A steel maker and a research centre have developed an automated
system for monitoring the quality of welds in industrial seam-welding.
An optical pyrometer senses the temperature of the weld as it emerges
from the machine. The weld is acceptable if the average and standard
deviation of the temperatures along its length both fall within set
values. Studies show excellent correlation with traditional inspection
techniques. The system can also detect any gradual deterioration in
the weld quality, indicating the need for preventative maintenance
of the welding machine.
When done correctly, welding is a strong
and reliable means of joining two pieces of metal and is widely
used in manufacturing industry. However, strict quality control
is essential if welds are to made to a high and consistent standard.
Traditionally, a weld is tested by visual inspection and by tapping
it with a hammer to check that the two metal edges have fused properly.
Though effective, these tests are qualitative and partly subjective.
While it is possible to monitor automatically various process parameters
such as electric current, pressure, speed, and so on, none of these
factors are directly related to the quality of the weld after it
The Steel Research Project started when Belgian steel-maker Cockerill
Sambre approached CRM, a metallurgical research centre in Liège,
for advice on how to avoid breakages in their annealing line at
Kessales. Steel strip is seam-welded together in lengths of about
one metre in a continuous and automated process. Faulty welds require
the process to be halted, and in some cases the plant may be shut
down for up to 40 hours. Cockerill Sambre wanted a fast and objective
method for assessing weld quality and also a means of monitoring
the condition of the welding machine to warn of impending trouble
before faulty welds started to appear.
A good weld requires the metal to be heated uniformly to melting
temperature, so CRM examined the possibility of monitoring the temperature
of the weld as it emerges from the machine. At the temperatures
involved, 600°C to 1000°C, the weld would emit strong infrared radiation,
so CRM decided to use an infrared camera to make a thermal image
of the newly made weld.
The camera is mounted to view the weld at a point 50 mm after the
melting position. It scans a strip 0.5 mm wide and 20 mm long across
the width of the weld. As the metal passes out of the machine, the
camera records many such 'profiles', building up a two-dimensional
contour map. The brightness of the infrared emission is directly
related to the temperature, so CRM was able to construct a 'thermal
map' showing how the temperature varies across and along the weld.
A typical map has about 300 profiles for a one-metre length of weld.
The next step was to use the map to assess the quality of the process.
CRM had to establish that the temperature recorded on the map did
indeed correspond to the strength of the weld. They made 150 sample
welds, of varying strengths, and submitted them to the standard
Erichsen test, in which a welded plate is squeezed into a cup until
it breaks. The tests showed that the temperature measured by the
camera was a good indicator of the strength of the weld.
Next, they compared the maps for 4,000 welds with the results of
traditional visual inspection and hammer tests. They found that
the thermal map for each weld could be characterised in a simple
way. First, the maximum temperatures of each profile are averaged
over the length of the weld to produce an average temperature for
the whole weld. Second, the variation of the maximum temperature
from profile to profile, defined by the standard deviation, is calculated.
A weld is deemed acceptable if its average temperature and the standard
deviation of the temperature fall within specified limits.
The computer keeps a record of the last 2,000 welds and alerts the
operator if there is any drift in weld temperature which may indicate
that the process is deteriorating. Appropriate maintenance can then
take place before the quality becomes unacceptable.
Although the system worked well, there were several practical problems
that needed to be overcome. The camera was bulky and cumbersome
and needed mirrors to view the weld without obstructing access for
maintenance. The scanning mechanism used rotating prisms which required
regular attention, and the optical window became contaminated with
smoke from the welding process and needed frequent cleaning. Being
so close to the welding point, the camera suffered from electrical
interference. It turned out to be an unnecessarily complicated and
very expensive solution to the problem.
When the original project finished in 1992, CRM sought further
EU funding to come up with a cheaper and more rugged version of
the system that could be developed into a marketable product. The
new equipment, which has been under development since 1994, uses
an optical pyrometer instead of a camera.
A pyrometer is a simple instrument that measures temperature by
directly sensing the amount of thermal radiation from an object.
Instead of a camera, a lens focuses radiation on to a 3-metre optical
fibre which transmits it to an infrared detector housed some distance
away in a strong cabinet. The signal from the detector is sent by
another optical link to the computer controlling the welding process.
The process operator is able to view the temperature 'signature'
of each weld and observe how it compares with the acceptance criteria,
which are the same as for the infrared camera system.
Unlike the camera, the pyrometer does not scan across the weld to
make a profile, but receives radiation from a fixed circular spot
10 mm across. This means that an average temperature across the
weld is sensed, rather than a detailed profile. One concern was
that the system would be vulnerable to imprecise positioning because
the electrodes on the welding machine are changed every week and
this alters slightly the position of the centre line of the weld.
However, this does not seem to be a problem in practice.
Not only is the pyrometer cheaper than the camera, but it is less
intrusive, the optical fibres are not susceptible to electrical
interference, and the whole system is simpler to install and easier
The partners are about to start work on a similar system for use
on a tinning line. Here the metal is so thin (0.1-0.3 mm) that the
hammer test would damage it. The lower temperatures, 350-500°C,
require a different kind of detector in the pyrometer, but essentially
the system will work in the same way.
Several enquiries about the system have come from interested companies,
but CRM, being a research centre, is not in the manufacturing business.
Instead, it intends to licence the system to the Belgian company
IRM for commercial manufacture. Potential customers include not
only steel makers, but a wide range of firms in the manufacturing
industry where quality control of welding is important.