Penetration Control in MIG Welding Using CCD Imaging

Literature Overview

This landmark paper, published in Welding Journal (Vol. 12, No. 1, 1991, pp. 53–58) by Pan Jiluan, Chen Qiang, and Wu Zhiqiang from Tsinghua University, describes a penetration control system for MIG/MAG welding that uses a CCD (Charge-Coupled Device) sensor to monitor the back-side temperature field of the weld in real time. The system operates without the need for calibration and employs a relay-type nonlinear control element to achieve stable and responsive penetration control. This work represents one of the earliest Chinese contributions to closed-loop welding process control using non-contact optical sensing.

Core Technical Concept

The fundamental challenge addressed is the control of weld penetration depth in single-sided MIG/MAG welding, where direct visual access to the root side of the joint is not available. Excessive penetration leads to burn-through and reduced fatigue strength, while insufficient penetration results in incomplete root fusion and potential leakage paths in pressure-containing applications. The authors' approach is to monitor the temperature distribution on the back side of the plate through a CCD camera and use this thermal information as the feedback signal for adjusting welding parameters (typically welding current and/or travel speed).

Technical Analysis of the Control Strategy

The use of a relay-type nonlinear control element is a particularly interesting and, in my view, quite elegant choice. In classical control theory, relay control is known for its robustness to model uncertainty and its ability to provide fast response, albeit with some oscillation around the setpoint. For welding penetration control, where the thermal system has significant time delays and the plant model changes as the weld progresses, a relay controller offers several practical advantages:

1. Model independence: The relay controller does not require an accurate mathematical model of the welding thermal system, which is notoriously difficult to establish due to the nonlinear nature of heat transfer, convective losses, and phase change phenomena.
2. Fast response: The binary switching action of the relay provides immediate correction when the penetration deviates from the target, which is essential for maintaining consistent root quality.
3. Simplicity of implementation: In the context of 1991 welding equipment, relay control could be implemented with relatively straightforward analog or digital circuitry, making the system more practical for industrial deployment.

The paper emphasizes that the system achieves both stability and fast response, which is often a trade-off in control system design. The key to achieving both lies in the specific characteristics of the welding thermal system: the thermal inertia of the base metal provides natural filtering that prevents excessive oscillation, while the relatively short thermal diffusion time in thin plates ensures that the back-side temperature responds quickly enough to be useful as a control signal.

Engineering Practice Implications

For pipe fitting manufacturing, penetration control is of paramount importance in several contexts:

• Butt-weld fittings: When manufacturing butt-weld fittings by welding pre-formed blanks, the root penetration must be controlled to ensure full fusion without excessive reinforcement that could create stress concentrations.
• Pipe-to-flange welding: In the fabrication of flanged pipe assemblies, the penetration of the fillet weld at the pipe-to-flange junction must be sufficient to achieve full strength while avoiding excessive heat input that could distort the flange face.
• Thin-wall pipe welding: For small-diameter ERW or HFW pipe fittings, where wall thicknesses may be as low as 1–2 mm, penetration control is critical to avoid burn-through.

The CCD-based approach described in this paper has the advantage of being non-contact and requiring no calibration, which makes it suitable for integration into automated welding cells. However, the practical implementation requires careful consideration of the optical path geometry, the viewing angle relative to the weld pool, and the shielding against arc light interference. The paper does not extensively discuss these practical aspects, but they are essential for any production deployment.

Study Insights and Reflections

Reading this paper in retrospect, I am struck by how far ahead of its time the approach was. The concept of using optical sensing for real-time welding process monitoring has become mainstream in recent decades, but in 1991, this was genuinely pioneering work. The authors' decision to use a CCD camera—which was already a mature technology in scientific imaging—rather than developing a custom sensor demonstrates good engineering pragmatism.

One reflection that occurs to me is the relationship between this work and the subsequent development of visual seam tracking and weld pool monitoring systems. The CCD-based penetration control described here is, in essence, the predecessor of modern visual servoing systems used in robotic welding. The same fundamental principle—extracting process-relevant information from optical images and using it for real-time control—has been extended to include seam tracking, weld pool shape monitoring, and defect detection.

The relay-type control element deserves particular attention. In my experience with welding control systems, the tendency has been to move toward increasingly sophisticated PID or model-predictive controllers. However, the simplicity and robustness of relay control make it particularly attractive for industrial applications where the control system must function reliably despite variations in material properties, joint geometry, and environmental conditions. This paper serves as a reminder that the best control strategy is not necessarily the most mathematically sophisticated one, but the one that best matches the practical requirements of the application.