Multiphase Bioreactor Design

relevant mechanisms in the cell. In the most simple version of such an analysis, all
metabolic reactions are lumped together in a single biochemical reaction describing the
overall cell metabolism. For instance, a biochemical aerobic reaction where one substrate,
ammonia and oxygen are consumed, producing biomass, one product, carbon dioxide and
water is stated in the following way:

When the elemental composition of all species involved is known it is possible to
evaluate the molar stoichimetric coefficients a, b, y z x and y by applying general mass
and energy balance principles (Reels, 1978; Erickson et al., 1978). The well- known yield
coefficients in biotechnology are related to these stoichiometric coefficients by simple
molar-to massbase transformations. The consumption and production kinetics of all the
species are linked together by the stoichiometry, and by the specific rates of growth, μ,
and of product formation, fp.
The growing availability of more detailed models of basic cell growth mechanisms, in
the form of structured and segregated models, poses a challenge to systems and control
scientists and engineers in the search of new methods for bioreactor optimisation and
control that can make use of such available models.

Identification

The use of kinetic models requires a previous identification of the parameters involved.
For instance, when using the Monod model (or alternatives as compiled in Table 3.1) the
maximum specific growth rate μmax, and the yield coefficient YX/S (or YP/S) must be
identified for the actual cultivation conditions. Unfortunately even for such a simple
kinetic relationship as the Monod model this identification is not straightforward,
requiring a careful experimental planning (Baltes et al., 1994; Munack, 1989).
Additionally the coupling of Monod-type kinetic models with mass balance equations
forms non-linear dynamical systems, which, depending on its structure, may not be
identifiable.

A New Perspective: Knowledge Integration Modelling

Modelling through knowledge integration is emerging as an alternative to the classical
modelling approach (Schubert et al., 1994a; Psichogios and Ungar, 1992; Thomson and
Kramer, 1994; Feyo de Azevedo et al., 1997, Simutis et al., 1997). Modelling through
knowledge integration aims at exploring all sources of a priori knowledge/information
about the process that should be optimally expressed and incorporated in the process

Multiphase bioreactor design 70