Control relevant model reduction: a reduced order model for ‘model IV’ fluid catalytic cracking units

Abstract

This paper extends the systematic model reduction approach of Robertson, developed for startup/shutdown operations, to regulatory/servo control applications. The ‘Model IV’ FCCU model developed by McFarlane et al. is used as a case study. To facilitate model reduction, the distributed parameter FCCU is firstly transformed into a lumped parameter representation prior to linearization. The lumping procedure presented is based on integral average properties, thereby overcoming the problem of a dramatic increase in the number of states whilst still retaining model accuracy. The subsequent linearization and the properties of the linear model are discussed. The ‘lumped’ parameter model is reduced by applying a structure preserving model reduction technique. The performance of the reduced order model, incorporated within a linear multivariable controller, was seen to lie within the preset bounds of the reference model. In contrast, the deficiency of traditional heuristic model reduction is also illustrated.