Experimental and Statistical Analyses of Dimensional Accuracy of Geometry-Dependent Features in Fiber Laser Cutting of CFRP

Abstract

This study investigates the influence of fiber laser cutting parameters on the dimensional accuracy of carbon fiber-reinforced polymer (CFRP) components. A full factorial design of experiments was employed considering laser power, cutting speed, and laser operating mode as input factors. Three laser power levels (800, 1600, and 2000 W), three cutting speeds (250, 500, and 1000 mm/min), and two operating modes (continuous wave, CW, and gated pulse, GP) were evaluated. Dimensional accuracy of 90 laser-cut specimens was assessed using three selected geometrical responses—the inner width (L2) and two internal feature heights (a1 and a3)—followed by analysis of variance and regression modeling. Laser power was identified as the dominant factor for L2 and a1, whereas a3 exhibited pronounced nonlinear behavior governed primarily by cutting speed and its quadratic contribution. The operating mode significantly affected L2 and a3, while its influence on a1 was statistically insignificant. The pronounced nonlinear behavior of a3 may be associated with the combined effects of heat accumulation, anisotropic heat transfer through the carbon-fiber reinforcement, and thermally induced degradation and recession of the polymer matrix. Differences in temporal energy delivery between CW and GP operation may influence the local thermal conditions and consequently affect material removal and the resulting position of the cut edge. The developed regression models achieved R2 values of 90.37%, 87.25%, and 92.22% for L2, a1, and a3, respectively. The calculated nominal line energy ranged from 48 to 480 J/mm and supported the observed tendency that higher power combined with lower cutting speed was generally associated with increased dimensional deviation. The results demonstrate that dimensional accuracy in fiber laser cutting of CFRP is strongly geometry-dependent and requires feature-specific process optimization rather than a single universal cutting condition.