tions in the outlet gas stream. On-line temperature measurements can be made
with thermocouples, and it is usually practical to have a number located at different positions with the bioreactor. Outlet gas concentrations can be measured
most conveniently on-line with paramagnetic oxygen analyzers and infrared
carbon dioxide analyzers, which have in fact been used in SSF processes for
many years, within control schemes regulating the inlet air flow rate and the
passage of recycled air through KOH solutions to remove CO 2 [168, 169]. Other
on-line measurements may be possible: a gas chromatograph with an automatic
sampler can be used to measure volatile end product concentrations in the
headspace gases [170]; on-line sensors can also be used to measure relative
humidities in the outlet air stream [171]; meaningful pH measurements may be
possible using normal pH electrodes, depending on the substrate properties [2];
and finally, in packed bed bioreactors, pressure drop measurements can give an
indirect indication of the amount of growth [172].
7.1
Application of Advanced Control Techniques
Simple control schemes in which a single input variable is used to manipulate a
single operating variable have been used in SSF for many years. However, the
degree of control that can be achieved by such control systems is limited due to
the complexity of the processes occurring within SSF bioreactors: airflow rates
and humidities can be controlled to influence the bed temperature, but this also
affects the level of water in the substrate bed. Sargantanis and Karim [164]
showed that, although temperature in a rocking drum bioreactor can be
controlled effectively using a single-input single-output algorithm, in which the
dry air flow rate is varied in direct response to the measured temperature, water
levels are best controlled by a multiple-input multiple-output scheme in which
both total wet weight of the bioreactor and the carbon dioxide evolution rate are
used to control the dry air flow rate and the water replenishment rate.
On-line measurements invariably contain noise, which may come from either
variations in the process itself or from the measuring equipment [173]. In
general it is necessary to process the data before it can be used in control
algorithms, using mathematical filtering procedures such as Kalman filtering or
Butterworth filtering to eliminate measurement noise [163, 164, 173]. The
control schemes which control the operating variables such as inlet air temperature, flowrate, and humidity usually cannot prevent significant variations
from occurring [167]. In the 50 kg stirred capacity bed of Fernandez et al. the
inlet air temperature was controlled by electric air heaters with a selective
on/off control algorithm and the humidity was controlled by steam addition
through an on/off solenoid valve. Inlet air flowrate was set manually. Variations
of ±10% RH occurred in the inlet air relative humidity and variations of 3 °C
occurred in the inlet air temperature. Further smoothing algorithms may also
be required to account for such variations in the values of these operating
variables when processing data from on-line measurements [173].
Biochemical Engineering Aspects of Solid State Bioprocessing
121
with thermocouples, and it is usually practical to have a number located at different positions with the bioreactor. Outlet gas concentrations can be measured
most conveniently on-line with paramagnetic oxygen analyzers and infrared
carbon dioxide analyzers, which have in fact been used in SSF processes for
many years, within control schemes regulating the inlet air flow rate and the
passage of recycled air through KOH solutions to remove CO 2 [168, 169]. Other
on-line measurements may be possible: a gas chromatograph with an automatic
sampler can be used to measure volatile end product concentrations in the
headspace gases [170]; on-line sensors can also be used to measure relative
humidities in the outlet air stream [171]; meaningful pH measurements may be
possible using normal pH electrodes, depending on the substrate properties [2];
and finally, in packed bed bioreactors, pressure drop measurements can give an
indirect indication of the amount of growth [172].
7.1
Application of Advanced Control Techniques
Simple control schemes in which a single input variable is used to manipulate a
single operating variable have been used in SSF for many years. However, the
degree of control that can be achieved by such control systems is limited due to
the complexity of the processes occurring within SSF bioreactors: airflow rates
and humidities can be controlled to influence the bed temperature, but this also
affects the level of water in the substrate bed. Sargantanis and Karim [164]
showed that, although temperature in a rocking drum bioreactor can be
controlled effectively using a single-input single-output algorithm, in which the
dry air flow rate is varied in direct response to the measured temperature, water
levels are best controlled by a multiple-input multiple-output scheme in which
both total wet weight of the bioreactor and the carbon dioxide evolution rate are
used to control the dry air flow rate and the water replenishment rate.
On-line measurements invariably contain noise, which may come from either
variations in the process itself or from the measuring equipment [173]. In
general it is necessary to process the data before it can be used in control
algorithms, using mathematical filtering procedures such as Kalman filtering or
Butterworth filtering to eliminate measurement noise [163, 164, 173]. The
control schemes which control the operating variables such as inlet air temperature, flowrate, and humidity usually cannot prevent significant variations
from occurring [167]. In the 50 kg stirred capacity bed of Fernandez et al. the
inlet air temperature was controlled by electric air heaters with a selective
on/off control algorithm and the humidity was controlled by steam addition
through an on/off solenoid valve. Inlet air flowrate was set manually. Variations
of ±10% RH occurred in the inlet air relative humidity and variations of 3 °C
occurred in the inlet air temperature. Further smoothing algorithms may also
be required to account for such variations in the values of these operating
variables when processing data from on-line measurements [173].
Biochemical Engineering Aspects of Solid State Bioprocessing
121