This study is aimed to develop control algorithms which can be effectively applied to pH or neutralization processes. The rigorous model of pH process is simplified on the assumption that the feed stream can be regarded as a weak acid or a acid mixture. Three control algorithms based on the simplified model are proposed as follows : 1) The nonlinear self-tuning algorithm (NSTR) employing the recursive least-squares method and the modified one-step-ahead control law is presented. To design NSTR, the feed stream is assumed to be a hypothetical weak acid with concentration of C$_1$ and dissociation constant of K$_a$ and the process model is transformed so that it is linear with respect to C$_1$, K$_a$ and manipulated variable. 2) The model-based adaptive control algorithm (MBAC), where the controlled variable is the state variable, employing the recursive least-squares method and the proportional-integral control law is presented. To design MBAC, the process model is transformed to be linear with respect to model parameters so that the model parameters and the state variable can be estimated on-line, assuming the feed stream as a 1-basic acid mixture. 3) The adaptive control algorithm with relay feedback (RFAC) of which the controlled variable is the state varable is presented. To design RFAC, the process model is transformed so that the model parameters can be obtained from relay feedback test and the state variable can be on-line estimated, assuming feed stream as a strong acid and 1-basic acid mixture. Simulation and experimental studies have been carried out in order to investigate the control performance of the proposed algorithms and the possibility of their applications to real processes. Dynamic simulation has been done for the pH control system corresponding to real processes. As a result of simulation, the control performances of NSTR and MBAC were found to be excellent. Especially, NSTR for the weak acid model designed for a single component system has shown the robustness for a multi-component system in spite of modelling error. RFAC also performs well for a single component system. However, RFAC would be hardly applied to time-varying processes because it requires a series of relay feedback test at the instant of feed change. Experiment to test the performances of NSTR and MBAC has been conducted for the acetic acid-caustic soda system(unbuffered system) and the acetic acid-phosphoric acid-caustic soda system(buffered system). NSTR has shown good control performance for both systems. However, MBAC works well for the acetic acid-phosphoric acid-caustic soda system, but not for the acetic acid-caustic soda system around at the set point of pH 7 where process gain is extrmely large. Thus, MBAC is recommended for buffered systems and NSTR for unbuffered systems because NSTR has more tuning parameters to handle than MBAC. Since the proposed algorithms are simple and efficient in comparison with the conventional algorithms, they are expected to be effectively used in real processes.
