Predicting laser weld variability in 304 stainless steel resulting from composition variation
Nicholas W. Finch, Todd A. Palmer, and Tarasankar DebRoy
Department of Materials Science and Engineering
The Pennsylvania State University
Keywords: Laser welding, 304 stainless steel, Composition variation, Numerical heat transfer model
Essential variables for controlling laser welding processes are typically developed through an extensive trial-and-error process. Tight controls and characterization of the material, welds, and machines used during this process enable high repeatability and high accuracy weld structures. However, this process yields essentially no predictive capability for assessing the impact that small, but allowable, changes in incoming material and processing variables have on weld geometry and defect generation. Although the standard for compositions of 304 and 304L are given by ASTM A240/A240M, the existing body of literature shows the regular occurrence of contaminants without an acceptable range having been established. Although extensive work is extant to detailing process variables such as defocus distance, travel speed, and laser spot size, very little work exists on establishing the effects of small variations about a set of process conditions. By using thermodynamic models for assessing composition variation on material properties and numerical models for assessing how changes in material properties affect weld geometry and properties, predictive assessments can be made for the impact that changes in material composition and process conditions will have.
Establishing the composition variation for 304 and 304L materials was done by using the measured composition from papers where laser welding was performed. Material properties for the literature compositions were calculated using a commercially available software, JMatPro V8. The properties pertinent to the numerical models used are solidus and liquidus temperatures, coefficient of thermal expansion, enthalpy, heat capacity, viscosity, and surface tension. The calculated material properties were then used in in-house numerical models for both conduction and keyhole mode welding. Static enhancement factors for liquid thermal conductivity and liquid viscosity are used to capture the enhancing effect of turbulence on these variables. The development of these models is detailed elsewhere. Parameters for conduction mode and keyhole mode welds are taken from the literature.
Results and Discussion
In looking at literature where 304/L material was laser welded, 16 alloying elements appeared. The specified elements in ASTM A240/A240M are Cr, Ni, Mn, Si, C, N, S, and P. Of the specified elements, underreporting of C, Si, S, N, and P was observed. Additional elements found are Mo, Cu, Nb, Co, Ti, V, and Al. Of the additional elements observed, molybdenum and copper appear most frequently. Material properties for the calculated compositions have been found to deviate from the analytical standards that have been used in prior work.
The thermophysical properties and final weld dimensions were found to have a strong dependence on the composition. The properties found to be the most sensitive to composition are solidus temperature, enthalpy of the solid at melting, and the specific heat at the solidus. All properties requisite for the models used showed compositional dependence.
To evaluate the relative effects of the alloying elements a statistical analysis was performed to determine what elements of the composition have a statistically significant influence on the composition. The elements found to have a statistically significant influence on weld dimensions are nickel, silicon, carbon, phosphorous, sulfur, and copper. Although the influence of boron was statistically significant, boron has outsized effects on the dimensions of weld pools and has been rejected as an outlier for composition and weld dimensions.
Although incoming material compositions may fall within industrial specifications, contaminant elements prove to be important to the thermophysical variables that affect the overall weld dimensions. Noted from the literature survey is an underreporting of C, N, S, and P, elements that have long been established as required variables for determining the chemical properties of stainless steels. The effects of contaminant elements may far outstrip that of specified elements. In considering the quantification of the variability of weld dimensions, the incoming material composition has a strong effect on the variability of weld dimensions. Numerical welding models and thermodynamic models can be used together with analytical tools to anticipate how changes in the incoming material composition will affect final weld dimensions and establish what alloying elements the process is sensitive to.