New developments in advanced welding

In-process monitoring techniques for laser welding

The use of optical energy for welding, in the form of a laser beam, offers a number of opportunities for sensing single defects in the process. Thus information is obtained that reflects the processes occurring and the quality of those processes. Figure 5.25 is a diagram indicating the signals that are available from the laser process for in-process monitoring.

For Nd:YAG laser welding systems, applications are partly stimulated by the commercial availability of optical fibre beam delivery systems. This facilitates the integration of laser welding with robotised methods and allows operation on three-dimensional workpieces. During the welding operation, it is very important to monitor the weld quality, reduce the quantity of scrap generated and avoid the possibility of weld failure. This is particularly important in such automotive applications as tailored blank welding or body in white welding, where weld quality and productivity are very important.

5.25 Opportunities for in-process monitoring of laser processes.2

There are a number of potential sensors for weld monitoring in both CO2 and Nd:YAG laser systems available on the market.24 These include optical sensors, which detect either the UV/visible light emitted from the plasma/ plume that forms above the workpiece during welding, or infrared black body emission from the melt pool. A typical Nd:YAG monitoring set-up system is shown in Fig. 5.26 and was developed by Prometec GmbH. This unit monitors the production process continuously, initiating an alarm for the operator and recording an alarm message when a defect from the normal process occurs. The camera, which is mounted in coaxial alignment to the laser beam, can be used with all current high-power laser types. It monitors and documents several critical process characteristics for example, the melt pool dimensions, penetration depth and gaps, as well as laser parameters. The monitoring system can be adapted to further specific manufacturing needs.

Various monitoring systems are in current use in the automotive industry for tailored blank welding. It has been estimated25 that over 55 million blanks are welded each year and this figure is growing rapidly. In pursuit of a solution to the problems of weight reduction and over-specification, the concept of tailored welded blank has become an area of particular interest. A tailored welded blank is produced by welding together two or more pieces of sheet to form a single sheet, which is than pressed into a shape. The following parameters play an important part in producing good quality welds

and it is very important to monitor them during welding. Any variation in them will affect the quality of the welds:

• focus position

• joint gap

• laser power

• spot positioning

• indentations (notches) on the edge of the material.

The welding monitor LWM900, commercially available from Jurca Optoelektronik GmbH, processes signals provided by a range of detectors. Each detector measures a different welding process signal that is very or partially independent of the other detected signals. The use of more than one detector is said to improve the correlation between the weld quality and the monitoring results. In production operation, the LWM900 requires several welds, made under optimised production conditions to ‘learn’ the characteristics of a ‘good’ weld. The LWM900 (using fuzzy logic) detects process disturbances as signal changes relative to the memorised weld reference. It then calculates the probability that an important weld defect has occurred. In such a case,
the calculated probability exceeds a pre-adjusted probability threshold, which is signalled to the machine controller. It is important that the ‘normal’ weld conditions are stable to ensure a stable baseline. If the nominally ‘good weld’ conditions vary from part to part, then the baseline will be noisy and it will be difficult to detect deviations from the norm. For research purposes the detected signals processed by the weld monitor can simply be analysed with respect to any given weld imperfection.

As can be seen from the schematic arrangement shown in Fig. 5.27, four detectors were utilised during the experimental trials. The ‘plasma detector’ sensed the ultraviolet radiation being emitted by the welding plasma/plume. The luminosity of the plasma/plume was then related to weld quality. The

detector for the ‘back-reflected laser light’ senses radiation with a wavelength of 1.06 mm being reflected from the workpiece during welding with a Nd:YAG laser. It is believed a direct relation exists between the amplitude of this signal and the keyhole geometry; the latter is itself correlated to laser intensity and welding speed. In addition, two ‘temperature sensors’ were aligned such that they detected the weld pool temperature by measuring infrared radiation. Changes in the thermal capacity of the workpiece during processing should strongly influence these detectors. Some of the output examples from this system will be highlighted.

Figure 5.28 shows detector traces recorded from a good weld produced under optimum welding conditions. A change in the vertical focus position can alter the welding performance. It is possible to detect the change in the focus position and a typical output is shown in Fig. 5.29. Another laser parameter that can affect the welding performance is the laser power. Reduction in the laser power might be caused during the production process through several possibilities such as laser or beam delivery system component failure or a general reduction of power over time as components become degraded. Typical outputs are shown in Fig. 5.30.

5.8.4 A review of joint tracking systems for laser welding

Owing to relatively low tolerance of laser welding to joint misalignment and gap, joint tracking systems offer the potential for improved quality assurance by allowing adjustment for small variations in joint position and fit-up brought about by standard engineering parts tolerances. Joint tracking systems for laser welding applications usually need to locate the joint to between 0.1mm for 1mm thick sheet and 0.4 mm for 6 mm thick plate. In addition, the joint tracking systems need to be able to operate at welding speeds of between 1 and 10m/min for typical plate and sheet applications. The generic differences between laser and arc welding result in the need for different joint tracking system specifications for laser welding. Five types of sensor are generally used for joint tracking.

Tactile sensors

Tactile sensors use a probe or stylus in direct contact with the workpiece; they position the welding torch at the proper location with respect to the joint by either mechanical or electromechanical means. Tactile sensors are of limited use in laser welding, primarily due to the requirement to locate in a mechanical corner and secondly because information is only provided on joint position.

Vision based sensors

Vision based sensors comprise the greatest percentage of joint tracking systems used today. There are several variants on these systems, but all are centred on the use of a charge-coupled device (CCD) image sensor. At present the sampling rates of the CCDs used in seam tracking equipment are insufficient for some laser welding applications. High-speed cameras are available, but currently at an inadequate size and a substantial cost penalty.

Ultrasonic sensors

Ultrasonic monitoring using a sensor in contact with the workpiece is potentially capable of simultaneous joint tracking in square edge butt joints. However, coupling of the sound waves to the work piece can only be achieved using gel, grease or water. In addition, the sensor needs to be accurately maintained at a constant distance from the weld pool. Airborne ultrasonic sensors are under consideration for future applications but development of suitable air coupled transducers is necessary.

Eddy current sensors

Eddy current sensors use an inductive coil, which sets up a magnetic field in the material and a detector to monitor the field strength in various positions. It is a non-contact device and produces a continuous signal that can be monitored. However, it can only be used with ferrous materials. A number of systems aimed primarily at arc welding applications are currently commercially available. In laser welding, the main concerns involve accuracy, due to local variations in field strength, which can be caused by variable surface quality, and through inherent signal interpretation problems.

Plasma sensors

Plasma sensors use the slight charge in the welding plume to measure the voltage drop from an isolated nozzle to the workpiece. This charge varies with plume intensity and therefore can be correlated with weld quality. It may, therefore, be possible to use this voltage drop to detect variations in joint position. Potential problems, however, lie in the resolution of the technique. The voltages are very low and variations are in the order of millivolts. It is questionable whether there are variations with mistracking and, if there are, whether they are detectable.