Experimental Investigation on the Effect of Laser Welding Parameters for P91 Steel Welding with Varying Shielding Gas Using Box-Behnken Design Methodology

In the current experiment, the fiber laser parameters were optimized for welding P91 steel using the Box-Behnken design (BBD). Bead-on-plate welds were produced with an 8000 W fiber laser welding machine, with the focal point, power, and welding speed of the laser varied under various shielding atmospheres. The input parameters were optimised to achieve the desired depth of penetration (Dp), underfill (U), top bead width (Tw), and heat-affected zone width (Hw), and their impact on output responses were analyzed using the analysis of variance (ANOVA). The results of bead-on-plate trial welds and analysis of variance demonstrated that welding speed has the most decisive influence on the depth of penetration (Dp), the width of top bead (Tw), and HAZ (Hw), while laser power has the most significant influence on underfill (U). The optimal solution for argon shielding resulted in a weld with a penetration depth of 5.5 mm, compared to full penetration weld with a narrow top bead and heat affected zone achieved in CO2 shielding atmosphere. The results of validation trials utilising the optimal parametric combination were found to be highly correlated with the expected values, confirming the mathematical model's adequacy. The high laser intensity used in CO2 shielding resulted in the formation of underfill with a depth of 0.456 mm. In both shielding atmospheres, the fusion zone microstructure exhibited untempered martensite in as-welded condition with a columnar lath structure and was devoid of delta-ferrite.

Automated summary

In this experiment, the fiber laser parameters were optimized for welding P91 steel, and the optimal solution was achieved. The results showed that welding speed had the most decisive influence on the welding effect, while laser power had significant influence on some welding characteristics. The obtained welding results were highly correlated with the expected values.