An application of adaptive pH-control algorithms based on physico-chemical modelling in a chemical waste-water treatment plant

A pH-process model for a chemical waste-water treatment plant is constructed by applying basic physico-chemical laws. The unknown composition of the wastewater is modelled by a composition of hypothetical chemical species which best describe the titration curves of real liquid samples. The dissociation parameters of the coagulant, aluminium sulphate, is also determined by using appropriate titration curves. The overall pH-model comprises a linear state model of concentrations and an implicit output model describing the physico-chemical equilibrium conditions. The estimation of the hypothetical concentrations is performed by a Kalman-type filtering algorithm obtained with the aid of a certain transformation. Using a non-linear transformation of the estimates, a deviation signal is generated and fed into the PI-controller. A feedforward control loop was connected to the algorithm and results of two simplifications of the algorithm are presented. After tuning by simulation, these algorithm variations we tested on-line in a chemical waste-water treatment plant. Comparisons of the control results are also made with the results of the permanently installed adaptive pH-controller.