Solid-State Phase Transformation and Residual Stress Simulation in Ti6Al4V Thin Plate TIG Welding
Literature Overview and Why It Matters
This paper from the China Ship Scientific Research Center and Shanghai Jiao Tong University addresses a problem I have wrestled with on the shop floor for decades: how do we predict what is going to happen inside a weld joint before we ever strike an arc? The authors tackle Ti6Al4V titanium alloy thin plate TIG welding, and they do it properly, by building a coupled thermal-metallurgical-mechanical finite element model that accounts for solid-state phase transformation. For anyone who has ever welded titanium pipe or plate for offshore, aerospace or marine service, this is not academic trivia. Titanium is unforgiving. It oxidises above roughly 600°C, it picks up hydrogen and oxygen readily, and its α/β phase balance dictates mechanical properties in a way that steel simply does not. The authors' approach, using the JMAK equation for diffusional transformation and the Koistinen-Marburger equation for martensitic transformation, is the right way to model what actually happens in the weld HAZ and fusion zone.
Core Technical Points and Model Development
The paper develops a mathematical model of solid-state phase transformation for Ti6Al4V, then implements it in Abaqus as a thermal-metallurgical-mechanical coupled analysis. The key input parameters are the phase transformation kinetics, the temperature-dependent thermophysical properties of each phase, and the transformation strain associated with the β→α and β→α′ transformations. The JMAK equation handles the diffusional β→α transformation, while the K-M equation handles the athermal β→α′ martensitic transformation. This distinction is critical because the cooling rate in TIG welding of a 1.6 mm thin plate is high enough to form significant amounts of α′ martensite, which has different mechanical properties and a different specific volume than the equilibrium α phase. The authors then validate their model against hardness measurements and residual stress measurements on actual TIG welded joints.
| Model Component | Equation or Approach | Purpose |
|---|---|---|
| Diffusional transformation | JMAK equation | Models β→α transformation during cooling |
| Martensitic transformation | Koistinen-Marburger equation | Models β→α′ athermal transformation |
| Coupling method | Thermal-metallurgical-mechanical FE in Abaqus | Links temperature history to phase fractions and stress state |
| Validation | Hardness testing + residual stress measurement | Confirms model accuracy |
| Plate thickness | 1.6 mm Ti6Al4V | Thin plate, high cooling rate regime |
Key Findings and Interpretation
The most important result is that the hardness distribution across the weld joint matches the simulated phase composition distribution. This is a strong validation because hardness is a direct proxy for the α/α′/β phase mix. The authors also demonstrate that the difference in thermophysical properties between α and β phases, and the difference in their proportions across the joint, significantly affects residual stress distribution and deformation. In practical terms, this means that if you change your welding parameters and shift the α/β balance, you are also shifting your residual stress field. For a thin-walled titanium pipe or a titanium plate structure, that can mean the difference between a part that stays within tolerance and one that distorts beyond repair.
Engineering Practice Insights
I have seen this play out in real production. When we weld thin titanium pipe for heat exchanger or offshore service, we often use a backing gas purge to prevent oxidation, but we also have to manage heat input carefully because the thin wall cannot tolerate excessive heat without distortion. The simulation approach in this paper gives us a tool to predict the phase balance and residual stress before we commit to a weld schedule. In practice, this means we can optimise the welding current, travel speed and interpass temperature to achieve the desired α/α′ ratio and a residual stress field that will not cause distortion or stress corrosion cracking in service. The authors' finding that α and β phase property differences dominate the residual stress response is a practical warning: do not assume that a single set of material properties will give you accurate predictions across the whole joint. You need to model the phase-dependent properties, or your residual stress predictions will be off by a significant margin.
Key Questions and Reflections
One question I would raise is whether the model captures the effect of hydrogen pickup on the phase transformation kinetics. In real titanium welding, even with good shielding, some hydrogen can be introduced, and hydrogen stabilises the β phase and can cause delayed cracking. The paper does not appear to address this. Another question is whether the model can be extended to multi-pass welding, where the thermal history is more complex and the phase transformation in the HAZ is affected by repeated thermal cycles. For thick titanium pipe or plate, this matters. My own experience says that the single-pass thin plate case is a good starting point, but the industry needs the multi-pass extension before we can fully trust the predictions for heavy-section titanium weldments.
Study Insights and Implications
This paper reinforces a lesson I learned the hard way: welding simulation is only as good as the material model behind it. If you ignore phase transformation, you will get residual stress predictions that are qualitatively wrong in sign and magnitude, especially in titanium and other materials with strong transformation effects. The authors have done the industry a service by demonstrating a rigorous coupled approach and validating it against experiment. For the pipe and fitting industry, where titanium is increasingly used in seawater piping, heat exchangers and offshore components, this kind of predictive capability is essential for qualifying weld procedures and avoiding costly rework.
Concluding Summary
The paper successfully demonstrates that a coupled thermal-metallurgical-mechanical finite element model, incorporating JMAK and K-M phase transformation kinetics, can predict the microstructure and residual stress distribution in Ti6Al4V thin plate TIG welds with good accuracy. The validation against hardness and residual stress measurements confirms the model's physical soundness. The key practical insight is that phase-dependent thermophysical properties and the α/β phase balance significantly influence residual stress and deformation, meaning that weld parameter optimisation must account for phase transformation effects. This work provides a solid foundation for optimising titanium welding process parameters and improving joint mechanical properties, and it deserves a place on the desk of any engineer responsible for titanium weld procedure qualification.
Zhuojin Pipe Fitting Co., Ltd