Zhuojin Pipe Fitting Co., LtdApplicability of Heat Source Models for TIG Welding Numerical Simulation
Literature Overview
The paper by Li Mengsheng and Wang Chuanbiao from Hefei University of Technology examines the applicability of different heat source models for numerical simulation of TIG welding. Published in the journal Re Gongyi Jishu (Hot Working Technology) in 2007, this study compares the Gaussian distribution function and the double-ellipsoid distribution function as heat source models for simulating the three-dimensional dynamic temperature field during TIG welding of plate butt joints. The study employs ANSYS finite element software and validates the simulation results against experimental measurements of thermal cycles and weld cross-sectional geometry.
Heat Source Models in Welding Simulation
Accurate heat source modeling is fundamental to the reliability of welding numerical simulations. The choice of heat source model directly influences the predicted temperature distribution, weld pool geometry, residual stress patterns, and microstructural evolution. Two primary heat source models are discussed in this paper:
| Heat Source Model | Mathematical Form | Spatial Distribution | Typical Application |
|---|---|---|---|
| Gaussian | f(r) = Q·exp(-r²/r₀²) | Circular symmetry around arc axis | Simple, isotropic heat input |
| Double-Ellipsoid | f(x,y,z) = different ellipsoidal shapes for front and rear | Asymmetric with front/rear distinction | Realistic representation of weld pool |
The Gaussian heat source model assumes a circular cross-section with heat intensity decreasing exponentially with radial distance from the arc center. While mathematically simple and computationally efficient, this model does not account for the asymmetric nature of the actual weld pool, where the front portion (ahead of the arc) is typically deeper and narrower while the rear portion (behind the arc) is shallower and wider due to the travel direction.
The double-ellipsoid model, proposed by Goldak, Akerman, and Bently in 1984, addresses this limitation by using two semi-ellipsoidal volumes—one for the front half of the weld pool and one for the rear half—with independently adjustable parameters. This model provides a more realistic representation of the weld pool geometry and heat flux distribution.
Simulation Methodology
The authors employed ANSYS finite element software to simulate the three-dimensional dynamic temperature field during TIG welding of plate butt joints. The simulation incorporated moving heat sources with the welding parameters matching the experimental conditions. The key simulation parameters included:
- Welding current: as specified in the experimental setup
- Travel speed: as specified in the experimental setup
- Heat input efficiency: calibrated to match experimental results
- Mesh size: sufficiently refined to capture thermal gradients near the weld zone
Two separate simulations were conducted—one using the Gaussian heat source model and one using the double-ellipsoid model—with all other simulation parameters kept identical. This controlled comparison isolates the effect of the heat source model on the predicted temperature distribution.
Validation Against Experimental Data
The simulation results were validated against two types of experimental measurements: thermal cycle curves and weld cross-sectional dimensions. Thermal cycle curves were obtained by embedding thermocouples at specific locations in the test plate and recording temperature versus time during the welding process. Weld cross-sectional dimensions were measured by sectioning the welded specimens and performing macrographic analysis.
The comparison revealed that the double-ellipsoid heat source model produced results that agreed well with the experimental measurements, while the Gaussian model showed notable deviations. Specifically, the Gaussian model tended to overpredict the width of the weld pool and underpredict the penetration depth, leading to inaccurate predictions of the thermal cycle curves at locations near the weld zone.
| Validation Parameter | Gaussian Model | Double-Ellipsoid Model | Experimental Measurement |
|---|---|---|---|
| Peak temperature accuracy | Moderate | Good | Reference |
| Weld width prediction | Overprediction | Good agreement | Measured |
| Penetration depth prediction | Underprediction | Good agreement | Measured |
| Thermal cycle curve match | Poor | Good | Measured |
Technical Interpretation
The superior performance of the double-ellipsoid model can be attributed to its ability to capture the asymmetric heat flux distribution inherent in TIG welding. As the arc travels along the joint, the molten metal ahead of the arc is continuously being added to the pool, creating a deeper, more concentrated heat zone in the front half. Behind the arc, the metal is solidifying and the heat flux spreads laterally, creating a shallower, wider zone. The double-ellipsoid model explicitly accounts for this front-rear asymmetry through its independent parameter sets.
The Gaussian model, by contrast, assumes symmetric heat distribution, which is physically unrealistic for any welding process with travel. This symmetry assumption leads to systematic errors in the predicted temperature field, particularly in the regions immediately surrounding the weld path where the thermal gradients are most critical for predicting microstructural evolution and residual stresses.
Engineering Practice Implications
For engineers using numerical simulation to optimize welding procedures, this study reinforces the importance of selecting an appropriate heat source model. The use of a Gaussian model may be acceptable for preliminary screening or conceptual studies, but for detailed predictions of weld geometry, thermal cycles, and residual stresses, the double-ellipsoid model or more sophisticated models should be employed.
The validation approach used in this paper—comparing simulation results against both thermal cycle data and weld geometry measurements—is a best practice that should be followed in all welding simulation work. Relying solely on one type of validation data can lead to overfitting or underfitting of the model parameters.
Study Insights and Reflections
This paper makes a clear and well-documented case for the use of the double-ellipsoid heat source model in TIG welding simulations. The methodology is sound, the validation approach is rigorous, and the conclusions are well-supported by the data. The study serves as an important reference for engineers who are beginning to incorporate numerical simulation into their welding procedure development workflows.
One limitation of the study is that it does not address more advanced heat source models that have been proposed in subsequent literature, such as the conical heat source, the hybrid Gaussian-ellipsoid model, or the multi-phase-field approach. Additionally, the study does not investigate the effect of mesh density on the accuracy of the simulation results, which is a critical factor in finite element analysis. Nevertheless, for the scope and purpose of the study, the conclusions are robust and practically relevant.
The key takeaway for practicing engineers is that heat source model selection is not a trivial decision—it directly affects the accuracy and reliability of simulation results. Investing time in proper model calibration and validation pays dividends in the form of more accurate predictions and more reliable welding procedure optimization.