[Skip to content]

TWI
Search our Site
.

Sensor systems for top-face penetration control

TWI Bulletin, November/December 1995

 

Paul Anderson
Paul Anderson

Paul Anderson is a Senior Research Welding Engineer in the Arc Welding Section of the Arc, Laser and Sheet Processes Department at TWI. He joined TWI in 1990 and has been actively involved with the development of gas-shielded arc welding processes. His recent activities include the development of shielding gases for the TIG welding of duplex stainless steels and assembly of a prototype top face penetration control system.

Paul Anderson reviews the available literature on sensor systems for the top-face monitoring of weld pool penetration in arc and laser welding and identifies candidate sensors for the development of a top-face feedback system for controlling weld penetration.



Problems are frequently experienced in ensuring consistent penetration of a joint, due to minor variations in material composition, component dimensions, joint fit-up or the surface condition of the component. In manual welding the welder can adjust the welding parameters, or his technique, to compensate for some of the variations. In mechanised and automated welding, the welding parameters are normally fixed prior to welding. A sensor which observes the back of the weld and modifies the welding parameters to ensure consistent penetration is commercially available. [1] However, for many applications, access to the back face of the weld is not possible.

There is a need for a sensor system to control accurately the weld bead penetration from the top-face, in both mechanised and automated welding. For the system to be widely applicable in the fabrication industry, it should ideally be able to accommodate:

  • Arc and laser welding processes
  • Flat, vertical and overhead welding positions
  • Butt and fillet welds
  • A broad range of materials, including C-Mn steels, stainless steels and aluminium alloys
  • Full and partial penetration welds

Of equal importance, the sensor should not make contact with the workpiece, or intrude into the welding area.

Sensor systems

Ultrasonic techniques

The depth of penetration of a weld pool can be measured by locating the fusion interface between the weld pool and parent metal using ultrasonic waves, generated by an angled piezoelectric transducer in contact with the surface, Fig.1. [2] Accurate measurements from the top face can only be obtained for plate thicknesses of greater than 10mm or less than 2mm. To ensure good coupling between the transducer and the workpiece, the workpiece must have a simple, uniform surface geometry with minimum irregularities. Distortion of the workpiece must be avoided. The transducers must track the weld pool, and component vibration or unexpected probe movement must be prevented.

Fig. 1. Ultrasonic penetration sensing. The deeper the weld pool, the shorter the total beam path
Fig. 1. Ultrasonic penetration sensing. The deeper the weld pool, the shorter the total beam path

Acoustic emission techniques

During welding, stresses which form in the workpiece due to the thermal gradient about the weld and changes in the volume of fused metal, generate acoustic emissions, which travel throughout the workpiece. As penetration changes from partial to full, the frequency spectrum of the acoustic emission changes. The acoustic emissions can be measured using a stationary acoustic transducer remote from the weld [3] or, for laser welding, a non-contact pressure wave sensor. [4] However, the relationship between the acoustic emissions and the weld penetration depth is not fully understood. Also, a number of microstructural features may influence the performance of these techniques.

Weld pool sag sensing techniques

As full penetration is achieved in the flat position, the weld pool surface begins to sag towards the back of the plate. This movement can be detected through the increase in arc length, [5] Fig.2, or the change in profile of the weld pool. However, the magnitude of the movement is very small, which limits the accuracy attainable. The accuracy is reduced as the workpiece thickness increases, and by oscillations within the pool due to pulsed welding current, power source ripple, and the addition of filler wire to the weld pool.

Fig. 2. Voltage sensing of weld pool slag
Fig. 2. Voltage sensing of weld pool slag

Weld pool oscillation sensing

When a pulse is applied to the weld pool, the surface of the pool vibrates. The oscillation frequency is dependent upon the weld pool size, and is higher in partial penetration welds than full penetration welds. [6] The oscillation frequency can be monitored by measuring fluctuations in either the arc voltage or the arc light reflected from the weld pool. Accuracy is reduced as the travel speed increases, and by power source ripple, slag islands, the presence of dirt, slag or oxide at the back of the weld pool and any addition of filler wire to the weld pool.

Weld pool viewing techniques

This technique assumes that if the weld pool size on the top-face remains constant, then the penetration is constant. [7] The weld pool may be viewed using a CCTV or infra-red camera, which can be incorporated into the welding torch, Fig.3. The performance in practice is promising, although the differences observed in the weld pool may not compensate for all combinations of variations due to joint fit-up, material composition or surface contamination. Accuracy is reduced by contributions from the arc light which necessitates the use of filters or selective viewing by blocking out the arc.

Fig. 3. Weld pool viewing with coaxial torch
Fig. 3. Weld pool viewing with coaxial torch

Thermal sensing techniques

If the thermal gradient around the weld is constant, this technique assumes that the weld pool penetration is constant.

Temperature may be measured using a contact thermocouple or an infra-red pyrometer, or a camera observing the workpiece. Thermal sensing is probably the most widely applied technology in practical use and can also provide seam tracking. [8] Unfortunately, these systems are particularly sensitive to variations in the surface condition and the emissivity of the material. The accuracy of camera-based systems is reduced by contributions from the arc light - this necessitates the use of filters or selective viewing by blocking out the arc.

Discussion

The range of applications within which the sensor systems still function is summarised in the Table. Although some sensors have been applied in specific applications, they all have limitations to their capabilities. None of the systems appeared to be capable of fulfilling all of the above criteria for the ideal sensor system.

Conclusions

There is currently no system commercially available which satisfies all the requirements of an ideal system. The range of application and accuracy of control of the reported systems can be summarised as follows:

  • Ultrasonic techniques are only suitable for plate thicknesses of less than 2mm or greater than 10mm. Accuracy of control needs to be improved before they can be used for practical applications.
  • Acoustic emission monitoring will function independently of welding position, process and joint type, but has still to be effectively demonstrated.
  • Weld pool sag sensing is limited to butt welds in the flat position and requires a highly accurate power source.
  • Weld pool oscillation frequency monitoring by optical systems can only be used for full penetration butt welds, and requires the back surface of the weld to be free of oxide and contamination.
  • Direct weld pool viewing functions independently of position, process and joint type, but requires accurate control of component thickness and joint fit-up.
  • Thermal sensing functions independently of position, process and joint type, but its accuracy is reduced by variations in the surface emissivity of the material.

Recommendations

Use of a combination of the generic sensor methods to overcome the limitations of individual methods should be evaluated. Possible combinations of weld pool viewing, thermal sensing and acoustic emission monitoring should be considered. These methods are not limited by the welding position, process and joint type, and the limitations of each method can be overcome using data generated by the other methods. As acoustic emission monitoring has yet to be effectively demonstrated in practice, the most promising option for monitoring and controlling weld penetration appears to be simultaneous pool viewing and thermal sensing, which can be carried out with a single infra-red camera.

Future work

A system based upon simultaneous pool viewing and thermal sensing with an infra-red camera, is at an advanced stage of development.

Capabilities of the reviewed sensors

Generic
technique
Sensing
method
Ability
proven
Sensitivity
to other
factors
Welding
processes
Joint
types
Full/
partial
penetration
Welding
positions
Ultrasonic Shear waves,
contact probe
Yes Low TIG,
MIG
Butt,
fillet
Partial All
Shear waves,
pseudo-
immersion
Yes Low TIG,
MIG
Butt,
fillet
Partial Flat only
Rayleigh
waves
No Low TIG,
MIG
Butt,
fillet
Full,
partial
All
Lamb
waves
Yes Low TIG,
MIG
Butt,
fillet
Full,
partial
All
Acoustic
emission
Contact
probe
No Medium TIG,
MIG,
Laser
Butt,
fillet
Full,
partial
All
Vapour
pressure
No Medium Laser Butt,
fillet
Full,
partial
All
Weldpool
sag
Voltage Yes Low TIG Butt Full Flat only
Displacement Yes Medium TIG,
MIG
Butt Full Flat only
Laser light
deflection
Yes Low TIG,
Laser
Butt Full Flat only
Weldpool
oscillation
Voltage No Medium TIG Butt Full Flat only
Light Yes Medium TIG Butt Full All
Weldpool
viewing
Photodiode
array
Yes High TIG,
MIG
Butt,
fillet
Full,
partial
All
Coaxial
viewing
Yes Medium TIG Butt,
fillet
Full,
partial
All
CCTV, pool
viewed during
pulse off
Yes Medium TIG,
MIG
Butt,
fillet
Full,
partial
All
Infra-red
camera
Yes Medium TIG,
MIG,
Laser
Butt,
fillet
Full,
partial
All
Thermal
sensing
Thermocouple Yes High TIG,
MIG,
Laser
Butt,
fillet
Full,
partial
All
Pyrometer No Low TIG,
MIG,
Laser
Butt,
fillet
Full,
partial
All
Infra-red
camera
Yes Medium TIG,
MIG,
Laser
Butt,
fillet
Full,
partial
All

Capabilities of the reviewed sensors continued

Generic
technique
Contact/
Non-contact
Cost Additional
capabilities
Comments
Ultrasonic Contact High Seam tracking,
identification of
defects
Component thickness >10mm.
Limitation on minimum component width.
Surface geometry must be simple.
Contact High Seam tracking,
identification of
defects
Component thickness >10mm.
Limitation on minimum component width.
Surface geometry must be simple.
Contact High Seam tracking,
identification of
defects
Limitation on minimum component width.
Surface geometry must be simple.
Contact High Seam tracking,
identification of
defects
Component thickness 0.1 to 2mm.
Limitation on minimum component width.
Surface geometry must be simple.
Acoustic
emission
Contact High Identification of
defects
Limitation on minimum component width.
Accuracy may be limited by variations in
microstructure and thermal or residual
stresses.
Non-
contact
High None  
Weldpool
sag
Non-
contact
Medium None Cannot be used with arc voltage control.
Non-
contact
Medium None Used with arc voltage control.
Non-
contact
High None Use of filler wire may reduce accuracy.
Weldpool
oscillation
Non-
contact
High None Requires accurate power source.
May be inaccurate where filler wire is used,
oxide is present on back of
weld or slag islands are present.
Non-
contact
High None May be inaccurate where filler wire
is used, oxide is present on back
of weld or slag islands are present.
Weldpool
viewing
Non-
contact
Low None Accuracy is reduced where
combinations of variations in the
component parameters occur.
Non-
contact
High Seam tracking Accuracy is reduced where combinations
of variations in the component parameters
occur.
Non-
contact
High Seam tracking Accuracy is reduced where combinations
of variations in the component parameters
occur.
Non-
contact
High Seam tracking,
identification of
defects
Accuracy is reduced where combinations
of variations in the component parameters
occur.
Thermal
sensing
Contact Low None Limit on minimum component size.
Surface geometry must be simple.
Non-
contact
High Seam tracking,
identification of
defects
Accuracy may be reduced by variations
in the emissivity of the surface.
Non-
contact
High Seam tracking,
identification of
defects
Accuracy may be reduced by variations
in the emissivity of the surface.

References

Author Title  
1 Harvey M D F and Lucas W: 'Just deep enough - automatic penetration control for TIG welding'. Bulletin 3, 1990, 50-52. Return to text
2 Siores E: 'Development of a real-time ultrasonic sensing system for automated and robotic welding'. PhD Thesis, Brunel University, December 1988. Return to text
3 Groenwald R A et al: 'Acoustic emission weld monitor system - data acquisition and investigation'. US Army Tank-Automotive Research and Development Command Report AD-AO85-518, October 1979. Return to text
4 Jon M C: 'Non-contact acoustic emission monitoring of laser beam welding'. Welding Journal 1985 64 (9) 43-48. Return to text
5 Lucas W and Maller R S: 'Automatic control of penetration in pulsed TIG welding'. TWI Members Report P/72/1975. Return to text
6 Madigan R B et al: 'Computer-based control of full penetration GTA welds using weld pool oscillation sensing'. lst Intl conf Computer Technology in Welding, Cambridge, 1987, 165-174. Return to text
7 Sugitani Y et al: 'Simultaneous control of penetration depth and bead height by controlling multiple weld parameters'. Welding International, 1990 4 (3) 194-199. Return to text
8 Chin B A et al: 'Infrared thermography for sensing the arc welding process'. Welding Journal, 1983 62 (9), 227s-234s. Return to text