Lasers in Manufacturing Conference 2015

Online Crack Detection During Laser Welding Using Passive

Thermography
Daniel Weller*a, Peter Stritta, Florian Fetzera, Rudolf Webera, Thomas Grafa
Cyrille Bezençonb, Jörg Simonb, Corrado Bassib
a
Institut fuer Strahlwerkzeuge (IFSW), Pfaffenwaldring 43, 70569 Stuttgart, Germany
b
Novelis Switzerland SA, Rte des Laminoirs 15, 3960 Sierre, Switzerland

Abstract

The non-destructive testing of laser welds within the process chain of industrial production is of prime importance for
the quality assurance. In this paper passive thermography was used to detect centerline cracks. These hot cracks are
likely to occur when welding car body sheets in an overlap configuration close to the edge. Welding at this position leads
to a strong heat accumulation. This results in a non-symmetric temperature field causing a heat flow from the edge into
the body across the weld seam. If the weld seam is split due to a centerline crack this heat flow is significantly disturbed.
It was found that this characteristic change of the heat flow is clearly measurable and can be utilized to detect hot
cracking during welding using an infrared camera.

Keywords: Laser Welding, 6000 Series Aluminum, Hot Cracking, Thermography, Process Monitoring

1. Introduction

Laser welding processes are well established in the automated mass production, such as in the
automotive industry. Since the introduction of the remote laser welding technology flexible weld shapes and
fast positioning are possible. To guarantee a constant product quality in high volume applications fast,
reliable and cost-effectively process monitoring is required.

* Corresponding author. Tel.: +49-711-68564146; fax: +49-711-68554146.

E-mail address: daniel.weller@ifsw.uni-stuttgart.de.
 Hot Cracks are one of the important weld defects to monitor in car body production. 6000 series
(magnesium and silicon based) aluminum alloys used as a lightweight material are susceptible to hot
cracking during laser welding. Centerline hot cracks are forming at the trailing end of the weld pool due to
an overload caused by tensile stresses and strains as mentioned by Pellini, 1952. This thermo-mechanical
load is due to the shrinkage during solidification and the temperature gradients during welding. The
temperature gradients and therefore the hot cracking sensitivity are influenced by the process parameters
and the weld position on the work piece as summarized by Cross, 2005. If the weld position on sheets with a
thickness of about 2.7 mm is below a distance of 10 mm to the edge the temperature gradient at the trailing
end of the weld pool is affected by a reflected heat flow from the edge, shown by Weller, 2013. This affected
temperature gradient leads to a changed thermo-mechanical load which is one reason for centerline hot
cracks in welds placed at distances between 3 to 6 mm from the edge of the work piece as investigated by
Hilbinger, 2000 or Stritt et al., 2012. Welds at these positions are typical for flange joints in car body
production.

2. The principle of measurement

In Fig. 1 the temperature field of a crack-free weld (a) and a weld with a centerline crack (b) is shown.
These temperature fields were calculated with finite element numerical welding process modelling using the
routines from Stritt et al., 2012. The applied heat source is calibrated by weld cross sections gained from
experiments done at different edge distances. The simulation is three-dimensional and takes temperature
dependence of heat conductivity, heat capacity and density into account. In the modelled temperature
fields, shown in Fig. 1, two 6000 series aluminum sheets with thicknesses of 1.2 mm on the top and 1.7 mm
on the bottom were processed. The welding process was modelled with a feed rate of 6 m/min, a laser
power of 5 kW, a spot diameter of 650 µm and an edge distance of 5 mm.

Fig. 1. (a) Temperature field of a crack-free weld calculated with a finite numerical welding process modelling; (b) Calculated
temperature field with a centerline crack; (c) Cross-section of an overlap weld from two 6000 series aluminum alloy sheets; Process
parameter for (a), (b), (c): P = 5 kW, v = 6 m/min, df = 650 µm, edge distance a = 5 mm, sheet thickness s = 1.2 mm (top) and s = 1.7 mm
(bottom).
 Part (c) of Fig. 1 shows a weld cross section of the corresponding experiment of two welded 6000 series
aluminum sheets with a typical centerline crack. As can be seen, this centerline crack splits the weld and
therefore prevents the heat from flowing from one side of the crack to the other. To implement the thermal
effect of a centerline crack into the modelling the temperature was set to ambient temperature t = 25°C
permanently and the heat conduction was set to zero permanently at the position of the crack after each
time step.
At the edge of the sheet the heat is reflected resulting in a non-symmetric temperature field as can be
seen for both temperature fields shown in Fig. 1. This causes a heat flow from the edge into the body of the
welded sheets across the weld seam. If a centerline crack occurs the heat flow across the weld is significantly
disturbed. This characteristic change of the temperature field will be used as a principle of measurement for
the method of online crack-detection proposed in this paper.

3. Experiments

3.1. Experimental setup

To prove the above introduced principle of crack diagnostics experiments using an infrared camera with a
frame rate of 50 Hz to analyze the arising temperature field were made. To reduce the dependence on the
local emissivity, the aluminum sheets were coated with a black graphite layer. All experiments were done
with a TruDisk 5001 laser and a fiber delivery with a 200 μm core diameter. The laser has a maximum output
power of 5000 W at a wavelength of 1030 nm. To focus the laser beam on the work piece a PFO 3D scanner
optics with a focal length of 450 mm was used. This yields a focus diameter of 650 μm. No shielding gas was
used.

3.2. Multi-alloy aluminum

Beside the thermo-mechanical effects described in chapter 1, the metallurgical properties of materials are
related to the hot cracking sensitivity. One reason for the crack initiation when welding 6000 series
aluminum alloys is the dendrite-solidification structure. If the residual melt is not able to fill up the volume
between the dendrite network in the solidification zone a crack forms by lack of material. Studies by Borland,
1960 and Feurer, 1977 have shown, that increasing the fraction of silicon in a 6000 series aluminum alloy
raises the ability to fill up these gaps between the dendrite structures. Furthermore welds with aluminum
alloys with a higher silicon fraction result in grain refinement and reduced solidification shrinkage and
therefore a lower hot cracking susceptibility as observed by Coniglio et al., 2008.

Fig. 2. Cross section of two welded AA6xxx series multi-alloy aluminum sheets. Process parameter: P = 5 kW, v = 6 m/min, df = 650 µm,
edge distance a = 5 mm
 Building upon that concept a common way to avoid hot cracks is the use of silicon-rich filler material
investigated by Coniglio et al., 2008.
For the experiments a standard AA6014 aluminum alloy and a multi-alloy aluminum, NF Ac-200RW by
Novelis were used. This fusion-bonded multi-alloy aluminum consists of a core material and two outer
cladding layers as can be seen in Fig. 2. The core material maintains all the advantages and required
properties of an AA6xxx monolithic alloy while the cladding has a high silicon content. During the welding
process the layers are mixing and the hot cracking sensitivity is reduced as proven by Bezençon et al., 2011.
This multi-alloy aluminum is designed for the remote laser welding technology because it makes an
additional filler wire for the welding process obsolete.
Using the multi-alloy and the monolithic aluminum alloy as materials for the experiments, enables a
crack-free and cracked weld seam at the same critical edge distance a = 5 mm which is required to compare
the arising temperature fields under equal conditions.

3.3. Experimental results

Fig. 3 shows measured temperature distributions in false colors based on the intensity measured by the
infrared camera. The color bar goes from blue (cold) to red (hot).

Fig. 3. Comparison of welds without (1) and with (2) cracks: (1a) single frame of infrared video in false colors; (1b) cross section of
welded sheets. (1c) surface plot of grey values from infrared video; (2a) single frame of infrared video in false colors; (2b) cross section
of welded sheets; (2c) surface plot of grey values from infrared video; Process parameter for (1), (2): P = 5 kW, v = 6 m/min,
df = 650 µm, edge distance a = 5 mm.
 Fig. 3 (1a) shows a frame of the infrared video at the end of the welding process for the crack-free weld
with multi-alloy aluminum. Fig. 3 (2a) shows the weld with a centerline crack for the standard monolithic
aluminum. Due to the destroyed graphite layer in the weld seam, the emissivity is decreased there and the
temperature appears lower in that region, even though this is actually not the case.
Comparing the frames (1a) and (2a) in Fig. 3 with the simulated temperature fields in Fig. 1 shows a good
qualitative agreement of the experiment with the expected temperature distribution by the numerical
welding process modelling. For the frame (2a) with centerline crack the dark red area on the edge side is
larger compared to the frame (1a) from the video of the non-cracked weld. The surface plots (1c) and (2c)
generated from these frames confirm this conclusion. The surface plot (2c) with the centerline crack shows a
constant high temperature level on the edge side whereas the surface plot (1c) for the non-cracked sample
shows a temperature decrease from front to the end of the weld. Therefore the difference of the
temperature fields is clearly measurable and can be utilized to detect centerline cracks at close-edge
welding.

4. Utilization of temperature signal to detect hot cracking at close-edge welding

To utilize the measured temperature signal to detect hot cracking during welding a certain criterion has to
be defined to decide whether there is a centerline crack detected or not. Two different possible approaches
will be presented.

4.1. Monitoring the temperature at two positions fixed in space

Fig. 4 (a) shows the position of the measuring point on the edge side and the measuring point on the body
side in the infrared video from the experiments. The measuring points were at the fixed position shown in
Fig. 4 (a).

Fig. 4. (a) Frame of infrared video in false colors of the weld with a centerline crack, P = 5 kW, v = 6 m/min, df = 650 µm, edge distance a
= 5 mm, AA6014 aluminum alloy; (b) The temperature in arbitrary units over time curves of the infrared videos for a measure point on
the edge side and on the body side for both cases the cracked and non-cracked weld.
 The temperature at these positions is plotted as a function of time for the complete weld in Fig. 4 (b) for
both cases cracked (left) and non-cracked (right). At t = 0.5 s the laser beam passes the measuring points and
a strong increase of the temperature occurs. On the edge side (red line) the temperature rises higher
compared to the temperature on the body side (blue line) due to the reflected heat on the edge.
After the pass by of the laser beam the cooling phase starts immediately. Comparing the two charts in Fig.
4 (b) shows a characteristic difference of the cooling curves. If there is no centerline crack the cooling
behavior of both measuring points look very similar whereas with a centerline crack the cooling rate of the
measuring point at the edge is obvious slower (red curve left). When the measuring points are positioned in
the middle of the weld length as shown in Fig. 4 (b) the slope of the cooling curve on the edge side was
found to be lowest. Therefore this position is best to detect a centerline crack due to the big difference in
temperature curves. Monitoring the temperature at two spatially fixed positions can be utilized very simple
with two pyrometers.

4.2. Monitoring the temperature in line with the welding process

Applying the introduced principle of crack detection on moving optics the temperature was measured in
line with the welding process. The temperature field was measured applying a moving coordinate system
according to the movement of the laser beam. Thereby three different phases occur during welding, the
starting phase, the steady state and the cooling phase. During the steady state the heat input is equal to the
heat loss which results in a steady temperature field. At the beginning of the weld and in the cooling phase
after the welding process the heat flow is varying yielding a varying temperature field.
In Fig. 5 temperature line plots along the black lines which are orthogonal to the weld of the calculated
temperature field in the steady state of the welding process are shown. The temperature curves compare
both cases, the cracked weld (red line) and the non-cracked weld (blue line).

Fig. 5. Transverse temperature line plots (b) along the black lines of the calculated temperature field (a) in the steady state of the
welding process. Process parameter: P = 5 kW, v = 6 m/min, df = 650 µm, edge distance a = 5 mm, sheet thickness s = 1.2 mm (top) and
s = 1.7 mm (bottom).
 The position of line plot (1) is at the trailing end of the weld pool where the centerline crack originates.
Except of the dip in the cracked temperature curve due to a point of discontinuity (described in section 2)
the two temperature signals are identical. Meaning there is no differentiation recognizable and therefore no
crack detection possible. At position (2) the heat flow of the line plot with the centerline crack is already
blocked and a temperature step at the position of the crack (dip in red curve) from the edge side to the body
side is noticeable. On the body side the temperature drops due to a lack of heat flow from the edge side. On
the edge side the temperature is rising for both cases due to the reflected heat on the edge. Shifting the
transverse line further to position (3) the heat accumulation increases and the temperature on the edge is
higher compared to the position at the center of the weld for the non-cracked case (blue line). At position (3)
the temperature step, marked with the symbol c, is higher compared to position (2). The higher this
temperature step, the better it can be detected and used as detection criterion for centerline hot cracks.
Fig. 6 (a) shows the temperature fields calculated at three different points in time for the condition with
centerline crack. The distance of the temperature measuring line (vertical black line) to the weld pool
corresponds to position (3) in Fig. 5. Fig. 6 (b) depicts the temperature curve at t = 0.3 s measured along the
black line shown in Fig. 6 (a). Again the temperature step at the centerline crack from the edge to the body
side is observable. To monitor the behavior of the height of the temperature step during the welding process
two measuring points, Tedge and Tbody, were set as shown in Fig. 6 (b). Fig. 6 (c) shows the temperature curves
of these two points over time with the three phases during the welding process. In the cooling phase the two
points stay at the last position of the weld process.
In the beginning of the welding process no crack occurs and the temperature of the two monitored points
are identical. At t = 0.16 s (1) the centerline crack starts to form as was seen in the experiment. Still no
difference between the two temperatures at the measuring points can be seen because they have not
reached the crack yet. At t = 0.2 s (2) the monitored points have the x-coordinate of the beginning of the
crack. From this point in time the heat flow from the edge to the body side is blocked.

Fig. 6. (a) Simulated temperature field at three different points in time; (b) Transverse line plot of temperature curve at t = 0.3 s; (c)
Temperature step at position of centerline crack over time. Process parameter: P = 5 kW, v = 6 m/min, df = 650 µm, edge distance
a = 5 mm, sheet thickness s = 1.2 mm (top) and s = 1.7 mm (bottom).
 The temperature on the body side stays constant whereas the temperature on the edge rises further. At
t = 0.3 s the steady state is reached and therefore the maximum temperature step at the monitored position.
After completion of the welding process the cooling rate on the edge side (red line) is lower compared to the
body side (blue line). This enhances the temperature step at the centerline crack between the body side and
the edge side.

5. Conclusion

The investigations of the temperature field during welding show that welding close to the edge leads to
heat accumulation which results in heat flow into the body. A centerline crack occurring during welding is
changing this heat flow and with it the temperature field significantly. It was found that this characteristic
change can be used as a principle for online crack-detection. To utilize this principle two methods were
discussed.
First, when monitoring the temperature at two spatially fix positions the behavior of the cooling rate can
be analyzed. If a centerline crack occurs during welding, this can be detected by a significant lower slope for
the cooling curve on the edge side compared to the body side. This principle can be utilized very easily
measuring the temperatures at these two positions with two pyrometers. However, it is not simply
practicable for remote welding applications with moving optics.
Second, to utilize the principle of measurement for moving optics the temperature signal was monitored
at a fixed position behind the welding pool. In that case the characteristic feature to detect a weld with a
centerline crack is a significant temperature step when monitoring the temperature across the weld from the
edge into the body. The level of this temperature step is dependent on the monitoring position. Shifting this
position away from the weld pool enhances the level of the temperature step. Yet, this will delay the time of
first detection by the monitor system. Monitoring the temperature at a fixed position behind the weld pool
enables a crack detection on-the-fly during the welding process using remote technology.

References

Bezençon, C., André, P.-D., Bassi, C., 2011, Remote Laser Welding of AA6xxx Multi-Layer FusionTM Material, in proceedings of
LATEST2 Joining Conference.
Borland, J.C., 1960, Generalized theory of super-solidus cracking in welds (and castings), British Welding Journal 7, 508-512.
Coniglio, N., Cross, C. E., Michael, T., Lammers, M., 2008, Defining a Critical Weld Dilution to Avoid Solidification Cracking in
Aluminum, Welding Journal, Vol. 87.
Cross, C.E., 2005, On the Origin of Weld Solidification, Hot Cracking Phenomena in Welds I, pp 3-18, Springer Berlin Heidelberg.
Feurer, U., 1977, Influence of alloy composition and solidification conditions on dendrite arm spacing, feeding and hot tearing properties
of aluminium alloys, Proceedings International Symposium on Engineering Alloys, 131-145
Hilbinger, R.M., 2000, Heissrissbildung beim Schweißen von Aluminium in Blechrandlage, Dissertation Bayreuth, Herbert Utz Verlag.
Pellini, W. S. 1952. Strain theory for hot tearing. Foundry 80: 125–199.
Stritt, P., Weber, R., Graf, T., Mueller, S., Weberpals, J.-P., 2012, New hot cracking criterion for laser welding in close-edge position, in
Proceedings of ICALEO.
Weller, D., Bezençon, C., Stritt, P., Weber, R., Graf, T., 2013, Remote Laser Welding of Multi-alloy Aluminum at Close-edge Position,
in Physics Procedia, volume 41, pages 164-168.