CA1148650A

CA1148650A - Classifying defects in structural materials pursuant to detection by means of ultrasonics

Info

Publication number: CA1148650A
Application number: CA000335642A
Authority: CA
Inventors: Heinz Schneider; Dieter Lather; Klaus Abend; Raimund Lang; Walter Sternberg
Original assignee: Mannesmann AG; Nukem GmbH
Current assignee: Vodafone GmbH; Nukem GmbH
Priority date: 1978-09-15
Filing date: 1979-09-14
Publication date: 1983-06-21
Also published as: IT1165301B; JPS5543493A; SU1061709A3; FR2436394A1; IT7925541A0; GB2034036B; FR2436394B1; DE2840748A1; GB2034036A

Abstract

CLASSIFYING DEFECTS IN STRUCTURAL MATERIALS
PURSUANT TO DETECTION BY MEANS OF ULTRASONICS

ABSTRACT OF THE DISCLOSURE

The wave form of an ultrasonic flaw echo is analyzed by generating from the signal a complex signal function and determin-ing the location of the peak in the complex function plane. The angle of the vector representing the peak is normalized by means of analogously processing reference echos, and the normalized peak displacement angle is used as discriminating criterion to separate harmless errors from harmful ones.

Description

ro~

2 . ~-~
The present invention relates/flaw and defect discrimina-4 tion pursuant to nondestructive ultrasonic testing of structural 5 material, particularly metals.
Ultrasonic testing involves primarily the detection of 8 the location of a defect on the basis of the transit time of an ¦ echo; i.e., the defect constitutes a discontinuity in the propa~
gation characteristics of an ultrasonic test signal, causing a 11 portion of that signal to be reflected; e.g., back towards the , 12 launching transducer which has been switched from the transmit 13 mode to the receive mode follo~ing the launch. The travel path 1~ of the launched signal and the retul~n path of an echo are geomet-ricàlly predeterminable on the basis of the geometry of the part 16 tested. The location of the transducer on the ob~ect and the ~;
17 orientation of transducers (direction of launching) to the test 18¦ object and its curface are supplem~ontal parameters. There,ore, 19 the time of occurance of any echo in reiation to the launch tlme is indicative generally and directly usable for finding the 21 location of the discontinuity causing that echo. In order to 22 exclude echos from "natural" boundaries in the error detection 23 process proper, one usually chooses certàin expectancy ranges 24 (loo~ing windows) and determines whether an echo does or does not occur in such a w.indow. On that basis, one distinguishes, for 26 exampler whether a defect is located near or in the ~surface of~
27 the test object, or in the interior thereof. That,distinction . .
. ~ : : . , ., :

can already be used as a criterion with regard to the type of 2 defect involved because certain types of defects are more preve-

3 lant near the surface, other types are more likely to occur in

4 ¦ the interior. Instrumental here is, further, the conduction of 51 tests in different test planes and from different directions, 61 primarily to localize the defect and to gain some information on 7 its extension and orientation. However, this type of evaluation is not discriminating enough for many purposes. It should be g observed that there is a need for identifying ~ny defect as to its type because some defects are relatively harmless, others 11 are extremely critical. Thus~ even a fairly large nonmetallic lZ inclusion may bë tolerable while a fine crack is not.

E~Taluàtion of a flaw or dPfect on the ~asis of a r~c~ived 15 ~ echo has been limited in the past to the determination of whether 16 ¦ or not a particular amplitude (test level) has been e~ceeded.
17 ¦ Of course, this approach requires extensive calibration because 18 ¦ the return signal amplitude depends on many other factors, 19 ¦ including the construction and operation of the transducer 20 I as transmitter; the characteristic of the transduber when 21¦ operated as receiver; the mode and manner of coupling the trans-22¦ ducer to the test ob~ect; the band uidth of the system, including 231 particularly the recelver circuit; the angle of incidence of 241 the launched test beam; etc. Another factor which influences 25¦ the amplitude of the echo is the condition under which the reflec-26¦ tion occurs at the boundary of the test object material and the 271 defect. >
281 . ' : . . . .

, ~VR.. ~ ,3~ ~ ~ !æ~

l Automated test equipment as it has been used is not capable 2 of providing an ade~uate indication and representation of the 3 specific type of defect involved. The amplitude, per se, just is 4 not an adequate indication. Thus, in practice, one has used e~perienced test personne] for inspecting the wave form of a 6 returned signal to arrive at a highly subjective conclusion as 7 to the cause of this return signal. The automated equipment 8 was, thus, used only for detection of (a) mere presence of a 9 defect,-and (b) its location; and, e.g., a loo~ at the envelope of the received signal did reveal, hopefullyj whether the ll defect is a crack, a slag inclusion, a void, etc.; and whether 12 it was harmless or not.

1~ It sho~ld be noted that in ar.othe~ type of testing (na~ely, testing by means of induced electrical eddy currents), one could 16 determine àny phase shift between the energizing (input) si~nal 17 and a resulting output signal. This phase shift does give some 18 indication as to the physical characteristics of the defect.
'19 . . ' ' .
This approach is not possible in ultrasonic testing, since ; 21 one cannot determine any phase between an electrical signal 22 which triggers launching of the ultrasonic test signal, and a 23 received ultrasonic signal and its electric replica.

~6 .

27 ~ `.

; -3-_ . ...
.' ` .
.

ro~ l . ~
l DESCRIPTION OF T~IE INVENTION

3 It is an object of the present invention to provide a ne~
4 method for discriminating among various types of defects in structural materials, particularly in metals, on the basis of 6- ultrasonic test signals.

81 It is a particular object of the present invention to 9 analyze ultrasonic return signals without requiring any reference to the phase of the launching signal.
11 ' _ .
12 It is, thus, a specific object of the present invention 13 to analyze an ultrasonic signal which results from interaction i~ of a launched test signal with a structural materia].

16 ¦ In accoxdance with the preferred embodiment of the present 17¦ invention, it is suggested to process the returned ultrasonic l8¦ signal after its interaction with, e.g., reflection on a defect l9 in structural material, by interpreting the signal as the real component of a complex s1gnal and generating the companion 21 imaginary component. The resulting function in the complex planè~
22 is processed to determine the value and location of the maximum 23 amplitude,~the location being defined by the angle in relation 24 to the real or imaginary function axis. That angle represents the displacement of the complex function peak from the read 26 signal peak. Particùlarly that peak displacement angle is an 27 indi tion of signal distortion a~ the re1e^ting interface .' :
, ro~

.
1 and that, in turn, is a criterion for classifying defects.
2 In addition, one provides a companion pair of amplitude and 3 angle data, using the same, or the same type of, 4 test equipment, but a ~nown type of interaction, e.g., a known reflecting surface known in the sense of producing 6 a definite response, i.e., return, to thereby eliminate equip-7 menl parameters from the result. The information now used 8 for clarifying defects is the difference between these two 9 angles as taken from the complex plane. The resulting normal-ized peak displacement angle is used as defect-classifying 11 criterion. It was found empirically that harmless defects 12 and harmfull dè~ects can be distinguished by determining 13¦ whether or not the normalized angle fà]ls into one or the 1~L other of a plurality of empi ically dètermined angles. This 15 ¦ lends itself directly to objectively operating automation.

191 . ' ' ' ~

27 .

~; -5-.. ~

~ '-~ ~

1 DESCRIPTION OF THE DRAWINGS .
2 _ 3 While the speciflcation concludes with claims, particu-4 larly pointing out and distinctly claiming the subject matter which is regarded as the invention, it is believed that the 6 inven~ion, the objects and features of the invention and 7 further objects, featuxes, and advantages thereof will be better 8 understood from the following description taken in connection with the accompanying dra~7ings in which ' ,, .
11 Figure 1 is a composite representation of three rëlated 12 diagràms a), b), and c), illustrating a flrst phase of the 13 procedure and process involved in practicing the present 14 invention, ..
16 Figure 2 is a perspective view of a three-dimensional 17 signal plot; and i8 ~
19 Figure 3 is a graph, showiny the envelopc of the signal as shown in Figure 2. -21 :
22 Proceeding now to the detailed description of the drawings and of the in~-entive process in accordance Wit}l the preferred 2~ embodimen~, the portion a) o~ Figure 1 illustrates an example of an echo as r~ceived by an ultrasonic transdùcer, operated 26 in the recei~e mode. This part a) of Figure 1 could.be inter- ~ ~ .
27 preted as a display on the screen o an oscilloscope. Such ;~
~D -6-' ~_ ...... ~ f~
~o~
. D-6689 1 a display is not, per se, part of the inventive process; a 2 received signal, if displayed, may look like the plot in 3 Figure 1, diagram a).

This particular signal may have been produced as an echo 6 of an ultrasonic test pulse which was launched by the ultra-7 sonic transducer then being operated in the transmit mode.
8 The transducer, its energizing and driving ¢ircuit for the 9 transmit mode, its operational mode controlj and the amplifier connected to the~transducer in the receive-mode are all . .
11 conventional; Moreover, it is also known, for example, to 12 digitize this analo~ signal and to store the resulting digital 13 signal for further processing.
14 . , The fu~ther processing is based on the principle that..
16 the function, as deli.niated by the signal (Figure 1, a)),.can 17 be interpreted as the real component RE of a complex function 18 RE ~ l. The (or, an) imaginary component IM of that func-19 tion is generated as quadrature by means of, for example, a ~
Fourier transformation, or a Hilbert Kern~l transformation, etc. :
21 Part b) of Figure 1 depicts the resulting imaginary component IM
22 generated via such a function transformation, being correspond-23 ingly associated with the real component as per Figure 1, 24 part a). :
. . :
26 . . ` ~.

28 . .
-7~

1:
''`~` .

:

_____ _~__ ~L
~o~yl~l , 1 The function IM, as per part b) of Figure 1, iS7 of course, 2 also a function in time. Accordingly, for eaeh instant on the 3 abseissa axis, one will now find a real component amplitude 4 (Figure 1, part a)) being associated with an imaginary component amplitude (Figure 1, part b)). By associating these values in a common, complex ~unction plane, one obtains a graphie repre-7 sentation as per part e) of Figure 1. The actual generation 8 of such a plot is not necessary for practiclng the invention;
9 the plot is shown here for the p~rpose of demonstration.
10 . .
11 Diagram e) of Figure 1 shows the location of the maximum 12 amplitude Am in the eomplex function plane and diagram. That 13 amplitude loeation can be represented by a vector which has 1~ a particular angle ~, e.g., rel~tive to the imaginary function axis. The angle-~ signifies that the absolu-te maximum, or pea~, 16 of the complex signal function does not coincide with the 17 maximum, or peak, of the signal as received. It is, therefore, 18 eonvenient to cail this angle ~ the peak displacement angle.
1~
These ~alues (Am,~) are used as a eriterion for identi-21 fying the deect which has caused the particular (real~~siynal.

22 For partieular purposes, the ultrasonic signal, as received, 23 amplified and digitized, is processed in a programmed computer which provides the function transform and generates there~rom the imaginary component for the complex function (Figure 2).

26 The program then determines the absolute maximum Am and, for . ' ': ' , .
.
'. ~ ' ' ' .~ ~

:

ropM 131 , 1 example, the peak displacement angle ~ relative to the imaginary 2 axis. T~ese values are suitably stored for further processing, 3 to be described hereaf~er.

Figure 2 shows real and imaginary components of the complex 6 signal as generated (in parts) and plotted as a function of time.
Figure 3 sho~^7s the envelope of the comple~ signal also as a function 8 of time. It was discovered that the complex signal and function 9 generated out of the r~ceived signal by interpreting it as a real 10 ¦ component of such a complex function is better suited for identify-11 ¦ ing the defect. As stated earlier, the real signal contour depends 12 ¦ on the acoustic impedance of the defect, but also on other factors, 13¦ and the complex function permits more readily the extraction of 1~¦ defec~-identiIyillg criteria, which are the maximum amplitude and~
primarily, its phase in the complex function plane. Particularly, 16 the displacement of the maximum amplitude in the complex plane is 17 indicatit~e of the signal distortion at the reflecting boun~ary due to 18 its acoustic properties. That angular displacement ~8, for 19 example, for reflection of an acoustic wave at a crack or at a non-metallic inclusion.
21 - ~
22 ` In order to eliminate from this processed measurlng result interfering components which relate to particular, even unique, 24 features of the test equipment itself, par-ticularly in conjunc-tion with the controlling transmitter circuit, one provides, 26 broadly, a reference standard by using the same (or the same -27 type of) electr:ic circuit and equipment. The reference standard ;20 is a~ echo signal generated under known condtions. ~or example,~ :~

: :

- ~ld.. ~

one uses here the echo generated by a true surface of the part, or of a similar part (front-wall echo or rear-wall echo). Or one can use a calibration wire placed across the transducer, or a calibra-tion standard, to which the transducer is coupled. Examples of this type of referen oe elements are shown in Canadian Patent applications Serial No. 284,167, filed August 5, 1977, or Serial No. 305,016, filed June 8, 1978. See also printed German Patent applications 2,635,982 and 2,726,400.
In either case of generating a reference, a signal contour 'O similar to the type shown in Figure l, part a), will be produced.
m at signal is (a) digitized, (b) processed as to the Fourier or Hilbert Kernel transformation, (c) further processed to generate the companion imaginary function component to obtain (d) the complex function. The maximum, or peak, amplitude and the corresponding peak displaoement angles in that ccmplex function plane are deter-mined.
These referenoe values are suitably stored also and pro-vide, in effect, equipment constants and parameters. It should be noted that in those cases, in which a rear-wall echo and/or a front-wall echo is available on a running basis, such an echo is then directly available during the same test cycle in which the echo occurs; these referen oe echos just occur at times different from de-fect echos. me referen oe signal in accordance with the Canadian application mentioned above (Serial No. 305,016) is also generated ', ~, . :, ~:

5~

rO~M 1~1 ~ D-6689 ., .' 1 during each test cycle. Thus, the signal representing such 2 a rear-wall echo can be computer-processed whenever the need 3 arises to generate on the 5pO~ the pair reference data (Am,~)ref .
This way, one eliminates, for example, varying coupliny con-ditions as bet~Jeen transducer and test ob~ect, signal drift

6 in the electronics, etc.

7 . i 81 The reference signal pair (Aml ~)ref.
9 ' the processing circuit to normalize the measuxin~ values.
10 ' The difference between the peak displacement angle ~, resulting 11 from the processing of the flaw echo signal as described 12 (Figure 1, part c)) and the reference angle generated analo-13 gously, is the nor~alized'pea~ displacement angle which represents 1 the signal distortion resulting from the specific acoustic conditions at the defect. Analogously, one may form a 16 normalized amplitùde.
17 ' 18 The differènce in the maximum amplitudes of the two 19 'complex functions ~measuring and reference) can be used directly 'for determining the relevant signal level for determining the ~i severeness of the defect (reject level). More important, 22 qualitatively, is the nor~alized peak dlspiacement angle.
23 This difference is usable as a criterion for identifying the 24 cause of the echo. The difference in the acoustic impedancès across the interface between the defect and the material pro-26 duces the reflection. But -this difference itself is'different 27 for different types of defects, such as laminations, crac~s,~` ~ ;

~ :

-11- , : i ' :

.

_E~ ., ______~
~o~

1 inclusions, pockets, etc. Th,us, the incident ultrasonic signal 2 is differently distorted upon reflection by these different 3 types of interfaces and discontinuities in acoustic impedance.
4 This di~ference in distortion is reflected in different normalized peak dlsplacement angles in the complex functlon plane. Thus, 6 one will generate in advance a variety of normalized peak 7 aisplacement angles for co~parison purposes. These several

8 no~mali2ed alphas are generated on the basis of kno~rn defect

9 types. More particularly, various pieces of structural material with different particular types o~ defects'can be prepared, 11 such as lamination~, cracks, seams, ~oids, nonmetallic inclusions, 12 etc.; preferably~ these defects have a large variety o~ shapes, 13 orlentation materials, etc. For each SUC}l type of defect, the a ¦ normalized pea~. displacement angle is generated, an~ the varloùs -151 ' angles are grouped into ranges. These ranges will ~e appropri-16 àtely defined as stored signals in the processing and computiny , " I7 facility. As the facility now generates a normali~ed peak ' 1 18 displacement angie from a true defect echo, a determinatlon is ; 19 made into which range that angle falls. The defect is now classified on that basis.
21 - ` -22 It should ~e noted that classifying the detected defects 23 in the msnner described above is primari~y of interest for the purpose of separating harmless imperfections from harmful defects and flaws. `
~'~ 26 ', `
27 ' ' , 28 , -, ~ -12~

, , ~OPbll~l . D-6689 ,, 1 It was found that there is a direct correspondence between 2 peak displacement angle ranges for normalized angles and the 3 severity of the type of defect involved. Cracks, l~mination.s, 4 seams, and the like, are deemed severe and lead more readily to product rejection. Their normalized peak dispiacement angles 6 are rather closely placed on the comparison scale, while less 7 severe defects, such as nonrnetallic inclusions, voids, or the like, 8 are readily distinguishable by a different, normalized peak dis-9 ¦ placement angle range. The boundary is, for example, with smaller angles constitutiny harmless defects.
11 ~ ., .
12 Thus, the procedure lends itself to automation in that it is ~- 13 determined whether the normalized alpha angle is, or is not, 14 in a dal-ger range, i.e., a ran~e signifying critical defects.

15- This determination can be made pursuant to the same program 16 which generates the normalized peak displacement ~ngIes out of - 17 a flaw echo signal.

18 ' 19 The information concerning peak displacement angles as a criterion for distinguishing harmfull defects from harmless ones ;~

21 should be supPlemented by the normalized amplitude as that 22~ am~litude is indicative of the siæe of the defects. ;

24 The peak disp].acement angle,as deflned, can broadly be interpreted as phase information. It is significant, however, ~;~ 2~ that one does not~determine dlrectly any phase in relation to 27 the launch cont:rol signal. The launch signal as such is not 2~ used in the determination of the peak displacement angle.
:~ -13-. .. ..
~ : .

_B~ ___ ~o~n ~1 D-6689 .
., . . .

1 The phase relation is an indirect or relative one, based upon 2 different conditions for reflecting an ultrasonic vibration.
3 .............................................. . .
.4 The invention is not limited to the embodiments described ~ above, but all changes and modifirations thereof not constituting .6 departures from the spirit znd scope of the invention are intended 8 to be included. .
9 ~, .
'10 . ' .' .
11 . ~ .
12 . . ~ .
13 . ~ .

15 I . . .
~6 ~
17 ~ :
8 ` .

2~ .
:
26 . ~
2~ ' ~ ~ ~ :;
28 -14~ :

;'

Claims

THE EMBODIMENTS OF THE INVENTION IN WHICH AN EXCLUSIVE
PROPERTY OR PRIVILEGE IS CLAIMED ARE DEFINED AS FOLLOWS:

1. Method for the automatic discrimination of defects during non-destructive material testing in which during testing ultrasonic echo signals received from the material are converted to electrical signals with the aid of a transducer, real components of said signals, received in pre-determined time domains, being transformed about a complex amplitude plane and in which, furthermore, characteristic values of the complex signal plane are determined and are placed in relation with the same characteristic values of the complex signal plane of a reference body obtained in the same way as for the material to be tested, whereby the difference of at least one chara-cteristic value represents a measure for the detected defect, characterized in that the characteristic values of the complex signal plane are the maxi-mum amplitude and the associated angle.