Membrane-based WFI generation The revised monograph (0169) in the

Ph. Eur. and the Q&A document of the

European Medicines Agency (EMA)
Regulatory requirements, process design and test results currently represent the only available
sources of information concerning the
requirements on membrane-based ge-
DISCUSSION OF REVISED PH. EUR. MONOGRAPH (0169) neration systems for water for injection
FOR WFI AND FINAL Q&A DOCUMENT OF THE EMA purposes (WFI). The statements and
specifications allow lots of room for
The road to revising the monograph (0169) in the Ph. Eur. interpretation. There is a big demand
It’s been a long journey. Discussions and attempts to use membrane-systems to ge- for clear guidelines in the industry.
nerate water for injection purposes (WFI) have existed for many years. In 2002, the This article discusses important aspects
Ph. Eur. monograph (1927) for highly purified water (HPW) was introduced. In the and requirements which arise from the
subsequent years there were further discussions and consultation. Monitored data afore-mentioned documents, taking
was collected for membrane systems and workshops were organized, before the into account the discussion of a work-
revised Ph. Eur. monograph (0169) for water for injection purposes was introduced shop of the ISPE-D/A/CH group “Phar-
in April 2017. mawasser- und Dampfsysteme” (phar-
You could say: Good things come to those who wait. In any case, the details in the re- maceutical water and steam systems)
vised monograph 0169 are brief and leave room for interpretation. Currently (as per in October 2017 about this issue.
Winter 2017/18) there are many opinions about this, some of which are contradicto- Furthermore, a system concept will be
ry. The reasons for this are diverse, and range from positive or negative experiences presented for a membrane-based WFI
in the past and firm convictions, to views which are sales-related, e.g. due to a lack generation system, based on the dis-
of expertise or access to particular technology. cussion points.
Using investigations and results from
The question and answer paper of the EMA the test operation of a system as per
The European Medicines Agency (EMA) has issued a question & answer (Q&A) docu- the presented concept, the reliability of
ment to provide clarifications and explanations, the final version of which was publis- the process is demonstrated and it is
hed on 1st August 2017. However, this document did not clarify all the questions, and shown that WFI can be safely genera-
even threw up additional questions. At the moment, there are no other applicable, ted with membrane-based systems.
supporting documents available for the EU region. At the end, the operating costs are com-
The EMA Q&A document serves as a temporary guideline until Annex 1 of the EU pared. This shows the attractiveness of
GMP Guideline, which is currently under revision, is available. However, it remains to membrane-based WFI generation.
be seen when the revised Annex 1 will be available, and how extensive and detailed
the information about WFI with (cold) non-distillation methods will be. This article was first published in Phar-
mind 01/2018 in the Editio Cantor
Involvement of the regional ISPE D/A/CH group and workshop in Penzberg Verlag.
The need for clear guidelines on the new monograph 0169 and the EMA Q&A docu-
ment is large. Consequently, the ISPE affiliates D/A/CH regional community of practi- Author: Andreas Minzenmay, BWT
ce “Pharmawasser- und Dampfsysteme” (pharmaceutical water and steam systems) Pharma & Biotech, Product Manager
held a specialist discussion and workshops with specialists from the industry over two for cold critical utility systems.
days at the start of October 2017. The objective was to work out the most important

-1-
points as a basis to issue a guideline for the generation of WFI
on the basis of membrane technology.
The work groups discussed process technology/the require-
ments on the final treatment stage, storage & distribution, qua-
lification & monitoring and sanitization.

Overview of the most important points of the Ph. Eur.,

EMA and the ISPE workshop
Experience from the semiconductor industry and with the phar-
maceutical quality of HPW show that membrane-based prepa-
ration systems can reliably generate WFI-quality. Correspon-
ding systems have been used in these areas for many years.
On the basis of this experience, the responsible experts also
came to the decision to permit the membrane-based generati-
on of WFI in the European Pharmacopoeia (Ph. Eur.). Combined EDI/UF module in a compact construction for dead-end operation
What are now the risks and what are the requirements from the (source or figures 1, 3, 6–8: the author/BWT Pharma & Biotech GmbH).
available documents, the Ph. Eur. monograph 0169 and the
EMA Q&A document? causes a standstill or heating of water, which encourages mi-
Reverse osmosis systems normally work at the surrounding tem- crobiological growth. The goal should therefore be to design
perature and are therefore prone to microbiological growth. the system so that the production operation is as continual as
The maintenance of microbiological quality, microbiological possible. Of course, the consumption of WFI in production
growth and the formation of biofilms is therefore identified as plays a key role here. Determining a consumption profile al-
the main risk for membrane-based WFI generation systems. lows the optimum design and coordination of generation ca-
The key conclusion from this is that the system is to be designed pacity, storage tank size and the capacity of the loop system.
and operated in such a way, that this risk can be minimized as Future expansions in capacity can be covered by modular sys-
much as possible. In the EMA Q&A document, a control strate- tems and, if necessary, by extra generation units. On the one
gy is required which identifies the risks and problems based on hand, this provides more safety with regards to microbiologi-
a risk assessment and derives corresponding measures from it. cal growth, and on the other hand, provides more operational
But what does a control strategy and a risk-based approach safety through redundancy.
specifically mean in the context of minimizing the microbiologi-
cal risk. In addition to the design of the system, the importance Pre-treatment
of monitoring, maintenance and qualification are highlighted. The EMA Q&A document highlights that selecting the correct
In the following section, some of the core points for the design pre-treatment procedure makes a key contribution to ensuring
of membrane-based WFI generation systems are addressed that the downstream reverse osmosis stage works properly.
and the discussion points from the ISPE workshop are dealt Doing so requires corresponding specialist know-how, to se-
with. lect the optimum process combination based on the required
fundamental data (e.g. feed water analysis). Some of the most
DESIGN OF MEMBRANE-BASED important pre-treatment techniques include:
WFI GENERATION SYSTEMS & DISCUSSION Multi-layer filtration for particle separation
POINTS FROM THE ISPE WORKSHOP Ultrafiltration for the separation of colloidal substances and
particles.
Process design UV irradiation for germ reduction or sanitisation, to break
The correct selection of the process and the system dimensions down oxidative substances or TOCs (comment: the design
is key to the proper functioning of a membrane-based WFI depends on the intended purpose)
generation system. Bisulfite dosage for removal of free chlorine
The selection of the process combination mainly depends on Active carbon filtration to removal of oxidative substances
the quality of the feed water. In addition to the normal para- or TOCs
meters, special attention is to be paid to the parameters which Softening to remove water hardness
influence the microbiological quality. They are mainly the Total Antiscalant(AS) dosage to stabilize the hardness or for si-
Organic Carbon (TOC), the temperature and the microbiology licic acid
(colony-forming units – CFU). According to the discussions at the ISPE workshop, the fol-
Fluctuations in the quality of the feed water must be observed lowing main process combinations are primarily used to ge-
and taken into account in the process combination. They may nerate WFI:
be seasonal fluctuations or fluctuations due to having different (Pre-treatment (Pre)) > Softening (E) > Reverse Osmosis (RO)
supply networks with different water quality. The availability of > Membrane degasification (MEG) > Electrodeionization
corresponding fundamental data is a key prerequisite for the (EDI) > Ultrafiltration (UF)
correct system concept. (Pre) > Dosage AS > RO > RO > MEG > EDI > UF
Another important aspect is the system capacity. Many phar- They are supplemented, if required, by extra process stages
maceutical water systems are oversized and designed with a (e.g. UV).
lot of “reserve capacity”. This means, however, that the system
is often on stand-by or circulates water internally. This in turn

-2-
Reverse osmosis and CO2 removal The EDI modules often used in the pharmaceutical industry are
While the WFI monograph 0169 states “…Reverse Osmosis either designed using a plate and frame construction or as a
may be single-pass or double-pass…” , in the EMA-Q&A do- spiral-wound module (Fig. 1). Table 1 shows the typical values
cument, the double-pass reserve osmosis process is highlighted which can be achieved after EDI.
as an extra barrier and assurance: „Use of double pass RO
membranes should be considered as an added assurance of Table 1 Expected values after EDI
the maintenance of the quality of the water produced“.
Double-pass osmosis is used, in particular, in combination with Conductivity at 25° C < 0.1 µS/cm
scale control agents. The benefit of this is the extra germ barrier TOC < 50 ppb
while at the same time, the water softener, which is often mi- Aerobic bacteria count < 1 CFU/ml
crobiologically critical, is eliminated and therefore reduces the Endotoxins < 0.25 EU/ml
microbiological risk.
It is often necessary to remove free CO2 to ensure the reliable The physical/chemical requirements of WFI are already safely
operation of the downstream EDI. The permitted CO2 concent- met after the EDI stage. The goal of the last process step is to
rations depend on the manufacturer. Membrane degasification capture microorganisms and endotoxins/pyrogens. The nano-
is a commonly-used process stage for CO2 separation. Air is filtration mentioned in the WFI monograph 0169 and the EMA
normally used as a stripping gas for CO2 removal. Q&A document was assigned a secondary importance in the
Furthermore, the ISPE workshop indicates that slightly-volatile ISPE workshop. In the last process stage, the UF was favored
hydrocarbons can be removed at the same time with vacuum- as a suitable process. The low cut-off of the nanofiltration does
supported operation. not offer any benefit, as the retention is ensured by UF (Fig. 2).
Unfavorable operating conditions, e.g. high operating pressure
Electrodeionization, Nanofiltration and Ultrafiltration and other issues such as the integrity testability, favor the UF.
The WFI monograph 0169 also states that: “…Reverse Osmo- The UF has been used successfully for many years in the pro-
sis, which may be single-pass or double-pass, coupled with duction of HPW as a tried and tested final preparation stage.
other appropriate techniques such as electro-deionisation, ul-
trafiltration or nanofiltration, is suitable.” Normally, the RO/ UF modules come in many different configurations such as hol-
MEG is connected to a downstream EDI, in which the residual low fiber, spiral wound and ceramic modules and can be ope-
desalination and the further reduction of TOCs, SiO2 and CO2 rated in either cross flow or dead end filtration modes.
is done.

Figure 2: Cut-offs of the different filtration procedures

(source: the author, based on: https://en.wikipedia.org/wiki/Membrane_technology).

-3-
For the separation of endotoxins/pyro- adressed and potential for improvement Q&A document mentioned 316l, PVDF
gens in pharmaceutical water, ultrafilters identified: and PP, whereby PP is certainly of minor
are typically used with a cut-off in the There were different solutions and an importance in membrane-based WFI ge-
range of 10,000 – 20,000 Da. Some of open discussion about the question of neration systems. Gaskets and membra-
the UF membranes available on the mar- how the membrane integrity should ne materials are selected according to
ket in pharmaceutical water applications be ensured in operation and when the thermal and chemical resistance.
have a cut-off of 6,000 Da. However, it and how often integrity tests are to
is to be questioned, to what extent these be performed. Sanitization/Cleaning
membranes exhibit a better retention of What maximum service life can be According to the EMA Q&A document,
endotoxins or pyrogens. A cut-off safety used, for example with regard to a an appropriate routine sanitization con-
buffer adjusted to the mode of operation cut-off increase on longer working cept is to be integrated as part of the
should, however, be provided, as par- times of hollow fiber UF modules? control strategy. A combination of chemi-
ticipants in the ISPE workshop reported As there is currently no uniform, cal and thermal sanitization (> 75 °C) is
cut-off changes (increases) to the hollow generally-applicable standard, each recommended.
fiber modules with long service lives. In manufacturer checks the cut-off based Depending on the selected pre-treatment
addition, the cut-off is also influenced by on its own standard. Here there is a process steps and any limited thermal re-
operating parameters such as pressure, desire for standardized cut-off verifi- sistance, thermal sanitization may not be
temperature and flow-rate. cation methods. possible with pre-treatment process steps.
The actual benefits and risks of UF are How is it ensured that there are no A corresponding risk assessment and the
therefore always to be viewed in the con- leachables or extractables (e.g. plas- use of suitable chemical sanitization pro-
text of the module type used, the method ticizer) and that this is certified by the cedures are necessary.
of operation, the service life and other manufacturer? For the microbiologically-critical softe-
marginal conditions. A hollow fiber UF The availability of corresponding do- ning stage and the downstream process
module operated in cross-flow may have cumentation (certificates and reports) stages (RO, MEG, EDI, UF), the hot water
a longer service life. With the increasing from the manufacturer. sanitization was deemed to be sensible
service life, the risk of a cut-off increase and necessary at the ISPE workshop. In
or a defect also increases, e.g. due to System design addition, there are other processes for
regular pressure fluctuations caused by It is not new knowledge that a 'hygi- the chemical sanitization of the softeners,
regularly switching cycles during ope- enic' system design reduces the risk if which also include the in-situ generation
ration, repeated integrity tests or saniti- microbiological growth. Consequently, of free chlorine during the regenerati-
zation. With a UF module operated in a GMP-conform design is of high im- on. In this context, we have to refer to
dead-end mode, this risk is lower due to portance. This also includes the minimi- the Biocides Regulation EU 528/2012,
the shorter service life. zation of dead legs, the optimization of which also covers biocides produced in-
There are a variety of test procedures to the pipe routing, drainability, the use of situ (in the example, free chlorine gene-
verify the integrity of the UF module de- pharma-compliant components, pharma- rated in an electrolytical process). The re-
pending on the type membrane construc- compliant connection technology, ade- spective approval processes are currently
tion: the pin-hole test, the gas diffusion quate surface quality for product wetted in progress and transition conditions ap-
test or other test procedures are possible. surfaces (roughness), turbulent flow, wel- ply. With the approval of licenses, the
At the ISPE workshop, the gas diffusion ding technology and much more. transition conditions will end, and the
test was discussed as one of the options. But even here, a risk assessment shows corresponding applications then need a
Here the UF membrane is pressurized that a hygienic design in the individual licence. If these processes are used, it is
with a test gas (sterile air) and the gas process stages, e.g. for water softening, therefore important to make sure that the
diffusion rate is measured and verified. only achieves a limited risk reduction. In manufacturer has access to the corres-
An automatic online integrity test on sys- water softening, the risk of contamination ponding license.
tems seems to be difficult, as it is associ- of the resin bed due to the large surface The suitable sanitization cycles to main-
ated with a high amount of instrument- of the ion exchange resin is considerab- tain the microbiological quality of the
based work. To test individual modules, ly greater than the risk of an incomplete system have to be verified during the
the test apparatus normally has to be hygienic design. In this specific example, qualification of the system. The determi-
manually connected. Often, several UF other solutions to reduce the risk are key nation of individual germ species is also
modules are connected on one rack. - such as ensuring operation is as con- noted in the EMA Q&A document, and
With this construction, individual modu- tinuous as possible, switching between should be included in the strategy to ad-
les have to be taken out of the system working and safety columns and regular just the sanitization concept if necessary.
to test them. It is recommended to find a sanitation. It also makes sense for individual parts of
technical solution where the integrity of With regards to materials, the following a system to be able to be sanitized. The
the UF module can be tested in built-in is noted in the EMA Q&A document: main part of the microbiological load is
condition. “The materials of construction must not kept back from the RO stage. It should
At the ISPE workshop there was a com- be reactive, additive or absorptive…”. be possible to sanitize them individually
mon consensus, that the UF is to be fa- In this respect, and also with regards to – i.e. more regularly – without unnecessa-
vorized as the last process stage. At the the surface requirements, no differences rily stressing the downstream stages.
same time, further questions have been can be seen to the general standard of
pharmaceutical water systems. The EMA

-4-
Monitoring on the maximum service life and route sanitization are menti-
In addition to the normal process parameters, the quality-rele- oned. If the maintenance work is performed by external com-
vant parameters TOC and microbiology are of key importance panies, the selection of a correspondingly-qualified partner is
for membrane-based WFI generation. TOC represents nutrition even more important.
for microorganisms. The removal of TOC and the correspon- The respective sanitization or replacement cycles are to be de-
ding monitoring is therefore of particular importance. termined during the qualification. The benefit of the hot water
While the WFI monograph 1069 prescribes general “regular sanitization is that it allows an automatic procedure, while in
monitoring of total organic carbon”, the EMA Q&A document the event of chemical sanitization, the rinsing of the sanitization
requires online TOC monitoring at different points within the chemicals has to be checked and qualified. With appropriate
system, based on the risk assessment. The quality of the feed monitoring and monitoring equipment – e.g. differential pressu-
water and the selected process combination go into the consi- re monitoring of flux for filters, RO or UF stage – it is possible
derations, about which points require online TOC monitoring. to recognize trends early and introduce preventative measures.
If the feed water quality is constant and known, TOC online With each individual measure, it must be ensured that the high
monitoring of the feed water is definitely not necessary, while requirements on WFI systems are met. For example, the repla-
it is indicated in the event of seasonal or other fluctuations. cement of gaskets and membranes should be documented in
Online monitoring of the TOC after the pre-treatment stage only a traceable way. Intelligent solutions will support this, for ex-
makes sense if the pre-treatment stage is designed for TOC ample seals which are individually packed for each installation
breakdown. The TOC retention is largely done at the RO stage. location and are marked with a QR code. This ensures that
Online TOC monitoring is therefore useful after the RO stage every seal is replaced as planned and the correct seal can be
and after the last process stage, to verify the WFI quality. It installed in every location – without having to try different types
should be noted that the current threshold value for TOC of from a box – and this is also recorded in the documentation as
500 ppb in pharmaceutical water systems is significantly un- “as maintained”.
dercut. TOC values of up to < 10 ppb are achieved. To what
extent the threshold still makes sense and is up-to-date, must be THE CONCEPT OF A MEMBRANE-BASED
questioned. WFI GENERATION SYSTEM
The EMA Q&A document highlights the importance of the Rapid
Microbiological Test Methods (RMM). The underlying methods As described in the section above, the optimum process combi-
are based on bioluminescence, laser-induced fluorescence and nation is always based on the different fundamental data and
other effects. For quite some time, devices have been on the marginal conditions. For the system concept described below, the
market which work according to these methods and which cal- basic values of feed water listed in Table 2 are used as the basis.
culate the cell count online (real time microbiology systems >
RMS). The option of online microbiological monitoring offers Table 2 Example of feed water quality
a lot of benefits, because the results are available online in
real time, without delay. In the event of deviations, appropriate Conductivity < 1000 µS/cm
measures can be taken immediately, without losing time on the
delays associated with conventional methods. This means, for SDI15 < 3 %/min
example, that the continual monitoring of WFI quality and the Opacity < 0.5 FTU
integrity of the last process stage (UF) are possible. TOC < 1.5 ppm
In any case, this test technique is still not very widely-used in Hardness < 270 ppm CaCO3
the pharmaceutical and biotech industry. For this reason, there SiO2 < 40 ppm
is little experience, and outstanding questions, for example re- CO2 < 200 ppm
garding the interpretation of the test results, the comparability Free chlorine < 2 ppm
with conventional methods (CFU calculation) and the valida- pH 7…7.5
tion. At the moment there is still some caution in the industry Temperature 12…18 °C
about using these devices, although industry representatives Microbiology < 100 CFU/ml
have been involved in the OWBA workgroup for many years (drinking water quality)
and promote the benefits and distribution of the online systems. Origin of water Ground water with moderate seasonal
Here, manufacturers are required to ease uncertainties and pro- fluctuations
vide appropriate information and help.
In general, it has to be determined, however, that these online With SDI < 3 and opacity of < 0.5 FTU, no specific pre-treat-
test methods (RMM/RMS) with the option to continually moni- ment steps are necessary with regards to colloidal substances
tor the generated WFI quality, lead to increased safety with in water or solids. Normally a pre-filter (90…100 µm) is in-
membrane-based WFI systems and in the future are certain to stalled to protect the system from rough particles and solids
grow in importance. (rust, sand etc.). Before the RO stage, the hardness either has
to be removed or kept in a solution using scale control agents
Preventive maintenance (AS), to avoid scaling on the RO membrane. As ion exchange-
The importance of (preventative) maintenance is highlighted in based softeners, as described in the previous section, are often
both the WFI monograph 0169 and in the EMA Q&A docu- microbiologically critical, they are not used to reduce the micro-
ment. Regular regeneration/back-flushing, preventative resin biological risk. Instead, a scale control agent is used. This has
replacement, replacement of filters, replacement of gaskets/ other benefits: it requires less space, the reclaim from the water
membranes, replacement of RO and UF membranes depending softening is no longer required, and thereby the environmental

-5-
Figure 3: Process diagram of a membrane-based WFI generation system.

pollution is also eliminated, no salt treatment. In any case, from The whole system can be sanitized with hot water at ≥ 80°C.
experience, regular chemical cleaning of the subsequent RO Chemical sanitization or cleaning is also possible. With re-
stage is necessary when using AS. gards to the operating method, AS is the first RO stage desig-
Free chlorine, as an oxidizing agent, damages membranes and ned with terminals for connection to an external CIP unit. Figure
therefore has to be broken down before the RO stage. This 3 shows the process diagram of the described concept.
can be done using active carbon, UV technology or bisulfite
dosage. Active carbon is critical regarding the microbiological EXPERIENCE FROM TESTS AND THE
risk. The breakdown of free chlorine by UV is associated with OPERATION OF A TEST SYSTEM
a comparatively high energy requirement and cost-intensive UV BASED ON THE PRESENTED SYSTEM
technology. Consequently, a redox-controlled bisulfite dosage
has been selected for this system concept. Depending on the Due to the fact that the three most important pharmacopoeias
marginal conditions, however, UV technology may also be pre- have only been harmonized since April 2017 and allow memb-
ferred. This also reduces the amount of germs. rane-based WFI generation, there have been very few WFI sys-
To protect the RO membranes from particle, a 5-µm filter is tems in the field build according to this concept to date. There
placed before the RO stage. is, however, experience from PW and HPW systems designed
In connection with the AS dosage, the RO stage is designed in a similar way. Table 3 shows the typical values which could
as a double-pass and the concentrate of the second stage is be sustainably achieved with the presented system concept.
recirculated. This allows high system yields and also allows
feed water with a higher salt content. At the same time there is Table 3 Expected values in WFI
a double germ barrier, which also means extra safety and risk Guideline value (µS/cm) at 25° C < 0.1
minimization.
The reduction of CO2 - to below the level for the operation of TOC (ppb) < 20
the EDI -–is done by membrane degasification. Air is used as Aerobic bacteria count (CFU/ml) <1
the stripping gas. Endotoxins (EU/ml) < 0.06
The residual desalination is done using EDI in a seal-free spi-
ral-wound technique. This technique achieves a very good se- With regards to a comparison of the microbiological risk of sof-
paration of the remaining CO2, SiO2, boron und TOC. The tening versus AS dosage, two tests were performed in 2 single-
avoidance of seals and dead spaces provides further microbio- pass RO test systems with identical construction. The systems
logical risk minimization. were operated simultaneously, whereby one of the systems had
In the final ultra-filtration stage, hollow fiber membranes with upstream conventional softening, and in the other system, AS
a cut-off of 15,000 DA are used, which are operated in dead- dosage was provided for hardness stabilization. The tests were
end mode. The UF module is flanged directly on the EDI module performed as part of a Master’s thesis in cooperation with a
(Fig. 1). This concept stands out due to its compact construction Swiss University of Applied Sciences. One of the means used
and low costs per m³ of generated WFI. The exchange of the to quantitively measure the microbiology was flow cytometry.
UF insert is done preventatively in the course of routine main- Flow cytometry is a fluorescence-based RMM procedure, where
tenance. The risk of signs of ageing (loss of integrity, cut-off the total cell count is calculated in TCC/ml. Samples were taken
displacements) is minimized by preventative replacement in 1-2 from the feed water, at the entry point to the RO systems – i.e.
year cycles. after the softening or after the AS dosage - and in the perme-

-6-
ate of both RO systems. Figures 4 and 5
show the microbiological comparison.
The measured values are on average 100
times higher than conventional plate tests.
Figure 4 shows that the values in the drin-
king water and with AS dosage are virtu-
ally identical, while in softened water the
values are much higher. This highlights the
higher microbiological risk of softening.
The microbiological quality of the perme-
ates of both test systems is of a similar
level. With the AS-operated systems, the
values are slightly lower, although the dif-
ference is not statistically significant. This
shows that even with high microbiologi-
cal loads, the first RO stage of the first
germ barrier already filters out the main
microbiological load.
The system concept shown for membrane-
based WFI generation in the section abo-
ve, contains a 3-fold membrane barrier. Figure 4: Microbiological data from 2 test systems (drinking water, drinking water with antiscalant, softened
To test the microbiological safety of this water; source of figures 4 and 5: Felix Frederic Thiele/BWT AQUA AG)
concept, challenge tests were perfor-
med on a test system. For this purpose,
at the entry to the system a concentra-
ted solution of bacteria from the Entero-
coccus faecium strain was delivered
in pulses and the microbiological flow
was calculated across the various pro-
cess stages of the system and over time.
In Fig. 6 the microbiological results cal-
culated using flow cytometry can be
seen at the various sampling points and
over time. The bacteria peak can be
clearly traced in mixing water (S2) and
in the RO concentrate (S7) which have
been loaded with bacteria. After the
first RO stage (S3) a one-off minimal
rise (to approx. 1,000 TCC/100 ml)
occurred after 3 min. which is not visi-
ble in the diagram due to the scaling. Figure 5: Microbiological data from 2 test systems (with the permeate “antiscalant” and the permeate
After the second RO stage (S4), after EDI “softened water”)
(S5.1, S5.2) and after the UF in WFI (S6)
there were no increased values. This confirms that the
first RO stage acts as a safe barrier even with a high-
er microbiological load. The second RO stage and the
downstream UF act as extra safety barriers. The main-
tenance of microbiological quality, the course of micro-
biological growth and the formation of biofilms over a
longer operating period represent further aspects of the
investigation. Long-term tests are being performed for
this on the test system, which were still running at the
time this article was written (as per 10/2017). The re-
sults will be reported once the test has been completed.

Figure 6: Microbiological data of a challenge test with Enterococcus faecium

-7-
COMPARISON OF OPERATING COSTS

To compare the operating costs, a sample calculation was made

with multi-effect distillation (MED) and a membrane-based WFI
generation system, according to the presented system concept,
based on a system output of 10 m³/h. The production marginal
conditions and basic costs listed in table 4 were used as a
basis.

Table 4 Basic data for the operating cost calculation

16 hours/day
Production 5 days/week
50 weeks/year
Drinking water 1.25 €/m³ Figure 7: Breakdown of operating costs of an 8-column MED system
Waste water 1.90 €/m³
Heating steam 41.50 €/t
Cooling water 0.032 €/kWh
Electrical energy 0.11 €/kWh
Softening salt 0.20 €/kg
Dosage chemical (AS) 15.10 €/kg

The different processes have different requirements on the feed

water quality. For the cost assessment it was assumed that the
MED is fed with permeate from a single-pass RO. The costs for
its generation also have to be considered in the cost assessment.
The membrane-based WFI generation system is supplied with
drinking water. In addition, the service life of the components
used plays a key role in the cost assessment. While the regular
replacement of seals and valve membranes is necessary in hot Figure 8: Breakdown of operating costs for a RO-based WFI generation system
processes, due to the thermal load, with membrane-based WFI
generation systems the service life of reverse osmosis membra- It should be noted that a comparison of operating costs always
nes, membrane degasification modules, EDI modules and ult- depends on the underlying marginal conditions. If the marginal
rafiltration modules have considerable influence. The following conditions change, the costs may alter. However, the compari-
service life periods were used as the basis: son clearly shows the large operating costs benefit of membra-
RO membranes: 5 years (the elements of the first RO pass ne-based WFI generation compared to the conventional MED
definitely have a shorter service life, while the elements in method.
the second-pass can have a longer service life.)
EDI and MEG: 6 years Literature
UF: 2 years [1] EUROPEAN PHARMACOPEIA 9.1, Monograph 0169,
Labor costs and system depreciation were not taken into ac- Water for Injections, 04/2017.
count in the operating cost calculation. [2] Questions and answers on production of water for
Table 5 shows the operating costs per m³ of generated WFI. injections by non-distillation methods – reverse osmosis and
biofilms and control strategies (final), European Medicines
Table 5 Operating costs per m³ of generated WFI, and per year Agency (EMA), 01 August 2017.
WFI capacity MED MED RO-based WFI [3] Jochen Schmidt-Nawrot. Revision der WFI-Monographie
10,000 l/h 6 columns 8 columns generation in der Europäischen Pharmakopöe, Pharm. Ind. 77, Nr. 11,
Per m³ 12.59 € 10.46 € 3.73 € 1640–1651 (2015).
Total per year 503,600 € 418,400 € 149,200 € [4] Second supplement to USP39-NF34, General Information,
(1231) Water for Pharmaceutical Purposes, United States
It is clear that the operating costs for distillation drop if the num- Pharmacopeial Convention, December 1, 2016, p. 8389.
ber of columns rises, as the process is more energy efficient. [5] Anthony Cundell, Oliver Gordon, Nick Haycocks, Joe
The largest cost pool is represented by the heating steam costs Johnston, Michelle Luebke, Neil Lewis, et al. Novel Concept
with 68 % (Fig. 7). That means, the lower the heating steam for Online Water Bioburden Analysis: Key Considerations,
costs, the better value of the WFI generation using MED. Applications, and Business Benefits for Microbiological
The operating costs of the membrane-based WFI generation Risk Reduction. www.americanpharmaceuticalreview.com/
are 30-36% in the example, and thereby three times lower than Featured-Articles/140513-Novel-Concept-for-Online-Water-
the costs for MED. The largest cost pool is the feed water costs Bioburden-Analysis-Key-Considerations-Applications-and-Busi-
at 47%, followed by the waste water costs at 20% and the ness-Benefits-for-Microbiological-Risk-Reduction/
energy costs at 19%. The share of maintenance costs at 12% is
significantly higher than with MED, mainly due to the necessary
replacement cycle for the afore-mentioned components (Fig. 8).

-8-