Hence, when dissolved oxygen concentration is controlled to a
fixed value, here 25%sat, kLa can be determined from the measured
oxygen transfer rates OTR and the known saturation concentration
O*.
OTR was determined from the volume fractions of oxygen and
carbon dioxide measured in the vent line of the reactor where
the mass flow of the gas throughput was controlled by a thermal
mass flow controller. Differences between the inlet and outlet
airflow rates were neglected. The solubility O* of oxygen in the
fermentation broth was estimated to O*=5.05 mgO2/kg considering
temperature, pressure and the initial salt composition of the
fermentation medium according to.Changing
salt concentrations during the fermentation, as monitored by
means of conductivity measurements, can be considered by means
of a correction term.
Most importantly, pO2 was kept tightly at the desired constant
value of 25%sat with a controller consisting of two parts, the first
manipulating the aeration rate, the second the stirrer speed. Both
profit from the gain scheduling approach. Typically kLa varies dramatically across the fermentation runs.
Fig. 1 depicts an example from the laboratory-scale reactor used
in this work. The kLa needed to satisfy the oxygen demand of the
culture is initially quite low and increases up to values in excess
of 2500 h−1 close to the end of the cultivation. In the last stage
of the process, the oxygen demand drops, as the production of
the recombinant protein lowers the metabolic performance of
the culture. A control strategy was chosen in which kLa is first
increased by enlarging the aeration rate at a relatively low stirrer speed and then by increasing the stirrer speed in such a way that
OTR is equal to the oxygen uptake rate OUR.
Of course, many different combinations of aeration rate and
stirrer speed will lead to the same constant pO2 level in such fermentations.
This fact is used to investigate the dependency of kLa
from aeration rate and stirrer speed.
The aeration rate was modulated between 0.5 vvm up to 4 vvm
in a triangular way for the particular E.coli cultivation with a fixed
periodof 10 min.Atthe same time,the stirrer speedis automatically
adjusted by the pO2 controller in order to keep the dissolved oxygen
level at the desired set point of 25%. The modulation was started
after reaching the pO2 set point for the first time. To achieve a suf-
ficient accuracy in the pO2 control, the controller parameters were
adapted online to the total oxygen uptake rate. This scheduling
variable is well correlated with the changes in the process dynamics.
In addition, a feedforward compensator within the controller
considered changes in the aeration and feed rates.
Since the amplitude ofthe aerationratemodulations was chosen
to be quite large, the pO2 controller could not completely wipe
out all pO2 fluctuation but it can keep it within narrow limits as
shown in Fig. 2. The mean stirrer speed rose with time since the
biomass increases and its oxygen demand as well. The changes in
the stirrer speed required to keep pO2 constant show quite low
amplitudes indicating that the sensitivity of its influence on the
OTR is much larger than the corresponding sensitivity with respect
to the aeration rate.
Changing gas throughput rates at such high amplitudes also lead
to changes in the dissolved CO2 concentration and consequently,because of a modulated CO2-stripping rate, to pH fluctuations
which the pH controller must cope with. This is shown in
Fig. 3.