| description abstract | Abstract. Erosion measurements inherently involve uncertainty due to the complex interaction of fluid and particle dynamics. Although a variation factor of 0.5–2 is typically considered acceptable between repeated tests, experimental erosion often shows deviations by orders of magnitude. This highlights a critical gap in the literature regarding the reliability and repeatability of erosion measurements. The present work addresses this challenge by applying statistical techniques to quantify and reduce uncertainty, thereby improving confidence in both experimental and computational fluid dynamics (CFD)-based erosion predictions. Gas-sand erosion experiments selected from the literature are repeated to validate and refine experimental data. A standard 3-in. (76.2 mm) stainless steel elbow with a bend radius to pipe diameter ratio (r/D) equal to 1.5 is used. Wall thickness losses are measured using fixed-mounted ultrasonic transducers at six locations on the outer wall for gas velocities of 15, 23, and 31 m/s and sand particle sizes of 75 and 300 μm. A statistical approach employing the 99% confidence interval is used to conservatively identify and assess statistical anomalies. Upper and lower bounds of erosion are presented to visualize the bands of uncertainty around the average profile. By applying this analysis, the variance of the experimental erosion values is reduced by 20–99% at different locations. Input, modeling, and numerical uncertainties related to CFD simulations are also quantified. Finally, experimental erosion profiles are compared with CFD results, and the propagation of experimental and CFD uncertainties is analyzed for increasing gas velocities and both sand particle sizes. | |
