Machine implemented analysis eddy current data

• US 4,763,274 A
• Filed: 06/24/1986
• Issued: 08/09/1988
• Est. Priority Date: 06/24/1986
• Status: Expired due to Term

• Current Assignee: Westinghouse Electric Company LLC (Brookfield Asset Management Incorporated)
• Original Assignee: Westinghouse Electric Corporation (Paramount Global (f/k/a ViacomCBS Inc.))
• Inventors: Savage, George A., Bone, David A., Taszarek, Bruce J., Ingraham, Ronald H., Junker, Warren R., Attaar, Mustan, Castner, Raymond P.
• Primary Examiner(s): Lall, Parshotam S.
• Assistant Examiner(s): Trans, V. N.
• Application Number:
  US06/878,095
• Time in Patent Office: 777 Days
• Field of Search: 364/481, 364/483, 364/484, 364/508, 364/551, 364/552, 364/580, 364/507, 324/223, 324/228, 324/229, 324/238, 324/83 Q, 324/220
• US Class Current: 702/38
• CPC Class Codes: G01N 27/9046 by analysing electrical signals
• CPC Class Codes: G01N 27/9093 Arrangements for supporting the sensor; Combinations of eddy-current sensors and auxiliary arrangements for marking or for rejecting

Abstract

A digital computer performs a complete eddy current analysis of the tubes of a steam generator from data input to generation of a final report. The computer extracts signatures generated by various structural features and tube defects from the data, breaks down composite signatures into their component parts and classifies each such signature by applying thereto a cascading set of experience based rules. Signatures classified as structural features are examined closely for the presence of colocated tube defects. A post processor correlates the location of the indication identified from the time based eddy current data with the known physical location of the structural features so that the position of the defects can be presented in terms of distances from fixed structures.

Claims

1. A method of machine implemented analysis of eddy current test data recorded in quadrature channels at spaced data points as a probe is passed through the tubes of a heat exchanger, comprising the steps of:

storing the eddy current data for the spaced data points in a data array in a programmed digital computer;

operating the digital computer to examine the data in the array point by point to identify lobes;

operating the digital computer to group the lobes into signatures;

storing a set of rules in the digital computer for characterizing as a function of selected characteristics of said signatures, structural features surrounding the tubes and defects in the tubes; and

operating the digital computer to apply said rules to said signatures to identify the structural features and defects and to generate an output identifying the structural features and defects identified.

2. The method of machine implemented analysis of claim 1 wherein said step of operating the digital computer to examine the data to identify lobes includes:

storing threshold values in the digital computer and operating the digital computer to sequentially point by point compare the data signals with the threshold values, and to identify as lobes the sequence of data points for which the amplitude of the data point signals exceed the threshold values.

3. The method of machine implemented analysis of claim 2 wherein said threshold values comprise quadrature components of a noise circle and wherein the comparing step comprises generating a data vector from the quadrature components of each data signal and comparing the magnitude of the data vector with a radius of the noise circle.

4. The method of machine implemented analysis of claim 3 including operating the digital computer to generate an average value of at least some of the data signals and to center the noise circle at said average value.

5. The method of machine implemented analysis of claim 4 wherein said average is a moving average comprising an initial average value of a selected number of data signals successively adjusted by the value of the data signal being examined.

6. The method of machine implemented analysis of claim 4 wherein a threshold factor weighted toward one axis is applied to the quadrature components of the eddy current data for each point for the comparison with the radius of the noise circle.

7. The method of machine implemented analysis of claim 2 wherein the step of operating the digital computer to group said lobes into signatures comprises operating the digital computer to count the number of data points between lobes and to group those lobes which are within a preselected number of data points from one another into a signature.

8. The method of machine implemented analysis of claim 7 further including the steps of operating the digital computer to count the number of data points in each signature, to break down signatures having more than a predetermined number of data points into component signatures and to apply said rules to said component signatures.

9. The method of machine implemented analysis of claim 8 wherein the step of breaking down a signature into component signatures comprises operating the digital computer to determine the phase angle of the lobes in said signature, and to group lobes having a predetermined phase angle relationship into a component signature.

10. The method of machine aided analysis of claim 9 wherein grouping lobes into component signatures comprise operating the digital computer to successively, beginning at one end of the signature, establish a first lobe as a base lobe, and to group all lobes having a phase angle within a prescribed number of degrees of a base lobe into a separate component signature.

11. The method of machine implemented analysis of claim 1 wherein at least some of said rules include evaluation of the phase angle of a signature and wherein the phase angle is calculated by operating the digital computer to determine the endpoints of the maximum amplitude of the signature and to measure the angle between a line joining the endpoints and a reference axis of the quadrature components.

12. The method of machine implemented analysis of claim 11 wherein the digital computer is operated to determine the endpoints of the maximum amplitude of the signature by selecting a data point as one endpoint and determining the other endpoint as the point which is the greatest distance in an X-Y plane of the quadrature components from said one endpoint, then determining a new position of the one endpoint by determining the point which is the farthest distance from said other endpoint and repeating the steps of determining new locations of the one and the other endpoints until the maximum distance between the one and the other endpoints is found.

13. A method of machine implemented analysis of multifrequency eddy current test data recorded in quadrature at spaced data points as a probe is passed through the tubes of a heat exchanger, said method comprising the steps of:

storing the multifrequency eddy current data for the spaced data points in a data array in a programmed digital computer; and

operating the digital computer to examine the data in the data array point by point to identify as lobes sequential data points which exceed preselected threshold values, to group lobes within a predetermined number of data points of each other into signatures, and as to each signature, to calculate the number of data points which define the signature, for signatures defined by less than a preselected number of data points to apply predefined rules to classify the signatures as one of a defined group of structural features, tube defects and irrelevant signals; and

for a signature defined by more than said preselected number of data points to apply additional predefined criteria to break the signature down into component signatures and to then apply predefined rules thereto to classify each of the component signatures as one of a selected group of structural features tube defects and irrelevant signals.

14. The method of claim 13 wherein the step of operating the digital computer to apply additional predefined criteria to break down a signature defined by more than said preselected number of data points, includes calculating the phase angle of each of the lobes within such a signature, and grouping successive lobes within preselected phase angle ranges into component signatures.

15. The method of claim 14 wherein the step of operating the digital computer to break down signatures defined by more than said preselected number of data points includes operating the digital computer to identify segments of the signature in a first selected frequency of said multifrequency eddy current data which overlap segments of a signature in a second selected frequency, to apply said predefined rules to classify overlapping segments of the first frequency which are defined by less than a designated number of data points as one of a defined group of structural features, tube defects and irrelevant signals, to divide overlapping segments of the first frequency which are defined by more than said designated number of data points into subsegments of a preselected number of data points each, and to then apply said predefined rules to said overlapping subsegments to classify them as one of a defined group of structural features, tube defects and irrelevant signals.

16. The method of claim 15 wherein the step of dividing overlapping segments of the first frequency includes operating the digital computer to determine the maximum amplitude of the subsegment, to determine the number of data points between the data point at which said maximum amplitude occurs and the end of said subsegment, and to adjust the number of data points in said subsegment when the end of said subsegment is within a predetermined number of data points of the maximum point.

17. The method of claim 15 including the step of operating the digital computer to apply selected predefined rules to each of the segments of the signature in the second frequency which are not overlapped by segments of a first frequency signature to classify them as one of a group of tube defects and irrelevant signals.

18. The method of claim 17 including the step of prior to classifying the unoverlapped segments of the second frequency signature, to operate the digital computer to breakdown such segments having more than a given number of data points by grouping successive lobes in such a segment which are within a given number of degrees of each other into a component signature and then applying said certain predefined rules to the component signatures to classify them as representative of one of a designated group of structural features, tube defects and irrelevant signals.

19. The method of claim 13 wherein the step of applying predefined rules to a signature defined by less than a preselected number of data points includes;

operating the digital computer;

to calculate the amplitude of said signature, to apply predefined rules to classify a signature with an amplitude greater than a preset value as one of a designated group of structural features and irrelevant signals, and to apply predefined rules to classify a signature with an amplitude less than said preset value as one of a designated group of tube defects and irrelevant signals.

20. The method of claim 19 further including operating the digital computer to examine signatures classified as structural features point by point in selected regions to extract component signatures therefrom and to apply said predefined rules to said component signatures to classify them as one of a selected group of tube defects and irrelevant signals.

21. A machine implemented method of identifying from point by point eddy current test data recorded in quadrature, a flaw occurring in the vicinity of another feature which produces a larger signature than the flaw, comprising operating a digital computer to:

select as component portions of the larger signature successive data points over which the rate of change of amplitude of the signature is less than a predetermined value;

measure the phase angle and amplitude of said selected portions of the signature;

identify as representative of flaws those selected portions of the larger signature which meet preselected criteria including phase angle and amplitude.

22. The method of claim 21 in which eddy current data is recorded point by point in quadrature in several frequencies and wherein said digital computer is operated to:

identify the data point limits of the signature produced by said another feature using eddy current data recorded in a first frequency; and

select said component portions of said larger signature in a second frequency only within the data point limits identified in said first frequency.

23. The method of claim 21 wherein the digital computer is further operated to determine the data point at which the maximum amplitude of said larger signature occurs, and to select component portions of said larger signal only within a predetermined number of data points of the data point at which the maximum amplitude occurs.

24. The method of claim 21 wherein said component portions of the larger signature are selected by operating the digital computer to:

construct piece by piece straight line segments extending between successive selected numbers of data points in said signature;

determine the phase angle of each of said segments; and

group successive segments having phase angles within a predetermined number of degrees of each other into a selected portion of the signature.

25. The method of claim 21 wherein said component portions of the larger signal are selected by operating the digital computer to:

construct triangles from groups of three data points each;

determine the area of each of said triangles and compare the same with a threshold area; and

group successive segments of the signature extending between data points forming said triangles having an area less than said threshold area into said selected portions of the signatures.

26. A method of eddy current testing for defects in tubes supported by support structures within a heat exchanger, comprising the steps of:

passing a probe through a tube and recording eddy current data signals for successive time spaced data points along the tube;

inputting the recorded eddy current data signals into a data file in a programmed digital computer;

programming into the digital computer a set of rules for analysis of the eddy current data to identify structural features of the heat exchanger and defects in the tubing;

operating the digital computer to analyze the eddy current data point by point and to apply said rules to identify structural features of the heat exchanger and defects in the tubes, if any;

inputting into the digital computer signals representative of the actual physical location of the structural features of the heat exchanger; and

operating the digital computer to compare the physical location signals for the structural features with the data points at which said structural features were identified, and to thus, as a function of data points between the identified structural features and defects, generate an output identifying the locations of said defects in relation to the actual physical location of the structural features.

27. A machine implemented method of identifying flaws in the tubes of a heat exchanger from multifrequency point by point eddy current data gathered by passing a probe through the tubes;

the method comprising the steps of;

inputting the multifrequency eddy current data into a digital computer;

operating the computer to examine the multifrequency data point by point to define as lobes sequential data points which exceed preselected threshold values and to group lobes within a preselected number of data points of the next lobe into signatures, to measure the amplitude, phase angle and direction of rotation of signatures in a first selected frequency and to identify as representative of a flaw a signature in said first frequency having an amplitude between preset limits, a phase angle of between about 10 to 15 degrees and about 165 to 170 degrees and a clockwise rotation.

28. A method of machine implemented analysis of eddy current test data recorded in quadrature at spaced data points, comprising the steps of:

storing the eddy current data for the spaced data points in an array in a digital computer;

operating the digital computer to establish a noise threshold as a function of selected quadrature components of the eddy current data;

operating the digital computer to point by point compare the eddy current data with said noise threshold and identified lobes as consecutive data points for which the value of the data points exceeds said noise threshold; and

operating the digital computer to group the lobes into signatures.

29. The method of machine implemented analysis of claim 28 where in the step of operating the digital computer to group said lobes into signatures comprises operating the digital computer to count the number of data points between lobes and to group those lobes which are within a preselected number of data points from one other into a signature.

30. The method of machine implemented analysis of claim 29 further including the step of operating the digital computer to count the number of data points in each signature and to divide signatures having more than a predetermined number of data points into component signatures.

31. The method of machine implemented analysis of claim 30 wherein the step of dividing a signature into component signatures comprises operating the digital computer to determine the phase angle of the lobes in said signature, and to group lobes having a predetermined phase angle relationship into a component signature.

32. The method of machine aided analysis of claim 31 wherein grouping lobes into component signatures comprises operating the digital computer to successively, beginning at one end of the signature, establish a first lobe as a base lobe, and to group all lobes having a phase angle within a prescribed number of degrees of a base lobe into a separate component signature.

33. The method of machine implemented analysis of claim 30 wherein the eddy current test data is recorded in multiple frequencies and wherein the step of dividing signatures defined by more than said preselected number of data points includes operating the digital computer to identify segments of the signature in a first selected frequency of said multifrequency eddy current data which overlap segments of a signature in a second selected frequency, and to divide said overlapping segments of the first frequency which are define by more than a designated number of data points into subsegments of a preselected number of data points each.

34. The method of machine implemented analysis of claim 33 wherein the step of dividing overlapping segments of the first frequency includes operating the digital computer to determine the maximum amplitude of the subsegment, to determine the number of data points between the data point at which said maximum amplitude occurs and the end of the subsegment, and to adjust the number of data points in said subsegment when the end of said subsegment is within a predetermined number of data points of the maximum point.

35. The method of machine implemented analysis of claim 34 including the step of operating the digital computer to determine the number of data points in segments of the signature in the second frequency which are not overlapped by the signature in the first frequency, and to group successive lobes in a segment in the signature of the second frequency having more than a preselected number of data points which are within a given number of degrees of each other into a component signature.

36. The method of machine implemented analysis of claim 28 wherein the digital computer is operated to determine the end points of the maximum amplitude of a signature by selecting a data point as one endpoint and determining the other endpoint as the point which is the greatest distance in an X-Y plane of the quadrature components from said one endpoint, then determining a new position of the one endpoint by determining the point which is the farthest distance from said other endpoint and repeating the steps of determining new locations of the one and the other endpoints until the maximum distance between the one and the other endpoints is found.

37. Apparatus for analyzing a tube of a heat exchanger comprising:

a probe and means for passing the probe through a tube to generate multifrequency point by point eddy current data in quadrature;

means for storing said point by point eddy current data;

means for electrically comparing said eddy current data point by point with threshold values to generate lobes, to group said lobes into signatures and to apply selected rules to said signatures to classify each of said signatures as one of a group of structural features, tube defects and irrelevant signals;

means for generating a graphical display simultaneously depicting tube length strip charts of a quadrature channel of selected frequencies of data, zoom strip charts of a selected portion of the data in selected quadrature components of selected frequencies, and X-Y plane lissajous patterns of the selected portion of the data for multiple designated frequencies; and

means for selecting said frequencies and said portions of the data.

38. Apparatus for analyzing point by point eddy current data recorded in quadrature from the tubes of a heat exchanger comprising:

means for storing said data point by point as electrical signals;

means for comparing said electrical data signals point by point with threshold signals and identifying as lobes sequences of data signals which exceed said threshold signals;

means for grouping said lobes into signatures;

means for applying preselected rules to said signatures to classify the same as one of a selected group of structural features, tube defects, and irrelevant signals; and

means for generating an output identifying said classification of each signature.