EVALUACIÓN DE LA ADMINISTRACIÓN DE SEGURIDAD DE LA INFORMACIÓN

Advances in geomics, proteomics, bioinformatics, and metabolitics, combined with high throughput screening techniques have greatly accelerated the drug discovery process. However, process development has remained essentially unchanged for the past five decades. A simultaneous approach to integrate both process development and drug discovery at the early stage o f drug development cycle is required to reduce the development time and cost. But at the early stage, only small quantities o f materials are available for evaluating the process performance. A new bioprocess development approach based on micro-titre scale bioprocess operations is proposed and studied in this thesis.

The new approach starts with experiments using the micro-titre plate, a miniature bioreactor, and the shake flask. The engineering performance o f all these reactors are evaluated in the thesis and compared to a 20L bioreactor at laboratory scale. Engineering characterisation o f three different types o f reactors is made with the aid o f CFD and the engineering performance in terms o f mass transfer and cell growth are compared using an experimental and theoretical basis.

A new miniature bioreactor with a diameter equal to that o f a single well o f a 24-well plate is designed. M ixing in the miniature bioreactor is provided by a set o f three impellers mechanically driven via a micro-fabricated electric motor and aeration is achieved with a single tube sparger. Parameter sensitive fluorophors are used with fibre optic probes for continuous monitoring o f dissolved oxygen tension and an optical based method is employed to monitor cell biomass concentration during fermentation.

The flow patterns in the miniature bioreactor from CFD analysis o f single-phase flow are typical to a multi-impeller stirred tank. In aerated conditions, at lower rotational speeds, gas bubbles move towards the shaft and the dispersion is very poor. By increasing the rotational speed, the dispersion can be improved. The evolution patterns are in good agreement with experimental observations. The energy dissipation rate is heterogeneously distributed in the reactor, and the highest energy dissipation rate occurs around the impeller regions. Volumetric mass transfer coefficients are predicted using Higbie's penetration model with the contact time obtained from the CFD simulations o f

Conclusions and recommendations

the turbulent flow in the bioreactor. Comparative data are provided from parallel experiments carried out in a 20L (15L working volume) conventional fermenter. Predicted and measured volumetric mass transfer coefficients are in good agreement in the miniature bioreactor and 20L bioreactor although the simulation data in the miniature bioreactor is underpredicted. The kLa data from experiments and simulations in the miniature bioreactor are in the range o f 100 hr’’ to 400 hr"\ typical o f those reported for large-scale fermentation. The fermentation process is evaluated by cultivating E. coli under 1 vvm and different rotational speeds. The cell growth profile is nearly the same as that in a 20L bioreactor.

The flow in the shake flask is characterised by the free surface model o f CFD. The gas- liquid interface changes are mapped as the reactor moves along the rotary platform. The flow patterns in the shake flask are found to be very complicated. The energy dissipation rate is non-homogeneous, and the highest energy dissipation rate occurs along the shaking wall. The power consumption is obtained by integrating the local energy dissipation over the entire working volume and compared to experimental observations by Buchs (2001). The energy consumption is found to be comparable to the experimental values. CFD predictions o f power consumption are more sensitive to shaking amplitude than to shaking frequency. The CFD predicted correlation o f power consumption as a function o f shaking frequency is supported by experimental data. The volumetric mass transfer coefficients obtained from H igbie’s penetration model are in the range o f experimental results. The correlations o f liquid mass transfer coefficient kL and the specific gas-liquid interfacial area as a function o f the shaking frequency are obtained and compared to those from experimental data. CFD predictions are in good agreement with experimental observations.

24-well and 96-well micro-titre plate reactors are investigated and different flow patterns are observed under different shaking conditions. “Out-of-phase” conditions are found in the 24-well reactor at lower shaking frequency. The liquid at the base o f the 24 well reactor remained stationary, and the mixing performance is improved by increasing shaking frequency or shaking diameter. In a 96 well reactor good mixing is achieved at a shaking frequency o f 1000 rpm and a shaking diameter o f 6mm because o f the existence o f horizontal and vertical mixing.

Conclusions and recommendations

The energy dissipation rates in the two reactors are found to be non-homogeneous. Most o f the energy is dissipated along the reactor wall. But in the 96-well reactor, the energy dissipation rate is relatively high at the base o f the reactor, which confirms good mixing in the reactor. The power consumption is strongly affected by the size o f the well and the volume o f shaken liquid as well as the shaking frequency and amplitude. The power consumption decreases in the 24-well reactor as the shaking frequency increases from 300 rpm to 800 rpm, and then increases as expected after the shaking frequency o f 800 rpm. In the 96-well reactor, the power consumption increases monotonically as the shaking frequency increases. The power input is found to be more sensitive to shaking diameter than shaking frequency in both micro titre plate reactors. The volumetric mass transfer coefficients are higher in the 96-well reactor than in the 24-well reactor under the same shaking conditions. The predicted mass transfer coefficients are in the same order o f magnitude as the experimental data. The mass transfer rate is more sensitive to changes in shaking diameter compared to shaking frequency.

High mass transfer rates in the miniature bioreactor are achieved and are comparable to a 20L bioreactor as the convective mass transfer prevails in both reactors. The power consumption in impeller-driving mixing tanks is higher than that in the shaking system. At the same shaking conditions, the mass transfer rate in a 96-well reactor is higher than that in a 24-well reactor; and the power consumption in a 96 well reactor is higher as well. The volumetric mass transfer coefficients in the shake flask are in the same order o f magnitude o f the 24-well reactor, but the power consumption is relatively higher. Overall, the power consumption can be correlated to the mass transfer performance in the different reactors.

Analysis o f the available data on cell growth suggests that all three small-scale systems (microwell plate, shake flask and miniature bioreactors) have the capacity for fermentation and cell culture. However, the growth profiles show differences in the three systems. The miniature bioreactor produces growth profiles that are closest to those in the conventional bioreactor under conditions o f equal energy dissipation rate allowing direct scale-up and/or scale down.

Conclusions and recommendations