User talk:AndreC
SAIChE Postgraduate Student Symposium Abstract
Abstract for Masters project up for discussion/comments.
Department of Chemical Engineering, University of Pretoria
The models used in multivariable predictive controllers (MPCs) are, typically, mathematically unconstrained. Certain MPC algorithms allow constraints to be imposed on the inputs or outputs of a system, but do so in a manner that is decoupled from the process model. This ad hoc specification can lead to infeasible and unrealistic constraints being specified for controllers. A method to determine and handle input/output constraints, taking their inherent interdependence into account, would therefore be an attractive addition to commercial MPC packages.
A rigorous and systematic approach to constraint management was developed, using the operability index (Vinson & Georgakis, 2000) as the basis for the design. The method supports linear and non-linear (polynomial) steady-state systems, and provides an interactive interface where the following actions can be taken and information obtained;
- effects of constraint changes on the corresponding and other input/output constraints,
- feasibility checks for given constraints,
- constraint type information,
- specification of constraint-set size (number of constraints to use) and
- optimal fitting of output constraints within the desirable output space.
Focus was placed on making the method rigorous with regards to the use of the process model and the clear distinction between constraint types. Ample feedback to the user was added to the interface to increase interactivity.
It was ensured that the outputs of the program were compatible with commercial MPC packages, such as Honeywell’s RMPCT® and AspenTech’s DMCPlus®. The aforementioned packages were used in conjunction with the developed software to test functionality and performance of the method. The method was applied to case studies from Anglo Platinum, the Tennessee Eastman sample problem and laboratory scale test rigs.
