Implementing an Extended Kalman Filter for estimating nutrient composition in a sequential batch MBBR pilot plant

Abstract

Online monitoring of water quality parameters can provide better control over various operations in wastewater treatment plants. However, a lack of physical online sensors, the high price of the available online water-quality analyzers, and the need for regular maintenance and calibration prevent frequent use of online monitoring. Soft-sensors are viable alternatives, with advantages in terms of price and flexibility in operation. As an example, this work presents the development, tuning, implementation, and validation of an Extended Kalman Filter (EKF) on a grey-box model to estimate the concentration of volatile fatty acids (VFA), soluble phosphates (PO4-P), ammonia nitrogen (NH4-N) and nitrate nitrogen (NO3-N) using simple and inexpensive sensors such as pH and dissolved oxygen (DO). The EKF is implemented in a sequential batch moving bed biofilm reactor (MBBR) pilot scale unit used for biological phosphorus removal from municipal wastewater. The grey-box model, used for soft sensing, was constructed by fitting the kinetic data from the pilot plant to a reduced order version of ASM2d model. The EKF is successfully validated against the standard laboratory measurements, which confirms its ability to estimate various states during the continuous operation of the pilot plant.

Implementing an Extended Kalman Filter for estimating nutrient composition in a sequential batch MBBR pilot plant