Numerical Simulation of TIG Weld Pool Morphology Evolution

Literature Overview

This paper, published in 2009 by Wu Jianhua from Shenyang Vocational and Technical College, addresses the numerical simulation of weld pool morphology evolution during gas tungsten arc welding (GTAW/TIG). The work appears in the journal "Hot Working Technology" and represents an early-stage contribution to computational modelling of the thermal-fluid dynamics within the molten weld pool. Understanding weld pool geometry is fundamental to predicting dilution rates, microstructure development, and residual stress fields — all critical parameters in cladding and overlay welding applications.

Core Technical Content

The study focuses on solving the coupled thermal-mass-momentum equations governing the weld pool behaviour under a TIG arc. The governing equations typically include:

Energy equation: Describes heat transfer by conduction, convection within the liquid pool, and heat input from the arc
Continuity equation: Mass conservation within the fluid domain
Momentum equation: Incorporates body forces such as Lorentz force, surface tension (Marangoni effect), buoyancy, and electromagnetic stirring

Key Modelling Parameters

Parameter	Typical Range in TIG	Role in Simulation
Arc current	80–300 A	Determines heat input magnitude
Arc voltage	12–25 V	Governs arc column geometry
Travel speed	200–800 mm/min	Controls heat input per unit length
Heat input	0.3–1.5 kJ/mm	Primary driver of pool geometry
Marangoni coefficient	-0.01 to -0.03 N/m/K (steel)	Drives surface convection
Surface tension gradient	dγ/dT (negative for Fe-S)	Pool surface flow direction

Interpretation of Technical Points

The weld pool morphology evolution is fundamentally governed by the competition between thermal conduction (which tends to create a shallow, wide pool) and Marangoni convection (which can deepen or flatten the pool depending on the surface tension gradient sign). In the numerical framework, the boundary conditions at the free surface of the weld pool are particularly challenging. The pressure boundary condition includes surface tension effects described by the Young-Laplace equation, while the thermal boundary accounts for radiation and convection losses to the surrounding atmosphere.

The evolution of pool geometry over time — from initial heating, through steady-state formation, to trailing solidification — represents a transient problem. The simulation must capture:

The initial thermal diffusion phase before melting begins
The steady-state pool shape achieved when heat input balances heat dissipation
The trailing edge solidification front and its relationship to grain growth direction

Process and Standards Analysis

In the context of weld overlay and cladding applications, the weld pool geometry directly determines:

Dilution ratio: The volume fraction of base metal in the deposited layer, which is critical for maintaining alloy composition in overlay welds
Bond line quality: The fusion line geometry affects the metallurgical bond between base and overlay
Microstructural gradient: Pool shape influences cooling rates and thus grain size, phase distribution, and hardness profiles

For overlay welding per NB/T 47014 and ASME IX, the dilution rate must be controlled within specified limits. A numerically predicted pool geometry allows engineers to optimise welding parameters before physical trials, reducing development time and material consumption.

Integration with Engineering Practice

In practical cladding operations, such as multi-pass overlay of Inconel 625 on carbon steel pressure vessels, the dilution in the first pass is typically 20–40%, decreasing to 5–15% in subsequent passes. Numerical simulation of pool geometry can predict these dilution values with reasonable accuracy (typically within ±5 percentage points), provided the material property inputs (thermal conductivity, specific heat, density) are temperature-dependent and the arc heat source model is properly calibrated.

A key insight from this type of simulation work is that the pool aspect ratio (depth-to-width ratio) is highly sensitive to the Marangoni coefficient. For low-sulphur steels where the surface tension gradient is positive (dγ/dT > 0), the pool becomes shallow and wide, while for high-sulphur steels with negative gradient, deep penetration is achieved. This has direct implications for choosing filler materials in overlay welding where controlled penetration is essential.

Key Questions and Reflections

Several questions arise from studying this work:

How sensitive are the simulation results to the assumed arc heat source model (Gaussian vs. double-ellipse vs. conical)?
What is the accuracy of the simulation when extended to multi-pass overlay scenarios with complex thermal histories?
How can the predicted pool geometry be validated experimentally — through cross-sectional macrographs, thermal imaging, or thermocouple arrays?

The 2009 timeframe of this publication places it in an era when computational resources were limited, and simulations were typically two-dimensional. Modern three-dimensional transient simulations with fully coupled electromagnetic-thermal-fluid models would provide significantly more accurate predictions, particularly for pool oscillation and droplet detachment phenomena.

Study Insights and Implications

This work represents an important foundational contribution to the understanding of TIG weld pool dynamics. For engineers engaged in cladding and overlay welding, the key takeaway is that weld pool geometry is not merely an outcome of welding parameters but can be predicted and controlled through proper understanding of the underlying physics. The numerical approach enables virtual optimisation of welding procedures, which is particularly valuable when working with expensive alloy systems such as nickel-based overlays where trial-and-error development is prohibitively costly. The methodology also provides a framework for troubleshooting defects such as undercuts, incomplete fusion, and excessive dilution by identifying the physical mechanisms responsible for each defect type.