Reprinted from PHARMACEUTICAL ENGINEERING®

The Official Magazine of ISPE

March/April 2010, Vol. 30 No. 2 Process Simulation Tools
www.ISPE.org ©Copyright ISPE 2010

This article
presents a case Design and Optimization of a Large
study focusing
on the design Scale Biopharmaceutical Facility Using
and optimization
of a large scale Process Simulation and Scheduling
biopharmaceutical
facility using
process
Tools
simulation and
scheduling tools. by Abdelaziz Toumi, Christian Jürgens, Carmen Jungo,
Bernd A. Maier, Victor Papavasileiou, and
Demetri P. Petrides

Introduction

T
very difficult to implement after the regulatory
he global competition in the biophar- approval of a new product.
maceutical industry and the increased Process development scientists have a
demand for affordable and effective short time window to optimize the process of a
medicines has shifted the industry’s promising new molecule. Similarly, engineering
focus on manufacturing efficiency. Therefore, teams face challenges within the design and
process development and design are gaining construction of new production lines and facili-
importance. For new products, it is crucial to ties required for manufacturing newly developed
minimize market entry time without compro- products. The challenges of both groups can be
mising product and process quality. This is lessened by the use of appropriate computer
particularly true for biopharmaceuticals for aids, such as process simulators and production
which it is commonly said that “the process scheduling tools.1,2,3,4
makes the product” and process changes are
Figure 1. Monoclonal
antibody production
flowsheet.

March/April 2010 PHARMACEUTICAL ENGINEERING 1

 Process Simulation Tools
The objective of our Large Scale Biotech (LSB) project was developed for the needs of this project.
to support the design of a new production facility at an exist- Our design project involved the modeling and optimization
ing manufacturing site of Merck Serono (Vevey, Switzerland). of a facility equipped with two production lines, each capable
The plant will initially be dedicated to the parallel production of producing a different MAb. Each line includes four produc-
of two different molecules, a Monoclonal Antibody (MAb) and tion bioreactors feeding a single purification train. The two
a fusion protein. Additional MAb and related molecules from production lines have their own independent main equipment,
the Merck Serono pipeline are expected to be manufactured but share tanks for media and buffer preparation. They also
in the same facility in the future. The limited space available share all utilities, such as steam, Water for Injection (WFI),
for the construction of the new facility made the design very Highly Purified Water (HPW), waste collection, and treatment
challenging and the project highly complex. A computerized systems, etc.
process model was developed at an early stage of the basic
design phase of the project to support all design activities Process Simulation Tools –
and facilitate scenario analysis and evaluation. This article Evaluation and Selection
describes the strategy followed for the development of the Computer-aided process design and simulation tools have
model, the challenges faced, and the benefits derived from been used in the chemical and petrochemical industries since
this effort. the early 1960s. Simulators for those industries have been
designed to model continuous processes and their transient
Monoclonal Antibody Production behavior. However, most biopharmaceutical products are pro-
Monoclonal Antibodies (MAbs) are large protein molecules duced in batch and semi-continuous mode. Such processes are
used to treat a wide variety of illnesses, such as rheumatoid best modeled with batch process simulators that account for
arthritis, psoriasis, Crohn’s disease, transplant rejection, and time-dependency and sequencing of events. In the mid 1990s,
a variety of cancers. They constitute the fastest growing seg- Aspen Technology, Inc. introduced Batch Plus (now called As-
ment in the biopharmaceutical industry. More than 20 MAbs pen Batch Process Developer) a recipe-driven simulator that
and fusion proteins are approved for sale in the United States targeted batch pharmaceutical processes. Around the same
and Europe5,6 and approximately 200 MAbs are in clinical time, Intelligen, Inc. (Scotch Plains, New Jersey) introduced
trials for a wide variety of indications.5,7 The market is grow- SuperPro Designer. The initial focus of SuperPro was on
ing by more than 15% per year and is expected to exceed $30 bioprocessing. Over the years, its scope has been expanded
billion in 2010.8,9,10,11 to include modeling of small-molecule Active Pharmaceutical
Figure 1 displays the flow diagram of a typical MAb process. Ingredients (APIs) and secondary pharmaceutical manufac-
The left-hand-side of the diagram displays the seed train (for turing processes. In 2005, Intelligen introduced SchedulePro,
inoculum preparation) and the production bioreactor(s). Such a production planning and scheduling tool. SchedulePro also
processes include several cell expansion steps as well as two functions as a modeling tool that facilitates design, debottle-
to three seed bioreactor steps to expand the volume of the necking, and capacity analysis of multi-product facilities that
inoculum. Cell growth and product formation in the produc- operate in batch and semi-continuous mode.
tion bioreactor takes usually 11 days. Considering the time Discrete-event simulators also have found applications
for cleaning and turnaround activities, the overall cycle time in the pharmaceutical industry, especially in the modeling
of the production bioreactors that operate in fed-batch mode is of secondary pharmaceutical manufacturing processes. Es-
around 14 days. That includes some idle time to synchronize tablished tools of this type include ProModel from ProModel
the cycle time and to accommodate batch to batch changes Corporation (Orem, Utah), Arena and Witness from Rockwell
in fermentation time in a way that a fixed amount of batches Automation, Inc. (Milwaukee, Wisconsin), and Extend from
are produced every week. After a production bioreactor run Imagine That, Inc. (San Jose, California). The focus of models
is completed, primary recovery is initiated, which typically developed with such tools is usually on the minute-by-minute
includes centrifugation for cell removal followed by filtration. time-dependency of events and on animation of the process.
The purification part of the process that follows usually in- Discrete event simulators are often used to evaluate the im-
cludes three chromatography steps, dia-filtration/concentra- pact of variation on step duration and random events, such
tion steps, and virus removal/inactivation steps. The overall as equipment failures and process delays. Material balances,
product recovery yield is around 70 to 80%. equipment sizing, and cost analysis tasks are usually out
Such processes utilize a large number of buffer and clean- of the scope of such models. Some of these tools are quite
ing solutions (usually 20 to 30) that must be prepared on time customizable and third party companies occasionally use
and be ready for delivery when required by the main process. them as platforms to create industry-specific modules. For
The preparation and storage of such buffers involve a large instance, BioPharm Services, Ltd. (Bucks, UK) have created
number of tanks. Most of the tanks are used for the preparation an Extend-based module with emphasis on biopharmaceutical
and storage of multiple solutions and require cleaning after processes.
each use. Estimating the number and size of such tanks is a Microsoft Excel is another common platform for creating
challenging task during the design of such facilities. Figure models for pharmaceutical processes that focus on material
1 does not display buffer preparation and holding activities. balances, equipment sizing, and cost analysis. Some compa-
However, such activities were taken into account in the models nies have even developed models in Excel that capture the

2 PHARMACEUTICAL ENGINEERING March/April 2010

 Process Simulation Tools
time-dependency of batch processes. This is typically done by a specific product. A single basic processing step is called a “unit
writing extensive code (in the form of macros and subroutines) procedure” as opposed to a “unit operation,” which is a term
in Visual Basic for Applications (VBA) that comes with Excel. used for continuous processes. The individual tasks contained
K-TOPS from Biokinetics, Inc. (Philadelphia, Pennsylvania) in a procedure (e.g., Transfer in, Ferment, Transfer Out, CIP,
belongs to this category. etc.) are called “operations.” A unit procedure is represented
Engineers at Merck KGaA (the parent company of Merck on the flowsheet with a single icon that represents the main
Serono) have had experience with chemical/pharmaceutical equipment used. Figure 2 displays the dialog through which
process simulators like Batch Plus and planning tools like operations are added to a vessel unit procedure. On the left-
Orion-Pi from Axxom Software AG (Munich, Germany) and hand side of that dialog, the program displays the operations
SimPlan from SimPlan AG (Munich, Germany). Batch Plus that are available in the context of a vessel procedure; on the
was initially considered for the project, but it was not finally right-hand side, it displays the registered operations for the
adopted because of its limited bioprocess modeling and ad- edited procedure. The two-level representation of processing
vanced scheduling capabilities. Instead, SuperPro Designer tasks (operations in the context of unit procedures) enables
and SchedulePro were selected because the combination users to describe and model batch processes in detail.
of the two tools satisfied both the modeling as well as the For every operation within a unit procedure, the simulator
scheduling objectives of the project. SuperPro Designer can solves a mathematical model representing the material and
effectively model the bioprocess recipes, which can then be energy balance equations. Equipment-sizing calculations are
exported to SchedulePro to generate representative production performed based on the results obtained by the material bal-
schedules for the combined operation of the two production ances. If multiple operations within a unit procedure dictate
lines, thus enabling visualization of the utilization of shared different sizes for a certain piece of equipment, the software
resources, such as buffer preparation tanks and utilities. reconciles the different demands and selects an equipment
Another reason for the selection of these tools was the fact size that is appropriate for all operations. The equipment is
that SuperPro and SchedulePro had already been adopted sized so that it is large enough (e.g., vessels are not overfilled
by the research and engineering departments at the Vevey during any operation), but not larger than necessary (in order
site of Merck Serono where the new facility was going to be to minimize capital costs). Equipment sizes also can be speci-
constructed. The adoption of common tools by multiple depart- fied by the user, in which case, the simulator checks to make
ments created a common platform of communication among sure that the provided size is adequate. For certain types of
the various teams and provided assurance that the start-up equipment, minimum size requirements also are taken into
and handover phases would be smooth. account in order to satisfy constraints, such as minimum
stirring volume in vessels.
Building a Model in a The outputs of batch process simulators include the fol-
Batch Process Simulator lowing:
The first step in building a simulation model is always the
collection of information about the process. In this case, draft • visual representation of the entire process
versions of process descriptions and block flow diagrams, which • material and energy balances
contained information about material inputs and operating • sizing of equipment and utilities
parameters, were available. Missing data forced the team to • estimation of capital and operating costs
make assumptions after consulting with the operations depart- • process scheduling and cycle time analysis
ment. Rough estimates were used at the start of the project
for unknown process parameters and operating times. As the
project progressed, the assumptions were updated several
times and were thoroughly documented in order to comprehend
and track the development of the various models.
The steps of building a batch process model are generally
the same for all batch process simulation tools. The best prac-
tice is to build the model step-by-step, gradually checking the
functionality of its parts. The registration of materials (pure
components and mixtures) is usually the first step. Next, the
flow diagram (Figure 1) is developed by putting together the
required unit procedures and joining them with material flow
streams. Operations are added to unit procedures (see next
paragraph for explanation) and their operating conditions
and performance parameters are specified.
In SuperPro Designer, the representation of a batch process
model is loosely based on the ISA S-88 standards for batch
recipe representation.12 A batch process model is in essence a
batch recipe that describes how to a make a certain quantity of Figure 2. Specifying the operations of a unit procedure.

March/April 2010 PHARMACEUTICAL ENGINEERING 3

 Process Simulation Tools
ulePro for the generation of the multi-product model. Within
SchedulePro, scheduling information imported from SuperPro
Designer related to processing tasks can be expanded in the
following ways:

• For every procedure, an equipment pool (instead of a

single equipment) can be declared representing the list
of alternative equipment that could potentially host that
procedure.
• Auxiliary equipment (e.g., rinse in place skids and transfer
panels) can be assigned, possibly through pools to opera-
tions.
• Flexible delays (i.e., the ability to delay the start of an
operation if the resources it requires are not available)
can be declared, thus relaxing the rigidity in executing a
Figure 3. Equipment occupancy chart. recipe.
• The general availability of resources in time can be declared
through a calendar.
• throughput analysis
• environmental impact assessment All these extra features proved very useful especially in
modeling the media and buffer preparation tasks. The multi-
With respect to process scheduling and cycle time analysis, product model offered us the ability to represent and visualize
the results are typically visualized with Gantt charts that the demand of shared resources, such as media and buffer
display equipment occupancy as a function of time - Figure preparation tanks, utility generation systems, and bio-waste
3. Equipment items grouped by type are listed on the y-axis treatment systems. The structure and boundaries of the multi-
and time is on the x-axis. The horizontal bars in the chart product model are shown in Figure 4.
represent occupancy of the corresponding equipment by a As soon as the multi-product model was constructed, it
procedure during a time interval. Different colors are used to was used to answer a wide variety of questions concerning
represent different batches. Multiple bars of the same color utility and raw material consumption, potential scheduling
on the same line represent reuse of a piece of equipment conflicts, and plant capacity issues.
within a batch, while bars of different colors correspond to
activities (unit procedures) of different batches. Scheduling Challenges Related to
conflicts arising from overlapping activities that share the Model Development and Validation
same equipment are displayed with multiple lines (one for The processes that were analyzed in this project have been
each conflicting activity) and exclamation marks on the y-axis. developed using a platform technology approach that aims
This type of chart enables engineers to resolve scheduling to standardize the number and the sequence of the produc-
conflicts and optimize the cycle time of the process. tion steps as well as the media and buffer solutions used.
All process parameters that affect product quality (e.g., bed
Modeling the Multi-Product Facility height of chromatography purification steps) were fixed by
After the SuperPro Designer models had been developed, the the end of process development. Such process parameters
individual process models (recipes) were exported to Sched- were not altered during the scope of this project. Instead,
the focus was on engineering parameters that affect capital
cost and capacity (e.g., number and size of vessels for buf-
fer preparation and storage, requirement for transfer lines,
cleaning skids, etc.).
Keeping the models up to date proved to be quite chal-
lenging because the design of the facility underwent many
changes. The collection of information concerning changes in
the processes and the general plant design is a tedious and
time-consuming task, due to the fact that many people are
involved. It would be advisable, for future practitioners, to
develop an appropriate information workflow and change-
management process that includes the simulation team,
thus enabling the members of the simulation team to have
constant access to the latest process and plant information.
The validation of the model was based on information
Figure 4. Structure and boundaries of the multi-product model. that was available to the team (e.g., process description, op-

4 PHARMACEUTICAL ENGINEERING March/April 2010

 Process Simulation Tools
erational experience based on past runs, analytical results,
etc.). The validation of the process parameters was based on
batch records from previous runs carried out by the process
development department. Values from existing processes were
used as a first approximation for operations that are similar
in other bioprocesses, such as buffer/media preparation and
CIP/SIP activities.
The modeling of the buffer preparation area was one of
the most challenging tasks of the simulation. That was due to
the fact that many constraints had to be taken into account
- Figure 5. In terms of main equipment, this area included
several buffer preparation vessels. The list of auxiliary equip-
ment included three closed powder transfer systems and two
Rinse-in-Place (RIP) skids. The model included interfaces to
the utilities that are used in buffer preparation and an in-
terface to the tank farm. The preparation of the 40 different
buffers required by the two processes was represented with Figure 5. Buffer preparation constraints.
40 different recipes. The large number of buffers required,
even though platform technology is adopted, is due to the in the future and there is a need for reduced purification
different physical properties of the two products (the first cycle times, the plant may switch to a three-shift operation
product is a monoclonal antibody and the second is a fusion for buffer preparation in order to accommodate the increased
protein). Modeling of buffer preparation and hold activities demands of the purification trains.
was particularly challenging because it involved numerous Sizing of WFI systems is simplified considerably using
connectivity constraints. For example, if a certain ingredient these tools. A WFI system consists of a still that generates
from the tank farm was required for the preparation of a the distilled water, a surge tank, and a circulation loop for
certain buffer, but not all preparation vessels were equipped delivering the material around the plant. Plant capacity may
with a supply line from the tank farm for this certain ingredi- be limited by any of the following:
ent, then some of the preparation vessels could not be used
for preparing that specific buffer. These constraints were • The plant cannot, on average, consume more water than
modeled by specifying appropriate equipment pools for the the still can generate.
various buffer preparation procedures. • The peak demand cannot exceed the capacity of the circu-
The handling of shift constraints also was quite challenging. lation system.
Since certain areas of the production facility were planned to • The surge vessel must be large enough to maintain capac-
operate in a two-shift-mode, appropriate outages (downtime) ity during peak demand.
had to be specified for the involved equipment, and flexible • Periodic circulation loop sanitization cycles may interrupt
delays had to be added to some of the operations. Using flexible all WFI draws.
delays, the tool was able to automatically shift the start of an
operation (or interrupt an operation) in order to accommodate Process simulation can provide reasonable estimates for the
facility downtime and/or unavailability of required resources. sizes of the still, the surge tank, and the pumping capacity of
The tool also is able to handle material supply, utility, and the circulation loop. Figure 6 displays the demand of WFI for
personnel constraints. However, such constraints add to the
complexity of the model and increase the computation time
significantly. If a problem is over constrained, the tool may
even fail to generate a meaningful solution.

Discussion of Results
The models were mainly used to size shared resources (e.g.,
utilities and media/buffer preparation tanks) and evaluate
various capacity scenarios. The impact of different shift
patterns on equipment demand for buffer preparation also
was evaluated. Using such tools it is easy to quantify the
trade-off between labor cost and capital investment when
management wants to decide whether buffers should only
be prepared during the day shifts or around the clock. The
former option involves lower labor cost, but higher capital
investment. However, it also constitutes a solution of higher
Figure 6. Instantaneous (red lines), 12-h averaged (blue lines), and
inherent capacity. More specifically, if product titers increase 12-h cumulative (green lines) WFI demand as a function of time.

March/April 2010 PHARMACEUTICAL ENGINEERING 5

 Process Simulation Tools
demand indicates the minimum pumping capacity for the
system (23,000 kg/h). The peak 12-h average rate provides
an estimate for the still capacity (10,600 kg/h) and the cor-
responding 12-h cumulative peak is an estimate of the surge
tank capacity of 128,000 L. The trade-off between still rate
and surge capacity can be examined by changing the aver-
aging time. Selecting a longer period predicts a larger surge
tank and a lower still rate. Figure 7 displays the inventory
profile of WFI in the surge tank (green lines) for a tank size
of 130,000 L and a still rate of 11,000 L/h. The still is turned
on when the level in the tank falls below 35% and it remains
on until the tank is full. The operation rate and frequency of
the still is depicted by the blue step-function lines.
Sizing of bio-waste treatment systems can be handled in
Figure 7. WFI inventory (green lines) and operating frequency of
still (blue lines).
a similar way. Such systems typically involve two tanks that
alternate in operation periodically (while one is receiving,
such a plant. The chart shows the instantaneous (red lines) the other is treating a batch of waste material). The peak
and the 12-h average (blue lines) demands. The chart also cumulative amount for the alternating period indicates the
shows the 12-h cumulative demand (green lines) that cor- minimum capacity of each tank.
responds to the y-axis on the right. The peak instantaneous The tools also were used to analyze the impact of buffer

Figure 8. File diagram representing the evolution of the scenarios: Buffer Hold (BH), Buffer Preparation (BP), Rinsing in Place (RIP), and
Sterilization in Place (SIP).

6 PHARMACEUTICAL ENGINEERING March/April 2010

 Process Simulation Tools
expiration times, shift patterns, equipment sizes, and number from the model. The final model also contained constraints for
of equipment items. Approximately, 35 different scenarios the delivery lines, the Rinse-in-Place (RIP) skids, the powder
were evaluated during the project and most of the scenarios transfer systems, the connectivity to the tank farm, and the
included major model updates. As the project evolved, the personnel resources, including shift patterns.
team’s understanding of the processes, the facility, the underly- Using the model, a number of potential bottlenecks mainly
ing links, and constraints improved, and the knowledge gain associated with cleaning equipment and delivery lines were
was used to improve the models. Figure 8 shows the evolution identified and resolved. Capacity analysis enabled the team
of the models up to scenario No. 15. to identify a number of opportunities for equipment savings.
As mentioned before, the initial stages of the project focused That approach worked especially well for areas with multiple
on the development of the SuperPro models of the two pro- parallel equipment items, such as media and buffer prepa-
cesses (one for each product). The SuperPro models were then ration. When analysis revealed that spare capacity existed,
combined in SchedulePro to generate the first multi-product resources were gradually removed from the equipment pool
model. Next, a rough model representing media preparation and feasibility rechecked. That eventually resulted in infea-
was added to the multi-product model. Two different options sible situations. Addition of an extra resource item led to the
for buffer preparation and holding were evaluated. Option optimal solution.
number one involved refilling of the buffer hold tanks after
every batch of the corresponding main process. That led to Return on Investment
a set of scenarios where the maximum number of buffer Table A summarizes the subjects that were analyzed and
preparation batches was performed (red scenarios in Figure the benefits that were derived from the use of simulation
8). Option number two involved the preparation of larger buf- tools. The core of the analysis was done during a period of 12
fer batches that could supply multiple batches of the main months. Besides the financial aspects, there were additional
process. That led to a set of scenarios where the minimum benefits that are hard to quantify, but are equally valuable. The
number of buffer preparation batches was performed (blue common language of communication that process simulation
scenarios in Figure 8). The final design evolved out of the brings to the different stakeholders was probably the most
blue set of scenarios. important qualitative benefit. The members of the various
The simulation of the process support areas was quite teams involved with plant design and operations were able
challenging and required an iterative approach. The buffer to communicate effectively despite the fact that they were
preparation area was initially represented with a simplified looking at the plant from different points of view: engineer-
model. Next, minimum cycle times for each process were ing vs. operations vs. maintenance. It was recognized that
specified and the tool was used to generate feasible solutions. the graphical presentations generated by such tools helped
Experienced manufacturing engineers were then asked to stakeholders to visualize the problems and come up with
evaluate the results and confirm that the generated solu- solutions more efficiently.
tions would work out in practice. For questionable solutions,
improvements were proposed involving rearrangement of Model Lifecycle Management and
existing equipment or installation of additional equipment. Hand-Over to the Operations Team
Then, the changes were incorporated into the model and The simulation work was intended to support the engineering
feasibility was checked once again. That worked very well for team during the detailed design phase. However, the simu-
the buffer preparation area and valuable results were gained lation model continues to live and evolve in the operations

No. Subject Initial Approach Benefits

1 Vessels for the buffer The initial number had been estimated using basic The detailed model enabled the team to eliminate one 2,500 L and
preparation area engineering assumptions and conservative design. two 8,000 L tanks, resulting in savings of more than $1.2 million
(€0.85 million).
2 Sharing of the bulk filtration The initial design assumed a bulk filtration unit for Simulation showed that sharing of the unit by the two production
unit each production line. lines is feasible, leading to savings of $1.4 million (€1 million).
3 Sizing of HPW and WFI The initial design was based on overall averaged The detailed simulation model enabled the team to size these
supply systems demand without taking into account the demand as a systems more accurately.
function of time.
4 Sizing of waste treatment The initial design was based on simplified spreadsheet The detailed simulation model enabled the team to size these
systems models. systems more accurately and reduce capital expenditures.
5 Tank farm sizing In the plant, basic chemicals are stored in the tank The detailed simulation model enabled the team to size the tanks and
farm. The number of tanks and their sizes had been the delivery lines more accurately and confirm the reliable supply of
estimated using crude spreadsheet models. these chemicals to the production lines.
6 RIP routing in buffer The initial piping design for this area was so crowded The process simulation model showed that this is achievable even
preparation and holding that the simulation team had been asked to evaluate with additional rinsing of the tri-blender (a closed introduction
areas the impact of an alternative piping design which uses system for buffer preparation).
fewer pipes and couples the usage of two RIP stations.
Table A. Subjects analyzed and benefits derived.

March/April 2010 PHARMACEUTICAL ENGINEERING 7

 Process Simulation Tools
department. The detailed model, which constitutes a virtual a design could work in reality. The final models have been
plant, was handed over to the operations team to help in handed over to the operations team to be maintained and for
preparing the personnel for the start-up of the plant and its future use. The scheduling models can be used for production
“routine” production schedule. and maintenance planning as well as scheduling in the future.
The model developed in SchedulePro by importing the Su- They might also prove valuable for bringing new products into
perPro Designer recipes of the two processes will be transferred the facility. The SuperPro process models might serve as basis
into the new production facility and serve as a basis for the of decision making for future process changes.
scheduling of the future production activities. However, many
details included in the model are not necessary for on-going References
scheduling purposes and lead to long calculation times (several 1. Petrides, D., Koulouris, A., and Lagonikos, P., “The Role of
minutes) every time a new production schedule is generated. Process Simulation in Pharmaceutical Process Develop-
Currently, a new “simpler” model is under development in ment and Product Commercialization,” Pharmaceutical
SchedulePro to support the scheduling of the future production Engineering, January/February 2002, Vol. 22, No. 1, www.
activities. Less detail will be specified in each unit procedure; ispe.org.
for example, the typical operations of a chromatography cycle
2. Petrides, D., Koulouris, A., and Siletti, C., “Throughput
(e.g., load, wash, elution, regeneration, etc.) will be lumped
Analysis and Debottlenecking of Biomanufacturing Fa-
into a “cycle” activity and consequently a chromatography
cilities, A Job for Process Simulators,” BioPharm, August
procedure will be represented as a sequence of the following
2002.
events: equilibration, cycle-1, cycle-2, … cycle-n, and sanitiza-
tion. Similar simplifications will be implemented in the proce- 3. Petrides, D., and Siletti, C., “The Role of Process Simulation
dures that represent buffer preparation and holding activities. and Scheduling Tools in the Development and Manufac-
The simplified model is intended to be used by the operations turing of Biopharmaceuticals, Proceedings of the 2004
department to: Winter Simulation Conference, Ingalls, R.G., Rossetti, M.
D., Smith, J. S., and Peters, B. A., eds., pp. 2046-2051.
• plan the activities during the start-up of the new produc-
4. Tan, J., Foo, D.C.Y., Kumaresan, S., and Aziz, R.A., “De-
tion facility
bottlenecking of a Batch Pharmaceutical Cream Produc-
• analyze the bottlenecks at full production capacity
tion,” Pharmaceutical Engineering, July/August 2006, Vol.
• analyze and schedule changeovers (change from one process
26 No. 4, pp. 72-82, www.ispe.org.
to another on a production line)
• consider the impact of equipment maintenance on produc- 5. Pharmaceutical Research and Manufacturers of America,
tion schedule Biotechnology Research Continues to Bolster Arsenal
• analyze the influence of a failure or delay of one step on Against Disease with 633 Medicines in Development,
the following steps of a batch and on the scheduling of PhRMA 2008 survey, www.phrma.org.
subsequent batches
6. Walsh, G., “Biopharmaceutical Benchmarks 2006,” Nat.
• understand interdependencies between shared areas and
Biotech, 24(7) 2006:769-775.
production lines
7. Farid, S.S., “Established Bioprocesses for Producing
Conclusions Antibodies as a Basis for Future Planning,” Advances in
When applied early, simulation tools can support plant design Biochemical Engineering/Biotechnology, (2006) 101: 1-42,
and technology transfer and can facilitate the communication published online 8 June 2006.
between the engineering and operations teams. In this project,
8. Pavlou, A.K., Belsey, M.J., “The Therapeutic Antibodies
process simulation was started early during basic engineer-
Market to 2008,” European Journal of Pharmaceutics and
ing and valuable results were obtained from the process
Biopharmaceutics, 59(3) 2005:389-396.
modeling effort. The insight that modeling provided for the
design of the support areas, such as buffer preparation and 9. Jagschies, G., “Where is Biopharmaceutical Manufactur-
holding, utilities, and equipment cleaning requirements, was ing Heading?,” BioPharm International, 21(10), October
of particular importance. In general, process simulation tools, 2008.
such as SuperPro Designer, are useful for understanding and
10. Farid, S.S., “Economic Drivers and Trade-Offs in Anti-
improving a process whereas process scheduling tools, such
body Purification Processes,” BioPharm International,
as SchedulePro, are beneficial for estimating equipment and
http://biopharminternational.findpharma.com/biop-
utility requirements for multi-product facilities. Scheduling
harm/Downstream+Processing/Economic-Drivers-and-
tools also facilitate production planning and scheduling of
Trade-Offs-in-Antibody-Purifi/ArticleStandard/Article/
operating facilities on an on-going basis. Future practitioners
detail/633295, 2 October 2009.
are advised to apply process simulation tools as early as pos-
sible in a project. That way, more synergies can be achieved. 11. Low, D., O’Leary, R., Pujar, N.S., “Future of Antibody Produc-
The use of process simulation in this biopharmaceutical tion,” Journal of Chromatography, B 848(1), 2007:48-63.
project was a success. It provided additional insights on how

8 PHARMACEUTICAL ENGINEERING March/April 2010

 Process Simulation Tools
12. Parshall, J., Lamb, L., Applying S88 – Batch Control from Crucell in Bern. In January 2009, she joined Merck Serono as
a User’s Perspective, Instrument Society of America (ISA), a Process Engineer where she is in charge of the technology
Research Triangle Park, North Carolina, USA, 2000. transfer and start-up of processes to the large scale biotech
production facility. She uses process simulation for in-depth
Acknowledgments analysis of process performance and for improved production
This work could not have been realized without the valuable planning. She holds a PhD in biological engineering from the
contributions of our integrated engineering and operation Swiss Federal Institute of Technology (EPFL) in Lausanne,
teams. The authors are grateful to Petra Wawra, Gabrielle Switzerland. She can be contacted by telephone: +41-21-923-
Wulf, Christophe Waeber, and Jörg Zink for contributing to the 2722 or by email: carmen.jungo@merckserono.net.
identification and evaluation of the different scenarios that Merck Serono S.A., Zone Industrielle B, CH-1809 Fenil-
were analyzed during the project. The authors express their sur-Corsier, Switzerland.
great appreciation to the project directors, Dirk Böhm and
Andre Overmeyer for sponsoring the project and assisting in Bernd A. Maier is a Technical Project
the estimation of the return of investment of the simulation Manager for investment projects at Merck
activities. Finally, the authors wish to thank Herve Broly, Head KGaA. He holds a Diploma and a PhD from
of Process Development at Merck Serono, for reviewing the RWTH Aachen University (Germany), both in
manuscript and substantially improving its quality. biochemical engineering. He can be contacted
by telephone: +49-6151-72-5435 or by email:
About the Authors bernd.a.maier@merck.de.
Abdelaziz Toumi is a Technical Project Merck KGaA, Frankfurter Strasse 250,
Manager for investment projects at Merck HPC: Q029/002, 64293 Darmstadt, Germany.
Serono. He holds a PhD in chemical engi-
neering from the University of Dortmund Victor Papavasileiou at the time of the proj-
(Germany). He can be contacted by telephone: ect was a senior applications engineer with
+41-21-923-2651 or by email: abdelaziz. Intelligen Europe. In his role, Papavasileiou
toumi@merckserono.de. was responsible for the management and
Merck Serono S.A., Zone Industrielle B, execution of process modeling and optimiza-
CH-1809 Fenil-sur-Corsier, Switzerland. tion projects in Europe. He holds an MS from
University College London and a Professional
Christian Jürgens works as a Develop- Doctorate in Engineering (PDEng) from
ment Engineer in the Process Development TU Delft both in biochemical engineering. He is currently
Department of Merck KGaA in Darmstadt, a Business Development Director with Avantium Pharma.
Germany. In his daily work, he is mainly Avantium is a CRO active in the pharmaceutical industry
engaged in the simulation of batch processes, specializing in solid state chemistry. He can be contacted by
data analysis, and process optimization. He email: victor_papavasileiou@yahoo.co.uk.
holds a diploma in chemical engineering from Avantium, Zekeringstraat 29, 1014 BV Amsterdam, The
Clausthal University of Technology. The focus Netherlands.
of his diploma thesis was on simulation of large scale, multi-
product biopharmaceutical facilities. He can be contacted by Demetri P. Petrides is the President of
telephone: +49-6151-72-4337 or by email: christian.juergens@ Intelligen, Inc. He has extensive experience
merck.de. in applying simulation and scheduling tools
Merck KGaA, Department AT-DE, Q29/002, Frankfurter to model, analyze, and optimize integrated
Str. 250, 64293 Darmstadt, Germany. biochemical, pharmaceutical, fine chemical,
and related processes. He holds a BS from
Carmen Jungo has several years of experi- National Technical University of Athens
ence in process development with specializa- (Greece) and a PhD from MIT, both in chemical
tion in upstream biotech process development. engineering. He is a member of ISPE, AIChE, and ACS. He
She has worked as a process development can be contacted by telephone: +1-908-654-0088 or by email:
scientist at pilot scale at the Dairy Research dpetrides@intelligen.com.
Station in Bern, as upstream project leader in Intelligen, Inc., 2326 Morse Ave., Scotch Plains, New Jersey
the R&D Department at Lonza in Visp, and as 07076, USA.
team leader virology process development at

March/April 2010 PHARMACEUTICAL ENGINEERING 9