Valve stiction estimation using global optimisation

Valve stiction is the most frequent cause of loop oscillations. Thus, detecting and quantifying this valve problem is essential to ensuring plant profitability. In this work, a novel one-stage procedure to estimate stiction parameters is proposed using a two-parameter stiction model. The optimisation problem is computed using a global optimisation algorithm. These two propositions make the stiction computation more efficient and computationally faster than the currently available method. The applicability of the proposed approach is illustrated using a large number of simulated and industrial valves. Moreover, to isolate the impact of each proposition, the novel method is compared with the currently available technique, which is based on a two-stage scheme. The results show that the global optimisation algorithm is more efficient than the direct search and genetic algorithms, as previously proposed by Jelali (2008). The two-stage procedure provides a better estimate of the apparent stiction, whereas the one-stage procedure provides a better slipjump value.