Submitted in fulfillment ofthe requirement for
Masters degree oftechnology (Mtech): Electrical engineering, 2002 / Fermentation is the process that results in the formation of alcohol or organic acids on
the basis of growth of bacteria, moulds or fungi on different nutritional media (Ahmed
et al., 1982). Fermentation process have three modes of operation i.e. batch, fed-batch
and continuous mode ofoperation. The process that interests a lot of control engineers
is the fed-batch fe=entation process (Johnson, 1989). The Fed-batch process for the
production ofyeast is considered in the study.
The considered yeast in the study is the Saccharomyces cerevisiae. It grows in both
aerobic and anaerobic environmental conditions with maximum product in the aerobic
conditions, also at high concentration of glucose (Njodzi, 2001). Complexity of fedbatch
fe=entation process, non-linearity, time varying characteristics, application of
conventional analogue controllers provides poor control due to problems in tuning
individual loops and the process characteristics. The problem for control of the fedbatch
process for the production of yeast is further complicated by the lack of on-line
sensors, lack of adequate models as a result of poorly understood dynamics. The lack
of on-line sensors results in the impossibility of tuning the analogue controllers in real
time. The process for propagation of yeast in aerobic conditions is considered in the
dissertation. The experiments are conducted at the University of Cape Town (VCT),
Department of Chemical Engineering with a bioreactor and bio-controller are
combined in a Biostat ® C lab scale plant (B. Braun Biotech International, 1996).
The bio-controller has built in PID controller loops for control variables, with the
ability to adjust the controller parameters i.e. P, D and I through the serial interface
(Seidler, 1996). Even though the used lab scale bio-controller has the ability to
monitor certain variables, the automation of the industrial bioreactors is still
developing slowly (Dochan and Bastin, 1990) with major problems experienced in
modelling and measuring important control variables on-line. This existing situation is
due to the characteristics of the fermentation processes as an object of control with
highly non-linear, non-stationary, slow dynamics and complex relationships between
variables. The existing control strategies on industry are based only of local PID
control of some easy for measuring variables. No computer systems for monitoring
and optimisation of the process (Morari and Stephanopoulos, 1980).
The dissertation is overcoming the mentioned above drawbacks by developing
methods, algorithms and programmes for building of a two layer system for optimal
control of the Biostat ® C pilot plant with the following subsystems:
~ Data acquisition,
~ Modelling and simulation,
~ Model parameter estimation,
~ Process optimisation,
~ PlO controller parameter tuning,
~ Real time control implementation
The system is based on LabVIEW™ and serial communication protocol. The interface
between the Bio-start and the host computer is through a standard communication
serial port. The development in the dissertation are described as follows:
Chapter 1 describes the necessity of the research discussed in the dissertation and
highlights comparison between the different approaches for modelling and control of
fed-batch processes for the production of yeast, (Johansson, 1993). The aim and the
objectives ofthe dissertation are stated and explained.
Chapter 2 describes the process as an object of control looking precisely at the
influence of the physiochemical variables on the biological variables. The relationship
is identified through the enzymes. The results from previous experiments are
discussed to illustrate the constraints associated with the control of the process under
study
Chapter 3 describes the different types of models as applicable to the dissertation. The
comparison between different types is highlighted. The derivation of the developed
yeast model using mass balance equations and rate laws is discussed and presented in
the chapter. The problem for simulation of the model is solved using Matlab and
LabVIEW™ programs.
Chapter 4 describes the formulation of the problem for estimation of the fed-batch
model coefficients and the method, algorithm and programme developed to solve this
problem.
In chapter 5 the optimal control problem is formulated and solved using optimal
control theory, the approach of the functional of Lagrange is used. The optimisation
layer problems are determined and based on the solutions of the previous upper layer
i.e. the model parameters from the adaptation layer. The optimal operation of the
process or yield of the yeast is based on some criteria for the production of biomass,
and some constraints over minimal and maximal values of the variables.
Decomposition method to solve the optimal control problem is developed on the bases
of an augmented functional of Lagrange and decomposition in time domain.
Algorithm of the method and program in Matlab are developed. Tuning of PID
parameters for the controllers in Biostat ® C is described based on the optimal
trajectories for the physiochemical variables obtained from the optimal control
problem solution.
Chapter 6 presents algorithms and programmes for monitoring and real time control
of fed-batch fermentation process for the production of yeast, using Personal
Computer, B Braun Biostat ® C Lab scale fermentation unit and LabVIEW™ as
driver software. Hardware and software parts of the control systems are described and
discussed. LabVIEW™ code is described.
Chapter 7 presents the users manual. The mam functionality of the developed
application and programmes is described and discussed.
The source code ofthe developed programmes is presented in chapter 8.
Chapter 9 presents the conclusion highlighting the developments in the dissertation as
well as the future work on the topic and the possible application of the developed
work in industry on the bigger scale fermentors.
The positive characteristics of the developed methods algorithms and programmes
are:
-7 The developed model incorporates the physiochemical variables in the
biological mass balance equations. In this way: the influence of the
enzymes over the biological variables is utilised and possibilities for
process optimisation is created.
The process can be optimised m both physiochemical and biological
variables.
The physiochemical variables can be used as control inputs to reach the
process optimisation.
Data acquisition system gIves good possibilities for analysis and
simulation ofthe process.
The optimal control of the process is achieved without using expensive online
sensors for measurement of the biological variables. This is why the
developed system is applicable to the existing hardware and software
control and measurement systems in industry.
The control system automates the operation of the lab scale fermentation
unit. It is safe, stable and operational.
| Identifer | oai:union.ndltd.org:netd.ac.za/oai:union.ndltd.org:cput/oai:localhost:20.500.11838/1169 |
| Date | January 2002 |
| Creators | Mkondweni, Ncedo S |
| Publisher | Peninsula Technikon |
| Source Sets | South African National ETD Portal |
| Language | English |
| Detected Language | English |
| Type | Thesis |
| Rights | http://creativecommons.org/licenses/by-nc-sa/3.0/za/ |
Page generated in 0.0029 seconds