This NSF GOALI project with Praxair and Aspen Technology is aimed at developing nonlinear modeling and control technology for cryogenic air separation plants. We have proposed a low-order dynamic modeling framework for cryogenic distillation columns using nonlinear wave theory. Essential column dynamics are modeled by tracking the composition wave front propagating through the column. The wave model is coupled to a detailed model of the integrated reboiler/condenser system derived from mass and energy balances. We have developed state and parameter estimation schemes that allow the wave model to be incorporated within nonlinear model-based control strategies. An extended Kalman filter (EKF) that generates estimates of the wave position and key parameters allows the wave model to be adjusted on-line to match the concentration profile produced by the Aspen simulator. A nonlinear model predictive controller that utilizes the EKF estimates outperforms the PID-based regulatory control strategy for simulated changes in the overhead nitrogen production rate.
Our more recent work has focused on the development of stage-by-stage balance models and reduced-order approximations for nonlinear model predictive control. Detailed models of the lower and upper columns of a double column air separation plant compare favorably to first-principles models developed with the commercial simulator Aspen Dynamics (Aspen). We have used these fundamental models to derive reduced-order models based on column compartmentalization and singular perturbation analysis. Our current work is focused on the incorporation of fundamental and reduced-order compartmental models into nonlinear model predictive controllers that provide dual nitrogen and oxygen composition control for the upper column.
Funding: National Science Foundation (CTS-0241211), Praxair and Aspen Technology
Students: Shoujun Bian (Ph.D.), Suabtragool Khowinij (M.S.) and Zhongzhou Chen (Post-Doc)
Collaborator: Drs. Lawrence Megan and Paul Belanger (Praxair)
Publications: