State estimation for a biological phosphorus removal process using an asymptotic observer

This study investigated the use of an asymptotic observer for state estimation in a continuous biological phosphorus removal process. The estimated states are the concentration of heterotrophic, autotrophic and phosphorus accumulating organisms, polyphosphate, glycogen and PHA. The reaction scheme describing the process was simplified from a combined ASM1-Delft model. Three examples were investigated: operation at steady state, operation at steady state with a random white-noise in the measurements and operation with a ramp disturbance. In each case, the estimation was quite accurate even if the convergence, driven by the dilution rate, was slow (from 15 to 60 days). The propagation of the measurement noise and a bias in the estimation of glycogen and PHA could be the result of the high condition number of one of the matrices used in the algorithm of the asymptotic observer for the aerated tanks.