Direct Numerical Simulations of Flow and Transport in Karst Conduits

Karst aquifers, with their intricate conduit networks, govern groundwater flow and contaminant transport. The strongly rough walls (k/D ≈ 0.1), irregular cross-sections, and bifurcating paths typical of karst conduits complicate prediction of flow behaviour, friction losses, and turbulence onset.

This work uses direct numerical simulations across a wide range of Reynolds numbers (Re =1-10,000) to analyse roughness-sensitive flow and transport quantities such as friction factors, centreline paths, velocity distributions, effective cross-sectional areas, and residence times in karst conduits. These quantities are essential for upscaling to network-scale models without resolving each conduit in detail. The geometries are reconstructed from LiDAR-based point clouds of real karst systems, preserving wall roughness and structural irregularities. An immersed boundary method combined with ray tracing ensures accurate wall representation.

Simulations reveal coexistence of different flow regimes depending on local features, with early transition at Re ≤ 1000. Velocity profiles are highly non-uniform, with localised separation driven by wall irregularities. Particle tracking shows accumulation in diverging regions and strong tailing in breakthrough curves. These results highlight the limitations of standard friction factor models and the need for new approaches that account for geometric complexity in heterogeneous wall-bounded flow systems.