Dynamics and control designs for internal thermally coupled distillation columns with different purities, Part 2: close loop analysis and control designs

Interaction between overhead and bottom composition loops of internal thermally coupled distillation columns (ITCDIC) with four purities are analyzed by means of relative gain array, condition number and singular value decomposition, which indicates that high- and very high-purity ITCDIC are well coupled and ill-conditioned systems with severe gain directionality problem. Based on the above dynamic behaviors, suitable control schemes of four different purity systems are studied. In low-purity system, internal model control (IMC) is better than decentralized proportion integral differential (D-PID) control in terms of response speed and remaining errors. In moderate-purity system, D-PID can effectively reject large load disturbances while IMC fails due to severe mismatch between model and plant. In high-purity system, neither IMC nor D-PID can provide satisfactory control so that modified IMC and multivariable PID with singular value decomposition scheme are presented respectively to improve control performances. Finally, in very high-purity systems, the combined feed-forward and feedback control scheme is proposed to deal with extremely sluggish responses to load disturbances.

Dynamics and control designs for internal thermally coupled distillation columns with different purities, Part 2: close loop analysis and control designs

Interaction between overhead and bottom composition loops of internal thermally coupled distillation columns (ITCDIC) with four purities are analyzed by means of relative gain array, condition number and singular value decomposition, which indicates that high- and very high-purity ITCDIC are well coupled and ill-conditioned systems with severe gain directionality problem. Based on the above dynamic behaviors, suitable control schemes of four different purity systems are studied. In low-purity system, internal model control (IMC) is better than decentralized proportion integral differential (D-PID) control in terms of response speed and remaining errors. In moderate-purity system, D-PID can effectively reject large load disturbances while IMC fails due to severe mismatch between model and plant. In high-purity system, neither IMC nor D-PID can provide satisfactory control so that modified IMC and multivariable PID with singular value decomposition scheme are presented respectively to improve control performances. Finally, in very high-purity systems, the combined feed-forward and feedback control scheme is proposed to deal with extremely sluggish responses to load disturbances.