An improved qualitative shape analysis technique for automatic detection of valve stiction in flow control loops

Valve stiction is a frequent cause of loop performance deterioration and in recent literature many techniques have been proposed for its automatic detection. Yamashita [(2006). An automatic method for detection of valve stiction in process control loops. Control Engineering Practice, 14, 503–510] proposes a method based on qualitative shape analysis, which is very effective for flow control loops where the valve position (or the proportional flow rate) is available as measurement.However, additional stiction patterns observed in industrial data are not included in the original algorithm. This paper provides an analysis and explanation of these patterns on the basis of valve type, loop dynamics and control system setup, by means of simulation and comparison with industrial data. The technique is then extended and applied to a plant data set in order to show the improved efficiency in detecting stiction.