Proceedings of the 6th International Conference on Process Systems Engineering (PSE ASIA)

25 - 27 June 2013, Kuala Lumpur.

Applying Process Systems Engineering for

Continuous Improvement in
Pharmaceutical Production
Hirokazu Sugiyamaa,b, Sandra Schinzela, Georg Mllerc, Rainer Schmidta
a

Pharma Technical Operations Bioloigcs, F. Hoffmann-La Roche, Ltd.,

Grenzacherstrasse 124, 4070 Basel, Switzerland
b
Department of Chemical System Engineering, The University of Tokyo,
7-3-1 Hongo, Bunkyo-ku, 113-8656 Tokyo, Japan
c
Site Engineering Basel/Kaiseraugst, F. Hoffmann-La Roche, Ltd.,
Grenzacherstrasse 124, 4070 Basel, Switzerland

Abstract
In the field of process systems engineering (PSE), pharmaceutical industry is recognized
as one of the future research targets [Reklaitis, 2007], and more intensive work is being
desired. With the aim of contributing to this research movement, this paper showcases
industrial application of PSE and related methods for improving pharmaceutial
production processes. The target here is a Roche new manufacturing facility of
Parenterals (i.e., injectable drug products) in Kaiseraugst called PKau, which started
commercial operations in 2012. Already during the facility start-up, we have been
applying PSE methods in improving PKau production processes considering quality,
finance as well as environment in a continuous manner. Three case studies are presented
in this paper: (1) reduction of product losses by applying simple mass flow analysis (2)
mitigating product quality using risk assessment tools, and (3) improving energy
consumption of the facility using Sanky diagram and multiobjective optimization
techniques.
Keywords: Pharmaceutical production, Parenterals, energy efficiency, risk analysis,
industrial case study

1. Introduction
Production processes of pharmaceuticals are facing wide range of pressures. Authorities
are constantly raising the level of GMP (Good Manufacturing Practice) requirements,
and at the same time, healthcare reformations are requesting higher cost-effectiveness.
In order to cope with such situation, companies need to apply various methods not only
from pharmaceutical sciences but also from other fields in an interdisciplinary manner.
Here, the strength of PSE can be explored for continually improving pharmaceutical
processes, which discipline was originally developed for chemical processes and not yet
fully known in pharmaceutical industry. On the side of academia, pharma industry is
recognized as one of the future expansion targets of PSE (Reklaitis, 2007). Up to now,
several industrial case studies have been presented with addressing the usefulness of
process modeling and statistical process control (e.g., Dassau et al., 2006, Dnnebier,
2008). However, more work needs to be done for establishing PSE as a support tool of
continuous improvement in pharmaceutical industry.

Applying Process Systems Engineering for Continuous Improvement in Pharmaceutical

Production
601

Figure 1 Parenterals production Kaiseraugst in Roche, Switzerland: (a) exterior of the

facility, (b) three product forms to be produced and (c) filling line of liquid vials with
isolator technology.
In this paper, we present three case studies of process improvement using PSE and
related methods. The target here is a new Roche manufacturing factory of Parenterals
(or injectables) in Kaiseraugst, Switzerland termed PKau. This facility, shown in Figure
1 (a), produces high-value biologics products aseptically in the forms of liquid vials,
prefilled syringes and lyophilized vials as depicted in (b). Isolator technology shown in
Figure 1 (c) is extensively applied in order to avoid contamination of injectable
products, and aseptic interventions are performed using isolator hand gloves.

2. Case Study 1: Product Loss Reduction

2.1. 2nd Order Heading
The aim of the first case study was to reduce losses of liquid vial products in PKau,
which consists of the following three batch process steps. The first is to compound the
product solution by mixing defined amounts of API, excipients and Water for Injection
(WFI). Secondly the solution is filled into primary packaging materials such as glass
vials, lyophilized if needed, and closed with rubber closures and caps in a sterile
environment. In the last step, 100% visual inspection is performed to sort out defectives
from good items which are sent to the adjacent packaging facility.
In these three steps, there are various causes of product losses, e.g. samples for quality
control taken in the filling step, or products rejected due to cosmetic defects in visual
inspection. These data are accessible in a modern data management system termed
Manufacturing Execution System (MES), however, stored in different units of [L] or
[#vial]. An expert team then created a indicator f for measuring contribution of
individual causes to the overall loss over batches i:
(1)
where
is the API mass [g] in the non-product output at step a in category b, i.e.
dead volume in hardware, samples due to GMP requirements and loss due to poor
operations. The values of m can be obtained by converting data in MES to g-API
equivalent. As an implementation, this indicator was applied to the 17 validation
batches produced at that time point, and the overall f was obtained as shown in Figure 2
(a). In this way, contribution of different causes to the overall loss has become
comparable, which helps management to focus on the most relevant ones for prevention.
As can be seen in Figure 1 (a), the largest contribution is minor defect or items not
fulfilling the required level of cosmetic appearance, e.g. scratch on vial outer surface,
and are thus rejected in visual inspection. The second highest is the product samples
taken for product release testing and various analyses for validation purposes. The third

602

H. Sugiyama et al

is the product solution losses during the machine pre-run phase before the filling
process starts. In this case, management decided to provide resources for improving the
first and third causes and not to focus on the second, because many samples were taken
puprposefully in the validation batches.
In this work, we present the improvement work done on the third cause, since the detail
of reducing cosmetic defects in the PKau processes is to be reported in a separate paper
(Sugiyama and Schmidt, 2013). At the sterile filling of liquid vials, the productcontacting parts of the system, such as product filter, pipes, buffer tank and filling
needles, are sterilized with highly purified steam and dried before the solution is
introduced. During the early start-up, a potential risk was identified that a tiny amount
of saturated steam could influence on the product concentration, and thus the vials filled
in the first takts have been discarded as machine pre-run during validation batch
production, as was seen in Figure 2 (a).
Parallel to the efforts to prevent this risk from the source, various product properties
were measured during the pre-run, in order to determine the degree of potential dilution.
Figure 2 (b) shows the product concentration of vials in the early pre-run phase in
comparison with the upper and lower specification limits (USL and LSL). The
robustness of the result has been quantified using process capability index Cpk:

(2)
where and represent estimates of sample mean and standard deviations. In the case
of Figure 2 (b), the obtained Cpk was larger than 2, which represents a process that has at
least six standard deviation ranges between mean and either of USL or LSL. The
product concentration after the analyzed takts can be expected to be even more robust
than the result in Figure 2 (b). Therefore in this case, management decided to discard
only the takts analyzed, and to declare the takts afterwards as regular product, which
corresponds to around 90% reduction of the pre-run discharge.

Figure 2 Results obtained for reducing product losses: (a) contribution of different loss
causes to the overall loss amount of 17 validation batches (b) concentration of a product
during early phase of machine pre-run.

Applying Process Systems Engineering for Continuous Improvement in Pharmaceutical

Production
603

3. Case Study 2: Managing Product Quality Risks

The second example is about managing product quality risks in the operation. Generally
in pharmaceutical production, there are various sources of potential errors, which could
lead to impact product quality, e.g., human errors. We applied Failure Mode and Effect
Analysis (FMEA), which is commonly used in quality engineering and also one of the
recommended methods from the health authorities [International Conference on
Harmonization, 2005]. In an FMEA of a working process, so-called failure modes, or
potential causes that could lead to unfavorable results, are first generated. Each failure
mode is then evaluated in terms of severity, possibility and detectability, and the socalled Risk Priority Number (RPN) is obtained as follows:
RPN = Severity Probability Detectability

(2)

These three parameters are usually evaluated using integer scores such as [2, 4, 6, 8, 10]
whereas the scoring system depends on the company policy. Table 1 shows an example
of an FMEA study on aseptic intervention in the filling isolator, where the hand glove
could be damaged by scissors used inside the isolator. The obtained RPN is in this case
240, which is classified as high risk according to the company criteria.
Table 1 Example of an FMEA on aseptic interventions in the filling isolator. *SOP
stands for Standard Operating Procedure.
Failure mode

Severity
[2, 4, 6, 8, 10]

Probability
[2, 4, 6, 8, 10]

Detectability
[2, 4, 6, 8, 10]

RPN

Isolator hand
gloves damaged
by an erroneous
use of scissors
inside the isolator

Product
contamination

In a certain
frequency

SOP* in place
for controlling
hand gloves

240
(high risk)

10

Figure 3 Results obtained for managing product quality risk: (a) number and
characterization of failure modes associated with asepctic operations in the filling
isolator before and after the improvement (b) example of an improvement

604

H. Sugiyama et al

Using FMEA, various potential risks have been identified in the entire PKau processes,
i.e., from material reception through compounding, filling, inspection up to product
dispatch. Figure 3 (a) shows the number of failure modes associated with the aseptic
operations during filling, which is one of the most critical operations from the entire
processes. In the initial evaluation, more than half of the failure modes were classified
as medium or high risk. This domination comes from the fact that the study was
performed during early start-up phases, and the process know-how was still under
development. For these risks, various counter measures have been implemented in a
step-by-step manner. For the example in Table 1, the scissors in the isolator were first
rounded, and then removed by modifying the operations that required scissors, as seen
in Figure 3 (b). After implementing this type of improvements, the initial evaluation has
been updated, which resulted in elimination of high risks and increase of low risk
modes, as seen in Figure 3 (a). Use of FMEA enabled prioritization of the error sources
that need to be improved, and guided continuous reduction of quality risks already
during the facility start-up.

4. Case Study 3: Energy Reduction

The third example is on the reduction of energy consumption, which is becoming
increasingly important in pharmaceutical industry nowadays. In order to identify
relevant processes, a Sankey diagram was developed as in Figure 4 (a) for visualizing
the different utilities supplied to PKau from the site service. Here, processes related to
production and ones associated with infrastructure are first differentiated, each of which
contains sub unit-processes in the second level. This hierarchical Sankey diagram
helped focusing on the relevant processes, which in this case was the infrastructure part,
and most prominently the Heating, Ventilation, and Air Conditioning (HVAC) system.
This system is of high importance in pharmaceutical manufacturing, and especially in
Parenterals production where the cleanliness of the working place is maintained with a
proper air conditioning. Production systems of utilities in PKau are also high
contributors such as production of highly purified steam or WFI, which are used for
cleaning and sterilizing production equipment as well as for production itself. Although
this Sankey diagram is based on the data during the start-up phase, a sensitivity analysis
revealed that even after the ramp up the dominance of infrastructure will remain.
With a focus on these infrastructure processes, an expert team was formed to generate
various alternatives in reducing energy consumption, yielding more than ten ideas.
These options were then evaluated with six simple evaluation indicators: energy saving
potential, running cost, investment, GMP risks, safety impact and workers comfort. In
this multiobjective evaluation, almost all options were in a trade-off relation, i.e., Pareto
optima, and a screening was necessary for narrowing down the alternatives. In this case
study, options with high GMP risks were first eliminated, and then the remaining four
Pareto optimal options were selected for further investigation. The second evaluation
was then performed using Net Present Value (NPV) and FMEA for covering the
financial benefits and all associated risks in a detailed manner. This step identified two
out of four options as Pareto optima, which were recommended for implementation.
One was to introduce flexible recirculation in the air conditioning system, which has
high NPV potential with a manageable risk. The other was to avoid unnecessary cooling
of WFI through operator training, which has low potential in both NPV and risks. In our
case, three alternatives including these two Pareto optima were realized, and which will
expectedly reduce energy consumption more than 10 % on the GJ basis as shown in
Figure 4 (b).

Applying Process Systems Engineering for Continuous Improvement in Pharmaceutical

Production
605

Figure 4 Results obtained for reducing energy consumption: (a) hierarchical Sankey
diagram of PKau processes for identifying improvement opportunities (b) comparison
of energy consumption before and after improvements on GJ basis.

5. Conclusions and Outlook

PSE and related methods for process simulation and optimization have been playing a
central role for the continuous improvement in In Roche Parenterals production in
Kaiseraugst. In the examples above, mass flow analysis, FMEA, Sankey diagram and
multiobjective optimization were applied in order to improve process performance
regarding API yield, quality risks and energy consumption, respectively. Use of simple
tools resulted in a large business impact. Moreover, the strength of PSE methods was
demonstrated especially in supporting more data-based prioritization and more rational
decision- making. For demonstrating such usefulness of PSE and related methods in the
pharmaceutical industry, it would be helpful to foster more intensive collaboration
between industry and academia and knowledge-sharing.

References
E. Dassau, I. Zadok, D. R. Lewin, 2006, Combining six-sigma with integrated design and control
for yield enhancement in bioprocessing, Ind. Eng. Chem. Res, 45, 8299-8309
G. Dnnebier, 2008, Troubleshooting and process optimisation by integrating CAPE tools and Six
Sigma methodology, Proceeding of 18th European Symposium on Computer Aided Process
Engineering (Ed. B. Braunschweig and X. Joulia). Elsevier, 943-948
G. V. R. Reklaitis, 2007, Perspectives on process systems engineering R&D in support of
pharmaceutical product/process development and manufacturing. Proceeding of 17th
European Symposium on Computer Aided Process Engineering (Ed. V. Plesu and P.S.
Agachi). Elsevier, 35-38.
H. Sugiyama, R. Schmidt, 2013, Business model of continuous improvement in pharmaceutical
production processes, submitted to ESCAPE 23
International Conference on Harmonisation (ICH). (2005). Quality risk management., ICH,
http://www.ich.org/fileadmin/Public_Web_Site/ICH_Products/Guidelines/
Quality/Q9/Step4/Q9_Guideline.pdf