Mathematical modelling of clean water treatment works

• Main Author: Akinmolayan, F.
• Published: University College London (University of London) 2017
• Subjects: 660
• Online Access: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.746515

One of the biggest operational risks to water companies arises from their ability to control the day-to-day management of their water treatment plants. With increasing pressures to remain competitive, companies are looking for solutions to be able to make predictions on how their treatment processes can be improved. This work focuses on mathematical modelling and optimisation of clean water treatment processes. The main motivation is to provide tools which water companies can use to predict the performance of their plants to enable better control of risks and uncertainties. Most modelling work within water operations has so far been based on empirical observations rather than on mathematically describable relationships of the process as will be considered in this work. Mathematical models are essential to describe, predict and control the complicated interactions between the different parts of the treatment process, a concept which is well understood within the process industry but not yet established within the water treatment industry. This work will also consider the level of modelling detail actually required to accurately represent a water treatment plant. This thesis develops the conceptual understanding of clean water treatment processes utilising first principles modelling techniques. The main objective of this work is the consideration of a complete mathematical model of an entire water treatment plant, which enables a wider view on how changes in one processing unit will affect the treatment process as a whole. The performance of the process models are first verified individually and are then combined to enable the simulation of a complete water treatment work. By using detailed modelling (especially gPROMS utilised in this work) requires specialist software knowledge. Without knowledge of advanced simulation tools or having a background in process modelling, the detailed models developed in this work would not be fully utilised if implemented in the water industry, if utilised at all. A systematic framework is presented for the development of simpler surrogate models that can be used to predict the effluent suspended solids concentration, for a given number of independent variables. This approach can provide valuable guidance in clean water treatment process design and operation, thus providing a tool to achieve better day-to-day performance management.