Computer Control of pH Reactor

Abstract

Physical inspection leads to the modelling of a pH-reactor by a linear, dynamic flow and mixing process followed by a static nonlinearity (Wiener model). Except in the case of a simple liquid composition, the process variables or parameters have to be estimated, on the basis of measured pH data. A process set-up provided with instrumentation for computer control is used, in order to test control algorithms with a process which was known hard to control. In one algorithm, the flow and mixing model is considered known and the chemical compounds, up to four in number, including their dissociation constants are likewise known. Their concentrations are estimated by means of a Kalman filter, from pH and flow measurements. The estimates are used for construction of the error signal for a PI-controller in such a way that the closed loop becomes linear which also requires a use of the static model of the chemical equilibrium. Pilot tests have shown that this algorithm gives far better responses than the conventional feedback control. The other algorithm uses an inverse overall process model obtained by combination of the flow and mixing model with a piecewise-polynomial approximation of the static nonlinearity. This model is applied to the development of a nonlinear controller based on the model reference adaptive technique. The method is applicable for tracking of a given pH variable and for regulation around a fixed reference value without prior knowledge of the process components or process dynamics.