A decision support system (DSS) for control of the coagulant dosage at the Görväln drinking water plant has been developed and implemented. The goal with the DSS was to enable the transition from manual to automatic control of coagulant dosage. The DSS is based on a multivariate statistical regression (PLS) model mimicking the operators’ manual dosage of coagulant and is based solely on UV-absorbance, colour, COD, TOC and conductivity in the raw water. By external validation with two years of historical data, the model was proven to provide a good estimation of the manual dosage. When the model was implemented for dosage control, the variation of the quality of the treated drinking water was significantly reduced as a result of quicker and correct response to changes in the raw water and at the same time the coagulant consumption was maintained. The results pave the way for future optimization of the coagulant dose, resulting in reduced coagulant consumption while still maintaining or even increasing the drinking water quality. The approach presented is expected to be able to give positive results on other drinking water plants as well.
The article describes the development and the effect of implementation of a mathematical model for control of the coagulant dosage at Görväln drinking water plant.