(174bj) Synthesizing Temperature Control System for Binary Distillation Columns
Based on the sensitivity analysis and minimum deviation methods, the synthesis and design of temperature inferential control systems for three ethanol/butanol binary distillation columns with, respectively, low, intermediate, and high product purities are addressed in this work. The controlled stages by the sensitivity analysis method usually exhibit quick dynamic responses but only in the case of low product purities can temperatures display satisfactory corresponding relationships with product compositions. For the controlled stages by the minimum deviation method, their temperatures usually display satisfactory corresponding relationships with product compositions. In the case of high product purities, because the controlled stages are quite near in locations to the ends of the distillation column, degraded process dynamics is introduced in the rectifying section and great sensitivity to pressure variations occurs in the stripping section, both worsening the performance of the temperature inferential control system. These outcomes demonstrate that neither the sensitivity analysis method nor the minimum deviation method can consistently yield the effective structure of the temperature inferential control system for a binary distillation column. Therefore, in order to achieve tight inferential control of product qualities, it is necessary to develop a new method that can compensate for changes in system characteristics. In order to compromise the static and dynamic behaviors in the rectifying section, the controlled stage should be selected between the top stage and the sensitive stage. The temperature difference between the two lowest stages in the stripping section should be used as the controlled variable to achieve the most effective compensation for pressure changes. The obtained results have confirmed that the new method not only reduces the steady-state deviation of the top product concentration, but also reduces the transient and steady-state deviations of the bottom product concentration compared with the sensitivity analysis and minimum deviation methods.