Eliminating Valve Stiction Nonlinearities for Control Performance Assessment

Abstract

Control performance assessment or CPA is a useful tool to establish the quality of industrial feedback control loops, but this requires establishing the minimum variance lower bound. While reliable algorithms have been developed for linear systems, common nonlinearities such as valve stiction require modifications to the basic strategy. If the valve gets stuck due to stiction, for stable plants the output will reach steady state until the valve again moves. During this time the nonlinearity due to stiction is essentially removed from the system, and it is possible to compute performance assessment indices in the standard manner. This paper describes an automated strategy to reliably identify these linear steady-state periods and subsequently compute the minimum variance lower bounds. The results of a simulation example illustrate that the proposed methodology is efficient and accurate enough for the classes of systems and nonlinearities considered to provide statistics for control performance assessment for linear systems with nonlinearities caused by valves.