Content uploaded by Thomas Grützner
Author content
All content in this area was uploaded by Thomas Grützner on Nov 27, 2019
Content may be subject to copyright.
Reducing Time to Market by Innovative
Development and Production Strategies
During the past years cost pressure on the European chemical companies grew
dramatically by increased competition from low-labor-cost BRIC countries on the
one side and the United States, favored by decreased energy costs due to shale gas
exploitation, on the other side. Raising costs for energy consumption and wages
come on top. Therefore, highly efficient procedures in both development and pro-
duction with the aim to decrease costs became necessary. A combination of tech-
nical solutions like new production concepts, e.g., hybrid processes, and organiza-
tional solutions like standardization and improved utilization of equipment are
needed. Some of the applied techniques are described in detail and their potential
in terms of time and cost reduction is demonstrated.
Keywords: 50 % idea, Process development, Smart production concepts, Time to market
Received: February 25, 2016; revised: April 18, 2016; accepted: July 20, 2016
DOI: 10.1002/ceat.201600113
1 Introduction
Increasing market competition, raising energy and raw material
costs over the past decades as well as the trend towards chemi-
cal products and processes with a higher complexity and rais-
ing customer demands are huge challenges for the chemical
industry. However, they have been the major drivers to steadily
search for potential improvements in the whole chain from
process development to the final production process. Although
these constraints affect chemical companies world-wide, the
impact on European companies is unlikely larger due to their
location in high-labor-cost countries. Environmental burdens
and faster product changes in the market have additional nega-
tive impact in this context.
Lonza dramatically encounters the global cost pressure by
competitors from BRIC countries and the United States since
the largest production and R&D facilities of the company are
located in Switzerland. The pressure is worsened by a strong
currency compared to those of the main markets that makes
the exports of goods even less beneficial. Therefore, the com-
pany is strongly interested in improving the complete chain
from route scouting over process development and engineering
until setting up and operation of the production plant. These
measures are absolutely crucial for surviving global competi-
tion and remaining a profitable company.
It is a well-received opinion that a package of different mea-
sures is required in order to maintain productivity and compet-
itiveness of the European chemical industry. These measures
include a wide range of approaches. Since this paper does not
aim to give a complete review of all the ideas in this context,
the mentioned points are more to show the variety of the strat-
egies both, industry and academia, are currently working on.
Technical solutions include new production concepts such as
the replacement of batch by continuous unit operations, micro-
technologies in reaction and purification, intensified unit oper-
ations (e.g., reactive distillation), dividing-wall distillation col-
umns [1–4], hybrid processes (e.g., coupling of membranes and
distillation [3, 4]), and flexible and slim production concepts
like standardized skid-mounted small-scale production plants
[5–7]. The technical aspects also cover all activities in the field
of improved modeling and simulation, especially predictive
models for thermodynamic behavior [8–10] and improved
optimization routines [11, 12] as well as automated laboratory
equipment and the application of process analytical technolo-
gies (PAT) [13].
Organizational solutions aim for slim and accelerated devel-
opment and innovation procedures for chemical processes.
Standardization and improved utilization of equipment belong
in this category as well as streamlined innovation and project
management structures.
All different approaches support the strategy of chemical
companies to reduce time to market, to enhance energy and
resource efficiency and establish flexible and slim productions.
In this context, the 50 % idea highlighted the time-to-market
reduction and gives evidence on the importance of accelerated
development cycles [14, 15].
The F
3
Factory (fast, flexible, future) consortium, a collabo-
ration of 25 partners from both academia and industry funded
by the European Commission’s 7th Framework Programme,
intensively investigated the impact of modular and standard-
ized container plants on efficiency, flexibility, and reduction of
capital expenditures (CAPEX) in chemical production. Seven
Chem. Eng. Technol. 2016,39, No. 10, 1835–1844 ª2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.cet-journal.com
Thomas Gru¨ tzner
1
Christian Schnider
1
Daniel Zollinger
2
Bernhard C. Seyfang
2
Niklaus Ku¨ nzle
1
1
Lonza AG Visp, R&T Center of
Excellence – Technology, Visp,
Switzerland.
2
Lonza AG Visp, R&T Front End
Development, Visp,
Switzerland.
–
Correspondence: Dr.-Ing. Thomas Gru
¨tzner (thomas.gruetzner@
lonza.com), Lonza AG Visp, R&T Center of Excellence – Technology,
Rottenstrasse 6, 3930 Visp, Switzerland.
Research Article 1835
industrial case studies have been carried out and proofed the
concept [7]. After expiring of the F3 project in 2013, the
INVITE research center carried on with the work in the field of
small-scale, standardized container plants [16]. Some compa-
nies established their own technology platforms for small-scale
production, e.g., Evonik’s Ecotrainer. Evonik uses this technol-
ogy to accelerate market entry and reduce market risks in the
time between product development and start of the large-scale
production [17]. The plants can be developed and built inde-
pendently from the future production site, which is another
benefit.
Lier et al. recently published a good overview highlighting
the pros and cons of versatile production concepts [18]. Even
though a number of industrial applications of modular plants
are published [19], the concept is not widely spread for chemi-
cal productions yet and can still be considered as niche applica-
tion. The reasons for this retention are not widely discussed in
the literature yet. The reluctance of chemical companies to
adopt new technologies might be an explanation to a certain
extent. Problems also occur when separation operations like
distillation are involved. In many cases, the necessary equip-
ment height does not fit to the size of standardized containers
and alternatives are often not available [18].
Different technical guidelines and standards are another
issue. Pressure tanks fulfilling, for instance, Swiss standards are
not automatically approved for operation in the UK. Recently,
the ENPRO initiative, which consists of leading German com-
panies and universities and is funded by the German Federal
Ministry for Economic Affairs and Energy, has been formed to
extend the modularization idea and link it to the research fields
of energy-efficient process technology, smart-scale production,
and big data [20]. The mentioned initiatives bear witness to the
importance of improvements in the field of more efficient pro-
duction concepts showing a high degree of flexibility.
Hence, from Lonza’s perspective, there is not the one
approach that fits all demands. Many different techniques have
to be applied in order to reach the goal of accelerated market
entry. Modular concepts are in this context only one piece of
the whole picture and the beneficial effects of flexible and mod-
ular production units can be almost neglected if the preceding
development process is not fast and flexible at the same time.
Lonza is a company that has world-scale as well as multipur-
pose plants in its portfolio and produces molecules from gram
to multi-thousand ton scale. As a consequence, Lonza had to
establish different solutions for the manifold demands resulting
from that diversity and covering the whole cycle from first lab
trials for concept proof over process development to buildup
and operate the production facility.
In the following section, examples are given that exemplarily
show Lonza’s effort to reduce time to market. Some of them are
already successfully implemented, others are work in progress.
Using the tools described below, time to market can be reduced
significantly. For custom-manufacturing projects, where com-
panies like Lonza produce a given molecule for their customer,
the time span for process development, tech transfer, and start-
up of the production ranges between six and twelve months.
The installation of complex separation equipment, e.g., a divid-
ing-wall distillation column, takes less than twelve months
from the first idea, equipment design, ordering of the equip-
ment, and commissioning until start-up. Compared to peers,
these times are considerably shorter and cannot be realized
without mind changes in the design and production process
including the implementation of new tools on all stages of the
development.
2 The Project Team
The efficient and fast processing of development projects is a
crucial issue. A broad range of mistakes can be made in this
field resulting in increasing development periods and cost,
respectively, or leading to a complete failure in the worst case.
It is a triviality that the project team should be capable to carry
out the required assignments throughout the life cycle of the
project. Therefore, the members of the project team must have
the technical, scientific, and organizational skills which are
required for the specific project.
Best practice at Lonza for development and research projects
are teams that consist of, at least, one organic chemist, one
engineer, one analytical chemist, one technical/scientific project
leader, the lab staff as well as an marketing and sales represen-
tative. The project leader has always a scientific background.
Typically, all members of the team, besides marketing and sales,
belong to the R&D department. Regarding the demands of the
project, the composition of the teams might be adjusted, e.g., a
biologist and a biotechnology specialist in the case of a fermen-
tation project are incorporated. From the very beginning, a
representative from operation is assigned to ensure that the
findings of the project team are always aligned with future pro-
duction capabilities. The project team persists over the whole
life cycle of the project, starting from early phase lab trials until
start-up of the production facility. The project life cycle is di-
vided into different sections. After the first evaluation of the
idea, the phase of project development starts followed by the
technology transfer phase which includes the engineering as
well. After each phase, the project is reviewed within a stage
gate process wherein a go/no-go decision is made (also indi-
cated in Fig. 3).
The described structure of the project team ensures that the
process know-how remains with the same people and no fric-
tion, caused by interfaces, cuts the knowledge and ruins valu-
able time resources. As the project moves onwards, additional
departments are involved in the project for a certain period of
time, e.g., specialists for safety, health, and environment, plant
engineering, legal, logistics, corrosion, packaging, or instru-
mentation. If more fundamental or long-term research
becomes necessary, the tool of academic collaborations is often
used. The described setup is illustrated in Fig. 1 and has proven
its capabilities to streamline development for many years.
Additionally, the process development work is accelerated by
the stringent application of a parallel engineering approach.
The chemical route and the downstream purification process
are developed at the same time, covering all experimental, sim-
ulation, and optimization steps as indicated in Fig. 2. The joint
structure of the process team, persisting of chemists and engi-
neers, facilitates this procedure. The results of the development
are in many cases directly transferred to the existing multipur-
pose production plants, skipping the piloting step. This
www.cet-journal.com ª2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eng. Technol. 2016,39, No. 10, 1835–1844
1836 Research Article
approach takes into account that the processes transferred to
the multipurpose production facilities are not fully optimized.
The optimization work on production scale is typically carried
out by the same project team during the first one or two pro-
duction campaigns.
Lonza gathered experience with this approach for one decade
now and managed to reduce the time and costs for process
development by more than 30 %.
3 Sharing-Best-Practice Approach
Maintaining technical and scientific expertise in the R&D
departments of chemical companies is a very important issue.
Otherwise, the demanded renewal of the product portfolio
and the production processes by the R&D departments can-
not be afforded. Therefore, companies have to find ways to
steadily ensure that their scientists are on the state-of-the-art
level or, in best case, above. Lonza is aiming to reach this
goal using a sharing-best-practice (SBP) approach. This is a
scientific parallel structure to the hierarchic line system
whose groups are linked to the product life cycles. As a
result, the scientists of similar professions are rather spread
over the whole R&D organization than gathered in one
organizational unit within the R&D department. To over-
come the problem of ‘‘dilution’’ of knowledge, each member
of the R&D department is assigned to one or two of the
existing SBP groups, according to the individual background
and regardless the location in the hier-
archy. The existing SBP groups are
Organic Chemistry, Analytical Chemistry,
Particle Technology, Thermal Process
Engineering, Reaction Technology, Process
Analytical Technology (PAT) and Chemo-
metrics, Project Management, Biotechnol-
ogy, and Laboratory Services.
Each of the groups is headed by a reputa-
ble scientist who is, within the scientific
hierarchy, on the same level as a group
leader in the line hierarchy and forms a
platform for internal and external experi-
ence exchange. The groups maintain exter-
nal networks to specialists in other compa-
nies and academia, and take care of the
scientific qualification and training of the
members. Moreover, they coordinate
research projects with universities and trig-
ger basic research projects. For that pur-
pose, a funding, independently from the
business units, is absolutely mandatory in
order to make long-term basic research
possible. The SBP groups are not directly
engaged in the project team described in
the previous section, but their members
are. In case of technical problems within a
project, the SBP groups are the main con-
tacts to start a joined trouble shooting and,
therefore, give indirect support to the
project teams. Generally, the structure of
SBP facilitates the intergroup know-how exchange and unifies
the different development processes.
4 Gain More Insight from Lab
Experiments
Modeling and simulation are two valuable tools to limit the
number of experiments that have to be performed in the labo-
ratory during process development. They not only allow
replacing experiments in certain cases, but they also facilitate
the selection of experiments. Solvent candidates can for exam-
ple be categorized using a model-based approach as it is
described in [21]. In this manner, work performed in the labo-
ratory and calculations that are done in-silico are considered to
be complementary. In the early phase of a project, when only a
few experiments have to provide the decision base for a go/no
go decision, such a combination of approaches is even more
important than in a later phase of a project.
In the early phase, one or more of the following goals should
be met by every experiment:
– confirm or correct the global expectation values from model-
ing and simulation
– generate a point of anchorage for ranking orders, e.g., of sol-
ubility or process concepts, obtained from modeling and
simulation
– obtain a first insight into the kinetic behavior of the reaction
that is investigated
Chem. Eng. Technol. 2016,39, No. 10, 1835–1844 ª2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.cet-journal.com
Figure 1. Project team setup as practiced at Lonza. While the core team remains over
the whole project, specialists from other departments or universities join the team as
necessary.
Figure 2. Parallel engineering approach within a project team.
Research Article 1837
– understand the thermal behavior of the reaction and get first
calorimetric information
– generate samples in order to set up the analytical method.
The goals cited above can be attained in various ways.
Experiments should be run in automated equipment with data
registration whenever possible. That way, experiments can also
be reviewed in a later phase when new insight is available. In
order to expand the spectrum of data that is registered, a con-
sideration of PAT can prove beneficial. Once such tools are rou-
tinely used, the benefits rapidly outplay the costs, even in an
early phase of the project. Laboratory experiments can be
understood more rapidly and optimized using online analytics
since less delays and costs occur compared to cases in which
offline analytics is used. Furthermore, a future large-scale pro-
duction can benefit from such a method, minimizing start-up
times and increasing throughput capacity. The fact that the
challenge of big data still needs efforts to result in maximum
benefit of the data thus acquired should not prevent the collec-
tion and proper storage of the data [22].
Thermal insight into reactions can also answer multiple
needs. On differential scanning calorimetry (DSC) scale, this
can lead to a first understanding of kinetics [23]. Even though
the information obtained from a heat flow detector is only a
global one, it will already lead to a considerably deeper under-
standing than a simple test tube trial. In addition, it is princi-
pally possible to ramp the temperature up to higher domains
after the reaction in order to learn about decomposition of the
samples. The resulting information can then be further used.
One option is to treat the resulting data with advanced thermal
analysis software. In this manner, the data from several experi-
ments can be combined and used to perform either safety
evaluations [24], process intensification, or optimization.
Treating the data in this way allows increasing the predictive
domain. Besides AKTS-Thermokinetics and AKTS-Thermal
Safety Software, AKTS [25] also offers a module that allows the
use of discontinuous data points. These can be used, e.g., for
shelf-life predictions, i.e., the period of save use of a product if
stored within defined temperature limits, based on accelerated
aging studies [26].
Based on the collected information, the decision between
process options in an early project phase is sound since the
information allows, e.g., to compare throughput comparison of
different concepts based on overall kinetics that has been deter-
mined using heat flow measurements.
Once the decisions have been taken in the early phase and
the project has been promoted to the next stage, there is again
a considerable need for experiments. Here, the combination of
goals and approaches can also prove beneficial. Examples are:
– automation for a series of experiments such as design of
experiments (DoE)
– adaptation of the size of DoE experiments in order to gener-
ate material for work-up development
– combination of sample productions with tests for additional
PAT techniques, e.g., those who could not be applied in the
early phase due to size/volume restrictions, or with corrosion
testing.
In order to clarify the approach, the following should be
mentioned: the broadest possible understanding with a limited
number of experiments in the early phase is not intended to
eliminate the need for experimental series such as a DoE. DoE
is very important to efficiently prepare the safety and also the
quality risk analysis for a new process. But most series of DoE
should take place once the process is known well enough to
limit the execution of larger experimental designs to one single
process concept.
5 Batch to Continuous and 24/7-Concept
In the past years, a lot of effort was made for batch processes to
be transformed into continuous processes [5, 27, 28], not only
on large production scale, where in bulk chemical industry this
approach has been common since long, but also on lab scale
both for more basic chemicals as for quite complex reaction
systems known from pharmaceutical and peptide business
[29, 30]. Although some challenges come with continuous reac-
tion mode, the benefit it generates already on lab scale is mani-
fold. On the one hand, continuous processes demand specific
hardware, e.g., pumps, reactors, and various sensors, on the
other hand, a process control system is needed in order to con-
nect, operate, and control the whole system. Once the initial
familiarization phase is successfully passed, the system can be
programmed to run various conditions on its own. In combi-
nation with process analytical technologies or an auto-sam-
pling unit, the presence of manpower is not mandatory any-
more. Whole experimental sets or large amounts of sample
preparation are possible overnight, during the weekend, or
even during vacation, provided that feed and receiving tanks
are large enough to cover the respective time period.
Basically, the working philosophy in the laboratory is shifted:
preparation of the experiment and evaluation of the results of
the prior experiment are performed during regular working
time. Running the experiment is then performed in the absence
of lab technicians or chemists, while the appropriate automa-
tion techniques control the process. Human resources can
therefore be utilized for more important and demanding tasks
whereas the automation system carries out the basic control of
the experiment. This strongly enhances the efficiency without
compromising jobs, which can be a real alternative to achieve
some savings and preserve jobs in economically difficult times.
The strategy of development described above can be
extended even further. Batch processes, which are meant to stay
batch processes on a production scale, could be investigated at
the lab scale in a continuous manner. Kinetics stay fully equiva-
lent if a plug-flow reactor is applied, so all the results received
from the lab on a continuous basis can be used for the scale-up
in the batch reactor. Therefore, even here the increase of effi-
ciency can take place. The highest level of efficiency will be
reached, when an algorithm determines the next best set of
conditions and then iteratively finds the optimum of the reac-
tion on its own [31, 32].
The switch to continuous mode can only be performed for a
minority of reaction systems, mainly due to missing equipment
for continuous handling of solids on a lab scale. Not only for
solids but generally the availability of reliable hardware for var-
ious conditions is not always given, e.g., high-pressure units,
extreme corrosion behavior, and multiphase reaction systems.
Nevertheless, it always makes sense to consider the switch to
www.cet-journal.com ª2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eng. Technol. 2016,39, No. 10, 1835–1844
1838 Research Article
continuous mode since the benefit
during development work and on
final production scale can be sub-
stantial as the following example of
a currently running process at Lon-
za clearly shows.
For an important new product
of Lonza a production process had
to be developed. Within the first
chemical reaction of the process, chlorosulfonic acid (CSOS)
reacts with chlorosulfonylisocyanate (CSI) forming bis(chloro-
sulfonyl)imide (ClSI) and carbon dioxide. The reaction is illus-
trated in Fig. 3. Besides the extreme corrosivity of the reaction
mixture, the decomposition onset temperature for ClSI is
180 °C followed by a pressure increase rate of > 50 bar min
–1
.
Initially, a semi-batch process with dosage and reaction times
of up to 12 h at 150 °C was developed. To produce the predicted
demand, the final production concept would include a 6-m
3
stirred vessel, which displays a significant hazardous potential
since the reaction temperature is near the onset of the ther-
mally unstable product. Extensive and costly security measures
were defined in order to conduct the reaction in a safe manner
on that large scale.
Before progressing into the realization phase, a feasibility
study in the lab was conducted. This showed that the process
could be performed continuously by increasing the tempera-
ture above 200 °C, which is beyond the onset temperature of
the product, and pressure of 80 bar. Due to the precise heat
control in the chosen plug-flow reactor and the almost instant
cooldown at the end of the reactor tube it is possible to reach
the desired product quality without having significant decom-
position despite the elevated temperatures. Therefore, the
desired reaction is finalized before the decomposition occurs.
The setup is depicted in Fig. 4. Because of the drastic intensifi-
cation of this process the reaction is finished within less than
10 min and the large-scale reaction volume could be reduced
down to 15 L. Security is increased whereas at the same time
the footprint of the facility is minimized, reducing both invest-
ment and operating costs dramatically.
6 Relying on Modeling and Simulation
Since many years, modeling and simulation are an integral part
within the process development work in the chemical and
pharmaceutical industry, covering reaction technology, separa-
tion units as well as entire processes. The design of single units
and the arrangement of the units within the process are carried
out computer-assisted. Sensitivity studies and process optimi-
zation are performed and at the end an optimal setup is
received. However, in many cases, the next step is the expensive
and time-consuming scale-up using mini or pilot plants. This
approach is certainly demanded for complex processes con-
taining a number of internal recycle streams including the risk
that traces of unwished and unknown side components of the
reaction accumulate over long times, finally resulting in the
collapse of the process. Also, for some complex unit operations
which cannot easily be simulated, e.g., reactive distillations,
piloting should be considered [33].
Lonza tries to avoid piloting in as many cases as possible,
even for complex unit operations such as extractive distillation
in a dividing-wall column (see example below). The entire unit
design is carried out by simulation studies and only a few key
lab trials are conducted in order to eliminate killer criteria
which cannot be simulated, such as solid precipitation, foam-
ing, or product decomposition [2]. The basis for the simulation
studies are reliable thermodynamic models that accurately
describe the physicochemical behavior of the involved compo-
nents in mixture. Unfortunately, very often the necessary data
is not available due to the fact that the molecules have never
been described before.
Chem. Eng. Technol. 2016,39, No. 10, 1835–1844 ª2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.cet-journal.com
Figure 3. Synthesis of ClSI.
Figure 4. Experimental setup of a continuously operated flow reactor.
Research Article 1839
Instead of determination of equilibrium data, predictive ther-
modynamic modeling is applied for the prediction of the data.
Many different models are available in commercially available
software packages, e.g., DDBST and ProPRED, to predict the
pure component data, usually by group contribution methods.
Phase equilibria are then calculated by predictive equations of
state (PSRK, VTPR) or by predictive g
E
-models (UNIFAC-
family, COSMO-RS). Kontogeorgis [34] gives a comprehensive
overview on the benefits and shortcommings of the abovemen-
tioned models. In the case of distillation, the model is checked
by a simple laboratory batch experiment. If the agreement
between model prediction and experimental result is suffi-
ciently good, the model is used for simulation. If not, the model
is adapted or, if it cannot be avoided, crucial data is determined
by experiments.
For optimizations of existing production processes a compa-
rable approach is used. Thermodynamic models are developed
to describe the behavior of a lab experiment or of existing
plants. In both cases, predictive models are used in cases where
no data for components are available. The models are then
compared to the measured data and, if necessary, adjusted. If
the correlation is satisfactory, optimizations are carried out
based on the simulations and are afterwards directly trans-
ferred to the plant. Using the example of distillation optimiza-
tion, this procedure is described in detail in [35].
The following example illustrates the procedure described
above. 4-Picoline, the desired product, and 2,6-lutidine have to
be separated but form an extremely narrow boiling system. The
boiling point at 0.1 bar for both components is 76 °C. There-
fore, simple distillation cannot be applied. The basic data of the
main components can be taken from Tab. 1.
It has to be mentioned that besides 4-picoline and 2,6-luti-
dine other compounds are present in the mixture (Fig. 5).
Extractive distillation carried out in a dividing-wall column
seemed to be a viable approach to separate the two components
and a comprehensive entrainer screening, based on COSMO-
RS, indicated that ethylene glycol is a promising candidate.
Ethylene glycol interacts stronger with 4-picoline (hydrogen
bond of the hydroxy group) compared to 2,6-lutidine (more
steric hindrance) and reduces the relative volatility of
4-picoline. Additionally, 2,5-lutidine should be separated from
the product. The boiling difference between the product and
2,5-lutidine is sufficient to draw the latter as a side product.
After the entrainer had been defined, the thermodynamic
model had to be established. As can be seen from Fig. 5, differ-
ent sources and predictive models are used to describe the
binary subsystems. Only in a few cases, experiments are con-
ducted to measure the vapor liquid equilibrium (VLE). Mostly,
UNIFAC was used. If reliable data was available from databanks
or literature, these were employed. Experience shows that the
accuracy of these predictive approaches is absolutely adequate
for an overwhelming number of systems and applications.
Due to financial and timing reasons no piloting of the pro-
cess was carried out. Instead, the necessary validation of the
thermodynamic model was conducted with a single lab-batch
distillation. In the lab experiment, the different characteristic
sections of the column (marked in different gray shades in
Fig. 6) were imitated in the batch distillation over the course of
time by altering the batch operation conditions. The experi-
mental approach is visualized in Fig. 6.
The left side of the dividing wall, which separates A from the
entrainer (D) and B, was approximated in the batch column by
charging the feed to the bottom vessel and feeding the entrainer
constantly to the top of the column (step 1). The distillate flow
showed high purities of component A and, thus, the principle
separation was proven. The bottom section of the extractive
dividing wall should separate the pure entrainer in the bottom
of the section and a mixture of B and C overhead. This was
approximated in step 2 by simply using the residue of step 1
(B, C, entrainer, no A). In the right section of the column com-
ponent B should be separated from C (step 3). Therefore, the
distillate of step 2 was used (only B and C) and distilled. Over-
head, pure B was received. Thus, batch distillation experi-
ments proved the principle of extractive distillation. The pre-
viously developed thermodynamic model was applied to
describe the three steps of this batch distillation. All trends
and compositions were in very good accordance with the ex-
perimental results. Therefore, the model was successfully vali-
dated and further piloting was disclaimed. The design of the
production-scale column (diameter, number of stages, heat
www.cet-journal.com ª2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eng. Technol. 2016,39, No. 10, 1835–1844
Table 1. Main components in the extractive distillation seperation task.
Component Boiling point at 0.1 bar [°C] Feed content [wt %] Structure
2,6-Lutidine (A) 76 3
4-Picoline (B, product) 76 86
2,5-Lutidine (C) 85.5 5
Ethylene glycol (D, entrainer) 140 –
1840 Research Article
duty, etc.) was based only on simulation studies with the
validated model.
The final setup of the extractive dividing-wall column is dis-
played in Fig. 7 and compared to the common setup in a two-
column train. The column is operated for two years now and
the plant data is in very good accordance to the simulations.
The time from idea to operation was below one year – a time-
span that would have never been realized if piloting had been
carried out. Beside the extractive dividing-wall column, other
complex columns, e.g., multipurpose dividing-wall columns,
have been designed in the same way, without piloting. All of
these columns perform in large-scale operation as predicted
[2].
7 Smart Multipurpose Production
Facilities
Multipurpose plants are applied for the production of different
products with various chemical processes, but using the same
Chem. Eng. Technol. 2016,39, No. 10, 1835–1844 ª2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.cet-journal.com
Figure 5. VLE matrix used to model the extractive distillation.
Figure 6. Experimental ap-
proach used for model vali-
dation.
Research Article 1841
fixed installed equipment, e.g., reactors, distillations, and
extractions. Campaign-wise production with durations of
2–6 months leading to 10–700 t of product is typical. In order
to be able to carry out different processes with the same units,
the plants and the operators have to comprise a high degree of
flexibility. Therefore, the installed units are rather general than
specific and only in rare cases optimal for one specific produc-
tion campaign.
Another issue are the multiple product changeovers
throughout the course of a year, accompanied in most cases
with process changes making large adaptions in the plant nec-
essary. This cleaning and changeover time has to be as short as
possible in order to maximize the plant utilization. In many
cases it is necessary to change the order of the used unit opera-
tions. That means, e.g., that tanks which are connected to a distil-
lation column in one process are part of an extraction column or
used as simple storage devices in another process. Obviously,
fixed piping between the equipment contradicts this require-
ment. Lonza overcame this problem by the application of cou-
pling stations. Fixed pipes from the entire plant equipment meet
at the coupling stations and can be interconnected by flexible
hoses according to the demand of the process (Fig. 8). As a result,
the order of equipment can be adapted in a very fast way [36].
The concept of flexible coupling is not really new, e.g., some
examples are given in [37]. However, it seems that this concept
is much more consequently applied at Lonza compared to peer
companies. As a result, typical changeover times between two
production campaigns including adaption, cleaning proce-
dures, and test runs do not exceed two weeks. At this point, it
has also to be mentioned that the short times for complex pro-
cess changes demand a very high flexibility as well as a high
level of knowledge from the staff. The operators have to be
carefully trained on a regular basis to ensure that their capabili-
ties are sustained.
www.cet-journal.com ª2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eng. Technol. 2016,39, No. 10, 1835–1844
Figure 7. (a) Standard two-column setup for an extractive distillation compared with (b) the setup of the extractive dividing-wall
column.
Figure 8. Flexible hoses as interconnections between fixed
equipment piping in a Lonza’s multipurpose-plant coupling
station.
1842 Research Article
Recently, another challenge of multipurpose production sites
has been tackled by Lonza. Since the future demands of the
processes are unknown, different unit operations have to be as
flexible as possible. For instance, a stirred-tank reactor which is
mainly used for reactions is also equipped with an external
heating and cooling device, with a vacuum pump for low-pres-
sure distillations, and with a phase separation system. Assum-
ing that every reactor is equipped with all additional devices, it
ensures a high flexibility but results in a low utilization rate for
particular parts of the equipment.
However, production can be streamlined without losing flex-
ibility by making the additional equipment mobile. For a
recently built production site in Visp, standardized mobile
equipment has been developed that can be connected to each
of the six reactors in the plant. Therefore, not every reactor has
to be fully equipped and as a direct result the investment has
been noteworthy reduced. The modules are flexible and stan-
dardized which means that every module consists of a trans-
portable rack, a process control unit, standard connectors for
utilities, electricity, and ethernet. After the connection to the
reactor, the equipment automatically communicates with the
process control system and can be directly used (plug & pro-
duce). Currently, phase separation modules, heating and cool-
ing modules, and decanter modules are available. This concept
does not shorten the flexibility of the plant, reduces the
change-over time between productions due to standardized
equipment, and reduces the CAPEX and maintenance costs
since a lower number of units exists overall.
8 Conclusions
To defend market share, chemical and pharmaceutical compa-
nies in high-income countries are forced to develop strategies
to maintain their competitiveness. Lonza as a Swiss chemical
company is exposed to the cost pressure since decades due to
high wages and a strong currency. Numerous measures were
applied to reduce costs, accelerate the development process,
and optimize and improve the flexibility of the production.
These measures can be divided into organizational and techni-
cal approaches. Lonza organizes the project management in
groups which consist of interdisciplinary specialists who take
care of the project over its whole life cycle, from first proof-of-
concepts to start-up in the production plant. The absence of
interfaces in this structure minimizes friction and know-how
drain and, therefore, supports the quality and pace of the pro-
cess development.
Technical approaches are manifold and cover intensified unit
operations, smart-scale production, batch-to-continuous
approaches, and the application of fully automated laboratory
equipment in combination with PAT to gain as much informa-
tion from lab trials as possible. This bundle of different
approaches allows short development times. Typically three to
six months are required for custom manufacturing projects,
where a production process is developed for a customer and
implemented into an existing multipurpose plant. For complex
unit operations, such as reactive distillations or dividing-wall
columns, less than twelve months are typical from the first idea
to the start-up of the plant. Large development projects where-
in an entire process together with a large-scale production
plant is developed typically need three to four years. These
times can be considered as very short for the chemical industry.
Last but not least, the mentioned techniques are worthless
without the mindset of the people. An open-minded, creative,
flexible, and cooperative atmosphere is absolutely mandatory
to generate high-quality products as fast as possible.
Acknowledgment
The authors gratefully acknowledge the contributions of
Dr.-Ing. Andreas Klein and Dr.-Ing. Daniel Staak (both Lonza
AG, Visp) in the fields of multipurpose production concepts
and dividing-wall columns and the fruitful discussions.
The authors have declared no conflict of interest.
References
[1] S. Becht, R. Franke, A. Geisselmann, H. Hahn, Chem. Eng.
Process. 2009,48 (1), 329–332.
[2] D. Staak, T. Gru
¨tzner, B. Schwegler, D. Roederer, Chem. Eng.
Process. 2014,75, 48–57.
[3] A. A. Kiss, J. Chem. Technol. Biotechnol. 2014,89 (4),
479–498.
[4] A. A. Kiss, Advanced Distillation Technologies: Design,
Control and Applications, John Wiley & Sons, Chichester
2013.
[5] T. Bieringer, S. Buchholz, N. Kockmann, Chem. Eng. Technol.
2013,36 (6), 900–910.
[6] J. Strube, R. Ditz, H. Fro¨hlich, D. Ko¨ster, T. Gru
¨tzner,
J. Koch, R. Schu
¨tte, Chem. Ing. Tech. 2014,86 (5), 687–694.
[7] Flexible, Fast and Future Production Processes, Final Report
F
3
Factory, Project No. 228867, European Commission Re-
search and Innovation DG, Brussels 2014. http://f3factory.
com/scripts/pages/en/newsevents/F3_Factory_final_report_
to_EC.pdf
[8] J. Ahlers, J. Gmehling, Ind. Eng. Chem. Res. 2002,41 (14),
3489–3498.
[9] M. Horsch, H. Hasse, Chem. Ing. Tech. 2014,86 (7),
982–990.
[10] Z. Xue, T. Mu, J. Gmehling, Ind. Eng. Chem. Res. 2012,
51 (36), 11809–11817.
[11] M. Bortz, J. Burger, N. Asprion, S. Blagov, R. Bo¨ttcher,
U. Nowak, A. Scheithauer, R. Welke, K. H. Ku
¨fer, H. Hasse,
Comput. Chem. Eng. 2014,60, 354–363.
[12] J. Burger, N. Asprion, S. Blagov, R. Bo¨ttcher, U. Nowak,
M. Bortz, R. Welke, K.-H. Ku
¨fer, H. Hasse, Chem. Ing. Tech.
2014,86 (7), 1068–1072.
[13] A. Chanda, A. M. Daly, D. A. Foley, M. A. LaPack, S. Mu-
kherjee, J. D. Orr, G. L. Reid, D. R. Thompson, H. W. Ward,
Org. Process Res. Dev. 2015,19 (1), 63–83.
[14] Die 50%-Idee – vom Produkt zur Produktionsanlage in der hal-
ben Zeit, Final Report, ProcessNet, Frankfurt, 2010. http://
processnet.org/processnet_media/Positionspapier+50+_+
Idee+final-p-1296.pdf
[15] T. Bott, G. Schembecker, Jahrestreffen PAAT-Fachgemein-
schaft, Weinheim, November 2009.
Chem. Eng. Technol. 2016,39, No. 10, 1835–1844 ª2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.cet-journal.com
Research Article 1843
[16] www.invite-research.com/
[17] J. Lang, F. Stenger, R. Schu
¨tte, Chem. Ing. Tech. 2012,84 (6),
883–884.
[18] S. Lier, D. Wo¨rsdo¨rfer, M. Gru
¨newald, ChemBioEng Rev.
2016,3 (1), 16–25.
[19] C. Bramsiepe, S. Sievers, T. Seifert, G. D. Stefanidis, D. G.
Vlachos, H. Schnitzer, B. Muster, C. Brunner, J. P. M. San-
ders, M. E. Bruins, G. Schembecker, Chem. Eng. Process.
2012,51, 32–52.
[20] www.enpro-initiative.de/
[21] A. Klamt, F. Eckert, M. Hornig, M. E. Beck, T. Bu
¨rger,
J. Comput. Chem. 2002,23 (2), 275–281.
[22] E. Horvitz, T. Mitchell, From Data to Knowledge to Action:
A Global Enabler for the 21st Century, Version 11,
Computing Research Association, Washington, D.C. 2010.
http://cra.org/ccc/resources/ccc-led-whitepapers/ (accessed
on August 19, 2015)
[23] B. Roduit, L. Xia, P. Folly, B. Berger, J. Mathieu, A. Sarbach,
H. Andres, M. Ramin, B. Vogelsanger, D. Spitzer, H. Mou-
lard, D. Dilhan, J. Therm. Anal. Calorim. 2008,93 (1),
143–152.
[24] B. Roduit, M. Hartmann, P. Folly, J. Therm. Anal. Calorim.
2014,117 (3), 1017–1026.
[25] a) www.akts.com/thermokinetics.html, b) www.akts.com/
thermal-safety.html (accessed on February 25, 2016)
[26] B. Roduit, M. Hartmann, P. Folly, Thermochim. Acta 2014,
579, 31–39.
[27] K. Plumb, Chem. Eng. Res. Des. 2005,83 (6), 730–738.
[28] D. M. Roberge, B. Zimmermann, F. Rainone, M. Gottsponer,
M. Eyholzer, N. Kockmann, Org. Process Res. Dev. 2008,
12 (5), 905–910.
[29] N. Kockmann, M. Gottsponer, D. M. Roberge, Chem. Eng. J.
2011,167 (2), 718–726.
[30] N. Kockmann, M. Gottsponer, B. Zimmermann, D. M. Ro-
berge, Chem. – Eur. J. 2008,14 (25), 7470–7477.
[31] J. S. Moore, K. F. Jensen, Org. Process Res. Dev. 2012,16 (8),
1409–1415.
[32] J. P. McMullen, K. F. Jensen, Org. Process Res. Dev. 2010,
14 (5), 1169–1176.
[33] A. Stankiewicz, Chem. Eng. Process. 2003,42 (3), 137–144.
[34] G. M. Kontogeorgis, G. K. Folas, Thermodynamic Models for
Industrial Applications: From Classical and Advanced Mixing
Rules to Association Theories, John Wiley & Sons, Chichester
2010.
[35] D. Staak, T. Gru
¨tzner, D. Zollinger, M. Caraucan, J. Chem.
Chem. Eng. 2011,5, 404–409.
[36] G. Heckmann, F. Previdoli, T. Riedel, D. Ruppen, D. Veghini,
U. Zacher, Chimia 2006,60 (9), 530–533.
[37] Mehrproduktanlagen (Ed: J. Rauch), Wiley-VCH, Weinheim
1998.
www.cet-journal.com ª2016 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eng. Technol. 2016,39, No. 10, 1835–1844
1844 Research Article
