| This thesis presents in a complete and detailed manner the modeling, simulation, identification and control of pH processes. The model is strictly based on the physical balance equations of mass and charge, and it is proven that such a model can be reduced to a minimal realization based on monoprotic equivalent substances, which makes the influent states observable. Identification is done in two different ways: identification of the concentrations using the Nonnegative Least Squares (NNLS) method when the influent components are known, and identification of the concentrations and dissociation constants using the Constrained Nonlinear Optimization algorithm in cascade with the NNLS when there is insufficient information about the influent stream. If the influent does not change or changes occur slowly, an Inline arrangement is proposed as a way of reducing cost. When influent changes are important a small CSTR is proposed as the best way to achieve reliable control. These identification approaches are implemented on-line, so that the model can be quickly updated to a new model resulting from changes in the number or concentration of influent components. Identification provides the inverse of the static nonlinear pH mapping, such a model is used in the feedback path to linearize the system, and a predictive control is used to regulate pH in the resulting nonlinear dynamic system. Parameters in the controller allow the adjustment of the receding horizon, the maximum actuator variation per sample, etc. Control and Identification were implemented using BridgeVIEW 2.0 and MATLAB in two Pentium II computers, experiments were done using a modified version of a LabVolt Analytic Station. |
