Curve Fitting and Optimisation of MIG Welding Current for Stainless Steel
Literature Overview and Core Findings
The paper by Quanyong Wang and Xiaojuan Wu from Shenyang Ligong University takes a mathematical approach to a practical problem: selecting the welding current for stainless steel MIG welding as a function of plate thickness. The authors collected welding current data from two different sources, plotted scatter diagrams, and fitted first, second and third-order polynomial curves using the least squares method.
The results show that the third-order curve provides the best fit for both data sources, and the authors verify through experiment that the fitted equation has some guiding significance for determining the welding current range. This is a useful contribution, but it is also a classic example of the difference between a mathematical fit and a physical understanding.
Interpretation of the Fitting Methodology
The least squares method is a standard tool for fitting curves to data, but the choice of polynomial order requires careful judgement. A first-order curve is a straight line, which implies that welding current increases linearly with plate thickness. A second-order curve allows for curvature, and a third-order curve allows for an inflection point.
In reality, the relationship between welding current and plate thickness is not a simple polynomial. At low thicknesses, the current is limited by the need to avoid burn-through. At high thicknesses, the current is limited by the need to achieve full penetration without excessive heat input. The transition between these regimes is not smooth, and a third-order polynomial may not capture the physics.
The paper's finding that the third-order curve is superior to the first and second-order curves suggests that the data contain at least two inflection points, which is consistent with the physical transition from a conduction-limited regime to a convection-limited regime in the weld pool.
Practical Application Limits
The practical value of a fitted curve equation is that it provides a starting point for welding parameter selection. If an engineer needs to weld a 6 mm stainless steel plate, they can plug the thickness into the fitted equation and obtain a current value. However, the equation must be validated against the actual welding procedure qualification.
In my experience, the welding current is only one of several interdependent parameters. The voltage, travel speed, wire feed rate, shielding gas flow, and torch angle all interact. A curve that fits current versus thickness is useful, but it does not tell you the voltage or the travel speed. A welding procedure specification must include all parameters, not just the current.
The paper also does not address the effect of joint geometry. A butt joint with a 60-degree V-groove requires different current than a fillet weld with a 5 mm leg size, even at the same plate thickness. The curve equation is a starting point, not a substitute for a properly qualified welding procedure.
Lessons from Real Projects
I have seen many cases where an engineer used a current value from a data sheet or a fitted equation without considering the joint geometry or the welding position, and the result was a defective weld. In one project, a stainless steel pipe with a 6 mm wall thickness was welded with a current that was appropriate for a flat position butt weld, but the weld was performed in the vertical-up position. The current was too high, the weld pool sagged, and the result was a lack of fusion and excessive reinforcement.
The correct approach is to use the fitted equation as a preliminary estimate, then perform a welding procedure qualification with the actual joint geometry, welding position, and production equipment. The qualified parameters are the ones that go into the production welding procedure specification.
Reflections on the Mathematical Approach
The paper is a good example of how mathematical tools can support, but not replace, engineering judgement. The third-order polynomial fit is statistically superior to the lower-order fits, but the physical meaning of the coefficients is not clear. The equation is a black box that interpolates between data points, and its predictive power outside the range of the original data is uncertain.
For a production engineer, the practical takeaway is to maintain a database of welding parameters from qualified procedures and use statistical tools to identify trends and outliers. The fitted curve is a useful visualisation, but the real value is in understanding why the current must increase with thickness, and how the other parameters must be adjusted in concert.
The paper's experimental verification is commendable, but it would be even more valuable if the authors had reported the joint efficiency and the defect rate for the welds made with the fitted current values. This would connect the mathematical fit to the actual quality of the weld, which is what ultimately matters on the shop floor.
Zhuojin Pipe Fitting Co., Ltd