Process Modelling and Control

Group focus

In order to be a world leader, the group focuses its research and industrial activities on those processes that make South Africa unique in the world and which can generate results of interest to the rest of the world. These include:

  • dynamics and control of very large gas circuits, such as the SASOL coal-from-gas process;

  • processing of indigenous woods in the paper and pulp industry;

  • process development in mining and minerals processing.

We are proud to be associated with the following industrial partners:

  • The SASOL group of companies – a world leader in coal-to-gas technology;

  • SAPPI-SAICCOR – producing 25% of the world's dissolved pulp;

  • MINTEK – developing sophisticated control systems for minerals processing unit operations;

  • CSIR – designing unique processes to add value to our mineral resources;

  • AMPLATS – producing precious metals.


Students can specialize in any of two areas:

  • process control;
  • process design.

See the Postgraduate information page for more information

Active projects as of 2019

Evert: A dynamic timeseries analysis environment

The Evert project is our platform for timeseries analysis of plant data. The eventual goal is to have all our algorithms operating from within this web-based environment.

Stochastic dynamical control within the context of switching probabilistic graphical models

The group is working on a novel formulation of stochastic dynamic control. The Master's report "Computationally efficient formulation of stochastic dynamical control within the context of switching probabilistic graphical models" documents the first work.

Network centrality analysis for fault finding

The group has expanded on the Google Pagerank algorithm and is exploring ways in which network centrality measures can be used to improve fault finding on plants.

Control of batch digesters used for the manufacturing of dissolved pulp

Controlling the degree of polymerization during the digestion process is extremely difficult, since no online measurement technique is available. We are using sophisticated modelling techniques like neural networks to infer this property and to control the process to produce at a consistently high quality. This project has enjoyed funding from Sappi, PITECH, and the THRIP-initiative of the NRF.

Amoss: Stochastic simulation of Sasol value chains

A simulator is being developed to simulate large-scale processes like Sasol's value chains using stochastic elements. This work has been presented at the Simul 2018 conference in Nice, France.

Stochastic simulation of chemical engineering systems in Modelica

Modelica is rapidly becoming an entrenched industry standard. We are developing blocks for Modelica which allow for simulation of chemical engineering systems. The system also includes stochastic elements.

Historical projects

Development of a Model Predictive Control software package

Model Predictive Control (MPC) techniques have proved to solve many complex control problems in the chemical process industries, where processes are highly nonlinear, contain considerable dead-time and have to operate within a constrained environment. We are developing a control package, which will be affordable to small- and medium-sized companies that often experience a need for sophisticated control, but cannot motivate the installation of expensive hard- and software systems.

Advanced Milling Control

In a highly competitive market, milling control needs to be highly optimized and above all, operator-friendly. Developing advanced milling control software to conform to the specific requirements of widely varying milling processes is the aim of this project. This project was run with the support of MINTEK.

Abnormal Situation Management (ASM)

The ability to detect the development of a possible crisis situation in a processing environment (and how to deal with it) is currently receiving a lot of attention in the world. Should it be possible to do this, substantial savings can be achieved on large chemical plants. Using state-of-the-art wavelet technology, we are working towards a solution. This project was run with the support and collaboration of SASTECH.

Modeling gas-solid reactions in rotary kilns

Gas-solid mass transfer in rotary kilns has traditionally been modelled as a mass transfer coefficient on the surface of the bed – the so-called "active layer". In this research, the entrainment of gas in the voids of the particles as they move through the kiln is taken into account, resulting in a novel and fundamental approach to the modelling of rotary kilns.

Exergy analysis of a cement manufacturing plant

The first law of thermodynamics neglects the degradation of energy quality and cannot give any clues as to whether energy is being used efficiently. In exergy analysis the first law is combined with the second law of thermodynamics to determine how efficient a process really is.

This powerful tool can be used to:

  • evaluate and compare various process alternatives in the conceptual design stage;
  • to identify unit operations in an existing process where modifications will be the most cost-effective.

Since cement production is an energy-intensive operation it is an ideal candidate for exergy analysis. This could lead to process improvements that would increase the exergetic efficiency. An increase in exergetic efficiency decreases the demand for fuels and therefore puts less pressure on nonrenewable resources.

The purpose of this project was to apply an exergy balance to the PPC Hercules plant in Pretoria. The project is done in collaboration with the Eindhoven University of Technology in The Netherlands.

Robust control of a triple-effect evaporator

The designed controller for this system must be able to control the process over its entire operating range. Given that a linearized model is used for controller design and that a linearized model is dependent on the operating point, it becomes necessary to consider the robustness of the controller. Robustness deals with the ability of a controller to still yield the desired output, despite mismatch between the model and actual process. Techniques are available to test the robustness of controllers. These techniques were applied to this case study, which was based on a biotechnological process.

The integration of modeling and control know-how is achieved by a thorough understanding of the behaviour of components in mixtures, unit operations and process dynamics. An excellent opportunity exists within an educational and research environment to develop and continuously improve tools to extend our understanding of such activities. The MATLAB/SIMULINK-environment provides the means to link these together into a useful set of tools that can be used to test plant-wide control strategies and process behaviour.

Published by Carl Sandrock

Copyright © University of Pretoria 2020. All rights reserved.

FAQ's Email Us Virtual Campus Share Cookie Preferences