Temperature Field Simulation of Dual-Wire Submerged Arc Overlay Using ABAQUS

Motivation and Technical Context

Submerged arc welding (SAW) is one of the most productive processes for weld overlay applications due to its high deposition rate, deep penetration, and ability to produce smooth, continuous weld beads. Dual-wire submerged arc welding (DW-SAW) further enhances productivity by depositing two wire electrodes simultaneously, effectively doubling the deposition rate while maintaining or even improving bead quality. However, the thermal behavior of DW-SAW overlay is significantly more complex than single-wire SAW, owing to the interaction of two heat sources, the asymmetric heat input distribution, and the dynamic evolution of the weld pool geometry. Understanding the temperature field distribution during DW-SAW overlay is essential for predicting microstructure evolution, residual stress development, and potential defects such as cracking, porosity, and lack of fusion.

The use of finite element analysis (FEA) with commercial software such as ABAQUS provides a powerful tool for simulating the thermal field during welding processes. By solving the heat conduction equation with appropriate boundary conditions and heat source models, FEA can predict the temperature distribution, cooling rates, and thermal history at any point in the workpiece. These predictions are invaluable for optimizing welding parameters, predicting microstructural outcomes, and assessing the risk of defects. The literature reviewed here presents a detailed FEA study of the temperature field during DW-SAW overlay, with particular attention to the effects of welding parameters on the thermal cycle and the resulting implications for overlay quality.

Finite Element Model Development

The FEA model described in the literature employs a moving heat source approach in ABAQUS, where the heat input from the welding arc is modeled as a distributed heat flux on the workpiece surface. The heat source model for DW-SAW is a dual-Gaussian distribution, where each wire electrode is represented by a separate Gaussian heat source with its own power, radius, and position. The two heat sources are offset from each other by a distance corresponding to the wire spacing, which is typically 15-25 mm for standard DW-SAW configurations. The total heat input is the sum of the individual wire powers, with the interaction between the two heat sources captured through the superposition of their thermal fields.

The thermal conductivity of the weld metal and base metal is modeled as temperature-dependent, using the Arrhenius equation for the solid phase and a modified model for the liquid phase that accounts for the reduced thermal conductivity of the molten pool. The density and specific heat are also defined as temperature-dependent functions. The boundary conditions include convection and radiation heat losses from the exposed surfaces, with a convection coefficient of approximately 5-10 W/(m^2-K) for still air and a radiation emissivity of 0.85 for oxidized steel surfaces. The initial temperature is set to 25 degrees Celsius, and the model is solved using an implicit time integration scheme with adaptive time stepping to capture the rapid temperature changes during the welding process.

Simulation Results and Analysis

The simulation results reveal several important features of the thermal field during DW-SAW overlay. First, the temperature distribution is highly asymmetric, with the peak temperature occurring at the midpoint between the two heat sources rather than directly beneath either wire. This is because the two heat sources partially overlap thermally, creating a zone of constructive interference where the temperatures from both sources add together. The peak temperature at the weld pool surface reaches approximately 2100 degrees Celsius, while the peak temperature in the depth direction is lower, approximately 1850 degrees Celsius, due to heat dissipation into the base metal.

Second, the cooling rate at the weld pool boundary is significantly higher than in single-wire SAW, primarily because the dual-wire configuration produces a wider and shallower weld pool with a larger surface area for heat loss. The maximum cooling rate from 800 to 500 degrees Celsius is approximately 25-35 degrees Celsius per second, which is within the range that promotes the formation of fine martensite in high-alloy overlay materials. This cooling rate is critical for predicting the microstructure of the overlay, as it determines the phase transformation kinetics and the resulting hardness and toughness.

Third, the thermal cycle at the bond line between the overlay and the base metal shows a peak temperature of approximately 1200-1400 degrees Celsius, which is below the melting point of the base metal but above the recrystallization temperature of austenitic stainless steel. This temperature range is sufficient to cause grain growth and phase transformation in the heat-affected zone of the base metal, which can affect the mechanical properties and corrosion resistance of the substrate. The simulation also predicts the residual thermal stress distribution, which shows tensile stresses in the overlay layer and compressive stresses in the base metal, consistent with the differential cooling between the overlay and substrate.

The simulation was validated by comparing the predicted temperature fields with experimental thermocouple measurements. Thermocouples were embedded at various depths and lateral positions in the workpiece, and the measured temperature-time histories were compared with the FEA predictions. The agreement was generally good, with deviations within 10-15% of the peak temperature and within 20% of the cooling rate. The larger deviations in cooling rate are attributed to the simplified boundary conditions used in the model, particularly the constant convection coefficient assumption, which does not fully capture the complex heat transfer during the solidification of the weld pool.

Process Optimization Insights

The FEA results provide valuable insights for optimizing the DW-SAW overlay process. The welding parameters that have the most significant influence on the thermal cycle are the travel speed, current, and wire spacing. Increasing the travel speed reduces the heat input per unit length, resulting in higher cooling rates and a narrower weld pool. Decreasing the wire spacing increases the thermal overlap between the two heat sources, raising the peak temperature but reducing the cooling rate at the bond line. The optimal wire spacing for a given application depends on the desired balance between deposition rate and thermal cycle control.

For overlay applications where a controlled cooling rate is critical for microstructure development, the FEA can be used to identify the parameter combinations that achieve the target cooling rate. For example, in overlaying a high-alloy austenitic stainless steel on carbon steel, a cooling rate of 20-40 degrees Celsius per second is desirable to promote full austenite transformation and prevent the formation of brittle delta ferrite. The FEA can predict the parameter window that achieves this cooling rate, reducing the need for extensive experimental trials.

Study Insights and Reflections

The use of FEA for welding process simulation represents a powerful tool for reducing development time and cost in weld overlay applications. By predicting the thermal cycle and residual stress distribution, FEA can guide the selection of welding parameters, consumables, and preheat temperatures before any experimental welding is performed. This is particularly valuable for DW-SAW overlay, where the complex thermal behavior makes empirical optimization time-consuming and expensive. The validation of the FEA model with experimental thermocouple measurements is essential to ensure the accuracy of the predictions, and the level of agreement achieved in this study provides confidence in using the model for process optimization.

One reflection from studying this work is the importance of developing accurate heat source models that capture the physics of the welding process. The dual-Gaussian model used here is a reasonable approximation for DW-SAW, but more sophisticated models that account for the electromagnetic interaction between the two wires and the fluid dynamics of the weld pool could provide even more accurate predictions. As computational capabilities continue to advance, multi-physics simulations that couple thermal, fluid, and solidification models will become increasingly accessible and will provide a more complete understanding of the welding process. The integration of FEA with metallurgical models for predicting phase transformation and microstructure evolution represents the next frontier in computational welding science.