Thermo-Mechanical Weld Process Model

• Thermo-mechanical weld process models are computational frameworks that predict coupled thermal and mechanical phenomena using partial differential equations and constitutive laws.
• They integrate thermal analysis, mechanical stress computations, and microstructural evolution to optimize weld parameters in demanding industrial applications.
• Advanced models employ inverse calibration and fully coupled solution strategies to ensure accurate predictions of weld integrity, residual stresses, and phase transformations.

A thermo-mechanical weld process model is a computational framework that predicts the coupled thermal and mechanical phenomena in a welded joint, accounting for heat generation and dissipation, transient temperature fields, phase changes, plastic flow, residual stress development, and—where required—the evolution of microstructure and mechanical properties. Such models are central to understanding weld integrity, process parameter optimization, and the reliable design of welded components in demanding industrial sectors.

1. Fundamentals and Governing Equations

The foundation of thermo-mechanical weld process modeling is a set of coupled partial differential equations reflecting mass conservation, momentum balance, and energy conservation under highly nonlinear, temperature- and phase-dependent constitutive behavior. In three dimensions, welding is typically simulated under the incompressibility constraint,

$\divergence \mathbf{u} = 0$

where $\mathbf{u}(\mathbf{x})$ is the velocity field in the Eulerian reference frame (Dawson et al., 2014).

Momentum balance neglects inertia and body forces: $\divergence\boldsymbol{\sigma} = \mathbf{0}$ where $\boldsymbol{\sigma}$ is the Cauchy stress tensor, and the stress is further decomposed into deviatoric and hydrostatic parts for plastic flow.

Energy conservation is cast as

$-\divergence[\kappa(T)\nabla T] + \rho c_p(T) \mathbf{u} \cdot \nabla T = \dot{\mathcal{Q}}$

where $\dot{\mathcal{Q}}$ is heat generation from plastic dissipation, $\rho$ is density, $c_p$ is heat capacity, and $\kappa$ is conductivity; all quantities are generally temperature-dependent. For processes utilizing phase-transforming alloys, energy conservation must be adapted to include latent heat and enthalpy changes associated with solid-state or melt-solid transitions (Wijnen et al., 2024).

Constitutive laws governing plastic flow are typically implemented via nonlinear viscoplasticity (e.g., Kocks–Mecking, Johnson–Cook, or Zener–Hollomon-type models), with internal state variables tracking evolving microstructural strength and saturation kinetics (Dawson et al., 2014, Li et al., 2021).

2. Boundary Conditions and Interface Physics

Boundary conditions must capture both mechanical and thermal loads at weld interfaces. At the tool-workpiece contact (e.g., friction stir welding), the normal velocity is set to zero, while tangential traction is prescribed to match measured torque: $$T_{\rm sim} = \mathbf{a}\cdot\int_{S_{\rm probe}} \mathbf{r}\times(\hat t\,\boldsymbol{\tau}_\mathrm{dir})\,dS$$ $\hat t$ is scaled so that $T_{\rm sim}=T_{\rm meas}$. A uniform temperature $T_{\rm interface}$ is enforced on the probe surface, driving heat flux into the workpiece via both frictional and plastic dissipation (Dawson et al., 2014).

For welding processes such as laser beam or arc welding, the heat input is modeled via prescribed temperature or moving heat source: $$\rho c(T)\frac{\partial T}{\partial t} = \nabla\cdot\big[k(T)\nabla T\big] + Q_{\text{source}}$$ where $Q_{\text{source}}$ reflects localized heating from the arc or laser. Convective and radiative losses on exposed surfaces use mixed boundary conditions: $$-k(T) \nabla T\cdot\mathbf{n} = h_c(T-T_\infty) + \epsilon \sigma_{\rm SB}[(T+273)^4 - (T_\infty+273)^4]$$ with prescribed values for heat transfer coefficient $h_c$, emissivity $\epsilon$, and ambient $T_\infty$ (Castro et al., 1 Jan 2026).

3. Coupled Thermal-Mechanical Solution Strategies

Weld process models commonly employ sequential or fully coupled solution strategies. In sequential approaches, a thermal analysis precedes the mechanical step, with the computed temperature field dictating thermal strains and locally temperature-dependent properties: $$\epsilon = \epsilon^e + \epsilon^p + \epsilon^{th}_{\rm mech}$$

$$\sigma = C(T):(\epsilon - \epsilon^p - \epsilon^{th})$$

The mechanical computation then yields plastic strains and residual stresses based on the evolving temperature distribution (Castro et al., 1 Jan 2026, Wijnen et al., 2024).

Fully coupled solutions tackle the nonlinear entanglement of heat generation and dissipation during rapid plastic flow (e.g., magnetic pulse welding, high-rate arc or laser welding) using monolithic Newton–Raphson strategies and spatial-temporal discretization via finite elements, with high-performance solvers (e.g., PETSc/GMRES with additive Schwarz preconditioning) applied for massive scale models (Bevilacqua et al., 12 Mar 2025, Proell et al., 2021).

4. Inverse Estimation, Error Quantification, and Parameter Calibration

A distinguishing feature of contemporary weld models is the use of inverse estimation procedures to determine interface conditions. By matching computed quantities (axial weld force, torque) to experimental measurements,

$$F_{\rm sim} = \int_{S_{\rm inlet}}\mathbf{t}\cdot\mathbf{n}\,dS$$

$$T_{\rm sim} = \mathbf{a}\cdot\int_{S_{\rm probe}}\mathbf{r}\times\mathbf{t}^f\,dS$$

one iteratively adjusts $T_{\rm interface}$ and traction magnitude until convergence criteria $$\Phi(T_{\rm interface}) < \varepsilon_F,\,|T_{\rm sim}-T_{\rm meas}|<\varepsilon_T$$ are satisfied (Dawson et al., 2014).

Multimesh error estimation under mesh refinement quantifies numerical convergence: $\varepsilon_h = \frac{|F_h - F_{h/2}|}{|F_{h/2}|}$ confirming linear rates of convergence and robustness when bridging experimental and simulated fields.

Calibration against laboratory temperature histories, microhardness, and residual stress maps is standard (Wijnen et al., 2024, Castro et al., 1 Jan 2026). Material property curves (thermal expansion, conductivity, yield stress vs. $T$) are interpolated from experimentally determined datasets or referenced standards (Bevilacqua et al., 12 Mar 2025, Wijnen et al., 2024).

5. Microstructural and Metallurgical Modeling Integration

Advanced models incorporate direct simulation of phase fractions (ferrite, pearlite, bainite, austenite, martensite) via isothermal/semi-empirical transformations (e.g., Kirkaldy–Li, Leblond–Devaux kinetics), and grain evolution (Pous–Romero) (Wijnen et al., 2024). Local mechanical properties are then mapped by rule-of-mixtures using phase fractions, enabling spatially resolved prediction of hardness, fracture resistance, and transformation-induced plasticity.

For hydrogen-containing environments, phase-field fracture and chemo-transport models are coupled to the thermal-metallurgical solution, capturing hydrogen diffusion, trapping, and hydrogen-assisted brittle failure using microstructure-dependent $G_c(C)$ and strain energy splits (Castro et al., 1 Jan 2026, Wijnen et al., 2024).

6. Application Examples and Validation

Thermo-mechanical process models have enabled quantitative agreement with:

• Friction stir welding torque, interface temperature, and power dissipation under varying weld speeds for Ti-5111 alloys (Dawson et al., 2014).
• Morphological predictions validated against SEM in Al/Cu magnetic pulse welding (unwelded, vortex, wavy, IM layers) with deformation bands and temperature fields matching experiment (Li et al., 2021).
• Laser beam welding strain localization and solidification crack predictions that reproduce CTW experimental bands and failure patterns via large-scale parallel FE computation (Bevilacqua et al., 12 Mar 2025).
• Multi-pass shielded arc girth weld residual stress and HAZ dimensions in X80 pipeline steel matching neutron diffraction and metallographic width maps (Castro et al., 1 Jan 2026).
• Weld region microhardness distributions and residual stresses in pipeline steels, with parametric studies demonstrating sensitivity to heat input, filler composition, and bead order (Wijnen et al., 2024).

7. Model Limitations, Extensions, and Prospects

Limitations persist regarding the inclusion of latent heat effects, phase transformations, large deformations, and explicit weld defect modeling. Coupled chemo-mechanical and fracture models are emerging to address hydrogen embrittlement and microstructure-sensitive failure under service environments (Castro et al., 1 Jan 2026, Wijnen et al., 2024). Mortar-based mesh-tying in macroscale process simulations enables flexible layer/region coupling with non-matching meshes, essential for additive manufacturing and complex multi-pass weld geometries (Proell et al., 2021).

In summary, thermo-mechanical weld process models provide a rigorous, experimentally validated pathway to predicting temperature fields, material flow, residual stresses, microstructural evolution, and their impact on weld integrity under both conventional and extreme welding protocols. They form the computational backbone of risk assessment, process design, and structural reliability in contemporary welded systems.