Masters Theses

Date of Award

8-1987

Degree Type

Thesis

Degree Name

Master of Science

Major

Chemical Engineering

Major Professor

Duane D. Bruns

Committee Members

C. F. Moore, T. S. Wang

Abstract

The performance of three alternative multivariable control strategies are investigated. The recently developed SVAC, singular value analysis controller, and MVPIR, multivariable tuning regulator, are compared to the conventional MVSISO, multivariable single - input single - output controller. These controllers can all be characterized as requiring a minimum amount of process information and being relatively straightforward to design. The MVSISO control technique is based on the steady state process gain matrix with the manipulated and controlled variables paired according to the relative gain array analysis. Similary, the SVAC development is based only on the process gain matrix^but accounts for steady state interaction by using matrices from the singular value decomposition of the gain matrix. In contrast, the MVPIR requires the process transfer function matrix or a dynamic state space model as the design takes into account the initial transient interaction, as well as, steady state interaction. The MVSISO controller is implemented using a diagonal PI controller. The SVAC also uses a diagonal PI controller in its structure but it is pre- and post-multiplied by singular value decomposition matrices. The MVPIR uses full P-mode and l-mode matrices.

In order to evaluate the potential of the controllers, a digital simulation study is carried out using four examples based on literature transfer function models excluding their deadtime. The examples were selected to exhibit various degrees of steady state interaction. The evaluation is made in terms of their response to set point changes with their tuning parameters optimized for the lAE or the lAE with a penalty for manipulated variable movement.

The simulation studies indicate that the three control schemes under study could successfully control non-interacting systems. However, if considerable interactions exist between loops, the MVPIR has the best performance, followed by SVAC. The effectiveness of MVSISO control system is reduced as the interactions between loops increases. Contrarily, the stronger the interactions, the more useful the MVPIR.

Download
Files over 3MB may be slow to open. For best results, right-click and select "save as..."

Share

COinS