20,793 kg/h

Feed composition (mass fraction)

Ethanol:water 0.1:0.9 kg/kg

Feed stream temperature 30 �CDistillate to feed ratio 0.1 kg/kg

Reflux ratio 3.4 kg/kg

Liquid split ratio (rL) 0:1 kg/kg

Vapour split ratio (rV) 0.4:0.6 kg/kg

Total reboiler duty (side reboiler and bottom reboiler) 25,775 kW

Condenser duty �12,964 kW

Ethanol recovery 99.81 %

Water recovery 99.99 %

Purity of bioethanol product 99.81/99.6 %wt/%mol

Purity of water by-product 99.8/99.9 %wt/%mol

Purity of ethylene glycol recycle 99.99/99.99 %wt/%mol


Figure 10.22 illustrates the simplified conventional flowsheet for methanol dehydration, along with

the ternary diagram of the process (Kiss and Ignat, 2013; Kiss, 2014). The equilibrium limited dehy-

dration reaction of vapourised methanol is carried out in a fixed-bed catalytic reactor. The outlet of the

reactor—consisting of DME, water and unreacted methanol—is cooled and subsequently distilled in

the first tower to yield pure DME. The unreacted methanol is separated from water in a second distil-

lation column and recycled back to the reactor. A major problem of this process is the high investments

costs for several units (e.g. reactors, distillation columns, heat exchangers) that require a large plant

footprint, as well as the associated energy requirements.

One approach to reduce the capital and operating costs is to integrate the two distillation columns

used for the DME purification and methanol recovery into only one DWC—as shown in Figure 10.23

(Kiss and Ignat, 2013). DME and water are separated as top and bottom end high-purity products

(>99.99%wt), while methanol accumulates towards the middle of the column, being withdrawn as

a side stream (>99%wt) and then recycled in the process. The reported process alternative requires

less equipment and 20% lower capital costs (TIC¼$1,412,490), with 28% savings in energy and TACs

(TAC¼$1,138,984 for a 100 ktpy plant), as compared to the conventional distillation sequence (Kiss

and Ignat, 2013).

Water(179.98 �C)

DME(44.40 �C)

Methanol (136.81 �C)

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

0.2

0.4

0.6

0.8Ternary diagram (10 bar)

RX

DC1DC2

Methanol(recycle)

Water

DC2

DME

DC1

Methanol

RX

FIGURE 10.22

DME production: simplified process flowsheet and ternary diagram (at 10 bar).

DME

Water

MethanolFeed

Steam

Methanol(recycle)

Water

Feed

DME

RDC DC

Water

FeedDME

Methanol(recycle)

R-DWC

FIGURE 10.23

Process alternatives based on single-step separation in a DWC, reactive distillation process, and all-in-one

reactive DWC.

43310.5 EXAMPLES

RD could be also used a process alternative, since it can successfully combine the functions of the

reactor and the DME purification column into one unit, as shown in Figure 10.23 (Kiss, 2014). How-

ever, a further improved alternative combining all functions into one unit is based on a reactive DWC

(Figure 10.23). Methanol is fed on top of the reactive zone where the solid catalyst is located, while

DME is produced as top distillate, water as bottom product, and the unreacted methanol as side-stream

product that is recycled.

Aspen Plus simulations embedding experimental results were performed using the rigorous RAD-

FRAC distillation unit. UNIQUAC-Redlich–Kwong was selected as the most adequate property

method in Aspen Plus, and the binary interaction parameters were validated against reported experi-

mental data (Kiss and Suszwalak, 2012a,b). Although there are no azeotropes present in this system, the

ternary diagram shows a small liquid split envelope hence the (reactive) distillation columns has to be

modelled using VLLE data.

The dehydration of methanol is an equilibrium limited reaction leading to DME and water. As ver-

ified experimentally, no side reactions occur at the operated conditions (Lei et al., 2011):

2CH3OH$CH3OCH3 +H2O (10.15)

The model of the catalytic distillation column includes also the experimentally determined intrinsic

kinetic model parameters previously reported by Lei et al., 2011, for the methanol dehydration over

an ion-exchange resin. The reaction takes place only in the liquid phase. Eley–Rideal and the equiv-

alent power–law models are both suitable for simulation purposes (Lei et al., 2011). The reaction rate,

determined for the temperature range of 391–423 K, is given by:

r¼ kWcat MeOH½ �m H2O½ �n (10.16)

k¼A exp �Ea= RTð Þð Þ (10.17)

where Wcat is the weight amount of catalyst (e.g. 15 kg of solid catalyst per stage), A is the Arrhenius

factor (A¼5.19�109 m3/kg-cat/s), Ea is the activation energy (133.8 kJ/mol), and m and n are the or-ders of reaction with respect to methanol and water (m¼1.51 and n¼–0.51).

The process was optimised in terms of minimal energy requirements, using the SQP method imple-

mented in Aspen Plus. The purity target was selected to be over 99.99%wt for both DME and water, but

no hard constraint was set on the purity of the unreacted methanol, as this stream is being recycled in the

process. The optimisation problem for theminimisation of the R-DWC reboiler heat duty is defined as:

Min Qð Þ¼ f NT,NF,NR,NRZ,NDWS,NDWC,NSS,RR,V,FSS,rV,rLð ÞSubject to y

!m � x

!m

(10.18)

where NT is the total number of stages, NF is the feed stage, NR is the number of reactive stages, NRZ is

the location of the reactive zone, NDWS is the number dividing-wall stages, NDWC is the location of the

dividing wall, NSS is the stage of the side-stream withdrawal, RR is the reflux ratio, V is the boilup rate,

FSS is the flowrate of the side-stream product, rL and rV are the liquid and vapour split, while ym and xmare vectors of the obtained and required purities for the m products. The design problem is a complex

optimisation problem with both discrete (NT, NF, NR, NRZ, NDWS, NDWC, NSS) and continuous (RR, V,FSS, rV, rL) decision variables. An additional objective function was used, Min NT (RR+1), to approx-imate the minimum of total annualised cost of distillation columns.


Figure 10.24 plots the temperature and liquid composition profiles in the R-DWC, while the key

parameters of the optimal R-DWC design are presented in Table 10.5 (Kiss and Suszwalak, 2012a,b).

Remarkable, the temperature difference between the two sides of the wall is very low—less than

15 �C—such conditions being feasible for practical applications, with little heat transfer expected

and negligible effect on the column performance. The reactive DWC unit has 35 stages in total, with

the reactive zone located on stages 8–31 on the feed side, and a common stripping section (stages

32–35) and a common rectifying zone (stages 1–7).

The methanol stream is fed on stage 8, at the top of the reactive zone—the feed side of the DWC

acting as the RD zone where the solid acid catalyst is present. High purity (99.99%wt) DME is

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Stage (-)

Stage (-)

Mol

ar fr

actio

n (-

)T

empe

ratu

re (

�C)

Side product sectionRD side of DWC

DME Methanol Water

40

60

80

100

120

140

160

180

200

0 5 10 15 20 25 30 35

0 5 10 15 20 25 30 35

Temperature PF side

Temperature R-DWC

FIGURE 10.24

Temperature and composition profiles along the reactive DWC (dashed line used for the side-product section,

while continuous line used for the main DWC section).

43510.5 EXAMPLES

delivered as distillate, while similar high-purity water is obtained as bottom product. The unreacted

methanol is collected as side product, and then recycled back to the process. The results reported show

that the R-DWC process has superior performances as compared to the conventional or RD process:

high energy savings of up to 60%, as well as around 30% lower capital investment costs (Kiss and

Suszwalak, 2012a,b). The methanol conversion in R-DWC is 50%, and the energy requirements are

only 560 kWh/t DME (Kiss, 2013a,b).

10.5.4 FATTY ESTERS SYNTHESISFatty esters are manufactured by the esterification of fatty acids with various alcohols, as requested by

applications. Typically, they result from the series C8 (caprylic), C10 (capric), C12 (lauric) and C14

(myristic) acids with low molecular alcohols, such as methanol, ethanol, n-propanol, i-propanol andn-butanol. Traditionally, these fatty esters are produced in batch processes employing homogeneous

catalysis, namely p-toluensulfonic acid. In order to shift the chemical equilibrium to completion an

excess of alcohol is employed. Water removal and alcohol recovery requires additional distillation

equipment. The post-treatment implies catalyst neutralisation and washing that generates waste. For

these reasons a technology based on continuous operation and solid catalyst is highly desirable. This

case study presents how to apply catalytic reactive distillation (CRD) to the synthesis of fatty acid esters

as a general multi-product continuous process. As representative species we consider the lauric (dode-canoic) acid and alcohols in the series C1–C8. The reversible chemical reaction consists of the ester-

ification of the fatty acid with alcohol in the presence of an acid catalyst, as follows:

Table 10.5 Design Parameters of the Reactive DWC for DME Synthesis (Kiss, 2013a,b)

Design Parameters Value Unit

Flowrate of feed stream 9 kmol/h

Temperature of feed stream 25 �CPressure of feed stream 10 bar

Number of stages 35 –

Stages reactive zone 8–31 –

Feed stage 8 –

Wall position (from/to stage) 8–31 –

Distillate to feed ratio 0.25 kmol/kmol

Reflux ratio 2.45 kmol/kmol

Liquid split ratio 0.65 kmol/kmol

Vapour split ratio 0.82 kmol/kmol

Operating pressure 10 bar

DME product purity >99.99/99.99 %wt/%mol

DME mass flowrate 103.653 kg/h

Methanol conversion 50.40 %

Water purity (bottom product) >99.99/99.99 %wt/%mol

Reboiler duty 58.70 kW

Condenser duty –37.13 kW


R1�COOH+R2�OH$R1�COO�R2 +H2O (10.19)

As undesired secondary reaction one may note the alcohol etherification:

R2�OH+HO�R2 $R2�O�R2 +H2O (10.20)

There is an active research in the field of superacid solid catalysts. Organic ion-exchange resins, such as

Amberlyst-15 and silica-supported Nafion SAC-13, are currently used for the esterification of low mo-

lecular acids, in general, below 120–130 �C. At higher temperatures the integrity of the resin-type cat-

alyst can be affected by swelling, as well as by thermal degradation. Preserving the colour of the

product is of greatest importance for cosmetic applications. For this reason, the use of inorganic super-

acid catalysts was investigated, such as sulphated zirconia, silica-supported heteropoly acids, zeolites,

etc. (Kiss et al., 2006). If the catalyst activity is water sensitive, the RD device should be designed such

to ensure operating conditions for water-free organic liquid.

We start by presenting comparatively the process design for the esterification of the fatty acid with

the highest and lowest boiler alcohols, 2-ethylhexanol (2-EtH) and methanol, respectively, where the insitu water removal is easier to handle. Then we consider the specific aspects raised by the intermediate

boiling alcohols forming azeotropes with water that can be broken by employing an entrainer. More

information can be found in the case study book of Dimian and Bildea (2008).

10.5.4.1 Project definitionThe objective is the design of a multi-product process for the manufacturing of fatty acid esters of high

purity by CRD. Lauric acid (LA) is selected as representative fatty acid, while the alcohols are 2-EtH,

methanol, n- and i-propanol. The reference production rate is 10 kmol/h, which in the case of 2-EtH

corresponds to 25 ktpy ester. All raw materials are of high purity. Because the product is aimed to cos-

metic applications, purity of more than 99.95% is required, which implies the same fatty acid conver-

sion. The design should ensure no water-phase formation inside the column, while the water

concentration in the organic phase should be limited below 1%mol.

10.5.4.2 Thermodynamic dataTable 10.6 gives the normal boiling points (nbp) of the key components, such as alcohols and esters.

Note that methyl laurate is about 33 �C lighter than that the respective acid, while the other esters are

heavier. Plotting the vapour pressure would show that over the range of nbp, from 65 to 300 �C, thevapour pressure of components spans four orders of magnitude, from 10�2 to 100 bars. Accordingly,

mastering the pressure in a multi-product operation mode is a challenging issue.

Due to the presence of water, the reactive mixtures exhibit a strong non-ideal character. Except

methanol, the other alcohols form azeotropes with water, as shown in Table 10.7. Both 2-EtH and

Table 10.6 Normal Boiling Points of Key Components for the Esterification of LA with Alcohols

Component MeOH NPA IPA 2-EtH LA

Methyl

Laurate

N-PropylLaurate

2-EtH

Laurate

Boiling

point (�C)64.7 97.2 82.5 184.6 298 267 302 334

43710.5 EXAMPLES

LA form heterogeneous azeotropes with water. Note that the solubility of LA and 2-EtH in water is

large, while the reciprocal solubility is extremely low. This behaviour is very beneficial for water sep-

aration. Clearly, the presence of azeotropes complicates the separation.

10.5.4.3 Reaction kineticsTwo contributions in enhancing the reaction rate can be distinguished: catalyst and auto-catalysis ef-

fects. The following kinetic expression based on activities can be formulated:

racid ¼ k�Aaacid + k�CCcat

� �aacidaalcohol�awateraester

Ka

(10.21)

whereKa is the equilibrium constant. Since concentrations are used in the simulation software, we consider

a simplifiedmodel as racid¼k1CalcCacid–k2CesterCwater. The kinetic constants depend on the type of alcohol.

Theoretically, the differences can be explained by the polar and steric effects of substituents, known in the

organic chemistry, and described by the Taft equation. Grecea et al. (2012) found that these effects apply

qualitatively for the esterification of fatty acid catalysed by a modified sulfated zirconia catalyst. Over the

series C1–C8 of normal alcohols the reaction rate decreases slightly with the chain length. It is important to

note that the reaction rate is considerably slower for iso-alcohols compared with normal alcohols. Thus, for

isopropanol the reaction rate drops by a factor of five when compared with n-propanol. In this example, we

take as reference the esterification of LA with 2-EtH (Omota et al., 2003). On the working domain the

experimental data are described by racid¼54.233 exp(�6691.84/T)� (CalcCacid�CesterCwater/4.672), with

the reaction rate in kmol/kg catalyst/s. For other substrates the reaction constant can be multiplied by 2 for

methanol, by 1.3 for n-propanol, and 0.25 for isopropanol.

The kinetics for the esterification of myristic acid with propanols using p-toluene sulfonic acid cat-alyst was reported by de Jong et al. (2009). For n-propanol (1) and isopropanol (2) the reaction rates arerespectively:

r1 ¼ 6:76�104 cat½ �exp �47,000= RTð Þð Þ A½ � B½ ��84:4 cat½ �exp �25,400= RTð Þð Þ E½ � W½ � (10.22)

r2 ¼ 3:35�105 cat½ �exp �58,900= RTð Þð Þ A½ � B½ ��2:18�103 cat½ �exp �45,900= RTð Þð Þ E½ � W½ � (10.23)

In the above relations the reaction rate is given in kmol/(m3 s), the activation energy in kJ/mol, [E], [A],

[B], [W] and [cat] are the concentrations of ester, alcohol, acid, water and catalyst in mol/l. The recom-

mended catalyst concentration is in the range of 0.15–0.2 mol/l.

Table 10.7 Azeotropic Data by the Esterification of Lauric Acid at Normal Pressure

Azeotrope

Water (1)/2-

EtH (2)

Heterogenous

Water (1)/Lauric

Acid (2)

Heterogenous

N-Propanol (1)/Water (2)

Homogeneous

Isopropanol (1)/

Water (2)

Homogeneous

T (�C) 99.1 99.9 87.7 80.2

Composition yaz,1¼0.968

x1 (w)¼0.9996

x1 (o)¼0.2411

yaz,1¼0.9999

x1 (w)¼0.9999

x1 (o)¼0.2502

yaz,1¼0.4330 yaz,1¼0.6810

Notes: w, water phase; o, organic phase.


10.5.4.4 Selectivity issuesSince the acid-to-alcohol ratio inside the RD column varies over three orders of magnitude, the super-

acidic catalyst might promote side reactions, the most probable being the formation of ethers. Exper-

imentally no by-products were detected at long contact time and higher temperatures (Omota et al.,

2003). However, in this project a limiting temperature of 200 �C is assigned for the temperature of

the bottom product.

10.5.4.5 Catalyst effectivenessTypically, the size of particles in CRD is between 0.85 and 1 mm. Methods for estimating the catalyst

effectiveness are available for simple irreversible reactions. With some approximations they can be

applied in the case of more complex reactions. In this case, the calculation shows that for particles

below 1 mm the diffusion resistance is negligible (Dimian and Bildea, 2008).

10.5.4.6 Chemical and phase equilibriumInside the reactive zone, C&PE occur simultaneously. Figure 10.25 presents the RCMs of simultaneous

phase and chemical equilibrium at normal pressure for the esterification of LA with 2-EtH and meth-

anol. Note that special coordinates are used, such as X1 (acid+water) and X2 (acid+ester), for repre-

senting all four component mixtures in a bi-dimensional diagram (Doherty and Malone, 2001).

In the first case, the esterification with 2-EtH (Figure 10.25, left), the RCM diagram shows the seg-

regation in two liquid phases, organic and aqueous, separated by a boundary connecting the two azeo-

tropes. There is also a third homogeneous region (only water phase) in the right corner, not visualised

because of the scale. The trajectories converged from the azeotropes of 2-EtH to the 2-EtH laurate,

which is the highest boiler. The heterogeneous region can be avoided by operating in the reaction zone

at the temperature above 100 �C. The RCM suggests that quantitative separation of water in top is pos-

sible, after condensation and decanting, since the solubility of 2-EtH in water is very small. The alcohol

is returned as reflux.

In the second case, esterification with methanol (Figure 10.25, right), the reactants are nodes, while

the products are saddles. Again, there is a heterogeneous region, but no azeotrope methanol/water. The

trajectories emerge from the methanol to LA, passing along the ester saddle. If the acid is completely

consumed by reaction, the top distillate will contain both water and methanol, while the fatty ester will

FIGURE 10.25

Reactive residue curve maps for the esterification of the lauric acid with 2-ethylhexanol (left side) and with

methanol (right side) (Omota et al., 2003).

43910.5 EXAMPLES

go to the bottom product. Thus, the RCM analysis indicated that esterification with methanol in a single

distillation column at infinite reflux is not possible. But if the acid is consumed completely by reaction,

the process can be carried out in a sequence of two columns, a first one for reaction, and a second one

for water removal and methanol recycling.

However, a finer analysis reveals other possibilities. The unfeasibility at infinite reflux is not nec-

essarily true at finite reflux. Omota et al. (2003) solved this issue by means of a tray-by-tray compu-

tation. If the boilup ratio is not too high, the column can produce in bottom a fatty ester of high purity

and deliver in top a vapour mixture of acid and water, from which water can be separated quantitatively

after condensation and decantation. Thus, the column behaves more as reactive absorber than RD. This

analysis was confirmed by the study of Steinigeweg and Gmehling (2003) regarding the esterification

of decanoic acid with methanol catalysed by Amberlyst-15 at 120 �C and 3 bar: getting higher conver-

sion needs a very small reflux. The conceptual design of a reactive absorber for the esterification of

fatty acids with methanol has been later studied by Kiss (2009) and with control implementation by

Bildea and Kiss (2011). The problem of recovery a useful slight methanol excess can be solved ele-

gantly by employing a DWC set-up, as reported by Kiss et al. (2012).

10.5.4.7 Case 1: Esterification with 2-ethylhexanolEquilibrium-based design: In a first approach, we consider an equilibrium based design. The RCM

analysis leads to the conceptual flowsheet presented in Figure 10.26. The acid and alcohol enter

counter-currently at the top and at the bottom of the reaction zone, respectively. The condition of bot-

tom temperature of 200�C can be realised working under vacuum at 32 kPa and diluting the product

with 12%mol alcohol. The bottom product goes to evaporator, from which the ester product is obtained

while the alcohol is recycled to CRD column. The top vapour is condensed and separated in two phases

after decantation: the water phase leaves the decanter as by-product since low alcohol solubility, while

the alcohol-rich phase is sent back as reflux to the column. Note that Figure 10.26 contains a make-up

device, which will enter later in discussion. The feasibility of above set-up can be evaluated by sim-

ulation. The RD column is modelled as reboiled stripper followed by top vapour condenser and three-

phase flash, with organic phase refluxed to column. The result is that only 3–5 reactive equilibrium

stages are necessary to achieve over 99% conversion. The stripping zone may be limited at two to three

stages, while the rectification zone to one to two stages.

Kinetic design: Next, we proceed with the simulation based on the assumption of kinetic controlled

chemical process, but instantaneous vapour–quid equilibrium. Because some sizing elements are

needed in simulation, namely, the holdup, the preliminary hydraulic design is always necessary.

Hydraulic design: For the esterification with 2-EtH the hydraulic design can be outlined as follows.

The ester production is 10 kmol/h (3120 kg/h). Assuming a reflux ratio of 0.5 leads to a mean liquid

flow of 3120�1.5¼4680 kg/h. Considering liquid density of 850 kg/m3 gives volumetric flow of

5.5 m3/h. Assuming linear liquid velocity of 10 m/h gives a column cross area of 0.55 m2 and diameter

of 0.837 m. We select Multipak-I, for which the catalyst fraction is c¼0.33. We assume NTSM¼3.

The density of the catalyst particles is about 1000 kg/m3. With a void fraction ECB of 0.3 the bulk cat-

alyst density is 1000� (1�0.3)¼700 kg/m3. The catalyst holdup is 0.55�1�0.33�1000/3¼60.5 l/

stage or 60.5�0.7¼42.3 kg catalyst/stage.

This preliminary design can be checked after simulation. For the middle of the column (stage 12), one

gets a liquid mass flow rate of 3700 kg/h at a reflux ratio of 0.274. At a temperature of 172 �C, the liquiddensity is 870 kg/m3 and liquid viscosity 1.03 cP. Using these data and catalyst characteristics in the


Eq. (10.12) leads to a velocity of the load point ofULP¼7.3 m/h. On the other hand, in the middle of the

column the fictive velocity isU¼3700/870/0.55¼7.75 m/h, just slightly above the load point. This result

shows that the liquid phase is employed efficiently. However, the examination of the liquid flow profiles

shows that the catalyst bag at the three or four top stages might be only partially wetted. The solution is

diminishing the catalyst load, or increasing slightly the reflux. Note that the computation of the liquid

holdup leads to hl¼0.176, in satisfactory agreement with the literature data (Hoffman et al., 2004), which

indicate a value of about 0.21. In the abovemean conditions the gas load factorF has the value of 0.81, for

which the pressure drop is less than 2 mbar/m with a total column pressure drop below 0.02 bar.

Simulation: In order to get robust convergence the RD column is simulated as a stripper. The top

vapour is condensed and separated in a two-phase decanter. Water is taken-off as aqueous phase, while

the organic phase is returned. The RD column has 24 stages, from which 20 reactive. LA is introduced

on the top reactive stage as liquid at 160 �C, while the 2-EtH enters as vapour at 0.5 bar at the bottom.

The RD column is operated under vacuum at 32 kPa.

Typical profiles are shown in Figure 10.27. The acid concentration in the liquid phase falls rapidly

over the first five stages, where most of the reaction takes place. The alcohol concentration is less

RDC

Reboiler

D-1

Purified ester

E-1

E-2

Sep1

Sep2

E-3

S-1

Reflux

Waterphase

CondenserVent

Fatty acid

Alcohol

Reactionzone

V-4

Alcohol recycle

Make-up heavy alcohol/entrainer

FIGURE 10.26

Conceptual flowsheet for the synthesis of fatty esters.

44110.5 EXAMPLES

sensitive, remaining almost constant. The water concentration in the liquid phase is maximum 3%mole

fraction on the top stage, but only 1000 ppm on the first reactive stages. The ester formation takes place

mainly on the first half of the reactive zone, the second being necessary to push the conversion over

99.9%. The ester production is at maximum over the first two stages. The temperature profile shows a

sharp increase to a plateau at about 170 �C.Optimisation: Sensitivity analysis indicates that for ensuring high productivity and purity the most

influential factors are catalyst holdup and reflux ratio. The catalyst distribution on stages may be also

seen as an optimisation variable, but the effect is rather small. The real advantage comes from a tech-

nological reason. The reaction rate is the highest on the first top stages, and therefore the catalyst could

deactivate here more rapidly. Placing less active catalyst gives a more uniform reaction rate. The re-

duction in productivity can be compensated for by only few reactive stages. Table 10.8 presents the

results of an optimised design, with the distribution of the catalyst activity. The purity specification

below 500 ppm is met, with a productivity of about 5 kg ester/kg catalyst. The pressure has a strong

effect. Increasing it from 0.32 to 0.5 bar will raise the reaction temperature by about 10 �C, sufficientfor doubling the reaction rate. Since the reaction is slightly exothermal, the reboiler duty is only needed

to compensate the differences in the enthalpy of components. Consequently, the reboiler duty is small

and has no effect on optimisation.

Detailed design: Hydraulic design should ensure high efficiency of the catalyst with respect to

the liquid and gas flows. A potential drawback would be the maldistribution. Actually, the liquid

should flow slowly but uniformly through the catalyst bed at a linear velocity of few mm/s. The

placement of special redistribution devices is necessary. Beside industry-proven methods, innova-

tive solutions can be found by taking advantage from the mixing and dispersion properties of the

packing itself. An example is ‘partially flooded beds’ designed by Montz and BASF (Olujic et al.,

2003) in which standard packing is combined with specially designed elements to promote bub-

bling, similar to that of a tray. In this way longer residence time can be ensured with minimum

backmixing.

0.0

0.2

0.4

0.6

0.8

1.0

0 4 8 12 16 20

Stages

Mol

ar fr

actio

n liq

uid

0

20

40

60

80

100

120

140

160

180

200

Tem

pear

ture

(�C

)

Lauric acid2-EthylhexanolWaterEsterT

FIGURE 10.27

Profiles in the catalytic RD column for fatty esters synthesis by kinetic modelling.


10.5.4.8 Case 2: Esterification with methanolAn original method that makes possible the synthesis of methyl fatty esters in a single RD column has

been proposed by Dimian et al. (2009). The in situ removal of water is solved by employing the dualesterification with methanol and a heavy alcohol immiscible with water, namely, 2-EtH, which acts

simultaneously as reactant and entrainer. The flowsheet depicted in Figure 10.26 remains valid, but

with the difference that the light alcohol is fed at the bottom side of the reactive zone, while the heavy

alcohol is added as make-up in the decanter. Figure 10.28 presents profiles of concentrations and tem-

perature. It can be observed that the heavy alcohol acts indeed as an entrainer, helping the water re-

moval in top, and as reactant over the top stages. The methyl ester reaction is pushed to the middle of

the column. Thus, with respect to the previous case the number of reactive stages is preserved too,

because of the higher reaction rate with methanol. As before, the separation zones are rather limited

at few stages at the top and bottom. The reaction temperature depends on the catalyst. In this example, a

lower temperature profile was chosen, at about 130 �C, for which the use of a resin ion-exchange cat-

alyst is possible. The operation takes place at nearly atmospheric pressure (1.5 bar), in contrast with

Table 10.8 Parameters of the Optimised Design

Parameter (Unit) Value

Number of theoretical stages 24

Reactive stages 22

NSTM 3

Diameter (m) 0.837

Catalyst holdup (kg/stage) 10 (2–5), 20 (6–11), 30 (11–14), 42 (15–23)

Total mass of catalyst (kg) 606

Lauric acid feed (kmol/h; �C) 10; 160

Alcohol feed (kmol/h; �C) 14; 160

Top vapour flow (kmol/h) 15

Reflux ratio 0.248

Temperature profile (�C) 113.5 (top) 151.8 (acid) 180 (maxi) 176.5 (alcohol) 182 (bottom)

Liquid flowa (kmol/h; kg/h; m3/h) 14.22; 3868.4; 4.2

Gas flowa (kmol/h; kg/h; m3/h) 14.08; 1866.2; 1392.4

Densities liquid/gas (kg/m3) 874.4/1.34

Liquid load (m3/m2/h) 7.62

Gas load (m3/m2/s) 0.7

Ester production (kmol/h) 9.996

Productivity (kg ester/kg catalyst/h) 5.15

Reboiler duty (kW) 35.6

aReference to the middle stage of the reactive section.

44310.5 EXAMPLES

6–12 bar with methanol alone, and 0.3 bar with 2-EtH. The two fatty esters can be obtained in the de-

sired ratio by adjusting the feeds. In a preferable operation mode the ratio of fresh feed reactants is acid:

methanol: 2-EtH 1:0.8:0.2. The above solution makes sense from economic viewpoint, because the

methyl ester is by far the most demanded, while 2-EtH is a cheap alcohol, in general a waste. The pro-

cess control implementation has been studied (Dimian et al., 2009). It has been demonstrated that the

concentration of methanol in top can bemaintained at minimum, such that there is no need of additional

column for methanol recovery.

10.5.4.9 Case 3: Esterification with propanolsWhen using propanols, the azeotrope formation makes necessary the recovery and recycling of alcohol.

An industrial process conducted in this way employs homogeneous catalyst (para-toluene sulphonicacid), which is lost after reaction (Bock et al., 1997). A more efficient solution can found by employing

an entrainer for breaking the azeotrope water/alcohol and superacid solid catalyst (Dimian et al., 2004;

de Jong et al., 2010). Besides, the entrainer has an enhancing effect on the reaction rate, by increasing

the amount of alcohol recycled to the reaction space. The selection of entrainer should fulfil some rules.

Suitable entrainers are ethers, esters, and hydrocarbons, as n-propyl acetate, di-propyl-ether and cyclo-hexane. A minimum amount of entrainer is necessary, corresponding to the azeotrope composition.

Hence, the entrainer enhances the water removal, ensuring simultaneously a larger internal recycle

of alcohol to the reaction zone. The comparison with a process without entrainer operating as

pseudo-absorber shows that the catalyst loading can be reduced up to 50%. The above approach

was confirmed by a study on design and control of a process for high-purity isopropyl palmitate by

RD using cyclohexane as entrainer, with substantial energy reduction (Wang and Wong, 2006).

The key result of the cited references is that process flowsheet presented in Figure 10.26 is preserved,

with the only observation that the top section should have sufficient number of stages, however limited,

to ensure efficient water removal.

At this point, it is useful to point-out the difference between the esterification with normal- and

isopropanol. The key issue is that in the second case the reaction rate is significantly lower, up to five

times, even with homogeneous catalysis. Consequently, a much more active heterogeneous catalyst

FIGURE 10.28

Column profiles for dual esterification of LA with methanol and 2-ethylhexanol (Dimian et al., 2009).


should be employed (Grecea et al., 2012), or at the limit a homogeneous superacid catalyst, but with

supplementary costs for catalyst and ester purification. Another alternative would be the conversion of

methyl esters by trans-esterification with iso-alcohols in the same CRD set-up.

Concluding remarks: The case study demonstrates that the manufacturing of fatty esters can be

done continuously in a multi-product catalytic distillation column, as displayed in Figure 10.26.

The design goal is achieving high purity, over 99.95%, and efficient water removal. The CRD column

is assembled from three sections: top rectification, reactive zone and bottom stripping, employing

structured packing. The reactants circulate in counter-current; the fatty acid is introduced at the top

and the alcohol at the bottom of the reactive zone, respectively. The simplest process operates with

heavy alcohol, since the water removal can be done by L-L decanting. In the case of methanol and

of light alcohols forming azeotropes with water, physical or reactive entrainers can be employed. Since

the solid catalyst exhibit similar activity for the series of normal alcohols C1–C8, the reactive zone is

practically the same. Employing iso-alcohols requires a muchmore active catalyst. Varying the column

pressure allows to adapt the temperature profile to reaction and separation requirements. The key issue

in practical implementation remains the availability of a solid superacid catalyst, which should be ro-

bust at long-time operation against leaching and swelling. Alternatively, a liquid catalyst can be used,

but with the drawback of additional equipment and operation costs.

10.6 SUMMARYSignificant cost reductions and high energy efficiency can be achieved by employing various ap-

proaches based on PI principles, for example, maximised effectiveness, driving forces and synergy.

There are several PI technologies that became success stories at industrial scale: SMs, CHEs, HiGee,

DWC and reactive separations.

Although many commercial applications of HiGee are known in absorption, stripping and reactive

precipitation only few commercial applications in distillation were reported so far. A key reason is that

several problems such as the dynamic seal, middle feed, liquid distributor and the multi-rotor config-

uration were not well addressed. In order to successfully solve these problems, a novel kind of HiGee

device was recently proposed and developed—the so-called RZB that contains a unique rotor. Remark-

able, the RZB fills the gap in HiGee distillation and it has the potential for a bright future in this di-

rection of PI (Wang et al., 2011).

CyDist uses a periodic operation mode that can bring new life in old distillation columns, providing

key benefits, such as: increased column throughput, low energy requirements and high separation per-

formance. Moreover, the column has more DoF that contribute to a good process control. The main

obstacle for the widespread implementation of CyDist is the periodic operation that requires special

training and extra safety measurements.

DWC is one of the best examples of proven PI technology in distillation, as it allows significantly

lower investment and operating costs (typically 25–40%)while also reducing the equipment and carbon

footprint.Considering the number andvariety of industrial applications,DWCcanalready be considered

as a success story about PI in distillation and it will certainly develop into a standard type of equipment in

the nearby future. Many applications are known today, mainly concerning separations of ternary mix-

tures. The development efforts focus nowadays towards the separation ofmore than three components or

applications of extractive, azeotropic and RD in a DWC. Nonetheless, DWC has some limitations as

well: operation at a single pressure, larger size as compared to any single column of a conventional

44510.6 SUMMARY

(e.g. direct or indirect) separation sequence, and larger temperature span as the energy required has to be

supplied and rejected at the highest (reboiler) and the lowest (condenser) temperature levels.

RD is nowadays an established unit operation in chemical process technology, being also the front-

runner in the field of PI. At present, there are a variety of models now available in the literature for

screening, analysis, design and optimisation of RD processes: residue curve maps (RCM) are

invaluable for initial screening and flowsheet development, EQ models have their place for initial

designs while NEQ models are used for the final design, development of control strategies, and

commercial RD plant design and simulation. RD brings key benefits to equilibrium limited chemical

systems, resulting in lower investment and operating costs, as well as reduced plant footprint. The

industrial applications of RD are flourishing as the scientific community and the technology providers

removed the main implementation barriers, developed heuristic process synthesis rules and expert

software to identify the techno-economical feasibility of RD (Shah et al., 2012; Kiss, 2013a,b).

REFERENCESAsprion, N., Kaibel, G., 2010. Dividing wall columns: fundamentals and recent advances. Chem. Eng. Process.

49, 139–146.

Bildea, C.S., Kiss, A.A., 2011. Dynamics and control of a biodiesel process by reactive absorption. Chem. Eng.

Res. Des. 89, 187–196.

Bock, H., Wozny, H., Gutsche, B., 1997. Design and control of a reaction distillation column including recovery

system. Chem. Eng. Process. 36, 101–109.

Boodhoo, K., Harvey, A. (Eds.), 2013. Process Intensification Technologies for Green Chemistry: Innovative

Engineering Solutions for Sustainable Chemical Processing. John Wiley & Sons, Chicester, UK.

de Jong, M.C., Feijt, R., Zondervan, E., Nijhuis, T.A., de Haan, A.B., 2009. Reaction kinetics of the esterification

of myristic acid with iso-propanol and n-propanol using p-toluene sulphonic acid as catalyst. Appl. Catal. Gen.

365, 141–147.

de Jong, M.C., Zondervan, E., Dimian, A.C., de Haan, A., 2010. Entrainer selection for the synthesis of fatty acid

esters by entrainer-based reactive distillation. Chem. Eng. Res. Des. 88, 34–44.

Dejanovic, I., Matijasevic, L., Olujic, Z., 2010. Dividing wall column—a breakthrough towards sustainable dis-

tilling. Chem. Eng. Process. 49, 559–580.

Dimian, A.C., Bildea, C.S., 2008. Chemical Process Design—Computer-Aided Case Studies. Wiley-VCH,Weinheim.

Dimian, A.C., Omota, F., Bliek, A., 2004. Entrainer-enhanced reactive distillation. Chem. Eng. Process.

43, 411–420.

Dimian, A.C., Bildea, C.S., Omota, F., Kiss, A.A., 2009. Innovative process for fatty acid methyl esters by dual

reactive distillation. Comput. Chem. Eng. 33, 743–750.

Doherty, M.F., Malone, M.F., 2001. Conceptual Design of Distillation Systems. McGraw-Hill, New York.

Edward, P.L., 2004. Handbook of Industrial Mixing-Science and Practice. JohnWiley & Sons, Hoboken, NJ, USA.

Ehrfeld, W., Hessel, V., Lowe, H., 2000. Microreactors: New Technology for Modern Chemistry. Wiley-VCH,

Weinheim, Germany.

Freund, H., Sundmacher, K., 2008. Towards a methodology for the systematic analysis and design of efficient

chemical processes—part 1: from unit operations to elementary process functions. Chem. Eng. Process.

47, 2051–2060.

Ghadrdan, M., Halvorsen, I.J., Skogestadad, S., 2011. Optimal operation of Kaibel distillation columns. Chem.

Eng. Res. Des. 89, 1382–1391.

Goetze, L., Bailer, O., Moritz, C., von Scala, C., 2001. Reactive distillation with Katapak. Catal. Today 69, 201–208.

Grecea, M.L., Dimian, A.C., Tanase, S., Rothenberg, G., 2012. Sulfated zirconia as a robust superacid catalyst for

multiproduct fatty acid. Catal. Sci. Technol. 2, 1500–1506.

Green, D.W., Perry, R.H., 2008. Perry’s Chemical Engineers’ Handbook, 8th ed. McGraw-Hill.


Halvorsen, I.J., Skogestad, S., 2003. Minimum energy consumption in multicomponent distillation. 1. Vmin dia-

gram for a two-product column. Ind. Eng. Chem. Res. 42, 596–604.

Halvorsen, I.J., Skogestad, S., 2011. Energy efficient distillation. J. Nat. Gas Sci. Eng. 3, 571–580.

Harmsen, G.J., 2007. Reactive distillation: the front-runner of industrial process intensification: a full review of

commercial applications, research, scale-up, design and operation. Chem. Eng. Process. 46, 774–780.

Harmsen, G.J., 2010. Process intensification in the petrochemicals industry: drivers and hurdles for commercial

implementation. Chem. Eng. Process. 49, 70–73.

Hesselgreaves, J.E., 2001. Compact Heat Exchangers: Selection, Design and Operation. Pergamon, Elsevier

Science, Oxford, UK.

Hoffman, A., Noeres, C., Gorak, A., 2004. Scale-up of reactive distillation columns with catalytic packings. Chem.

Eng. Process. 43, 383–395.

Kaibel, G., 1987. Distillation columns with vertical partitions. Chem. Eng. Technol. 10, 92–98.

Kays, W.M., London, A.L., 1998. Compact Heat Exchangers, third ed. Krieger Pub Co., Malabar, FL, USA.

Kiss, A.A., 2009. Novel process for biodiesel by reactive absorption. Sep. Purif. Technol. 69, 280–287.

Kiss, A.A., 2013a. Advanced Distillation Technologies—Design, Control and Applications. John Wiley & Sons,

Chicester, UK.

Kiss, A.A., 2013b. Novel applications of dividing-wall column technology to biofuel production processes.

J. Chem. Technol. Biotechnol. 88, 1387–1404.

Kiss, A.A., 2014. Distillation technology—still young and full of breakthrough opportunities. J. Chem. Technol.

Biotechnol. 89, 479–498.

Kiss, A.A., Bildea, C.S., 2011. A control perspective on process intensification in dividing-wall columns. Chem.

Eng. Process. 50, 281–292.

Kiss, A.A., Ignat, R.M., 2012. Innovative single step bioethanol dehydration in an extractive dividing-wall column.

Sep. Purif. Technol. 98, 290–297.

Kiss, A.A., Ignat, R.M., 2013. Revamping dimethyl ether separation to a single-step process. Chem. Eng. Technol.

36, 1261–1267.

Kiss, A.A., Suszwalak, D.J.-P.C., 2012a. Enhanced bioethanol dehydration by extractive and azeotropic distilla-

tion in dividing-wall columns. Sep. Purif. Technol. 86, 70–78.

Kiss, A.A., Suszwalak, D.J.-P.C., 2012b. Innovative dimethyl ether synthesis in a reactive dividing-wall column.

Comput. Chem. Eng. 38, 74–81.

Kiss, A.A., Dimian, A.C., Rothenberg, G., 2006. Solid acid catalysts for biodiesel production—towards sustain-

able energy. Adv. Synth. Catal. 348, 75–81.

Kiss, A.A., Segovia-Hernandez, J.G., Bildea, C.S., Miranda-Galindo, E.Y., Hernandez, S., 2012. Reactive DWC

leading the way to FAME and fortune. Fuel 95, 352–359.

Krishna, R., 2002. Reactive separations: more ways to skin a cat. Chem. Eng. Sci. 57, 1491–1504.

Lei, Z., Zou, Z., Dai, C., Li, Q., Chen, B., 2011. Synthesis of dimethyl ether (DME) by catalytic distillation. Chem.

Eng. Sci. 66, 3195–3203.

Luyben, W.L., Yu, C.C., 2008. Reactive Distillation Design and Control. John Wiley & Sons, Hoboken, NJ, USA.

Maleta, V.N., Kiss, A.A., Taran, V.M., Maleta, B.V., 2011. Understanding process intensification in cyclic dis-

tillation systems. Chem. Eng. Process. 50, 655–664.

Muller, M., Hubsch, U., 2005. Dimethyl ether. In: Ullmann’s Encyclopedia of Industrial Chemistry, seventh ed.

Wiley-VCH, Weinheim.

Noeres, C., Kenig, E.Y., Gorak, A., 2003. Modelling of reactive separation processes: reactive absorption and

reactive distillation. Chem. Eng. Process. 42, 157–178.

Olujic, Z., Kaibel, B., Jansen, H., Rietfort, T., Zich, E., Frey, G., 2003. Distillation column internals/configurations

for process intensification. Chem. Biochem. Eng. Quart. 17, 301–309.

Omota, F., Dimian, A.C., Bliek, A., 2003. Fatty acid esterification by reactive distillation. Part 1: equilibrium-based de-

sign. & Part 2—kinetics-based design for sulphated zirconia catalysts. Chem. Eng. Sci. 58, 3159–3174, 3175–3185.

Patrut, C., Bildea, C.S., Lita, I., Kiss, A.A., 2014a. Cyclic distillation – design, control and applications. Sep. Purif.

Tech. 125, 326–336.

447REFERENCES

Patrut, C., Bildea, C.S., Kiss, A.A., 2014b. Catalytic cyclic distillation – a novel process intensification approach in

reactive separations. Chem. Eng. Process. Process Intensification 81, 1–12.

Petlyuk, F.B., 2004. Distillation Theory and Its Application to Optimal Design of Separation Units. Cambridge

University Press, New York, USA.

Ramshaw, C., 1983. HiGee distillation—an example of process intensification. Chem. Eng. Lond. 389, 13–14.

Rao, D.P., Bhowal, A., Goswami, P.S., 2004. Process intensification in rotating packed beds (HiGee): an appraisal.

Ind. Eng. Chem. Res. 43, 1150–1162.

Reay, D., Ramshaw, C., Harvey, A., 2013. Process Intensification—Engineering for Efficiency, Sustainability and

Flexibility, second ed. Butterworth-Heinemann, UK.

Reddy, K.J., Gupta, A., Rao, D.P., 2006. Process intensification in a HIGEE with split packing. Ind. Eng. Chem.

Res. 45, 4270–4277.

Reschetilowski, W. (Ed.), 2013. Microreactors in Preparative Chemistry: Practical Aspects in Bioprocessing,

Nanotechnology, Catalysis and More. Wiley-VCH, Weinheim, Germany.

Sanders, J.P.M., Clark, J.H., Harmsen, G.J., Heeres, H.J., Heijnen, J.J., Kersten, S.R.A., van Swaaij, W.P.M.,

Moulijn, J.A., 2012. Process intensification in the future production of base chemicals from biomass. Chem.

Eng. Process. 51, 117–136.

Shah, R.K., Kraus, A.D., Metzger, D., 1990. Compact Heat Exchangers. Taylor & Francis, USA.

Shah, M., Kiss, A.A., Zondervan, E., de Haan, A.B., 2012. A systematic framework for the feasibility and technical

evaluation of reactive distillation processes. Chem. Eng. Process. 60, 55–64.

Stankiewicz, A., 2003. Reactive separations for process intensification: an industrial perspective. Chem. Eng. Pro-

cess. 42, 137–144.

Stankiewicz, A., Moulijn, J.A., 2004. Re-engineering the Chemical Processing Plant: Process Intensification.

Marcel Dekker, New York.

Steinigeweg, S., Gmehling, J., 2003. Esterification of a fatty acid by reactive distillation. Ind. Eng. Chem. Res.

42, 3612–3619.

Stichlmair, J.G., Fair, J.R., 1998. Distillation, Principles and Practice. Wiley-VCH, New York.

Sundmacher, K., Kienle, A. (Eds.), 2003.ReactiveDistillation: Status and FutureDirections.Wiley-VCH,Weinheim.

Sundmacher, K., Kienle, A., Seidel-Morgenstern, A. (Eds.), 2005. Integrated Chemical Processes: Synthesis, Op-

eration, Analysis, and Control. Wiley-VCH, Weinheim.

Taylor, R., Krishna, R., 2000. Modelling reactive distillation. Chem. Eng. Sci. 55, 5183–5229.

Thakur, R.K., Vial, Ch., Nigam, K.D.P., Nauman, E.B., Djelveh, G., 2003. Static mixers in the process industries—

a review. Chem. Eng. Res. Des. 81, 787–826.

Thiel, C., Sundmacher, K., Hoffmann, U., 1997. Residue curve maps for heterogeneously catalysed reactive dis-

tillation of fuel ethers MTBE and TAME. Chem. Eng. Sci. 52, 993–1005.

TU Dormund 2013. www.fvt.bci.tu-dortmund.de/cms/en/research/research_topics/Process_intensification_by_

unit_operation_design/index.html (last accessed 20.10.13.).

van Gerven, T., Stankiewicz, A., 2009. Structure, energy, synergy, time—the fundamentals of process intensifi-

cation. Ind. Eng. Chem. Res. 48, 2465–2474.

Wang, S.-J., Wong, S.H., 2006. Design and control of entrainer-aided reactive distillation for fatty ester produc-

tion. Ind. Eng. Chem. Res. 45, 9041–9049.

Wang, G.Q., Xu, Z.C., Ji, J.B., 2011. Progress on Higee distillation—introduction to a new device and its industrial

applications. Chem. Eng. Res. Des. 89, 1434–1442.

Wirth, T. (Ed.), 2013. Microreactors in Organic Chemistry and Catalysis. Wiley-VCH, Weinheim, Germany.

Yildirim, O., Kiss, A.A., Kenig, E.Y., 2011. Dividing wall columns in chemical process industry: a review on

current activities. Sep. Purif. Technol. 80, 403–417.


Documents

[Computer Aided Chemical Engineering] Integrated Design and Simulation of Chemical Processes Volume 35 || Process Intensification