50
Chapter 3
These three strategies (which are called “closed loop control” strategies by control engineers) are
used together with a more or less detailed mathematical model of the process, which occurs in
the bioreactor. As in any textbook treatment of stirred tank reactors it is assumed that the effluent
concentrations of biomass and metabolic products are identical to those found at any point in the
reactor. The reactor is then called an
ideal bioreactor.
Unless the steady state to be explored is
unstable all the four strategies discussed above can be used to reach a desired steady state. The
choice of strategy should ideally depend on the sensitivity of the steady state to changes in the
variable used to establish the control policy. The strategy, which gives the highest sensitivity,
should be chosen - although for practical reasons most investigations are carried out with only
one control policy to obtain the steady state at all different dilution rates investigated.
As seen in Fig. 3.2 the biomass concentration in the effluent from a steady state aerobic
fermentation with the yeast
S. cerevisiae
on glucose is virtually independent o f
D
at small values
of
D.
Consequently a steady state at a small
D
cannot be accurately fixed in a turbidostat. The
substrate concentration
s
in the effluent may well vary significantly on a relative scale (we shall
see in chapter 7 that
s
is proportional to
D
at small
D)
but it is difficult to measure the small
substrate concentration accurately enough to fix v or
D
at the desired value. Operation of the
bioreactor as a chemostat is therefore the preferred strategy. At a steady state close to the so-
called “wash-out” the substrate concentration has increased significantly and the biomass
concentration has decreased from its high and almost constant value at small
D.
Here the
turbidostat is working very well since even small variations in
D
give rise to large changes in
x,
and the steady state is pinpointed by basing the control on a given set point for
x.
The chemostat
is totally unsuited near wash out, but the pH-stat is also very satisfactory since the rate of proton
production is strongly coupled (perhaps even proportional) to the value of biomass production.
Around the so called “critical dilution rate”
Dctit
where ethanol production sets in (see Example
7.3)
the ethanol concentration
p
(or the ethanol production rate) depends strongly on
D.
Hence a
productostat is the ideal control strategy whereas both control of
x
or of the respiratory quotient,
the ratio between oxygen consumption and carbon dioxide production, are less sensitive, and the
chemostat is unsuited since one cannot control v accurately enough to obtain a steady state with a
desired
p.
It always takes patience to reach a steady state in a stirred tank continuous reactor and the time
constant for the transient between one steady state and the next varies with the steady state. It
takes on the order of five holding times, i.e., 5 • D 1, to attain a new steady state and the measured
rates can be far off their true steady state values if the approach to the new steady state is not
within 95-99%. The time between steady states is not wasted since the transient itself contains
much information on the physiology of the organism. The transient time is definitely a function
o f the control strategy used to fix the next steady state. Postma
et al.
(1989) found that around
£>CTt of an aerobic yeast fermentation on glucose the chemostat strategy had a transient time close
to 50 holding times whereas Lei (2001) obtained a steady state within 5 holding times using the
productostat control strategy.
previous page 73 Bioreaction Engineering Principles, Second Edition  read online next page 75 Bioreaction Engineering Principles, Second Edition  read online Home Toggle text on/off