RETRACTED: Review on Valve Stiction. Part I: From Modeling to Smart Diagnosis

Abstract

Abstract:Valve stiction is indicated as one of the main problems affecting control loop performance and then product quality. Therefore, it is important to detect this phenomenon as early as possible, distinguish it from other causes, and suggest the correct action to the operator in order to fix it. It is also very desirable to give an estimate of stiction amount, in order to be able to follow its evolution in time to allow the scheduling of valve maintenance or different operations, if necessary. This paper, in two parts, is a review of the state of the art about the phenomenon of stiction from its basic characterization to smart diagnosis, including modeling, detection techniques, quantification, compensation and a description of commercial software packages. In particular, Part I of the study analyzes the most significant works appearing in the recent literature, pointing out analogies and differences among various techniques, showing more appealing features and possible points of weakness. The review also includes an illustration of the main features of performance monitoring systems proposed by major software houses. Finally, the paper gives indications on future research trends and potential advantages for loop diagnosis when additional measurements are available, as in newly designed plants with valve positioners and smart instrumentation. In Part II, performance of some well-established methods for stiction quantification are compared by applications to different industrial datasets. OPEN ACCESSProcesses 2015, 3 423

RETRACTED: Di Capaci, R.B.; Scali, C. Review on Valve Stiction. Part I: From Modeling to Smart Diagnosis. Processes 2015, 3, 422–451

“…We have become aware that a review paper [1] recently published in Processes contained a considerable amount of identical text and a similar structure to a previously published review paper [2], although some novel content was presented, including new citations and a section on smart diagnosis. In agreement with the authors, we-the Editor-in-Chief and Publisher-have decided to withdraw the paper.…”

Data-Based Method To Diagnose Valve Stiction with Variable Reference Signal

“…The literature contains a significant number of methodologies to detect stiction, which can be classified into the following: 4 limit cycle patterns-based, 5,6 nonlinearity detection, 2,7 cross-correlation function-based, 8 and waveform shapebased detection. 9−12 With the exception of limit cycle patterns-based methods, the above-mentioned techniques can detect a sticky valve using only the process variable (PV) and controller output (OP), meaning that no information on valve stem position (MV) is required.…”

Generic framework for valve stiction detection and compensation with ANFIS-activated dual-mode MPC

“…Model-based approaches allow compensative measures to be taken to maintain higher levels of controller performance once system parameters have been identified. However, the inherent complexity of model-based methods makes them less robust and more difficult to implement at an industrial level [31].…”