MONITORING STEADY-STATE GAINS IN LARGE PROCESS SYSTEMS

Stationary process gains are critical model parameters to determine targets in commercial MPC technologies. Consequently, important savings can be reached by acceding to an early prevention system capable of indicating when the actual process moves away from the modeled dynamics, particularly when the actual process gains are no more represented by those included in the model identified during commissioning stages, A subspace identification method is used here to determine the state-space model matrices that help to define a gain matrix estimator. The pursued goal is a monitoring method capable to identify those gains of a multivariable model that start a mismatching condition, or that show tendencies to change already known mismatches with the actual process. The anticipated knowledge of these events should prevent process engineers about the eventual reliability of targeting optimal process conditions with wrong gain estimations, and thus, help to localize the dynamic relationships for which an updating identification would be necessary.