Blind identification of stiction in nonlinear process control loops

Stiction in control loops lead to oscillations and loss of productivity. Timely detection of stiction in control valves can be used in scheduling valve maintenance and deployment of compensation techniques to reduce their impact. Several approaches have been proposed for detection of stiction. Data-based approaches use unique shapes of the PV and OP data to identify stiction. Approaches based on nonlinearity detection have also been used to identify stiction. Yet another approach uses a Hammerstein model structure identification for detecting stiction. Most of these approaches are restricted to linear processes. In this paper, possible approaches to detect stiction in nonlinear process control loops are discussed.