Stiction identification in nonlinear process control loops

Nearly 20–30% of all process control loops oscillate due to stiction resulting in productivity losses. Thus, detection and quantification of stiction in control valves using routine operating data is an important component of any automated controller performance monitoring application. Many techniques have been proposed for the detection and quantification of stiction. However, most of the approaches assume that the underlying process is linear; very little work is available for nonlinear processes. In this paper, Volterra model-based technique is investigated for the detection of stiction in closed-loop nonlinear systems. The advantages of the proposed method are: (i) it can be used to detect stiction in nonlinear systems and (ii) requires no prior information on whether the loop is linear or nonlinear. Results obtained from simulation and industrial case studies demonstrate the utility of the proposed methodology.