Subscribe to our newsletter to receive the latest news and events from TWI:

Subscribe >
Skip to content

In-process weld quality monitoring of laser and hybrid laser-arc fillet welds in 6-12mm C-Mn steel (June 2007)

   
G. Shi, P. Hilton and G. Verhaeghe

Paper presented at Proceedings of the Fourth International WLT-Conference on Lasers in Manufacturing 2007 (LIM2007) , Munich, 18 - 22 June 2007

Abstract

This paper describes the result of work carried out at TWI to assess the effectiveness of a range of commercially available photodiode based sensors for detecting, in real-time, engineered imperfections, when laser welding and hybrid laser-MAG welding in the T-joint configuration in medium to thick section (6-12mm) steels. Welding procedures were developed for welding fully penetrating T-joints between 6mm thickness C-Mn steel stiffeners and 8mm thickness C-Mn steel base plates and between 8mm thickness C-Mn steel stiffeners and 12mm thickness C-Mn steel base plates, using autogenous Nd:YAG laser welding and hybrid Nd:YAG laser-MAG welding. Three laser process monitoring sensors(sensitive in the infrared and ultraviolet wavelength and at the Nd:YAG laser wavelength) were integrated with commercial laser welding equipment. Weld imperfections, likely to occur during laser and hybrid laser-MAG welding of stiffened structures were simulated by varying joint fit-up conditions and welding parameters, to establish the performance of the monitoring sensors.

1 Introduction

Laser beam welding, as a substitute for conventional welding, provides advantages in terms of high speed, high productivity and low heat input. However, the accuracy in part fit-up required for autogenous laser welding is not always possible to achieve in today's production processes, which require flexibility and automation, while maintaining a high standard of weld quality. More tolerant welding procedures and/or process control must be integrated into automated welding systems to maintain weld quality. The combination of an electric arc with the laser beam can significantly improve the gap bridging capability of the laser process, but whatever process is used, a first stage in any adaptive process control, is to be able to effectively monitor the welding process in order to provide the necessary feedback parameters. For in-process monitoring of laser welding, most current techniques employ a variety of sensors to monitor electromagnetic signals from the molten pool during welding, with the objective of correlating the output from the sensor to features such as weld penetration, weld pores or pin holes, and the weld shape. [1-2]

The availability of all the monitoring equipment has certainly contributed to the industrial acceptance of laser welding, particularly in thin sheet metal fabrication, such as found in the automotive industry, where laser welding is accepted as a process capable of achieving high weld quality with high productivity. But this is not yet the case for medium to thick section fabrications, where laser welding is still not widely exploited. Quality control procedures in these thicknesses are still predominantly based on post-weld destructive testing, which only allow spot checks of the manufactured parts. With the development of new high power laser sources and increased use of advanced processes such as hybrid laser-arc welding for large section fabrication, a reliable and on-line weld monitoring system is useful for laser and hybrid laser-arc process monitoring and quality control.

This paper describes the result of work carried out at TWI as part of a EU funded project. The main objective of work was to assess the effectiveness of a range of commercially available photodiode based sensors to detect, inreal-time, engineered factors causing weld imperfections, when welding T-joints in medium to thick section (6-12mm) steels.

2 Experimental work

2.1 Materials

The steel plate used in this work was 6mm, 8mm and 12mm thick S355J2G3 steel, in accordance with EN10025. The steel plate was laser cut into samples 300mm long and 50mm wide. All the samples were sandblasted to remove surface scale.One edge of each plate was milled to facilitate close-fitting square-edge T-joints. A 1.0mm diameter A18 C-Mn steel filler wire (EN440) was used for the hybrid laser-MAG welding trials.

2.2 Simulation of weld imperfections

Based on knowledge of laser welding and hybrid laser-MAG welding and discussions with end users of laser welding, potential parameter variations, which might be expected to arise in a production, were established. These factors, associated with process parameters, joint preparation and fit-up, are likely causes of different weld imperfections. In different ways, these imperfections were engineered into the T-joint welds to examine the detection sensitivities of the chosen sensors.

For autogenous laser welding, the main weld imperfections investigated were lack of side-wall fusion, lack of root fusion, incomplete penetration, surface pores, porosity, undercut and spatter. These imperfections were simulated by joint face contamination, variations in joint fit-up and variations in laser process parameters. Welding trials were conducted on 6mm to 8mm T-joints to examine whether the sensors could detect the engineered factors resulting in weld imperfections. The engineered factors included:

  • Addition of metal debris (a small piece of silver solder wire) in the joint.
  • Localised contamination with grease.
  • Change of laser focus position relative to the joint surface.
  • Change of laser beam position on the stiffener.
  • Change of plume suppression gas flow rate.

For hybrid laser-MAG welding, the main weld imperfections investigated were lack of side-wall fusion, lack of root fusion, incomplete penetration, surface pores, porosity, under cut, incorrect weld toe, excess weld metal and spatter. These imperfections were simulated by changes to the workpiece geometry, joint face contamination, joint misalignment, hybrid welding configuration and hybrid process parameters. Welding trials were conducted on 8mm to 12mmT-joints to examine whether the sensors could detect the engineered factors resulting in weld imperfections. The engineered factors included:

  • Addition of metal debris (a small piece of silver solder wire) in the joint.
  • Localised contamination with grease.
  • Tapered 0-1.2mm joint gap along the joint line.
  • Change of laser focus position relative to the joint surface.
  • Change of laser beam position on the stiffener.
  • Change of arc current (wire feed rate) during welding.
  • Change of arc voltage during welding.
  • Switch off of the arc process during hybrid welding.

2.3 Experimental set-up

The autogenous laser welding trials were carried out using a 4kW CW Nd:YAG laser, manufactured by Trumpf. The laser beam was delivered via a step index optical fibre, 600µm in diameter, to an output housing utilising a 200mmfocusing lens, to produce a nominal minimum spot size of 0.6mm in diameter. A Servo-Robot seam tracker was used to follow the joint seam during welding. A plume suppression gas jet (argon) was angled at 40° to the surface of the workpiece, trailing the laser beam, with an impingement point 1mm above the laser beam focus. A high pressure air knife was used to protect the laser optics during welding. The actual housing and seam tracker were mounted onto a JS30Kawasaki robot, traversing the arrangement over the specimen, held in a stationary jig, during welding.

Hybrid laser-MAG welding trials were carried out using the same laser, in combination with a Lincoln PowerWave 450 synergic MIG/MAG welding power source. The same beam delivery and focusing optics used in the autogenous laserwelding trials were used in all the hybrid laser-MAG welding trials. The MAG torch was attached to the laser output housing to provide an arc travel angle of 30° and a 16mm contact tip to workpiece distance.

For hybrid laser-MAG welding, the shielding gas used through the MAG torch was an Ar-20% CO 2 mixture. The gas flow rate through the MAG torch, was 15l/min for all the hybrid laser-MAG welding trials. No other shielding gases were used.

All the welding trials were carried out using the same laser power of 4kW. The joints were degreased with acetone and clamped prior to welding. The samples were tack welded in the jigging arrangement, using the MAG process, to maintain the required joint fit up and reduce movement of material during welding.

2.4 In-process monitoring of the welding

Three different sensors, manufactured by Precitec Optoelektronik GmbH, were used in this work.

The first sensor, described by Precitec as the 'plasma' sensor, detects radiation in the wavelength range less than 600nm. When welding steel with Nd:YAG laser light a high temperature, thermally excited gas or 'plume' is expected rather than an ionized plasma. [3,4] In practice, these plumes are very energetic in the visible part of the spectrum and hence it was hoped that any fluctuations in the Nd:YAG generated plume, due to problems during welding would be visible using the plasmasensor. The second sensor, referred to as the 'temperature' sensor, is sensitive in the infrared part (1100-1800nm) of the spectrum and was arranged to pick up signals from the molten weld pool. The third sensor, referred to as the'back reflection' or 'reflectivity' sensor, detects a narrow band of radiation centred around the wavelength of 1.06µm ( i.e. the laser wavelength) reflected from the workpiece during processing.

The 'plasma' detector was positioned in the cabinet of the laser source to detect radiation from the weldpool passing back through the quartz optics of the process head, and eventually reaching the output end of the laser beam delivery optical fibre. The 'temperature' and 'back reflection' sensors were mounted on the side of the beam focusing head, signals from the weldpool reaching these sensors via the laser beam focusing lens and a dichroic mirror, positioned between this lens and the collimating lens as shown schematically in Fig.1.

spgsjune07f1.gif

Fig.1. Photodiode based multi-sensor system (plasma, temperature and back reflected laser light)

The signals from the sensors were amplified as necessary and fed into a Precitec LMW900 weld monitor. The monitor allowed the signals to be displayed as amplitude versus time graphs, for each sensor; the time axis corresponding to distance along the weld. The amplitude of the sensor signals is displayed in arbitrary units. The frequency of sampling was 5kHz.

3 Experimental results and discussion

3.1 Autogenous laser welding with in-process monitoring

Welding trials were initially carried out to develop optimised conditions for achieving fully penetrating welds in 6-to-8mm T-joints. Imperfection free reference welds were produced and the sensor signals received were recorded during welding. These signals were then compared to welds made containing imperfections assess the performance of the sensors.

Figure 2 shows the recorded sensor signals from a typical autogenous laser weld produced using the reference laser welding conditions. The particular weld corresponding to the trace in Fig.2, had a smooth top and under bead and was free from visible imperfections. No obvious step changes or peaks in the signals from the various sensors were seen. Radiographic examination of the weld indicated that it was free of pores larger than 0.5mm diameter, no assessment of pores less than this having been made.


spgsjune07f2a.jpg

Fig.2. Sensor signals from the temperature, and plasma sensors, arising from the reference autogenous laser welding conditions for a 6mm stiffener plate welded to an 8mm base plate:

a) Weld top bead;

spgsjune07f2b.gif

 

b) Temperature sensor response;

 

spgsjune07f2c.gif

c) Plasma sensor response

Figure 3 shows the sensor signals received from a T-butt weld produced using the reference set of welding conditions but with silver solder wire added to two areas in the joint. Both the recorded plasma signal and temperature signal showed changes corresponding to the placement of the wire, which resulted in the formation of surface pores, which can be seen in Fig.3a.


spgsjune07f3a.jpg

Fig.3. Sensor signals received from a 6mm stiffener welded to an 8mm base plate made with two small pieces of silver solder introduced at the joint line:

a) Weld top bead;

 

spgsjune07f3b.gif

 

b) Temperature sensor response;

 

spgsjune07f3c.gif

 

c) Plasma sensor response

 

Tests were also carried out to simulate different imperfections by changing process parameters and joint fit-up conditions during laser welding, such as changes in laser beam focus (causing incomplete penetration), beam to joint misalignment (causing incomplete penetration and lack of fusion), and variation in plasama suppression gas flow rate (causing incomplete penetration).

It was noted that, for this particular joint/material combination, the temperature sensor detected most of the changes in joint fit up, process parameters and contamination of the joint face. It provided responses to added solderwire and grease (causing surface pores and excessive spatter), with step changes in the amplitude of the signal. The responses to the changes in characteristics of the optical signals caused by addition of wire and grease to the joint were different. The temperature sensor showed a step decrease in the area with added solder and a step increase in areas with applied grease. The temperature sensor was also sensitive to variations in laser focus position and changes in laser beam position on the stiffener plate (caused incomplete penetration and lack of root fusion in welds). This sensor also responded to changes in laser focus (causing incomplete penetration), with gradual increase in the amplitude of the signal as the laser was moved away from the optimised focus position. A similar behaviour was observed when the position of laser beam on the stiffener plate was varied. The sensors could detect a changes in emitted radiation caused by a change of laser beam position on the stiffener plate, (resulting in lack of penetration) when the weld top bead showed no obvious changes.

For autogenous welding the plasma sensor showed a similar capability to detect variations in emitted radiation caused by changes in joint fit-up and surface condition as the temperature sensor, apart from not being able to detect the results of adding grease (causing excessive weld spatter). However, the responses from the plasma sensor to some of the engineered anomalies were different to those from the temperature sensor in terms of the magnitude of variation in recorded intensity, with respect to the signals from the reference weld. For example, the temperature sensor exhibited an increase in amplitude to changes in laser focus position (causing incomplete penetration). In contrast, the plasma sensor responded to the same change with a decrease in amplitude.

It is also interesting to note that the temperature sensor could clearly detect changes in the optical signals associated with adding grease (resulting in excessive weld spatter), whereas the plasma sensor showed little response to this. The combination of the plasma sensor and the temperature sensor was able to detect changes in the emitted radiation corresponding to all the simulated imperfections studied.

3.2 Sensor performance during hybrid laser-MAG welding

Similiar to autogenous welding, hybrid laser-MAG welding trials were carried out to develop optimised welding conditions for achieving fully penetrating welds on 8-to-12mm T-joints. Sensor signals recorded from the reference weld and welds with engineered imperfections were compared to assess the perfromace of these sensors for monitoring weld imperfections dring hybrid laser-MAG welding.

Figure 4 shows the recorded sensor signals from a typical weld produced using the reference hybrid laser-MAG welding conditions. The weld had a smooth top and under bead and was free from visible imperfections. A smooth response in signal was recorded for the full length of the weld. Radiographic examination indicated that the weld was free of pores, greater than 0.5mm diameter. The weld quality was classified as Class B in accordance with BS ENISO13919:1 - 1997, in terms of profile and porosity levels.


spgsjune07f4a.jpg

 

Fig.4. Sensor response from a hybrid laser-MAG arc weld made using the reference welding conditions, for an 8mm thickness stiffener welded to a 12mm thickness base plate:

a) Weld top bead;

 

spgsjune07f4b.gif

 

b) Temperature sensor response;

 

spgsjune07f4c.gif

c) Reflectivity and plasma sensor response.

Figure 5 shows sensor signals received from a hybrid weld produced by changing the axial position of the beam focus along the beam incident axis from 2mm at the weld start to 12mm at its end. It is clear that all three sensors were sensitive to this change in laser focus position. The amplitude of the signals from the sensors gradually increased as the laser beam moved away from the optimised focus position. Visual examination of the resulting weld indicated that the weld bead profile was similar to that of the reference weld for a laser beam focus position from 2 to 4mm above the stiffener plate, i.e. for the first third of the weld. There was little change in the sensor signals in this part of the weld ( Fig.5a). In the second third of the pass, the weld still showed full penetration, but on the top of the weld, incorrect weld toe could be seen. In this region, the amplitude of all three detectors can be seen to rise. In the final part of the weld, the top bead became wider and more concave and weld penetration was eventually lost. In this region, the amplitude of the signals continued to rise for all three detectors, but no obvious step-change in signal could be seen on the change from full to partial penetration.

spgsjune07f5.jpg
 

Fig.5. Sensor response from a hybrid laser-MAG weld made at the reference conditions but changing the axial position of the beam focus from 2mm away from the joint position at the start of the weld, to 12mm away fromthe joint position at the end of the weld. (8mm thickness stiffener and 12mm thickness base plate):

a) Weld top bead;
b) Weld under bead;
c) Temperature sensor response;
d) Reflectivity and plasma sensor response

Tests were also conducted to simulate various imperfections by introducing contamination into the joint, variations in joint fit-up and hybrid laser-MAG welding parameters, to induce undercut and irregular weld bead profile, beam to joint misalignment caused incomplete penetration and lack of fusion, variations in wire feed rate to introduced excess weld metal, variations in arc voltage to introduce weld spatter and irregular weld profile and changes in shielding gas flow rate to induce porosity. The results show that the hybrid laser-MAG process exhibited larger tolerance to joint contamination and joint gap, compared with autogenous laser welding. The added grease, (causing excessive spatter in autogenous laser welding), did not result in any observable imperfection during hybrid laser-MAG welding. Similarly, no imperfections were found in the hybrid laser-MAG weld produced with 0 to1.2mm variable gap. The sensors did not show any changes in these two cases. In addition, during the hybrid welding trials, the reflectivity sensors were not saturated as was the case for autogenous laser welding and it is therefore possible to include results from this sensor in this discussion. A possible explanation of the attenuated signals reaching the reflectivity sensor in hybrid welding is that the introduction of the MAG process enhanced the absorption of the back reflected laser energy. This phenomenon has not been confirmed however, as part of this work.

Results from the plasma and temperature sensors during hybrid laser-MAG welding indicated differences when compared to the results from these sensors in the autogenous welding trials. If the results for the reference welds are compared, then for hybrid laser-arc welding the signals from the plasma detector appear to be of a lower relative intensity, but are also less noisy. In comparison, for hybrid laser-MAG welding the signals from the temperature detector seem to be slightly higher in average intensity, and noisier, than those found during autogenous welding.

For hybrid welding, the plasma sensor exhibited clear responses to changes in joint fit-up, surface condition and process parameters causing undercut, incomplete penetration, excessive weld metal, excessive weld spatter and possibly internal porosity. The largest response from the plasma sensor was from the added silver solder, causing regions of undercut, and changes in arc current (wire feeding rate) resulting in the formation of excess weld metal. Changes in laser beam (and MAG torch) position with respect to the joint line could lead to lack of root fusion and incomplete penetration in the weld. This change could be detected by the plasma sensor. When the axial position of the beam focus was changed, it was possible to detect changes in the sensor signal corresponding to the change from an acceptable top bead weld profile to an irregular weld profile, and from an irregular weld profile, to partial penetration. This behaviour was similar to the results for autogenous welding.

For hybrid welding, the temperature sensor exhibited clear responses to factors causing undercut, lack of penetration, excess weld metal and excessive weld spatter. In a similar manner to the response from the plasma sensor, the largest response by the temperature sensor was from changes in arc current, causing excess weld metal and variations in laser focus, causing lack of penetration.

The temperature sensor, however, did not show any response to changes in the position of the laser beam with respect to the joint line, resulting in lack of root fusion. When the beam focus position with respect to the stiffener was kept constant but the focus position away from the base plate was varied from 0 to 5mm, the plasma and reflectivity (to a lesser extent) sensors measured changes corresponding to the appearance of lack of fusion on the top bead whilst still achieving full penetration, to the condition where the beam was so high on the stiffener that no fusion was achieved in the base plate, effectively resulting in a fully penetrating bead-on-plate run. For this particular trial, no change at all was registered by the temperature sensor. The plasma sensor, exhibited a very small increase in the amplitude of its signal when lack of root fusion occurred. This indicates that for this particular joint configuration, factors causing lack of side-wall fusion will be more difficult to identify with these sensors, compared to the other imperfections tested in this work.

It has been claimed [5] that the reflected laser light, ie that light detected by the reflectivity sensor, is an indicator of weld penetration, as the amount of reflected light correlates with the aspect ratio of the keyhole. If a constant diameter keyhole is maintained, the aspect ratio of the keyhole only depends on the keyhole depth. The deeper the keyhole, the more keyhole surface is available for absorption of the laser beam. For this particular joint/material combination, ie a steel T-joint, the maximum speed for achieving fully penetrating welds was when the laser beam was focused 2.0mm from the stiffener plate surface and 1.0mm away from the base plate surface. If the argument is used that when the laser beam is moved away from an optimized position, the weld penetration (keyhole depth) decreases, then this should correspond to an increase in reflected light. The reflectivity sensor has clearly picked up the changes in signal caused by variations in laser focus position, resulting in incomplete penetration in the weld, as shown in Fig.5. The reflectivity sensor did not indicate any significant changes when the laser focus position was kept unchanged, (unless it detected some other anomaly).

4 Conclusions

The following conclusions could be draw from this work:

For both thicknesses of stiffener (6mm and 8mm), it was possible to produce a set of reference welding conditions, for both the autogenous and hybrid laser-MAG processes, which provided repeatable, imperfection-free, fully penetrating welds, for comparison purposes. In this series of trials, the welding speed using the hybrid laser-MAG process was no greater than that found using the autogenous process, but the hybrid laser-MAG process was found to be much more tolerant to joint gap and joint misalignment, than the laser process.

The laser process monitoring system has shown itself effective to be an indicator of various factors causing weld imperfections during autogenous laser welding. The plasma sensors were able to pick up factors causing surface pores, incomplete penetration, and undercut during laser welding. The temperature sensor showed a similar behaviour to the plasma sensor, in terms of response to weld imperfections caused by changes in beam to joint alignment, but gave a larger response to joint contamination (causing excessive weld spatter) than the plasma sensor. The largest response seen from both the plasma and temperature sensors was from factors causing surface breaking pores and incomplete penetration.

For this particular joint/material combination, it was also possible to detect various factors causing weld imperfections during hybrid laser-MAG welding. The temperature and plasma sensors were able to detect the results of changes in arc parameters (current, voltage and shielding gas flow rate) causing excess weld metal, excessive weld spatter and incorrect weld toe, changes in beam to joint alignment and variation in axial focus position, resulting in undercut and incomplete penetration. The plasma sensor also showed responses to insufficient gas shielding, causing porosity in the weld. Both sensors exhibited the largest response to variations in arc parameters causing excess weld metal and spatter. The reflectivity sensor showed significant response to changes in laser focus position, resulting in incomplete penetration.

Bibliography

  1. Jokinen T, Vihervä T, Riikonen H and Kujanpää V: 'Welding of ship structural steel A36 using a Nd:YAG laser and gas-metal arc welding'. Journal of Laser Applications 2000 12 (5) 185-188.
  2. Jokinen T, Vihervä T, Riikonen H and Kujanpää V: 'Welding of ship structural steel A36 using a Nd:YAG laser and gas-metal arc welding'. Journal of Laser Applications 2000 12 (5) 185-188.
  3. Ales A: 'On-line monitoring of laser welded steel tailor blanks', TWI Members Report, 698/2000.
  4. Greses J and Hilton P A: 'Spectroscopic studies of plume/plasma in different gas environments', Proceedings ICALEO 2001.
  5. Farson D F and Kim K R, 'Optical and acoustic emissions in laser welding', ICALEO 1998, pp.C1-C10, 1998.

For more information please email:


contactus@twi.co.uk