Narrow-Gap TIG Swing Welding: Simulation, Parameter Optimisation, and the Undercut Problem
Literature Overview and Core Findings
This paper by Gao Hui, Qu Jiajun and Li Xianhui from Beijing Institute of Petrochemical Technology addresses a problem that has cost me more sleepless nights than I care to remember: undercut in narrow-gap TIG swing welding. The authors have taken a systematic approach, combining Abaqus thermal-mechanical sequential coupling simulation with response surface methodology (RSM) to develop a multi-parameter collaborative control strategy for the swing welding process. The work is grounded in a double-ellipsoid heat source model that captures the dynamic oscillating behaviour of the torch, which is essential because the thermal input in narrow-gap welding is anything but static.
The key technical insight is the establishment of a dynamic swing model that accounts for the torch's lateral movement, the dwell time at the side walls, and the interaction between the arc and the groove geometry. The authors analysed how swing width, welding speed and side-wall dwell time influence the temperature field, residual stress distribution and molten pool behaviour. They then employed a Box-Behnken experimental design to screen the optimal parameter combination, followed by actual single-pass multi-layer welding trials to validate the simulation results. The findings indicate that a swing width of 16 mm, a welding speed of 0.75 mm/s and a side-wall dwell time of 2 s produce welds with minimal undercut tendency, good fusion between layers and parent metal, and improved hardness uniformity across the weld.
Technical Interpretation and Engineering Practice
Let me tell you what this means for those of us who actually run the plant. Narrow-gap TIG welding is the workhorse for thick-wall components in pressure vessels, pipe fittings and critical structural joints where access is restricted and quality demands are high. The gap is typically 10 to 20 mm, and the torch must oscillate to ensure the side walls are adequately fused. The undercut defect is a classic failure mode in this process, and it is not merely a cosmetic issue. Undercut acts as a stress raiser, reduces the effective load-bearing cross-section, and can initiate fatigue cracks in service. In pressure-retaining applications, an undercut exceeding the allowable limits per ASME Section VIII or API 1104 is grounds for rejection, and rework on a thick-wall component is expensive and time-consuming.
The simulation approach taken by the authors is particularly valuable because it allows the welding engineer to visualise the thermal history and residual stress state before committing to a physical trial. In my experience, the double-ellipsoid heat source model, when properly calibrated with the actual arc efficiency and torch geometry, provides a remarkably accurate prediction of the fusion zone profile and the HAZ extent. The dynamic swing model adds another layer of fidelity by accounting for the fact that the heat input is not uniformly distributed along the oscillation path. The dwell time at the side walls is critical: too short and you get lack of fusion at the side wall; too long and you risk excessive heat accumulation, grain growth and distortion.
The RSM optimisation is a sound statistical approach. The Box-Behnken design is efficient for three-factor experiments, and the response surface allows the engineer to identify the optimal operating window rather than just a single point. This is crucial in production, because welding parameters drift, and you need a robust window, not a knife-edge setting. The optimal parameters identified in this study — swing width 16 mm, welding speed 0.75 mm/s, dwell time 2 s — are consistent with what I have seen on the shop floor for similar groove geometries and material thicknesses. The hardness uniformity improvement is a good indicator of consistent thermal input and proper interpass temperature control.
Key Lessons and Practical Implications
Let me share some hard-won lessons from the field that complement this study. First, the swing width must be matched to the groove geometry. If the swing width is too narrow, the arc does not reach the side walls, and you get lack of fusion. If it is too wide, the arc extends beyond the groove and you get excessive spatter, poor shielding gas coverage and undercut at the toe. The 16 mm swing width in this study is for a specific groove configuration; for a wider gap, you would need to increase it proportionally.
Second, the welding speed of 0.75 mm/s is on the slow side, which is typical for narrow-gap TIG. This means the heat input per unit length is relatively high, and interpass temperature control becomes critical. In multi-layer welding, if the interpass temperature is not maintained within the specified range, you risk excessive grain growth in the HAZ and reduced toughness in the weld metal. I have seen cases where the interpass temperature was not monitored, and the resulting weld failed impact testing at -20°C.
Third, the dwell time at the side walls is the parameter that most directly influences side-wall fusion. In practice, I have found that a dwell time of 1.5 to 2.5 s is the sweet spot for most narrow-gap applications. Too short and the side wall is not adequately melted; too long and you get a "washing" effect that can lead to undercut. The authors' finding that 2 s is optimal aligns well with my experience.
Let me also emphasise the importance of the actual welding trial. Simulation is a powerful tool, but it is only as good as the input data and the assumptions. The authors validated their simulation with actual welding trials and metallographic examination, which is exactly what I would insist on before implementing any parameter change in production. The hardness testing across the weld cross-section is a good way to verify that the thermal input is uniform and that the microstructure is consistent.
Study Insights and Implications
This study provides a solid theoretical basis for the optimisation of narrow-gap swing TIG welding. The combination of thermal-mechanical simulation and RSM optimisation is a methodology that I would recommend for any welding process where multiple parameters interact and where trial-and-error is too costly. The key takeaway for the practising engineer is that the swing parameters — width, speed and dwell time — are not independent variables. They must be optimised together to achieve the desired weld quality.
In terms of practical application, I would add a few caveats. The study was conducted on a specific material and thickness, and the optimal parameters may not transfer directly to other alloys or thicknesses. The weld joint design, the shielding gas composition and flow rate, and the torch geometry all influence the process. I would recommend that any plant adopting these parameters conduct a qualification test per the applicable code, such as ASME Section IX or ISO 15614, before using them in production.
The issue of residual stress is also worth noting. The simulation showed the residual stress distribution, but the actual residual stress in a thick-wall weld is influenced by the restraint of the surrounding structure, the welding sequence, and the post-weld heat treatment (PWHT). In pressure vessel fabrication, PWHT is often required to relieve residual stresses, and the simulation results can help in determining the appropriate PWHT schedule.
Conclusion
This study is a valuable contribution to the field of narrow-gap TIG welding. It demonstrates a rigorous methodology for process optimisation that combines simulation with statistical design and experimental validation. The identification of the optimal swing parameters — 16 mm width, 0.75 mm/s speed and 2 s dwell time — provides a practical reference for welding engineers working on thick-wall components. The key insight is that the swing parameters must be optimised as a set, not individually, to achieve defect-free welds with consistent mechanical properties. For those of us in the industry, this is exactly the kind of work that moves the needle on weld quality and productivity.
Zhuojin Pipe Fitting Co., Ltd