Drinking water treatment plants (DWTPs) face significant challenges due to anthropogenic pressure and climate change. To cope with that, DWTPs managers should adjust their treatment units. The decision-making involved in this process is complex and deals with environmental, economic and health factors (e.g. disinfection by-products (DBPs). Digitalisation of DWTPs open new pathways. Remote sensing generates huge databases that, together with the experience of process operators and managers, allow to develop mathematical models to predict water quality parameters and recommend operational set-points. Those models can be integrated in environmental decision support systems (EDSSs), which are software tools to optimise the decision making process and reduce the time in which a decision in taken.
Two case studies were selected: the full-scale DWTPs of Llobregat and Ter, which supply drinking water to Barcelona and its metropolitan area. The main challenges that have been tackled in this work include:
1. Control of pre-oxidation process – The setting the potassium permanganate rate was modelled at DWTP Llobregat using artificial neural networks (ANNs). Several ANNs and linear regression models were compared and a comprehensive methodology for parameter estimation, uncertainty and sensitivity analysis was applied. In parallel, an approach based on case-based reasoning model was developed, and both models were integrated in an EDSS.
2. Formation of DBPs - Trihalomethanes (THMs) formation models were compared and calibrated with field-scale data of Llobregat DWTP. Then, the operation of an electrodialysis reversal system was modelled and process knowledge was incorporated to assess the quality of water at two critical points of the distribution network. In a second study, a fuzzy inference system was developed for Ter DWTP. This process consisted of a sequential dose of sodium hypochlorite and chlorine dioxide. Validation at full scale during 6 months was positive 85.6% of the time.
3. Microbiological safety - A key process indicator (KPI) was developed to assess the overall plant microbiological safety. The framework for quantitative microbiological risk assessment was adapted to real-time approximating some risk-based metrics such as the disability adjusted life years (DALY). This KPI was integrated in a Supervisory Control and Data Acquisition (SCADA) system, which also alerts the users about the consequences of certain operation practices or treatment failures.
Different tools were developed to integrate the developed models into an EDSS for the validation at full-scale.
© 2001-2024 Fundación Dialnet · Todos los derechos reservados