A simple method for detection of stiction in control valves

IFAC Proceedings Volumes

Shah

Huang

2001

This article addresses the detection and diagnosis of oscillations in process measurements. Principal components analysis (PCA) can be applied to the power spectra for detection of groups of oscillatory process measurements. In cases where a well defined periodic oscillation is detected then diagnosis of the root cause is possible using a signature that grows stronger closer to the source. Two candidates for the signature are the distortion factor and a non-linearity statistic based on surrogate data. Their performance is compared using experimental data and an industrial data set.

Section: Introductionmentioning

confidence: 99%

Detection of Distributed Oscillations and Root-Cause Diagnosis

IFAC Proceedings Volumes

Shah

Huang

2001

“…The index Y→X in (1) indicates that the influence of sequence Y on sequence X is measured. The reverse dependency X→Y is calculated by exchanging x and y of the joint and conditional PDFs (Fig 9(b)).…”

Section: The Basis Of the Transfer Entropy Methods Is The Determinatiomentioning

confidence: 99%

“…Root causes of plant-wide oscillations due to sticking valves have been studied extensively and solutions now exist for that case [1][2][3][4][5], however other causes remain to be addressed including oscillations emerging in processes with recycle, also known as the snowball effect. The transfer entropy method looks promising for diagnosis of this type of problem.…”

Section: Introductionmentioning

confidence: 99%

Cause and effect analysis of chemical processes - analysis of a plant-wide disturbance

Bauer

IEE Seminar on Control Loop Assessment and Diagnosis

Meaburn

2005

In continuous chemical processes, disturbances in the process conditions can propagate widely and cause secondary upsets in remote locations. The aim of this paper is to apply some recent data-driven methods for detection and diagnosis of process disturbances using historical process data that have been proving successful in a range of applications. An industrial case study is presented in which a plant-wide control system disturbance caused by the presence of a recycle was successfully located and then verified by further plant testing.

“…Taha et al [11] gave a diagnostic procedure based on valve characteristics while [7] distinguished limit cycle oscillation caused by non-linearity in a control loop from linear effects such as oscillatory tuning. Horch [12] showed that the covariance of the controller input and the process input would be an odd function in the presence of a valve with hysteresis. Xia and Howell [13,14] gave an analysis of loop status that determined if an individual loop was affected by long or short term transients or a slow or fast oscillatory disturbance.…”

Section: Oscillation Detection and Diagnosismentioning

confidence: 99%

“…Diagnosis of the root cause may be achieved using techniques described elsewhere [7,[10][11][12]15]. A diagnosis reached on the basis of plant data can direct maintenance effort towards the likely root cause so that a definitive diagnosis using, for instance, a manual valve test can be reached promptly.…”

Section: Diagnosismentioning

confidence: 99%

Detection of multiple oscillations in control loops

Journal of Process Control

Huang

Zhang³

2003

212

169

This article addresses the detection of oscillations in measurements from chemical processes including the case when two or more oscillations of different frequency are present simultaneously. The presence of oscillations in selected frequency ranges is determined using a new method based on the regularity of the zero crossings of filtered autocovariance functions. The work is motivated by and illustrated with industrial data that exhibit multiple plant-wide oscillations. Issues of practical implementation in an automated tool are discussed. #