Method and system for assessing the quality of adhesively bonded joints using ultrasonic waves

Abstract

A method is provided for assessment of quality of an adhesively-bonded lap joint, wherein the joint includes a first metal plate, a second metal plate and an adhesive therebetween. The method includes sending ultrasonic waves normally to the surface of a sample outside of the joint where the sample has a first sample metal plate with the same properties as does the first metal plate at an assessment point of the joint. Reflected waves from the sample joint as a reference waveform are recorded. Wideband ultrasonic waves are sent normally to the surface of the joint at the assessment point. Reflected waves of the ultrasonic waves from the joint are recorded. A waveform of the reflected waves from the joint and reference waveform are analyzed to determine an informative parameter. The informative parameter is compared with a threshold value to assess quality of the joint.

Claims

1. A method for assessment of quality of an adhesively-bonded lap joint, the joint including a first curved metal plate, a second curved metal plate and an adhesive, the method including the steps of: a) sending ultrasonic waves normally to the surface of a sample outside of the joint, where the sample has a first sample curved metal plate with the same properties as does the first curved metal plate at an assessment point of the joint, and wherein the sample has a second sample curved metal plate with the same properties as does the second curved metal plate, and wherein the sample has an adhesive having a thickness that varies along the sample; b) recording reflected waves from the sample as a reference waveform; c) sending wideband ultrasonic waves normally to the surface of the joint at an assessment point; d) recording reflected waves of the ultrasonic waves from the joint; e) analyzing a waveform of the reflected waves from the joint and reference waveform to determine an informative parameter; and f) comparing the informative parameter with a threshold value to assess quality of the joint.

2. The method of claim 1, wherein the informative parameter is a measure of the deviation of the waveform reflected from the joint and the reference waveform.

3. The method of claim 2 further including the step of determining a defect is present based upon a determination that the deviation of the waveform from the joint and the reference waveform is below a threshold.

4. The method of claim 1, further including the step of multiplying the reference waveform with an exponentially decaying function having specific exponential factor and subtracting the product from the waveform recorded from the joint to obtain a difference.

5. The method of claim 4 further including the step of monitoring polarity of the difference to obtain an indication of the defect in the joint.

6. The method of claim 1 further including the step of compensating for time shift, overall amplitude variation and variation of a reverberation period of the waveform.

7. The method of claim 1 further including the step of translating a transducer along a surface of the joint and generating an image map of the quality of the joint.

8. The method of claim 7 further including the step of high pass filtering the image map and monitoring a level of irregular output to determine a presence of a foam-like defect.

9. The method of claim 1 where the sample has a first sample metal plate with properties similar to the first metal plate at the assessment point of the joint.

10. The method of claim 1 where the sample is a portion of the first metal plate outside the joint.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) FIG. 1 shows a schematic of the adhesive joint and ultrasonic transducer with defects in the adhesive joint.

(2) FIG. 2 is a schematic of the acoustic wave propagation in adhesively bonded joint.

(3) FIG. 3 is a sample with gradually varying adhesive thickness.

(4) FIG. 4 shows the delays of the responses S.sub.1 and S.sub.2 recorded for the sample shown in FIG. 3.

(5) FIG. 5 is a block diagram of the processing.

(6) FIG. 6A shows the experimental waveforms (grey curves) measured for the joint.

(7) FIG. 6B shows the experimental waveforms (grey curves) measured for the joint.

(8) FIG. 7 shows the block-scheme of the processing algorithm for detection defects #2, 3.

(9) FIG. 8 shows an estimation of the period of the ultrasound reverberations in the first plate.

(10) FIG. 9 shows a set of calibration samples.

(11) FIG. 10 shows curved calibration samples.

(12) FIG. 11 shows the structure of the s(x,t) data recorded for the sample with defect #4.

(13) FIG. 12 shows the structure of the s(x,t) data presented in FIG. 11 after high frequency filtration over x.

DESCRIPTION OF A PREFERRED EMBODIMENT

(14) FIG. 1 shows a schematic of a system 10 according to one embodiment of the present invention. The system 10 includes a transducer 12 connected to a computer 14 having a CPU 16, storage 18 (such as memory, RAM, hard drives, or some combination of working memory and mass storage, in electronic and/or optical, magnetic or other format). The computer 14 may have a user interface 20 (graphical user interface, input devices, display, etc).

(15) The adhesive joint being measured includes a first plate 22, a second plate 24 (generally parallel to and spaced from the first plate 22) and an adhesive 26 between the first plate 22 and the second plate 24. The typical thicknesses of these materials are in the range of 0.7-2 mm.

(16) FIG. 1 also shows the defects in the adhesive joint i.e. voids 31, 32, 33 and a foam-like defect 34. The defects detected by the proposed method are: voids (or delamination) 31 at the first plate/adhesive interface; voids (or delamination) 32 inside of the adhesive; voids (or delamination) 33 at the adhesive/second plate interface; and foam-like defect 34 (consists of numerous small air bubbles).

(17) There are two approaches of the testing procedure provided by the system 10. The first approach is applicable in the case when the interfaces of the adhesive 26 can be approximated within the transverse dimension of the interrogating ultrasonic beam with planes which are parallel to the surfaces of the plates 22, 24. The second approach is applicable in the case when the adhesive 26 has highly non-flat adhesive/air interface within the transverse dimension of the interrogating ultrasonic beam (foam-like defect) or adhesive has approximately flat interface inclined with respect to the surfaces of the plates 22, 24 or when the first and second plates 22, 24 are not parallel.

(18) First Approach.

(19) In this case the wave propagation can be explained by the model presented in FIG. 2. In this model the interfaces between transducer 12, plates 22, 24 and adhesive 26 are parallel and the ultrasonic wave experiences multiple reverberations within the layers and transitions the interfaces between them. In the waveform received by the transducer in pulse-echo mode it is possible to select the responses S1, S2, S3. Response S1 is completely produced by the waves reverberating in the first plate 22. There is only response S1 when the transducer 12 is located over no adhesive area or the defect 31 is presented.

(20) The responses S2 and S3 are produced by the waves which propagate in first metal sheet 22 and adhesive layer 26 and in all three layers, respectively. FIG. 2 shows scheme of the wave propagation. The amplitude of S1 is much larger than the amplitudes of S2 and S3.

(21) The time shift of these responses and reverberation periods depends on the thicknesses of the layers and sound velocities in the plates 22, 24 and the adhesive 26. Since the thicknesses of the layer can have arbitrary values, responses are overlapped in time and generally it is not simple to distinguish them in the output waveform:
s(t)=s.sub.1(t)+s.sub.2(t)+s.sub.3(t).

(22) However in some special case it is possible to observe these responses separately. FIG. 3 shows sample with gradually varying thickness of the adhesive h(x). FIG. 4 shows the delays of the responses S.sub.1 and S.sub.2 recorded as a function of position of the transducer x. Since responses S1 and S2 delayed on time which is proportional to the varying thickness of the adhesive layer, they look in the image as tilted lines.

(23) For defect #1 (no adhesive area) the responses S2 and S3 are absent and there is the response S1 only.

(24) The responses S2 and S3 are presented in cases of a good joint and in presence of the defects 32 and 33. To distinguish the defects 32 and 33 and good joint, the response S2 should at least be separated.

(25) In case of the foam-like defect 34 the response S2 does not have a regular structure.

(26) FIG. 4 shows the delays of the responses S.sub.1 and S.sub.2 recorded for the sample shown in FIG. 3.

(27) According to the invention the analysis of the signal is produced as follows and as illustrated by the block scheme presented in FIG. 5. The contact of adhesive 26 with the rear surface of the first plate 22 is tested. This procedure is based on the comparison of the waveform s(t) measured in the point of interest with the previously recorded reference waveform sat) from the outside of the joint for the bare first plate 22. The similarity of the current and reference waveforms can be considered as an evidence of the absence of adhesive 26 in the test point. To compare these measured and reference waveforms the cross correlation coefficient or deviation parameter r can be used. The deviation parameter r is defined as follows:

(28) $r={\left[{}_{t_1}^{t_2}{\left(s\left(t\right)-s_{R1}\left(t\right)\right)}^2\mathrm{dt}\right]}^{1\text{/}2}$

(29) First, in step 50, reference waveform s.sub.R1(t) from the outside of the joint for the bare first plate 22 is measured and stored. Then measurement of the waveform s(t) is measured in the point of interest in step 52. Then a deviation parameter is calculated based upon a comparison of the reference waveform and the measured waveform in step 54. Then the deviation parameter should be compared with the threshold r.sub.0. If r>r.sub.0, the difference between current and reference waveforms is large, then it should be assumed that there is contact with adhesive 26 at the rear surface of the first plate 22 in step 56. If r<r.sub.0, the difference between current and reference waveforms is small, then it should be assumed that there is no adhesive 26 at the rear surface of the first plate 22 in step 58.

(30) If the processing presented above shows that there is contact with adhesive at the first interface the second step is to detect defects 32, 33. To do this, the response s.sub.1(t) should be subtracted from the waveform s(t). The response s.sub.1(t) can be measured experimentally using a special calibration specimen which materials are identical to the material of the specimen under the test, the thicknesses of the first plates 22 are equal and the thickness of the adhesive 26 is large enough to be sure that all possible responses S2 and S3 are negligibly small. Alternatively s.sub.1(t) can be estimated by applying additional damping to the reference waveform:
s.sub.1(t)s.sub.R2(t)=s.sub.R1(t).Math.|R.sub.12|.sup.t/T

(31) where T is the period of the reverberations in the first plate 22. The coefficient R.sub.12 is the reflection coefficient of the ultrasonic wave at the interface between the first plate 22 and adhesive 26. It can be estimated theoretically using handbook parameters of the materials, and as a result of the experimental calibration procedure.

(32) FIGS. 6A and 6B show the experimental waveforms (grey curves) measured for the joint which consists of the steel plates 22, 24 with the thickness of h.sub.1=1 mm and epoxy-based adhesive 26. FIG. 6 shows the waveforms s(t) (grey lines 62) and s.sub.2(t)s(t)s.sub.R(t) (black lines 60). FIG. 6A illustrates a good joint and FIG. 6B illustrates a defect 32, 33 (FIG. 1). Thickness of the adhesive 26 (FIG. 1) is 0.4 mm. The reflection coefficient was estimated to be |R.sub.12|0.9. The results of subtraction of the estimated response s.sub.1(t) from the initial waveforms give the residual which can be treated as a response s.sub.2(t)s(t)s.sub.R2(t). The estimated responses s.sub.2(t) are presented in FIG. 6 as black lines 60. The polarity of s.sub.2(t) is coincide with the polarity of the initial waveform in case of the good joint (graph A) whereas the polarity of s.sub.2(t) is inversed in case of the disbond at the second adhesive/second plate interface (graph B) This inversion takes place due to negative value of the reflection coefficient at the rear adhesive/air interface in presence of the disbond.

(33) FIG. 7 shows the block-diagram of the processing algorithm for detection of defects 32, 33 (FIG. 1). The reference waveform s.sub.R2(t) is measured or estimated in step 68. The waveform under test s(t) is measured in step 70. The response s.sub.2(t) is estimated as s(t)s.sub.R2(t) in step 72. The polarity of s.sub.2(t) is estimated in step 74. If the polarity of s.sub.2(t) is inverted, there is determined to be a defect 32 or 33 in step 76. If the polarity of s.sub.2(t) is non-inverted, it is determined to be a good joint in step 78.

(34) The robustness of the defect detectability depends on many factors including curvature and roughness of the surfaces, variations of the thickness of the plates and the adhesive layer, non-stability of the acoustical contact between the transducer and the sample. These factors cause the time shift of the recorded waveform, decreasing of its amplitude and changes in the period of the reverberations in the first plate. As the result the divergence between the reference signal and the responses s.sub.1(t) of the measured waveforms increases and the accuracy of the estimation of the response s.sub.2(t) decreases. To compensate for the time shift and overall amplitude variation, the time delay and amplitude of the first pulse reflected at the transducer-plate interface were measured and then the waveforms is normalized in amplitude and aligned in time domain. The influence of this thickness variation can be reduced by recording the reference signal in the vicinity of the testing point. This, however, is not always possible due to restricted access to the joint; neither is frequent repetition of this setup procedure convenient in the case of long joints.

(35) FIG. 8 shows an estimation of the period of the ultrasound reverberations in the first plate. To compensate variations in the reverberation period, it is proposed that after amplitude normalization and time alignment the waveforms be scaled in the time domain by T.sub.0/T, where T and T.sub.0 are the periods of reverberations for the measured waveform and reference signal, respectively. To avoid interference caused by the echoes from the adhesive and second metal layers, the period T is measured as the time delay between the leading fronts of the pulses reflected at the front and rear surfaces of the first plate (FIG. 8).

(36) The proper threshold value r.sub.0, time gate [t.sub.1, t.sub.2], damping coefficient R.sub.12, and other parameters which used in the proposed technique can be optimized by comparing the ultrasonic data and the results of the destruction test of a set of samples. Alternatively a set of specially prepared calibration samples can be used (FIG. 9).

(37) FIG. 9 shows a set of calibration samples. The calibration samples A and B have the gradually varying thickness of the adhesive layer and the thicknesses h.sub.1, h.sub.2 and material identical to those of the parts which should be tested. The s(x,t) data recorded as a function of the time t and longitudinal position of the transducer x cover almost all possible thicknesses of the adhesive layer h at given thicknesses of the plates h.sub.1, h.sub.2. An example of the s(x,t) data is shown in FIG. 4.

(38) The calibration sample C has area with very thick adhesive. The waveform measured at this area can be used as estimation of the reference waveform s.sub.R2(t) or can be used for adjustment of the damping parameter R.sub.12.

(39) FIG. 10 shows curved calibration samples. Using curved calibration samples as shown in FIG. 10 it is possible to determine how the output of the processing varies with the curvature of the specimen and knowing this established relationship and the curvature of the sample under the test it is possible to adjust the output parameter and increase the detectability of defects.

(40) The technique presented above can be applied for the testing of the small area of the sample which is approximately equal to the dimension of the interrogating ultrasound beam. Also it is applicable in the cases when the testing is produced by means of manual or robotic translation of the probe or by means of electronic switching of elements of multi-element ultrasonic probe to generate B- or C-scans.

(41) Second Approach

(42) This approach should be used when the adhesive has highly non-flat adhesive/air interface within the transverse dimension of the interrogating ultrasonic beam (foam-like defect) or adhesive has approximately flat interface inclined with respect to the surfaces of the plates or when the first and second sheets are not parallel.

(43) FIG. 11 shows the structure of the s(x,t) data recorded for the sample with foam-like defect #4 produced by presence of air bubbles and channel in the adhesive. In the joint there are the steady response S1 and the weak response S2 which time delay and amplitude are not regular. It is possible to detect the response S2 applying a high frequency filtration of the s(x,t) data over the spatial coordinate x (FIG. 12). Similar, it is possible to detect the response S2 in cases when the rear surface of the adhesive layer is flat but substantially not parallel to the first plate. In these cases the inversion of the polarity of the selected response S2 can be used as an evidence of the void presence.

(44) For a good joint and for relatively large defects 31, 32, 33 the responses S1 and S2 are slowly varying functions over x and the output of the spatial high pass filter is small.

(45) In accordance with the provisions of the patent statutes and jurisprudence, exemplary configurations described above are considered to represent a preferred embodiment of the invention. However, it should be noted that the invention can be practiced otherwise than as specifically illustrated and described without departing from its spirit or scope.