Electromagnetic testing of a welding area using a magnetic sensor array

Abstract

This paper presents an electromagnetic testing method for the inspection of cracks in an overlay welding area. The overlay welding method is usually used to repair failure welds in a structure in which the components are difficult and expensive to replace. The inspection system uses an air-core exciting coil to produce an electromagnetic field in the overlay welding area and a linearly integrated giant magnetoresistance (GMR) sensor array (LIGiS) to measure the distribution of the electromagnetic field around the cracks; thus, the cracks can be inspected. The LIGiS is fabricated using 32 GMR sensor elements arrayed at 0.6 mm intervals. Artificial cracks with different sizes, orientations, and locations on the crisp surface of an overlay welding sample are tested to verify the effectiveness of the proposed method.

Keywords

LIGiS eddy current testing nondestructive testing overlay welding

1. Introduction

The overlay welding method is used to repair welds that appear as damages, such as stress corrosion cracks, during operations. It is widely used in gas/petroleum pipeline systems and in the reactor vessel in power generators. The replacement of new welding parts, in some cases, could involve considerable time, complications, and require the shutdown of the whole the system. Therefore, if the crack is in within a criterion and the weld can still be used after repair, overlay welding is applied over the crack in the welding area. However, frequent inspection and monitoring of the overlay welding using nondestructive testing (NDT) methods are required to ensure the safety and integrity of the system [1]. Several NDT methods such as X-ray testing, ultrasonic testing, and eddy current testing have been used. X-ray testing methods can be applied for any surface roughness condition, including the crisp surface of the overlay welding. The crack can be detected by advanced fusion techniques such as neural networks [2], adaptive network-based fuzzy inferences [3], and principle component analysis [4]. Ultrasonic testing methods such as linear and nonlinear phased array techniques have been applied to detect deep cracks [5, 6]. However, they are limited to the detection of surface and subsurface cracks owing to the crisp surface of the overlay welding, and these methods require couplant materials to transmit the ultrasound waves. Eddy current testing methods can detect surface and subsurface cracks without the need of a couplant material. Array eddy current probes using multiple pancake coils as the exciting and sensing elements have been developed [7, 9]. An exciting coil produces an eddy current in the inspected material, and this eddy current is distorted owing to presence of a crack. The distorted eddy current will be sensed by the changes in the coil’s impedance. Thus, the crack can be inspected.

In this research, we propose an inspection technique based on eddy current testing. A linearly integrated giant magnetoresistance (GMR) sensor array (LIGiS) is used instead of the pancake coils to directly measure the distribution of the magnetic field around the cracks [10, 11]. An air-core exciting coil, to which an alternating current is supplied, is used as the magnetic source. The magnetic field of the coil produces an eddy current in the inspected material. If there is a crack in the material, the eddy current will be distorted and it concentrates around the crack. The secondary magnetic field, which is produced by the distorted eddy current, will be measured by the LIGiS; thus, the crack can be inspected. The size of the GMR sensor is compact, allowing a high spatial resolution for arraying. In this research, 32 GMR sensors arrayed at 0.6 mm intervals are used. To verify the effectiveness of the proposed technique, artificial cracks on a three-overlay welding layer specimen was tested. The effects of the size, orientation, and location of the cracks will be analyzed.

2. Experimental setup

Figure 1 shows a sample of the overlay welding specimen with artificial cracks on the rough welding surface. Three layers of nickel-based filler alloy 52 M have been overlay welded on an Inconel alloy 690 base material. The welding line has a width of approximately 9 mm and the distances from the valleys to the peaks are 0.3–0.7 mm. The total thickness of the overlay weld is approximately 3.2 mm. Artificial cracks with 10 mm lengths, 0.15 mm widths, and 0.5–2.5 mm depths were formed on the valleys and peaks with different orientation angles, as shown in detail in Table 1.

Figure 1.

Overlay welding specimen with artificial cracks.

Table 1

Sizes and locations of the artificial cracks on the overlay welding specimen, unit [mm]

No.	Width	Length	Depth	Angle	Location	No.	Width	Length	Depth	Angle	Location
1	0.15	10.0	0.5	45 ${}^{\circ}$	Peak	8	0.15	10.0	1.0	90 ${}^{\circ}$	Peak
2	0.15	10.0	1.0	45 ${}^{\circ}$	Peak	9	0.15	10.0	2.5	90 ${}^{\circ}$	Peak
3	0.15	10.0	2.5	45 ${}^{\circ}$	Peak	10	0.15	10.0	0.5	0 ${}^{\circ}$	Valley
4	0.15	10.0	0.5	0 ${}^{\circ}$	Peak	11	0.15	10.0	1.0	0 ${}^{\circ}$	Valley
5	0.15	10.0	1.0	0 ${}^{\circ}$	Peak	12	0.15	10.0	2.5	0 ${}^{\circ}$	Valley
6	0.15	10.0	2.5	0 ${}^{\circ}$	Peak	13	0.15	1.5 $\times$ 4	0.5	0 ${}^{\circ}$	Peak
7	0.15	10.0	0.5	90 ${}^{\circ}$	Peak	14	0.15	1.5 $\times$ 4	0.5	90 ${}^{\circ}$	Peak

The setup of the sensor probe on the specimen is shown in Fig. 2. The sensor probe has a LIGiS placed under an air-core exciting coil and has a motor support for scanning. The exciting coil is made of 0.2-mm-diameter copper wire with 140 turns. An AC current of 200 mA at 10 kHz is supplied to the exciting coil which produces a RMS magnetic field density of 0.659 mT on the top of the coil in the axial direction (X-direction). The LIGiS has 32 GMR sensor elements (bare-chip type NVE AA004) arrayed at 0.6 mm intervals to form a sensing length of 19.2 mm. The GMR sensors are oriented at 45 ${}^{\circ}$ to the array length to prevent their saturation by the magnetic field of the exciting coil. Thus, the LIGiS is sensitive at 45 ${}^{\circ}$ of its length. Owing to the crispness of the welding surface, the lift-off varies from 0.2–1 mm.

Figure 2.

Inspection setup: a) sensor probe on the specimen, b) Configuration of the sensor probe including an exciting coil and a GMR sensor array.

Figure 3.

Block diagram of the inspection system.

The block diagram of the inspection system is simplified in Fig. 3. The I/O controller of the NI-USB 6255 device is used to control the power supplied to the exciting coil, LIGiS, and signal processing circuits. The output signal of the LIGiS is processed using high-pass filters with cut-off frequencies of 284 Hz, differential-type amplifiers with gains of 58 dB, and root-mean-square (RMS) circuits for conversion to a DC signal that is digitized by the A/D converter of the NI-USB 6255, finally appearing as a software in a PC. The sensitivity of the measurement system is 55.7 mV/ $\mu$ T. The motor is controlled by software via a USB port and scans at a speed of 14.6 mm/s.

Figure 4.

Experimental results for the artificial cracks with different orientations. The magnetic image of the cracks are formed by the distortion of the eddy current around the cracks.

3. Experimental results and discussions

Figure 4 shows the experimental results of the artificial cracks on the overlay welding specimen. The result ( $\Delta V_{\textit{RMS}}$ ) is the differential of the RMS voltage of the LIGiS ( $V_{\textit{RMS}}$ ), as expressed in Eq. (1) and is further filtered by a band-pass filter in the software. The left-column images indicate the crack orientations and the distribution of the eddy current around the cracks. It is observed that the crack orientation can be determined by the crack image caused by the distribution of the eddy current. The 90 ${}^{\circ}$ -crack images (#7, #8, #9) have a negative data group between two positive data groups and are distributed in the vertical direction that is the same as that of the crack orientation. This is similar for the multi-short crack (#14) as well, but the image has several groups distributed in the vertical direction because of the multi-short cracks. The image of the 45 ${}^{\circ}$ -cracks (#1, #2, #3) also has three data groups oriented at 45 ${}^{\circ}$ that is the same as the orientation of the crack. The image of the 0 ${}^{\circ}$ -cracks (#4, #5, #6) orients in the horizontal direction (0 ${}^{\circ}$ ) and the data concentrates on the two tips of the crack. Because the eddy current concentrates around the crack tips, as shown in the image on the left, the image of crack #13 has negative and positive data groups distributed along the horizontal direction, similar to the distribution of the short cracks.

$\displaystyle\Delta V_{\textit{RMS}}(i,j)=V_{\textit{RMS}}(i+1,j)-V_{\textit{% RMS}}(i,j),$ (1)

where $i$ and $j$ are the scan data index and sensor index, respectively.

Table 2

Evaluation results of the crack lengths and depths

		Length						Depth
Crack #		#4	#5	#6	#10	#11	#12	#4	#5	#6	#10	#11	#12
Repeated experiment	1 ${}^{\text{st}}$	9.47	10.33	10.05	9.47	9.47	9.76	0.42	1.13	2.42	0.51	0.81	2.52
	2 ${}^{\text{nd}}$	9.76	10.33	10.05	9.18	10.05	9.47	0.44	1.11	2.59	0.44	1.15	2.52
	3 ${}^{\text{rd}}$	9.47	10.33	10.33	9.76	10.05	9.76	0.32	1.21	2.40	0.52	1.09	2.44

Figure 5.

Section view signal: a) different depth cracks at the peak of the welding area and b) cracks at the peak and valley of the welding area.

The 45 ${}^{\circ}$ - and 90 ${}^{\circ}$ -cracks (#1 $\sim$ #3, #7 $\sim$ #9, #14) are on the peaks of the welding lines with 0.3–0.7 mm distances to the valleys; thus, the crack depths are reduced at the crack tips. Therefore, the evaluation of the lengths and depths of these cracks is limited. For 0 ${}^{\circ}$ -cracks situated on the peaks (#4 $\sim$ #6) and valleys (#10 $\sim$ #12), the crack depths could be same because the height is the same. Thus, the evaluation of the length and depth of these cracks is possible. Figure 5 shows the horizontal section view signal at the center of the 0 ${}^{\circ}$ -cracks. The negative peak signal on the left and the positive peak signal on the right indicate the tips of the crack. The crack length is the distance between these two peaks signal, as shown in Fig. 5a. In addition, the amplitude of the positive peak increases as the crack depth increases. Thus, the crack depth can be also evaluated by the positive peak amplitude. The section view signal of the same-depth (2.5 mm) cracks on peak (#6) and valley (#12) of the welding lines is shown in Fig. 5b. The amplitude of the valley-crack (#12) is significantly smaller than that of the peak-crack (#6). This is mainly because of the different lift-offs between the LIGiS and the crack, which is approximately 0.5 mm.

Table 2 shows the evaluation of the depths and lengths of the 0 ${}^{\circ}$ -cracks in the three repeated experiments. The length and depth were estimated by linear relationship with the peak-peak distance and the peak-peak value of the crack signal, respectively [12, 13]. The estimation errors are $\pm$ 0.33 and $\pm$ 0.21 mm for length and depth, respectively.

4. Conclusions

This paper has proposed an inspection technique for an overlay welding area based on eddy current testing. A linearly integrated giant magnetoresistance (GMR) sensor array (LIGiS) has been used to measure the distribution of the electromagnetic field around the cracks. The usage of a GMR sensor array in this study, with a 0.6 mm array interval, enables a high spatial resolution to be obtained for the measurement. Artificial cracks with a minimum size of 0.1 $\times$ 0.5 $\times$ 10 mm on an overlay welding specimen have been inspected by the proposed technique. Furthermore, the orientation, length, and depth of the crack could also be evaluated.

Footnotes

Acknowledgments

This research was funded by the Korea Institute of Energy Technology Evaluation and Planning (KETEP 20171520101610). We are grateful for the support.

References

Chu

H.A.

Young

M.C.

Chu

H.C.

Tsay

L.W.

and Chen

, Hot cracking susceptibility of Alloy 52M weld overlays onto CF8 stainless steel, Journal of Nuclear Materials433 (2013), 419–423.

Kasban

Zahran

Arafa

El-Kordy

Elaraby

S.M.S.

and Abd El-Samie

F.E.

, Welding defect detection from radiography images with a cepstral approach, NDT & E International44 (2011), 226–231.

Zapata

Vilar

and Ruiz

, An adaptive network-based fuzzy inference system for classification of welding defects, NDT & E International43(3) (2010), 191–199.

Mirapeix

Garcia-Allende

P.B.

Cobo

Conde

O.M.

and Lopez-Higuera

J.M.

, Real-time arc-welding defect detection and classification with principal component analysis and artificial neural networks, NDT & E International40 (2007), 315–323.

Dennis

Latiolais

and Thigpen

, Weld Overlay Examination Using Ultrasonic Phased Array Technology, in: 6th International Conference on NDE in Relation to Structural Integrity for Nuclear and Pressurized Components, Oct. 2007, Budapest, Hungary.

Diaz

A.A.

Crawford

S.L.

Cinson

A.D.

and Anderson

M.T.

, An Evaluation of Ultrasonic Phased Array Testing for Reactor Piping System Components Containing Dissimilar Metal Welds, U.S. Nuclear Regulatory Commission, Technical Letter Report, JCN N6398, Task 2A, Nov. 2009.

Todorov

Nagy

Levesque

Ames

and Na

, Inspection of laser welds with array eddy current technique, in: AIP Conference Proceeding 1511, 1065, 2013, doi: 10.1063/1.4789161.

Mook

Michel

and Simonin

, Low frequency eddy current arrays with video clock, in: 10th European Conference and Exhibition on Non-destructive Testing, June 7-11, 2010, Moscow, Russia.

Yamamoto

Takagi

and Uchimoto

, Extraction of crack indications from ECT signals using signal phase characteristics of a multi-coil probe, International Journal of Applied Electromagnetics and Mechanics33 (2010), 1179–1184.

10.

Angani

C.S.

Kim

and Lee

, Non-destructive testing of inclusions in cold-rolled strip steels using hall and giant magnetoresistance sensor arrays, International Journal of Applied Electromagnetics and Mechanics45 (2014), 747–753.

11.

Kim

Lee

and Park

, Detection and evaluation of backside crack in spot-welding component of heat-resistant alloy using electromagnetic field distortion, International Journal of Applied Electromagnetics and Mechanics (2016), doi: 10.3233/JAE-162146.

12.

Jun

and Lee

, Nondestructive evaluation of a crack on austenitic stainless steel using a sheet type induced current and a Hall sensor array, Journal of Mechanical Science and Technology22 (2008), 1684–1691.

13.

Hwang

Lee

and Kwon

, The application of a differential-type Hall sensors array to the nondestructive testing of express train wheels, NDT & E International42 (2009), 34–41.