Automatic Control of Bioprocesses - Measurement, Monitoring, Modelling, and Control of Bioprocesses - page 38

employed to control cultivation processes. In order to realize closed-loop control,
reliable system measurements are vital. However, the application of closed-loop
control is still rare, due to many reasons:
In many cases important process variables can be determined online only with
excessive effort. They become available delayed by a dead time as well as lag
elements and also discontinuously. Most bioprocesses are batch or fed-batch
processes; therefore, one has to deal with a transient (not stationary) process,
where the automation task is to provide an optimal environment for the micro-
organism. The typical goals of automation of bioprocesses are to:

• Compensate failure of any kind
• Minimize energy and raw materials
• Maximize yield and product quality
• Guarantee safe operation
• Prevent substrate, overflow metabolite, or product inhibition
• Ensure well-directed induction and repression of enzyme production
• Prevent high shear stress
• Present an optimal environment for the organism for growth as well as

production

With the help of standard control algorithms, some of these goals can already
be achieved. Basic bioreactor equipment often includes control algorithms for the
volume, temperature, pH, dissolved oxygen, and addition of antifoam agents.
However, these basic controllers are not always sufficient for special applications.
In this chapter, state-of-the-art bioprocess automation and recent progress are
discussed. An overview of the discussed application is presented in Table 1 .

2 Controller Design

2.1 Direct Measurements

Especially for closed-loop control purposes, measurements are fundamental. For
bioprocesses, in situ measurements such as temperature, pH, dissolved oxygen
concentration (DO), optical density, and pressure and at-line measurements such
as the exhaust gas composition are performed most frequently and can be used as
input variables for a controller [ 1 , 2 ]. At-line measurements based on spectro-
photometric, mass-spectrometric, HPLC, GC, and flow injection analysis (FIA)
systems are applied less frequently for online measurements and even less often as
input variables for a controller [ 1 - 3 ]. Due to the fact that direct measurements of
important variables such as growth rate, substrate uptake rate, and carbon dioxide
production rate are missing, soft sensors have been established to enable a kind of
indirect measurement that can provide access to relevant variables using different
techniques.