Prediction of the saturated hydraulic conductivity based on X-ray CT analysis of a permeable mixture bonded with a polyurethane binder

Soil sealing by pavements disrupts the natural hydrological cycle, reducing infiltration and increasing the risk of urban flooding. Permeable pavements offer a sustainable solution not widely implemented in cities due to their limited structural performance. This limitation was addressed with the development of a Polyurethane-Bound Permeable Mixture (PBPM), which demonstrated enhanced structural performance and high drainage capacity. Since the primary function of permeable pavements is water drainage, accurate hydraulic conductivity (Kv) prediction is essential for optimizing urban drainage infrastructure; nevertheless, no previous study has applied pore network modeling (PNM) simulations to permeable pavements. This study evaluates Kv prediction for PBPM by comparing Darcy’s averaging models and PNM simulations, both based on microstructural parameters extracted from X-ray computed tomography. Kv was measured and predicted for three permeable mixtures: two polyurethane-based mixtures with maximum aggregate sizes of 8 mm (PBPM8) and 5 mm (PBPM5), and one bitumen-based mixture with a maximum aggregate size of 8 mm (PA8). Darcy’s averaging models showed a mean deviation of 19% from experimental Kv values but exhibited strong sensitivity to the selected representative pore diameter, leading to large variability among models. In contrast, PNM showed a higher mean deviation (28%) but greater consistency, reflected by stable prediction trends and systematic improvement in accuracy as pore and throat size inputs were refined. These results indicate that while Darcy-based models may achieve higher accuracy in isolated cases, PNM provides a more robust and reliable framework due to its capacity for progressive Kv prediction refinement. • Hydraulic conductivity (Kv) prediction for a polyurethane-based permeable Mixture. • Kv prediction based on Darcy´s averaging models (DAM) and Pore Network Model (PNM). • Kv predictions within the same order of magnitude as measurements. • DAM was more accurate, but PNM was more consistent and can improve further. • Further studies are needed to enhance the accuracy of PNM.