(371m) On the Detection of Valve Non-Linearities in Closed Loop Linear Systems

Authors: 
Srinivasan, B., Columbia University
Nallasivam, U., Clarkson University
Rengaswamy, R., Texas Tech University


Oscillations in closed-loop systems lead to poor product quality, degradation in productivity and huge profit loss. The presence of sticky valves in control loops is a major cause for oscillation in closed-loop systems. Therefore, it is essential to develop methods for identification of stiction and compensation techniques to reduce the impact of stiction in control valves. It is well known that presence of stiction in control valves introduces nonlinearities in the closed-loop system. Several approaches utilize this information to detect stiction in linear closed-loop systems. Recently, there is a growing interest to develop methods that can identify stiction in closed-loop systems with a nonlinear process. However, a thorough analysis on: (i) the conditions under which stiction can be identified in linear closed-loop systems and (ii) the applicability of stiction detection techniques to nonlinear systems is not available. In this work, frequency domain analysis of closed-loop systems is performed from which conditions for stiction detection in linear and nonlinear processes are derived. Results from simulation and industrial case studies support the arguments presented in this article.