Modelling a water purification process for quality monitoring

This paper deals with a quality engineering problem introduced by ‘Waterlaboratorium Noord’ (WLN) situated at the Netherlands. In-terest lies in determining an optimal sampling frequency that provides suycient information on the water quality in a drinking water purifica-tion plant. The water purification plant that is studied consists of two aeration and filtration processes and a clear water reservoir where wa-ter is saved until distribution to households. One of the main processes during these filtration processes is iron removal. A stochastic model is proposed that describes the decreasing eects on iron concentration after the filtration processes by multiplicative eects. This model is combined with an ordinary dierential equation to model the amount of iron in the clear water reservoir that fluctuates due to the quality of the incoming filtrated water and the varying water demand. In this way the iron concentration levels in the dierent compart-ments of a water purification plant can be simulated. Range and fluc-tuations approximate those of the observed data. Hence a realistic benchmark for detecting anomalies is obtained. Keywords: quality engineering, sampling frequency, stochastic model, dierential equation, water quality