Zhuojin Pipe Fitting Co., LtdK-TIG Horizontal Welding Penetration State Identification Under Variable Parameters
Literature Overview
The research by Zhao Shizhan, Shi Yonghua, Li Bohan and Xu Weidong, published in The Welding Journal (2026, Vol. 47, No. 1, pp. 1-10), addresses a significant practical challenge in automated welding of large stainless steel vertical cylindrical tanks: maintaining keyhole TIG (K-TIG) penetration during horizontal circumferential welding. Funded by the National Key R&D Program (2023YFC2809800) and Guangxi provincial research programs, this work develops an OCR-SVM-based penetration identification model and investigates how variable welding current and travel speed affect penetration state and identification accuracy.
Core Technical Content
K-TIG welding offers the remarkable advantage of single-sided welding with double-sided penetration, eliminating the need for groove preparation. This is particularly attractive for large stainless steel storage tanks where groove machining is impractical or uneconomical. However, during horizontal circumferential welding of large vertical tanks, gravity disrupts the dynamic force balance maintaining the keyhole, leading to penetration instability and potential quality degradation.
Influence of Parameter Variation on Penetration State
The study systematically varied welding current and travel speed to assess their individual and combined effects on penetration quality.
| Variable Parameter | Effect on Penetration | Relative Influence |
|---|---|---|
| Welding current variation | Direct impact on keyhole stability and penetration depth | More significant |
| Travel speed variation | Affects heat input distribution and keyhole maintenance | Less significant |
The finding that welding current variation exerts a more pronounced influence on penetration state than travel speed variation is consistent with fundamental arc physics: current directly determines arc power density, which governs keyhole formation and maintenance, whereas travel speed primarily affects the duration of heat input per unit length.
OCR-SVM Penetration Identification Model Performance
The optical character recognition-support vector machine (OCR-SVM) model was evaluated under variable parameter conditions.
| Operating Condition | Identification Accuracy |
|---|---|
| Variable current | 92.29% |
| Variable speed | 91.50% |
The identification accuracy under both variable conditions exceeds 91%, demonstrating robust performance. The key geometric features identified as critical for penetration state recognition are the width and area of both the molten pool and the keyhole.
Keyhole Force Balance in Horizontal Welding
The fundamental challenge in K-TIG horizontal welding lies in the disruption of keyhole force equilibrium by gravitational effects. During vertical welding, the keyhole is maintained by a balance of arc pressure, surface tension, and plasma shear forces acting in the axial direction. When the weld position rotates to horizontal, gravity introduces a lateral force component that must be compensated by the arc dynamics to maintain stable keyhole geometry. The inability to fully compensate for this gravitational disturbance leads to asymmetric keyhole collapse on the lower side of the weld, resulting in incomplete penetration or backside deficiency.
Engineering Practice Implications
For engineers involved in the fabrication of large stainless steel storage tanks, this work provides actionable insights for automated K-TIG welding implementation:
- The OCR-SVM model offers a practical real-time monitoring solution for penetration quality assessment during horizontal welding, enabling immediate corrective action when penetration deviation is detected.
- Welding current should be the primary parameter adjusted for penetration control during horizontal welding, given its more significant influence on keyhole stability.
- The identification of molten pool and keyhole width and area as critical geometric features provides a clear basis for sensor selection and image processing algorithm design.
Process Development Recommendations
Based on the study findings, the following process development strategies are recommended:
- Develop position-dependent current profiles that increase current at horizontal positions to compensate for gravitational keyhole disruption.
- Implement real-time molten pool monitoring with image acquisition at rates sufficient to capture keyhole dynamics during parameter transitions.
- Establish qualification procedures that include horizontal position testing at multiple circumferential locations to verify consistent penetration quality.
- Incorporate the OCR-SVM model into the welding control system for closed-loop penetration feedback, enabling automatic current adjustment.
- Conduct systematic studies on the interaction between travel speed and current at horizontal positions to define the full process window.
Key Questions and Reflections
The identification accuracy of approximately 92% under variable parameter conditions, while impressive, raises the question of what the remaining 8% misclassification represents in practical terms. In a production environment welding large storage tanks, even a small fraction of misidentified penetration states could lead to significant rework costs if undetected. The question of model robustness under edge cases—such as sudden parameter changes, sensor contamination, or ambient light interference—deserves further investigation.
Additionally, the study focuses on identification accuracy but does not extensively address the control response time. In a closed-loop system, the delay between penetration state detection and parameter adjustment could be critical, particularly at higher travel speeds where the weld moves rapidly through the problematic horizontal position. The integration of prediction capability into the control system, rather than purely reactive correction, could improve penetration quality.
Another important consideration is the generalizability of the findings. The study was conducted on stainless steel storage tanks, but the gravitational effects on keyhole stability would apply to any large-diameter cylindrical vessel fabricated by K-TIG. Extending the process window studies to other materials and geometries would broaden the engineering applicability.
Reference Value and Outlook
This work represents a meaningful contribution to the automation of K-TIG welding for large cylindrical structures. The development of a penetration identification model with over 91% accuracy under variable parameter conditions provides a foundation for real-time quality control in automated welding systems. The identification of welding current as the dominant parameter for penetration control during horizontal welding offers a clear and practical process control strategy. For engineers in the storage tank and pressure vessel fabrication industry, the demonstrated feasibility of K-TIG horizontal welding with automated penetration monitoring opens the possibility of significantly reducing manufacturing costs by eliminating groove preparation and enabling single-sided welding. The OCR-SVM approach, while not perfect, provides a practical technology that can be further refined through continued research and field deployment.