TIG Welding Molten Pool External Excitation Oscillation and Its Relationship with Penetration
Literature Overview and Historical Context
This paper from Harbin Institute of Technology, published in 1989, is a classic piece of work that laid the foundation for what later became known as weld pool oscillation sensing or pool resonance control. At the time, automated TIG welding was still in its infancy, and one of the biggest obstacles to full automation was the lack of a reliable, real-time method for detecting weld penetration. A welder can judge penetration by looking at the pool, but a machine cannot. The authors, Zhang Jiuhai, Yang Chunli and Wang Qilong, set out to find a way to make the pool itself tell us whether it has penetrated.
The idea is elegant in its simplicity. If you give the molten pool a mechanical impulse, it will oscillate at a natural frequency that depends on its size and geometry. As the pool grows, the natural frequency drops. If you can measure the oscillation frequency, you can infer the pool dimensions, and from that, the penetration state. The challenge is how to excite the pool and how to detect the oscillation without interfering with the welding process.
Technical Points and Experimental Findings
The authors built a transistor power supply that could superimpose a controlled current pulse on the welding current. This pulse creates a sudden change in arc pressure, which pushes down on the pool surface. When the pulse ends, the pool rebounds and oscillates. The oscillation causes the arc length to change, which in turn causes the arc voltage to fluctuate. By monitoring the arc voltage signal, the authors were able to detect the pool oscillation frequency.
The key relationships they established are summarised below.
| Parameter | Relationship Observed |
|---|---|
| Pool oscillation frequency vs. pool top width | Frequency decreases as top width increases |
| Pool oscillation frequency vs. pool back width | Frequency decreases as back width increases |
| Pool oscillation frequency vs. plate thickness | Different frequency ranges for different thicknesses |
| Arc length effect | Longer arc length reduces the detectability of oscillation |
| Pulse current amplitude | Higher amplitude gives stronger oscillation signal |
| Pulse current duration | Optimum duration exists for maximum oscillation amplitude |
The most significant finding is that the oscillation frequency correlates with both the top and back width of the pool. This means that by measuring the frequency, you can infer the back width, which is the parameter that indicates penetration. If the back width is zero, the weld has not penetrated. If the back width exceeds a certain value, the weld is fully penetrated. The frequency gives you a continuous, real-time signal that can be used as a feedback variable for controlling heat input.
The authors also studied the effect of arc parameters on the oscillation itself. They found that arc length is critical: a longer arc damps the signal because the arc becomes less stiff and the voltage change per unit of pool displacement is smaller. This has practical implications for sensor placement and arc length control.
Practical Implications for the Shop Floor
This work is directly relevant to anyone involved in automated or robotic TIG welding, particularly for pipe welding where root pass penetration is critical. In my experience, the root pass is where most penetration-related defects occur. If you have a real-time signal that tells you whether the root has penetrated, you can adjust the current or travel speed before the defect forms, rather than discovering it after the fact with radiography.
The practical implementation requires three elements:
- A power source capable of superimposing a controlled current pulse on the welding current. This is not available on standard TIG machines, so it requires either a custom-built power supply or a modification to an existing one.
- A signal detection system that can extract the arc voltage fluctuation and process it to obtain the oscillation frequency. This involves filtering out the welding current noise and isolating the oscillation component.
- A control algorithm that relates the frequency to penetration and adjusts the welding parameters accordingly.
The authors demonstrated the feasibility of this approach in the laboratory, but they did not close the loop into a full automatic control system. That was left for subsequent researchers. From a production standpoint, the key question is whether the oscillation signal is robust enough to be used in a real welding environment, where there are electromagnetic interference, arc instability and variations in fit-up.
Key Questions and Reflections
One issue that the paper does not address is the effect of weld pool geometry on the oscillation mode. A pool is not a simple mass-spring system; it has multiple modes of oscillation, and the frequency that is detected may depend on the excitation method and the sensor position. In practice, the detected frequency may be a combination of modes, which complicates the interpretation.
Another issue is the effect of surface tension and viscosity on the oscillation. These properties vary with temperature and alloy composition, so the frequency-pool size relationship is not universal. A calibration curve developed for one material may not be valid for another. This is a significant limitation for a production system that must handle multiple alloys.
The paper also does not address the effect of weld pool contamination or oxidation on the oscillation behaviour. A dirty pool will have different surface tension and viscosity, which will shift the frequency. In a real production environment, cleanliness is variable, so the sensor signal will drift.
Study Insights and Practical Recommendations
This paper is a reminder that the weld pool is a dynamic system with measurable physical properties. Too often, we treat welding as a black box and rely on post-weld inspection to catch defects. The authors showed that the pool itself can provide a rich source of information for real-time control.
For a production engineer considering this approach, I would offer the following advice:
- The technique is most useful for root pass control in pipe welding, where penetration is critical and difficult to observe visually.
- Calibration is essential. You must establish the frequency-pool size relationship for each material, thickness and joint configuration.
- The arc length must be held constant for the signal to be reliable. This requires a stable torch position and a consistent work distance.
- The pulse current and duration must be optimised for each application. Too small a pulse will not excite the pool; too large a pulse will disturb the weld.
- Expect to spend significant time on signal processing. The arc voltage signal is noisy, and extracting the oscillation frequency requires careful filtering.
Concluding Summary
This 1989 paper from Harbin Institute of Technology established the fundamental relationship between weld pool oscillation frequency and pool dimensions, demonstrating that the natural frequency of the pool decreases as the pool grows. By using a pulsed current to excite the pool and monitoring the resulting arc voltage fluctuation, the authors showed that it is possible to detect penetration in real time. The technique is elegant but requires careful control of arc length and pulse parameters, and the frequency-pool size relationship must be calibrated for each material and joint configuration. While the authors did not close the loop into a full automatic control system, their work provided the foundation for subsequent research on weld pool oscillation sensing. For anyone involved in automated TIG welding of pipe or critical structures, this paper is essential reading, as it demonstrates that the weld pool can be used as a sensor for penetration control.
Zhuojin Pipe Fitting Co., Ltd