Available online at www.sciencedirect.

com

ScienceDirect
Procedia Engineering 86 (2014) 469 – 475

1st International Conference on Structural Integrity, ICONS-2014

Non-Destructive Evaluation of Friction Stir Welded Joints by

X-ray Radiography and Infrared Thermography
T. Saravanana,*, B.B. Lahiria, K. Arunmuthua, S. Bagavathiappana, A.S.Sekharb,
V.P.M. Pillaib, J.Philipa, B.P.C. Raoa and T. Jayakumara
a
Metallurgy and Materials Group,
Indira Gandhi Centre for Atomic Research, Kalpakkam-603102, India
b
Department of Optoelectronics, University of Kerala, Trivandrum-695581, Kerala, India
*
E-mail ID: tsara@igcar.gov.in

Abstract

Digital X-ray radiography and infrared thermography techniques are used for evaluation of the quality of the friction stir welded
aluminum butt joints and aluminum-zinc coated steel dissimilar lap joints. Digital frame integration and gradient operation based
image processing techniques are used on the radiography images which ensured 48% increase in the signal-to-noise ratio. The
effects of various welding parameters like tool rotation; travel speed etc. on the quality of the weld are studied. A sub-surface
tunnel defect along the weld line of a butt-joint is detected using infrared thermography technique and it is observed that the rate
of temperature decay is lower for the defect regions. Using lock-in thermography technique, the optimum frequency is
determined and the defect depth is quantified at this frequency. Improved visualization and contrast sensitivity are achieved using
adaptive single plateau based histogram equalization on the acquired infrared images.
© 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
© 2014 The Authors. Published by Elsevier Ltd.
(http://creativecommons.org/licenses/by-nc-nd/3.0/).
Peer-review under responsibility of the Indira Gandhi Centre for Atomic Research.
Peer-review under responsibility of the Indira Gandhi Centre for Atomic Research
Keywords: Friction stir welding, Digital X-ray radiography, Infrared thermography, Lock-in thermography, Image processing

1. Introduction

Weld joints are the origins of structural weakness in maximum cases and must be routinely inspected to
ensure structural integrity of the fabricated components. Friction stir welding (FSW) is gaining popularity in
engineering industries. FSW is a solid state joining process where simultaneous presence of forging pressure and
frictional heating causes the metal pieces to fuse together to form weld joints and the microstructural characteristic
remains mostly unchanged. FSW offers a number of advantages like less porosity, shrinkage and distortion, absence
of melting and any filler materials resulting in less weld contamination and less number of process variables to
control. Friction stir welded joints show superior mechanical properties because of the re-crystallized fine and equi-

1877-7058 © 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/3.0/).
Peer-review under responsibility of the Indira Gandhi Centre for Atomic Research
doi:10.1016/j.proeng.2014.11.060
470 T. Saravanan et al. / Procedia Engineering 86 (2014) 469 – 475

axed grains in the stirred zone. FSW is associated with some unique defects such as kissing bonds, cavity or groove
like defects (sub-surface tunnel defects) due to insufficient heat input, voids, porosities, lack of bonding, large
masses of flash due to excessive heating and abnormal stirring induced cavities [1, 2]. Non-destructive evaluation
(NDE) of FSW joints is essential as defects in weld joints may render a part of the joint unable to meet required
merits and their presence may cause premature failure of the joints. Hence, there is a strong need for inspection and
monitoring of FSW joints. Farley et al. qualitatively described the importance of NDE in weld inspection by the
following relation [3].

Probability of weld failure = probability of flaw occurring × probability of NDE missing the
flaw × probability of flaw growing

In this paper, digital X-ray radiography (DXRG) and infrared thermography (IRT) techniques are used for
defect detection and quality evaluation of two types of friction stir welded joints, viz. aluminum butt joint and
aluminum and zinc coated steel dissimilar lap joints [4, 5]. Defect-free butt joints of 3003 Al alloy to mild steel
plates with 3 mm thickness were made using FSW process and the effects of welding speed, rotation speed and tool
shoulder diameter on the micro-structure and strength of weld was reported [6]. Chen et al have reported that the
surface state of the steel (zinc-coated steel, brushed finish steel and mirror finish steel) influences the mechanical
strength of the FSW joint for AC4C Al alloy and Steel dissimilar friction stir lap joints [7]. Aluminum alloy welds
produced by FSW process has been evaluated by X-ray radiography, conventional ultrasonic NDE procedure and
phased array ultrasonic technique for varied welding parameters such as FS tool rotational and traverse speed [8].
Radiography is based on the differential absorption of radiation on its passage through the matter, whereas, in IRT,
infrared rays (wavelength lies between 0.75-1000 μm) emitted by an object is detected by an infrared detector and
the temperature of the object is measured in a non-contact way from the intensity of the emitted infrared waves. The
objective of the present study is to develop a DXRG procedure for dissimilar aluminum and zinc coated steel FSW
lap joints and to explore ways for improved signal to noise ratio (SNR) and for enhancing the defect detection limit.
This methodology uses digital frame integration for acquisition of data to increase the SNR followed by high-pass
filtering to sharpen the image by gradient operation and contrast adjustments to detect micro defects in FSW joints.
A sub-surface tunnel defect along the weld line of a friction stir welded butt joint is detected using active IRT
techniques. Using lock-in thermography (LI-IRT) technique, the optimum frequency is determined and defect depth
is quantified at this frequency. Point operation based image processing techniques are used for enhancing the
contrast of the LI-IRT images. Improved visualization and contrast sensitivity is achieved through adaptive single
plateau histogram equalization of the acquired LI-IRT images.

2. Materials and Methods

2.1 Materials

Dissimilar lap joints (150 × 100 mm) were fabricated using 2 mm thick aluminum (grade 6061) and zinc
coated steel (SS 316) sheet of 1 mm thickness. The lap joints were manufactured using a friction stir welder (Model:
RM 1A-0.7) under displacement control mode. The weld joints were produced for a dwell time of 2 s and for two
different depths of 2.3 and 2.45 mm, respectively. The dimension of the friction stir welded aluminum (grade 6061)
butt joint is 250 × 145 × 3 mm. IRT was performed on this specimen. The specimen surfaces were black painted to
enhance the emissivity.

2.2 Experimental Method

DXRG of the weld joints were carried out using a 450kV Balteau constant potential X-ray unit (focal spot
size 1.2 mm) and flash scan FS35 Thales flat panel detector was used as the X-ray detector (127 μm pixel pitch).
The radiography exposure parameters are given in Table-1. The X-ray images were acquired after an exposure time
of 4.2 seconds i.e. 1 frame (frame time) and by integration of a number of single frame of X-ray images. VI3
software was used to integrate the multiple frames of data into one image.
 T. Saravanan et al. / Procedia Engineering 86 (2014) 469 – 475 471

Table 1 X-ray radiography exposure parameters for FSW joints

Parameters Dissimilar lap joint Aluminum butt joint

Voltage (kV) 135 125
Current (mA) 1 1
Exposure time (s) 4.2/105 4.2/105
Number of frames 1/25 (integration) 25 (integration)
Source –to-Object 1000 950
Distance (SOD) (mm)
Object –to- Detector Contact 50
Distance (ODD) (mm)
Magnification 1.0 1.05

For conventional active IRT the specimen was heated using one 1 kW halogen lamps and the surface
temperature evolution was observed in the transient domain while natural cooling using a FLIR SC 5000 infrared
camera (spectral range: 2.0-5.1 μm). Temperature was recorded from the blind side of the specimen. The camera has
indium antimonide (InSb) detector with a two dimensional array of 320×256 elements. The detector elements are
cooled using Stirling cycle and thermal sensitivity of the camera is better than 25 mK. The infrared camera was
positioned at a distance of 700 mm away from the specimens in such a way that the axis of the camera coincides
with the axis of the specimen. For LI-IRT, the specimen was heated by sinusoidally modulated heat waves from two
1 kW halogen lamps kept at 300 mm away from the specimen. For generation of sine waves of a single frequency, a
programmable frequency generator (HM 8131-2, Hameg) was used. LI-IRT images were acquired using ALTAIR
LI software. Several excitation frequencies were used to determine the optimum frequency. As heating of aluminum
specimen using optical excitation is difficult due to its high thermal diffusivity, prior to lock-in thermography,
appropriate pre-heating was done.

3. Results and Discussion

Figures 1 (a & b) show the radiographic images obtained with 1 and 25 frames exposure, respectively for a
typical FSW joint made at a rotational speed of 500 min-1 and travel speed of 50 mm min-1 for a plunger depth of
2.30 mm. From Fig.1a, on the weld joint location, the normalized SNR was measured as 114, whereas, after
integration (Fig. 1b) SNR increased to 169. The radiographic image in Fig. 1a is not acceptable, as the ASTM E
2737 accepts the images with SNR > 130. As an image quality indication ASTM 2-B wire penetrameter was pasted
on the source side of the specimen as per the standards. Using the modified digital radiography procedure, the 63
μm thick IQI was clearly seen. Figure 2a shows the radiographic image of a FSW joint for the welding parameters
of 1000 min-1 rotational speed and 50 mm min-1 travel speed and penetration depth of 2.45 mm. From Fig. 2a, it is
observed that the heat energy produced is optimal which resulted in a sound (defect-free) FSW joint. These welding
parameters resulted in reduction of thermo-mechanically affected zone (TMAZ) thickness in the FSW plate which
facilitated the improved contrast of TMA zone in the radiography image. Figure 2b shows the radiography image of
a FSW joint with a tool rotational speed of 1500 min-1 under identical welding and X-ray exposure parameters to
that of Fig. 2a. At the same travel speed of 50 mm min-1, due to higher thermal gradient at higher tool rotational
speed of 1500 min-1, small cavities were present in the weld as indicated in Fig. 2b. At a constant travel speed of 50
mm min-1, with increase in rpm, energy input increases from 88.9 kJ mm-1 to 223.8 kJ mm-1, the Z force decreases
from 6 kN to 3.5 kN and the torque decreases from 15 Nm to 11.2 Nm. Thus, the material was more softened with
increasing energy input, forming a more sticking condition at 1500 rpm and 50 mm min-1. Under these conditions
material reaches to a state of abnormal stirring with a very low downward force and torque which may lead to the
formation of cavity.
472 T. Saravanan et al. / Procedia Engineering 86 (2014) 469 – 475

(a) (b)
Fig. 1 Digital X-ray radiography image of a FSW joint with rotational speed of 500 min-1 and travel speed of 50 mm
min-1 for a plunger depth of 2.30 mm (a) exposure performed at 1 frame (b) exposure with 25 frames integration.

(a) (b)
Fig. 2 Radiography image of the FSW joint with a travel speed of 50 mm min-1 for a plunger depth of 2.45 mm with
weld tool tip rotational speed of (a) 1000 min-1 and (b) 1500 min-1.

(a) (b)

Fig. 3 X-ray radiography image of the FSW joint with rotational speed of 500 min-1 and travel speed of 50 mm min-1
(a) plunger depth of 2.30 mm (b) plunger depth 2.45mm

Figure 3a shows the radiography image of the FSW lap joint for the welding parameters of 500 min-1
rotational speed and 50 mm min-1 travel speed for a penetration depth of 2.30 mm. It can be observed from Fig. 3a
that the voids and porosities are present. The formation of defects was due to poor material flow-ability as the
frictional heat input produced is less in the case of 2.30 mm penetration depth conditions. Figure 3b shows the
radiography image of FSW joint for the welding parameters of 500 min-1 rotational speed and 50 mm min-1 travel
speed for a penetration depth of 2.45 mm. From Fig. 3b, it is observed that a lack of bonding type defect is present
as indicated by arrows at the root side. For the welding parameters of 500 min-1 rotational speed and 50 mm min-1
travel speed, the heat energy produced was insufficient to plasticize the weld zone. Thus, a cold weld was formed at
the weld joint.
 T. Saravanan et al. / Procedia Engineering 86 (2014) 469 – 475 473

Figure 4a shows the temperature decaay as a function of time for the defect and defect-free reegion of the
aluminum butt weld. It can be seen from Fig. 4a 4 that the rate of temperature decay is slower in the deefect region
compared to the defect-free region. This is attrributed to the fact that the defect region is devoid of aluuminum and
filled with air as probability of contamination is very low in friction stir welding. Thermal conductivity of air is
several times lower than that of aluminum and hence,
h the temperature of the defect region decreases at a slower rate.
The temperature decay data was modeled with an a exponential equation. The time constants of temperaturre decay for
the defect and defect-free regions are found too be 0.00646 and 0.00671 s, respectively, which confirrms that the
temperature of the defective region indeed decaays at a slower rate. For quantification of defects using LI-IRT, the
thermal diffusion length [μ; μ= (2Į/Ȧ) 0.5, whhere Į is the thermal diffusivity and Ȧ is the angularr excitation
frequency] must be of the order of defect depths. Therefore, defects of a particular depth are visible with h maximum
contrast by a thermal wave of a fixed frequenncy. A range of excitation frequencies (9-13 Hz) were tried t for the
aluminum specimen and the optimum frequencyy, where the defect can be demarcated with maximum co ontrast, was
determined. Figure 4b shows the variation of phhase contrast (i.e. phase difference between the defect and d defect-free
regions) with excitation frequency. As can be seen
s initially the phase contrast increases with excitation
n frequency,
attains a maximum at 11.1 Hz and thereafter, decreases
d to a very low value. From the thermal migratio on length at
optimum frequency, defect depth (1.8μ) was esttimated as 2.79 mm.


Fig. 4 (a) Temperature decay as a function of tiime for the defect and defect-free region of the aluminum
m butt joint.
(b) Phase contrast as a functionn of excitation frquency for the aluminum butt joint.

Figure 5a shows the phase image of the aluminum specimen acquired at 11.1 Hz excitation freq quency. The
defect can be clearly seen from the image. Figurre 5b shows a 2D colour map of the phase image, where the t presence
of the defect can be distinctly identified. Figuure 5c shows the magnified view of the defect from diigital X-ray
radiography and lock-in thermography, respectively. The length and width of the defects were determined after
suitable pixel calibration within an error marginn of ±3.6%. Estimated (and actual) length and width of th he defect are
59.00 (60.00) and 0.53 mm (0.55 mm), respecctively. Point operations based image enhancement techn niques were
developed for improving the contrast sensitivityy of the infrared images. Conventional histogram equalizzation is the
most widely used point operation technique but b suffers from serious drawbacks like over equalizaation of the
background, blurriness of edges and reduced SN NR. For non-Gaussian histogram images (which are very common in
infrared thermography) with large background and a small target, single plateau histogram equalization tecchnique was
implemented [9]. In single plateau histogram m equalization technique, the equalization of the bacckground is
controlled using a plateau threshold value whichh is in general image dependent. An adaptive algorithm was w used for
automatic determination of this threshold value and gray levels greater than this are truncated to this thresshold value.
Once through and recursive adaptive single plaateau histogram equalization was implemented and Fig. 6 shows the
results for the aluminum specimen. The variaation of intensity contrast with the image processing teechniques is
shown in Fig. 6e and it can be seen that once thhrough and recursive technique enhances the intensity con ntrast by 80
and 10%, respectively compared to the conveentional histogram equalization technique. This clearly shows that
474 T. Saravanan et al. / Procedia Engineering 86 (2014) 469 – 475

plateau histogram equalization based processing of infrared thermography enhances the contrast and enables
visualization of defects. This technique is highly suitable for defect detection in weld joints using IRT.

Fig. 5 (a) Phase image of the aluminum butt joint specimen at 11.1 Hz lock-in frequency. (b) 2D colour map of the
phase image. (c) Magnified view of the defect region from XRG and LI-IRT images.

Fig. 6 (a) Original phase image, (b) conventional histogram equalized image, (c) once through single plateau
histogram equalized image, (d) recursive single plateau histogram equalized image and
(e) Comparison of intensity contrast achieved by various image processing algorithms.

4. Conclusion

A digital X-ray radiography procedure is developed to detect micro-pores and voids in aluminum-zinc
coated steel dissimilar friction stir weld lap joints. The studies confirmed that by using this procedure, signal to
noise ratio increases by 48%. This methodology has been successfully used to study the effect of welding
parameters like rotational speed, travel speed and penetration depth. A sub-surface tunnel defect along the weld line
was detected using infrared thermography and it was observed that the rate of temperature decay was slower for the
defect region. Using lock-in thermography, defect depth was quantified from the thermal diffusion length.
Compared to conventional histogram equalization, intensity contrast was increased by 80% using single plateau
histogram equalization of the acquired lock-in thermography images.

5. Acknowledgement

The authors thank Professor T. K. Pal and Mr. Hrishikesh Das, Jadavpur University, India for fruitful
discussions and technical help.
 T. Saravanan et al. / Procedia Engineering 86 (2014) 469 – 475 475

6. References

[1] Y.S. Sato, H. Takauchi, S.H.C. Park, H. Kokawa, Characteristics of the kissing-bond in friction stir welded
Al alloy 1050, Mat. Sci. Eng. A, 405 (2005) pp 333-338.
[2] Y.G. Kim, H. Fujii, T. Tsumura, T. Komazaki, K. Nakata, Three defect types in friction stir welding of
aluminum die casting alloy, Mat. Sci. Eng. A, 415 (2006) pp 250-254.
[3] J.M. Farley, J.L. Thompson, B.J. Dikstra, Nondestructive Testing to avoid weld failures: a review, in: J.D.
Harrison (Ed.) International Conference on Weld Failures, The Welding Institute, London, England, 1988.
[4] T.Saravanan, H. Das, K. Arunmuthu, J. Philip, B.P.C. Rao, T. Jayakumar, T.K. Pal, Evaluation of
dissimilar friction stir lap joints using digital X-ray radiography, Sci. Technol. Weld. Joining, 19 (2014)
pp 125-132.
[5] B.B. Lahiri, S. Bagavathiappan, T. Saravanan, K.V. Rajkumar, A. Kumar, J. Philip, T. Jayakumar, Defect
Detection in Weld Joints by Infrared Thermography, in: International Conference on NDE in Steel and allied
Industries, NDESAI, 2011, Jamshedpur, India, 2011, pp 191-197.
[6] M.Dehghani, S .A.A.Akbari Mousavi and A.Amadeh, Effects of welding parameters and tool geometry on
properties of 3003-H18 aluminum alloy to mild steel friction weld, Trans.Nonferrous Met.Soc.China, 23
(2013) pp 1957-1965.
[7] Y.C Chen and K.Nakata , Effect of the Surface State of Steel on the Microstructure and Mechanical
Properties of Dissimilar Metal Lap Joints of Aluminum and Steel By Friction Stir Welding,Metal. and Mat.
Trans.A, 39A(2008) pp 1985-1992.
[8] Vidya Joshi, Krishnan Balasubramainam and Raghu V Prakash, Study of defects in friction welded A A
5083 by radiography, ultrasonic and phased array unltrasonic technique, Proc. National. Seminar on Non-
destructive Evaluation, NDE 2011, Dec.8-10, 2011.
[9] Wang Bing-jian, Liu Shang-qian, Li Qing, Zhou Hui-xin, A real-time contrast enhancement algorithm for
infrared images based on plateau histogram, Infrared Phys. Techn., 48 (2006) pp 77-82.