Thesis (DTech (Electrical Engineering))--Cape Peninsula University of Technology, 2009

The problem for effective and optimal control of wastewater treabnent plants is very
important recently because of the increased requirements to the qualitY of the effluent
The activated sludge process is a type of wastewater process characterized with
complex dynamics and because of this proper control design and implementation
strategies are necessary and important for its operation. Since the early seventies, when
a major leap forward was made by the widespread introduction of dissolved oxygen
control, little progress has been made.
The most critical phase in the solution of any control problem is the modelling stage. The
primary building block of any modem control exercise is to construct and identify a
model for the system to be controlled. The existing full Activated Sludge Model 1
(ASM1) and especially University of Cape Town (UCT) models of the biological
processes in the activated sludge process, called in the thesis biological models, are
highly complex because they are characterised with a lot of variables that are difficult to
be measured on-line, complex dependencies and nonlinear interconnections between
the biological variables, many kinetic parameters that are difficult to be determined, .
different time scales for the process dynamics.
The project considers reduction of the impact of the complexity of the process model
over the methods for control design and proposes a solution to the above difficulties by
development of a reduced model with small number of variables, but still with the same
characteristics as the original full model for the purposes of real time. control design.
Then the research problem can be stated in the following way: To develop mathematical
methods, algorithms and programmes for transformation of the complex biological and
mass balance models of the activated sludge process into simpler reduced models of
the process convenient to be used for optimal parameter estimation, control design and
control implementation in real- time. The thesis chapters describe the solution of this
problem:
? Chapter 1 considers the problems for the complexity of the mathematical models
of the activated sludge process and states the thesis aim and objectives,
hypothesis, assumptions, methodology.
? Chapter 2 provides a study of the activated sludge technology and its
characteristics as an object of control. The complex relationships between the
process variables are described. The place of the problem for modelling and
parameter estimation in the plant multi-layer control hierarchy is stated.
? Chapter 3 defines a system within a state space, important criteria for process
model development and various model types are given. The wastewater
treatment biological models such as UCT biological model and Activated Sludge
model No. 1 are described. The similarities and differences between these two
models are discussed. The default values for kinetic, stoichiometric and switching
functions in the UCT model and ASM1 model are given. The processes and
processes' rates of the UCT model Peterson matrix and the ASM1 Peterson
matrix are explained in a detailed manner. Three wastewater treatment plant
structures viz., Benchmark, MitcheU's Plain and Athlone are introduced and
described. Their mass balance equations are derived for every tank and
component.
? Chapter 4 elucidates the need for reduced model development of both the ASM1
and UCT models. The assumptions for model reduction are also studied. The
dissolved oxygen concentration is considered as a control input for the reduced
models. The ASM1 and UCT reduced biological models are developed. The
mass balance equations for the three wastewater treatment plants are derived on
the basis of the reduced UCT and ASM1 biological models. The matrices and
vectors for each of the plants are separately built and the reduced model
characteristics are compared and contrasted.
? Chapter 5 gives answer to the question which parameters to be estimated on the
basis of sensitivity study of the model parameters. It examines the changes of
the model variables according to the changes in the model parameters. State
space models of sensitivity for the three considered plants are developed and
used for calculation of the sensitivity functions in Matlab environment.
Parameters with large values of the sensitivity functions are found.
? Chapter 6 states the problem for parameter estimation and proposes a method
for its solution based on time domain decomposition.The aim is to determine
values of model parameters that provide the best fit to measured data, based on
least squares criterion. Matlab software is developed and applied to solve the
problem for the considered plants for the ASM1 and UCT biological models.
? Chapter 7 gives a theoretical analysis of the problem for state and parameter
estimation. Different approaches are compared in order to find the techniques
. suitable for the activated slUdge process characteristics. Kalman filter is selected
as the most convenient one.
? Chapter 8 gives short comparative analysis of the existing simulators of the
activated sludge process. SIMBA simulator is selected and its structure is
described. Simulation of the mass balance full and reduced models of the
considered plants is provided and described. Discussion of the results is done,
? Chapter 9 investigates the Sensitivity function simulation and Parameter
Estimation for the Benchmark process and the Athlone process model based on
the reduced ASM1 and UCT biological model.
? Chapter 10 gives the conclusion and future research directions.