Figure 8: Principle of a flow
rate gradient for optimization of productivity and resin utilization.
(A) set-up of sensors in the chromatographic workstation (B) change of
flow rate vs time or volume loaded onto the column and breakthrough
curves at low and high residence time blue line high residence time,
redline low residence time, (C) resin utilization productivity and yield
of an unoptimized process and a process with model predictive control,
data from (Eslami et al., 2022a; Eslami et al., 2022b).
The system can be also used for continuous chromatography. Then two
columns are operated in tandem, while one column is loaded and the
second column is washed, eluted and regenerated. If the wash, elution,
and regeneration cycle is longer than loading then four columns must be
installed. The selection of column number follows the same rules as for
counter current chromatography. Furthermore, the velocity gradient
approach can be extended to wash, elution and regeneration and thereby
even higher productivity could be obtained. During loading in counter
current chromatography two columns are interconnected and therefore the
maximal system pressure and flowrate is constrained by this situation.
This is not the case when columns are operated in parallel with velocity
gradients during loading.
How fully integrated continuous biomanufacturing may look like was been
previously proposed by the Morbidelli group (Feidl et al., 2020). They
propose a process wide control of the integrated process. The unit
operations are individually controlled and process wide by a supervisory
system. The challenge is to integrate all unit operations and to
maintain a constant mass flow and to avoid propagation of process errors
or deviations. Therefore, always surge tanks between unit operations are
placed to dampen fluctuations in flow and pressure and also might be
used as intermedia storage when a unit operation stops working.
Fluctuations flow rate and concentrations are inherent when a
conventional batch manufacturing process is rendered into a continuous.
Then preferably perfusion culture is connected to a counter current
chromatography. Constants harvest conditions in perfusion culture could
be only expected if be fully growth arrested. This is not the case and
therefore the excess of cells must be removed from the bioreactor by a
so-called bleeding leading to a change in harvest flow rate. The
concentration is simultaneously changing. Therefore without proper
control these fluctuations would propagate through the entire process
and eventually amplify and lead to a derailed process. Model predictive
control ensures either constant harvest flow or concentration
(Pappenreiter et al., 2022).
Conventional feedback control is not sufficient, because it is designed
for rapid changes in process conditions, but in perfusion cell culture
conditions change slowly (Zhao et al., 2015) . Then harvest is sent to a
surge thank and then captured by a counter current chromatography. The
loading flow rate is constant but elution after the columns is saturated
leads to a periodically change of mass flow and in order to warrant a
constant contact time in the virus inactivation the elution fractions
are pooled and then very often subjected to a batch virus inactivation,
although many continuous virus inactivation methods have been
established. By following such a process design the benefits of
continuous biomanufacturing are reduced in part automation and on-line
control and real-time release become very difficult to implement. Surge
tanks have a deteriorating effect RTD and wide RTD makes a process slow
and batch definition difficult(Lali et al., 2022). In an ideal process
the mass flow is never interrupted, and surge tanks are avoided. The
perfusion bioreactor produces a constant harvest flow and can be further
processed by flocculation and precipitation and polished with flow
through chromatography methods. After concentration adjustment by e.g.,
single pass tangential flow filtration a continuous virus filtration can
be added as recently suggested (Figure 9).