ARQUITECTURAS DE RED DE LOS SISTEMAS PACS

There is much published data on the fermenter design for containment (Pennman, 1989;

Hambleton et al, 1991; Lever and Hambleton, 1992), however, few, if any of these

studies use a quantitative approaeh. Therefore it was apparent that there was a need for a programme to quantitatively assess the release of miero-organisms from a fermentation

or from other process equipment (Ferris et al, 1995; Noble et al, 1997; Noble et al,

1999). Noble et al (1997) monitored the release of E. coli into the fermenter exhaust gas

from a 2 L fermenter using a sampling eyelone-QPCR methodology. It was found that over the time eourse o f a 5.5 hour period, 3 x 1 0 ^ process eells were released into the exhaust gas stream. If there were more data available on the seale o f accidental and incidental release from a proeess then performanee criteria could be written more eonfidently and as a consequence eontained bioprocesses designed with referenee to a measurable set of performance parameters.

The influenee of the operating parameters of a fermentation on the release of micro­ organisms into the exhaust gas was investigated at different seales. It was found that

E.coli release rates inereased with increasing agitation and aeration and were highest in

the largest seale (250 L) vessel. The total number o f S. cerevisiae cells released was

found to be approximately one order of magnitude lower than the observed release rates for E.coli under identical experimental eonditions. The increase in the settling velocity that occurs according to Stokes Law as the particle diameter increases (at a fixed density difference between the particle and the air) suggests that at the air flow rates employed

S.cerevisiae eells are too large to be effectively aerosolised.

Chapter 7. Summary and Conclusions

Throughout this study it was found that micro-organisms were released into the exhaust gas in only very low numbers. At an agitation rate o f 500 rpm and an aeration rate of

1.5 L min’\ approximately 1.5 x 10^ E.coli cells were released per minute from a 2L

fermenter. Based on this result it can be estimated that 5 x10^ cells will be released into the exhaust gas stream over the course of a 5.5 hour fermentation. This is equivalent to the loss of 3.3 pL o f broth liquid at harvest. The estimated number of cells released is

slightly higher than that observed by Noble et al 1997. However, the estimate was based

on the extrapolation o f release measured at stationary phase, when the cell concentration is at its highest and it has been shown (Figure 5.8) that the higher the broth cell concentration, the larger the number of cells released.

The installation of the Turbosep onto a 250 L fermenter was found to reduce the

microbial concentration in the exhaust gas by approximately 6 orders of magnitude

whilst recirculating foam. In the absence o f foam the microbial concentration in the exhaust gas was reduced by approximately 4 orders o f magnitude. The number o f cells

released from the Turbosep per m^ of exhaust gas was approximately 6 x 10^ whilst

recirculating foam and 1 x 10^ in the absence of any foam. As the number released is low, particularly in relation to the total process volumes concerned, containment systems should be designed to reduce the accidental release rather than the incidental. An accidental foam-out, which released 1 mL of fermentation broth, would allow 1 x lO’® cells to escape into the processing environment. Based on a fermentation time of 5.5 hours, the release from this single accident is equivalent to the cumulative incidental release from 764 fermentations.

The Turbosep mechanical foam separator has been shown to be an efficient device for separating foams emerging from highly aerated fermentation processes. However the fouling of the foam probe interferes with the normal antifoam dosing strategy and causes variation in the volumes of anti-foam administered between batches of the same fermentation. A novel control strategy for the addition of anti-foam into a fermenter fitted with a Turbosep has been described. Differential pressure measurements across the Turbosep were linked to the provision and regulation of antifoam addition. Using the Turbosep in conjunction with the differential pressure control strategy allowed the addition of antifoam to be totally automated and dependent on the requirements of the

Chapter 7. Summary and Conclusions

fermentation at any time. The Turbosep was determined to reduce the antifoam

requirements of a 6000 L E.coli fermentation by approximately 6 6 %, leading to a 16 %

increase in productivity without changing any o f the process operating parameters. As the Turbosep manages the foaming process rather than eliminating it through the excessive use of antifoam, light foaming can be harnessed as a means of effective gas transfer. An increased working volume is also possible as foam is allowed to pass into the exhaust gas pipe without compromising the fermentation, thus further increasing the productivity of the process.

Computational fluid dynamics (CFD) has been proven to be an effective way to model the performance o f the Turbosep in terms of particle collection efficiency. Performance curves were produced with approximately the same shape and dso as those obtained by experiment. Furthermore the predicted pressure drops were in excellent agreement with the measured data. The CFD model was able to predict the salient features of the Turbosep flow field, thus providing a better understanding o f the fluid dynamics of the device. It was originally believed that the particle collection mechanics of the Turbosep would be similar to that of any other cyclone. However, CFD has demonstrated that particle capture is more likely to be based on sedimentation according to Stokes Law rather than centrifugal force. The use of the CFD technique to predict the fluid flow and particle penetration allowed a systematic investigation of the influence of the main design parameters on the Turbosep performance to be investigated in a very cost effective manner and provide valuable guidance in the process o f optimising the design of the Turbosep. O f the new designs considered, the greatest improvement in collection efficiency resulted from the removal of the vortex arrestor. On a linear scale this improvement was greatest for smaller particles, but the absolute efficiencies did not approach 100% as they did for the larger particles. The predicted particle collection efficiency of the new design was at least 30 % higher than the original design.