Fuzzy Control of Continuous Stirred Tank Reactor

Continuous Stirred Tank Reactor (CSTR) involves complex reactions, highly nonlinear and very hard to control by conventional methods. Fuzzy controller is introduced in this research to control the CSTR. A powerful design and simulation tool which is MATLAB and SIMULINK has been used for evaluation and test. In this research, the process considered is the decomposition of hydrogen peroxide (H2O2) using Fe3+ as the catalyst. The product is water (H20) and dissipated oxygen (O2). The main task is to maintain the temperature and concentration in the CSTR around the working point, in spite of disturbances by manipulating the flowrate of inlet or outlet stream and coolant temperature. Two fuzzy controllers have been designed and tested based on Mamdani and Sugeno inference mechanisms. Performance comparisons have been made; first between the two controllers and second between the best fuzzy controller and previous researches on the similar topic but using different techniques. The overall performance of both Mamdani and Sugeno based fuzzy controller are excellent. The tests results showed that Sugeno and Mamdani based fuzzy controllers have no significant difference in performances due to slow process occurred in the CSTR. However, Sugeno based fuzzy controller slightly improves the response time of Mamdani based fuzzy controller by 20%. Steady state error is also reduced to less than 1%. Furthermore, the simulation time is also reduced. Fuzzy control provides a simple technique but yet the achievements are comparable to other more complicated and time consuming techniques. Various tests have been conducted throughout this research to show the robustness of both fuzzy controllers