
Comparison of advanced distillation control methods. Second annual report
Publication year - 1996
Language(s) - English
Resource type - Reports
DOI - 10.2172/402401
Subject(s) - selection (genetic algorithm) , sensitivity (control systems) , control (management) , fractionating column , distillation , computer science , configuration design , reliability engineering , control theory (sociology) , engineering , machine learning , artificial intelligence , chromatography , chemistry , operating system , electronic engineering
Detailed dynamic simulations of three industrial distillation columns (a propylene/propane splitter, a xylene/toluene column, and a depropanizer) have been used to study the issue of configuration selection for diagonal PI dual composition controls. ATV identification with on-line detuning was used for tuning the diagonal PI composition controllers. Each configuration was evaluated with respect to steady-state RGA values, sensitivity to feed composition changes, and open loop dynamic performance. Each configuration was tuned using setpoint changes over a wider range of operation for robustness and tested for feed composition upsets. Overall, configuration selection was shown to have a dominant effect upon control performance. Configuration analysis tools (e.g., RGA, condition number, disturbance sensitivity), were found to reject configuration choices that are obviously poor choices, but were unable to critically differentiate between the remaining viable choices. Configuration selection guidelines are given although it is demonstrated that the most reliable configuration selection approach is based upon testing the viable configurations using dynamic column simulators