Automatic Control of Bioprocesses - Measurement, Monitoring, Modelling, and Control of Bioprocesses - page 46

Fig. 3 Scheme of a cascade process control system

3.2.2 Multiple-Input Single/Multiple-Output Control

Wahab et al. [ 20 ] applied a multiple-input single-output controller for DO and
nitrate control in a wastewater treatment process (WWT). They carried out
extensive simulation studies on a nonlinear model to demonstrate the superior
performance concerning set-point tracking and disturbance robustness.
Another MISO approach is given by Jenzsch et al. [ 7 ], representing a nonlinear
adaptive controller based on multiple inputs (oxygen uptake rate, carbon dioxide
production rate, and base consumption) to estimate the specific growth rate. They
compare the results of their generic model control with the control performance
employing only a PI controller. The generic model control shows better perfor-
mance due to the model-based feedforward part and online adjusted control
parameters obtained from the state estimation.
Cascade PI/PID controllers are employed to increase the precision of the PI
principle, mostly for nonlinear control problems. Typically a so-called slave
(inner) control loop is nested within a master (outer) control loop, as shown in
Fig. 3 .
Biener et al. [ 21 ] used this principle to precisely control the temperature in the
reactor, considering a controller cascade for the reactor jacket and the reactor
interior. The temperature inside the reactor is the main target of this control.
Therefore, it is the outer cascade circle that is called the master loop. The tem-
perature of the reactor jacket is used for the inner, slave loop. The state observer
for the process control uses a heat balance equation that calculates the specific
growth rate from the heat flow supposed to result from cell metabolism. Based on
the specific growth rate estimation, they formulate a control law for the substrate
feed rate. This controller design was employed in high-cell-density cultivation
(HCDC) of E. coli producing green fluorescent protein. The authors suggested that
the method is advantageous for HCDC because of the high heat flow due to the
high cell density and describe a gain of sensitivity with increasing biomass. The
specific growth rate can be controlled just below the critical growth rate where
overflow metabolites occur. The authors demonstrate that no other measurement is
necessary except DO concentration to guarantee an aerobic milieu. The heat flow
generated by the cells and therefore the specific growth rate can be estimated
reliably. Since the method only uses easily and quickly measurable process vari-
ables, they suggest its potential application in standard industrial bioprocesses.
They recently applied the described method to a Saccharomyces cerevisiae