BioSafety Journal Pontificia Universidad Católica de Valparaíso ISSN: 1366 0233 Vol. 1, Num. 1, 1995

BioSafety
Volume 1, Paper 2 (BY95002), April 26th 1995
Online Journal, URL -http://bioline.bdt.org.br/by

Containment aspects of couplings and connections for biotechnology plant

G Leaver, I W Stewart, and J S Deans*

Industrial Biosafety Project, Biotechnology Services, AEA Technology, 353 Harwell, Oxfordshire.

*Current Address, Health and Safety Executive, Rose Court, Southwark Bridge Road, London.

Code Number: BY95002 
Size of Files:    
Text:  58K  
Graphics:   Line Drawings (Gif) - 89K

SUMMARY

There is much interest in the design of bioprocess plant to satisfy the biosafety requirements of regulations and associated standards. This paper focuses on containment testing of couplings and connections which are often critical components of a bioprocess plant. Four couplings and four seal types were investigated by repeated sterilisation/cooling cycle duties together with preliminary work on the effect of contact with cleaning agents. The overall result was that containment was retained although seals distorted in many cases which would make cleaning-in- place difficult. One failure was noted on a coupling which was attributed to the seal being unsuited for the intended application. The results provided evidence that couplings used in bioprocess applications should offer adequate protection for all levels of containment requirements, provided that the seals are routinely tested and replaced as part of an overall maintenance schedule. It is suggested that the seals should always be tested before use at higher containment requirements.

1. INTRODUCTION

1.1 Regulatory Issues

The design of process plant and equipment for the large scale manufacture of products based on biotechnology techniques, has received greater prominence in recent years. Much of the interest stems from the application of Genetically Modified Organisms (GMOs) deployed in fermentation and cell culture techniques to produce useful products. The technique of in vitro genetic modification, deployed originally some 20 years ago, initially brought fears that the technique could potentially be dangerous. Thus a pro-active approach was taken by the industry to ensure the technique was deployed in a safe manner. The national and international regulatory activity reflected the pro-active and cautious approach to the application of such organisms. As more knowledge has been gained over the years, a balanced approach to their implementation has been developed. In Europe, the European Commission directive on Contained Use of Genetically Modified Organisms (GMOs) (EC, 1990) together with a parallel Directive on Deliberate Release has been the main driving force for the current legislation in EC countries concerning the safe handling of GMOs.

Another growing area of biosafety activity is the European Committee for Standardisation (CEN) Technical Committee (TC) 233). Four working groups are established two of which are most relevant to biotechnology production processes. The work of CEN TC233 is mandated by the European Commission and is providing further detail on developing a common approach to biosafety compliance (Kirsop, 1993).

In general, biosafety principles for both laboratory and production processes have focused on identifying hazard groups of GMOs then relating these to appropriate containment levels and codes of practice. However, the guidelines are qualitative. For example, terms such as minimise release or prevent release are used depending on the hazard group. Qualitative guidelines are favoured by many who prefer the inherent flexibility. Others prefer a more structured approach whereby examples of engineering features required for each level of containment are specified. Such an approach has been published in Germany (BG Chemie, 1989).

A debate between flexible and prescriptive guidance has developed. Within working group 4 of CEN TC 233, the aim has been to produce biotechnology equipment standards which specify levels of equipment performance and methods of measuring its performance. The working group is not principally concerned with specifying equipment design to meet this performance. Nevertheless, there has been a need expressed to specify equipment based on its intended application. For example, how do you specify a category 2 fermenter?

Chapman (1989) suggested a framework for containment design related to a fermenter specification. The concept was based on increasing the mechanical engineering design containment features depending on the hazard category of the GMO. Leaver and Hambleton explored this concept in more detail and reviewed many of the engineering designs and solutions commonly in use on fermenters. A particular point which has aroused considerable discussion and debate is the issue of static seals. Chapman has suggested double seals some with steam tracing. For example, steam traced double seals have been reported to be used on fermenter top lids in some instances.(Walker et al, 1987).

Titchener-Hooker et al (1993) focused on the implications for static seals and the associated process plant design using the Chapman framework. Details on the problems associated with double sealed systems are described by the authors. Leaver and Hambleton (1992) also recognised the practical limitations of providing a secondary back up to a static seal and recommended that further work was undertaken to test the effectiveness of single seal systems and thus provide better guidance. The debate on effective sealing is particularly concerned with arrangements for higher containment systems. The consensus is that the majority of industrial GMO processes will deploy low hazard GMOs (group I) and the principles of Good Industrial Large Scale Practice will be satisfied by single sealed systems. Thus the issue is whether single static sealed systems will be adequate for GMOs classified as group II category.

1.2 Couplings and Connections

In this work, we set out to provide further guidance on the effectiveness of single sealed systems used on couplings and connections. These are often critical components in biotechnology and their failure could result in significant problems such as product contamination and/or breach of containment.

They are used for making temporary or permanent joints between components and bioprocessing equipment. In the food industry, the main selection criteria of a high quality coupling is based on the need to prevent ingress of material into the coupling and contamination of the product (CEN, 1991). This work also set out to provide selection guidance for process biotechnology to limit the egress of biological material into the workplace environment from a joint or coupling.

The most effective connection for pipe-work to prevent both ingress and egress, is to butt-weld connections using an orbital welder (Leaver and Hambleton, 1992). This method has been widely used in the construction of plant for the food and pharmaceutical industries. It is usually an automated process performed in situ by a welding head as it rotates around an argon-filled section of pipe. The welds formed have the same thickness as the pipe wall and have a smooth finish which does not require polishing. The main advantage of butt-welding is that it produces permanent joints which are crevice-free and impermeable to process fluids and microorganisms (Matthew Hall, 1987).

Butt-welding was used on a high containment fermenter (Hambleton et al, 1991). The aim here was to limit the number of points of potential seal failure. The main disadvantage is that servicing of some components requires cutting through the pipe-work and re-welding with the new component. This is an example where it is important to know whether a single sealed system could be just as effective.

2. COUPLINGS AND SEAL TYPES TESTED

We consulted with members of the Industrial Biosafety Project (Leaver, 1994), took advice on the types of couplings most widely used by members, and enquired where further advice was required. The couplings chosen were the IDF, DIN 11851 and clamp types. We also chose to examine RJT couplings which are widely used in the brewing and dairy industries since it was envisaged they were most likely to fail. Thus the RJT would provide us with a means of evaluating the test protocol.

All four types of coupling were tested at one size - 25.4 mm pipe outside diameter or the imperial equivalent.

All seals and couplings were supplied by Memtech Ltd. (Morriston, Swansea, UK) based on a specification that they should withstand steam sterilisation at 121C.

2.1 Description of Couplings

2.1.1 Ring Joint Type (RJT)

This type of coupling is recommended for use where piping systems are frequently dismantled (BS 4825: Part 5: 1991). A half-section of an assembled RJT coupling is shown in Figure 1.

Figure 1 Ring Joint (RJT) Coupling

The seal sits within a groove of the male half of the coupling. The coupling does not need to be accurately aligned when the connection is made and can withstand a degree of "rough handling". Whilst RJT couplings appear to be widely used in the brewing and dairy industry in cases where pipe-work is manually cleaned (Matthew Hall, 1987), they are not known to be used much for biotechnology applications.

2.1.2 International Dairy Federation (IDF)

BS 4825: Part 4: 1977 recommends that the IDF type of coupling is used where CIP is normally practised. The coupling is shown half-section in Figure 2.

Figure 2 International Dairy Federation (IDF) Coupling

Barnsley (1990) stated that IDF fittings were used on their large scale cell culture systems with seals that give them a "reasonably clean inner surface".

2.1.3 Clamp Type

Clamp type fittings are recommended for use where either CIP or frequent dismantling is necessary (BS 4825: Part 5: 1991).

Figure 3 Clamp Type Coupling

Care is needed with certain designs and sizes of these fittings to ensure that they are not over-tightened which could cause the seal to distort. The pipe-work must be accurately aligned to prevent mechanical distortion of the seal when making the connection.

Clamp type couplings are widely used in the biotechnology industry. Some user companies have indicated a preference for clamp fittings rather than screw type couplings because in the event of a spill, screw threads cannot be decontaminated effectively. In our experience, companies who manufacture products for the USA find that clamp type couplings comply with the US Food and Drug Administration (FDA) inspector requirements. Hence the choice of couplings is greatly affected by these considerations.

2.1.4 DIN 11851

The DIN 11851 coupling is similar to the IDF fitting described above. In the standard fitting, there is a crevice and CEN (1991) suggested that this type of fitting was not suitable for use in food processing machinery where CIP is necessary. A non-standard collared sealing ring is available which produces a flush surface at the pipe joint (Matthew Hall, 1987).

Figure 4 DIN 11851 Coupling

2.2 Description of Seals Tested.

Four different seal materials were tested in the couplings.

2.2.1 EPDM (Ethylene propylene dien rubber)

EPM is a rubber manufactured as a copolymer of Ethylene and Propylene. EPDM is produced by using a third monomer. It is heat resistant to a maximum of 200 C in water and steam and is chemically resistant to many organic and inorganic acids, alkali cleaning fluids and many polar solvents such as alcohols and ketones.

2.2.2 NBR (Acryl nitrile butadiene rubber)

Nitrile rubber is the general term for Acryl nitrile butadiene rubber mixed polymer. The Acryl nitrile content in commercial products varies between 18 and 50% and influences the elastomers properties. NBR is heat resistant to 100 C, and while it will withstand 120 C, its lifetime would be expected to be reduced. It is chemically resistant to many diluted acids, alkalis and salt solutions at low temperatures, but NBR is not compatible with polar solvents.

2.2.3 FPM (Fluorocarbon rubber)

Fluorocarbon rubber is heat resistant to approximately 200 C and up to 250 C with a reduced life. It is chemically resistant to most of the inorganic acids and organic solvents but has limited resistance to Sodium hydroxide and organic acids such as acetic acid and butyric acid.

2.2.4 PTFE (Polytetrafluor ethylene)

Polytetrafluor ethylene is fully compatible with most acids, alkalis, salts and solvents encountered in bioprocess operations. It is heat resistant up to 260 C. PTFE is often supplied filled with glass or other material to enhance specific physical properties. All commercial grades of PTFE have a high degree of hardness.

2.3 Design of the Test Method

A standard ageing test, DIN 53508 (Parker-Pradifa, 1986) consists of measuring the physical properties of a sample of the seal material, before and after subjecting it to a known constant temperature for a known time. A minimal change in hardness, tensile strength, elongation at break and tensile stress indicate a good resistance to ageing.

We reasoned that this standard test would not mimic the expected working duty of seals used in biotechnology plants. With the exception of pressure homogenisers, most biological processes are operated at low pressure, < 3bar g, and moderate temperature, < 140 C, (Titchener-Hooker et al, 1993). What is more relevant is that the seals are subjected to cycles of sterilisation, cooling and cleaning. We also reasoned that testing the seal in situ rather than a sample of the seal material was more representative of the expected working duty since it experiences the full range of temperature gradients and pressure loadings.

The series of experimental tests described in this report were therefore designed to emulate working conditions of the couplings in biotechnology process plants. The seals assembled in the couplings were aged by carrying out repeated sterilisation (steam at 121 C and 1 bar) and cooling (water at 25 C and 1 bar) cycles. After ageing the seals, the couplings were subjected to an emission test where liquid was pumped through the coupling at pressures up to 4 bar. Any liquid emitted at the coupling would be collected and measured over a known time period. The final step was to carry out a pressure hold leak test to determine whether this test could detect any ageing effects in the seals of the couplings.

Seals were also subjected to ageing via contact with two cleaning solutions commonly used in biotechnology, sodium hydroxide (1M) and Terg-a-zyme. The latter is an enzyme powder detergent. Reverse osmosis (RO) water and tap water were also used, RO water was used because it is considered 'aggressive', ie it has a high potential for leaching substances from the sealing material, and the tap water to serve as a control. After contact for 24 hours with the liquids, the couplings and seals were subjected to a pressure hold test in order to determine whether this test could detect any difference in the seals of the couplings.

The following combinations of couplings and seals were used to test for the effects of sterilisation/cooling and cleaning fluids (Table 1).

_________________________________________________________

      Table 1 Coupling/Seal Combinations Tested
  
__________________________________________________________
COUPLING                                SEAL
__________________________________________________________

                           NITRILE   PTFE    EPDM    FPM
__________________________________________________________

RJT                           x       sc 
IDF                           x
CLAMP                                         x      x
DIN                           x               sc
_________________________________________________________

2.4 Description of Test Methods

Figures 5 and 6 show the two test circuits used in the experimental programme. The test coupling was butt-welded to 1 inch (25.4 mm) stainless steel pipe. Clamp type couplings were butt-welded to the ends of the pipe which together formed the test piece. The test piece was connected by the clamp fittings to the test circuit using high quality seals constructed of EPDM.

Figure 5 Steam Sterilisation Circuit

Figure 6 Pressure/Leak Test Circuit

This strategy was used so that any loss of containment due to unscrewing could be determined.

2.4.1 Sterilisation/Cooling Tests

Figure 5

2.4.2 Contact With Cleaning Fluids

Each test piece (coupling and seal) was filled with the cleaning fluid under test and left for 24 hours at room temperature. The test piece was agitated several times to ensure full contact between the seal and the cleaning fluid. After contact with the cleaning agent, the coupling was rinsed with "tap" water and left to dry before carrying out further pressure hold tests. A new seal was used for each treatment.

These cleaning fluid contact tests were designed as preliminary investigations.

2.4.3 Pressure Hold Leak Test

Before and after repeated sterilisation cycles and cleaning tests, the test piece was subjected to a pressure hold test in the circuit shown in Figure 6. The test piece was removed from the circuit shown in Figure 5 and the "high quality" EPDM seals, connecting the test piece to the circuit via the clamp fittings, were removed. These seals were only used to connect the test pieces in the ageing tests and were replaced as was deemed necessary by the experimenter. For connection to the pressure hold test circuit, another set of "high quality" EPDM seals were used to connect up the test piece. These seals were reserved for use in the pressure test circuit only. In this way, the pressure test was relevant to the "aged" seal in the test piece. A blanking piece was connected to the outlet of the test piece and nitrogen gas was fed into the pressure vessel.

Prior to the pressure test, the blanking piece was moved to the inlet of the test piece. The system was pressurised and any change in pressure was noted. This pre-check was undertaken to check that the connecting pipework of the test apparatus was leak tight and was carried out routinely over the duration of the test programme.

In later tests, the pressure vessel was not used and nitrogen was fed directly into the test coupling assembly.

By removing the pressure vessel, the volume in the system was reduced by 10 to leave a test volume of 0.1 . The reduced volume enabled a more accurate measure of pressure loss to be determined.

The initial test pressure used was 408 kPa (60 psig). The coupling was held at that pressure for 15 minutes before allowing the pressure to decay. This allowed for thermal equilibrium to occur. It also allowed time for any gas absorption into the seals to occur and for the seal to "bed in".

Typically, the pressure decay was measured over 15 minutes. The pressure test was carried out 3 to 6 times on the same seal.

The pressure data was converted to an equivalent critical orifice leak diameter (Leaver 1991). By doing so, an index of leak tightness which normalises the gas test volume can be calculated.

For nitrogen gas (and air to a close approximation), the critical orifice leak diameter can be calculated from:

      _                    _   0.5
     |                      |
     |     V        | Po |  |
d =  |   ------  ln |----|  |
     |   120 t      | P  |  |
     |_                    _|
where

d  - critical orifice leak diameter (m)
ln - natural log function
P  - absolute pressure at time t (Pa)
P0 - absolute pressure at time 0 (Pa)
t  - duration time of pressure test (s)
V  - total volume under test pressure (m^3)

From the series of pressure tests, the average value of the critical orifice leak diameter was calculated together with its standard deviation. Statistical significance tests were also carried for leakage data on seals before and after treatment by sterilisation/cooling and cleaning agents. The details are described elsewhere (Deans et al, 1993). The purpose of this analysis was to establish whether changes in the pressure testing data were due to experimental variations or due to a real effect from ageing tests.

3. RESULTS

The pre-checks to test for leak tightness of the pressure hold test circuit and the blanking port revealed very low gas leakage. More significant gas leakage was observed from the test pieces themselves.

3.1. Sterilisation/Cooling Tests

A summary of the tests undertaken is provided on table 2. All pressure tests registered a change of pressure over time and thus an equivalent critical orifice leak diameter could be calculated. However, with the exception of the RJT/PTFE test, none of the test pieces released detectable liquid or aerosol at any point during the repeated sterilisation cycles. The table also shows the results of the significance tests.

_____________________________________________________________

Table 2. Summary of Pressure Test Data for Sterilisation /
Cooling Cycle Tests, 
_____________________________________________________________
Coupling and    No of            Leak diameter (um) Standard
seal            sterilisation    deviation in brackets
                cycles
_____________________________________________________________
                               Before     After    Significant
                                                   change?
                              ________________________________

RJT/Nitrile      117            18 (1.4)   25 (2.0)   YES: +
RJT/PTFE          32            16 (3.2)   206 (10)   YES: +
IDF/Nitrile       48            16 (0.8)   23 (5.3)   YES: +
IDF/Nitrile      160            27 (10.1)  29 (2.0)   NO
Clamp/EPDM        41            20 (0.6)   17 (2.3)   NO
Clamp/EPDM       163            15 (2.5)   14 (0.4)   NO
Clamp/FPM        158            15 (0.9)   13 (0.01)  YES  -
DIN/EPDM          65            19 (2.3)   15 (0.5)   YES  -
DIN/EPDM         161            16 (3.7)   14 (0.2)   NO
DIN/Nitrile     1425            16 (4.2)   13 (0.1)   NO
______________________________________________________________
YES +    statistically significant increase   
YES -    statistically significant decrease
NO       no significant change

In contrast, a gross leakage of water of the order of 100-200 mls/min at 1.5 bar g, was noted for the RJT/PTFE test with a corresponding gas leak tightness value of 206 um.

A test to destruction was attempted, to obtain a gas leak tightness value at the point of failure. This was carried out for the DIN/Nitrile test piece. Up to 1425 cycles were carried out representing 44 days of "round the clock" testing. The leak diameter did not alter significantly over this time, and no liquid leakage was seen. The physical condition of the seal was examined at this point. On removal from the coupling, the seal disintegrated and hence the test could not be re-commenced.

Visual examination of the seals after repeated sterilisation revealed that most of the seals had undergone some change. Some of the seals had become flattened eg the nitrile and PTFE seals. The seals fitted in the clamp couplings suffered from "creep" forming a lip protruding into the inner bore. This situation is illustrated by Figure 7.

Figure 7 Clamp Type Coupling. Schematic of Seal Creep Following Repeated Sterilisation/Cooling cycles

3.2 Tests by Contact with Cleaning Agents

The results are summarised on table 3. All the leak diameters are means of three pressure tests. ______________________________________________________________

Table 3. Summary of Pressure Test Data for Cleaning
Agent Tests,   
______________________________________________________________

Coupling      Treatment             Leak diameter (um) 
and seal                     Standard deviation in brackets

______________________________________________________________ 
                                 Before     After  Significant
                                                     change?
RJT/Nitrile  Blank (tap  water)  20 (0.6)  17 (0.2)   YES -
             1M NaOH             14 (0.1)  19 (2.0)   YES 
             Terg-a-zyme         16 (0.7)  17 (4.6)   YES +
             RO water            21 (5.1)  14 (0.3)   YES -
IDF/Nitrile  Blank (tap water)   17 (1.2)  13 (0.4)   YES -    
             1M NaOH             13 (0.2)  14 (0.2)   YES +
             Terg-a-zyme         15 (1.2)  16 (0.5)   NO
             RO water            15 (0.9)  17 (0.3)   YES +
Clamp/EPDM   Blank (tap water)   15 (0.4)  15 (0.4)   NO
             1M NaOH             14 (0.1)  14 (0.2)   NO
             Terg-a-zyme         13 (0.1)  14 (0.7)   YES +
             RO water            14 (0.3)  15 (1.8)   NO
Clamp/FPM    Blank (tap water)   14 (0.8)  14 (0.1)   YES -
             1M NaOH             14 (0.3)  14 (1.0)   NO
             Terg-a-zyme         13 (0.2)  15 (0.6)   YES +
             RO water            14 (0.01) 13 (0.3)   NO
DIN/Nitrile  Blank (tap water)   13 (0.2)  13 (0.1)   No
             1M NaOH             17 (0.9)  26 (1.0)   YES +
             Terg-a-zyme         13 (0.4)  15 (0.2)   YES+
             RO water            13 (0.1)  14 (0.1)   NO

______________________________________________________________

YES +     statistically significant increase 
YES -     statistically significant decrease 
NO        no significant change.

The change in leak diameters were relatively small. The largest change was an increase in leak tightness diameter from 17 to 26 um for the DIN/Nitrile when exposed to 1M NaOH. There was no noticeable change to the physical appearance of the seal when the coupling was dismantled.

4. DISCUSSION

4.1 Interpretation of Results

The data provides encouraging evidence that single sealed coupling devices are capable of "preventing release" in the majority of cases. The gas leakage data suggested that changes were occurring to the seals during the tests. In some cases, the seals were less leak tight to the test gas, in other cases, the opposite was true. There were also cases where changes were too small to be statistically significant (based on a 95% confidence limit, Deans et al, 1993).

The Parker Precision O-Ring Handbook (Parker-Pradifa, 1986) states the gas leakage rate though an O-ring as the sum of:

(a) Diffusion through the O-ring (b) Leakage across the seal face caused by surface finish irregularities which are not filled by the seal.

Gas seal permeabilities based on a typical equivalent orifice leak size of 22 um are calculated in appendix 1. With a test volume of 0.1 l , there was significant change in recorded pressure over the test period of 15 minutes. The gas permeabilities calculated from this data were several orders of magnitude higher than published values of gas permeabilities for elastomers. Therefore, it would seem that the leakage data is due to surface finish irregularities not filled by the seal. An important result was that containment from the process fluid was still retained for these gas leakage values.

Increases in the gas leakage rates, from sterilisation/cooling, could be attributed to a number of factors including:

*    lack of seal heat resistance, 
*    unscrewing of the coupling,
*    distortion of the seal.

The only seal failure experienced in these tests was the RJT coupling fitted with a PTFE seal giving rise to a significant spill or breach of containment.

We were not entirely sure of the precise reason why this occurred. During the tests, it was noticeable that the nut had to be tightened back to its starting torque value. Examination of the seal after it had been removed from the coupling revealed that it had become flattened along the face which makes contact with the female part of the coupling. The PTFE seal is harder than the other seals tested and a higher torque would be required to get a comparable compression and sealing effect. With a flattened seal face, the coupling would require even more torque since the force is applied over a greater surface area of seal. The couplings were tightened back to 27 Nm (20 lbf ft) if they were obviously loose. This would normally be more than sufficient. Although it was mentioned earlier that the coupling does not need to be accurately aligned when making the connection, the hardness of the PTFE seal could result in less tolerance to mis-alignment particularly when the seal has changed shape due to temperature cycling. PTFE has a high thermal coefficient of expansion which would result in expansion and shrinkage during temperature cycles. It is not surprising therefore that retightening of the coupling was required.

As was mentioned in the commentary on design of the tests, the RJT fitting was chosen since it was envisaged it might fail and thus provide us with a means of evaluating the test protocol. It is unlikely that a RJT fitting would be used in the biopharmaceutical industry or bioprocess operations where containment and/or product quality was important. PTFE is unlikely to be suitable seal material for such couplings.

For the cases where the calculated leak diameter decreased as a result of repeated sterilisation cycles, the implication was that the connection has somehow become more leak tight. When this occurred, the seals in question were made either from EPDM or FPM. These seal materials are soft and they may have swollen due to fluid absorption making the seal less permeable to gas. Another reason may be that any microleaks in the seal/metal surface interface may have been reduced as the seal distorted and bedded into the assembly.

Repeated sterilisation was shown to change the physical appearance of most of the other seals. Although these differences may have influenced the gas leaktightness, there was no evidence of loss of containment. In contrast, because the seals deviated from the flush finish, there was likely to be a deleterious effect on the coupling's cleanability. In some cases, seal shrinkage would lead to crevices. With the clamp type fittings, the seal protruded into the inner bore also leading to difficulties decontaminating the dead space developed at the seal.

Even when the nitrile seal was subjected to a relatively high number of sterilisation cycles, it proved very difficult to get a containment failure. The seal at this point was fragile yet no liquid escaped. The number of cycles used equated to approximately 4 years of daily sterilisation.

The tests contacting the seals with various cleaning agents also affected the gas leak tightness of the couplings and seals under test. However, the tests were only performed under a limited challenge to the seals and hence there was no obvious change to their visual appearance. With a decrease in gas leaktightness, the contact fluid could have dissolved a proportion of the plasticizers in the elastomer. Conversely when the coupling become more leak tight, the contact fluid could have caused the seal to swell increasing the number of surface irregularities filled by the seal.

Overall, these ageing tests were designed to be representative of the envisaged duties of a coupling within bioprocessing. An in situ test was likely to be more relevant rather that an ageing test based on a sample of seal material. This was borne out by the seals distorting yet maintaining containment.

The test for loss of process fluid was undertaken at a pressure of 4 bar. This is a relatively low pressure compared with other process duties for seals. Arguably considerably higher pressures could be used to provide higher assurance on the containment integrity. Where higher pressure duties are envisaged, then testing at higher liquid pressures would be helpful. There is not likely to be any real advantage increasing the pressure of the gas leakage test since detectable gas leakage occurred. An important issue is correlating physical tests such as pressure testing, to likely releases of bioprocess material. This is recognised by CEN TC 233 as a subject for future research. A worst case estimate of liquid loss is 0.2 ml/minute (appendix 1). This assumes a single leak path and ignores surface tension resistances to flow. In the tests here, it is implied that the microleaks and associated surface tension would be sufficient to prevent liquid flow.

The one remaining consideration is whether microbial (bacterial) migration could occur even though no aerosol or liquid loss was detected. Bacterial based tests have been developed for food processing equipment and plant where the ingress of microorganisms into the food product is the primary concern. A bacterial leak tightness test has been developed and published by the European Hygienic Equipment Design Group (EHEDG, 1993). This test uses a small motile bacteria. Serratia marcescens was chosen because it is a pigmented organism which is easy to identify. It is able to penetrate through small holes and crevices which are difficult to detect by physical methods.

Bacteria can cross gaps as a result of surface tension and capillarity effects. Investigations of microbial contamination of tin cans during cooling in water, have suggested that a minimum microbial concentration of microorganisms on the outside is necessary for contamination to occur, due to defects in the can sealing (Keith Brown, Campden Food RA, Private Communication). There is no work at present which has investigated either migration of bacteria outwards (egress) or correlated bacteria tightness with a physical leak test.

4.2 Implications for Process Plant Design and Operation

The results suggest that a user would need to replace seals as part of a general maintenance schedule before the seals reached the observed state of distortion. This would be carried out more for reasons of cleanability and product quality rather than for reasons of loss of containment. Single seals on higher containment process plants, designed to prevent release, appear to be the option in the USA according to Titchener-Hooker et al, 1993. The authors also reveal that a company in Europe use single sealed systems for a plant designed to prevent release. The decision was justified by routine environmental monitoring which to date had never detected the release of process organism or product.

Our results with couplings indicated that it is important to select the correct seal material particularly with respect to its hardness and thermal expansion. The material needs to be flexible to provide the necessary sealing effect. One possible concern is the unloosening effect of the coupling from repeated sterilisation particularly for seals having a high thermal expansion. This effect is not that well characterised and some further test work would be worthwhile for added assurance.

For a plant designed to prevent release, there needs to be increased emphasis on the validation and monitoring of containment. A pressure test or other leak test prior to start up, backed up by environmental monitoring during operation, would be a responsible strategy. It is suggested that the seals should always be tested before use at higher containment requirements.

As part of environmental monitoring, checks on surface contamination of equipment for the process organism by swabbing should be considered. This would address concerns on microorganism migration to the outside. A related issue is setting an appropriate pass/fail criterium for a pressure test. In this work, the results suggested that setting an equivalent leak size in the region of 20 um, would ensure no loss of liquid or aerosol. This is equivalent to a leak rate of 2.3 x 10^-2 Pa m^3/s. If the test criterium were based entirely on the theoretical permeation of test gas through the elastomer, then we calculate that the pass/fail criterium would be of the order of 0.2 um equivalent leak size (which is equivalent to a leak rate of 1.4 x 10^-6 Pa m^3/s). This is a more stringent test but should address the uncertainty on microorganism migration and growth. A more sensitive detection method using a tracer gas would be required. The issue of microorganism migration and corresponding physical test methods is an area where further work is needed.

Pressure testing is often carried out as a pre-check to reduce contamination problems with fermenters. Many operating companies have found that contamination problems can be reduced by such practice. In-situ validation and monitoring of couplings and other sealed systems have been successfully undertaken with ultrasonic leak detector devices (Leaver, 1994). This is particularly useful as a quick check for components which may have become loosened. In our opinion, such practice would "minimise release" to the workplace environment.

As part of the maintenance procedure, operating companies should visually inspect the seals' damage and changes in their shape. This would provide helpful baseline data on their useful lifetimes.

There are additional factors to be considered with respect to coupling selection. With quick release couplings such as clamp type, there is the possibility that they could be inadvertently dismantled either by operators or by unauthorised personnel. More "permanent" couplings, such as flanged type or screw type would have less risk of this happening since separate tools would need to be made available. These aspects would need to be considered as part of a risk assessment. Consideration also needs to be given to the effect of pipework flexibility on the performance and maintenance of the seal. Long lengths of hard piping on both sides of a coupling can have unpredictable effects on the seal behaviour due to the differing expansion rates of the pipe and seal. If there is insufficient provision for axial movement of the pipe, then it can be physically very difficult to dismantle a coupling for maintenance purposes. The ability to replace seals and re-align them without damage, is then significantly reduced.

It is important to ensure compatibility between the components forming a coupling. The authors have experienced problems of clamp fitting incompatibility when assembling a component to the exhaust gas line of a fermenter. The component was received welded to its respective halves of the clamp fitting. Unfortunately, the other halves of the coupling connected to the pipe, although specified as the same diameter, were from a different manufacturer. The result was a less than satisfactory sealing arrangement due to differences in the two halves making the coupling connection. The seal became misshapen and significant air leakage occurred around the seal. It was not immediately apparent that there were differences in the coupling halves.

The overall implication from this work is that single sealed couplings should offer adequate protection for all levels of containment requirements provided they are specified and installed correctly, regularly tested depending on the containment required, and the seals replaced on a regular basis commensurate with their use. Much of this would probably be undertaken routinely, particularly in the biopharmaceutical industry, to ensure that the product is protected.

5. CONCLUSIONS

The commercialisation of process biotechnology using GMOs and cell culture techniques is dependent to a significant extent on a balanced regulatory climate. European standards are being developed to underpin and elaborate on the regulations. Bioprocess equipment standards will be based on their performance validation and measurement. There is a need for suitable test methods and data to be used for design and implementation of bioprocess plant.

Couplings and connections are often critical components of a bioprocess plant. Of the types tested, most of them retained containment when challenged by repeated sterilisation/cooling cycle duties and contact with cleaning agents. In many cases, the seals distorted which could make cleaning-in-place difficult. The gas pressure testing carried out on the seals indicated that gas flowrates were higher than that expected based on permeability though seals. A worst case calculation of predicted liquid release was not observed in the tests. These data provide a basis for further work on correlating physical test methods to potential process material release.

The question of microorganism migration, across small gaps, has been raised. Further investigation adapting methods such as the EHEDG bacteria tightness test would be highly useful. The results would allow a realistic test criterium to be specified.

The in situ test method was more representative of the seal duty compared with a standard ageing test on the seal material. It was unlikely that the results of the standard ageing test could be used to assess the containment properties of a coupling. Liquid pressure testing at higher pressures may provide higher assurance on the containment ability.

The work provided encouraging evidence that single sealed couplings should offer adequate protection at all levels of containment requirements, provided that the seals are routinely tested and replaced as part of an overall maintenance schedule. It is suggested that the seals should always be tested before use at higher containment requirements.

Organisations are encouraged to share data on the performance of seals on bioprocess plant. In particular further work on the effect of cleaning agents would be helpful to supplement the preliminary results here. It would also be helpful to provide further data on the effect of repeated sterilisation on the unscrewing effects of a coupling and its subsequent need to be re-tightened.

ACKNOWLEDGEMENTS

Funding for this work was from the UK Department of Trade and Industry, Biotechnology Unit and is gratefully acknowledged. The work was undertaken under the auspices of the Industrial Biosafety Project which has since transferred from Warren Spring Laboratory to AEA Technology. The Industrial Biosafety Project provides an important forum on biosafety technology and the authors will be pleased to provide more details on request.

7. REFERENCES

BG Chemie, (Berufsgenossenschaft der chemischen Industrie) 1989. "Safe Biotechnology, Industrial Operation, Equipment and Standard Practices", Guidelines M 057e, 3/89,ZH 1/343e, Jedermann-Verlag Dr Otto Pfeffer oHG, Postfach 1031 40, 6900 Heidelberg 1, Germany.

Barnsley J H,(1990) "Containment of Large Scale Cell Culture Fermenter Systems". Paper presented at IPSE Conference, Containment in the Pharmaceutical Industry, Stratford-upon-Avon, 12-13 June.

BS 4825, 1976. "Specification for Stainless Steel Pipes and Fittings for the Food Industry. Part 3. Clamp Type Couplings". British Standards Institution.

BS 4825, 1977. "Specification for Stainless Steel Pipes and Fittings for the Food Industry. Part 4. Screwed Type Couplings". British Standards Institution.

BS 4825, 1991. "Stainless Steel Tubes and Fittings for the Food Industry and Other Hygienic Applications. Part 5. Specification for Recessed Ring Joint Type Couplings". British Standards Institution.

Chapman C, 1989. "Client Requirements for Supply of Contained Bioreactors and Associated Equipment" in Salusbury T T (Ed) "Proceedings of the DTI/HSE/SCI Symposium on Large Scale Bioprocessing Safety, 30 November and 1 December, 1988". Warren Spring Laboratory Report Number LR 746 (BY), Stevenage.

CEN, 1991. "Food Processing Machinery Safety and Hygiene Requirements Basic Concepts - Part 2: Hygiene Requirements". Draft European Standard, European Committee for Standardisation, TC153, July 1991.

Deans JS, Stewart IW and Williams A, 1993. "Containment of Couplings and Connections" Warren Spring Laboratory Report CR 3820. AEA Publications, Culham, Oxfordshire.

EHEDG update, 1993 "A Method for the Assessment of Bacteria Tightness of Food Processing Equipment", Trends in Food Science and Technology, June 1993, vol 4, pp 190-192. Full report (Doc 7) from EHEDG, DA Timperley, Campden Food and Drink Research Association, Chipping Campden, Gloucestershire, GL55 6LD.

European Commission, 1990."Council Directive on the Contained Use of Genetically Modified Microorganisms" (90/219/EEC, Brussels 20 March 1990).

European Commission, 1990."Council Directive on the Deliberate Release into the Environment Genetically Modified Organisms" 90/220/EEC, Brussels 20 March, 1990.

Hambleton P, Griffiths J B, Cameron D R and Melling J, 1991. "A High Containment Polymodal Pilot Plant Fermenter - Design Concepts". J Chem Tech Biotechnol 50, pp 167-180.

Kirsop B, 1993. "Development of Equipment Standards in Biotechnology" Pharm Tech, 5 (8) Sep 93, pp 36-44.

Leaver G "Measuring and Monitoring Containment in Bioprocessing Equipment", Hazards XI - New Directions in Process Safety, 16-18 April 1991, UMIST, Manchester. IChemE Symposium Series No 124, pp 349- 361, Hemisphere Publishing Corp, ISBN 0 85295 269 4.

Leaver G and Stewart I W, 1990."Aerosol Formation from Containment Breach of Bioprocess Plant". Proceedings Fourth Annual Conference on Aerosols, Their Generation, Behaviour and Applications. University of Surrey, 9 - 11 April, The Aerosol Society.

Leaver G and Hambleton P, 1992. "Designing Bioreactors to Minimise or Prevent Inadvertent Release into the Workplace and Natural Environment". Pharm Tech Int 4 (3) 18-26.

Leaver G, 1994. "Interpretation of Regulatory Requirements to Large Scale Biosafety - the Role of the Industrial Biosafety Project", Ch. 11 in Biosafety in Industrial Biotechnology, Hambleton P, Melling J, and Salusbury T T (Eds). Blackie Academic and Professional ISBN 0 7514 0204 4, pp 213-239.

Matthew Hall, 1987. "Sterilisable Pipe and Fitting Connectors". Matthew Hall Engineering (Southampton) Ltd, PPFB/7/1987, Available from DTI MT Division, Ashdown House, 123 Victoria Street, London.

McMaster , 1982, Nondestructive Testing Handbook, Vol 1 Leak Testing ASNT/ASM ISBN 0-87170-125-1.

Parker-Pr„difa (1986) "Precision O-Ring Handbook" Parker-Pr„difa GmbH, Postfach 40, D-7127 Pleidelsheim, Germany.

Titchener-Hooker N J, Sinclair P A, Hoare M, Vranch S P, Cottam A, Turner M K ,1993. "The Specifications of Static Seals for Contained Operations: An Engineering Appraisal" Pharm Tech Europe, October, pp 26-30.

Walker P D, Narendranathan T J, Brown D C, Woolhouse F, and Vranch S P, 1987 "Containment of Micro-organisms during Fermentation and Downstream Processing", in Separations for Biotechnology, Verall M S and Hudson, M J (Eds), Ellis Horwood, Chichester, Ch.38, pp469-479.

APPENDIX 1.

Estimation of gas permeability rates and equivalent liquid leakage

Typical values of critical orifice values were 15 to 30 um calculated from gas pressure testing data for couplings that did not emit liquid at 400 k Pa test pressure.

A typical pressure test when the test volume was 0.1 , was a change in pressure from 401 to 190 kPa gauge occurring over 15 minutes. Inserting the strict SI values into equation 1.

      _                                   _  0.5
     |                                    |
      |0.1 x 10^-3      |(4+1) x 10^5   | |   
d =  |-----------    ln |-------------  | |                (2) 
         
     |120 x 15x 60      |(1.9+1) x 10^5 | |             
     |_                                  _|

The above data can also be expressed as a leak rate which is the rate of change of pressure multiplied by the test volume

         |  Po - P  |
    L =  | -------- | V                                    (3) 
         |     t    |


     | (401 - 190) x 10^3  |          
L =  | --------------------|  0.1 x 10^-3                  (4)
     |       15 x 60       |

L is the average leak rate over the pressure test. Leak rates are dependant on the test pressure used and can be converted to a standardised leak rate to allow inter-comparison with tests carried out under different conditions.

Permeabilities of elastomers are expressed in terms of gas at a specific temperature passing per unit time through a specified area of material with a certain thickness under a defined differential pressure usually 1 atmosphere or 100 kPa). The permeabilities increase significantly with temperature.

Values of permeability for different elastomers have been reported by McMaster (1982). The air permeability values quoted range from 0.32 x 10^6 to 45 x 10^6 (Pa m^3/s)/(m^2/m). The permeabilities are expressed as a gas leak rate at 100 kPa differential pressure divided by the elastomer surface area per unit depth.

Thus it is possible to make estimates of the gas permeability through the coupling seal on the initial assumption that the leak rate was due to permeation through the seal.

For the leak data described above, the average differential pressure was (401 + 190)/2 = 295 kPa. It is assumed that the gas permeation rate is proportional to the differential pressure (Parker-Pradifa, 1986). Therefore, the leakage rate at the differential pressure of 100 kPa would be approximately three times less,

L(1bar) = 7.6 x 10-3 Pa m^3/s.

The coupling type was the IDF fitting shown in Figure 2. The exposed surface area of the elastomer is the inner circumference of the pipe multiplied by the axial dimension of the seal (the thickness of seal butting up to the two halves of the coupling. The thickness of the seal is the radial distance from the flush surface to the hexagonal nut.

Thus the exposed surface area divided by the seal thickness (S in SI units) is

                pi d1 ba              
            S = ---------                                  (5) 
                   br

            pi 22 x 10^-3 x 3.5 x 10^-3 
     S =   ----------------------------                    (6) 
                     5 x 10^-3

Thus the permeability (F) is the leak rate divided by S which results in

F = 7.6 x 10^-3 / 4.8 x 10^-2 = 0.16 (Pa m^3/s)/(m^2/m)

This permeability is many orders of magnitude higher than permeabilities quoted in the literature.

Thus, it is inferred that gas leakage across the seal face is caused by surface finish irregularities which are not filled by the seal.

Equivalent liquid leakage rates.

Equivalent liquid leak rates at operating conditions can be estimated from gas leakage rates at test conditions by

         u (Pu-Pd)liq          
Q = 2 L --------------------                               (7)
         u liq (Pu^2 - Pd^2)

Q = estimated volumetric flow rate at operating conditions (m^3/s)

u liq = liquid viscosity at operating conditions

(Pa s) u = gas viscosity at test conditions (Pa s)

(Pu-Pd)liq = upstream and downstream pressures for liquid flow at operating conditions (Pa)

(Pu-Pd) = upstream and downstream pressures for gas flow at test conditions (Pa)

In this situation, the estimated liquid flow at testing conditions of 400 kPa (4 bar) is of interest.

                      1.82 X 10^-5 (400 -100) x 10^3

Q = 2 x 2.3 x 10^-2  -------------------------------------    
(8)

                     0.001 [(295 x 10^3)^2-(100 x 10^3)^2] 

Q = 3.3 x 10-5 m^3/s

This flowrate is equivalent to 0.2 ml/minute. The correlation with gas leakage worst case assumption of a single leak path and ignores surface tension effects.

Since no liquid leakage was observed in practice, it is inferred that the calculation over estimates the liquid flowrate. It is implied therefore that the microleaks (surface finish irregularities not filled by the seal) and associated surface tension would be sufficient to prevent liquid flow.

Published by Bioline Publications and Science and Technology Letters.

Copyright held by the authors.

Editorial Office: biosafe@biostrat.demon.co.uk

The following images related to this document are available:

Line drawing images