Detection and Quantification of Stiction

• US 20070288103A1
• Filed: 06/30/2005
• Published: 12/13/2007
• Est. Priority Date: 07/02/2004
• Status: Active Grant

• Current Assignee: Matrikon Inc. (Honeywell International Inc.), University of Alberta
• Original Assignee: Matrikon Inc. (Honeywell International Inc.), University of Alberta
• Inventors: Shook, David, Thornhill, Nina, Shah, Sirish, Choudhury, Ali
• Granted Patent: US 8,145,328 B2
• Time in Patent Office: Days
• US Class Current: 700/35
• CPC Class Codes: G05B 23/0229 knowledge based, e.g. expert systems; genetic algorithms

Abstract

An effective non-intrusive data-based monitoring method may reduce the cost of control loop performance maintenance by screening and short-listing those control loops or valves that need maintenance. The invention comprises a data-based, model-free, non-invasive method that can automatically detect and quantify stiction present in control valves. The method does not require the implementation of any additional valve travel test or, as commonly known, bump test of the control loop. The method may detect and quantify stiction using controlled variable (pv), controller output (op) and set point (sp) data. It does not require valve positioner (mv) data.

Claims

1. A non-invasive method for detecting stiction in a control loop comprising the steps of:

(a) obtaining pv and op signals from the control loop;

(b) detecting non-linearity in the control loop;

(c) if non-linearity is detected, filtering the pv and op signals to obtain filtered pv_f and op_f signals and obtaining a pv_f-op_f relationship;

(d) determining if the non-linearity arises from valve stiction by determining if a geometrical shape can be fitted onto the pv_f-op_f data.

2. The method of claim 1 further comprising the step of obtaining sp data from the control loop and using it to detect non-linearity in the control loop.

3. The method of claim 1 or 2 wherein the geometrical shape is an ellipse or a rectangle.

4. The method of claim 1 or 2 further comprising the step of quantifying the apparent amount of stiction by measuring the maximum width of the cycles of pv_f-op_f plot.

5. The method of claim 2 wherein non-linearity in the control loop is detected by the use of the sensitivity of the normalized bispectrum or bicoherence to the non-linear interactions that may be present in the control error signal (sp-pv).

6. The method of claim 1, 2 or 5 wherein the pv and op signals are filtered using a frequency domain Wiener filter.

7. The method of claim 1 or 2 wherein the control loop is non-linear if both the Non-Gaussian Index (NGI) and the Non-Linearity Index (NLI) are greater than 0.

8. The method of claim 4 wherein the apparent amount of stiction is determined by C-means clustering, fuzzy c-means clustering or fitted ellipse techniques.

9. The method of claim 1 or 2 further comprising the step of segmenting the data and choosing the data segment with regular oscillations to produce the pv_f-op_f plot.

10. A method for automatically detecting stiction in an industrial process, comprising the steps of:

(a) obtaining pv and op data from the process;

(b) determining whether or not the control error signal (sp-pv) is non-Gaussian and non-linear, using pv as a proxy for the sp data;

(c) if the control error signal is both non-Gaussian and non-linear, obtaining the pv and op relationship;

(d) determining from the pv-op data whether or not the process is suffering from stiction by determining whether the pv-op data conforms to a limit cycle.

11. The method of claim 10 wherein sp data is also obtained from the process and is directly used to determine whether or not the control error signal is non-Gaussian and non-linear.

12. The method of claim 10 or 11 wherein the pv and op data are filtered prior to step (d).

13. The method of claim 12 wherein the data are filtered with a frequency domain filter.

14. The method of claim 10 or 11 wherein the determination of whether or not the pv-op plot conforms to a limit cycle is performed by fitting a suitable geometry onto the pv-op plot.

15. The method of claim 14 wherein the suitable geometry comprises an ellipse or a rectangle.

16. The method of claim 10 or 11 further comprising the step of quantifying the amount of apparent stiction.

17. The method of claim 16 wherein the quantification step comprises the step of measuring the width of the cycles of the pv_f-op_f data in the direction of op_f.

18. The method of claim 17 wherein the width of the cycles is measured by a C-means clustering, a fuzzy c-means clustering or a fitted ellipse method.