PD and PID Fuzzy Logic Controllers. Application to Neutralization Processes
Systematic methods and tools for managing the complexity
Process Control (T4-8P)
Keywords: Process control, neutralization, fuzzy logic, pH-control
pH control plays an important role in several industrial processes such as wastewater treatment, coagulation and precipitation, ore flotation, biotechnology and electrochemistry. Neutralization is subjected to many difficulties due to its inherent non linearity and its high sensitivity to small perturbations around the equivalence point. The neutralization curve depends on the acid base system and the buffering capacity changes with the type and concentration of other compounds such as strong acids and common ion salts. In addition, the streams that are being neutralized are frequently complex, unknown and of varying composition. These special characteristics lead to the great number of strategies about pH control that have been reported in the literature. So, several alternatives to classical PID controllers for pH control have been considered, including different types of linear and nonlinear models as well as advanced artificial neural networks (ANN). All these controllers are difficult to tune in the presence of lag, delay and other real factors. They also require precise information about the neutralization process and other particular data about actual factors and, usually, are effective only for small perturbations that, specially, do not involve strong buffering changes. The fuzzy logic control (FLC) has been applied to control diverse processes during the last years. The fuzzy logic can be considered as a way of non-mathematical control that avoids the lacking of flexibility of the binary-Boolean logic. It is based on some heuristic rules defined from the control workers’ experience.
The aim of this work is to compare the performance of two FLCs for the pH control in a neutralization process. One FLC is proportional derivative controller (PD) that considers the error (set point minus actual pH) and the error derivative as the two input variables of the controller. The other controller considers the pH error, pH error derivative and pH cumulative error (PID) as the input variables of the controller. For both controllers, the output variable is the valve stem position, which regulates the flow rate of the neutralizing agent stream. The controller performance is evaluated by two response specifications: settling time and overshoot. The pH response obtained during the neutralization plant start up and also under strong changes in the input stream characteristics during a normal operation. The process used to test the controller performance is the continuous neutralization of an aqueous stream of a mixture of acetic and propionic acids with an aqueous solution of sodium hydroxide.
The neutralization vessel is a continuous stirred tank reactor (CSTR). The control system is a feedback loop in which the measured variable is the pH inside the CSTR, the manipulated variable is the flow rate of the alkaline stream and the control action is based on a FLC. The PD-FLC has 5, 3 and 7 membership functions for the error, error derivative and valve stem, respectively. The PID-FLC has 7, 3, 3 and 7 membership functions for the error, error derivative, error integral and valve stem, respectively. The valve stem position after each sampling is calculated by defuzzification, using the “centroid” method. The study has been made currently by the numerical simulation of the neutralization process. In further studies the implementation in a real pilot plant will be made to confirm and improve the control system here presented. The simulation has been made by using the LabView software (National Instrument Co., version 7.1). The program contains, for each chemical compound, the mass conservation equations and the whole system electro neutrality equation; also, it contains the control and fuzzy algorithms and an interface of Virtual Instrument (VI) for a friendly use of the whole system.
The PD-FLC provides better results than the PID-FLC. When the PD-FLC is used, the settling times of the response range between 0.28 and 0.9 min, values which are lower than the values obtained with a previous ANN based controller. The overshot of the pH response to perturbations that drives to a pH decreasing is lower than 0.5 pH units. When the perturbations drive to a pH increasing, the overshoots are greater than 3 pH units. High overshoots are typically due to perturbations that move the system to states very close to the electroneutality point, which has very high gain. The system response is then very fast and the action control is not sufficiently fast to correct the deviations from the set point. Now, we are studying a PID-FLC controller in order to compensate the fast response of the system under these conditions.
See the full pdf manuscript of the abstract.
Presented Tuesday 18, 13:30 to 15:00, in session Process Control (T4-8P).
