Mixing intensity and spatial strength variability in laboratory deep-mixed columns

Deep mixing method (DMM) column quality is commonly characterised by the coefficient of variation (COV) of unconfined compressive strength, which cannot distinguish depth-wise variability from local horizontal heterogeneity. Internal strength variability was characterised for laboratory deep-mixed columns prepared at four blade revolution counts, T2, T4, T6 and T8 (2000, 4000, 6000 and 8000 rev/m), using a miniature DMM apparatus on kaolin clay. Needle penetration tests at 20 positions on 11–12 slices per column provided spatially resolved quNP data. Mean quNP did not increase monotonically with blade revolution count, whereas overall COV decreased from 0.168 at T2 to 0.089 at T8. A two-component variance decomposition revealed distinct spatial patterns: local horizontal heterogeneity dominated at T2 and T4, depth-structured variability dominated at T6, and both components were substantially reduced at T8. Normal and lognormal assumptions were not consistently supported across mixing intensities, with tentative support for a normal approximation found only at T8. Vertical correlation lengths were comparable to or shorter than the 3 cm slice spacing, and horizontal autocorrelation was negligible at the present grid resolution. Overall COV alone is therefore insufficient for DMM quality characterisation; depth-structured columns require denser vertical sampling, whereas locally heterogeneous columns require greater horizontal coverage.