Comprehensive methodology for detection and diagnosis of oscillatory control loops

In recent years, significant research has been performed on control loop oscillation diagnosis; there are still many unresolved issues such as indecisive or wrong diagnostics for certain regions of process dynamics, failure in the case of multiple faults occurring simultaneously and, absence of quantification of root-cause. In this work, a novel methodology has been devised to bridge these gaps in oscillation diagnosis. Power Spectral Density (PSD) and Auto Correlation Function (ACF) based oscillation detection method followed by a model based approach for identifying and quantifying the root-cause of the oscillations is proposed. The application of each component of the algorithm is described with appropriate example cases. Finally, overall methodology is validated on industrial and simulation case studies for various cases of stiction, poor controller tuning, and external disturbances. (C) 2009 Elsevier Ltd. All rights reserved.