Configuring Effectively Double Temperature Difference Control Schemes for Distillation Columns

The conventional method for configuring double temperature difference control (DTDC) schemes relies on sensitivity analysis (SA) and singular value decomposition (SVD) analysis, respectively, to determine sensitive and reference stages. Since no considerations are given at all to the interactions between the synthesized double temperature differences (DTDs) and to the coordination between the upper and lower temperature differences (TDs) subtracted in each synthesized DTD, the conventional method may lead to DTDC schemes that fail to secure tight product quality control of distillation columns. In this article, a novel method is proposed that employs a newly defined metric in our recent work, the averaged absolute variation magnitudes (ASVM), to determine the two reference stages in each DTDC loop. The ASVM measures the variations of TDs between the sensitive stage and the remaining ones with the assumption of complete rejection of all disturbances concerned and can thus reflect the inherent characteristics of coupling between the controlled product qualities. For each DTDC loop, while the first reference stage should be chosen to cope with its coupling with the other control loops, the second reference stage should be to coordinate the two TDs involved, thereby yielding a favorable effect to the inference of the controlled product qualities. Four example systems, including one conventional distillation column separating a binary mixture of ethanol and butanol, two conventional distillation columns separating a ternary mixture of ethanol, propanol, and butanol, and one dividing-wall distillation column separating a ternary mixture of ethanol, propanol, and butanol, are used to assess the proposed method by means of in-depth comparison with the conventional method. While they display comparable dynamic performances, the former leads to considerably smaller steady-state deviations in the controlled product qualities than the latter. These striking outcomes demonstrate evidently that the proposed method can be a promising alternative for the pursuit of tight temperature inferential control of various distillation columns.