Multiphase Bioreactor Design

PID control or even on/off control can be used successfully for controlling some
environmental quantities in bioreactors. Temperature is usually controlled with a P(I(D))
controller. Pressure is normally controlled with an on/off control strategy. pH is normally
controlled with P(ID)-control but can be controlled satisfactorily with on/off control as
well. These are the simplest control loops normally found in bioreactor control. For other
control loops it is difficult to implement PID control due to the non-linear and time-
varying nature of the process. This essentially means that the problem is to find a set of
PID parameters that guarantees stability and acceptable tracking properties for the entire
operating range.

Advanced model based control

Model-based control can be classified as linear or non-linear depending on the type of
model it is based on. Additionally, the control system may be adaptive if on-line
measurements are used to tune the model (controller) parameters. They may also be
predictive if the model is used to predict the process dynamic behaviour for a given time
horizon. Many applications of all these control strategies were reported in the literature in
recent years with predominance of linear adaptive and/or predictive control. An overview
of many of these control strategies is given by Chattaway et al. (1993). For those cases
where model uncertainty is a significant constraint, as it is the case of waste water
treatment reactors, robust control is receiving increasing attention (e.g. Georgieva and
Feyo de Azevedo, 1999). Within the non-linear control class black-box non-linear control
using neural networks and Fuzzy control have been catching an enormous attention (e.g.
Montague and Morris, 1994; Shi and Shimizu, 1992; Glassey et al., 1997).
In many control problems more or less complex controller designs are required,
however this is not always necessary. Sometimes the non-linear and time-varying
characteristics of the process can be captured with sufficient accuracy by a mechanistic or
hybrid model within the operating region. In this situation the application of a Direct
Model Control

Figure 3.9 Direct model control.

Multiphase bioreactor design 82