Modeling for control of fedbatch fermentation processes

1
Modeling for control of fed-batch fermentation
processes

• S. Valentinotti, C. Cannizzaro,
• M. Rhiel, D. Bonvin, U. von Stockar
• Laboratoire dAutomatique and
• Laboratoire de Génie Chimique et Biologique
• Swiss Institute of Technology, Lausanne

2
S. cerevisiae

• Well known microorganism
• Used for making wine, beer, bread etc.
• Easy to use in the lab

3
Bioreactor
X biomass G glucose E ethanol V volume
F Glucose feed rate Gin Inlet substrate
concentration
4
Reactions
(2)
(3)
5
Objective

• Maximize biomass productivity
• grow cells as fast as possible
• with the highest biomass/substrate yield

6
Bottleneck principle(Sonnleitner et al. 1986)
7
Control objective
Process objective
maximize biomass productivity
8
Simplifications
9
Simplifications

• Assumption 3 dynamics of S are fast, i.e., S is
  in
• quasi steady state

10
Two linear models
System Ethanol production (integrator)
Disturbance Biomass growth (unstable pole)
11
Simplified models
12
Disturbances

• External signals on which we have no influence
• Wind gusts airplane
• Waves boat
• Sun skin
• We wish to supress its effect
• Sun solar cremes
• Rugged terrain Suspension
  systems

13
Internal Model Principle
 Effect of disturbances on the system output can
only be eliminated if the process is operated in
closed loop and the disturbance model is
included in the controller denominator  
14
Block diagram
15
Experimental setup
16
Experimental Results (ER)
17
ER biomass
18
ER rates
19
ER RQ
20
Extensions
Metabolite control

• M could be
• oxygen
• glucose
• or

21
Conclusions

• Modeling for control
• Complex nonlinear description transformed into
• two simple linearmodels
• Models capture process characteristics
• vehicles for control design

• Controller
• Disturbance rejection via IMP
• Adaptation to different growth regimes

• Extensions
• Methodology could be used for controlling other
  metabolites
• Same control strategy used for underflow
  metabolite control

22
(No Transcript)
23
Simplifications
Assumption 2 biomass production in Reaction 1
is much higher than in Reaction 2
Since reaction 1 is saturated
Constant growth rate
Growth rate m is independent of F
24
Mass balances
25
Controller
26
extras
27
Control strategy
28
Implementation problem
Difficult to measure volume online
Assumption
Will control E (g/l) instead of VE (g)