Abstract
Temperature control is widely perceived to be superior to direct composition control for the control of dividing-wall distillation columns (DWDCs) due to its advantages in dynamic characteristics. However, because of the limited estimation accuracy to the controlled product purities, the former cannot eliminate the steady-state errors in the maintained product purities as completely as the latter. In order to reduce the steady-state deviations in the maintained product purities, an effective temperature control method is proposed in the current article by means of a kind of simple but effective product quality estimator (PQE). For the proposed PQE, temperatures of three stages located in the controlled column section (TI1, TI2, and TI3) are employed as inputs, and a linear sum of these three inputted stage temperatures (α × TI1 + β × TI2 + γ × TI3) is given as output. A genetic algorithm with an elitist preservation strategy is used to optimize the locations of the three stage temperatures and the values of α, β, and γ to ensure the estimation accuracy of the PQE. Concerning the controls of two DWDCs, i.e., one Petlyuk DWDC separating an ethanol/propanol/butanol ternary mixture and one Kaibel DWDC separating a methanol/ethanol/propanol/butanol quaternary mixture, the effectiveness of the PQE is assessed through comparing the performance of the temperature inferential control scheme using the PQE and the double temperature difference control scheme. According to the dynamic simulation results obtained, the former control scheme displays not only smaller steady-state deviations in the maintained product purities, but also better dynamic characteristics as compared with the latter control scheme. This result fully demonstrates that the proposed PQE can be a useful tool for the temperature inferential control of the DWDC.