Industrially relevant modeling, analysis and tuning for process control
Date of Completion
This research focuses on industrially relevant modeling, analysis and tuning for process control. Modeling includes a nonlinear model derived as a schedule of linear models. Analysis includes robustness plots and quantitative robustness metrics derived from the Bode, Revised Bode and Nyquist stability criteria. Tuning includes extensions of the Internal Model Control (IMC) tuning correlations to the proportional-integral-derivative with filter (PID with Filter) controller forms found in commercially available industrial controllers. ^ The derived robustness plots complement the widely used Bode and Nyquist plots with an intuitive visual representation of the stable and unstable regions over a range of process model parameters. Robustness plots and metrics, namely the proposed Robust Stability Factor (RSF), are derived for a general understanding of the robust stability of the IMC tuning correlations. These plots and metrics allow users of the IMC tuning correlations to specify their desired level of robust stability. A novel contribution of this work is the determination, based upon RSF values, that plant-model mismatch in the process time constant can significantly impact robust stability. ^ Various control architectures are studied for improving control of integrating (or non-self regulating) processes. Specifically, an internal cascade control structure is proposed. The tuning procedure for this structure is analogous to traditional cascade architectures. The inner controller stabilizes the process to allow tuning of the outer controller using well-understood techniques for self regulating processes. This structure is particularly beneficial because it allows the use of a conventional Smith predictor on the outer controller. Without a Smith predictor, this internal cascade structure is shown to be equivalent to a single-loop set point weighted controller. Thus, the performance of a single-loop control structure may be significantly improved by implementing either set point weighting or set point filtering.^
Arbogast, Jeffrey Elliott, "Industrially relevant modeling, analysis and tuning for process control" (2007). Doctoral Dissertations. AAI3276603.