Purpose:　The　macropore　structure　and　seepage　characteristics　profoundly　influence　the　stability　of　granite　residual　soil　(GRS)　slopes.　However,　accurately　predicting　the　permeability　of　undisturbed　GRS　(U-GRS)　is　challenging　owing　to　its　complex　and　susceptible　pore　structure.　Aims　and　methods:　Employing　X-ray　computed　tomography　(CT)　technologies,　a　three-dimensional　(3D)　pore　structure　of　U-GRS,　was　established.　Permeability　prediction　for　U-GRS　samples　was　conducted　using　three　simulation　methods,　namely,　the　pore　network　model　(PNM),　finite　element　method　(FEM),　and　the　lattice　Boltzmann　method　(LBM),　along　with　two　empirical　models　(EMs)—specifically,　Kozeny–Carman　(K–C)　and　Katz–Thompson　(K–T)　models.　Subsequently,　the　methods　were　comparatively　analyzed　for　calculating　efficiency　and　accuracy.　Finally,　a　piecewise　permeability　prediction　model　(PPPM)　for　U-GRS　based　on　the　CT-LBM　was　proposed.　Results:　The　ranking　of　permeability　estimation　methods　in　terms　of　accuracy　was　as　follows:　LBM　＞　PNM　＞　FEM　＞　EMs.　Substantial　disparity　was　observed　in　the　permeabilities　obtained　using　both　FEM　and　EMs　compared　to　other　methods,　which　exhibited　a　deviation　of　up　to　six　orders　of　magnitude.　The　PPPM　demonstrated　smaller　prediction　deviations　than　the　EMs,　with　its　accuracy　influenced　by　the　strategy　for　selecting　calculation　parameters.　Conclusion:　The　CT-LBM,　which　uses　real　pore　structures,　was　employed　to　estimate　the　permeability　of　U-GRS.　The　PPPM,　established　based　on　this　method,　was　found　to　be　applicable　for　estimating　U-GRS　permeability.　Graphical　Abstract:　(Figure　presented.).　©　The　Author(s),　under　exclusive　licence　to　Springer-Verlag　GmbH　Germany,　part　of　Springer　Nature　2024.