ZHUOJIN-LOGOZhuojin Pipe Fitting Co., Ltd
Zhuojin Pipe Fitting Co., Ltd
STEEL PIPE · FITTING · WELDING  |  NOTES FROM A 30-YEAR INDUSTRY VETERAN

Literature Study Note: Three-Dimensional Dynamic Simulation of TIG Welding Temperature Field and Infrared Temperature Measurement

1. Literature Overview and Core Contribution

This paper by Lei Yucheng et al. from Jiangsu University addresses a fundamental problem that every welding engineer eventually confronts: how do we accurately describe the temperature field during rapid heating and cooling cycles in tungsten inert gas (TIG) welding? The authors establish a three-dimensional dynamic finite element model for the moving arc, implement it using ANSYS with APDL programming, and validate the simulation against infrared thermography measurements taken during actual welding of aluminum alloy. The key technical innovation lies in the careful treatment of temperature-dependent thermophysical properties, radiation and convection boundary conditions, and phase transformation effects, combined with a double-ellipsoid heat source model that captures the asymmetric distribution of heat input from the moving arc.

The significance of this work extends far beyond the laboratory. In my thirty years in the steel pipe and pipe fitting industry, I have seen countless cases where welding procedures were qualified on the basis of trial-and-error alone, without any fundamental understanding of the thermal history that the weld metal and heat-affected zone (HAZ) experience. The temperature field is the master variable that controls everything downstream: phase transformation kinetics, grain growth, residual stress development, distortion, and ultimately the mechanical properties and corrosion resistance of the joint. If you cannot describe the temperature field with reasonable accuracy, you cannot predict the microstructure, and if you cannot predict the microstructure, you cannot predict the performance of the joint in service.

2. Technical Interpretation and Key Parameters

2.1 Heat Source Model Selection

The paper employs a double-ellipsoid heat source model, which is a significant improvement over simpler Gaussian surface heat source models. The double-ellipsoid model, originally proposed by Goldak, divides the heat input into front and rear ellipsoidal regions to account for the fact that the heat distribution ahead of the arc is steeper than behind it. This asymmetry is critical for moving arc welding because the trailing portion of the weld pool experiences a different thermal gradient than the leading edge. The parameters of the double-ellipsoid model are summarized below.

Parameter Symbol Typical Value for TIG Welding Physical Meaning
Front ellipsoid length a_f 1.5-3.0 mm Depth of heat concentration ahead of arc
Rear ellipsoid length a_r 4.0-8.0 mm Depth of heat concentration behind arc
Width parameter b 2.0-4.0 mm Lateral spread of heat source
Depth parameter c 1.0-3.0 mm Penetration depth of heat source
Arc efficiency η 0.60-0.75 for TIG Fraction of electrical energy transferred to workpiece

The arc efficiency value is particularly important in practice. For TIG welding of aluminum alloys, the efficiency typically falls in the range of 0.60-0.75, but this can vary significantly with shielding gas composition, arc length, and electrode geometry. In my experience with stainless steel pipe welding, the efficiency tends to be slightly lower, around 0.50-0.65, because the higher thermal conductivity of aluminum means more heat is conducted away from the weld pool. Getting this value wrong by 10-15% can shift the predicted weld pool dimensions by a similar percentage, which directly affects the accuracy of the simulation.

2.2 Temperature-Dependent Material Properties

The paper correctly emphasizes the importance of considering temperature-dependent thermophysical properties. This is a point that many engineers overlook when they first attempt thermal simulation. The table below illustrates the magnitude of variation for a typical aluminum alloy.

Temperature (°C) Thermal Conductivity (W/m·K) Specific Heat (J/kg·K) Density (kg/m³)
20 167 896 2700
200 180 940 2680
400 190 1010 2650
600 195 1100 2600
660 (melting) 200 1180 2550

The density variation from 2700 kg/m³ at room temperature to 2550 kg/m³ at melting represents a 5.5% change, which may seem small but has a measurable effect on the thermal field because the volumetric heat capacity (density × specific heat) changes by more than 20% across this range. If you assume constant properties, you will overestimate the heat required to raise the temperature of the material, and your predicted weld pool dimensions will be smaller than reality.

2.3 Boundary Conditions and Phase Transformation Effects

The paper accounts for radiation and convection heat losses from the workpiece surface, as well as the latent heat of fusion during phase transformation. These are essential boundary conditions that many simplified models neglect. The radiation heat loss follows the Stefan-Boltzmann law and becomes significant at high temperatures. For a weld pool at 1500°C, the radiation heat flux can be on the order of 10^5 W/m², which is not negligible compared to the heat input from the arc. Convection losses are typically modeled using a combined heat transfer coefficient that accounts for both natural convection and the forced convection from the shielding gas flow. The latent heat of fusion for aluminum is approximately 397 kJ/kg, which acts as a thermal buffer during melting and solidification. Ignoring this effect will cause the predicted cooling rate to be too fast, which in turn leads to overestimation of the HAZ hardness and underestimation of the weld pool size.

3. Infrared Temperature Measurement: Practical Considerations

The paper uses a SAT-HY6800 infrared pyrometer with a filter to exclude arc radiation. This is a critical practical detail that I have seen misapplied in many shop floor situations. The arc emits intense radiation across a broad spectrum, including the infrared range that the pyrometer is designed to detect. If you do not filter out the arc radiation, your temperature readings will be dominated by the arc itself rather than the workpiece surface. The filter must be selected based on the specific emission characteristics of the arc and the emissivity of the workpiece material.

For aluminum alloys, the emissivity varies significantly with surface condition. A polished aluminum surface has an emissivity of only 0.04-0.06, while an oxidized or anodized surface can have an emissivity of 0.3-0.5. This means that the same physical temperature can produce very different infrared readings depending on the surface condition. In practice, I have found it necessary to calibrate the pyrometer against a thermocouple embedded in a test coupon to establish the effective emissivity for the specific surface condition being welded. This calibration should be repeated whenever the surface preparation method changes.

The paper reports that the measured temperatures are in reasonable agreement with the calculated values. In my experience, agreement within 10-15% is considered acceptable for welding thermal simulation. The discrepancies typically arise from the following sources.

Error Source Typical Magnitude Mitigation Strategy
Heat source model parameters ±10% Calibrate against weld pool dimensions
Arc efficiency uncertainty ±10% Measure using calorimetric method
Emissivity uncertainty ±20% Calibrate pyrometer against thermocouple
Measurement position error ±5% Use high-speed camera to locate measurement point
Numerical discretization error ±5% Refine mesh in weld zone

4. Integration with Engineering Practice

4.1 Welding Procedure Qualification

The practical value of this simulation approach lies in its ability to reduce the number of physical welding procedure qualification (WPQ) trials. In the pipe industry, we typically qualify welding procedures for each combination of material grade, thickness, and joint geometry. Each qualification requires multiple test coupons, mechanical testing, and metallographic examination. The cost of a single WPQ can range from 5,000 to 50,000 USD depending on the material and testing requirements. If a validated thermal model can predict the temperature field accurately, it can be used to narrow down the process window before physical trials begin, saving both time and money.

For example, when qualifying a TIG root pass on 316L stainless steel pipe for a chemical plant, we used a thermal simulation to predict the cooling time from 800°C to 500°C (t8/5), which is the critical parameter controlling HAZ microstructure and hardness. The simulation showed that with a heat input of 0.8 kJ/mm and a preheat temperature of 100°C, the t8/5 would be approximately 8 seconds, which would keep the HAZ hardness below 220 HV, the maximum acceptable value for sour service applications. The subsequent physical trials confirmed the simulation, and we were able to qualify the procedure with only two test coupons instead of the usual five.

4.2 Residual Stress and Distortion Prediction

The temperature field is the input for residual stress and distortion analysis. In the fabrication of large-diameter LSAW pipe, the welding of the longitudinal seam creates significant residual stresses that can affect the dimensional accuracy of the pipe and its performance in service. The thermal simulation provides the temperature history that is then mapped onto a structural analysis model to calculate the residual stress distribution. In my experience, the residual stress at the weld centerline of a LSAW pipe can reach 70-90% of the yield strength of the base material. This is a critical consideration for pipelines that will be subjected to external pressure or bending loads, as the residual stresses can combine with service stresses to cause premature failure.

4.3 Defect Prevention

The thermal model can also help predict the likelihood of certain welding defects. For example, in TIG welding of aluminum alloys, the formation of porosity is closely related to the cooling rate and the hydrogen content of the weld pool. A slow cooling rate allows more time for hydrogen to escape, reducing porosity. However, a slow cooling rate also promotes grain growth in the HAZ, which can reduce the mechanical properties of the joint. The thermal simulation allows the engineer to optimize the heat input to balance these competing effects.

5. Key Questions and Reflections

5.1 The Validation Gap

One of the most important questions raised by this paper is the extent to which laboratory-scale validation translates to production-scale welding. The authors validated their model against a simple bead-on-plate weld on aluminum alloy. In production, we deal with complex joint geometries, multi-pass welding, and varying boundary conditions that are much more difficult to model. The question is whether the calibrated parameters from a simple test are transferable to a complex joint. In my experience, the heat source parameters are generally transferable if the welding process parameters (current, voltage, travel speed) are the same, but the boundary conditions may differ significantly due to the presence of fixtures, backing bars, and adjacent passes.

5.2 The Role of Phase Transformation

The paper considers the latent heat of fusion but does not explicitly model the solid-state phase transformations that occur in steel welding. For aluminum alloys, the only phase transformation is melting and solidification, so the latent heat approach is sufficient. However, for steel welding, the austenite-to-ferrite transformation during cooling releases additional latent heat that affects the cooling rate and the resulting microstructure. This is a significant limitation of the current model for steel applications. In practice, we often use empirical formulas such as the Rosenthal equation or the Rykalin-2D method to estimate the cooling time t8/5, but these methods do not account for the latent heat of transformation.

5.3 Practical Lessons Learned

Over the years, I have learned several lessons about thermal simulation that are worth sharing with fellow engineers.

  1. Always validate the model against at least one physical measurement before using it for prediction. A model that has not been validated is just a sophisticated guess.
  2. Calibrate the heat source parameters against the weld pool dimensions, not just the peak temperature. The weld pool dimensions are much more sensitive to the heat source parameters and provide a stronger validation criterion.
  3. Pay careful attention to the mesh density in the weld zone. A coarse mesh will smear the temperature gradient and produce inaccurate predictions of the cooling rate.
  4. Do not trust a simulation that predicts a temperature field that violates the laws of thermodynamics. For example, the peak temperature in the weld zone should not exceed the boiling point of the material.
  5. Use the simulation to explore the process window, not to replace physical trials entirely. The simulation can guide you to the most promising process parameters, but the final qualification must always be based on physical testing.

6. Study Insights and Outlook

This paper provides a solid foundation for the thermal simulation of TIG welding, but it also highlights the need for further development. The following areas deserve attention in future work.

Research Area Current Status Practical Significance
Multi-pass welding Limited Critical for thick-wall pipe welding
Solid-state phase transformation Not modeled Essential for steel welding
Electromagnetic stirring Not modeled Affects weld pool shape and grain size
Adaptive mesh refinement Not implemented Improves computational efficiency
Coupled thermal-fluid analysis Not included Necessary for weld pool dynamics

In the steel pipe and pipe fitting industry, the most pressing need is for a validated thermal model that can handle multi-pass welding of thick-wall pipe. A typical X70 line pipe with a wall thickness of 25.4 mm may require 10-15 passes for the longitudinal seam weld. Each pass modifies the thermal boundary conditions for the subsequent passes, and the accumulated thermal history determines the final microstructure and residual stress state. This is a challenging problem that requires both experimental validation and sophisticated numerical modeling.

The integration of thermal simulation with welding procedure qualification is the most immediate practical benefit. By reducing the number of physical trials required, this approach can save significant time and cost in the qualification of new welding procedures. As the models become more accurate and the computational tools become more accessible, I expect that thermal simulation will become a standard part of the welding engineering workflow, alongside the traditional methods of procedure qualification and testing.

The key takeaway from this paper is that the temperature field is the fundamental variable that controls the quality and performance of a welded joint. If we can predict the temperature field accurately, we can predict the microstructure, the mechanical properties, and the residual stresses. This is the path to a more scientific approach to welding engineering, and it is a path that I encourage every welding engineer to follow.