Zhuojin Pipe Fitting Co., LtdAdaptive Torch Height Tracking for Vessel Inner Wall Overlay Welding
Literature Overview
This paper, published in Manufacturing Technology & Machine Tools (2025, Vol. 1, pp. 141-147) by Zhang Ning and colleagues from the School of Mechanical Engineering at Xinjiang University, addresses a critical challenge in the automated repair welding of vertical pressure vessel inner walls: real-time torch height tracking during overlay welding operations. The research is supported by the Natural Science Foundation of Xinjiang Uygur Autonomous Region (2022D01C391) and the Xinjiang University Excellent Doctoral Innovation Project (XJU2022BS092). The authors propose a Model Reference Adaptive Control (MRAC) algorithm enhanced with an integral term, combined with an Infinite Impulse Response (IIR) filter for welding current signal processing, to achieve precise torch height regulation during horizontal-position overlay welding on curved vessel surfaces.
Core Technical Approach
The fundamental problem addressed here is that during overlay welding on the inner wall of a vertical vessel, the torch must follow a curved surface at varying distances, and the welding current serves as the primary feedback signal for torch height deviation. Traditional PID controllers struggle with the nonlinear and time-varying characteristics of this process. The proposed solution introduces three key innovations:
- Model Reference Adaptive Control (MRAC) with Integral Term: The integral component is embedded within the adaptive control law to improve the controller's disturbance rejection capability, specifically against fluctuations in welding current caused by arc instability, surface roughness, and material inconsistencies.
- IIR Filter for Current Signal Conditioning: The raw welding current signal, which contains high-frequency noise from arc phenomena, is processed through an IIR filter to extract meaningful trend information without introducing excessive phase lag.
- Horizontal Welding Torch Height-Current Mathematical Model: A specific transfer function relating torch height deviation to welding current change is established for horizontal-position welding, accounting for the gravity-driven arc deflection and molten pool behavior unique to this orientation.
Process Parameters and Control Performance
| Parameter | Specification |
|---|---|
| Control Method | Integral-term MRAC |
| Signal Processing | IIR filter on welding current |
| Welding Position | Horizontal (F-position equivalent on curved surface) |
| Maximum Tracking Error | ±0.34 mm |
| Feedback Variable | Welding current |
| Control Variable | Torch height (via servo motor) |
The maximum tracking error of ±0.34 mm is a significant achievement for overlay welding applications. In engineering practice, overlay weld quality is highly sensitive to torch height variations: a deviation of even 0.5 mm can lead to significant changes in dilution ratio, weld bead geometry, and microstructure homogeneity. For vessel inner wall repair, where the overlay layer must provide corrosion resistance, wear resistance, or stress relief, maintaining consistent bead width and penetration is essential.
Engineering Practice Insights
From a practical standpoint, this research addresses a pain point that many maintenance welding operations encounter. Vessel inner wall repair often involves large-diameter vertical vessels (such as reactors, distillation columns, and storage tanks) where manual torch height control is impractical and existing automated systems frequently suffer from poor tracking accuracy on curved surfaces.
Key Considerations for Implementation
- Surface Preparation: The effectiveness of current-based height tracking depends on a consistent baseline surface. Rust, scale, and uneven machining marks can introduce current fluctuations unrelated to torch height.
- Welding Parameter Stability: The mathematical model relating height to current assumes relatively stable welding parameters (current, voltage, speed). Significant parameter drift during long welds must be compensated.
- Curvature Effects: On highly curved surfaces, the torch standoff distance varies not only with height but also with the angle of incidence. The model must account for the geometric relationship between the torch axis and the local surface normal.
- Multi-Pass Overlay: For thick overlay layers, the torch height reference must be updated between passes to account for the previous pass's buildup height.
FMEA Analysis of Potential Failure Modes
| Failure Mode | Cause | Effect | Detection Method | Countermeasure |
|---|---|---|---|---|
| Loss of tracking | Severe surface irregularity | Porosity, lack of fusion | Visual inspection, UT | Surface pre-machining, adaptive gain scheduling |
| Current sensor noise | Arc instability | Erroneous height correction | Signal monitoring | IIR filter tuning, signal averaging |
| Servo motor lag | Mechanical friction, backlash | Overshoot, oscillation | Response time measurement | Backlash compensation, motor maintenance |
| Model mismatch | Welding parameter change | Systematic bias | Error trend analysis | Online model identification |
Study Insights and Implications
This work represents a meaningful advancement in adaptive control for welding automation. The integration of the integral term within the MRAC framework is particularly noteworthy because it addresses a known limitation of pure MRAC: the inability to eliminate steady-state error caused by unmodeled disturbances. In overlay welding, where the baseline current is affected by material composition variations, surface condition, and ambient conditions, this integral action provides essential robustness.
The choice of welding current as the feedback signal is pragmatic—current sensors are robust, inexpensive, and inherently coupled to the arc process. However, engineers should be aware that current-based tracking has inherent limitations: it cannot distinguish between height changes and other factors that affect current (such as wire feed rate variations or consumable condition changes). In high-criticality applications, supplementary sensors (such as arc voltage or capacitive proximity sensors) may be warranted.
The ±0.34 mm tracking accuracy reported here meets the requirements for most overlay welding quality standards, including those specified in ASME B31.3 and API 579 for overlay repair welds. Future work should explore extending this approach to multi-axis torch tracking on complex geometries and integrating real-time microstructure prediction models to further enhance quality assurance.