This study utilises parallel discrete element method (DEM) simulations of cone penetration tests (CPTs) and pile load tests to investigate the application of direct CPT-based methods for predicting the base capacity of bored piles in layered soils. To reduce boundary effects, a coupled DEM-finite difference method (FDM) model is constructed to simulate pile load tests. The study focuses on the scale effect of pile diameter on the correction factor αb and the effectiveness of existing qc averaging methods in layered soils. Two pile diameters and three soil layering conditions, featuring a single silt layer interbedded within sand at varying depths, are considered in the simulations. Results show that both soil layering and pile displacement influence the scale effect. At small settlements (s) up to 0.3 times the pile diameter (D), the scale effect is insignificant, except when a soft layer is directly above the pile tip. At larger settlements, particularly when s > 0.5D, piles with smaller diameters show more pronounced reductions in base resistance due to a weak layer closely beneath the tip. Among the four methods evaluated, the BD-18 qc averaging method produces more consistent αb values across various soil profiles and pile diameters. Microscopic analysis reveals that soils above the tip contribute less to the pile base resistance at s/D = 0.1 than to the cone tip resistance. Additionally, the strength mobilisation levels of soils in most soil layering conditions appear similar at s/D = 0.1, supporting the proposal of a constant αb value in direct CPT-based methods.