Abstract
Dissimilar metal welding (DMW) is needed in several economic sectors, including transportation infrastructure, nuclear power, aerospace, automotive, and defense. One area where DMW could have a significant impact within the Commonwealth of Virginia is bridge construction and repair, where the use of stainless steel could be used to enhance corrosion resistance and strength. Decreased maintenance costs are expected to help offset the significant upfront expense of a stainless-steel construction or repair. DMW would enable even those upfront costs to be decreased by only using stainless steel in select locations. However, complications can arise due to differences in the physical properties of stainless and traditional carbon steels. The present research is focused on the effective joining of a relatively low-cost stainless steel (ASTM A709 50CR) to conventional carbon steels (A709-Grade 36 structural and A709-Grade 50W weathering steels) using austenitic and austenitic-ferritic stainless steel filler materials (309L & 312, respectively). In industry, a range of welding processes, geometries, and heat inputs are employed, which can lead to solidification and cold cracks if not properly selected and configured. The main objective of this study is to identify metallurgical aspects behind the observed cracking and provide engineering solutions that practitioners in the field can use to avoid such weld failures. The influences of chemical composition, phase fractions, microstructural evolution, thermal contraction (shrinkage), and grain size were all considered within three interrelated sections.
The first section concerns the mitigation of specific cracking scenarios in dissimilar metal welds. The fusion zone of DMW exhibits complex phase evolution governed by dilution and solidification modes. With increasing dilution from the carbon steel side, the Creq/Nieq ratio is decreased, shifting the solidification mode from ferritic-austenitic to fully austenitic. Solidification cracking was primarily localized to fully austenitic regions near the carbon steel interface, where suppression of delta ferrite limited impurity accommodation. During terminal solidification, these impurity-enriched grain boundaries, which are mechanically weak due to low-melting eutectics, were unable to withstand the tensile stress generated by constrained solidification shrinkage. Cracks are intergranular, propagating along high-angle grain boundaries as verified by electron backscatter diffraction (EBSD). Kernel average misorientation (KAM) mapping highlights the presence of localized plastic deformation at crack tips, which enabled the shrinkage-induced stress to be estimated as ~270 MPa based on principles of fracture mechanics. Cold cracking was identified in the root pass where the fusion zone contacted the carbon steel backing plates. This backing plate configuration led to rapid heat extraction as well as carbon enrichment (dilution of welds), which destabilized austenite and promoted martensite formation.
Based on these mechanistic insights, mitigation strategies were developed. Solidification cracking was eliminated by reducing heat input, increasing welding speed, and modifying the joint design to a single-V groove with backgouging. This series of engineering changes effectively minimized dilution and stabilized sufficient delta ferrite to avoid low melting point eutectic formation. In addition, by removing the carbon steel backing plate, cold cracking was resolved, and this eliminated the rapid cooling and carbon contamination, which gave rise to the formation of brittle martensite in the root pass. Together, these combined engineering modifications effectively controlled phase evolution and shrinkage-induced internal stress, enabling defect-free dissimilar metal welds.
The second section concerns improvements in the accuracy of delta ferrite predictions. While it can be undesirable in some cases, delta ferrite is also well known for enhancing resistance to solidification cracking in stainless steel welds. Its presence is particularly important in austenitic-ferritic steels, where phase balance directly influences the solidification cracking behavior. Accurate prediction of delta ferrite is therefore critical for ensuring weld quality and structural reliability. Over the last 80 years, empirical constitutional diagrams, most notably the Schaeffler and WRC-1992 diagrams, have been widely used to estimate ferrite content based on Cr and Ni equivalent composition values. However, these models have significant limitations: the Schaeffler diagram omits N and Cu and overestimates the ferrite-stabilizing role of Si, often resulting in ferrite overprediction, while the WRC-1992 diagram excludes Si and Mn entirely. Both systems fail to account for the non-linear influence of C and N in the austenite phase at higher concentrations, leading to severe overestimation of austenite and underprediction of delta ferrite, particularly in higher carbon-containing stainless steel filler metals like 312 (0.15% C), resulting in errors exceeding 40–60%. Slope-based models, which derive element coefficients from the austenite solvus line, i.e., the γ/(γ+α) phase boundary in binary systems of iron and alloying elements, have been proposed, but they consider only binary interactions and ignore kinetic and multi-element effects. These limitations also lead to large discrepancies (50%) in ferrite prediction for higher carbon-containing stainless steels. Therefore, both approaches fall short in accurately predicting delta ferrite in such compositions.
To address these shortcomings, this study introduces a hybrid approach. Slope-based analysis was used to assess the austenite and ferrite-stabilizing tendencies of individual elements, while empirical coefficients were derived directly from experimental results on 309L and 312 dissimilar welds performed in this research. These coefficients account for actual phase partitioning behavior and the nonlinear influence of C and N in the austenite phase observed through electron microscopy and CALPHAD simulations. For low-carbon stainless steels (C ≤ 0.08 wt.%), the refined equivalents are Creq = Cr + Mo + 0.7Nb and Nieq = Ni + 35C + 20N + 0.25Cu + 0.5Mn. For stainless steels with higher carbon contents (0.08 < C ≤ 0.15 wt.%), where the concentration of carbon in the austenite phase does not increase proportionally with increasing bulk C content, the Nieq was revised to Ni + 7(C + N) + 0.25Cu + 0.5Mn. The modified model showed high prediction accuracy for 80 different stainless steels of ASTM A959−19, including higher carbon-containing grades, where traditional diagrams exhibited significant errors, demonstrating improved reliability across a broad compositional range.
The final section documents the effects of filler metal type on welding processes. Literature suggests that cored wires, which contain internal flux and metal powder, facilitate a projected metal transfer mode by promoting droplet detachment and more uniform solidification compared to solid wire. However, a complete understanding of how solid and metal-cored wires influence solidification cracking in dissimilar metal welds remains limited. In flux-cored arc welding (FCAW) and shielded metal arc welding (SMAW) processes, flux-cored wire exhibited no solidification cracking, while solid wire in SMAW showed minor cracks near the carbon steel interface. FCAW also results in a narrower heat-affected zone (HAZ) and lower geometric dilution than SMAW under similar parameters. In the submerged arc welding (SAW) process, cracking was more pronounced when solid wire was used. Metal-cored wire, by contrast, enabled crack-free welds at 36 kJ/in and travel speed of 18 in/min, whereas solid wire required lower heat input (31 kJ/in) and faster travel speed (20 in/min) to avoid cracking. Features such as a finer, delta ferrite-containing structure, more tortuous (longer) grain boundaries, narrower HAZs, and smoother fusion lines are more prevalent in metal-cored welds, which contribute to enhanced resistance against solidification cracking. These findings further highlight the well-known fact that both filler metal wire type and welding process play a critical role in mitigating cracking in dissimilar metal welds.