In comparison to other disciplines such as physics or engineering, sensors useful for in
situ monitoring of biotechnological processes are comparatively few and they measure
physical and chemical variables rather than biological ones (Locher et al., 1992). The
reasons are manifold but, generally, biologically relevant variables are much more
difficult and complex than others (e.g. temperature, pressure). Another important reason
derives from restricting requirements, namely,
- sterilisation procedures
- stability and reliability over extended periods
- application over an extended dynamic range
- no interference with the sterile barrier
- insensitivity towards protein adsorption and surface growth
- resistance towards degradation or enzymatic break down.
Biomass concentration is of paramount importance both to scientists and engineers. It is a
simple measure for the available quantity of a biocatalyst. It is definitely an important
key variable because it determines—simplifying—the rates of growth and/or product
formation. Almost all mathematical models used to describe growth or product formation
contain biomass as a most important state variable. Many control strategies involve the
objective of maximizing biomass concentration (though it remains to be discussed
whether this is always wise).
The product is almost the only reason why a bioprocess is run. One is interested in
maximising profit which depends directly on the concentration and/or volumetric
productivity and/or on the purity of the product. It is therefore interesting to know these
values. The classical methods to determine product concentrations are typically off-line
laboratory methods and the above statements for substrate determinations are valid here,
too.
The interest and pressure for developing new monitoring techniques, particularly non-
invasive ones, come from both scientists and process engineers. The former aim mainly
at a better understanding of physiology and its regulation. The latter search for robust and
reliable forms of process operation in order to achieve high global quality: volumetric
productivity, product purity and yield as well as high reproducibility.
More and more non-invasive techniques are invented and developed. Some are
experimentally based, others are model based. Many of them are presently state of the art
and not yet state of routine.
A binding link between measurement and control is modelling. Mechanistic
(deterministic) models can promote the understanding of biological and physico-chemical
process mechanisms. A sound and verified basis of experimental analysis is necessary to
create this type of model. Very often they are not available. In many instances, only a few
relevant variables are known (or measurable) and then simplified mechanistic models or
black box-type models are employed. These simplified models, sometimes tailor-made,
are useful in describing and predicting typical trajectories and patterns, and very
importantly, in creating versatile and efficient control algorithms.
Modelling most effectively links theoretical and experimental knowledge, aiming at
describing and improving the understanding of the process, namely,
- the basic understanding of molecular or simply functional mechanisms of microbial
and cellular physiology;
New methodologies for multiphase bioreactors 3 59