ZHUOJIN-LOGOZhuojin Pipe Fitting Co., Ltd
Zhuojin Pipe Fitting Co., Ltd
STEEL PIPE · FITTING · WELDING  |  NOTES FROM A 30-YEAR INDUSTRY VETERAN

Vision-Based Molten Pool Detection for Copper TIG Hardfacing: Imaging Challenges and Solutions

The Need for Process Monitoring

In TIG welding of copper, the control of the molten pool is critical because copper has a high thermal conductivity and a low melting point, and the pool can easily become too large or too small, leading to lack of fusion, burn-through, or excessive dilution. In a production environment, the welding is typically performed by a skilled operator who visually monitors the pool through a welding helmet. However, for automated welding or for quality control purposes, there is a need for a vision-based system that can monitor the pool in real time and provide feedback to the welding controller. The paper addresses this need by investigating the imaging of the copper molten pool in the near-infrared band, using a charge-coupled device (CCD) camera with various narrow-band filters.

Optical Challenges of Copper Welding

The imaging of a copper molten pool is fundamentally different from the imaging of a steel molten pool. Copper has a high reflectivity in the visible and near-infrared bands, which means that the pool surface reflects the surrounding environment rather than emitting radiation from its own surface. The emissivity of molten copper is low, typically in the range of 0.1 to 0.3, compared to 0.4 to 0.6 for molten steel. This means that the radiation emitted by the copper pool is weak, and it is easily overwhelmed by the radiation from the welding arc and by the reflections from the surrounding surfaces.

The paper identifies the near-infrared band as the most suitable for imaging the copper pool, because in this band the ratio of the pool radiation to the arc radiation is more favorable than in the visible band. The welding arc emits strongly in the ultraviolet and visible bands, but the emission in the near-infrared band is relatively lower, so by filtering out the visible and ultraviolet radiation, the camera can capture the pool radiation with less interference from the arc. The paper also notes that the copper pool is more difficult to image than the steel pool in the near-infrared band, which is consistent with the lower emissivity of copper.

Imaging System Design and Filter Selection

The paper describes the design of multiple composite narrow-band filter systems, based on different imaging mechanisms. The key parameters are the central wavelength and the bandwidth of the filters. The filters are selected to pass only the near-infrared radiation from the pool while blocking the visible and ultraviolet radiation from the arc. The paper compares the imaging results at different observation windows and identifies the optimal window that provides the clearest image of the copper pool.

Filter Type Central Wavelength (nm) Bandwidth (nm) Image Quality
Narrow-band 800 - 900 10 - 20 Good contrast, some arc interference
Composite narrow-band 900 - 1000 20 - 30 Best contrast, minimal arc interference
Wide-band 700 - 1100 400 Poor contrast, strong arc interference

The optimal filter is a composite narrow-band system that passes radiation in the 900 to 1000 nm range, which is in the near-infrared band. This filter provides the best contrast between the pool and the surrounding base metal, and it minimizes the interference from the arc. The paper also notes that the imaging is affected by the surface condition of the copper; a clean, polished surface will reflect more radiation and produce a brighter image, while an oxidized surface will absorb more radiation and produce a darker image.

Comparison with Steel Molten Pool Imaging

The paper compares the imaging of the copper pool with the imaging of the steel pool and finds that the copper pool is more difficult to image in the near-infrared band. This is due to the lower emissivity of copper, which results in a weaker radiation signal from the pool. The steel pool, with its higher emissivity, produces a stronger signal that is easier to distinguish from the arc radiation. The paper also notes that the copper pool has a different geometry than the steel pool; the copper pool is more circular and has a more defined boundary, while the steel pool is more elongated and has a less defined boundary.

The practical implication of this finding is that the vision system for copper welding must be designed with a more sensitive camera and a more selective filter than the system for steel welding. The camera must have a high quantum efficiency in the near-infrared band, and the filter must be narrow enough to block the arc radiation while passing the pool radiation. The system must also be calibrated for the specific copper alloy being welded, because the emissivity varies with the alloy composition and the surface condition.

Practical Implications for Automated Welding

From a production standpoint, the ability to image the copper molten pool is a prerequisite for automated TIG welding of copper components. Without a reliable vision system, the welding must be performed by a skilled operator, which limits the production rate and introduces variability in the weld quality. The paper's finding that the near-infrared band provides the best imaging conditions is a useful guide for the design of automated welding systems, but it is not the whole story.

In my experience, the vision system must be integrated with a control system that can adjust the welding parameters in real time based on the pool image. The control system must be able to detect the pool size, the pool shape, and the pool position, and then adjust the welding current, the travel speed, and the wire feed rate to maintain the desired pool characteristics. This is a challenging control problem because the pool dynamics are fast and the relationship between the welding parameters and the pool characteristics is nonlinear. The paper does not address the control aspect, but it provides the foundation for the vision system that the control system would use.

The paper is a solid contribution to the field of welding process monitoring, and it addresses a real problem in the welding of copper. The finding that the copper pool is more difficult to image than the steel pool is a useful reminder that the imaging system must be tailored to the specific material being welded.