Optimal operation of integrated processing systems: Part II: Closed-loop on-line optimizing control

A method for tracking the economically optimal operating conditions of a chemical process in the presence of constraints is developed. The technique is based on an on-line search rather than a fundamental model. The most profitable operating point is found by fitting a dynamic model of the process based on data obtained from experimental moves on the plant. This model is used to compute gradients of the economic objective and of the constraints so that a direction of economic improvement inside the allowed operating region of the plant is always obtained. Constraint violations during the transients are prevented by a multivariable regulator. A new regulation method (Internal Model Control) is used which permits explicit handling of constraints and which can be made robust against modelling errors. This combined optimization/regulation approach is tested in a demonstrative simulation example and shown to be reliable for following a moving optimum and safely handling complex constraint moves.