This NSF GOALI project is a joint effort between the University of Massachusetts/Amherst and Praxair aimed at the development and implementation of nonlinear model predictive control (NMPC) technology for cryogenic air separation plants. The planned methodology involves the construction of low-order distillation column models based on nonlinear wave theory. The wave modeling approach will be applied to the problem of triple column plant used to separate nitrogen, oxygen and argon. The low-order models will be compared to a rigorous cryogenic plant simulator developed using the Aspen Technology suite of dynamic simulation tools. The nonlinear wave-modeling framework will be used to develop NMPC technology specifically designed for cryogenic air separation plants. The planned control strategy will be evaluated utilizing the Aspen simulator as a surrogate for an actual cryogenic separation plant. The project will culminate with the implementation of the triple column NMPC controller at the Praxair Laporte, Texas plant site.
Cryogenic gas separation plants produce large quantities of liquid and/or gaseous products for the chemical and steel manufacturing industries. Domestic consumption of electricity by industrial gas producers is estimated to be over $700 million per year. Continued deregulation of the electric utility industry will lead to more frequent and unpredictable changes in electricity costs. This will dictate fundamental changes in the operating philosophy of gas separation plants. Large production rate changes and more frequent startups/shutdowns will be required to take full advantage of time-varying utility costs. Consequently, small improvements in automatic process control have the potential for substantial economic benefits. Preliminary economic analysis for a representative plant indicates that the proposed control technology has the potential to recover $120,000/year of products lost during startups. The research will provide Praxair with a control system prototype for the 60 cryogenic air separation plants in the United States and will provide impetus for other domestic gas suppliers to implement nonlinear control technology.
