Nonlinear Model Reduction and Control for Integrated Process Systems

Research:

This project involves the development and evaluation of nonlinear model reduction and control methods for integrated process systems, specifically, networks of reaction and separation units with large material and/or energy recycle. Such networks typically exhibit strongly nonlinear dynamics which is induced by the recycle and evolves in a slower time scale (longer time horizon) compared to the dynamics of the individual units. This time scale separation lends itself naturally to a control paradigm which:

Distinguishes between 'distributed' control objectives for individual process units (fast time scale) and 'supervisory' control objectives for the overall network (slow time scale), and

Dictates compensating for the nonlinearities induced by recycle at the supervisory control level.

Within this framework, the main technical tasks will be the derivation of low-order nonlinear models of the core network dynamics induced by recycle structures and the design of nonlinear supervisory controllers on the basis of these models.

The methods to be developed will: (i) enhance fundamental understanding of the dynamics and control of highly integrated process systems, (ii) enable improving their operation via the design of effective supervisory control schemes, and (iii) complement existing plant-wide control methodologies and facilitate a more efficient integration of real-time optimization and supervisory control.

Impact:

This work could result in more efficient and hence more commercially viable operation of chemical plants. Process systems that could benefit from the results of this work include reaction-separation networks and individual multi-stage processes with 'internal' material recycle and/or heat integration, such as high purity and reactive distillation columns. The project also involves the development of software tools and a dedicated web site which should further enhance the dissemination of the results and the infrastructure for research and education.