NMR-based nondestructive evaluation of silica fume-modified coal gangue coarse aggregate concrete pore fractals

Quantitative non-destructive evaluation (QNDE) should convert non-destructive signals into engineering performance indicators, rather than only detect defects. This study used low-field nuclear magnetic resonance (NMR) to quantify the pore-volume distribution of silica fume-modified coal gangue coarse aggregate concrete (CGCAC) and relate it to compressive strength. The framework converts T2 relaxation signals into pore radius, pore volume, segmented fractal dimensions, and strength-related prediction parameters. Using the Menger sponge volume-fractal model, multi-scale pore fractal characteristics were evaluated and compared with global fractal fitting, microstructural observations, and regression models. The dominant pore size was 100–1000 nm, and pore volume ranged from 7.35 to 41.38 cm³. Reducing the water-to-binder ratio and adding silica fume decreased pore volume, with 10% silica fume showing the strongest refinement effect. Compared with global fitting, segmented regression improved fitting accuracy, especially for fine pores, coarse pores, and macro voids. Compressive strength positively correlated with fractal dimension. Among linear, quadratic, and power-law models, the linear model balanced accuracy and robustness best, with adjusted R² values above 0.83 and t- and F-test p-values below 0.01. These results indicate that NMR-derived fractal parameters can serve as quantitative non-destructive indicators for evaluating and predicting CGCAC mechanical performance.